Enhanced methionine cycle suppresses naïve CD8+ T-cell maturation

The metabolic pathways controlling naive CD8+ T (Tn) cell maturation following thymic egress remain mostly undefined. This is important because immature Tn are a major component of peripheral immune tolerance in newborns and under lymphopenia. In this study we demonstrate that TMRM, a mitochondrial membrane potential marker, could be applied to rapidly identify an immature Tn cell population in the periphery. Applying this marker to perform metabolic and proteomic analysis, we show that immature Tn cells maintain accelerated methionine cycle in respect to mature Tn. This unique metabolic state was associated with restricted Tbx21 locus and diminished immune response in vitro and in vivo. Following our findings, we demonstrate that inhibition of methionine cycle leads to rapid functional maturation of Tn and recovery of immune response to stimuli. Our work provides insight into the way the rate of methionine cycling regulates T cell maturation, opening a path for metabolic manipulation of immune tolerance.

The mechanism governing naive CD8 + T-cells (Tn) post-thymic maturation in the periphery is 55 incompletely understood (Boursalian et al., 2004;Cunningham et al., 2018a;Fink, 2013; 56 Hogquist et al., 2015). In particular, the role of metabolism in regulating the transition into 57 mature Tn remains unknown. Understanding the pathways critical for T cell maturation is 58 important because immature Tn are thought to be a major component of immune tolerance 59 adapted to provide protective immunity in newborns and during lymphopenia. (Cunningham 60 et al., 2018a). 2016b). We wanted to characterize a marker that could identify immature naive CD8 + T cell 112 populations independent of tissue specificity or time elapsed from thymic output. We chose 113 to use TMRM, a mitochondrial polarization dependent stain, since it is an efficient tool for 114 identification of distinct effector T cell and macrophages populations in humans and mice 115 (Sukumar et al., 2016a). To normalize our findings, we also measured the TMRM intensity of 116 Tn cells following oligomycin (complex V inhibitor) or Carbonyl cyanide-p-117 trifluoromethoxyphenylhydrazone (FCCP, uncoupler) treatment. These values were used as 118 reference for maximal or minimal TMRM staining, respectively. Flow cytometry analysis of 119 Tn obtained from wild type (WT) mice spleens and stained with TMRM revealed a bipolar 120 TMRM staining pattern, wherein the major subpopulation showed decreased TMRM staining 121 (dpT) and the minor nearly maximal TMRM staining (hpT) (Figures 1A-B, Figure S1A). To 122 exclude the possibility that the differences observed in polarization state stems from a 123 transient phenomenon, we assessed the capacity of the two Tn subpopulations to regain 124 their original phenotype following mitochondrial membrane uncoupling. Tn cells were treated 125 with FCCP to depolarize their mitochondria and reach minimal TMRM staining. FCCP 126 treated Tn were then washed and allowed to recover from FCCP-induced uncoupling. 127 Following the recovery period, Tn were restrained with TMRM and analyzed by flow-128 cytometry ( Figure S1B). We found that after FCCP withdrawal, TMRM staining was identical 129 to that seen in untreated cells (Figures S1C). These results indicate that the differences in 130 TMRM staining pattern observed in Tn is linked to a stable metabolic divergence. 131 To characterize the two Tn populations, we compared them by Forward Scatter (FSC) and 132 Side Scatter (SSC)-proxies for cell size and granularity, respectively-, and saw no 133 differences ( Figure S1 D-E). We found that the two Tn populations were indistinguishable in 134 size, or granularity. To examine whether the Tn populations identified are related to 135 differences in T cell receptor (TCR)-repertoire, we analyzed TMRM staining in Tn cells from 136 OT1 or PMEL17, a TCR transgenic mice, giving rise to a single CD8 + TCR clone. Similar to 137 WT mice, Tn cells from OT1 and PMEL17 mice gave rise to two distinct subpopulations 138 (Figure S1 F-G). 139 Naïve T cell populations vary between compartments (Van Den Broek et al., 2018). We 140 therefore next compared the proportion of hpT and dpT cells in secondary and primary 141 lymphatic tissues. We found that -similar to the spleen-Tn from secondary lymph nodes and 142 blood had a majority of dpT cells ( Figure 1A-B and S1H-I, S1K). In contrast, the vast majority 143 of thymic single positive CD8 + cells (SP8) had hyperpolarized mitochondria ( Figure 1C-D, 144 S1J-K). 145 To examine the relevance of our findings to human immunity, we next analyzed TMRM 146 staining of human T cells. As observed in mice, human resting Tn cells (CD28 hi , CD45RA + ) 147 gave rise to bipolar TMRM staining (Figures S1L, S1N), and staining was higher in 148 thymocytes than Tn from peripheral blood or adenoids (Figure S1M-N). Together these 149 findings suggest that as in mice, human hpT relative abundance relates to the rate of thymic 150 output. 151 Age-related thymic regression is associated with a decline in Tn cell output (George and 152 Ritter, 1996;Lynch et al., 2009;Ma et al., 2013). To support the link between hpT and 153 thymic output we examined the ratio of hpT to total Tn in mice grouped by age. We found 154 that in newborn mice, hpT are the majority (~80%) of the total Tn cells. This population markedly declined with age, reaching approximately 5% of the Tn cell population in aged 156 mice ( Figure 1E-G). In line with these results, the hpT/Tn ratio substantially increased in 157 adult mice following lymphocytes recovery from T cell depletion ( Figure 1H-J). 158 Based on our results, we hypothesized that TMRM staining corresponded to Tn maturation 159 state. We used ATAC-seq to compare the chromatin landscape of SP8 thymocytes, hpT and 160 dpT cells. Both principal component analysis ( Figure 1K) and gene clustering analysis 161 ( Figure 1L) revealed distinct chromatin accessibility landscapes among the cells, with hpT 162 cells intermediate between SP8 thymocytes and dpT cells. 163 To assess whether the hpT also shares functional similarities with their thymic progenitors, 164 we evaluated cytokine production of SP8 thymocytes, hpT and dpT in response to activation 165 using anti-CD3ε and anti-CD28 agonist antibodies. In line with the ATAC-seq data, the 166 percentages of cytokine-producing hpT and SP8 thymocytes were similar and reduced in 167 comparison to cytokine-producing dpT ( Figure 1M-P). Overall, our results so far indicate that 168 TMRM-based mitochondrial polarization staining distinguishes between immature-hpT and 169 mature-dpT naïve CD8 + T cell subsets. 170 Immature-hpT are hyporesponsive to diverse stimuli relative to mature-dpT 171 Immature CD8 + T cells support protective immunity in lymphopenic environment and during 172 chronic inflammations due to their reduced responsiveness to diverse immune stimuli 173 To ensure that T cell functional assays are not influenced by TMRM related toxicity, we 177 measured Tn activation in vitro in the presence or absence of high TMRM concentration 178 (100 μM). In line with previous reports, the presence of TMRM in the medium had no effect 179 on cell proliferation or elevation of activation markers ( Figures S2A-B). Next, we compared 180 the response of both Tn subsets to homeostatic proliferation signals. Immature-hpT and 181 mature-dpT cells isolated from congenic Ly5.2 and Ly5.1 mice, respectively, were adoptively 182 transferred at 1:1 ratio into Rag1 -/recipient mice and monitored for proliferation in the 183 peripheral blood at different time points (Figure 2A). Five days following the transfer, dpT-184 donor cells (Ly5.1) rapidly increased their abundance in the total Tn-donor pool relative to 185 hpT-donor cells (Ly5.2) ( Figure 2B). This ratio between the two competing donor subsets 186 remained constant also twenty-five days after the transfer ( Figure 2B). These results suggest that the mature-dpT responds better to homeostatic expansion cues relative to the 188 immature-hpT. 189 To further explore the lag in the immature-hpT response in a lymphopenic environment, we Donor cells were then analyzed by flow cytometry to correlate the levels of proliferation and 194 TMRM intensity ( Figure 2C). In line with our hypothesis, hpT-donor cells that depolarized 195 their mitochondria, proliferated to a greater extent than hpT-donor cells that retained their 196 increased mitochondrial polarization levels ( Figure 2D-E). 197 Next, we tested whether the immature-hpT population also maintains a reduced 198 responsiveness to activation stimuli compared to that of the mature-dpT population. Isolated 199 Cell-Trace labeled hpT and dpT cells from the spleens of OT1 transgenic mice were 200 stimulated with various concentrations of ovalbumin-derived SIINFEKL peptide. We found 201 that mature-dpT cells proliferate at a significantly higher rate in a peptide-concentration-202 dependent manner in comparison to the immature-hpT ( Figure 2F). Similarly, the killing 203 capacity of the dpT population was substantially higher than that of its hpT counterpart, 204 reaching a two-fold increase at most ratios of effector vs. target cells ( Figure 2G respect to the mature-dpT ( Figure 3B). Finally, we used proteomic analysis to quantify the 234 levels of mitochondrial-associated proteins, in both immature-hpT and mature-dpT. We 235 identified seventeen enriched mitochondrial associated proteins in dpT vs. only four in hpT, 236 demonstrating that mature-hpT maintain increased levels of mitochondrial associated 237 proteins in respect to the immature-hpT population ( Figure 3C). Together these observations 238 demonstrate that the immature-hpT hold decreased mitochondrial mass and respiratory 239 capacity in respect to the mature-dpT. 240 Next, we examined the metabolic profile of each population using metabolomic analysis. We 241 focused on metabolic circuits that were previously shown to be important to T cell function 242 and fate, including TCA and glycolysis. We found higher abundance of Acetyl-CoA in the 243 mature-dpT in respect to the immature-hpT, a metabolite associated with increased 244 mitochondrial activity and acetylation events (Pietrocola et al., 2015). In contrast, succinate, 245 a metabolite which was previously reported to stabilize HIF1a (Tannahill et al., 2013), was 246 higher in the immature-hpT in comparison to the mature-dpT ( Figure 3D). Analysis of the 247 glycolytic pathway, a hallmark of T cell biology (Fox et al., 2005; Gerriets and Rathmell, 248 2012), revealed differences in two intermediates between the two populations. Specifically, 249 we found that phosphoenolpyruvate, a metabolite important for effector T cell function (Ho et 250 al., 2015), was higher in the mature-dpT populations ( Figure 3E). In contrast, both cellular 251 and secreted lactate were higher in the immature-hpT population ( Figure 3E, S4D). 252 Collectively, these findings demonstrate that the immature-hpT maintains a distinct 253 metabolic network in respect to the mature-dpT subset.

mature-dpT 256
We next attempted to identify a link between the functional and the metabolic differences 257 observed between the two Tn subsets. Further analysis of proteomics data from the two Tn 258 subsets highlighted several key metabolic circuits that are known to affect cellular function 259 ( Figure 4A-B). As expected, mitochondrial, electron transport chain and fatty acid oxidation 260 proteins were enriched in the mature-dpT population in respect to the immature-hpT. 261 Likewise, proteins related to the amino acid metabolism and the folate cycle, shown to be 262 critical for T cell activation (Miyajima, 2020), were also enriched in the mature-dpT subset 263 ( Figure 4A-B). In contrast, the immature-hpT demonstrated higher enrichment of proteins 264 related to glycolysis, nucleotide metabolism and the methionine cycle ( Figure 4A A substantial constraint to investigating Tn maturation process is the luck of simple and 355 available markers, independent of tissue localization or time elapsed from thymic egress. 356 We therefore first investigated whether we could apply TMRM to distinguish immature Tn 357 from the total peripheral Tn population. In mature mice, we identified two populations; the 358 minor subpopulation (hpT), maintained a substantially higher mitochondrial membrane 359 potential relative to the major naïve CD8 + T cell subpopulation, dpT. Using numerous 360 assays, we identified a strong positive association between thymic output and the 361 proportions of hpT in the total naïve CD8 + T cells pool. We also found that the larger dpT 362 maturates from the hpT progenitor subset. To verify that TMRM could be used as a probe for 363 immature hpT we conducted comparative ATAC-seq profiling of thymic SP8, hpT and dpT 364 splenocytes. This assay revealed that hpTs represents an immature subset in transition 365 between SP8 and mature-dpT. The ability to identify immature-Tn using a simple and 366 commercially available mitochondrial probe, opened an opportunity for us and others to 367 better characterize immature-hpT and establish the cues regulating its abundance and 368 function. 369 We next applied our system to examine the functional differences between immature-hpT 370 To inspect whether hpT functional hyporesponsiveness is related to distinct metabolic 378 networks, we quantified both metabolic and mitochondria related proteins as well as 379 metabolites extracted from immature-hpT and mature-dpT. Likewise, we tested 380 mitochondrial capacity using mtDendra2 mice, quantification of mtProteins and Seahorse 381 analysis. We found that immature-hpT maintain decreased mitochondrial mass, proteins, 382 and respiratory capacity in respect to mature-dpT. Proteomics and metabolomics analysis 383 has highlighted a marked difference in the metabolic network in the two subsets. Immature-384 hpT had increased levels of succinate, a proinflammatory metabolite, and lactate in 385 comparison to mature-dpT. These findings suggest that mitochondrial biogenesis and 386 rewiring, which are critical for T cell activation, may also be important for Tn maturation. 387 The mature-dpT subset maintained increased levels of PEP, a metabolite associated with 388 improved effector function (Ho et al., 2015). In addition, we found an increased acetyl-CoA abundance in mature-dpTs relative to immature-hpTs. Acetyl-CoA is central for chromatin 390 and protein acetylation and may be central for acquiring the mature phenotype. In addition, 391 our analysis revealed an increase in abundance of DNMT3, a methionine cycle enzyme, in 392 immature-hpT relative to mature-dpT. Motivated by this finding, we examined the rate of 393 methionine cycle by SFGN inhibition. Following inhibition, we observed a significant increase 394 in SAM signal in treated immature-hpT in comparison to untreated immature-hpT control. For the experiments with Sinefungin the cells were pre-incubated with 30µM Sinefungin for 486 3h and then seeded into plates coated with anti-CD3 as described above with addition of 487 soluble anti-CD28, rhIL-2, anti-IFNγ and rmIL-12 (2ng/ml, Peprotech). Intracellular levels of 488 T-bet, IFNγ and TNF were analyzed after 36 h of culture. 489

In vitro T cell S35-Met Cytotoxicity assay 490
The ability of CD8 + T cells to kill target cells was directly measured using the standard S35- shaped plates at 37ºC for 5 h. Following incubation, plates were centrifuged (1600rpm, 5min, 496 4ºC) and supernatants (50µl) were collected and transferred to opaque Opti-plates 497 (Packard). Wells were added with 150µl scintillation liquid (Packard) and analyzed by a β-498 counter (Packard). The total amount of labeled cells was determined by adding 100μl of 499 0.1M NaOH to an equal amount of targets (5000/well) that were not added with effectors. 500 Spontaneous release was determined by measuring radioactive readings (as above) in wells 501 that were given identical treatment as the experimental wells, but were not added with CD8 + 502 T cells. Final specific lysis was calculated as follows: ((radioactive reading -spontaneous 503 release) / (total labeling -spontaneous release))*100 = specific lysis. 504

T cells depletion assay 505
T cells were depleted by injecting i.p. to 8-week-old mice, with anti-Thy1.2 antibody. Eight 506 days post antibody injection, the polarization levels of the newly emerging CD8 + T cells in 507 the spleen of these mice were measured by TMRM staining using flowcytometry analysis. 508

Targeted metabolic analysis 518
Isolated hpT and dpT cells were cultured in 96 well plate (1× 10 6 cells/well), suspended in 519 RPMI supplemented with 10% dialyzed Fetal Bovine Serum and 100μM Alanine with or 520 without 30µM Sinefungin. Following 5 or 24 h, were then extracted for metabolomics LC-MS 521 analysis. min, starting at 20% aqueous (20 mM ammonium carbonate adjusted to pH.2 with 0.1% of 538 25% ammonium hydroxide) and 80% organic (acetonitrile) and terminated with 20% 539 acetonitrile. Flow rate and column temperature were maintained at 0.2 ml/min and 45°C, 540 respectively, for a total run time of 27 min. All metabolites were detected using mass 541 accuracy below 5 ppm. Thermo Xcalibur was used for data acquisition. TraceFinder 4.1 was 542 used for analysis. Peak areas of metabolites were determined by using the exact mass of 543 the singly charged ions. The retention time of metabolites was predetermined on the pHILIC 544 column by analyzing an in-house mass spectrometry metabolite library that was built by 545 running commercially available standards. 546

Statistical analysis 586
The statistical significance of differences was determined by ANOVA. Differences with a P 587 value of less than 0.05 were considered statistically significant. Graph prism and Perseus 588 programs were use. MS data was normalized by ranking, when applicable, non-values were 589 plugged with replicates mean to prevent zeros bias. were considered and the required FDR was set to 1% at the peptide and protein level. 612 Protein identification required at least 3 unique or razor peptides per protein group. The 613 dependent-peptide and match-between-runs options were used. 614