Magnesium sensing via LFA-1 regulates CD8+ T cell effector function Jonas Lötscher1, Adrià-Arnau Martí i Líndez1,2,3,19, Nicole Kirchhammer4,19, Elisabetta Cribioli5,6,20, Greta Giordano5,6,20, Marcel P. Trefny4,20, Markus Lenz7, Sacha I. Rothschild8,9, Paolo Strati10, Marco Künzli11, Claudia Lotter12, Susanne H. Schenk12, Philippe Dehio1, Jordan Löliger1, Ludivine Litzler11, David Schreiner11, Victoria Koch4, Nicolas Page13, Dahye Lee1, Jasmin Grählert1, Dmitry Kuzmin14,15, Anne-Valérie Burgener1, Doron Merkler13, Miklos Pless9,16, Maria L. Balmer1,17,18, Walter Reith2, Jörg Huwyler12, Melita Irving5,6, Carolyn G. King11, Alfred Zippelius4,8, Christoph Hess1,3,21, * 1 Department of Biomedicine, Immunobiology, University of Basel and University Hospital of Basel, 4031 Basel, Switzerland 2 Department of Pathology and Immunology, Faculty of Medicine, University of Geneva, 1211 Geneva, Switzerland 3 Department of Medicine, CITIID, Jeffrey Cheah Biomedical Centre, University of Cambridge, Cambridge CB2 0AW, United Kingdom 4 Department of Biomedicine, Cancer Immunology, University of Basel and University Hospital of Basel, 4031 Basel, Switzerland 5 Ludwig Institute for Cancer Research, University of Lausanne, 1066 Epalinges, Switzerland 6 Department of Oncology, University Hospital of Lausanne, 1011 Lausanne, Switzerland 7 University of Applied Science Northwestern Switzerland, Institute for Ecopreneurship, 4132 Muttenz, Switzerland 8 Division of Medical Oncology and Comprehensive Cancer Center, University Hospital Basel, 4031 Basel, Switzerland 9 Swiss Group for Clinical Cancer Research, 3008 Bern, Switzerland 10 Department of Lymphoma and Myeloma, University of Texas MD Anderson Cancer Center, 77030 Houston, TX, USA 11 Department of Biomedicine, Immune Cell Biology, University and University Hospital of Basel, 4031 Basel, Switzerland 12 Department of Pharmaceutical Sciences, Pharmaceutical Technology, University of Basel, 4056 Basel, Switzerland 13 Department of Pathology and Immunology, University of Geneva, Geneva, Switzerland 14 Hornet Therapeutics Ltd, London SW1Y 5ES, UK 15 Department of Medical Oncology, Yale School of Medicine, 06510 New Haven, CT, USA 16 Department of Oncology, Cantonal Hospital Winterthur, 8400 Winterthur, Switzerland 17 Department for Biomedical Research (DBMR), University Clinic for Diabetes, Endocrinology, Clinical Nutrition and Metabolism, Inselspital, University of Bern, 3008 Bern, Switzerland 18 Diabetes Center Berne (DCB), 3010 Bern, Switzerland 19 These authors contributed equally 20 These authors contributed equally 21 Lead contact * Correspondence: chess@uhbs.ch SUMMARY The relevance of extracellular magnesium in cellular immunity remains largely unknown. Here, we show that the co-stimulatory cell-surface molecule LFA-1 requires magnesium to adopt its active conformation on CD8+ T cells, thereby augmenting calcium flux, signal transduction, metabolic reprogramming, immune syn- apse formation, and, as a consequence, specific cytotoxicity. Accordingly, magnesium-sufficiency sensed via LFA-1 translated to the superior performance of pathogen- and tumor-specific T cells, enhanced effec- tiveness of bi- specific T cell engaging antibodies, and improved CAR T cell function. Clinically, low serum magnesium levels were associated with more rapid disease progression and shorter overall survival in CAR T cell and immune checkpoint antibody-treated patients. LFA-1 thus directly incorporates information on the composition of the microenvironment as a determinant of outside-in signaling activity. These findings conceptually link co- stimulation and nutrient sensing and point to the magnesium-LFA-1 axis as a therapeu- tically amenable biologic system. GRAPHICAL ABSTRACT 2 INTRODUCTION Low dietary Mg2+ intake and hypomagnesemia have been implicated in the pathophysiology of various diseases, including infection and cancer (Costello and Nielsen, 2017; Larsson et al., 2012; Qu et al., 2013; Ravell et al., 2018; Sakaguchi et al., 2014; Saris et al., 2000; Schulze et al., 2007; Sojka and Weaver, 1995). Only limited experimental data are available exploring how organismal Mg2+ abundance may affect the immune system. It has been reported, however, that in mice fed a Mg2+-deficient diet, (1) metastatic spread of carcinoma cells was accelerated (Nasulewicz et al., 2004), and (2) immune responses against influenza were impaired due to insufficient inducible T cell kinase (ITK) activity (Kanellopoulou et al., 2019). Whether in these models an effect of extracellular Mg2+ was contributing to shaping adaptive immune responses has not been assessed. CD8+ T cells are a key component of the adaptive immune system, playing a critical role in recognizing and eliminating infected or malignantly transformed cells. Immune surveillance of peripheral tissues by effector- memory (EM) CD8+ T cells is a challenging task, requiring adaptation to constantly changing microenvironments with highly variable nutrient and oxygen content (Bantug et al., 2018; Lötscher and Balmer, 2019). Whether and, if so, how Mg2+ availability in tissues is sensed and integrated functionally by CD8+ T cells has not been investigated. RESULTS First, we explored how Mg2+ affected memory CD8+ T cell function in vitro, using complete media containing 1.2 mM Mg2+ versus no Mg2+ as the only difference. Analysis of metabolic flux profiles revealed that rapid induction of glycolysis of human EM CD8+ T cells (defined as CD45RA and CD62L) activated by injection of soluble anti-CD3 and anti-28 monoclonal antibodies (mAb) was blunted in the absence of Mg2+. This deficit was reverted when adding back Mg2+ just prior to activation (Figure S1A). By contrast, Mg2+ did not affect activation-induced upregulation of glycolysis in naive CD8+ T cells (defined as CD45RA+ and CD62L+; Figure S1B). Next, we assessed the dose effect of Mg2+ in EM CD8+ T cells upon stimulation with anti-CD3 and CD28 mAb, staining for markers of early and late activation (CD69 and CD25, respectively), metabolic adaptation (CD71, CD98), and degranulation (CD107a). These experiments demonstrated a strict Mg2+ dose response, plateauing at 0.6– 1.2 mM, that is, in the physiologic range of plasma concentrations reported for Mg2+ (Figure S1C) (Akizawa et al., 2008; Costello et al., 2016; Lowenstein and Stanton, 1986; Zhan et al., 2014). The difference in the capacity of memory CD8+ T cells to upregulate these phenotypic markers in the presence versus absence of Mg2+ is summarized in Figure S1D, left panel. Cytokine release followed the same pattern, with decreased production under Mg2+-depleted conditions in EM cells (Figure S1D, middle panel). EM CD8+ T cell viability was not affected by the lack of Mg2+ in the culture media (Figure S1D, right panel). Analogous to induction of glycolysis, no Mg2+- related effect was observed for naive CD8+ T cells for each of these measures (Figure S1E, left and middle panels), and also, in naive cells, viability was not affected by the lack of Mg2+ in culture media (Figure S1E, right panel). The immediacy with which Mg2+ affected reactivity of EM CD8+ T cells (Figure S1A) suggested the possibility that a Mg2+-sensitive target (or targets) expressed on the cell surface could (co-)affect T cell function. Clustering of activated T cell blasts was also reduced in the absence of Mg2+, lending further support to the idea that Mg2+ was impacting a target (or targets) on the cell surface (Figure S1F). This notion was also supported by the lack of 3 A B C Immune Leukocyte synapse adhesion LFA-1 expression unstimulated (199) (216) 6 183 191 5 **** 3 2 **** ****4 6784 0 1 105 naive CD8 1 2EM CD8 15 ITGAL 5 PHA-Blast 1 1 0 10 2 10 3 10 4 10 5 10 6 naive EM PHA CD11a - FITC CD8 CD8 Blast Metal ion 36 Differentially binding expressed (6841) (memory>naïve)(152) D PHA-Blast: PHA-Blast: PHA-Blast: PHA-Blast: PHA-Blast: Total LFA-1 (TS2/4) LFA-1 extension (KIM127) Open headpiece (m24) Phospho-FAK Phospho-ERK1/2 25 8.0 60 60 15 stim. 20 unstim. 6.0 40 40 15 10 10 4.0 20 20 5 5 2.0 0 0 0 1.2 0.6 .12 .06 2 001 1. 2 .6 12 06 2 0 .2 .6 2 6 2 0 2 6 2 6 2 0 .2 .6 2 6 2 00 0 . 0 0. 0.0 0.0 1 1 0 .1 .0 01 1. 0. .1 .0 01 1 0 0.1 0.0 .010 0 0. 0 0 0. 0 mM Mg2+ mM Mg2+ mM Mg2+ mM Mg2+ mM Mg2+ PHA-Blast: PHA-Blast: PHA-Blast: PHA-Blast: Glucose uptake TNF CD107a CD69 80 6.0 stim. 20 60 unstim. 60 15 4.040 40 10 20 2.020 5 0 0 0.0 1.2 0.6 .12 .06 12 0 1.2 0.6 .12 .06 12 0 1.2 0.6 .12 .06 12 00 0 . 2 .6 2 6 2 0 0 0 0. 0 0 0. 0 0 .0 1 0 0.1 0.0 .010 0 mM Mg2+ mM Mg2+ mM Mg2+ mM Mg2+ E PHA-Blast: FAK phosphorylation F PHA-Blast: LFA-1 extension G PHA-Blast: open headpiece upon TCR stimulation ± BIRT377 and ± Mn2+ ± BIRT377 and ± Mn2+ 80 2.5 **** 1.6**** *** **** **** *** **** **** CD3/28 - 1.2 mM 2.0 ** *** 60 CD3/28 - 0 mM 1.5 ** 1.4 ** **** 1.2 40 1.2 1.0 1.0 20 0.8 0.8 0 0.6 0.6 ol 7 3 + + 2+r 7 4 18 /2 2 rol 77 43 18 /2 2 mM Mg mM Mg 2+ nto RT 3 1 / 1 n tV 3 1 / 1 n C I X TS 1 A- n 1 - LF M TCo IR XV TS LF A M B R B R - + + + + + CD3/28 - + + + + + CD3/28 CB CB - - + - - + BIRT377 - - + - - + BIRT377 H PHA-Blast: FAK phosphorylation I PHA-Blast: Glycolytic activity J PHA-Blast: Degranulation ± BIRT377 and ± Mn2+ ± BIRT377 and ± Mn2+ ± BIRT377 and ± Mn2+ 50 **** 50 25 **** 1.2**** 0 1.2 mM**** 0 mM **** **** **** 40 **** 0 1.2 mM + BIRT377 200 Mn2+ * 40 0 Mn2+30 + BIRT377 ** 0.815 ** ** 20 10 30 0.4 10 5 0 20 0 0.0 mM Mg2+ 0 50 100 150 mM Mg2+mM Mg2+ Time (minutes) - + + + + + CD3/28 − + − − + BIRT377 - + + + + + CD3/28 - - + - - + BIRT377 - - + - - + BIRT377 Figure 1 4 % phospho-FAK high gMFI 2-NBDG (x10^3) gMFI TS2/4 (x10^4)397 % phospho-FAK397 high 1.2 1.2 1.2 0 0 + Mn 0 2+ + Mn 2+ % TNF positive gMFI KIM127 (x10^3) ECAR (mpH min-1) ± Mg 2+o ± r B MIR n 2+ C T3D 73 7/28 % CD107a positive gMFI m24 (x10^3) gMFI - KIM127 (x10^5) 1.2 gMFI CD11a (x10^4) 1.2 ECAR (mpH min-1) 1.2 0 1 0. 0 2 0 + 1 M.2 n0 2+ + Mn 2 gMFI CD69 (x10^4)0 + % phospho-FAK397 high 0 Mn 0 2+ Mn 2+ gMFI - m24 (x10^3) gMFI - CD107a (x10^4) 1.2 1 1. .2 2 % phospho-ERK1/2 pos. 1 1 .. 22 1 0.2 0 + M 0 n0 2+ 0 ++ MM n 2n 0 2 + + + Mn 2+ Figure 1. Sensing of extracellular magnesium through LFA-1 shapes T cell activation (A) Venn diagram of GO-term lists and differentially expressed proteins between CD8+ T cell subsets (memory > naive). (B) Representative CD11a expression profile on human naive and EM CD8+ T cells as well as human PHA blasts as determined by flow cytometry (left panel) and a summary bar graph (right panel). (C) Schematic of LFA-1 conformational states and antibodies used to probe them. (D) Flow cytometric analysis of human PHA blasts ± TCR stimulation at indicated Mg2+ concentrations, detecting LFA-1 expression independent of conformational state (TS2/4 binding: first panel, upper row); extended LFA-1 (KIM127 binding: second panel, upper row); LFA-1 open headpiece conformation (m24 binding: third panel, upper row); phosphorylation of FAK (fourth panel, upper row); phosphorylation of ERK1/2 (fifth panel, upper row); uptake of 2-NBDG, indicating glucose/nutrient uptake (first panel, lower row); TNF production (second panel, lower row); degranulation as indicated by upregulation of CD107a (third panel, lower row); and expression of the activation marker CD69 (forth panel, lower row). PHA blasts established from n = 5–7 healthy donors were used. (E) FAK phosphorylation in PHA blasts after activation in medium ± Mg2+ using BIRT377 (a I allosteric antagonist), XVA143 (a/b I-like allosteric antagonist), TS1/18 (blocking antibody-binding b2 chain), CBR-LFA1/2 (activating antibody-binding b2 chain), or Mn2+. The mean signal of each condition was compared with the mean control signal of 1.2 or 0 mM Mg2+, respectively. (F–J) Assessment of human PHA blasts in medium ± Mg2+; ± Mn2+; ± BIRT377, as indicated, with regard to activation-induced LFA-1 extension (KIM127 binding) (F); headpiece opening (m24 binding) (G); FAK phosphorylation (H); glycolytic activity— metabolic flux traces from a representative experiment (left panel), summary (right panel) (I), and degranulation (J). Each symbol represents an individual healthy donor. Data are presented as mean ± SD, except left panel of (I), where symbols indicate mean ± SEM. Statistical significance was assessed by one-way ANOVA with Sidak’s multiple comparison test (B and E–J). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 5 evidence for rapid Mg2+ flux into Mg2+-deprived CD8+ T cells, which was in contrast to Ca2+ flux tested in the same setting (Figure S1G). Guided by these phenotypic and functional data, we overlaid genes from gene ontology (GO) lists in a Venn diagram, inputting the following terms: metal ion binding (parent term of Mg2+ binding), immune synapse (CD3/28 activation), and leukocyte adhesion (clustering experiments), together with a published list of proteins preferentially expressed by EM over naive CD8+ T cells (differential impact of Mg2+) (van Aalderen et al., 2017). This approach identified ITGAL (the gene encoding the integrin subunit alpha L chain, CD11a) as a candidate gene plausibly linked to the input criteria (Figure 1A). CD11a combines with CD18 to form the leukocyte function-associated antigen 1 (LFA-1). LFA-1 is an integrin that is involved in T cell activation via immune synapse formation as well as in leukocyte trafficking and extravasation (Hogg et al., 2011). Upon TCR stimulation, inside-out signaling mediates LFA-1 conformational changes, which regulate LFA-1 ligand-binding affinity and the molecule’s capacity for outside-in signaling. Binding of Mg2+ to metal-ion-dependent adhesion sites (MIDAS), one each on CD18 and CD11a, impacts these conformational changes. Specifically, MIDAS binding of Mg2+ modulates the interchain allosteric relay (stalk extension) and external ligand affinity (headpiece conformation) (Schürpf and Springer, 2011; Zhang and Chen, 2012). The overall expression of LFA-1, which can be assessed using mAb TS2/4, was lowest on naive CD8+ T cells, higher on EM CD8+ T cells, and highest on T cell blasts (Figure 1B). Memory cell-selective Mg2+ dependency of metabolic reprogramming and activation and Mg2+ dependency of clustering of T cell blasts thus aligned with the selection of ITGAL in our GO-term integration approach. The molecular conformation of LFA-1 can be probed by staining conformational epitopes. Specifically, KIM127 recognizes extended LFA-1, whereas m24 binds LFA1 when the headpiece is in its open position (Figure 1C). These probing antibodies were used to assess how LFA-1 extension and headpiece opening related to activation (i.e., inside-out signaling) across a range of Mg2+ concentrations in T cell blasts. Cell-surface expression of LFA-1, irrespective of conformation (i.e., mAb TS2/4 staining), was similar in nonactivated and activated cells and independent of Mg2+ availability (Figure 1D, first panel, upper row). LFA-1 extension (KIM127 binding) was detected on activated cells only and regulated by Mg2+ (Figure 1D, second panel, upper row). LFA-1 headpiece opening (m24 binding) was likewise detected on activated cells only and also modulated by Mg2+ (Figure 1D, third pane, upper row). The effect of Mg2+ on LFA-1 headpiece opening was also observed in primary human CD8+ T cells (Figure S1H). In a Mg2+-modulated manner, T cell activation further induced proximal and distal signaling activity (focal adhesion kinase (FAK) and extracellular signal-regulated protein kinase 1/2 (ERK1/2) phosphorylation, respectively) (Giannoni et al., 2003; Schaller et al., 1994), nutrient (2-NBDG) uptake, TNF production, and expression of CD107a and CD69 (Figure 1D, fourth and fifth panels, upper row and lower row). We then aimed to differentiate whether extension or headpiece opening was the critical event enabling LFA-1 outside-in signaling. To that end, effects of LFA-1 inhibitors (XVA143, BIRT377, and TS1/18) and LFA-1 activators (CBR LFA-1/2 and Mn2+) were tested regarding proximal LFA1 outside-in signaling (FAK phosphorylation) in activated T cell blasts, both in the presence and absence of Mg2+. In the presence of Mg2+, XVA143 (allosteric antagonist inhibiting the ligand-binding I domain while stabilizing extended conformation), BIRT377 (allosteric inhibitor stabilizing bent conformation), and the mAb TS1/18 (allosteric inhibitor stabilizing the closed headpiece) all suppressed activation-induced FAK phosphorylation to levels that were observed when Mg2+ was lacking (Figure 1E). Extending LFA-1 with the activating mAb CBR LFA-1/2 in the presence of Mg2+ increased FAK 6 phosphorylation above the level observed in the presence of Mg2+ alone, and yet, extension in the absence of Mg2+ did not increase LFA-1 outside-in signaling (Figure 1E). Mn2+, which binds both MIDAS, functions as a universal activator of integrins (Dransfield et al., 1992b; San Sebastian et al., 2006; Vorup-Jensen et al., 2007), was largely compensating for the absence of Mg2+ in terms of FAK phosphorylation (Figure 1E). Of note, Mn2+ did not affect the cell-intrinsic glycolytic or mitochondrial functional capacity of T cell blasts - core metabolic activities that were also similar in presence versus absence of Mg2+ in the culture media (Figures S1I and S1J). From these experiments, we concluded that extension was required but not sufficient to trigger FAK phosphorylation, and that headpiece opening - which was fully dependent on Mg2+ (Figure 1D) - was necessary for FAK phosphorylation to occur. Birt377 stabilizes LFA-1 in its bent state by eliminating the basal interaction between CD11a and CD18 (Moore et al., 2018; Salas et al., 2004). Therefore, adding Mn2+, which binds to the beta I-MIDAS with high affinity, is expected to extend LFA-1 even in the presence of Birt377 while keeping the headpiece closed - which is indeed what we observed (Figures 1F and 1G). In this scenario, the extension of LFA- 1 did not drive FAK phosphorylation, corroborating the notion that headpiece opening was a requirement for FAK phosphorylation to be initiated, while extension alone was not sufficient (Figure 1H). The same applied to downstream events regulated by LFA-1, namely activation-induced glycolysis and degranulation (Figures 1I and 1J). Intercellular adhesion molecules (ICAMs) are the ligands of LFA-1, and high-affinity LFA-1-ICAM interactions promote immune synapse formation. LFA-1 can also interact with ICAMs in cis (Fan et al., 2016), as well in homotypic cell-cell interactions (Fan et al., 1993; Sabatos et al., 2008). To define how ICAM binding contributed to LFA-1 outside-in signaling in the presence or absence of Mg2+ in our in vitro system, we assessed the surface expression of ICAMs on T cells. On T cell blasts, expression of ICAM-3 was highest, followed by that of ICAM-1 and ICAM-2 (Figure S1K). Primary human naive CD8+ T cells expressed less ICAM-3 than EM CD8+ T cells; expression of ICAM-3 on EM CD8+ T cells and T cell blasts was similar (Figure S1L). We then used blocking antibodies to prevent ICAMs from interacting with LFA-1, both in the presence and absence of Mg2+. Blocking of ICAM-3 moderately reduced LFA-1 headpiece opening upon activation of T cell blasts (Figure S1M) and decreased the interlinked upregulation of CD69 (Figure S1N). These experiments established Mg2+ as an important regulator of EM CD8+ T cell function. Mg2+ was further identified to impact both extension and headpiece opening of LFA-1 on activated T cells, with headpiece opening being critical to outside-in signaling and downstream T cell activation. Immune synapse formation of expanded tumor-specific T cells is regulated via the Mg2+-LFA-1 axis Next, we generated NY-ESO-1-specific T cells using a rapid expansion protocol (REP T cells) (Dudley et al., 2002; Thomas et al., 2018). LFA-1 surface expression on human NY-ESO-1specific REP T cells was as high as on T cell blasts (PHA blasts) (Figure 2A). REP T cell killing activity was augmented by Mg2+ across a spectrum of target peptide (9c) concentrations used for pulsing tumor target cells - most efficiently so at low 9c concentrations (Figure S2A). Also in this antigen-specific setting, Mg2+ was required for LFA-1 headpiece opening, and extension of LFA-1 in the absence of headpiece opening was insufficient to initiate FAK phosphorylation, degranulation, and target cell killing (Figure 2B), as well as ERK1/2 as well as c-Jun activity (Figures S2B and S2C). Engagement of ERK1/2 and c-Jun through the LFA-1-Mg2+ regulatory axis was confirmed by immunoblot analyses in both 7 A B REP T cells and T2 target cells co-culture: LFA-1 LFA-1 FAK extension open headpiece phosphorylation Degranulation Cytotoxicity LFA-1 expression 15 ** *** 80 4.0 **** 25 *** ** * *** *** *** ** *** ** ** 80 ** * 60 20 naive CD8 10 3.0 6015 EM CD8 2.0 40 10 40 PHA-Blast 5 1.0 20 5 20 REP T cell 3 4 5 6 7 10 10 10 10 10 0 0.0 0 0 0 CD11a - FITC 2+g g2 + n2 + 77 2+ 2+ + 7 + + + 7g g n2 7 2 2 2 7 2 + 2+ 2+ 77 g2 + 2+ 2+ 7 M M M T3 M M M T3 Mg Mg Mn 3 g g n g n 3 7 M M + 3 IR M M + IR M M + IR T M M M T M + M IR M M M M T m m M B M + IR .2 0 m + .2 m m M B m m M B0 m m M B m m M B 1 0 M 1 0 m +M 1.2 0 m M + .2 0 m M + 0 1 0 1.2 0 0 m + 2 m m M . .2 m m m 1 1 1.2 1.2 1.2 C LFA-1: D Merge open headpiece Phospho-Tyrosine Merge Centrosome Perforin unpulsed unpulsed 1.2 mM 1.2 mM unpulsed unpulsed 0 mM 0 mM 9c 9c 1.2 mM 1.2 mM 9c 9c 0 mM 0 mM 9c 9c 0 + Mn2+ 0 + Mn2+ 9c 9c 1.2mM 1.2mM + BIRT377 + BIRT377 REP T cells: open headpiece REP T cells: Tyrosine phosophorylation REP T cells: Perforin polarization REP T cells: Centrosome polarization 2.0 **** **** 1.05 ** 1.05 n.s. ******** 400 ********1.5 ** ** 1.00 * n.s. 300 1.00 0.95 1.0 200 0.900.95 0.5 0.85 100 0.0 0.90 0.80 1.2 0 1.2 0 0 1.2 mM Mg2+ 0 1.2 1.2 0 0 1.2 mM Mg2+ 1.2 0 1.2 0 0 1.2 mM Mg2+ 1.2 0 1.2 0 0 1.2 2+ + Mn2+ + Mn2+ 2+ mM Mg + Mn + Mn2+ - - + + + + 9c peptide - - + + + + 9c peptide - - + + + + 9c peptide - - + + + + 9c peptide - - - - - + BIRT377 - - - - - + BIRT377 - - - - - + BIRT377 - - - - - + BIRT377 Figure 2 8 m24 objects T cell-1 gMFI KIM127 (x10^4) pan-phospho Tyrosine signal itensity in T cell gMFI m24 (x10^4) median of polarization ratio per field of view % phospho-FAK397 high % LAMP1 pos. median of polarization ratio per field of view % Caspase-3 pos.T2 cells Figure 2. The magnesium-LFA-1 axis regulates immune synapse formation of tumor-specific human T cells (A) Representative histogram of CD11a expression on human naive and EM CD8+ T cells, PHA blasts and NY-ESO-1-specific REP T cells. (B) Flow cytometric assessment of REP T cells after incubation with T2 target cells that were pulsed with cognate peptide. REP T cells were analyzed with regard to LFA-1 extension (first panel), LFA-1 headpiece opening (second panel), FAK phosphorylation (third panel), and degranulation (fourth panel). In the fifth panel, co-culture-induced caspase-3 activity in T2 target cells (i.e., apoptosis) is shown. Representative of n = 2 independent experiments with n = 5–7 healthy donors; fourth panel n = 1 healthy donor and 4 technical replicates. (C) Upper panel, representative confocal projections of REP T cells co-cultured with T2 target cells, labeled for headpiece- open LFA-1, and pan-tyrosine phosphorylation. Scale bars indicate 10 mm. Lower left panel, quantitation of headpiece-open LFA-1 within confocal 3D stacks (symbol represents an individual field of view). Lower right panel, quantitation of median fluorescence intensity of tyrosine phosphorylation in REP T cells (n = 203–344 cells per condition, pooled from n = 1–2 field of views). (D) Upper panel, representative confocal projections of REP T cells co-cultured with T2 target cells. Cells were labeled for g- tubulin (centrosome) and perforin. Lower left panel, quantitation of perforin, lower right panel, quantitation of centrosome polarization (symbol represents an individual field of view). Data are presented as mean ± SD, except for the lower right panel of (C), where data are presented as median ± IQR. Statistical significance was assessed by repeated- measures one- way ANOVA with Sidak’s multiple comparison test (B), one-way ANOVA with Sidak’s multiple comparison test for the lower left panel of (C and D), and lower right panel of (D). Kruskal-Wallis test with Dunn’s multiple comparison test was used for the lower right panel of (C). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n.s., nonsignificant. 9 human T cell blasts (Figures S2D and S2E), as well as primary human EM CD8+ T cells (Figure S2F). We then visually explored immune synapse formation between NY-ESO-1-specific T cells and tumor cells. These experiments identified early antigen-specific LFA-1 headpiece opening at sites of close physical interaction between REP T cells and tumor targets, which was Mg2+ dependent, Mn2+ rescuable, and Birt377 sensitive (Figure 2C). Phospho-tyrosine signal intensity at REP T cell-target cell interaction sites also mirrored LFA-1 headpiece opening across all conditions (Figure 2C). Maturation of the immune synapse is characterized by centrosome polarization and focused delivery of cytolytic granules, two events that the Mg2+-LFA-1 axis also contributed to (Figure 2D, imaging performed after 35–45 min of REP T cells target cell co-incubation). LFA-1 is essential for Mg2+ to shape cytotoxic T cell activity and immune synapse formation, in vitro To genetically probe the suggested Mg2+-LFA-1 regulatory axis, experiments using LFA-1 deficient (LFA-1-/-) cytotoxic T cells (CTLs) and LFA-1-/- Jurkat T cells were performed (validation of k.o.; Figure S3A). First, we repeated metabolic flux analyses using LFA-1-/- CTLs. These experiments confirmed LFA-1 as the mediator through which Mg2+ exerted its effect on glycolysis (Figure 3A). 2-NBDG (glucose/nutrient) uptake assays - which were complemented with Mn2+ and BIRT377 - also aligned with the metabolic flux studies (Figure S3B). TCR stimulation mediates influx of Ca2+ into the cytosol, where it functions as a second messenger (Trebak and Kinet, 2019). Depletion of extracellular Mg2+ blunted early (white bar) as well as sustained (black bar) influx of Ca2+ in wild-type (WT) OT-I CTLs, a finding phenocopied independently of extracellular Mg2+ in LFA-1-/- OT-I CTLs (Figure 3B). The same requirement for LFA-1 for both early and late Ca2+ flux was also observed in WT versus LFA-1-/- Jurkat cells (Figure S3C). As a proxy for early immune synapse-initiated signaling, we further analyzed TCR ligation-induced tyrosine phosphorylation, 2 min postactivation using confocal microscopy. In LFA-1-/- Jurkat T cells, TCR-induced tyrosine phosphorylation was markedly reduced, independently of extracellular Mg2+ (Figure S3D). In WT Jurkat T cells, Mg2+ restriction had the same effect (Figure S3D). Aligning with these genetic data, in anti-CD3 mAb activated T cell blasts, phospho-tyrosine signaling was augmented by Mg2+ in a BIRT377-sensitive manner, and the absence of Mg2+ was partially rescued by Mn2+ (Figure S3E). LFA-1 further recruits and amplifies TCRmediated ERK1/2 signaling at the immune synapse (Li et al., 2009) and Mg2+ and partially Mn2+ increased also ERK1/2 activity in CTLs in an LFA-1-dependent manner (Figure 3C). Of note, Mg2+ and Mn2+ per se did not induce p-ERK1/2 in unstimulated T cells, nor did genetic deletion or LFA-1 negatively affect pERK1/2 signaling upon supra-physiologic activation with PMA and ionomycin (Figure S3F). We then investigated the impact that the Mg2+-LFA-1 system had on more downstream events in activated CTLs. CD69 and degranulation were similarly induced in activated LFA-1 competent and LFA-1-/- CTL cells; yet, as expected, only in LFA-1 competent cells they were modulated by Mg2+, and partially also by Mn2+ (Figures 3D and 3E). Specific cytotoxic activity in contrast - which is a highly coordinated process requiring timed and polarized release of cytotoxic granules - was severely blunted in LFA-1-/- CTLs, while regulated via LFA-1 in a Mg2+ dependent, partially Mn2+ rescuable, and Birt377-sensitive manner in LFA-1 competent counterparts (Figure 3F). This suggested that both high-affinity binding of Mn2+ to the LFA-1 MIDAS and lack of LFA-1 were associated with dysregulated immune cell function, plausibly due to uncoordinated immune synapse formation in both scenarios. Using OVA variant peptides (affinity for OT-I: G4 < H7 < R7), the importance of the Mg2+-LFA-1 system in regulating proximal LFA-1-driven 10 A Murine CTLs: Glycolytic activity B Murine OTI: Calcium Flux 30 120 WT: 1.2 mM *** 6000 ****CD3/28 LFA-1-/-: 1.2 mM 10 OVA**** WT: 1.2 mM LFA-1 -/-: 1.2 mM -/- -/- ****WT: 0 mM LFA-1 : 0 mM WT: 0 mM LFA-1 : 0 mM 100 **** 8 n.s. 20 n.s. 4000 6 80 4 60 10 2000 2 40 0 0 0 0 50 100 150 1.2 0 1.2 0 mM Mg2+ 0 200 400 600 800 1.2 0 1.2 0 mM Mg2+ Time (minutes) WT LFA-1 -/- Time (sec) WT LFA-1-/- C Murine CTLs: Phospho-ERK1/2 D Murine CTLs: CD69 **** n.s.WT LFA-1-/- 50 WT LFA-1-/- **** ***** n.s. 3.0 **** n.s. 40 **** n.s. **** n.s. 1.2 mM 1.2 mM 30 0 mM 0 mM 1.5 20 0 + Mn2+ 0 + Mn2+ 1.2 mM 10 1.2 mM + BIRT377 4 5 4 50 10 10 0 10 10 + BIRT377 3 4 5 3 4 510 10 10 10 10 10 0 0.0 p-ERK1/2 - AF647 WT LFA-1-/- CD69 - APC WT LFA-1-/- CD3/28 CD3/28 E Murine CTLs: Degranulation F Murine CTLs: Cytotoxicity 40 **** n.s. 100 **** WT LFA-1-/- **** n.s. WT LFA-1-/- **** 30 **** n.s. 80 * n.s. n.s. 1.2 mM 1.2 mM 60 20 n.s. 0 mM 0 mM 40 0 + Mn2+ 10 0 + Mn2+ 1.2 mM 1.2 mM 20 + BIRT377 4 5 4 50 10 10 0 10 10 + BIRT377 3 4 5 6 3 4 5 610 10 10 10 10 10 10 10 0 0 CD107a - Pe/Cy7 WT LFA-1-/- Caspase-3 activity - AF488 WT LFA-1-/- CD3/28 PHA Murine OTI co-culture: Phospho-FAK G 50 WT: 1.2 mM unpulsed G4 H7 R7 OVA WT: 0 mM 40 LFA-1-/-: 1.2 mM 30 n.s. WT: 1.2 mM LFA-1-/-: 0 mM WT: 0 mM 20 LFA-1-/-: 1.2 mM 10 LFA-1-/-: 0 mM 10 4 10 5 10 6 10 4 10 5 10 6 10 4 10 5 10 6 10 4 10 5 10 6 10 4 10 5 10 6 0 p-FAK - AF488 - G4 H7 R7 SIINFEKL Murine OTI co-culture: Phospho-ERK1/2 H WT: 1.2 mM unpulsed G4 H7 R7 OVA 60 WT: 0 mM LFA-1-/-: 1.2 mM -/- WT: 1.2 mM 40 LFA-1 : 0 mM WT: 0 mM LFA-1-/-: 1.2 mM 20 LFA-1-/-: 0 mM 0 10 4 10 5 0 10 4 10 5 0 10 4 10 5 0 10 4 10 5 0 10 4 10 5 0 p-ERK1/2 - AF647 - G4 H7 R7 SIINFEKL I Murine OTI co-culture: Degranulation 60 WT: 1.2 mM unpulsed G4 H7 R7 OVA WT: 0 mM 40 LFA-1 -/-: 1.2 mM WT: 1.2 mM LFA-1-/-: 0 mM WT: 0 mM 20 LFA-1-/-: 1.2 mM LFA-1-/-: 0 mM 0 10 3 10 4 10 5 10 6 0 10 3 10 4 10 5 10 6 0 10 3 10 4 10 5 10 6 0 10 3 10 4 10 5 10 6 0 10 3 10 4 10 5 10 6 0 LAMP1 - AF647 - G4 H7 R7 SIINFEKL J Murine OTI co-culture: Cytotoxicity 100 WT: 1.2 mM unpulsed G4 H7 R7 OVA WT: 0 mM 80 LFA-1-/-: 1.2 mM n.s. WT: 1.2 mM 60 LFA-1-/-: 0 mM WT: 0 mM 40 LFA-1-/-: 1.2 mM 20 LFA-1-/-: 0 mM 10 3 10 4 10 5 10 6 10 3 10 4 10 5 10 6 10 3 10 4 10 5 10 6 10 3 10 4 10 5 10 6 10 3 10 4 10 5 10 6 0 Figure 3 Caspase-3 activity - AF488 - G4 H7 R7 SIINFEKL 11 ECAR (mpH min-1) % CD107a pos. % phospho-ERK1/2 pos. ECAR (mpH min-1) MFI - Calbryte (Ft F -10 ) % Caspase-3 pos. EL4 cells % LAMP1 pos. % phospho-ERK1/2 pos. % phospho-FAK 397 high % Caspase-3 pos. EL4 cells gMFI - CD69 (x10^4) AUC (Calbryte Ft F -10 ) **** **** **** **** * ** **** **** **** **** Figure 3. LFA-1 is required to regulate T cell activation and cytotoxicity, in vitro (A) Glycolytic activity upon activation of murine WT and LFA-1-/- CTLs. Representative experiment (left panel), summary of DECAR (basal versus max.) (right panel) from n = 4 mice. (B) TCR stimulation induced calcium flux in WT and LFA-1-/- OT-I CTLs. Representative trace (left panel), and quantification of area under the curve (AUC) of individual wells (right panel). Representative of n = 3 independent experiments. (C–E) Representative histograms (left panels) and quantified results from n = 3 mice with three technical replicates each (right panels): (C) Phosphorylation of ERK1/2; (D) surface expression of CD69, and (E) surface expression of CD107a on murine WT and LFA-1-/- CTLs. (F) Caspase-3 activity in EL4 target cells after co-culture with murine WT and LFA-1-/- CTLs. The representative histogram shows caspase-3 activity in EL4 cells assessed by flow cytometry (left panel), results are quantified from n = 3 mice, with two to three technical replicates each (right panel). (G–J) Representative histograms (left panels) and quantified results with n = 4 (technical replicates) (right panels). (G) Phosphorylation of FAK; (H) phosphorylation of ERK1/2; (I) surface expression of CD107a, on murine WT and LFA-1-/- OT-I CTLs. (J) Cytotoxicity assay using WT and LFA-1-/- OT-I CTLs, co-cultured with cognate antigen-pulsed EL4 target cells. Representative histogram shows caspase-3 activity in EL4 cells assessed by flow cytometry (left panel), the right panel depicts quantified results from n = 4 technical replicates. Results are representative of n = 2 independent experiments. Data are presented as mean ± SD right panel of (A–J). In the representative metabolic flux traces, left panel of (A), symbols indicate mean ± SEM. Statistical significance was assessed by one-way ANOVA with Sidak’s multiple comparison test (A–F), and ordinary two-way ANOVA with main effects only and Tukey’s multiple comparison test (G–J). *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, n.s., nonsignificant 12 signaling (FAK activity), distal LFA-1-impacted signaling (ERK1/2 activity), degranulation, and specific target cell lysis was confirmed across a spectrum of TCR affinities (Figures 3G–3J). Notably, only OVA, with its very high affinity for OT-I, triggered limited lysis independently of Mg2+ and LFA-1 (Figure 3J). LFA-1 is essential for Mg2+ to shape T cell activation and effector function, in vivo Our goal was then to test how genetic deletion of LFA-1 affected T cell function in relation to Mg2+ abundance in vivo. First, we established a dietary Mg2+ depletion model. Organismal extracellular magnesium abundance was mapped both during steady state and after 2 weeks of dietary magnesium restriction (Figure S4A). Intriguingly, in mice fed a normal diet, extracellular magnesium distribution was heterogeneous between anatomic sites - with serum, lymph nodes, and muscle being magnesium rich, while, relatively speaking, spleen, liver, and, even more pronounced so, subcutaneous fatty tissues were naturally magnesium low (Figure 4A). Mg2+ depletion was readily induced through dietary restriction (Figure 4A), and short-term Mg2+ low diet did not affect the general health status of mice (data not shown), and the body weight was comparable between cohorts (Figure S4B). Complete blood counts revealed only small changes in platelet and erythrocyte numbers (Figure S4C, left panel), as well as in hemoglobin levels and erythrocyte volumes (Figure S4C, middle panel). Differential white blood cell counts showed no alterations in Mg2+ depleted animals (Figure S4C, right panel). Further flow cytometric analyses indicated that Mg2+ depletion also had no relevant impact on CD8+ T cell frequency, distribution of phenotypic subpopulations, and steady-state expression of the activation marker CD69 (Figures S4D–S4F). The percentage distribution of other lymphocyte subpopulations was likewise unaffected (Figure S4G). Using this model, LFA-1 WT and LFA-1-/- mice were then placed on normal diet versus magnesium low diet for 2 weeks and injected with anti-CD3ε F(ab0)2 fragments to polyclonally activate CD8+ T cells (Figure 4B). In a strictly LFA-1-dependent manner, mice with normal organismal Mg2+ abundance increased expression of the activation marker CD69 on CD8+ T cells. By contrast, only modest upregulation of CD69 was detected on CD8+ T cells from LFA-1-/- mice, irrespective of diet (Figure 4C). Next, the impact of Mg2+ on WT and LFA-1-/- OT-I T cells was tested in an MC38-OVA (murine colon cancer cell line) model (Figure 4D). On day 9 after tumor implantation, that is, when tumors became palpable, WT or LFA-1-/- OT-I CTLs were adoptively transferred, and tumors were injected every second or third day with either 3 mM MgCl2 or 3 mM NaCl (Figure 4D). Intratumoral (i.t.) Mg2+ application efficiently increased local Mg2+ abundance (Figure S4H, left panel). By contrast, Mg2+ concentrations in draining lymph nodes were variably affected and serum Mg2+ concentration remained unchanged (Figure S4H, middle and right panels). In this setting, the most efficient control of cancer growth was observed in mice transferred with WT OT-I T cells and supplemented with Mg2+ i.t. Tumor growth in mice transferred with LFA-1- /- OT-I T cells was insensitive to i.t. Mg2+ application and largely identical to the growth seen in mice transferred WT OT-I T cells but injected with NaCl i.t. (Figure 4E). In vivo magnesium availability impacts infection control In order to directly monitor the effect of organismal magnesium abundance on the antigen-specific development of cytotoxic activity of memory CD8+ T cells, in vivo killing assays were performed (Figure 5A). Antigen-specific clearance of target cells was lower in Mg2+-depleted hosts, with cytotoxicity partially rescued by co-injecting magnesium together with target cells (Figure 5B). To explore the effect of magnesium on memory CD8+ T cells fighting intracellular infection, a Listeria 13 A Serum [Mg2+] Peritoneum [Mg2+] Interstitial Fluid [Mg2+] 4 **** 0.10 * 800 ** * * 3 600 Ctrl diet Mg2+ low diet 2 0.05 400 1 200 0 0.00 0 iet die t et ie t Spleen Lymph Liver Muscle Subcutaneous trl d ow i d rl d w Nodes FatC l t o2+ C l g 2 + M Mg B C Polyclonal CD8 T cell activation with anti-CD3ε F(ab')2 fragment 80 **** *** n.s. 60 Unstim. control WT: Ctrl diet 40 WT: Mg2+ low diet LFA-1-/-: Ctrl diet 20 LFA-1-/-: Mg2+ low diet 0 10 3 10 4 10 5 0 CD69 - BV650 Ctrl Low Ctrl Low Mg2+ Diet WT LFA-1-/- D E Tumor volume: WT or LFA-1-/- OTI CTLs WT: MgCl 2 i.t. 1000 WT: NaCl i.t. LFA-1-/-: MgCl 2 i.t. n.s. LFA-1-/-: NaCl i.t. 500 0 10 15 20 25 Days post-tumor injection Figure 4 14 mg dL-1 mg dL-1 ppb[Mg] μg[Tissue]-1 Tumor Volume (mm3) % CD69 pos. **** **** Figure 4. LFA-1 is required to regulate CD8+ T cell activation and cytotoxicity in vivo (A) Magnesium levels in serum (left panel), peritoneal lavage (middle panel), and interstitial fluids (right panel) from mice placed on Mg2+ reduced or matching control diet for 2 weeks. Results were pooled from two to three independent experiments, with n = 4–7 mice each. ppb, parts per billion. (B) Schematic of experimental design. (C) CD69 expression on CD8+ T cells from the spleen of LFA-1-/- or WT mice. Representative flow histogram (left panel) and pooled data (right panel). Representative results from n = 2 independent experiments. (D) Schematic of experimental design. (E) Tumor growth curves with n = 6–7 mice per group. Each symbol represents an individual mouse (A and C), and data are presented as mean ± SD (A,C), and mean ± SEM (E). Statistical significance was assessed by unpaired two-tailed Student’s t test in the left and middle panel of (A), unpaired two-tailed Student’s t test with Holm-Sidak corrected multiple comparison test (A, right panel), one-way ANOVA with Sidak’s corrected multiple comparison test (C), and two-way ANOVA with Bonferroni correction for multiple analysis (E). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 15 monocytogenes-OVA (LmOVA) intraperitoneal (i.p.) infection model was used (Figure 5C). Dietary depletion of Mg2+ reduced the efficiency of memory CD8+ T cell effector function also in this system, with spiking the i.p. bacterial inoculum with Mg2+ immediately improving bacterial clearance (Figure 5D). Of note, no impact of Mg2+ per se on bacterial viability or growth was observed (Figure S5A). In line with these immune control data, the peritoneal abundance of Granzyme B was lower in Mg2+depleted hosts and significantly rescued in mice receiving a Mg2+-spiked bacterial inoculum (Figure 5E). Notably, the innate cytokine IL-6 followed an inverse pattern, with markedly increased peritoneal and serum concentrations in mice less able to control infection (Figures 5F and S5B) - plausibly reflecting increased innate inflammation due to a higher bacterial burden. Aligning with better immune control, expression of CD107a and CD69 was higher on OVA-specific CD8+ T cells from Mg2+ competent than from Mg2+-depleted mice (Figures 5G and 5H). Spiking the bacterial inoculum with Mg2+ rescued CD8+ T cell reactivity also in this readout by trend (Figures 5G and 5H). The proportion of antigen- specific memory CD8+ T cells in the peritoneal cavity was comparable between cohorts, indicating that recruitment of antigen-specific cells to the site of infection was intact in mice fed a Mg2+-depleted diet (Figures S5C and S5D). Intratumoral magnesium delivery improves memory CD8+ T cell-mediated anti-tumor immunity Adoptive transfer experiments in NSG mice demonstrated that Mg2+ improved cancer-directed effector activity of CTLs in an LFA-1-dependent manner (Figure 4E). In immunocompetent mice harboring OVA-specific memory CD8+ T cells (i.e., immunized with OVA), injection of Mg2+ i.t. likewise improved local control of MC38 tumor growth. Injection of Mg2+ i.t. in OVA-naive mice did not affect tumor growth, and neither did contralateral i.t. injection of NaCl (Figures 6A and 6B). Of note, no clinical signs or symptoms of autoimmunity were observed in Mg2+ treated mice (data not shown), and anti-nuclear antibodies (ANAs) screening assays returned negative results across cohorts (Figure S6A). OVA immunization induces OVA-specific memory T cells, including memory CD8+ T cells that play a key role in tumor rejection (Enamorado et al., 2017; Han et al., 2020), and depletion experiments established that Mg2+ exerts its effect via CD8+ T cells (Figure 6C). Using flow cytometry, we then enumerated and phenotyped tumor-infiltrating immune cells. Notably, the number of tumor-infiltrating CD8+ T cells was increased in the Mg2+-treated group (Figure 6D), whereas the abundance of other immune cell subsets was not affected significantly (Figures S6B–S6I). Aligning with their increased number, more Mg2+-treated CD8+ T cells expressed Ki67 (Figure 6E, left panel). In addition, more Mg2+ exposed CD8+ T cells contained Granzyme B and expressed the activation marker CD25 (Figure 6E, middle and right panels). Further reflecting increased activation, PD-1 and TIM-3 were also significantly more often (co-)expressed on Mg2+-treated CD8+ T cells (Figure 6F). Of note, in vitro re-stimulation experiments at day 3 and day 6 post-initiation of i.t. Mg2+ injections did not find evidence of dysfunction (exhaustion) of CD8+ T cells (data not shown). Since expression of PD-1 was higher on CD8+ T cells isolated from MgCl2-treated as compared with NaCl-treated tumors, we next assessed how combining MgCl2 injection with PD-1 blockade was impacting tumor-directed memory CD8+ T cell function (Figure 6G). Tumor control in mice receiving both intratumoral MgCl2 as well as PD-1 blockade was superior compared with all other treatment regimens, with MgCl2 alone again improving immune control significantly (Figure 6H). Aligning with these tumor 16 A B In vivo killing assay OVA-pulsed Control 100 ** ** 80 naive 60 Control diet 40 Mg2+ low diet 20 Mg2+ low diet + Mg2+ i.p. 10 3 10 4 10 5 10 6 0 CellTrace Violet Ctrl Low Low Mg2+ Diet − − + Mg2+-spiked i.p. injection C D Peritonitis: CFU E Peritonitis: Granzyme B F Peritonitis: IL-6 104 6*** * 4 * 10 * 0.068 3 103 105 2 102 104 1 101 0 103 Ctrl Low Low Ctrl Low Low Ctrl Low Low Mg2+ Diet − − + − − + − − + Mg2+-spiked i.p. injection G H Peritonitis: Degranulation Peritonitis: CD69 80 100 ** 0.077** 0.077 90 60 Control diet Control diet 80 Mg2+ low diet Mg2+ low diet 70 40 Mg2+ low diet Mg2+ low diet + Mg2+ i.p. 3 4 0 10 10 + Mg 2+ i.p. 603 3 4 5 -10 0 10 10 10 CD107a - PE/Cy7 20 CD69 - APC 50 Ctrl Low Low Ctrl Low Low Mg2+ Diet − − + − − + Mg2+-spiked i.p. injection Figure 5 17 % CD107a pos. of Tetramer+ CFU pg ml-1 % specific lysis % CD69 pos. of Tetramer+ pg ml-1 Figure 5. Organismal magnesium depletion via dietary restriction impairs memory CD8+ T cell-mediated cytotoxicity (A) Schematic of experimental design. (B) Target cell clearance was evaluated in the spleen. Representative flow histogram (left panel), and summarized data from n = 2 independent experiments, with n = 4–9 mice each (right). (C) Schematic of experimental design. (D) Graph depicts bacterial burden in peritoneal lavage, data summarized from n = 3 independent experiments with n = 2–6 mice each. Abundance of Granzyme B (E) and IL-6 (F) in corresponding peritoneal fluids. (G and H) Expression of the degranulation marker CD107a (G), and the activation marker CD69 (H) on tetramer-positive memory CD8+ T cells retrieved from the peritoneal cavity. Representative flow histogram (left panels), and summary data (right panels), from n = 5 mice each. Representative of n = 2 independent experiments. Each symbol represents an individual mouse, data are represented as median ± IQR (B and D–H) dashed line in (D) indicates detection limit. Statistical significance was assessed by unpaired two-tailed Student’s t test (B), and one-way ANOVA with Sidak’s corrected multiple comparison test (D–H). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 18 growth data, survival of animals injected with MgCl2 up to day 26 was significantly improved compared with the NaCl-injected control group, both as a sole intervention and when combined with PD-1 blockade (Figure 6I). Liposomal packaging might be a translational strategy to deliver Mg2+ into tumors since liposomes accumulate at sites with leaky vasculature beds (Franco et al., 2021; Nakamura et al., 2016), and i.t. application of MgCl2- packed liposomes did replicate the effects induced by solute injections (Figures 6J–6L). Encouraged by these data, we tested the effect of i.p. applied liposomes on tumor growth. Specifically, we compared NaCl-loaded versus MgCl2-loaded liposomes coupled to PD1 mAb. The concentration of Mg2+ applied i.p. with liposomal Mg2+ packaging was controlled for and confirmed to be in a physiologic range (Figure S6J). I.p. administration of Mg2+-loaded liposomes enhanced tumor control in a manner additive to PD-1 blockade alone (Figures 6M–6O). Mg2+ has translational potential across a spectrum of emerging treatment modalities in oncology Blinatumomab has improved the clinical outcome of select B cell malignancies (Franquiz and Short, 2020). Notably, hypomagnesemia occurs in 12%-22% of patients treated with Blinatumomab (Coyle et al., 2020; Kiyoi et al., 2020; Topp et al., 2015). Blinatumomab-mediated cytotoxicity, when used in its reported therapeutic range (230-620 pg mL1) (Topp et al., 2015), was strongly, and in a dose-dependent manner, dependent on the availability of Mg2+ (Figure 7A). Aligning with the cytotoxicity data, the effect of Blinatumomab on cell clustering (Figure 7B), activation-induced LFA-1 headpiece opening (Figure S7A), as well as FAK phosphorylation (Figure S7B) was likewise Mg2+ sensitive. Stabilizing the headpiece of LFA-1 in its closed position, using the mAb TS1/18, almost completely abrogated Blinatumomab-mediated cytotoxicity, irrespective of Mg2+ availability (Figure 7C). Inversely, combining Blinatumomab with the LFA-1 extending mAb CBR LFA-1/2 increased target cell killing both in the presence and absence of Mg2+ (Figure 7C). To explore the relevance of Mg2+ in the context of CAR T cell therapy, human CD19 or prostate-specific membrane antigen (PSMA) targeting CAR T cells were used (Neelapu et al., 2017; Santoro et al., 2015). Introducing a CAR did not alter LFA-1 expression (Figure 7D) or dependency of these cells on Mg2+ both with respect to glycolytic reprogramming and phenotypic changes upon conventional activation via ligation of CD3 and CD28 (Figures S7C and S7D). We then went on to assess the Mg2+ dependency of CAR-driven cytotoxicity. In line with our previous findings with REP T cells, the Mg2+-LFA-1 regulatory system reproduced in CD19 CAR T cells (Figure 7E), and timelapse killing assays demonstrated that Mg2+ restriction severely blunted PSMA CAR- specific killing of target cells (Figure 7F). Concomitantly, the release of IFN-g by anti-PSMA CAR T cells was reduced in Mg2+-deplete conditions (Figure 7G). Tumor rejection and survival were also improved in a model where CAR T cells were adoptively transferred into mice harboring established tumors that were then Mg2+ enriched by i.t. injections (Figures 7H and 7I). We then wanted to test whether dietary Mg2+ restriction, thus reducing systemic and intratumoral Mg2+ abundance (Figures S7E and S7F), affects CAR T cell-mediated cytotoxicity (Figure 7J). Indeed, dietary Mg2+ restriction was sufficient to negatively impact CAR T cell-mediated tumor rejection (Figure 7K). Food intake was comparable between the cohorts, and yet, tumor-bearing mice fed a Mg2+-depleted diet had a lower body weight by day 25 of the experiment (Figures S7G and S7H), suggestive of a more catabolic state. While the impact of (mild) intratumoral Mg2+ depletion through dietary restriction was 19 A B Tumor volume: Tumor volume: non-immunized pre-immunized 30 30 NaCl NaCl MgCl 20 2 MgCl 20 2 n.s. 10 10 0 0 0 2 4 6 8 10 12 14 16 0 2 4 6 8 10 12 14 16 Days post treatment Days post treatment C Tumor volume: D E F CD8 depletion CD8 T cells CD8: Ki67 CD8: GrzB CD8: CD25 CD8: PD-1 TIM-31500 MgCl2 + Isotype 20.0 * 15.0 * 6.0 * 10.0 0.0718 100.0 * ** n.s. NaCl + αCD8a 1000 80.0MgCl2 + αCD8a n.s. 15.0 8.0 NaCl 10.0 4.0 MgCl 6.0 60.0 2 10.0 500 4.0 40.0 5.0 2.0 5.0 2.0 20.0 0 0.0 0.0 0.0 0.0 0.0 10 15 20 25 30 l l l l l l l l 3l o hi lo C 2 C 2 C 2 C 2 - -3 -3 Days post tumor injection a gC a C a C a CN M N Mg N Mg N Mg TI M hi hi T IM IM 1 1 1l o T - - - PD PD PD G H Tumor volume: I Survival: Mg2+ and PD-1 blockade Mg2+ and PD-1 blockade 2000 NaCl + IgG2A 1500 100MgCl2 + IgG2A NaCl + αPD-1 MgCl2 + αPD-1 1000 NaCl + IgG2A MgCl2 + IgG2A ***50 NaCl + αPD-1 500 MgCl2 + αPD-1 ** 0 0 5 10 15 20 25 0 10 20 # 30 Days post tumor injection Days post tumor injection (# Day 26: End of treatment) J K Tumor volume: L Survival: 1000 Liposomal Mg 2+ Liposomal Mg2+ Lipo(NaCl) 800 Lipo(MgCl2) 100 Lipo(NaCl) MgCl2 Lipo(MgCl2) 600 MgCl2 400 50 200 0 0 10 15 20 25 0 20 40 60 Days post tumor injection Days post tumor injection M N Tumor volume: O Survival: Immunoliposome Immunoliposome 800 PBS anti-PD1-Lipo(NaCl) 600 anti-PD1-Lipo(MgCl2) PBS 100 anti-PD1-Lipo(NaCl) anti-PD1-Lipo(MgCl2) 400 50 200 0 0 5 10 15 20 0 10 20 30 40 50 Days post treatment Days post treatment Figure 6 20 Tumor volume (mm3) cells per g tumor (x10^7) cells per g tumor (x10^7) Relative Tumor Volume (to first treatment day) Tumor Volume (mm3) Tumor volume (mm3) Tumor volume (mm3) cells per g tumor (x10^7) cells per g tumor (x10^7) **** Relative Tumor Volume (to first treatment day) Percent survival Percent survival Perrcent survival cells per g tumor (x10^7) * * *** **** **** ** ******* *** Figure 6. Intratumoral magnesium administration improves memory CD8+ T cell-mediated anti-tumor immunity (A) Schematic of experimental design. (B) Tumor growth curves in non-immunized mice (n = 20) (left panel), and immunized mice (n = 19) (right panel). Results were pooled from two independent experiments, with n = 9–10 mice each. (C) Tumor growth curves in ± CD8-depleted mice (n = 9–11). Results were pooled from two independent experiments, with n = 4–6 mice each. (D) Absolute numbers of tumor-infiltrating CD8+ T cells. (E) Number of tumor-infiltrating CD8+ T cells positive for Ki67 (left panel), Granzyme B (middle panel), and CD25 (right panel). (F) Cell number of tumor-infiltrating CD8+ T cells expressing PD-1 and TIM-3. (G) Schematic of experimental design. (H and I) (H) Tumor growth curves (n = 13-14 mice), and (I) survival (n = 13–14 mice). Results were pooled from two independent experiments, with n = 6-7 mice each (H and I). (J) Schematic of experimental design. (K and L) (K) Tumor growth curves (n = 5–6 mice), and (L) survival (n = 5–6 mice). (M) Schematic of experimental design. (N) Tumor growth curves (n = 11–13). Results were pooled from two independent experiments. (O) Survival. Representative of n = 2 independent experiments with n = 5–7 mice per group. Of note, experiments (B, H, and I) and (C–F and K–O) were conducted at different animal facilities to probe for robustness. Data are presented as mean ± SEM (B, C, and H), median ± IQR with each symbol representing one mouse (D, E, and F). Statistical significance was assessed by two-way ANOVA with Bonferroni correction (B, C, H, K, and N), unpaired two-tailed Student’s t test (D and E), two-way ANOVA with Sidak corrected multiple comparison test (F), and log-rank Mantel-Cox test (I, L, and O). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001, n.s., nonsignificant. 21 less pronounced than in the opposite experiment where Mg2+ was injected into the tumors (Figures 4E, 6B, 6C, 6H, 6K, and 7I), the effect was still clear. This prompted us to retrospectively assess the relationship between serum Mg2+ levels and clinical outcomes in a CAR T cell trial and in an immune checkpoint inhibitor study. The CAR trial included a cohort of 100 patients with refractory B cell lymphoma treated with CD19directed CAR T cells (Axicabtagene Ciloleucel), of which four had to be excluded from the retrospective analysis due to incomplete Mg2+-serum testing (Figure S7I, Study Diagram). Patients were classified into two strata according to the mean Mg2+ level between days 5 and +3 of treatment (n = 5 measurements available for each patient). An arbitrary cut-off was set at 1.7 mg dl-1 for assigning patients into normoversus hypomagnesemia groups (Figure 7L, left panel, and Figure S7K). Baseline characteristics of these retrospectively assigned study populations were similar, including age, ECOG performance status, and disease stage (Figure S7J). Although the number of patients with a mean Mg2+ level of <1.7 mg dl-1 was low, overall survival and median progression-free survival for these patients were reduced as compared with patients with normal serum Mg2+ levels (Figure 7L, right panel, and Figure S7L). We next explored how organismal Mg2+ abundance was associated with outcome in a cohort of non- small cell lung cancer (NSCLC) patients enrolled in an immune checkpoint inhibitor trial (SAKK16/14) (Rothschild et al., 2021). From a total of 67 initially enrolled patients, two had to be excluded, leaving 65 that were treated with an anti-PD-L1 mAb (Durvalumab) in addition to neoadjuvant chemotherapy (Figure S7M, study diagram). Any detection of hypomagnesemia during the course of the trial (n = 2-23 measurements/study participant) assigned an individual to the hypomagnesemia group. This stratification strategy well discriminated the mean Mg2+ levels across all available measurements (Figure 7M, left panel). Also in this clinical trial, baseline characteristics were similar between the two retrospectively assigned groups (Figure S7N). Pathological complete response and overall survival (Figure 7M, middle and right panels), as well as radiographic esponse and event-free survival (Figures S7O and S7P), were all reduced in patients with hypomagnesemia. While these retrospective analyses have many limitations, in the context of our experimental data, the findings aligned with the concept that Mg2+, by increasing LFA-1 outside-in signaling activity, may contribute to the clinical efficacy of CAR T cells and endogenous cancer-directed T cells in human patients. DISCUSSION LFA-1 inside-out signaling, or activation, has previously been shown to modulate T cell function by modifying the molecule co-stimulation activity (i.e., outside-in signaling) (Gérard et al., 2021; Wang et al., 2009). Our experiments now demonstrate that, upon activation of LFA-1, extracellular Mg2+ increases LFA-1 outside-in signaling in a dose-dependent manner by contributing to extension and headpiece opening. Mg2+ exerts its effects on the conformation of LFA-1 both by binding to the beta I- and the alpha I-MIDAS, with the precise contribution of the respective interactions yet to be resolved (Schürpf and Springer, 2011; Sen et al., 2018; Shimaoka et al., 2003). Through this activity, extracellular Mg2+ defines the outcome of memory CD8+ T cell- dependent immunity. Unsurprisingly, also in light of our data, complete pharmacological inhibition of LFA-1 causes severe immunosuppression (Goodman and Picard, 2012). However, it is interesting to note that, when using a mAb that stabilizes extended LFA-1 or when transferring T cells expressing constitutively extended LFA- 1, T cell responses are severely impaired both in vitro and in vivo (Dransfield et al., 1992a; Semmrich et al., 2005). 22 A B C Blinatumomab: in vitro cytotoxicity 1.2 mM 0 mM Blinatumomab: in vitro cytotoxicity with LFA-1 modulating antibodies 80 80 1.2 mM **** 0.12 mM **** **** 60 0.012 mM 60 **** **** 0 mM *** *** **** ****n.s. 40 40 n.s. ** 20 20 0 0 0 30 300 3000 typ e 8 e 8 e 8 e 8 1/2 /1 p 1/2 /1 p 1/2 /1 p 1/2 /1 Blinatumomab [pg ml-1] o A- S1 t y - 1 ty - 1 ty - 1 Is LF T Is o FA TS Iso L LF A TS Iso FA SL T BR BR R R C C CB CB 1.2 mM 1.2 mM + 0 mM 0 mM + Blinatumomab Blinatumomab D CAR T cell: E LFA-1 expression Anti-CD19 CAR T cell: in vitro cytotoxcity F Anti-PSMA CAR T cell: in vitro cytotoxicity G Anti-PSMA CAR T cell: IFNγ 1.5 8.0 1.1 **** CAR T cells [0.6 mM] *** *** n.s. 6.0 CAR T cells [0 mM] 1.0 1.2 mM 1.0 PHA-Blast UTD T cells [0.6 mM] 0.9 0 mM 4.0 UTD T cells [0 mM] UTD T cell 0.8 0 + Mn2+ 0.5 anti-CD19 1.2 mM 2.0 0.7 CAR T cell 10 2 10 3 10 4 10 5 6 3 4 510 + BIRT377 0 10 10 10 CD11a - FITC Caspase-3 - AF405 0.0 0.0 0.6 M M 2+ 737 0 8 16 24 32 40 48n 0.6 0 mM Mg 2+ m .2 0 m M TIR Timer (hours)1 0 + + B mM 1.2 H I Anti-CD19 CAR T cell in vivo - intratumoral injections: Tumor volume Survival 1000 Saline - MgCl2 i.t. 100 CAR T cells - NaCl i.t. CAR T cells - MgCl2 i.t. Saline - MgCl2 i.t. CAR T cells - NaCl i.t. 500 50 CAR T cells - MgCl2 i.t. 0 0 5 10 15 20 25 30 10 20 30 40 Days post tumor injection Days post tumor injection J K Anti-PSMA CAR T cell - in vivo tumor rejection 2000 Saline Ctrl diet 1500 UTD T cells Ctrl diet CAR T cells Ctrl diet 1000 Saline Mg2+ low diet ** UTD T cells Mg2+ low diet CAR T cells Mg2+ low diet 500 0 5 8 12 15 19 22 26 29 33 Days post tumor injection L CAR T cell Trial - Axicabtagene ciloleucel in B cell Lympohma: M SAKK16/14 Trial - Durvalumab in NSCLC: Mean serum levels Overall survival Mean serum levels Pathologicalcomplete response Overall survival 3.0 **** 100 Mean magnesium level 1.5 **** 40 100 >1.7 mg dl-1 29.2 <1.7 mg dl-1 30 2.5 1.0 50 20 50 2.0 0.5 9.7 10 Normomagnesemia Hypomagnesemia 1.5 0 0.0 0 0 l-1 -1g d m mg dl 0 5 10 15 20 25 normal low normal low 0 20 40 >1.7 <1.7 Time (months) Magnesium level Magnesium level Time (months) Figure 7 23 Serum Magnesium (mg dl-1) % Caspase-3 pos. target cells Probability of Survival Caspase-3 pos. Ramos Blinatumomab Control cells norm. to 1.2 mM Tumor Volume (mm3) ** Serum Magnesium (mg dl-1) Tumor Volume (mm3) Tumor Cell Killing (Green Area - μm well-1) (x10^6) % of patients % Caspase-3 pos. target cells Percent survival Probability of Survival * Fold change IFNγ release ** ** * **** **** **** **** Figure 7. Extracellular magnesium impacts functionality of immunotherapeutic modalities in vitro and in vivo (A) Box plots representing flow cytometric assessment of caspase-3 activity in Ramos target cells after co-culture with PHA blasts, n = 5 healthy donors at Mg2+ and Blinatumomab concentrations as indicated. (B) Representative brightfield images of Blinatumomab-induced cell aggregation between PHA blasts and Ramos target cells in dependence of Mg2+. Scale bars indicate 250 mm. (C) Caspase-3 activity in LCL target cells after co-culture with PHA blasts ± Blinatumomab (300 pg mL1), in dependence of Mg2+ plus CBR LFA-1/2, TS1/18, or isotype control mAb, as indicated (n = 4 healthy donors). (D) Representative histograms of CD11a expression on PHA blasts, untransduced T cells, and anti-CD19 CAR T cells. (E) Representative histograms (left panel), and results quantified from n = 2 healthy donor with three technical replicates (right panel) of caspase-3 activity in Ramos target cells, after co-culturing with anti-CD19 CAR T cells, ± Mg2+, ± Mn2+, and ± BIRT377, as indicated. (F) Cytotoxicity assay with anti-PSMA CAR T cells and UTD T cells co-cultured with PSMA+ PC3-PIP cell line. Pooled results using cells generated from n = 6 healthy donors and from n = 2 independent experiments are shown. (G) Abundance of IFNg in cell culture supernatants corresponding to conditions depicted in (F) (n = 8, 3 independent experiments). (H) Schematic of experimental design. (I) Tumor growth curves (left panel) and survival (right panel) from n = 6–7 mice per group. (J) Schematic of experimental design. (K) Tumor growth curves. Representative of n = 2 independent experiments with n = 6 mice per group. (L) Stratification of patients according to mean serum magnesium levels > 1.7 mg dl1 versus <1.7 mg dl1 between day 5 and day +3 of adoptive cell therapy, n = 5 measurements per patient (left panel). Each symbol represents one individual. Overall survival of patients stratified according to normal and low Mg2+ levels (right panel). (M) Comparison of mean serum magnesium levels after stratification according to occurrence of R1 hypomagnesemia- measurement during the trial (n = 2–23 measurements per patient) (left panel). Complete pathological response (middle panel), and overall survival (right panel) according to this stratification. NSCLC, non-small cell lung cancer. Data are presented as median ± IQR (A and C), mean ± SD right panel of (E), (G), mean ± SEM (F and K) and left panel of (I), or median ± 95% CI left panel of (L and M). Statistical significance was assessed by two-way ANOVA with Dunnett corrected multiple comparison test (A), two-way ANOVA with Tukey’s multiple comparison test (C), one-way ANOVA with Sidak’s multiple comparison test right panel (E), repeated-measures two-way ANOVA with Tukey’s multiple comparison test (F), unpaired Student’s t test (G) and left panel of (L and M), two-way analysis of variance (ANOVA) with Bonferroni correction left panel of (I) as well (K), and log-rank Mantel-Cox test in right panels of (I, L, and M). *p < 0.05, **p < 0.01, ***p < 0.001, **** p < 0.0001, n.s., nonsignificant. 24 This underscores the complex interconnection between LFA-1’s structural conformation, its outside-in signaling activity, and the function of T cells - a system tuned, as we show here, by Mg2+ in the extracellular space. In an intricate manner, extracellular Ca2+ and Mn2+ coordinate with Mg2+ in stabilizing beta integrin heterodimer complexes as well as in regulating conformational changes. Specifically, Ca2+ at physiologic concentrations has an inhibitory effect on integrin-ligand binding, while low Ca2+ abundance functions synergistically with Mg2+ in mediating cell adhesion (Dransfield et al., 1992b; Labadia et al., 1998; Onley et al., 2000). Ca2+ exerts these modulatory effects by binding to two adjacent metal coordination sites flanking the beta I MIDAS, termed Ligand-induced metal-binding site (LIMBS) and adjacent to MIDAS (ADMIDAS) (Chen et al., 2003; Mould et al., 2003). The potent integrin activator Mn2+, in contrast, operates by binding MIDAS motifs and by competing with Ca2+ for LIMBS and ADMIDAS binding (Chen et al., 2003; Valdramidou et al., 2008). While our data demonstrate that modifying Mg2+ in the extracellular space is sufficient to impact T cell effector maturation in vivo, it will be interesting to study how Mg2+ integrates its function with other divalent cations across physiologic and pathologic concentrations. The compartmentalized distribution of Mg2+ in healthy mice plausibly creates tissue-specific differences in the TCR signal strength required to activate memory CD8+ T cells that are engaged in immune surveillance of Mg2+ high versus Mg2+ low tissues. It will be of value to explore whether compartment-specific effects on the CD8+ T cell activation threshold not only affect infection control and cancer immune control but also affect the occurrence of autoimmunity. Improving conventional cancer therapy by manipulating nutrient availability through dietary intervention is a field of growing interest (Kanarek et al., 2020). For example, experimental data suggest beneficial effects of repetitive fasting cycles in mice, the rationale being to target the high metabolic requirements of cancer cells (Lee et al., 2012). When aiming to translate such deprivation strategies to clinical medicine, they need, however, to be carefully balanced against the danger of imposing nutrient deficiencies that may negatively impact on host immunity. Indeed, only 2 weeks of dietary depletion of Mg2+ was sufficient to affect both tumor and infection immune control in mice, and low serum Mg2+ levels were associated with an unfavorable outcome both in CAR T cell and anti-PDL1 mAb-treated human cancer patients. Of note, hypomagnesemia is present in up to a fifth of Blinatumomab-treated individuals (Kiyoi et al., 2020), a quarter of CAR T cell recipients (Abramson et al., 2020; Locke et al., 2019; Siddiqi et al., 2019), and in up to 90% of patients receiving platin-based chemotherapy (Lajer and Daugaard, 1999; Liu et al., 2019; Oronsky et al., 2017). Inversely, a recent study in patients with acute myeloid leukemia showed that higher serum magnesium levels after allogeneic stem cell transplantation were associated with a decreased incidence of relapse but with a higher risk of acute graft-versus-host disease (Angenendt et al., 2021). In conclusion, our data identify extracellular Mg2+ as an immune modulator, regulating T cell activation via LFA- 1 MIDAS binding. LFA-1 thus senses Mg2+ and integrates its abundance in its co-stimulatory function. Through this axis, Mg2+ defines the threshold of T cell activation in a subset selective manner. 25 LIMITATIONS OF THE STUDY Autochthonous tumors are more heterogeneous in architecture, immune cell landscape, perfusion, etc., than the cancer models used in this study, and our findings, therefore, cannot be generalized. We also have not analyzed the long-term consequences of enhancing LFA-1 outside-in signaling, which will require slow growth, i.e., more chronic cancer models. Furthermore, how organismal depletion of Mg2+ affects the efficiency of CAR T cells to attack established solid tumors remains to be defined. It is notable, however, that stratifying human cancer patients according to serum Mg2+ abundance was discriminating study participants with regard to relevant endpoints in both a CAR T celland an immune checkpoint trial. Yet, the retrospective nature of these analyses limits the significance of the associations detected. Therefore, as instructed by our findings, prospective controlled trials will need to clarify the clinical relevance of Mg2+ as an immunomodulatory agent in cancer medicine. ACKNOWLEDGMENTS J. Loet. was supported by SNSF grant 323530_171148 as well grants from the Freiwillige Akademische Gesellschaft Basel and Nikolaus und Bertha Burckhardt-Bu ̈rgin-Stiftung; P.D. by SNSF grant 183980; M.L.B. by SNSF grant PMPDP3_171261/1, PCEFP3_194618 / 1, and a Novartis Foundation grant 17C141; A.Z. by SNSF grant 320030_188576; M.I. by SNSF grant 310030_204326; and C.H. by SNSF grants 31003A_172848, FZEB-0180487, and 310030_192677. We thank Ricardo Mancuso for helpful discussion and scientific input; Klaus Ley and Markus Sperandio for providing reagents; Veronica Richina for help with the hematology analyzer; and Priska Grünig, Tobias Öttl, and Michael Mayr for organizational help during the revision. AUTHOR CONTRIBUTIONS J. Lötscher designed, performed, and analyzed most experiments and wrote the manuscript; A.M.L., N.K., G.G., E.C., M.T., M.L., M.K., P.D., J. Löliger, L.L., C.L., S.H.S., N.P., D.S., V.K., D.L., J.G., and A.-V.B. performed and analyzed experiments; S.I.R., P.S., and M.P. provided clinical data; D.K., M.L.B., D.M., J.H., M.I., W.R., C.G.K., and A.Z. supervised and coordinated experiments; C.H. designed, supervised, and coordinated the study and wrote the manuscript. All authors reviewed the manuscript and approved its final version. DECLARATION OF INTERESTS J. Lötscher and C.H. are inventors on a patent relating to this study filed by the University of Basel (EP20/191392.8), which is being developed by a start-up company (Hornet Therapeutics Ltd - Scientific Founder: C.H.). J.G. is an employee of Hornet Therapeutics. The authors declare no other competing financial interests. 26 REFERENCES Abramson, J.S., Palomba, M.L., Gordon, L.I., Lunning, M.A., Wang, M., Arnason, J., Mehta, A., Purev, E., Maloney, D.G., Andreadis, C., et al. (2020). Lisocabtagene maraleucel for patients with relapsed or refractory large B-cell lymphomas (transcend NHL 001): a multicentre seamless design study. Lancet 396, 839–852. Akizawa, Y., Koizumi, S., Itokawa, Y., Ojima, T., Nakamura, Y., Tamura, T., and Kusaka, Y. (2008). Daily magnesium intake and serum magnesium concentration among Japanese people. J. Epidemiol. 18, 151–159. Angenendt, L., Hilgefort, I., Mikesch, J.H., Schlüter, B., Berdel, W.E., Lenz, G., Stelljes, M., and Schliemann, C. (2021). Magnesium levels and outcome after allogeneic hematopoietic stem cell transplantation in acute myeloid leukemia. Ann. Hematol. 100, 1871–1878. Article Bantug, G.R., Galluzzi, L., Kroemer, G., and Hess, C. (2018). The spectrum of T cell metabolism in health and disease. Nat. Rev. Immunol. 18, 19–34. Chen, J., Salas, A., and Springer, T.A. (2003). Bistable regulation of integrin adhesiveness by a bipolar metal ion cluster. Nat. Struct. Biol. 10, 995–1001. Costello, R.B., Elin, R.J., Rosanoff, A., Wallace, T.C., Guerrero-Romero, F., Hruby, A., Lutsey, P.L., Nielsen, F.H., Rodriguez-Moran, M., Song, Y., and Van Horn, L.V. (2016). Perspective: the case for an evidence-based reference interval for serum magnesium: the time has come. Adv. Nutr. 7, 977–993. Costello, R.B., and Nielsen, F. (2017). Interpreting magnesium status to enhance clinical care: Key indicators. Curr. Opin. Clin. Nutr. Metab. Care 20, 504–511. Coyle, L., Morley, N.J., Rambaldi, A., Mason, K.D., Verhoef, G., Furness, C.L., Zhang, A., Jung, A.S., Cohan, D., and Franklin, J.L. (2020). Open-label, phase 2 study of Blinatumomab as second salvage therapy in adults with relapsed/refractory aggressive B-cell non-Hodgkin lymphoma. Leuk 61, 2103–2112, Lymphoma. Dransfield, I., Cabañas, C., Barrett, J., and Hogg, N. (1992a). Interaction of leukocyte integrins with ligand is necessary but not sufficient for function. J. Cell Biol. 116, 1527–1535. Dransfield, I., Cabañas, C., Craig, A., and Hogg, N. (1992b). Divalent cation regulation of the function of the leukocyte integrin LFA-1. J. Cell Biol. 116, 219–226. Dudley, M.E., Wunderlich, J.R., Robbins, P.F., Yang, J.C., Hwu, P., Schwartzentruber, D.J., Topalian, S.L., Sherry, R., Restifo, N.P., Hubicki, A.M., et al. (2002). Cancer regression and autoimmunity in patients after clonal repopulation with antitumor lymphocytes. Science 298, 850–854. Enamorado, M., Iborra, S., Priego, E., Cueto, F.J., Quintana, J.A., Martínez-Cano, S., Mejías-Peérez, E., Esteban, M., Melero, I., Hidalgo, A., and Sancho, D. (2017). Enhanced anti-tumour immunity requires the interplay between resident and circulating memory CD8+ T cells. Nat. Commun. 8, 16073. Fan, S.T., Brian, A.A., Lollo, B.A., Mackman, N., Shen, N.L., and Edgington, T.S. (1993). CD11a/CD18 (LFA-1) integrin engagement enhances biosynthesis of early cytokines by activated T cells. Cell. Immunol 148, 48–59. Fan, Z., McArdle, S., Marki, A., Mikulski, Z., Gutierrez, E., Engelhardt, B., Deutsch, U., Ginsberg, M., Groisman, A., and Ley, K. (2016). Neutrophil recruitment limited by high-affinity bent b2 integrin binding ligand in cis. Nat. Commun. 7, 12658. Franco, M.S., Gomes, E.R., Roque, M.C., and Oliveira, M.C. (2021). Triggered drug release from liposomes: Exploiting the outer and inner tumor environment. Front. Oncol. 11, 623760. Franquiz, M.J., and Short, N.J. (2020). Blinatumomab for the treatment of adult B-cell acute lymphoblastic leukemia: toward a new era of targeted immunotherapy. Biologics 14, 23–34. Gérard, A., Cope, A.P., Kemper, C., Alon, R., and Köchl, R. (2021). LFA-1 in T cell priming, differentiation, and effector functions. Trends Immunol 42, 706–722. Giannoni, E., Chiarugi, P., Cozzi, G., Magnelli, L., Taddei, M.L., Fiaschi, T., Buricchi, F., Raugei, G., and Ramponi, G. (2003). Lymphocyte function-associated antigen-1-mediated T cell adhesion is impaired by low molecular weight phosphotyrosine phosphatase-dependent inhibition of FAK activity. J. Biol. Chem. 278, 36763–36776. 27 Goodman, S.L., and Picard, M. (2012). Integrins as therapeutic targets. Trends Pharmacol. Sci. 33, 405–412. Han, J., Khatwani, N., Searles, T.G., Turk, M.J., and Angeles, C.V. (2020). Memory CD8+ T cell responses to cancer. Semin. Immunol. 49, 101435. Hebeisen, M., Oberle, S.G., Presotto, D., Speiser, D.E., Zehn, D., and Rufer, N. (2013). Molecular insights for optimizing T cell receptor specificity against cancer. Front. Immunol. 4, 154. Heberle, H., Meirelles, G.V., da Silva, F.R., Telles, G.P., and Minghim, R. (2015). InteractiVenn: A web-based tool for the analysis of sets through Venn diagrams. BMC Bioinformatics 16, 169. Hogg, N., Patzak, I., and Willenbrock, F. (2011). The insider’s guide to leukocyte integrin signalling and function. Nat. Rev. Immunol. 11, 416–426. Huwyler, J., Yang, J., and Pardridge, W.M. (1997). Receptor mediated delivery of daunomycin using immunoliposomes: Pharmacokinetics and tissue distribution in the rat. J. Pharmacol. Exp. Ther. 282, 1541–1546. Kanarek, N., Petrova, B., and Sabatini, D.M. (2020). Dietary modifications for enhanced cancer therapy. Nature 579, 507–517. Kanellopoulou, C., George, A.B., Masutani, E., Cannons, J.L., Ravell, J.C., Yamamoto, T.N., Smelkinson, M.G., Jiang, P.D., Matsuda-Lennikov, M., Reilley, J., et al. (2019). Mg2+ regulation of kinase signaling and immune function. J. Exp. Med. 216, 1828–1842. Kiyoi, H., Morris, J.D., Oh, I., Maeda, Y., Minami, H., Miyamoto, T., Sakura, T., Iida, H., Tuglus, C.A., Chen, Y., et al. (2020). Phase 1b/2 study of Blinatumomab in Japanese adults with relapsed/refractory acute lymphoblastic leukemia. Cancer Sci 111, 1314–1323. Krutzik, P.O., and Nolan, G.P. (2003). Intracellular phospho-protein staining techniques for flow cytometry: Monitoring single cell signaling events. Cytometry A 55, 61–70. Labadia, M.E., Jeanfavre, D.D., Caviness, G.O., and Morelock, M.M. (1998). Molecular regulation of the interaction between leukocyte function-associated antigen-1 and soluble ICAM-1 by divalent metal cations. J. Immunol. 161, 836–842. Lajer, H., and Daugaard, G. (1999). Cisplatin and hypomagnesemia. Cancer Treat. Rev. 25, 47–58. Larsson, S.C., Orsini, N., and Wolk, A. (2012). Dietary magnesium intake and risk of stroke: A meta-analysis of prospective studies. Am. J. Clin. Nutr. 95, 362–366. Lee, C., Raffaghello, L., Brandhorst, S., Safdie, F.M., Bianchi, G., Martin-Montalvo, A., Pistoia, V., Wei, M., Hwang, S., Merlino, A., et al. (2012). Fasting cycles retard growth of tumors and sensitize a range of cancer cell types to chemotherapy. Sci. Transl. Med. 4, 124ra27. Li, D., Molldrem, J.J., and Ma, Q. (2009). LFA-1 regulates CD8+T cell activation via T cell receptor-mediated and LFA-1-mediated ERK1/2 signal pathways. J. Biol. Chem. 284, 21001–21010. Liu, W., Qdaisat, A., Soliman, P.T., Ramondetta, L., Lopez, G., Narayanan, S., Zhou, S., Cohen, L., Bruera, E., and Yeung, S.J. (2019). Hypomagnesemia and survival in patients with ovarian cancer who received chemotherapy with carboplatin. Oncologist 24, e312–e317. Locke, F.L., Ghobadi, A., Jacobson, C.A., Miklos, D.B., Lekakis, L.J., Oluwole, O.O., Lin, Y., Braunschweig, I., Hill, B.T., Timmerman, J.M., et al. (2019). Longterm safety and activity of axicabtagene ciloleucel in refractory large B-cell lymphoma (ZUMA-1): a single-arm, multicentre, phase 1-2 trial. Lancet Oncol 20, 31–42. Lötscher, J., and Balmer, M.L. (2019). Sensing between reactions how the metabolic microenvironment shapes immunity. Clin. Exp. Immunol. 197, 161–169. Lowenstein, F.W., and Stanton, M.F. (1986). Serum magnesium levels in the United States, 1971–1974. J. Am. Coll. Nutr. 5, 399–414. Moore, T.I., Aaron, J., Chew, T.L., and Springer, T.A. (2018). Measuring integrin conformational change on the cell surface with super-resolution microscopy. Cell Rep 22, 1903–1912. Mould, A.P., Barton, S.J., Askari, J.A., Craig, S.E., and Humphries, M.J. (2003). Role of ADMIDAS cation-binding site in ligand recognition by integrin alpha 5 beta 1. J. Biol. Chem. 278, 51622–51629. 28 Nakamura, Y., Mochida, A., Choyke, P.L., and Kobayashi, H. (2016). Nanodrug delivery: Is the enhanced permeability and retention effect sufficient for curing cancer? Bioconjug. Chem. 27, 2225–2238. Nasulewicz, A., Wietrzyk, J., Wolf, F.I., Dzimira, S., Madej, J., Maier, J.A., Rayssiguier, Y., Mazur, A., and Opolski, A. (2004). Magnesium deficiency inhibits primary tumor growth but favors metastasis in mice. Biochim. Biophys. Acta 1739, 26–32. Neelapu, S.S., Locke, F.L., Bartlett, N.L., Lekakis, L.J., Miklos, D.B., Jacobson, C.A., Braunschweig, I., Oluwole, O.O., Siddiqi, T., Lin, Y., et al. (2017). Axicabtagene Ciloleucel CAR T-cell therapy in refractory large B-cell lymphoma. N. Engl. J. Med. 377, 2531–2544. Onley, D.J., Knight, C.G., Tuckwell, D.S., Barnes, M.J., and Farndale, R.W. (2000). Micromolar Ca2+ concentrations are essential for Mg2+-dependent binding of collagen by the integrin alpha 2beta 1 in human platelets. J. Biol. Chem. 275, 24560–24564. Oronsky, B., Caroen, S., Oronsky, A., Dobalian, V.E., Oronsky, N., Lybeck, M., Reid, T.R., and Carter, C.A. (2017). Electrolyte disorders with platinum-based chemotherapy: mechanisms, manifestations and management. Cancer Chemother. Pharmacol. 80, 895–907. Qu, X., Jin, F., Hao, Y., Li, H., Tang, T., Wang, H., Yan, W., and Dai, K. (2013). Magnesium and the risk of cardiovascular events: A meta-analysis of prospective cohort studies. PLoS One 8, e57720. Ravell, J., Otim, I., Nabalende, H., Legason, I.D., Reynolds, S.J., Ogwang, M.D., Ndugwa, C.M., Marshall, V., Whitby, D., Goedert, J.J., et al. (2018). Plasma magnesium is inversely associated with Epstein-Barr virus load in peripheral blood and Burkitt lymphoma in Uganda. Cancer Epidemiol 52, 70–74. Rothschild, S.I., Zippelius, A., Eboulet, E.I., Savic Prince, S., Betticher, D., Bettini, A., Früh, M., Joerger, M., Lardinois, D., Gelpke, H., et al. (2021). SAKK 16/ 14: durvalumab in addition to neoadjuvant chemotherapy in patients with stage IIIA(N2) non-small-cell lung cancer-a multicenter single-arm phase II trial. J. Clin. Oncol. 39, 2872–2880. Sabatos, C.A., Doh, J., Chakravarti, S., Friedman, R.S., Pandurangi, P.G., Tooley, A.J., and Krummel, M.F. (2008). A synaptic basis for paracrine interleukin-2 signaling during homotypic T cell interaction. Immunity 29, 238–248. Sakaguchi, Y., Fujii, N., Shoji, T., Hayashi, T., Rakugi, H., and Isaka, Y. (2014). Hypomagnesemia is a significant predictor of cardiovascular and non-cardiovascular mortality in patients undergoing hemodialysis. Kidney Int 85, 174–181. Salas, A., Shimaoka, M., Kogan, A.N., Harwood, C., von Andrian, U.H., and Springer, T.A. (2004). Rolling adhesion through an extended conformation of integrin alphaLbeta2 and relation to alpha I and beta I-like domain interaction. Immunity 20, 393–406. San Sebastian, E.S., Mercero, J.M., Stote, R.H., Dejaegere, A., Cossío, F.P., and Lopez, X. (2006). On the affinity regulation of the metal-ion-dependent adhesion sites in integrins. J. Am. Chem. Soc. 128, 3554–3563. Santoro, S.P., Kim, S., Motz, G.T., Alatzoglou, D., Li, C., Irving, M., Powell, D.J., Jr., and Coukos, G. (2015). T cells bearing a chimeric antigen receptor against prostate-specific membrane antigen mediate vascular disruption and result in tumor regression. Cancer Immunol. Res. 3, 68–84. Saris, N.E., Mervaala, E., Karppanen, H., Khawaja, J.A., and Lewenstam, A. (2000). Magnesium. An update on physiological, clinical and analytical aspects. Clin. Chim. Acta 294, 1–26. Schaller, M.D., Hildebrand, J.D., Shannon, J.D., Fox, J.W., Vines, R.R., and Parsons, J.T. (1994). Autophosphorylation of the focal adhesion kinase, pp125FAK, directs SH2-dependent binding of pp60src. Mol. Cell. Biol. 14, 1680–1688. Schmid, D.A., Irving, M.B., Posevitz, V., Hebeisen, M., Posevitz-Fejfar, A., Sarria, J.C., Gomez-Eerland, R., Thome, M., Schumacher, T.N., Romero, P., et al. (2010). Evidence for a TCR affinity threshold delimiting maximal CD8 T cell function. J. Immunol. 184, 4936–4946. Schulze, M.B., Schulz, M., Heidemann, C., Schienkiewitz, A., Hoffmann, K., and Boeing, H. (2007). Fiber and magnesium intake and incidence of type 2 diabetes: a prospective study and meta-analysis. Arch. Intern. Med. 167, 956–965. Schürpf, T., and Springer, T.A. (2011). Regulation of integrin affinity on cell surfaces. EMBO J 30, 4712–4727. 29 Semmrich, M., Smith, A., Feterowski, C., Beer, S., Engelhardt, B., Busch, D.H., Bartsch, B., Laschinger, M., Hogg, N., Pfeffer, K., and Holzmann, B. (2005). Importance of integrin LFA-1 deactivation for the generation of immune responses. J. Exp. Med. 201, 1987–1998. Sen, M., Koksal, A.C., Yuki, K., Wang, J., and Springer, T.A. (2018). Ligandand cation-induced structural alterations of the leukocyte integrin LFA-1. J. Biol. Chem. 293, 6565–6577. Shimaoka, M., Salas, A., Yang, W., Weitz-Schmidt, G., and Springer, T.A. (2003). Small molecule integrin antagonists that bind to the b2 subunit I-like domain and activate signals in one direction and block them in the other. Immunity 19, 391–402. Siddiqi, T., Soumerai, J.D., Dorritie, K.A., Stephens, D.M., Riedell, P.A., Arnason, J.E., Kipps, T.J., Gillenwater, H.H., Gong, L., and Dubovsky, J.A. (2019). Rapid undetectable MRD (uMRD) responses in patients with relapsed/refractory (R/R) chronic lymphocytic leukemia/small lymphocytic lymphoma (CLL/ SLL) treated with lisocabtagene maraleucel (liso-cel), a CD19-directed CAR T cell product: updated results from transcend CLL 004, a phase 1/2 study including patients with high-risk disease previously treated with ibrutinib. Blood 134, 503. Sojka, J.E., and Weaver, C.M. (1995). Magnesium supplementation and osteoporosis. Nutr. Rev. 53, 71–74. Thomas, R., Al-Khadairi, G., Roelands, J., Hendrickx, W., Dermime, S., Bedognetti, D., and Decock, J. (2018). NY- ESO-1 based immunotherapy of cancer: current perspectives. Front. Immunol. 9, 947. Topp, M.S., Gökbuget, N., Stein, A.S., Zugmaier, G., O’Brien, S., Bargou, R.C., Dombret, H., Fielding, A.K., Heffner, L., Larson, R.A., et al. (2015). Safety and activity of Blinatumomab for adult patients with relapsed or refractory B-precursor acute lymphoblastic leukaemia: A multicentre, single-arm, phase 2 study. Lancet Oncol 16, 57–66. Trebak, M., and Kinet, J.P. (2019). Calcium signalling in T cells. Nat. Rev. Immunol. 19, 154–169. Valdramidou, D., Humphries, M.J., and Mould, A.P. (2008). Distinct roles of beta1 metal ion-dependent adhesion site (MIDAS), adjacent to MIDAS (ADMIDAS), and ligand-associated metal-binding site (LIMBS) cation-binding sites in ligand recognition by integrin alpha2beta1. J. Biol. Chem. 283, 32704–32714. van Aalderen, M.C., van den Biggelaar, M., Remmerswaal, E.B.M., van Alphen, F.P.J., Meijer, A.B., Ten Berge, I.J.M., and van Lier, R.A.W. (2017). Label-free analysis of CD8+ T cell subset proteomes supports a progressive differentiation model of human-virus-specific T cells. Cell Rep 19, 1068–1079. Vorup-Jensen, T., Waldron, T.T., Astrof, N., Shimaoka, M., and Springer, T.A. (2007). The connection between metal ion affinity and ligand affinity in integrin I domains. Biochim. Biophys. Acta 1774, 1148–1155. Wang, Y., Li, D., Nurieva, R., Yang, J., Sen, M., Carren ̃ o, R., Lu, S., McIntyre, B.W., Molldrem, J.J., Legge, G.B., and Ma, Q. (2009). LFA-1 affinity regulation is necessary for the activation and proliferation of naive T cells. J. Biol. Chem. 284, 12645–12653. Zhan, Y., Chen, R., Zheng, W., Guo, C., Lu, L., Ji, X., Chi, Z., and Yu, J. (2014). Association between serum magnesium and anemia: China health and nutrition survey. Biol. Trace Elem. Res. 159, 39–45. Zhang, K., and Chen, J. (2012). The regulation of integrin function by divalent cations. Cell Adh. Migr. 6, 20–29. 30 KEY RESOURCE TABLE REAGENT or RESOURCE SOURCE IDENTIFIER Antibodies Mouse Anti-Human CD3 BioLegend Cat# 300333 Mouse Anti-Human CD28 BioLegend Cat# 302943 Armenian Hamster Anti-Mouse CD3 BioLegend Cat# 100359 Syrian Hamster Anti-Mouse CD28 BioLegend Cat# 102121 Mouse Anti-Human CD11a/CD18 (LFA-1) (clone m24), BioLegend Cat# 363402 unlabeled, AF488, PE Cat# 363406 Cat# 363404 Mouse Anti-Human CD11a (clone TS2/4), unlabeled, FITC BioLegend Cat# 350602 Cat# 350604 Mouse Anti-Human CD18 (clone KIM127), unlabeled M. Sperandio, LM custom University Munich and InVivo Biotech Services GmbH Mouse Anti-Human CD18 (clone CBR LFA-1/2) BioLegend custom Mouse Anti-Human CD18 (clone TS1/18) BioLegend Cat# 302116 Mouse Anti-Human CD54 (ICAM-1) ThermoFisher Cat# 16-0549-82 Mouse Anti-Human CD102 (ICAM-2) ThermoFisher Cat# BMS109 Mouse Anti-Human CD50 (ICAM-3) ThermoFisher Cat# BMS111 Mouse IgG1 κ Isotype control BioLegend Cat# 400166 Mouse IgG1 κ Isotype control ThermoFisher Cat#16-4714-82 Mouse IgG2a κ Isotype control ThermoFisher Cat#16-4724-82 InVivoMab Anti-Mouse CD3ε F(ab')2 fragment BioXCell Cat# BE0001-1FAB InVivoMab Anti-Mouse PD-1 (CD279) BioXCell Cat# BE0146 InVivoMAb Anti-Mouse CD16/CD32 BioXCell Cat# BE0307 InVivoMab rat IgG2a isotype control BioXCell Cat# BE0089 Rabbit Anti-Human/Mouse phospho-FAK (Tyr397) ThermoFisher Cat# 700255 Phospho-Tyrosine (P-Tyr-1000) MultiMab Rabbit mAb mix Cell Signaling Cat# 8954 Technology Phospho-p44/42 MAPK (Erk1/2) (Thr202/Tyr204), Rabbit Cell Signaling Cat# 4370 mAb Technology Phospho-c-Jun (Ser73), Rabbit mAb Cell Signaling Cat# 3270 Technology p44/42 MAPK (Erk1/2), Rabbit mAb Cell Signaling Cat# 137F5 Technology c-Jun, Rabbit mAb Cell Signaling Cat# 60A8 Technology β-Actin, Mouse mAb Cell Signaling Cat# 8H10D10 Technology Recombinant Anti-gamma Tubulin Abcam Cat# ab179503 Purified Mouse Anti-Human Perforin BD Biosciences Cat# 556434 Goat anti-Mouse IgG (H+L) Highly Cross-Adsorbed Secondary ThermoFisher Cat# A-11031 Antibody, Alexa Fluor 568 Goat anti-Rabbit IgG (H+L) Highly Cross-Adsorbed Secondary ThermoFisher Cat# A-21245 Antibody, Alexa Fluor 647 Goat anti-Mouse IgG1 Cross-Adsorbed Secondary Antibody, ThermoFisher Cat# A-21121 Alexa Fluor 488 Mouse anti-ERK1/2 Phospho (Thr202/Tyr204), AF647 BioLegend Cat# 369504 Hamster Anti-Mouse CD3, BUV805 BD Biosciences Cat# 749276 Rat anti-mouse CD4, BUV496 BD Biosciences Cat# 612952 Hamster Anti-mouse CD8a, eFluor 450 eBioscience Cat# 48-0081-82 31 Rat Anti-mouse CD11b, APC/Cyanine7 BioLegend Cat# 101226 Armenian Hamster Anti-Mouse CD11c, FITC BioLegend Cat# 117306 Rat Anti-Mouse CD19 Antibody, BB515 BD Biosciences Cat# 564509 Rat Anti-Mouse CD25 Antibody, PE-Cyanine5.5 eBioscience Cat# 35-0251-82 Rat Anti-Mouse CD45, BUV395 BD Biosciences Cat# 565967 Armenian Hamster Anti-Mouse CD80 Antibody, FITC BioLegend Cat# 104706 Hamster Anti-Mouse CD103, BV650 BD Biosciences Cat# 748256 Armenian Hamster Anti-Mouse CD80, Brilliant Violet 605 BioLegend Cat# 104729 Rat Anti-Mouse CD206, Brilliant Violet 711 BioLegend Cat# 141727 Hamster Anti-Mouse CD183 (CXCR3), BUV737 BD Biosciences Cat# 741895 Rat Anti-Mouse F4/80, Alexa Fluor 647 BioLegend Cat# 123122 FOXP3, rat mAb, APC eBioscience Cat# 17-5773-82 Granzyme B, mouse mAb, PE-eFluor 610 Inivtrogen Cat# 61-8898-82 Ki-67, rat mAb, Alexa Fluor 532 eBioscience Cat# 58-5698-82 Rat Anti-mouse CD11a (LFA-1alpha), Super Bright 436 ThermoFisher Cat# 62-0111-82 Rat Anti-Mouse Ly-6G, BUV563 BD Biosciences Cat# 612921 Rat Anti-Mouse Ly-6c, PerCP BioLegend Cat# 128028 Rat Anti-Mouse I-A/I-E (MHCII), BV510 BioLegend Cat# 107636 Rat Anti-Mouse CD335 (NKp46), BV563 BD Biosciences Cat# 741435 Rat Anti-Mouse CD279 (PD-1), Brilliant Violet 785 BioLegend Cat# 135225 Rat Anti-Mouse CD274 (PD-L1), Brilliant Violet 421 BioLegend Cat# 124315 Rat Anti-Mouse TCF7/TCF1, Alexa fluor 700 R&D Systems Cat# FAB8224N Mouse Anti-Mouse CD366 (TIM-3), BB700 BD Biosciences Cat# 747619 Rat Anti-Mouse CD8a, FITC BioLegend Cat# 100706 Rat Anti-Mouse CD11b, PE/Cyanine5 BioLegend Cat# 101210 Armenian Hamster Anti-Mouse CD11c, PE/Cyanine5 BioLegend Cat# 117316 Rat Anti-Mouse/Human CD45R/B220, PE/Cyanine5 BioLegend Cat# 103210 Rat Anti-Mouse F4/80 BioLegend Cat# 123112 Rat Anti-Mouse CD69, APC BioLegend Cat# 104514 Rat Anti-Mouse CD107a (LAMP-1) BioLegend Cat# 121620 Tetramer/PE - H-2 Kb OVA (SIINFEKL) Tetramers core facility, N/A University of Lausanne Rat Anti-Mouse CD11a, FITC BioLegend Cat# 101106 Rat Anti-Mouse CD11a/CD18, Pacific Blue BioLegend Cat# 141014 Rat Anti-Mouse CD107a (LAMP-1), PE/Cyanine7 BioLegend Cat# 121620 Rat Anti-Mouse CD4, R718 BD Biosciences Cat# 566939 Rat Anti-Mouse CD8, BUV805 BD Biosciences Cat# 612898 Rat Anti-Mouse CD45, APC-Cy7 BioLegend Cat# 103115 Armenian Hamster Anti-Mouse CD69, Brilliant Violet 650 BioLegend Cat# 104541 Armenian Hamster Anti-Mouse TCRβ, Brilliant Violet 711 BioLegend Cat# 109243 Armenian Hamster Anti-Mouse CD11c, Brilliant Violet 650 BioLegend Cat# 117339 Armenian Hamster Anti-Mouse CD69, PE/Cyanine5 BioLegend Cat# 104510 Armenian Hamster Anti-Mouse CD80, FITC BioLegend Cat# 104706 Armenian Hamster Anti-Mouse TCR γ/δ Antibody, PE BioLegend Cat# 118108 Mouse Anti-Mouse NK-1.1, APC BioLegend Cat# 108710 Rat Anti-Mouse Anti-Mouse Ly-6G/Ly-6C (Gr-1), Brilliant BioLegend Cat# 108457 Violet 510 Rat Anti-Mouse CD11a, PE/Cyanine7 BioLegend Cat# 153108 Rat Anti-Mouse CD19, PE/Cyanine5 BioLegend Cat# 115510 Rat Anti-Mouse CD62L, Pacific Blue BioLegend Cat# 104424 32 Rat Anti-Mouse CD8, PE BioLegend Cat# 100708 Rat Anti-Mouse CD4, Brilliant Violet 650 BioLegend Cat# 100555 Rat Anti-Mouse F4/80, PE/Cyanine7 BioLegend Cat# 123114 Rat Anti-Mouse Ly-6C, FITC BioLegend Cat# 128005 Rat Anti-Mouse/Human CD11b, PE BioLegend Cat# 101208 Rat Anti-Mouse/Human CD44, FITC BioLegend Cat# 103006 Mouse Anti-Human CD54 (ICAM-1), PE BioLegend Cat# 353106 Mouse Anti-Human CD102 (ICAM-2), PE BioLegend Cat# 328506 Mouse Anti-Human CD50 (ICAM-3), PE BioLegend Cat# 330005 Mouse Anti-Human CD25, APC BD Biosciences Cat# 340939 Mouse Anti-Human CD45RA, Pacific Blue Beckmann Cat# 2H4LDH11LDB9 Mouse Anti-Human CD62L, APC ImmunoTools Cat# 21279626 Mouse Anti-Human CD69, PerCP, FITC BioLegend Cat# 310928 Cat# 310904 Mouse Anti-Human CD71, PE BioLegend Cat# 334106 Mouse Anti-Human CD107a (LAMP-1), Alexa Fluor 647 BioLegend Cat# 328612 Mouse Anti-Human CD98, FITC BioLegend Cat# 315603 Mouse Anti-Human TCR V β13.1, FITC, PE/Cyanine7 BioLegend Cat# 362404 Cat# 362406 Mouse Anti-Human TNF, APC, PE BioLegend Cat# 502912 Cat# 502909 Blincyto (Blinatumomab) Amgen N/A Bacterial and virus strains Listeria monocytogenes expressing chicken Ovalbumin E. Palmer, University of N/A (AA134–387) Basel NY-ESO-1 TCR lentiviral vector, codon optimized, pairing Provided to M. Trefny N/A optimized: pRRL 131 (WT) T2A 1xATG Cys from M. Hebeisen and N. Rufer, University of Lausanne Anti-CD19-CD28z-T2A-copGFP Provided to M. Trefny N/A from W. Schamel, University of Freiburg Biological samples Human Peripheral Blood Buffy Coat Blood Donation Center N/A Basel and Lausanne, Switzerland Human AB+ male serum Blood Donation Center N/A Basel, Switzerland Chemicals, peptides, and recombinant proteins BIRT377 Tocris XVA143 Roche N/A NY-ESO-1 peptide 9C: SLLMWITQC >95% purity EZ Biolabs custom Ovalbumin EndoFit InvivoGen Cat# vac-pova-100 OVA (257-264) Peptide Fragment Eurogentec Cat# AS- 60193 OVA-G4 Peptide, SIIGFEKL, OVA (257-264) Variant Eurogentec Cat# AS-64384 OVA (257-264) Variant, SIIQFERL, Q4R7 Eurogentec custom OVA-Q4H7 Peptide, pQ4H7, SIIQFEHL, OVA (257-264) Variant Eurogentec Cat# AS-64402 DAPI Sigma Aldrich Cat# D9542 Phalloidin-iFluor 555 Reagent ABCAM Cat# ab176756 CellTrace Far Red Cell Proliferation Kit ThermoFisher Cat# C34564 CellTrace Violet Cell Proliferation Kit ThermoFisher Cat# C34557 CellTrace CFSE Cell Proliferation Kit ThermoFisher Cat# C34554 33 Mag-Fluo-4, AM, cell permeant ThermoFisher Cat# M14206 Fluo-4, AM, cell permeant ThermoFisher Cat# F14201 Fura Red, AM, cell permeant ThermoFisher Cat# F3021 Calbryte 520 AM AAT Bioquest Cat# 20653 BioTracker NucView 488 Green Caspase-3 Dye Merck Millipore Cat# SCT100 BioTracker NucView 405 Blue Caspase-3 Dye Merck Millipore Cat# SCT102 NucView 488 Caspase-3 Enzyme Substrate Biotium Cat# 10402 CellEvent Caspase-3/7 Green Detection Reagent ThermoFisher Cat# C10423 Incucyte Caspase-3/7 Dye for Apoptosis Sartorius Cat# 4440 2-NBDG Abcam Cat# ab235976 Proleukin (Aldesleukin, recombinant IL-2) Novartis N/A Human IL-7 Mylteni Biotec Cat# 130-095-363 Human IL-15 Mylteni Biotec Cat# 310-095-765 Zombie UV Fixable Viability Kit BioLegend Cat# 423108 Zombie Aqua Fixable Viability Kit BioLegend Cat# 423102 Zombie Green Fixable Viability Kit BioLegend Cat# 423112 Zombie Red Fixable Viability Kit BioLegend Cat# 423110 LIVE/DEAD Fixable Blue Dead Cell Stain Kit ThermoFisher Cat# L34962 LIVE/DEAD Fixable Near-IR Stain Kit ThermoFisher Cat# 15519340 CD8 MicroBeads human Mylteni Biotec Cat# 130-045-201 CD8a (Ly-2) MicroBeads mouse Mylteni Biotec Cat# 130-117-044 LS Cloumns Mylteni Biotec Cat# 130-042-401 T cell stimulation and expansion kit Mylteni Biotec Cat# 130-091-441 Cas9 QB3 MacroLab, UC N/A Berkeley EasySep Mouse Naïve CD8+ T Cell Isolation Kit Stem Cell Technologies Cat# 19858 EasySep Mouse CD8+ T Cell Isolation Kit Stem Cell Technologies Cat# 19853 EasySep Mouse CD4+ T Cell Isolation Kit Stem Cell Technologies Cat# 19852 Critical commercial assays P4 Primary Cell 4D-Nucleofector X Kit L Lonza Cat# V4XP-4024 Cell Line Nucleofector Kit V Lonza Cat# VCA-1003 ELISA MAX Deluxe Set Human IFN-γ BioLegend Cat# 430104 LEGENDplex Human Th1 Panel (5-plex) BioLegend Cat# 741036 LEGENDplex Mouse Th1 Panel (5-plex) BioLegend Cat# 740025 ProQuantumMouse Granzyme B Immunoassay Kit Inivtrogen Cat# A44238 Mouse ANA (Anti-nuclear Antibody) ELISA Kit Hoelzl Biotech Cat# MBS7606315-96 Deposited data N/A Experimental models: Cell lines Jurkat, Clone E6-1 ATCC RRID:CVCL_0367 HEK-293T ATCC RRID:CVCL_0063 T2 A. Zippelius, University RRID:CVCL_2211 of Basel Ramos A. Zippelius, University RRID:CVCL:0597 of Basel LCL G. Bantug, University of Basel EL4 A. Zippelius, University RRID:CVCL_0255 of Basel PC3-PIP Provided to M. Irving by N/A A. Rosato, University of Padua 34 MC38-OVA Provided to A. Marti RRID_CVCL_XY96 from P. Romero, University of Lausanne Experimental models: Organisms/strains C57BL/6NRj Animal Facility N/A University of C57BL/6NCrl Animal Facility N/A University of Basel, Charles River, Janvier OT-I (B6.129S6-Rag2tm1Fwa Tg(TcraTcrb)1100Mjb) Taconic, Animal Facility Cat# 2334 University of Basel LFA-1 KO (B6.129S7-Itgaltm1Bll/J) Jackson Laboratory, Cat# 005257 Animal Facility University of Basel NSG (NOD.Cg-PrkdcIl2rgSzJ) Animal Facility N/A University of Basel and University of Lausanne Oligonucleotides Hs.Cas9.ITGAL.1.AA (TGCCCGACTGGCACTGATAG) Integrated DNA N/A Technologies IDT Mm.Cas9.ITGAL.1.AB (CACATAGTTGATGGCACGAA) Integrated DNA N/A Technologies IDT RISPR-Cas9 Negative Control crRNA #1 Integrated DNA Cat# 224163224 Technologies IDT Alt-R CRISPR-Cas9 tracrRNA Integrated DNA Cat# 222427350 Technologies IDT Recombinant DNA N/A Software and algorithms FlowJo BD Biosciences https://www.flowjo.co m/ GraphPad Prism GraphPad Software Inc. https://www.graphpa d.com/scientific- software/prism/ ImageLab Biorad https://www.bio- rad.com/ FiJi National Institute of https://imagej.net/sof Health tware/fiji/ Huygens - Huygens Remote Manager - Deconvolution N/A https://www.huygens- rm.org/wp/ OMERO - The Open Microscopy Environment OME https://www.openmic roscopy.org/ Imaris Bitplane https://imaris.oxinst.c om/ BioRender N/A https://biorender.com / Other AIN-76A Rodent Diet Research Diets, Inc. Cat# D1001 AIN-76A Rodent Diet Without Added Magnesium Research Diets, Inc. Cat# D16601 Lead contact Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Christoph Hess (ch818@cam.ac.uk, or chess@uhbs.ch). 35 Material availability This study did not generate new unique reagents. Data and code availability This study did not generate or analyze unique datasets or code. EXPERIMENTAL MODELS AND SUBJECT DETAILS Cell Lines Primary T cells were obtained from Buffy coats of healthy donors (Blood donor center, University Hospital Basel or Blood donor center Lausanne). Jurkat T cells (Clone E61, TIB-152) and HEK-293T were originally purchased from ATCC. T2, Ramos and EL4 cells were kindly provided by A. Zippelius (University of Basel). MC38-OVA cells were provided by P. Romero (University of Lausanne). PC3-PIP cell lines were provided by A. Rosato (University of Padua, Padova). Lymphoblastoid cell lines (LCLs) were provided by G. Bantug (University of Basel). For the culture of primary human CD8+ T cells, PHA-induced T cell blasts, Jurkat T cells, PC3-PIP, Ramos, LCL and T2 cells, RPMI-1640 medium (Invitrogen) was supplemented with heat-inactivated 10% fetal calf serum (HI FCS, Gibco), 50 U mL-1 penicillin (Invitrogen) and 50 μg mL-1 streptomycin (Invitrogen). Human REP T cells were expanded in AIM V medium (ThermoFisher) mixed 1:1 with RPMI-1640 (Invitrogen) supplemented with 10% human HI AB serum, 50 U mL-1 penicillin (Invitrogen) and 50 μg mL-1 streptomycin, 1 mM pyruvate (Gibco), 1% MEM Non-Essential Amino Acids (Gibco), 1% GlutaMAX (Gibco) and 3,000 U mL-1 human recombinant IL-2 (Proleukin, Novartis). Murine T cells und EL4 cells were kept in RPMI-1640 medium containing 10% HI FCS, 100 U mL-1 penicillin, 100 μg streptomycin, 0.29 mg mL-1 L-glutamine, 50 μM 2-Mercaptoethanol (Invitrogen). 293T human embryonic kidney (HEK-293T) were cultured in RPMI-1640 supplemented with 10% HI FCS, 2 mmol l- glutamine, 100 μg mL-1 penicillin and 100 U mL-1 streptomycin (all purchased from Invitrogen). MC38-OVA cells were maintained in RPMI-1640-Glutamax medium supplemented with 10% FCS, 50 U mL-1 penicillin and 50 μg mL-1 streptomycin, 1 mM sodium pyruvate, 50 μM 2-Mercaptoethanol and under geneticin selection (0.4 mg mL-1 G418). All reagents were purchased from Gibco. Magnesium-free medium was made in-house, using double distilled water supplemented according to manufacturer’s instruction with RPMI-1640 amino acid solution (Sigma Aldrich), RPMI-1640 vitamin solution (Sigma Aldrich), 1% GlutaMAX (Gibco), 25 mM HEPES (Gibco), 2 g L- 1 sodium bicarbonate (Sigma Aldrich), 2g L-1 glucose (Sigma Aldrich), 100 mg L-1 calcium nitrate (Sigma Aldrich), 400 mg L-1 potassium chloride (Sigma Aldrich), 6 g L-1 sodium chloride (Sigma Aldrich), 800 mg L-1 sodium phosphate dibasic (Sigma Aldrich), 1 mg L-1 Glutathion (Sigma Aldrich), 50 U m-1 penicillin and 50 μg mL-1 streptomycin and 10% HI dialyzed FCS (dFCS, Gibco). For functional readouts, the medium was either supplemented, as indicated, with 1.2 mM MgCl2 or 0.05 mM MnCl2 or left untreated (=0 mM Mg2+). Cells of every condition were washed initially twice in magnesium-free medium prior to any functional read out. Low background Mg2+ values in self-made medium was verified by ICP-MS (data not shown). Mice Information on sex and age of the mice used in each experimental setting can be found in the corresponding section. C57BL/6NRj, C57BL/6NCrl and NOD.Cg-PrkdcIl2rgSzJ (NSG) mice were bred and housed at specific pathogen free (SPF) conditions at the Universities of Basel or Lausanne. MHC class I-restricted OVA- specific T cell receptor (B6.129S6-Rag2tm1Fwa Tg(TcraTcrb)1100Mjb) (OT-I) transgenic, B6.129S7-Itgaltm1Bll/J 36 (LFA-1 KO) were originally purchased from Taconic (OT-I) or Jackson Laboratories (LFA-1 KO), and thereafter bred and housed at SPF conditions at the University of Basel. For some experiments, C57BL/6NCrl were purchased from Charles River as well as Janvier, and maintained at SPF conditions and acclimatized for 1 week prior to experiments at the animal facilities of the Universities of Geneva or Basel, respectively. Anti-CD19 CAR T cell experiments were conducted with NSG mice at the University of Basel. Anti-PSMA CAR T cell experiments were conducted with NSG mice which were bred and housed in a specific and opportunistic pathogen-free animal facility in the Oncology Department of the University of Lausanne. All experiments were conducted in accordance to the Swiss Federal Veterinary Office guidelines and were approved by the Cantonal Veterinary Office (Canton of Basel-Stadt, Vaud and Geneva). All cages provided free access to food and water. During experimentation, all animals were monitored at least every other day for signs of distress and, if required, body weight was measured three times a week. Mice were sacrificed at the endpoint by carbon dioxide overdose. Human Data The retrospective analysis of diffuse large B cell lymphoma (DLBCL) cohort treated with Axicabtagene ciloleucel (Axi-cel) was approved by the Institutional Review Board of MD Anderson Cancer Center and conducted in accordance with institutional guidelines and the principles of the Declaration of Helsinki. All included patients provided written informed consent. The SAKK16/14 trial (NCT02572843) was conducted in accordance with the principles of the Declaration of Helsinki. The protocol was approved by the ethics committee of each participating site. Written informed consent was obtained from all patients. METHOD DETAILS Chemicals LFA-1 inhibitor studies were performed using BIRT377 (Tocris) and XVA143 (Roche). Both inhibitors were used at 10 μM. TCR independent activation of T cells were conducted with PMA (50 ng mL-1; Sigma Aldrich) and ionomycin (1 μg mL-1; Sigma Aldrich). All chemicals were aliquoted in DMSO and stored at -20°C until used. aliquoted in DMSO and stored at -20°C until used. Flow cytometry Either a BD Fortessa LSR II (BD Bioscience), Cytek Aurora (Cytek Biosciences) or Cytoflex S (Beckmann) flow cytometer were used for flow cytometry. For analysis of surface markers, T cells were harvested at indicated time points after activation in vitro, washed once in cold PBS and, if required, stained with Fixable Viability Dyes for 15 min at 4°C. Surface markers were stained with appropriate antibodies for 20 min at 4°C. For stainings of cell suspension from murine organs, cells were additionally pre-incubated with anti-mouse Fc block 10 µg mL-1 (anti-CD16/CD32, BioXCell). For intracellular TNF staining, cells were activated for 4 h. During the final 2 h of activation, cells were treated either with brefeldin A solution (BioLegend) to block cytokine secretion. Cells were then washed and fixed for 20 min at RT (fixation/permeabilization solution, BD Biosciences) and washed with permeabilization buffer (BD Biosciences) prior to staining for 45 min and further washing before acquisition. For analysis of protein phosphorylation, T cells were stimulated as indicated and stained as previously described (Krutzik and Nolan, 2003). Briefly, for assessment of phosphorylation of ERK1/2, c-Jun and FAK, T cells were activated for 45 min either by CD3/28 Ab, antigen-pulsed target cells or PMA/Ionomycin. Cells were fixed by adding 8% Paraformaldehyde (PFA) (ThermoFisher) directly into the culture medium to obtain a final 37 concentration of 4% PFA. Cells were incubated for 15 min at RT, washed with FACS buffer, followed by permeabilization with ice cold methanol at 4°C for 5 min. After washing with FACS buffer, cells were stained at room temperature for 30 min, washed and acquired. For evaluation of activation-induced LFA-1 conformation, T cells were cultured in respective medium for 45 min ± activation (CD3/28 Ab or antigen-pulsed target cells). For probing of open headpiece anti-human CD11a/CD18 (clone m24) was directly added to the medium and incubated for 15 min at 37°C followed by incubation on ice for 20 min. Cells were then washed twice in FACS Buffer and fixed in 2% PFA, incubated at room temperature for 20 min and washed with FACS Buffer before acquisition. For assessment of LFA-1 extension, the anti-human CD18 (clone KIM127) was used. The cells were activated for 45 min, fixed directly in medium with PFA at a final concentration of 2%, washed and then incubated with KIM127. For 2-NBDG uptake assays, previously activated T cells were co-incubated for last 45 min of experiment directly with a final concentration of 20 μM 2-NBDG (Abcam). Cells were washed twice in FACS buffer before acquisition. The following antibodies were used: Human in vitro activation: CD11a (FITC, BioLegend), CD18 (PE, BioLegend), CD25 (APC, BD Biosciences), CD45RA (Pacific Blue, Beckmann), CD62L (APC, Immuno Tools), CD69 (PerCP, FITC, BioLegend), CD71 (PE, BioLegend), CD107a (AF647 and PE-Cy7, both BioLegend), CD98 (FITC, BD Bioscience and BioLegend), ICAM-1 (PE, BioLegend), ICAM-2 (PE, BioLegend), ICAM-3 (PE, BioLegend), m24 epitope LFA-1 (PE, BioLegend), phospho-c- Jun (Ser73, unlabeled, Cell Signaling Technology), phospho-ERK1/2 (Thr202/Thr204, AF647, BioLegend), phospho- FAK (Tyr397, unlabeled, ThermoFisher), goat anti-rabbit IgG (AF488, ThermoFisher), goat anti-mouse IgG1 (AF488, ThermoFisher), TCR V β13.1 (FITC and PE-Cy7, BioLegend), TNF (PE, APC, BioLegend), Viability Dye (Aqua Zombie, BioLegend; Zombie Green, BioLegend). Murine in vitro activation: CD11a (FITC, BioLegend), LFA-1 (BV421, BioLegend), CD8a (FITC, BioLegend), CD69 (APC, BioLegend), CD107a (PE-Cy7, BioLegend), phospho-ERK1/2 (Thr202/Thr204, AF647, BioLegend), phospho- FAK (Tyr397, unlabeled, ThermoFisher), goat anti-rabbit IgG (AF488, ThermoFisher), Viability Dye (Aqua Zombie, BioLegend). Murine immune cell subset characterization in vivo (comprising 3 individual stainings): CD4 (BV650, BioLegend), CD8 (PE, BioLegend), CD11a (PE/Cy7, BioLegend), CD11b (PE, BioLegend), CD11c (BV650, BioLegend), CD19 (PE/Cy5, BioLegend), CD44 (FITC, BioLegend), CD62L (PB, BioLegend), CD69 (PE/Cy5, BioLegend), CD80 (FITC, BioLegend), F4/80 (PE/Cy7, BioLegend), Gr-1 (BV510, BioLegend), Ly6-C (FITC, BioLegend), NK1.1 (APC, BioLegend), TCR γ/δ (PE, BioLegend), LIVE/DEAD Fixable Near-IR (ThermoFisher) Polyclonal CD8 T cell activation in vivo with anti-CD3ε F(ab’)2 fragment: CD4 (R718, BD Bioscience), CD8 (BUV805, BD Bioscience), CD45 (APC-Cy7, BioLegend), CD69 (BV650, BioLegend), TCRβ (BV711, BioLegend), LIVE/DEAD Fixable Blue Dead Cell Stain Kit (ThermoFisher). Murine peritonitis model: CD8 (FITC, BioLegend), CD11b (PE-Cy5, BioLegend), CD11c (PE-Cy5, BioLegend), CD69 (APC, BioLegend), CD107a (PE/Cy7, BioLegend), B220 (PE-Cy5, BioLegend), F4/80 (PE-Cy5, BioLegend), Tetramer H-2 Kb OVA (PE, Tetramer core facility, University of Lausanne), Viability Dye (Zombie Red, BioLegend,). MC38-OVA tumor model: CD3 (BUV805, BD Biosciences), CD4 (BUV496, BD Biosciences), CD8 (eFluor450, eBioscience), CD11b (APC-Cy7, BioLegend), CD11c (FITC, BioLegend), CD19 (BB515, BD Biosciences), CD25 (PE- Cy5.5, eBioscience), CD45 (BUV385, BD Biosciences), CD80 (BV605, BioLegend), CD103 (BV650, BD Biosciences), 38 CD206 (BV711, BioLegend), CXCR3 (BUV737, BD Biosciences), F4/80 (AF647, BioLegend), FoxP3 (APC, eBioscience), GzmB (PE-eFluor610, Inivtrogen), Ki67 (AF532, eBioscience), LFA-1 (SB436, ThermoFisher), Ly-6G (BUV563, BD Biosciences), Ly-6c (PerCP, BioLegend), MHCII (BV510, BioLegend), NKp46 (BUV563, BD Biosciences), PD-1 (BV785, BioLegend), PD-L1 (BV421, BioLegend), TCF-7 (AF700, R&D Systems), Tim-3 (BB700, BD Biosciences), Zombie UV Fixable Viability Kit (BioLegend). Human naïve and memory T cell isolation Blood samples were obtained from healthy male and female donors (18-65 years old) as buffy coats after written informed consent (Blood donor center, University Hospital Basel). Peripheral blood mononuclear cells (PBMCs) were isolated by standard density-gradient centrifugation protocols (Lymphoprep; Fresenis Kabi). MACS beads and LS columns (both Milteny Biotec) were used to sort CD8+ positive T cells. The positively selected CD8+ T cells were incubated with APC anti-CD62L mAb (ImmunoTools) and Pacific Blue anti-CD45RA (Beckman Coulter). Naïve and EM CD8+ T cells were identified as CD62L+ CD45RA+ and CD62L– CD45RA– populations, respectively. Cell sorting was performed using a BD FACSAria III or BD influx cell sorter (BD Bioscience). Cells were rested for 24 h at 37°C prior to further experiments. Metabolic assays A Seahorse XF-96e extracellular flux analyzer (Seahorse Bioscience, Agilent) was used to determine the metabolic profile of cells. T cells were plated (2×105 cells/well) onto Celltak (Corning, USA) coated cell plates. Experiments were carried out in unbuffered, serum- and Mg2+-free self-made medium. Medium was reconstituted with 1.2 mM MgCl2 or 0.05 MnCl2 as indicated in individual experiments. Reconstitution of cations was either present from beginning of experiment or applied onto plated cells via the instrument’s multi-injection port. All following concentration represent final well concentrations of indicated substance. Human T cells were activated by injection anti-CD3 Ab (1 µg mL-1) and anti-CD28 Ab (10 µg mL-1). Murine T cells were activated by injection anti-CD3 Ab (5 µg mL-1) and anti-CD28 Ab (2.5 µg mL-1). Glycolytic activity was quantified by subtracting maximal ECAR from baseline ECAR-measurements. Mitochondrial perturbation experiments were carried out by sequential addition of oligomycin (1 µM, Sigma Aldrich), FCCP (2 µM, Carbonyl cyanide 4-(trifluoromethoxy) phenylhydrazone, Sigma Aldrich), and rotenone (1 µM, Sigma Aldrich). Glycolysis stress test was performed in medium as described above but devoid of glucose. Sequential injections of glucose (10 mM, Sigma Aldrich), oligomycin (1 µM) and 2-deoxy-glucose (50 mM, Sigma Aldrich). Oxygen consumption rates (OCR, pmol/min) and extracellular acidification rates (ECAR, mpH/min) were monitored in real time after injection of each compound. Flow cytometry-based Mg2+ and Ca2+ flux assay Naïve and EM CD8+ T cells were washed and incubated in Mg2+ or Ca2+ free medium for 1 h, followed by loading with MagFluo4 (ThermoFisher) or Fluo4 (ThermoFisher) and FuraRed (ThermoFisher) at a final concentration of 2 µM, in Mg2+- or Ca2+-free Dulbecco's phosphate-buffered saline (DPBS, ThermoFisher) for 30 min at 37°C. The cells were washed twice in DPBS, and 0.5-1×106 cells were acquired using an AccuriC6 flow cytometer (Becton Dickinson). The buffer was reconstituted with 3 mM MgCl2 or 3 mM CaCl2 at the indicated time points. Generation of human T cell blasts (PHA-blasts) 39 PBMCs were obtained as described above and activated with 10 μg mL-1 Phytohaemagglutinin (PHA, ThermoFisher) and 300 U mL-1 human recombinant IL-2 (Proleukin, Novartis). PHA-blasts were expanded by adding fresh IL-2 every 3-4 days. In vitro activation primary human T cells and PHA-blasts Unless stated otherwise, human EM CD8+ T cells and PHA-blasts were activated in presence of plate-bound anti- CD3 Ab (HIT3a, BioLegend) at 1 μg mL-1 and soluble anti-CD28 Ab (CD28.2, BioLegend) at 5 μg mL-1. Naïve CD8+ T cells were activated with in-house generated anti-CD3/anti-CD28 coated microbeads. Polybead microspheres (4.5 mm, Polyscience Eppenheim) were incubated with 1 μg anti-CD3 Ab and 10 μg anti-CD28 Ab. T cells were plated at 2×105 cells per well in flat bottom 96 well plates (Greiner Bio One) in self-made medium supplemented with 10% dFCS and indicated supplementation of cations or LFA-1 inhibitor. For analysis of surface markers, primary human T cells were activated for 24 h, PHA-blasts for 4 h. For phospho-flow and LFA-1 conformational states, T cells were activated for 45 min. When using inhibitory or activating antibodies or chemical LFA-1 inhibitors, T cells were pre-incubated with respective treatment for 20 min before activation. Venn Diagram Venn diagram visualizes the following gene lists: metal ion binding (GO:0046872), leukocyte cell-cell adhesion (GO:0007159), immunological synapse (GO:0001772) and differentially expressed protein between naïve and memory CD8+ T cells (memory>naïve) (van Aalderen et al., 2017). Venn diagram was made with InteractiVenn online tool (Heberle et al., 2015). ICAM blockade PHA blasts were pre-incubated with respective anti-ICAM or isotype control antibody (anti-ICAM-1 HA58, anti- ICAM-2 CBR-IC2/2, anti-ICAM-3 CBR-IC3/1 and matching isotype control, all from ThermoFisher) at a concentration of 10 μg mL-1 for 20 min at room temperature prior to activation. NY-ESO Peptides NY-ESO-9c peptide (SLLMWITQC) was purchased in >95% purity from EZ Biolabs. Lyophilized peptides were resuspended at 10 mM in sterile dimethyl sulfoxide (DMSO) and stored at -20°C until further use. T Cell Receptor Construct for REP T cells The lentiviral construct encoding for the codon-optimized WT LAU155 NY-ESO-1 T cell receptor a and b chains under an hPGK promotor separated by an IRES domain was kindly provided by M. Hebeisen and N. Rufer at the University of Lausanne (Hebeisen et al., 2013; Schmid et al., 2010). This TCR has a KD=21.4 µM for its endogenous NY-ESO-1 SLLMWITQC peptide. Generation of lentivirus for REP T cells To generate lentivirus, 2.5×106 low passage HEK293T cells were cultured in DMEM medium (ThermoFisher) and seeded into a 15 cm tissue-culture treated dish. After 3 days, 2nd generation LTR-containing donor plasmid, packaging plasmid pCMV-delta8.9 and the envelope plasmid VSV-G were mixed at a 4:2:1 ratio in non- supplemented Opti-MEM (ThermoFisher) and sterile filtered. This solution was then mixed with polyethyleneimine 25 kDa (Polysciences Inc.), also diluted in Opti-MEM at a DNA:PEI ratio of 1:3. 28 µg of DNA was transfected per 15 cm dish. After 2 days, supernatants were collected from cells (exchange medium) and filtered through a 0.45 μm PES filter. Supernatants were stored for 1 day at 4°C until the second batch of 40 supernatant was collected 24 h later. The supernatant containing lentiviral particles was concentrated by ultra- centrifugation at 40,000 x g for 2 h at 4°C, resuspended in 0.1% BSA in PBS, and frozen to -80°C until further use. Transduction of human T Cells for REP T cell production To generate NY-ESO-1 TCR specific T cells, human healthy donor PBMC were thawed and washed in PBS. CD8+ T cells were then isolated using the CD8+ microbeads (Miltenyi) according to the manufacturer’s instructions on an AutoMACS (Myltenyi). Isolated cells were washed and resuspended in medium supplemented with 150 U mL- 1 IL-2 and plated at 1.5 mio mL-1. CD8+ T cells were then activated at a 1:1 ratio with activation beads from T cell activation and expansion kit (Miltenyi) according to manufacturer’s instructions. 24 h later, NY-ESO-1 TCR lentiviral particles, produced as described above, were added at a multiplicity of infection (MOI) of 2. Cells were then expanded every 2 days with fresh medium and replenishing 50 UmL-1 IL-2 for 5 days. NY-ESO-1 TCR positive T cells were sorted using a FACSAria III or FACS SorpAria (BD) and re-stimulated with NY-ESO-9c peptide. A cell density of 0.5–2×106 cells mL-1 was maintained for expansion and 3,000 U mL-1 IL-2 replaced ever third day. After 1 week of expansion, cells were either stored in liquid nitrogen or further expanded and subsequently used for functional read outs as described below. REP T cells: activation in vitro and cytotoxicity assays REP T cells were incubated with T2 target cells in flat bottom 96 well-plate at a 1:1 ratio (4-6×104 each). In order to distinguish the different cell populations, REP T cells were labeled with CTV and T2 target cells with carboxyfluorescein diacetate succinimydyl ester (CFSE, Invitrogen) or CFTR (Invitrogen). Prior to co-incubation, T2 target cells were pulsed with NY-ESO-9c peptide at 10-8 M for 30 min in magnesium-free medium before being washed and re-suspended with REP T cells in magnesium-free medium supplemented with 10% dFCS at indicated cation or LFA-1 inhibitor concentration. For all co-incubation experiments, cells were allowed to sediment without centrifugation. For analysis of protein phosphorylation, co-incubation was terminated after 25 min as described above. In these experiments, Mn2+ concentration was 0.5 mM instead of 0.05 mM as in all other experiments. For degranulation assays, an anti-CD107a-AF647 Ab was added directly into culture medium throughout the entire co-incubation. After 4 h, cells were harvested, washed in cold FACS Buffer and gently fixed with PFA 2% for 15 min at room temperature. Cytotoxicity was examined with NucView 488 fluorogenic caspase- 3 substrate (Biotium). Fluorogenic caspase substrate was added to wells at the beginning of co-incubation at final concentration of 1 µM. After 45 min, cells were washed in FACS Buffer and gently fixed with PFA 2% for 15 min at room temperature. Imaging of immune synapse with confocal microscopy Pictures from immunofluorescence imaging were recorded on a Nikon Ti with a Yokogawa CSU-W2 spinning disk module on a Photometrics 95B (22 mm back-illuminated sCMOS) camera. A Nikon CFI Apo Lambda 60x objective or Nikon CFI Apo TIRF NA 1.49 100x objective was used with 1.515 oil mounted samples. Diode-pumped solid- state lasers at 405, 488, 561, and 647nm were used together with filters for DAPI (ET460/50nm), AF488 (ET525/50nm), AF555 (ET630/75nm) and AF647 (ET700/75nm) with a Quad BS Dichroic mirror. If required, raw nd2 format image stacks were deconvoluted using Huygens using a theoretical point spread function classical maximum likelihood estimation using 100 iterations and a quality stop criterion of 0.05. Immunofluorescent staining of immune synapse components of REP T cell/T2 cell conjugates 41 REP T cells and T2 target cells were individually labeled with CTV or CFSE. T2 cells were loaded with 10-7 M NY- ESO-9c peptide for 30 min. T2 and REP T cells were washed three times in serum- and Mg2+-free medium, mixed 1:1 and resuspended to a final concentration of 2.5×106 cells mL-1. Cell suspension was incubated 5 min at RT, before aliquoting in 50 μl onto glass multi-well slides (ThermoFisher) and incubating for indicated time. Cell conjugates were fixed for 20 min at RT in 4% methanol-free PFA (Sigma Aldrich), permeabilized with 0.1% Triton- X100 (Sigma Aldrich) in PBS for 5 min and quenched with 50 mM Glycine (Sigma Aldrich) in PBS for 20 min. Fixation was followed by blocking in 1% bovine serum albumin (Sigma Aldrich) in PBS (blocking buffer) for 45 min at 4°C. Primary antibodies were then incubated in the same blocking buffer for 1 h at RT or overnight at 4°C. Samples were then washed four times with blocking buffer, followed by incubation with secondary antibodies in blocking buffer at room temperature for 1 h. Slides were then mounted with Prolong Diamond Antifade Mountant (ThermoFisher) and analyzed after 24 h of curing. Of note, for visualization of extended LFA-1, m24 antibody was added directly in multi-well slides during co-incubation and cells were fixed after 10 min. The following reagents and antibodies were used for staining: CD11a/CD18 (clone m24, BioLegend), gamma Tubulin (clone EPR16793, Abcam), perforin (clone δG9, BD Bioscience), phospho-tyrosine P-Tyr-1000 (#8954S, Cell Signaling Technology), secondary goat anti-mouse (AF568, ThermoFisher), secondary goat anti-rabbit (AF647, ThermoFisher) and DAPI (Sigma Aldrich) Image analysis and quantitation of REP T cell-T2 target cell conjugates For analysis of extended LFA-1 and pan phospho-tyrosine intensity in REP T cells, confocal Z-stacks were acquired and analyzed using Imaris software (Bitplane). REP T cell volumes were identified using the surface-tool: source channel 4 (DAPI=CTV), 5 μm diameter cut-off with 0.369 μm surface detail and > 1689 voxels. Fluorescent signal of source channel 3 (AF555=LFA-1 m24 staining) outside of REP T cell-surface was masked and extended LFA-1 was quantified with surface-tool: source channel 5 (masked AF555), 1.39 μm diameter cut-off with 0.369 μm surface detail and > 154 voxels. Number of LFA-1 m24-objects were normalized to REP T cell number per field of view. Pan phospho-tyrosine intensity was determined by plotting median intensity of source channel 2 (AF647) per REP T cell volume identified as described above. For analysis of perforin and centrosome polarization, position of REP T and T2 target cells in confocal Z-stacks were characterized with spot-tool: Source channel 1 (AF488=CFSE) with the following parameters 10 μm estimated diameter and spot classification with quality threshold > 27.4 was used to identify T2 target cells and source channel 4 (DAPI=CTV) with 8 μm estimated diameter and spot classification with quality threshold > 37.8, for REP T cells respectively. For detection of centrosome and perforin granules in REP T cells surface-tool was applied on source channel 4 (DAPI=CTV), 5 μm diameter cut-off of with 0.369 μm surface detail and > 2014 voxel cut-off. Fluorescent signal of source channel 1 (AF488=perforin) and source channel 2 (AF647=centrosome) outside of defined REP T cell surface was masked and positions of respective objects were identified with spot-tool: 0.8 μm estimated diameter and spot classification with quality threshold > 1365 for perforin and 1.84 μm estimated diameter and spot classification with quality threshold > 637 for centrosome. X, Y, Z values of each individual spot (cells, perforin granule and centrosome) were subjected to further analysis using script in R. For each spot we identified its parent T cell and for each T cell its nearest tumor cell using the RANN package. T cell-cancer cell interactions were considered if their centers were between 3 and 25 µm apart, which reduces distant non-interacting cells 42 and artefacts. Then the spot-to-cancer-cell distance was divided by the cell-to-cell distance to calculate the polarization ratio for each interaction pair. Median polarization ratios between conditions and field of views were compared in Graphpad Prism. Immunoblot analysis Activated memory T cells and PHA-blasts were lysed in RIPA buffer (ThermoFisher) containing protease- and phosphatase-inhibitors (Roche, #05 892 970 001 and #04 906 837 001), and protein concentrations determined with a BCA protein assay kit (ThermoFisher). Whole-cell lysates were denatured with 4x Laemmli buffer and separated by 4-20% SDS-PAGE and transferred to nitrocellulose or PVDF membranes (Biorad). Membranes were probed with the following primary antibodies: P-p44/42 MAPK (Erk1/2, Thr202/Tyr204, #D13.14.4E, Cell Signaling Technology), P-c-Jun (S73, #D47G9, Cell Signaling Technology), p44/42 MAPK (Erk1/2, #137F5, Cell Signaling Technology), c-Jun (#60A8, Cell Signaling Technology) and Actin (#8H10D10, Cell Signaling Technology). Blots were then stained with HRP-conjugated anti-rabbit or anti-mouse (both from Jackson ImmunoResearch Laboratories) secondary antibodies. SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher) according to the manufacturer’s instructions and ChemiDoc Imaging System (Biorad) were used for detection and FiJi software for quantification. CRISPR-Cas9 editing of murine OT-I cells Single cell suspensions were made from lymph nodes and spleens harvested from OT-I mice (male and female, 6-10 weeks, equal distribution of sex and age). 2×106 mL-1 OT-I lymphocytes were resuspended in medium containing 100 ng mL-1 OVA257–264 peptide (Eurogentec) and 100 U mL-1 of IL-2 (Proleukin) and incubated for 48 h. OT-I cells were nucleofected with the P4 Primary Cell 4D-Nucleofector (Lonza) according to manufacturer’s instructions using 4D-Nucleofector (Lonza). Briefly, 1×107 activated OT-I T cells were resuspended in 100 μl of Nucleofector Solution and combined with RNP solution. crRNAs were selected from predesigned CRISPR-Cas9 guide RNAs Tool from IDT. Product ID and sequences are listed in Supplemental Table 1. Per reaction, 900 pmol crRNA (IDT) or 900 pmol negative control crRNA #1 (IDT) were mixed with 900 pmol trRNA (IDT) in nuclease-free duplex buffer (IDT), annealed at 95°C for 5 min and added to 300 pmol μM Cas9 (QB3 MacroLab, UC Berkeley) followed by incubation at room temperature for at least 10 min. An appropriate nucleofector program was applied. OT-I lymphocytes were rested in Mouse T Cell Nucleofector Medium (Lonza) for 12 h and then washed and seeded in fresh medium at 106 mL-1 in round bottom 96 well-plates with 500 U mL-1 IL-2. Fresh IL-2 was added on a daily basis. Knock-out efficiency was validated by flow cytometry and purified by cell sorting. CRISPR-Cas9 editing of human Jurkat T cells Jurkat T cells were nucleofected using the AMAXA cell line V nucleofection kit (Lonza) according to manufacturer’s instructions using 2b Nucleofector (Lonza). Briefly, 1×106 Jurkat T cells were resuspended in 100 μl of Nucleofector Solution and combined with RNP solution. crRNAs were selected from predesigned CRISPR- Cas9 guide RNAs Tool from IDT. Product ID and sequences are listed in Supplemental Table 1. crRNA (IDT) or negative control crRNA #1 (IDT) and trRNA (IDT) were mixed at a 1:1 ratio to a final concentration of 50 μM in nuclease-free duplex buffer (IDT), annealed at 95°C for 5 min and added to 40 μM Cas9 (QB3 MacroLab, UC Berkeley) followed by incubation at room temperature for at least 10 min. An appropriate nucleofector program 43 was applied. Knock-out efficiency was validated by flow cytometry and purified by cell sorting. Jurkat T cells were initially expanded for 1 week and then stored in liquid nitrogen until further use. Plate reader-based Ca2+ flux assay Murine OT-I CTLs were loaded with Calbryte 520 AM (AAT Bioquest) and Jurkat T cells were loaded with Fluo4 (ThermoFisher). Both calcium indicator dyes were used at a final concentration of 2 µM in Mg2+-free self-made medium for 30 min at 37°C. Cells were washed twice and plated at 5×105 per well (OT-I CTLs) and 2×105 per well (Jurkat T cells) in a black flat bottom 96 well-plate (Greiner BIO-one) which had been precoated with Poly-D- Lysine (Gibco) in case of OT-I CTLs or collagen (ThermoFisher) for Jurkat T cell respectively. An additional incubation for 15 min at 37°C allowed cells to adhere and Fluo4 probe to de-esterified completely. OT-I CTLs were stimulated with 10 µM OVA257–264 peptide (SIINFEKL, Eurogentec) and Jurkat T cells with 10 µg mL-1 anti- CD3. Fluorescence intensity over time was measure with a Tecan Spark M10 plate reader. The mean of fluorescent signal intensity was normalized to unstimulated baseline values. Murine CTLs: differentiation and cultivation Single cell suspensions were made from lymph nodes and spleens harvested from C57Bl/6 and LFA-1 KO mice (male and female, 6-10 weeks, equal distribution of sex and age). Naïve CD8+ T cells were isolated using a magnetic bead-based negative selection kit following the manufacturer’s recommendations (easySEP, Stem Cell Technologies). Naïve T cells (2×105 per well) were plated in presence of 5 μg anti-CD3 Ab (plate-bound) and 1μg anti-CD28 Ab (soluble; both from BioLegend) for 2 days in presence 100 U mL-1 of IL-2 (Proleukin). Cells were washed and seeded in fresh medium at 106 mL-1 in round bottom 96 well-plates with 500 U mL-1 IL-2. A cell density of 0.5–2 × 106 cells mL-1 was maintained for expansion and IL-2 was replaced on a daily basis. Functional read outs were carried out 7-19 days after initial activation and in the absence of IL-2. Murine CTLs: In vitro activation and cytotoxicity assay CTLs of WT or LFA-1 KO C57/Bl6 were activated for with plate-bound anti-CD3 Ab (145-2C11, BioLegend) at 0.05 μg mL-1 and soluble anti-CD28 Ab (37.51, BioLegend) at 1 μg mL-1 at 2×105 cells per well in a flat bottom 96 well plate. For stainings of surface activation markers, CTLs were activated for 8 h and for evaluation of 2-NBDG uptake, cells were activated for 6 h. For analysis of ERK1/2 phosphorylation, CTLs were stimulated with 5 μg mL- 1 anti-CD3 Ab and 1 μg mL-1 anti-CD28 for 20 min prior to fixation and permeabilization. For cytotoxicity assays with WT or LFA-1 KO C57/Bl6 derived CTLs, cytotoxicity was evaluated with NucView 488 fluorogenic caspase-3 substrate (Biotium). Prior to co-incubation, CTLs were labeled with CellTrace Violet (CTV, Invitrogen) and EL4 target cells with CellTrace Far Red (CFTR, Invitrogen). Both cell population were then co-incubated at a CTL- target cell ratio of 3:1 (1.5×105 CTLs and 5×104 EL4 target cells) in presence of 10 μg mL-1 PHA for 4 h in a flat bottom 96 well plate. Caspase-3 substrate was added for final 45 min of incubation at final concentration of 1 μM. Cells were harvested, washed in FACS Buffer and fixed with PFA 2% for 15 min at RT prior to analysis by FACS. For cytotoxicity assays with OT-I derived CTLs were labeled with CTV, EL4 target cells were labeled with CFTR and pulsed with OVA257–264 peptide (SIINFEKL, Eurogentec) or the altered peptide ligands R7 (SIIQFERL, Eurogentec), H7 (SIIQFEHL, Eurogentec) or G4 (SIIGFEKL, Eurogentec) at 1 μM for 30 min. EL4 target cells were washed 3 times prior to co-incubation at a CTL-target cell ratio of 3:1 (1.5×105 CTLs and 5×104 EL4 target cells) 44 in a flat bottom 96 well plate for 4h. Cytotoxicity was quantified with CellEvent Caspase-3/7 Green Detection Reagent (Invitrogen, ThermoFisher) as described above. Immunofluorescent staining of immune synapse components in Jurkat T cells and PHA blasts Cells were activated on species-appropriate anti-CD3-coated (1 μg mL-1 for human cells) glass-coverslips (ThermoFisher) for 2 min for phosphor-tyrosine analysist. Cell conjugates were fixed for 20 min at RT in 4% methanol-free PFA (Sigma Aldrich), permeabilized with 0.1% Triton-X100 (Sigma Aldrich) in PBS for 5 min and quenched with 50 mM Glycine (Sigma Aldrich) in PBS for 20 min. Fixation was followed by blocking in 1% bovine serum albumin (Sigma Aldrich) in PBS (blocking buffer) for 45 min at 4°C. Primary antibodies were then incubated in the same blocking buffer for 1 h at RT or overnight at 4°C. Samples were then washed four times with blocking buffer, followed by incubation with secondary antibodies in blocking buffer at room temperature for 1 h. Slides were then mounted with Prolong Diamond Antifade Mountant (ThermoFisher) and analyzed after 24 h of curing time at RT. The following reagents and antibodies were used for staining: phospho-tyrosine P-Tyr-1000 (multiple monoclonal antibodies, unlabeled, Cell Signaling Technology), secondary goat anti-mouse (AF488, ThermoFisher), secondary goat anti-rabbit (AF647, ThermoFisher), Phalloidin-iFluor (AF555, Abcam) and DAPI (Sigma Aldrich). Image analysis and quantitation of phospho-tyrosine signal intensity For measurement of pan-phospho tyrosine signal intensity, FiJi software was used. Series of confocal Z-stacks were displayed as maximum intensity z-projections using Z project- tool. Individual cells were selected manually with freehand selection-tool according to cell boundaries indicated by phalloidin staining and fluorescent intensity was measured accordingly with measure-tool. Magnesium-restricted diet Magnesium-restricted diet and matching control diet, based on the purified ingredient rodent diet AIN-76A, were purchased at Research Diets Inc. (USA). Sample collection of tissue interstitial fluid Healthy C57BL/6NRj mice (male and female, 6-10 weeks, groups were sex and age adjusted) were kept on magnesium-restricted diet or corresponding control diet for 2 weeks. At day 14, liver, spleen and peripheral lymph nodes, muscle (musculus quadriceps femoris) and subcutaneous fat (flank) were aseptically removed. Organs were weighed, 50-300 ul PBS added (adjusted according to weight), and gently centrifuged at 300 x g for 8 min. Organ supernatants were recovered and stored at -80°C. Peritoneal fluid was harvested upon injection if 10 mL sterile PBS into peritoneal cavity; peritoneum was gently massaged and then aseptically opened by incision and lavage was collected. Tumor interstitial fluid of MC38-OVA tumors were collected 1 h after intratumoral injection of either 50 μl 3 mM MgCl2 or 50 μl 3 mM NaCl. Tumors were aseptically excised, weighed and supernatant was recovered as described above. Tissue interstitial fluid of tumors from in vivo CAR T cell experiments were collected from UTD and saline control mice reaching ethically acceptable end point. Tumors were aseptically excised, weighed and supernatant was recovered as described above. Magnesium measurements with ICP-MS Samples of varying volumes (serum 5 μl; Muscle 10 μl; liver, lymph node, peritoneal fluid, subcutaneous fat and tumor 25 μl; spleen 50 μl) were added to 200 μl 67-69% HNO3 (VWR Chemicals; NORMATOM® grade; LOT 45 1119100) and incubated at 95°C for 2 h. Digestates were filled up to 5 mL volume using ultra-pure H2O (<18mΩ; Merckmillipore, 115333) and stored at 4°C. Samples were analyzed using triple quadrupole inductively coupled plasma mass spectrometry (qqq-ICP-MS) on an 8800 system (Agilent, Basel, Switzerland), using general-purpose operational settings. The system was operated in single quad mode using helium as collision gas and quantification done on 24Mg+. To account for matrix effects, 103Rh was used as the internal standard. Organ Mg2+ levels were normalized to tissue weight. Blood count mice Blood samples (70uL) were harvested in EDTA-coated tubes and diluted with 210uL NaCl 0.9%. Samples were run on the ADVIA 2120i hematology analyzer (Siemens) and analyzed by the multi-species program. Polyclonal CD8 T cell activation in vivo with anti-CD3ε F(ab’)2 fragment Healthy C57BL/6NRj and LFA-1 KO mice (male and female, 6-10 weeks, groups were sex and age adjusted) were put on AIN76A magnesium-restricted or control diet for two weeks. On day 14, each mouse received 0.5 µg anti- CD3ε F(ab’)2 fragment (BioXCell) per i.v. tail-vein injection. After 6 h, spleens were harvested splenocyte suspensions were analyzed by flow-cytometry. In vivo killing assay C57BL/6NRj mice (male and female, 6-10 weeks, groups were sex and age adjusted) were immunized against OVA at day -19. In a first experiment, mice were immunized s.c. with 100 μg OVA protein (Invivogen) and 50 μg of CpG-B (Invivogen). In a second experiment, mice were immunized i.v. with 5×103 CFU LmOVA. On day 0, mice were put on magnesium-restricted or control diet for two weeks. On day 14, target splenocytes were harvested from naïve, syngeneic mice and either loaded with OVA257–264 peptide (Eurogentec) or left untreated. Unloaded control splenocytes were labeled brightly with 2.50 µM CTV (CTVbright) and OVA257-264-loaded splenocytes dimly with 0.25 µM CTV (CTVdim). Cells were then counted, mixed at 1:1 ratio in PBS, and 2-4×106 total target splenocytes were injected i.v. into pre-immunized and naive control mice. Additionally, 200 µl of 3 mM MgCl2 and 200 µl of 3 mM NaCl were administered i.p., respectively. After 12 h, spleens were harvested and splenocyte suspensions were analyzed by flow-cytometry. Percentages of specific in vivo killing were calculated as [1-(% CTVdim naive / % CTVbright naive) / (% CTVdim immunized / % CTVbright immunized)]×100. Listeria peritonitis model C57BL/6NRj mice (male and female, 6-10 weeks, groups were sex and age adjusted) were immunized i.p. with 5×104 CFU Listeria monocytogenes expressing the OVA-peptide (Listeria monocytogenes expressing chicken Ovalbumin (AA134–387) originally gifted from Prof. Ed Palmer, University of Basel). After 19 days, mice were put on Mg2+-restricted or matching control diet for 2 weeks. Mice were then re-infected i.p. with 5×105 CFU LmOVA. Bacterial inoculum was either spiked with 3 mM MgCl2 or 3 mM NaCl diluted in 200 μl ddH2O. Mice were sacrificed 12 h post infection and peritoneal fluid was harvested. Peritoneal fluid was harvested upon injection if 10 mL sterile PBS into peritoneal cavity; peritoneum was gently massaged and then aseptically opened by incision and lavage was collected. Peritoneal fluid was then plated on BHI agar-plates and colonies counted upon 24 h of incubation. Remaining peritoneal fluid were centrifuged at 300 x g for 10 min and supernatant recovered. Peritoneal fluid and sera were frozen at -80°C prior to further analysis. For flow cytometric analysis of memory CD8+ T cell compartment, mice were scarified 20 h post infection, and peritoneal 46 fluid was harvested. MACS beads and LS columns (both Milteny Biotec) were used to enrich cell suspension for CD8+ T cells. Cells were stained as described above and analyzed by FACS. Cytometric bead array Cytokine concentrations in cell culture supernatants, peritoneal fluids and serum were determined using the LegendPlex cytrometric bead array (CBA) Th1-Pannel (human and mouse, both from Biolegend) according to manufacturer’s instructions. Perforin quantification Perforin concentrations in peritoneal fluids were determined using ProQuantumMouse Granzyme B Immunoassay Kit (Inivtrogen) according to manufacturer’s instructions. Murine MC38-OVA tumor model – set up Unless stated otherwise, C57BL/6NCrl (female, 6 to 12 weeks old) were used for experiments. In assays with pre-immunized mice, mice were immunized 19 days before tumor implantation by subcutaneous injection of 100 μg of OVA protein (Invivogen) and 50 μg of CpG-B ODN 1826 (Eurogentec), resuspended in 100 μL of PBS. For tumor implantation, mice were inoculated subcutaneously onto the flanks with 0.5×106 MC38-OVA cells, resuspended in 100 μL of PBS. In bilateral tumor experiments, mice received 50 μL intra-tumoral injections of either 3 mM NaCl or 3 mM MgCl2 (both diluted in ddH2O). Injections of NaCl solution was applied in left flank tumor, whereas MgCl2 solution was injected in contralateral tumor. I.t. injections were initiated once tumors were palpable, usually between day 5 and 10 after tumor injection. Injections were repeated every third day. Tumor size was quantified using a caliper and tumor volume was calculated using a rational ellipse formula (α2 × β × π/6, α being the shorter axis and β the longer axis). In all survival experiments, mice were withdrawn from the study after any tumor dimension had reached a length greater than 15 mm. Murine MC38-OVA tumor model – with NSG mice On day 0, 0.25×106 MC38-OVA cells were implanted subcutaneously onto the flank of NSG mice (female mice, 6-8 weeks old). On day 9, 2.5×106 LFA-1-/- or WT OT-I CTLs were transferred by tail-vein injection. CRISPR-Cas9 gene editing was carried out as described above. From day 9 onwards, intratumoral 3 mM NaCl or 3 mM MgCl2 injections were initiated and repeated every second to third day. Murine MC38-OVA tumor model – in vivo CD8+ T cell depletion For CD8 depletion experiment, mice were immunized with OVA, as described above, and inoculated with 0.5×106 MC38-OVA cells unilaterally on the flank. Intratumoral injections of either 3 mM NaCl or 3 mM MgCl2 were initiated, and repeated every third day as tumors became palpable. CD8 T cells were depleted by administering anti-CD8a Ab (53-6.72, BioXCell) at 10mg kg-1 i.p. once per week. Murine MC38-OVA tumor model – in vivo PD-1 blockade For PD-1 blockade experiments, mice were immunized with OVA, as described above, and inoculated with 0.5×106 MC38-OVA cells unilaterally on the flank. As tumors became palpable - at day 5 - intratumoral injections of either 3 mM NaCl or 3 mM MgCl2 were initiated, and repeated every third day for 8 cycles. Mice were additionally injected i.p. with isotype control (IgG2a) or anti-PD-1 Ab on day 9, 12, and 15 post-tumor implantations, at a dose of 200 μg per mouse diluted in 100 μL of pH-matched PBS (according to manufacturer’s 47 recommendations). The antibodies used were: anti-PD-1 IgG2a Ab (clone RMP1-14) or IgG2a isotype control Ab (clone 2A3, both purchased from BioXCell). Murine MC38-OVA tumor model – flow cytometry analysis of tumor-infiltrating immune cells Tumor tissue was isolated from mice, weighed and minced using razor blades. Tissue was then digested using accutase (PAA), collagenase IV (Worthington), hyaluronidase (Sigma), and DNAse type IV (Sigma) for 60 min at 37 °C with constant shaking. The cell suspensions were filtered using a cell strainer (70 μm). Precision Counting beads (Biolegend) were added before staining to quantify the number of cells per gram of tumor. Single cell suspensions were blocked with rat anti-mouse FcγIII/II receptor (CD16/CD32) blocking antibodies ('Fc-block') and stained with live/dead cell-exclusion dye. Cells were then incubated with fluorophore-conjugated antibodies directed against cell surface antigens, washed and resuspended in FACS buffer (PBS+2% FBS). For intracellular/intranuclear antigens, cells stained with cell surface antibodies were fixed and permeabilized using Foxp3/transcription factor staining buffer (eBioscience) prior to incubation with antibodies directed against intracellular antigens. Murine MC38-OVA tumor model – liposome and immunoliposome application Liposomes were prepared using the thin film-hydration method (Huwyler et al., 1997). The lipid composition consisted of DSPC (1,2-distearoyl-sn-glycero-3-phosphocholine, Avanti Polar Lipids, Alabaster, AL), Cholesterol (3β-Hydroxy-5-cholestene, 5-Cholesten-3β-ol, Sigma Aldrich, Schaffhausen, Switzerland), DSPE-PEG2000 (1,2- distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000], Avanti Polar Lipids, Alabaster, AL), DSPE-PEG2000-Mal (1,2-distearoyl-sn-glycero-3-phosphoethanolamine-N- [maleimide(polyethyleneglycol)-2000], Avanti Polar Lipids, Alabaster, AL), and DiI (DiIC18(3), Sigma Aldrich, Schaffhausen, Switzerland) at a molar ratio of 69:27:2.5:0.5:1. Lipids were mixed in ethanol and a homogenous thin film was prepared using a Rotavapor A-134 (Büchi, Switzerland). Dried lipid films were stored under vacuum overnight and rehydrated with either 150 mM NaCl or 100 mM MgCl2 at 70°C with constant stirring. Liposomes were extruded 13x through polycarbonate filters with an average pore diameter of 100 nm (Nucleopore, Whatman, North Bend, OH). Amicon Ultra-4 centrifugal filter units (MWCO 10 kDa) were used to concentrate the liposomes and to exchange the buffer to PBS pH 7.2. The liposomes were used within 4 h for anti-PD-1 conjugation via SATA-maleimide conjugation chemistry. Briefly, anti-PD-1 antibodies (clone RMP1-14, BioXCell) were functionalized with 10x molar excess of SATA (N-succinimidyl-S-acetylthioacetate, Fisher Scientific, Reinach, Switzerland) to introduce sulfhydryl groups according to the supplier’s recommendations. The reactive sulfhydryl group on the antibody was then conjugated to the maleimide moieties on the liposome for 2 h at RT (ratio SH: maleimide = 1:4). Free maleimide groups were blocked by incubating with a 10x molar excess L- Cysteine for 30 min at RT. The anti-PD-1-conjugated liposomes were purified by gel filtration chromatography (Sepharose CL4B, elution buffer PBS pH 7.2). Fractions containing liposomes were pooled and concentrated using Amicon Ultra-4 centrifugal filter units (MWCO 10 kDa, Merck Millipore Ltd., Tullagreen, Carrigtwhohill, Co Cork IRL). Average size, size distribution and particle concentration were determined by dynamic light scattering using a Zetasizer Ultra (Malvern Panalytical, Volketswil, Switzerland). For intratumoral liposome injection, 4×1012 of MgCl2-loaded liposomes and 2.6×1012 of NaCl-loaded liposomes were used. For immunoliposome 48 experiments, 100 μg of anti-PD-1 Ab coupled to MgCl2- or NaCl-loaded liposomes were injected i.p. at indicated time points. Anti-nuclear antibody (ANA) quantification ANA concentrations were determined in murine serum 25 days after the start of i.t. NaCl or MgCl2 injection as well in age- and sex-matched untreated control mice, using the mouse anti-nuclear antibody ELISA kit (Hoelzl Biotech) according to manufacturer’s instructions. Co-culture assay with Blinatumomab Blinatumomab (Amgen) was derived from the leftover of infusions. Human PHA-blasts were incubated with Ramos or LCL target cells in flat bottom 96 well-plate at a 0.5:1 ratio (6.5×104 PHA blasts and 1.3×105 target cells) at indicated Blinatumomab concentrations. In order to distinguish the different cell populations, PHA-blasts were labeled with CTV and target cells with CFTR Invitrogen. Anti-CD18 antibodies, CBR-LFA-1/2 (unlabeled, Ultra-LEAF, BioLegend), TS1/18 (unlabeled, Ultra-LEAF, BioLegend) as well Mouse IgG1 κ Isotype control (unlabeled, Ultra-LEAF, BioLegend) were used at a final concentration of 10 μg mL-1. For all co-incubation experiments, cells were allowed to sediment without centrifugation. For analysis of protein phosphorylation, co-incubation was terminated after 60 min with fixation of of 4% PFA and methanol as described in more detail above. For assessing LFA-1 headpiece opening, m24 antibody was directly added to the cell culture medium and stained for 30 min on ice before washing and subsequent fixation with 2% PFA. Cytotoxicity was quantified after 3.5 h with CellEvent Caspase-3/7 Green Detection Reagent (Invitrogen, ThermoFisher) as described above. Caspase substrate was added for final 45 min of incubation at final concentration of 2 μM. Cells were harvested, washed in FACS Buffer and fixed with PFA 2% for 15 min at RT prior to analysis by FACS. Production of anti-CD19 CAR T cells for in vitro experiments 24 h before transfection, HEK-293T cells were seeded (3.8×106 cells 10 mL-1 media). All plasmid DNA was purified using the Endotoxin-free Plasmid Maxiprep Kit (Sigma). HEK-293T cells were transfected with 1.3 pmol psPAX2 (lentiviral packaging plasmid) and 0.72 pmol pMD2G (VSV-G envelope expressing plasmid) and 1.64 pmol of pCAR-CD19CAR-p2a-EGFP (Creative Biogene) using Lipofectamine 2000 (Invitrogen) and Optimem medium (Invitrogen, Life Technologies). The viral supernatant was collected 48 h after transduction. Viral particles were concentrated using PEG precipitation and stored at -80°C. Blood samples (Blood donor center, University Hospital Basel) were obtained from healthy donors after written informed consent. PBMCs were isolated by standard density-gradient centrifugation protocols (Lymphoprep; Fresenius Kabi). CD4+ and CD8+ T cells were positively selected using magnetic CD4+ and CD8+ beads (Miltenyi Biotec). Purified CD4+ and CD8+ T cells were cultured in R10AB. CD4+ and CD8+ T cells were plated into a 24-well cell culture plate and stimulated with anti- CD3 and anti-CD28 mAb-coated beads (Miltenyj, T cell activation & expansion kit) in a ratio of 1:1 in medium containing IL-2 (150 U mL-1). T cells were transduced with lentiviral particles at 18-22 h after activation in media containing Polybrene (6 μg mL-1, Millipore). Every second day medium was replaced with fresh IL-2 (150 U mL- 1). Five days after transduction GFP+ cells were sorted enrich CD19-CAR+ cells and magnetic beads were removed from non-transduced cells. Cells were further expanded for 3 days in medium containing IL-2 (150 U mL-1) before the target cell killing assay. Cytotoxicity assay with anti-CD19 CAR T cells 49 CD8+ anti-CD19 CAR T-cells were incubated with Ramos target cells at a 0.1-0.33:1 ratio (0.5-1.5×104 CAR T cells and 5×104 Ramos Target cells). Ramos cells had been labeled with CFTR prior to co-incubation. Cells were allowed to sediment without centrifugation in flat bottom 96 well-plate and incubated for 3 h. Cytotoxicity was quantified by flow cytometry using BioTracker NucView 405 Blue Caspase-3 Dye (Sigma-Aldrich). Primary human T cell transduction for anti-CD19 CAR T cell generation used in in vivo experiments To generate human CD8+ CAR-T cells, we isolated CD8+ T cells from heathy donor PBMCs using the CD8 human microbead MACS kit according to manufacturer’s instruction. Cells were then cultured in RPMI-1640 (Sigma) with 10% heat inactivated human male AB+ serum with 1mM Sodium Pyruvate (Sigma), 2mM Glutamine (contained in RPMI formulation), 10 mM HEPES (Gibco), 5mM beta-mercapto-Ethanol (Gibco), 1% PenicilinStreptomycin (Sigma). Cells were stimulated on the same day with 1:1 ratio of CD3/CD28 beads (Human T cell Activation and Expansion kit, Miltenyi) and 150 U mL-1 rh-IL-2 (Proleukin). The next day, cells were collected into a falcon tube and 4 µg mL-1 Polybrene (Sigma) was added together with VSV-g pseudotyped lentivirus encoding an anti-human-CD19-FMC63vH chimeric antigen receptor with a CD28 transmembrane domain and a CD28 and CD3z signaling domain with a c-terminal T2A self-cleaving copGFP protein (anti-CD19-CD28z-T2A- copGFP). The cell and lentiviral mixture was centrifuged for 90 min at 1000 g (spinfection) and the resuspended and plated for 24 h at 37°C. Then cells were expanded 1:2 every 2 days for 2 iterations with fresh medium and 50 U/mL rh-IL-2. GFP+ CD8+ (CD8-APC SK1 clone, Biolegend) live (DAPI, Sigma) cells were then sorted using a BD FACS Melody, washed in medium and the plated at 1.5 Mio mL-1 in fresh medium and 50 U mL-1 rh-IL-2 for another 24h. Cells were then counted, washed by centrifugation at 500g 3 min in PBS and transferred in PBS intravenously to female NSG mice (6-8 weeks old) subcutaneously injected 5 days before with 0.5 Mio Raji (ATCC) in 50% Matrigel (Corning, standard formulation). Intratumoral injections of either 3 mM NaCl or 3 mM MgCl2 were initiated 7 days after tumor implantation, and repeated every second to third day. Recombinant lentivirus production for anti-PSMA CAR T cells High-titer replication-defective lentivirus was produced and concentrated by ultracentrifugation for primary T cell transduction. Briefly, 24 h before transfection, HEK-293 cells were seeded at 10 × 106 in 30 mL of medium in a T-150 tissue culture flask. All plasmid DNA was purified using the Endo-free Maxiprep kit (Invitrogen, Life Technologies). HEK-293T cells were transfected with 7 μg pVSV-G (VSV glycoprotein expression plasmid), 18 μg of R874 (Rev and Gag/Pol expression plasmid) and 15 μg of pELNS transgene plasmid, using a mix of Turbofect (ThermoFisher) and Optimem medium (Invitrogen, Life Technologies, 180 μl of Turbofect for 3 mL of Optimem). The viral supernatant was collected 48 h after transfection. Viral particles were concentrated by ultracentrifugation for 2 h at 24,000 x g and resuspended in 400 μl medium, followed by immediate snap freezing on dry ice. Primary human T cell transduction for anti-PSMA CAR T cell generation Primary human T cells were isolated from the peripheral blood mononuclear cells of healthy donors (HDs; prepared as buffycoats or apheresis filters). All blood samples were collected with informed consent of the healthy donors, and genetically engineered with ethics approval from the Canton of Vaud, Switzerland. PBMC were obtained via Lymphoprep (Axonlab) separation solution, using a standard protocol of centrifugation. CD4 and CD8+ T cells were isolated using a magnetic bead-based negative selection kit following the manufacturer’s 50 recommendations (easySEP, Stem Cell technology). Purified CD4 and CD8+ T cells were cultured at a 1:1 ratio and stimulated with anti-CD3 and anti-CD28 Ab coated beads (Invitrogen, Life Technologies) at a ratio of 1:2 T cells to beads. T cells were transduced with lentivirus particles at 18-22 h after activation. Human recombinant IL-2 (h-IL-2; Glaxo) was replenished every other day for a concentration of 50 IU mL-1 until 5 days after stimulation (day + 5). At day + 5, magnetic beads were removed, and h-IL-7 and h-IL-15 (Miltenyi Biotec) were added to the cultures at 10 ng mL-1 replacing h-IL-2. A cell density of 0.5-1 × 106 cells mL-1 was maintained for expansion. Rested engineered T cells were adjusted for equivalent transgene expression before all functional assays. Cytotoxicity assay with anti-PSMA CAR T cells Cytotoxicity assays were performed using the IncuCyte Instrument (Essen Bioscience). Briefly, 1.25×104 PC3-PIP target cells were seeded in flat bottom 96-well plates (Costar, Vitaris). Four hours later, rested T cells (no cytokine addition for 48 h) were washed and seeded at 2.5 × 104 per well, at a 2:1 effector to target ratio in self- made medium supplemented with 10% dFCS and ±0.6 mM MgCl2. No exogenous cytokines were added during the co-culture period. IncuCyte Caspase-3/7 (Essen Bioscience) was added at a final concentration of 5 μM in a total volume of 200 μl. Internal experimental negative controls were included in all assays, including co- incubation of untransduced (UTD)-T cells and tumor cells in the presence of IncuCyte Caspase-3/7 reagent to monitor spontaneous cell death over time. As a positive control, tumor cells alone were treated with 1% triton solution to evaluate maximal killing in the assay. Images of total green area per well were collected every 2 h of the co-culture. The total green area per well was obtained by using the same analysis protocol on the IncuCyte ZOOM software provided by Essen Bioscience. Cytotoxicity is reported as total area under the curve of the fluorescence driven by incorporation of cytotoxic green reagent in dead target cells (green area per μm2). All data were normalized by subtracting the background fluorescence observed at time zero (before any cell killing by CAR T cells) from all further time points. Cytokine release assay of anti-PSMA CAR T cells Cytokine release assays were performed by co-culture of 5×104 T cells with 5×104 target cells per well in 96-well round-bottom plates, in duplicate, in a final volume of 200 μl of self-made medium supplemented with 10% dFCS and 0.6 mM MgCl2 or without Mg2+ supplementation. After 24 h, the co-culture supernatants were collected and tested for the presence of IFN-γ by commercial enzyme-linked immunosorbent assay kits according to the manufacturer’s protocol (BioLegend). Anti-PSMA CAR T cell in vivo experiment Male NSG mice of 10-12 weeks were put on Mg2+-restricted or matching control diet 5 days prior to tumor injection and kept on respective diet throughout the experiment. 5×106 PC3-PIP tumor cells were injected subcutaneously. After 5 days, intravenous injection of saline solution or 2×106 T cells (UTD or CAR T cells) were adoptively transferred intravenously. Tumor volume was monitored twice per week. The animals were monitored daily and the tumors were calipered every other day. Tumor volumes were calculated using the formula V=1/2(length × width2), where length is the greatest longitudinal diameter and width is the greatest transverse diameter determined via caliper measurement. Quantification and statistical analysis 51 Statistical significance was tested for using Prism 9.0 (GraphPad Software, USA). P values of less than 0.05 were considered statistically significant. Graphical Illustrations All graphical illustrations were created with BioRender.com. 52 SUPPLEMENTARY FIGURES 53 A EM CD8: Glycolytic activity ± Mg 2+ B Naive CD8: Glycolytic activity ± Mg 2+ 20 1.2 mM 20 ** 20 1.2 mM 20 0 mM ** 0 mM0 1.2 mM 15 15 15 0 1.2 mM 15 n.s n.s 10 10 10 10 5 5 5 5 0 0 0 0 0 50 100 150 200 mM Mg2+ 0 50 100 150 200 2 0 2 mM Mg2+ Time (minutes) 1.2 0 1.2 Time (minutes) 1. 1. 0 0 CD3/28 CD3/28 C D EM CD8 T cells: EM CD8: Mg2+ titration and surface activation markers Surface activation markers Cytokine release Viability 1.2 mM 100 **** **** **** **** **** 104 ** ** ** 0 mM unstim. 1.0 100 80 1.2 mM 1.2 mM0 mM 0 mM 103 0 mM 75 0.012 mM CD69 60 0.06 mM 2 0.5 CD71 10 50 0.12 mM CD98 40 0.6 mM CD107a 101 25 1.2 mM CD25 20 2.4 mM 0.0 0 100 0 CD69 - PerCp .4 .2 .6 12 06 12 0 im . CD69 CD25 CD107a CD71 CD98 IFN TNF IL-2 unstim. CD3/28 2 1 0 0. 0. 0.0 stun CD3/28 CD3/28 CD3/28 stim. - [mM Mg2+] E Naive CD8 T cells: Surface activation markers Cytokine release Viability 100 104 1.2 mM 1.2 mM 100 1.2 mM 0 mM 80 0 mM 0 mM103 75 60 102 50 40 10120 25 0 100 0 CD69 CD25 CD107a CD71 CD98 IFN TNF IL-2 unstim. ActivationBeads Activation beads Activation beads F G H PHA-Blast: CD3/28 activation-induced clustering CD8 T cells: Magnesium Flux CD8 T cells: Calcium Flux LFA-1: open headpiece 1.5 MgCl2 1.5 CaCl2 1.6 **** 1.4 1.0 1.0 1.2 naive CD8 T cells 0.5 naive CD8 T cells 1.0 EM CD8 T cells EM CD8 T cells 0.8 0.5 0.0 0 100 200 0 100 200 EM Naive 1.2 mM 0 mM Time (sec) Time (sec) CD3/28 I PHA-Blast: Glycolytic stress test J PHA-Blast: Mitochondrial stress test 80 30 60 Glucose Oligomycin 2-DG 1001.2 mM Oligomycin FCCP Rotenone 1.2 mM 0mM 0mM 60 800 mM 0 mM 20 1.2 mM+ 0.05 mM Mn2+ 40 2+0 mM + 0.05 mM Mn60 40 0 + Mn2+ 40 10 20 20 1.2 mM 0 mM 20 0 + Mn2+ 0 0 0 0 0 20 40 60 80 Glycolysis Glycolytic 0 20 40 60 80 basal ATP- leak maximal SRC non- Time (minutes) reserve Time (minutes) coupled mitochondrial M N K L PHA-Blast CD3/28 activation PHA-Blast CD3/28 activation PHA-Blast unstimulated: ICAM expression ICAM-3 expression unstimulated with ICAM blockade: with ICAM blockade: LFA-1 headpiece opening CD69 upregulation 40 ****15 **** **** ** n.s. 10.0 **** n.s. * *** **** **** Isotype **** 20 0 ICAM-1 10 naive CD8 7.5 0 ICAM-2 EM CD8 5.0 -50 ICAM-3 5 PHA-Blast -20 0 10 3 10 4 10 5 10 6 10 3 10 4 10 5 10 6 10 7 2.5 ICAMs - PE ICAM3 - PE -40 0 0.0 -100 1 2 3 naive EM PHA 1 2 3 e -1 -2 -3 e AM - AM - AM - - CD8 CD8 TCB AM AM - AM - p oty AM AM AM t yp IC IC IC IC IC C Is C C C Is o I I I I 1.2 mM 0 mM 1.2 mM 0 mM Supplementary Figure 1 54 ECAR (mpH min1) CD3/28 stim. - [Mg2+] gMFI ICAM-3 (x10^4 ) ECAR (mpH min-1) % positive T cells of resp. marker norm. to 1.2 mM ECAR (mpH min-1) ± M % positive g 2C +D3/28 MFI MagFluo4 Normalized to T0 ECAR (mpH min-1) gMFI ICAM-3 (x10^5) OCR (pmol min-1) % positive pg ml -1 ECAR (mpH min-1) gMFI m24 %-change -1 to 1.2 mM Isotype Ctrl. MFI Fluo3 FuraRed ± Mg 2 C +D3/28 OCR (pmol min-1) % Zombie Aqua neg. pg ml -1 gMFI CD69 %-change ECAR (mpH min-1) to 1.2 mM Isotype Ctrl. gMFI(m24) norm. to unstim % Zombie Aqua neg. Figure S1. Extracellular Mg2+ promotes activation of memory T cells, related to Figure 1 (A and B) Glycolytic activity of human CD8+ T cell subsets upon injection of anti-CD3 and anti-CD28 antibody in medium containing 1.2 mM Mg2+, 0 mM Mg2+, or medium which was reconstituted from 0 mM to 1.2 mM Mg2+ just prior to activation (0 / 1.2 mM Mg2+). (A) Results for human effector memory (EM), and (B) for human naive CD8+ T cells. Traces of a representative metabolic flux experiment (left panels), and pooled data of n = 5 independent experiments (right panels). (C) Flow cytometric analysis of the indicated surface activation markers on human EM CD8+ T cells activated at indicated Mg2+ concentrations for 24 h. Representative flow histogram of CD69 expression (left panel), and pooled data from two healthy donors (right panel). (D) Surface activation markers on EM CD8+ T cells after activation for 24 h. Pooled data from n = 5 independent experiments (left panel). Abundance of inflammatory cytokines in corresponding cell culture supernatants. The dashed line indicates detection limit (middle panel). Viability of EM CD8+ T cells (right panel). (E) Surface marker expression on activated naive CD8+ T cells after 24 h. Pooled data from n = 4 independent experiments (left panel). Abundance of inflammatory cytokines in corresponding cell culture supernatants. The dashed line indicates detection limit (middle panel). Viability of naive CD8+ T cells (right panel). (F) Representative brightfield images of activation-induced clustering of PHA blasts after 6 h. Scale bars indicate 50 mm. (G) Flow cytometric analysis of Mg2+ influx (left panel) and Ca2+ influx (right panel) into naive and EM CD8+ T cells, using fluorescent probes. (H) Flow cytometry-based assessment of TCR activation-induced LFA-1 headpiece opening (as detected by mAb m24 binding) on human EM and naive CD8+ T cells. Pooled data of n = 4 independent experiments using cells from 1 to 2 individual healthy donors (I) Assessment of glycolysis and glycolytic reserve of human PHA blasts. One representative experiment (left panel) and pooled data from two independent experiments (right panel) are shown. (J) Mitochondrial perturbation assay of human PHA blasts. Left panel depicting a representative experiment, summary bar graph in the right panel, representing calculated basal respiration, ATP-coupled respiration, leak respiration, maximal respiration, and spare respiratory capacity (SRC), as well nonmitochondrial respiration (pooled from two independent experiments). (K) Representative ICAM1-3 expression profile on PHA blasts as determined by flow cytometry (left panel) and summary bar graph (right panel). (L) Representative ICAM-3 expression profile on human naive and EM CD8+ T cells as well human PHA blasts, as determined by flow cytometry (left panel) and summary bar graph (right panel). (M and N) Flow cytometric assessment of LFA-1 headpiece opening (M), and expression of CD69 (N) on PHA blasts activated in presence of ICAM blocking mAb, as indicated (not normalized for ICAM expression). Each symbol represents an individual healthy donor. Data are presented as mean ± SD in (H, M, and N), in the right panel of (A–C and I–L) and the left as well as right panels of (D and E) or median ± IQR middle panel of (D and E). In representative metabolic flux analyses symbols indicate mean ± SEM in the left panel of (A, B, I, and J). Traces of Mg2+ or Ca2+ flux experiments indicate mean of respective fluorescent intensity (G). Statistical significance was assessed by repeated-measures one-way ANOVA with Sidak’s multiple comparison test right panel of (A, B, and I–L) as well as (M and N), unpaired two-tailed Student’s t test with Holm-Sidak corrected multiple comparison test left panel and right panel of (D and E) and (H), and Mann-Whitney test middle panel of (D and E). * p < 0.05, ** p <0.01,***p < 0.001, **** p < 0.0001, n.s. = nonsignificant. 55 A Fold increase cytotoxicity Co-culture: cytotoxicity 50 2.5 ** *** **** **** 40 2.0 30 20 1.5 10 0 -6 -7 -8 -9 9c peptide [M] - 10^-6 10^-7 10^-8 10^-9 9c peptide [M] 0^1 10 ^ 10 ^ 10 ^ B Co-culture: phospho-ERK1/2 C Co-culture: phospho-c-Jun 20 ** 25 **** * **** ** **** 15 20 1.2 mM 1.2 mM 15 10 0 mM 0 mM 10 0 + Mn2+ 5 0 + Mn2+ 5 1.2 mM 0 1.2 mM 0 + BIRT377 4 50 10 10 + 7 + BIRT377 3 4 5 6 10 10 10 10 mM mM 2+ 7 p-ERK1/2 - AF647 n 2 37 m M mM n 37 1.2 0 + M IR T p-c-Jun - AF488 .2 0 + M RT M 1 B B I m +M 0 m M + 0 m 2 m M 1.2 1. D PHA-Blast CD3/28 activation: phospho-ERK1/2 E PHA-Blast CD3/28 activation: phospho-c-Jun 2+ * 2+ 2.0 * M + M n M 7 * n M m 37T 2.5m mM + M M 7 M m M .2 R M M M .2 m 7 6 1 I 6 m IRT 3 ** ** 0 m 0.0 1.2 0 m + B 0 m 0.0 1.2 m 1 1.5 0 + B p-ERK1/2 1.5 p-c-Jun 1.0 ERK1/2 c-Jun 0.5 0.5 Actin Actin 0.0 mM M 2+ 7 2+ 7 0 6 m 2 m M n 37 mM mM mM n 37 .0 1. + M RT I 0 .06 1.2 + M T 0 mM + B 0 R mM + B I 0 M 0 M 1.2 m 1.2 m F EM CD8: Time course CD3/28 activation ± Mg2+ 0’ 5’ 15’ 45’ 0h 6h 24h 0 mM Mg2+ 2+ p-ERK1/2 p-ERK1/2 1.2 mM Mg p-c-Jun Actin Actin Supplementary Figure 2 56 Fold change to 0 mM p-ERK1/2 / ERK1/2 % T cells p-ERK1/2 pos. % Caspase-3 pos.T2 cells % T cells p-c-Jun pos. p-c-Jun / c-Jun Figure S2. The Mg2+-LFA-1 axis regulates cytotoxicity and phosphorylation of ERK1/2 and c-Jun, related to Figure 2 (A) Caspase-3 activity in T2 target cells after co-incubation with NY-ESO-1-specific REP T cells. Left panel depicts fold increase of caspase-3 positive T2 cells in 1.2 mM Mg2+, the right panel illustrates the corresponding percentage of caspase-3 positive T2 cells. (B and C) Representative histograms (left panels), quantified results of n = 1 healthy donor with 4 technical replicates (right panels). Phosphorylation of ERK1/2 (B), and c-Jun (C) in human REP T cells after incubating with cognate-peptide pulsed T2 target cells (D and E) Immunoblot analyses of ERK1/2 phosphorylation (D) and c-Jun phosphorylation (E) in activated human PHA blasts. Left panel depicting immunoblots that were probed for the indicated phosphorylated and total proteins, as well as actin. Right panel depicting summarized quantitation with 3–5 healthy donors. (F) Representative immunoblots of primary human EM CD8+ T cells. Immunoblots were probed for phosphorylated ERK1/2 and c-Jun, as well actin. Each symbol represents an individual healthy donor. Bars indicate mean ± SD (A–E), and statistical significance was assessed with unpaired two-tailed Student’s t test with Holm-Sidak corrected multiple comparison test right panel of (A), one-way ANOVA with Sidak’s multiple comparison test right panel of (B–E). *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001. 57 A Murine CTLs: Jurkat T cells: Murine OT-I CTLs: LFA-1-/ - validation LFA-1-/ - validation LFA-1-/ - validation unstained naive OT-I unstained Ctrl Guide - pre sort Isotype Ctr. Ctrl Guide - post sort WT Ctr. Guide - WT KO Guide - pre sort LFA-1-/- KO Guide - LFA-1-/- KO Guide - post sort 2 3 4 5 6 10 10 10 10 10 2 3 4 510 10 10 10 2 3 4 5 610 10 10 10 10 CD11a - PB450 CD11a - FITC CD11a - Pe/Cy7 B Murine CTLs: Nutrient uptake C Jurkat T cells: Calcium Flux WT LFA-1-/- 8 **** n.s. 2.0 1500CD3 1.2 mM WT ** **** **** 0 mM WT *** 6 **** * 1.2 mM LFA-1-/- ** 1.5 0 mM LFA-1-/- 1250 n.s. 1.2 mM 0 mM 4 0 + Mn2+ 1.0 1000 1.2 mM 2 10 3 10 4 10 5 3 4 5 + BIRT377 10 10 10 2-NBDG 0 0.5 750400 800 1200 2+ WT LFA-1-/- 1.2 0 1.2 0 mM MgTime (sec) CD3/28 WT LFA-1 -/- -p- e dinD npa rosi n i llo I Merge Ty h a APP D Jurkat T cells: Pan phospho-Tyrosine WT 5 *1.2 mM * n.s. 4 WT 0 mM 3 2 LFA-1-/- 1.2 mM 1 0 LFA-1-/- 1.2 0 1.2 0 mM Mg2+ 0 mM WT LFA-1-/- CD3-coated plate E -p-n ine din pa i os oall I Merge rTy Ph DA P unstim. 1.2 mM PHA-blast: Pan phospho-Tyrosine unstim. 3 ** **** ** 0 mM 2 CD3 1.2 mM 1 CD3 0 mM 0 1.2 0 1.2 0 0 1.2 mM Mg2+ + Mn2+ CD3 - - + + + + CD3-coated plate 0 + Mn2+ - - - - - + BIRT377 CD3 1.2 + BIRT377 F Murine CTLs: phospho-ERK1/2 Murine CTLs: phospho-ERK1/2 10 1.2 mM 100 0 mM8 0 mM + Mn2+ 1.2mM + BIRT377 6 80 4 2 60 0 WT LFA-1-/- WT LFA-1-/- unstimulated PMA/Ionomycin Supplementary Figure 3 58 % phospho-ERK1/2 pos. MFI - 2-NBDG (x10^3) % phospho-ERK1/2 pos. Pan phospho-Tyrosine Pan phospho-Tyrosine signal itensity (x10^3) signal itensity (x10^3) MFI (FLuo4 F -1t 0 ) AUC (FLuo4 F -1t 0 ) Figure S3. Extracellular Mg2+ regulates Ca2+ flux and immune synapse formation through LFA-1, related to Figure 3 (A) Representative histogram comparing CD11a surface expression on CTLs derived from murine WT and LFA-1-/- CTLs (left panel), WT (treated with negative control guide RNA) and LFA-1-/- Jurkat T cells (middle panel), as well naive OT-I cells and CRISPR-Cas9-edited OT-I CTLs before and after purification by cell sorting (right panel). (B) Glucose uptake by activated murine WT and LFA-1-/- CTLs. Representative flow histogram profile (left panel) and summary graph from n = 3 mice with three technical replicates each (right panel). (C) TCR stimulation induced calcium flux in WT and LFA-1-/- Jurkat T cells. Representative trace (left panel), and quantification of area under the curve (AUC) of three independent experiments in duplicates (right panel). (D) Left panel, representative immunofluorescence images of WT and LFA-1-/- Jurkat T cells stimulated on anti-CD3 Ab- coated coverslip. Images are displayed as confocal projections of 3D stacks. Scale bars indicate 10 mm. Right panel, quantitative analysis of fluorescent signal intensity (n = 62–71 cells/condition, pooled from two different fields of view). (E) Left panel, representative immunofluorescent images of PHA blasts stimulated on ±anti-CD3-coated coverslip. Cells were stained for pan-tyrosine phosphorylation, actin, and nuclei. Images are displayed as confocal projections of 3D stacks. Scale bars indicate 10 mm. Right panel, quantitative analysis of fluorescent signal intensity (n = 91–146 cells/condition, pooled from 2 to 4 different field of views). (F) Flow cytometric analysis of ERK1/2 phosphorylation in murine WT and LFA-1-/- CTLs. Left panel shows unstimulated baseline ERK1/2 phosphorylation, and right panel depicts phosphorylation status upon PMA and ionomycin treatment. Quantified results of n = 3 mice in duplicates each. Data are presented as mean ± SD right panel of (B and C) and both panels (F), median ± IQR right panel of (D and E). Traces of representative calcium flux, left panel of (C), indicate mean of duplicate wells. Statistical significance was assessed by one-way ANOVA with Sidak’s multiple comparison test right panel of (B and C) and both panels (F), Kruskal-Wallis test with Dunn’s multiple comparisons test was applied for (D and E). *p < 0.05, **p <0.01, ***p <0.001, ****p < 0.0001, n.s., nonsignificant. 59 A B Body weight 120 110 100 90 80 rolnt lo w tro l. ow Co l g2 + on + M C Mg 2 Day 7 Day 14 C Complete blood count Erythrocyte parameters Differential white blood cell count 500 350 100 Ctrl diet Ctrl diet Ctrl diet 400 Mg2+ low diet 300 Mg2+ low diet 80 Mg2+ low diet 300 250 200 60 60 40 40 6 20 20 4 150.2 10 2 0.1 5 0 0.0 ) 0WBC PLT RBC L- 1 ) -1 ) fL) g) - 1 L ( p L te hil l l ( te h i hi C (g (L VB T C CH C (g y y ho c ptro cno ino p so p LU HG u s a HC M M MC H mp e Mo o B Ly N E D E Steady-state CD69 expression F Steady-state CD69 expression CD8 T cell distribution CD8 Phenotype: Spleen on CD8 T cells in Spleen CD8 Phenotype: Peritoneum on CD8 T cells in Peritoneum 20 100 1.5 100 1.5 Ctrl diet Ctrl diet Ctrl diet Ctrl diet Ctrl diet Mg2+ low diet 80 Mg2+ low diet Mg2+ low diet 80 Mg2+low diet Mg2+ low diet 15 1.0 1.0 60 60 10 40 40 0.5 0.5 5 20 20 0 0 0.0 0 0.0 Spleen Peritoneum Naive EM CM Naive EM CM Naive EM CM Naive EM CM G Spleen: Immune cell subsets Peritoneum: Immune cell subsets 60 80 Ctrl diet 60 Ctrl diet50 Mg2+ low diet Mg2+ low diet 40 40 20 30 15 6 10 4 5 2 0 0 lls .1 Cs e s g te s hil es lls /1 p t e 1. 1 Cs es ph il tes B ce CR K D ha c y N ro cy c C R K D ag o y T rop on o ute lo B T N ph utr loc ac M N ran u cro Ne u M a ra n G M G H Mg2+ concentration upon i.t. application Tumor Lymph node Serum 400 ** 5000 6 300 4000 4 3000 200 2000 2 100 1000 0 0 0 l aC Cl 2 C l l 2 Cl l 2 N a CMg N Mg N a C Mg Supplementary Figure 4 60 ppb[Mg] μg[Tissue]-1 % pos. of live singlets % pos. of alive single cells x10 9 cells L-1 x1012 cells (RBC) L-1 ppb[Mg] μg[Tissue]-1 % of CD8 pos. mg dL-1 % pos. of live singlets gMFI CD69 (x10^4) % weight from baseline percentage (%) % of CD8 pos. gMFI CD69 (x10^4) Figure S4. Immune phenotype of short-term organismal Mg2+ depletion in uninfected mice, related to Figure 4 (A) Schematic of experimental design. (B–G) Heathy Bl/6 mice were placed on Mg2+ low and matching control diet for 2 weeks without further intervention. (B) Percentage weight change compared with individual baseline value after 7 and 14 days. Pooled data from n = 5 independent experiments. (C) Left panel, complete blood count representing absolute numbers of white blood cells (WBC), platelets (PLT), and red blood cells (RBCs). Middle panel, quantitation of red blood cell parameters: Hemoglobin (HGB), hematocrit (HCT), mean cell volume (MCV), mean cell hemoglobin (MCH), mean cell hemoglobin concentration (MCHC). Right panel, white blood cell subsets in peripheral blood are shown as percentage distribution. Pooled data from n = 2 independent experiments. (D) Abundance of CD8+ T cells in spleen and peritoneal cavity. (E and F) Flow cytometric characterization of CD8+ T cell subsets: naive, effector memory (EM) and central-memory (CM). Composition of CD8+ T cell subsets in spleen (E, left panel) and peritoneal cavity (F, left panel). Quantification of CD69 expression on CD8+ T cell subsets in spleen (E, right panel) and peritoneal cavity (F, right panel). (G) Analysis of relative abundance of immune cell subpopulations in spleen (left panel) and peritoneal cavity (right panel). (H) Magnesium levels in interstitial fluid of tumor (left panel), draining lymph node (middle panel) and serum (right panel) after intratumoral application of NaCl or MgCl2. Each symbol represents an individual mouse. Data is presented as mean ± SD (B–G) and median ± 95% Cl (H). Statistical significance was assessed with unpaired two-tailed Student’s t test (H). ** p < 0.01. 61 A LmOVA: LmOVA: B Peritonitis: C D Viability CFU Serum IL-6 Peritonitis: Tetrame r + Peritonitis: Tetramer+ 10000 600 4 2.5**** **** 0.066 0.24 0.51 100 1000 3 2.0 400 1.5 100 2 90 1.0 200 10 1 0.5 80 1 0 0 0.0 NaCl MgCl2 NaCl MgCl2 Ctrl Low Low Ctrl Low Low Ctrl Low Low Mg2+ Diet − − + − − + − − + Mg2+-spiked i.p. injection Supplementary Figure 5 62 % Bacteria Syto9 pos. CFU (x 10^6) pg ml-1 % pos. Tetramer+ CD + 44 of CD8 T cells Tetramer+ CD44+ CD8 T cells (x103) Figure S5. Dietary restriction of Mg2+ impairs memory CD8+ T cell-mediated control of bacterial infection, related to Figure 5 (A) Viability of LmOVA in 3 mM NaCl or 3 mM MgCl2 inoculum solution used for in vivo infection studies (left panel), and number of colony forming units (CFUs) after plating the respective bacterial solution on BHI plates (right panel). (B–D) LmOVA infection experiments. Each symbol represents an individual mouse. (B) Bar graph representing serum IL-6 concentrations. Relative abundance (C) and absolute cell numbers (D) of OVA-tetramer specific memory CD8+ T cells retrieved from the peritoneal cavity. Data are presented as mean ± SD left panel of (A). Median ± IQR is shown in right panel of (A) and (B–D). Statistical significance was assessed with unpaired two-tailed Student’s t test left panel of (A), Mann- Whitney test right panel of (A) and one-way ANOVA with Sidak corrected multiple comparison test (B–D). **** p < 0.0001. 63 A Antinuclear antibody 10 8 6 4 2 0 ntr ol l i. t. i.t . Co a C N gC l 2 M B C D E F CD4 T cells B cells Granulocytes Macrophages cDC 4.0 0.2092 8.0 0.2807 20.0 0.9009 8.0 0.6377 6.0 0.4399 3.0 6.0 15.0 6.0 4.0 2.0 4.0 10.0 4.0 2.0 1.0 2.0 5.0 2.0 0.0 0.0 0.0 0.0 0.0 Cl l 2 Cl l 2 Cl l 2 Cl l 2 Cl Na gC Na gC Na gC l 2 M M M N a gCM N a C Mg G Conventional H Conventional I CD8 : Tregs ratio CD4 T cells CD4: Ki67 1.5 0.8 5.0 0.6936 0.4369 0.9237 0.6 4.0 1.0 3.0 0.4 2.0 0.5 0.2 1.0 0.0 0.0 0.0 Cla Cl 2 aC l Cl 2 C l a Cl 2N Mg N Mg N Mg J Magnesium concentration per injection 1.0 **** 0.99 0.8 **** 0.6 0.4 0.2 0.0 1 ) ) D- aC l Cl 2 nti -P (N Mg a -Li po (ipo D- 1 1-L i-P PD - t - an an ti Supplementary Figure 6 64 cells per g tumor (x10^8) cells per g tumor (x10^8) mM Mg2+ ng ml -1 cells per g tumor (x10^8) cells per g tumor (x10^7) Ratio cells per g tumor (x10^8) cells per g tumor (x10^9) cells per g tumor (x10^8) Figure S6. Effect of Mg2+ on ANA production and intratumoral immune phenotype, related to Figure 6 (A) Concentration of anti-nuclear antibodies (ANAs). Each symbol represents an individual mouse, n = 5–10 per group. (B–I) Cell number of tumor-infiltrating CD4+ T cells (B), B cells (C), granulocytes (D), macrophages (E), conventional dendritic cells (F), conventional CD4+ T cells (G), Ki67+ conventional CD4+ T cells (H), as well the CD8:TReg ratio (I). Each symbol represents an individual mouse, n = 5 per group (B–I). (J) Magnesium concentration per injection of respective treatment modality. Data are presented as median ± IQR (A–I) and mean ± SD (J); dashed line in (A) indicates detection limit. Statistical significance was assessed by unpaired two-tailed Student’s t test (B–I) and one-way ANOVA with Sidak corrected multiple comparison test (J). **** p < 0.0001. 65 66 Figure S7. Association of Mg2+ abundance and efficacy of immune therapeutic modalities in vitro and in vivo, related to Figure 7 (A and B) Flow cytometry-based assessment of activation-induced LFA-1 headpiece opening on PHA blasts (A), as well FAK phosphorylation in PHA blasts (B), after co-culture with Ramos cells at a Blinatumomab concentration of 300 pg mL1 (n = 5 healthy donors) (C) Activation-induced glycolytic activity of human CAR T cells. Representative experiment (left panel), quantification of summarized data (right panel) from n = 3 healthy donors (D) Left panel, representative histograms of surface activation markers on CAR T cells. Right panel shows quantitation of respective staining from n = 3 healthy donors (E and F) Magnesium abundance determined by ICP-MS in serum from mice in CAR T cell tumor-rejection experiment after 14 days on respective diet (E). Magnesium abundance in interstitial fluid of tumors from UTD and saline control mice reaching ethically acceptable end point (F) (G) Diet consumption of mice in CAR T cell tumor-rejection experiment (H) Body weight after 24 days of respective diet relative to individual starting value (E–H) Each symbol represents one mouse (I) Study diagram and patient exclusion criteria (J) Table summarizing baseline characteristics. DLBCL, diffuse large B cell lymphoma; PMCL, primary mediastinal large B cell lymphoma; TFL, transformed follicular lymphoma. (K) Comparison of patients stratified according to mean serum magnesium levels between day 5 and day +3 of treatment in normal Mg2+ group (>1.7 mg dl1) and low Mg2+ group (<1.7 mg dl1), n = 5 measurements per patient. Each symbol represents individual patient (L) Progression-free survival (M) Study diagram and patient exclusion criteria (N) Table summarizing baseline characteristics (O and P) Comparison of radiographic response (O), and event-free survival (P) in normoversus hypomagnesemic patients with NSCLC. Data are presented as mean ± SD in (A, B, E, G, H, and K) as well right panels of (C and D), median ± IQR (F) and in representative metabolic flux analyses symbols indicate mean ± SEM in the left panel of (C). Statistical significance was determined with unpaired Student’s t test (A, B, E–G, and K), one-way ANOVA with Sidak’s multiple comparison test (C), unpaired two-tailed Student’s t test with Holm-Sidak corrected multiple comparison test (D and H) and log-rank Mantel-Cox test (L and P). * p < 0.05, ** p < 0.01,***p < 0.001, **** p < 0.0001. 67