NFAT signaling is indispensable for persistent memory responses of MCMV-specific CD8+ T cells

Cytomegalovirus (CMV) induces a unique T cell response, where antigen-specific populations do not contract, but rather inflate during viral latency. It has been proposed that subclinical episodes of virus reactivation feed the inflation of CMV-specific memory cells by intermittently engaging T cell receptors (TCRs), but evidence of TCR engagement has remained lacking. Nuclear factor of activated T cells (NFAT) is a family of transcription factors, where NFATc1 and NFATc2 signal downstream of TCR in mature T lymphocytes. We show selective impacts of NFATc1 and/or NFATc2 genetic ablations on the long-term inflation of MCMV-specific CD8+ T cell responses despite largely maintained responses to acute infection. NFATc1 ablation elicited robust phenotypes in isolation, but the strongest effects were observed when both NFAT genes were missing. CMV control was impaired only when both NFATs were deleted in CD8+ T cells used in adoptive immunotherapy of immunodeficient mice. Transcriptome analyses revealed that T cell intrinsic NFAT is not necessary for CD8+ T cell priming, but rather for their maturation towards effector-memory and in particular the effector cells, which dominate the pool of inflationary cells.


Introduction
The development of immune memory to viruses is usually studied in the context of primary virus infections, which leads to a potent immune response that clears virus infection.This immune memory protects the host during reinfection and generates an anamnestic response upon secondary infections [1,2].Chronic viruses, on the other hand, have developed immune evasion strategies that allow them to persistently maintain themselves in the host [3][4][5].Viral persistence can induce chronic inflammatory conditions and exhaust the adaptive host immune system over time, in particular the CD8 + T cell compartment [6].The exhaustion of virus-specific CD8 + T cells is characterized by poor effector cytotoxic activity, impaired cytokine production, and sustained expression of inhibitory receptors, such as programmed cell death-1 (PD-1) [7].In other cases, such as herpesvirus infection, the chronic virus infection is associated with long-term persistence of functional T cells despite continuous antigen stimulation [8][9][10].
Human cytomegalovirus (HCMV) is a prototypic herpesvirus that induces long-term chronic latent infection in human host.HCMV is of particular concern as a major cause of morbidity and mortality in transplant patients [11,12].Normally in humans, the progressive expansion and persistence of virus-specific CD8 + T cells in the context of chronic CMV infection leads to a massive enrichment of an oligoclonal pool of CD45RA + CCR7 -effector memory (TEMRA) cells.These cells lose the expression of classic memory markers like CD28 and IL-7R.This phenomenon is collectively referred to as 'memory inflation' [13].The murine CMV (MCMV) infection faithfully replicates the functional phenotype of inflationary T cell responses in mouse models and has been widely used to model human CMV infection [14].
CMV-specific persistent CD8 + T cell responses are maintained by continuous production of terminally differentiated effector cells that are characterized by downregulation of costimulatory receptors CD28, CD27, and also of the IL-7 receptor alpha chain (CD127) [15,16].These cells express high levels of KLRG1 and the NK cell-inhibitory molecule NKG2A [15].Another molecule expressed on inflationary cells is the chemokine receptor CX3CR1, a fractalkine receptor that is not expressed on conventional memory T cells generated following acute infections [17].Interestingly, conventional and inflationary CD8 + T cell responses emerge in parallel against different viral epitopes of CMV within the same host [16].This dichotomy is explained by differences in gene expression [18] and by the differential processing of persistent antigens, which require processing by the constitutive proteasome [19] in non-hematopoietic cells [20].The persistent CD8 + T cell responses are not induced by crosspresentation, but only if the target epitope is presented directly on the virus-infected cells [19].The conventional and inflationary CD8 + T cells also differ in their costimulatory requirements, where CD27 and CD28 costimulation is crucial for conventional CD8 + T cell responses but dispensable for persistent responses [21,22].The lack of CD27 and CD28 costimulatory molecules on persistent CD8 + T cells and their stimulation by direct antigen presentation suggest that direct TCR ligation is decisive for these responses, but this has not been demonstrated.
T cell receptor (TCR) ligation by antigen and the resulting TCR-calcium-calcineurin axis represent a central signaling pathway for T cell activation.The TCR-mediated increase in intracellular Ca 2+ and subsequent activation of calcineurin is responsible for nuclear factor of activated T cells (NFAT) signaling, i.e. its nuclear translocation and transactivation of target genes [23].NFAT family consists of five members first known as transcriptional regulators of T cell development, maturation and activation.Among these, NFATc1 to NFATc4 are activated by calcium-calcineurin signaling [24], of which only NFATc1 to NFATc3 are expressed in thymocytes and T cells.Among those, NFATc1 (also called NFAT2) and NFATc2 (NFAT1) play dominant roles in T cell activation and function.All NFAT members share a DNA binding domain that allows them to bind to a consensus DNA motif [25].In CD4 + T and Treg cells, different NFAT members have overlapping as well as distinct roles [26][27][28][29].In addition, NFATc1 and NFATc2 have been shown to regulate CD8 + T cell differentiation following acute LCMV infection [30], whereas especially NFATc1 is important for chronic LCMV-specific CD8 + T cells responses and inducing CD8 + cell exhaustion [31].However, the short isoform of NFATc1, NFATc1/αA [32] is highly upregulated during chronic LCMV infection [33].Despite this, the role of NFAT signaling or even a possible distinct role of NFATc1 versus NFATc2 in maintenance of persistent T cells responses during chronic CMV infections has not been explored so far.
In this study, we characterized the role of NFAT signaling in CMV-specific responses during acute and chronic infection.We show that NFAT-deficient CD8 + T cells can mount robust CMV-specific responses following acute infection, but NFATc1 is indispensable for maintaining persistent CD8 + cell responses during the chronic phase.Furthermore, NFATc1 and NFATc2 differentially regulate the CD8 + cell transcriptome that leads to differences in T cell differentiation during acute and chronic infection phases.The loss of NFATc1 in virus-specific CD8 + T cells induces a memory phenotype and a reduction of effector CD8 + T cells.Moreover, the combined loss of NFATc1 and NFATc2 enforces a stronger memory phenotype in form of an accumulation of central memory T cells.Interestingly, despite defects in T cell activation and differentiation, CD8 + T cells with selective ablation of NFATc1 or NFATc2 efficiently controlled the virus in immunocompromised animals.However, CD8 + T cells with combined loss of NFATc1 and NFATc2 showed reduced antiviral capacity.These results strongly indicate that TCR-mediated NFAT signaling is crucial in maintaining persistent T cell accumulation during chronic virus infection.
MCMV is derived from pSM3fr-MCK-2fl clone 3.3 BAC [36].MCMV-ie2-SIINFEKL was generated by 'en passant' mutagenesis as described previously [37] from the BAC-derived mouse cytomegalovirus pSM3fr-MCK-2fl clone 3.3.SIINFEKL peptide was inserted at C-terminus of ie2 ORF in MCMV genome.The viruses were reconstituted by BAC transfection in MEF cells.After reconstitution, the virus was propagated on M2-10B4 cells.Virus stocks were prepared according to previously described protocol [38].In brief, virus was pelleted from supernatants of infected cells (26000 x g for 3.5 h).Subsequently, the pellet was re-suspended in VSB buffer (0.05 M Tris-HCl, 0.012 M KCl, and 0.005 M EDTA, adjusted to pH 7.8) and then purified by centrifugation through a 15% sucrose cushion in VSB buffer (53000 x g), and a subsequent slow centrifugation step (3000xg, 5 min) to remove cellular debris.Animals were infected by injecting 10 6 plaque forming units (PFU) of MCMV via intraperitoneal or intranasal route.

In vitro virus titration
Animal organs were isolated and homogenized in DMEM medium supplemented with 5% FCS.MEFs were infected with organ homogenates diluted in DMEM.The infection was enhanced by centrifugation for 30 minutes at 1000 x g.The plates were incubated for another 30 minutes at 37˚C, 5% CO2.The organ homogenates were then removed, and cells layered with DMEM supplemented with 5% FCS and methylcellulose.The plates were incubated for 4 days at 37˚C, 5% CO2 and the plaques were quantified by visual inspection under inverted microscope.

Flow cytometry
Spleens and lymph nodes were collected and mashed through a 70-μm cell strainer to generate a single-cell suspension, followed by erythrocytes lysis with ACK buffer.Peripheral blood samples were collected by retro-orbital bleeding and were lysed directly.Before lungs isolation, lungs were perfused via the right heart ventricle with approximately 5 mL PBS to remove circulating blood.The lungs were cut into small pieces and digested for 30 min at 37˚C with 0.9 mg/mL collagenase type IV, 0.2mg/mL dispase and 40 ug/mL DNase I in complete RPMI medium (RPMI-1640 supplemented with 2.5% FCS, 100 U/ml penicillin and 100 U/ml streptomycin).Finally, the mononuclear cells were purified by gradient centrifugation over 30% Percoll.

Peptide stimulation assay
The SIINFEKL peptide (H-2K b -restricted) was synthesized and HPLC purified (95% purity) at the HZI peptide-synthesis platform.For intracellular staining, single cell suspensions were incubated with 1 μg/ml SIINFEKL peptides in complete RPMI medium for 1 h at 37˚C followed by brefeldin A (Golgiplug; BD Pharmingen) addition and further 5 h incubation.For intracellular staining, surface-labelled cell suspensions were fixed using eBioscience Foxp3/ Transcription Factor Staining Buffer Set or eBioscience IC Fixation Buffer (both from Thermo Fisher).Cell suspensions were stained with following antibodies at 4˚C for 30 min: anti-GzmB (clone QA16A02), anti-IFN-γ (clone XMG1.2) and anti-TNF-α (clone MP6-XT22).After washing with FACS buffer, the cells were analyzed using flow cytometer.

Adoptive T cells transfer
OT-I cells were isolated from spleen of CD90.1 OT-I mice using Naive CD8a + T Cell Isolation Kit (Miltenyi Biotec) according to manufacturers' guidelines.10 4 naïve NFATc1 KO, NFATc2 KO, NFATc1c2 DKO or WT OT-I cells were adoptively transferred into 8-12 week old recipient animals via tail vein injection.Next day, the animals were infected intraperitoneally with 10 6 PFU of MCMV-ie2-SIINFEKL.

RNA isolation and transcriptional analysis
OTI cells were sorted from spleen and LNs at 7 days post infection (dpi), and total RNA was extracted using RNeasy Plus Mini kit (Qiagen) according to manufacturer's instructions.
Quality and integrity of total RNA was controlled using 5200 Fragment Analyzer System (Agilent Technologies).The RNA sequencing libraries were generated using NEB Next Single Cell/ Low Input RNA Library Prep Kit for Illumina (NEB) according to manufacturer's protocol and sequenced on a NovaSeq 6000 sequencer (Illumina) using NovaSeq 6000 S1 Reagent Kit (100 cycles, paired end run 2 × 50 bp) with an average of 2.5 × 10 7 reads per sample.The GEO accession number for all RNA-seq data reported in this paper is GSE228527.Read quality of sequenced libraries was evaluated with FastQC.Sequencing reads were aligned to the reference mouse genome assembly GRCm38 using hisat2 [39].Reads aligned to annotated genes were quantified with featureCounts [40].Read counts were further processed with DESeq2 for quantification of differential gene expression [41].A log-fold change bigger than one and a false discovery rate cut-off of 5% was employed to select significantly over-and under-represented genes.Geneset enrichment analysis was performed using fgsea and clusterProfiler R packages.Analysis was performed on DEGs (p-val < 0.05) from the comparison between NFAT KO and WT OTI cells.The number of permutations was set to 1000.

NFAT KO animals fail to mount persistent CD8 + T cell responses during chronic infection
Animals with T cells that lack NFATc1 (Nfatc1 fl/fl .Cd4cre, earlier named Nfat2 fl/fl .Cd4cre), NFATc2 (NFATc2 -/-) or both (Nfatc1 fl/fl .Cd4cre.Nfatc2 -/-; i.e.DKO) were inoculated with MCMV, whereupon blood CD8 + T cell response kinetics were longitudinally monitored to define the persistence of virus-specific T cells.We used peptide-MHC tetramers against immunodominant epitopes of MCMV [16] to label the peripheral blood mononuclear cells (PBMCs), and thus measured CD8 + T cell responses via flow cytometry (representative gating strategy in S1 Fig) .M45-epitope-specific CD8 + T cells responded strongly to acute infection, followed by a rapid contraction phase after the first week (Fig 1A).NFATc1 KO and NFAT DKO animals showed impaired M45-specific responses at 7 dpi.Inflationary M38-specific responses do not contract, but rather accumulate steadily during chronic infection [16,42].We observed a highly significant reduction in long-term M38-specific CD8 + T cell responses of animals lacking NFAT molecules.The total CD8 + T cell number was reduced in the peripheral blood of animals lacking NFATc1 in T cells (Fig 1A).However, the overall CD8 + and CD4 + T cell response in the spleen of chronically (90 dpi) infected animals remained unchanged in the absence of NFATc1, whereas Nfatc2 -/-animals showed elevated numbers of CD8 + T cells in the spleen ( Nevertheless, NFAT KO animals showed similar levels of MCMV replication during acute infection, suggesting that acute infection control is not impaired in these animals (Fig 1D).The latent load of virus was also assessed in relevant sites of virus latency such as the spleen and salivary glands [43,44], and was not altered in NFAT KO animals over the values seen in WT control animals (Fig 1E).Thus, the lack of CD8 + T cell response or persistence in NFAT KO animals was not due to reduced virus replication or latent persistence.

NFATc1 signaling is required for persistent CD8 + T cell responses
In the conventional Nfatc2 -/-mice, not only T cells are NFATc2-deficient.We generated mixed BMC mice and further created a competitive environment for WT versus NFAT-deficient lymphocytes.Bone marrow cells from NFATc1 KO, NFATc2 KO, NFAT DKO or WT littermates were mixed 1:1 with congenic WT bone marrow cells and transferred to C57BL/6J animals whose hematopoietic cells were ablated by lethal irradiation (Fig 2A).Animals were  infected 3 months post bone marrow transplant and CD8 + T cell responses were tracked.The NFAT-deficient (CD45.1 -/-CD45.2+/+ ) CD8 + T cell reconstitution levels were similar to WT (CD45.1 +/-, CD45.2 +/-) controls (Fig 2B and 2C).NFATc2 KO CD8 + T cells showed a modestly increased frequency among total CD8 + T cells in line with previous reports [45], whereas the NFATc1 showed an inverse pattern.The frequency of CD45.2 +/+ T cells remained stable or showed modest reduction during acute infection (Fig 2C).However, the fraction of NFATc1 and NFAT DKO CD8 + T cell populations was significantly reduced during latency for both epitopes tested, where the contraction occurred by 14 days post infection and the populations were stably reduced thereafter.
Interestingly, M45-specific responses in PBMCs of mixed BMC animals were not reduced in absence of NFATc1 (Figs 2D and S3A).This was in contrast with the phenotype observed in NFAT KO animals, where a significant reduction in M45-specific cell number was observed in NFATc1 KO and NFAT DKO animals (Fig 1A).Similarly, NFATc2 KO populations showed a two-fold increase in M45-specific CD8 + T cells of BMC mice and no difference in conventional NFATc2 KO animals, which suggests that NFAT signaling in non-CD8 + T cells may have contributed to differences in phenotypes between BMC and KO animals.M38 and m139-specific cell responses were significantly reduced in the CD8 + T cell populations lacking NFATc1 signaling, because the steady accumulation of antigen-specific cells observed during chronic infection was absent in NFATc1 KO and NFAT DKO cells in absolute (Fig 2D) and relative terms (S3A Fig) .At times of virus latency, a significant reduction of M38 and m139-specific responses in NFATc1 KO and NFAT DKO cells was observed in spleen and lungs, although similar numbers of antigen-specific T cells were observed across all populations in the mesenteric lymph nodes (LNs) (Fig 2E).The congenic and co-transferred WT populations in BMC animals (defined as CD45.1 +/-CD45.2+/-) showed consistent levels of CD8 + T cell responses during chronic infection regardless of co-transferred NFAT-deficient cells (S3B Fig), suggesting similar levels of latent virus genomes and antigen presentation in the BMC animals.Thus, WT cells mounted normal responses, whereas the T cells that lack NFATc1 signaling were unable to maintain persistent responses in the analyzed tissue compartments.These data suggest that NFATc1 is necessary for maintaining persistent CD8 + T cell responses during chronic infection.

Intrinsic NFAT signaling regulates CD8 + T cell differentiation during chronic infection
Following MCMV infection, CD8 + T cell persistence is maintained by continuous production of terminally differentiated effector T cells that are characterized by downregulation of costimulatory receptors (CD27 and CD28) and upregulation of killer cell lectin-like receptor subfamily G member 1 (KLRG1) [15,46].To investigate the phenotype of NFAT KO CD8 + T cells during the chronic phase, we used the mixed BMC model.The chronically infected animals were sacrificed at 90 dpi to determine the expression of activation and memory markers on CD8 + T cells.NFAT DKO CD8 + T cell populations showed reduced expression of CD44 and CD11a and enhanced expression of CD27 and CD62L (S4  S5B).The frequency of the splenic CM population was also somewhat increased in NFATc1 KO cells, but this was not statistically significant (Fig 3C , 3D).Next, we checked the CM phenotype among the virus-specific CD8 + T cells.A high percentage of M45-specific cells showed a CM phenotype at 90 dpi in all conditions, which is consistent with the conventional and non-inflationary nature of this epitope [15].In line with previous studies [15,16], M38-specific CD8 + T cells showed predominantly an effector phenotype, but in NFAT DKO mice, the CM frequency was significantly increased (Fig 3E).The absolute number of CM cells was similar among all groups in different organs (S5C Fig) , arguing that the increase in the frequency of CM in NFAT DKO CD8 + T cells was only relative and due to a reduction of effector cells.
Since fewer KLRG1 + CD27 -CD8 + T cells are generated in absence of CXCR3 [47,48], we next investigated if impaired effector generation from NFAT KO populations matched an impaired CXCR3 expression.Interestingly, we observed that NFAT DKO CD8 + T cells expressed high levels of CXCR3 in all tested organs (Figs 3F and S4), which indicated that NFAT deficiency does not interfere with CXCR3 upregulation upon CD8 + T cell activation, but rather with effects subsequent to CXCR3 expression in early T cell activation.Thus, we analyzed the fractalkine receptor CX3CR1, another important chemokine receptor that has been used as a marker for differentiated effector T cells.CX3CR1 is upregulated on inflationary CD8 + T cells during chronic infections [17,49], and we observed that NFAT DKO cells did not upregulate CX3CR1 expression (Figs 3G and S4), which is in line with the memory-like phenotype observed in the same cells (Fig 3A and 3B).Similar to the KLRG1 + CD27 -phenotype, the frequency of CX3CR1 + T cells among NFATc1 KO population was not significantly changed, but the absolute count of the CX3CR1 + CD8 + T cells significantly decreased in NFATc1 KO in spleen and lungs (S5D Fig) .These data indicate that the lack of NFAT signaling altered the CD8 + T cell differentiation during chronic phase, where the loss of NFATc1 alone led to a significant decrease of effector cells, but the loss of both c1 and c2 increased the fraction of cells with a memory phenotype.Thus, the lack of inflationary CD8 + T cells responses was associated with altered T cell differentiation.

Cell-intrinsic NFAT signaling regulates CD8 + T cell differentiation following acute infection
Our data suggested that NFATc1 signaling promotes KLRG1 + CD27 -T cell differentiation and memory inflation during chronic virus infection.In contrast, data from our BMC model showed that the lack of NFAT signaling did not impair early epitope-specific responses in peripheral blood (Fig 1D).To ascertain in more detail the effect of NFAT signaling in antigenspecific CD8 + T cells following acute MCMV infection, we adoptively transferred TCR transgenic OTI T cells on NFATc1 KO, NFATc2 KO, NFAT DKO or WT background to congenic animals that were then infected with MCMV-ie2-SIINFEKL [50].Akin to the BMC model, the (CD44 + CD11a + ) cells (left).Mean CX3CR1 + CD8 + T cells population from blood, spleen and lungs at 90 dpi are shown as mean ± SEM, each dot is a mouse.Data are pooled from two experiments and each dot represents one mouse, n�7.Statistically significant differences are highlighted; *, p < 0.05; **, p < 0.01; ***, p < 0.001; (Mann-Whitney U Test); mean ± SEM values are plotted. https://doi.org/10.1371/journal.ppat.1012025.g003

PLOS PATHOGENS
lack of NFATc1, NFATc2 or both did not impair CD8 + T cell responses (Fig 4A ), but in contrast to it, not only NFATc2 KO, but also NFAT DKO OTI responses were significantly larger than WT response.Absolute numbers of CM OTI cells were increased in absence of NFATc1 or NFATc2, while the relative frequency of CM cells was significantly increased only in absence of NFATc1, but not in absence of NFATc2 (Fig 4B).The lack of NFAT signaling in OTI cells resulted in reduced KLRG1 + CD27 -OTI frequency following acute infection (Fig 4C).The KLRG1 -CD27 + cell frequency did not change significantly but CD8 + T cells with defective NFAT signaling showed a pronounced increase in KLRG1 + CD27 + cells, which represent a transitory state.Hence, our data suggest that the loss of NFATc1 appears to restrict the transition of the CM phenotype to more mature forms, and this is further enhanced in absence of NFATc2, suggesting non-redundant and distinct functions of NFATc1 and NFATc2.
To understand how NFAT signaling regulates CD8 + T cell differentiation, we performed detailed transcriptional profiling of NFATc1 KO, NFATc2 KO, NFAT DKO and WT OTI T cells isolated from spleen of acutely infected animals.NFATc1 and NFATc2 KO T cells showed distinct transcriptional signatures with some overlaps (Fig 4D and 4E).We observed a clear under-representation of transcripts related to effector function in NFATc1 KO CD8 + T cells.These transcripts code for molecules such as granzyme A, perforin, CX3CR1, KLRG1 and various other killer-like receptors.Other transcripts that are associated with central memory T cells, such as Id3, Ccr7, Il7r and Sell, were significantly upregulated in NFATc1 KO T cells.Transcriptional profile of NFATc1c2 DKO CD8 + T cells showed a pattern similar to one from NFATc1 KO T cells (S6A Fig) .A total of 4115 and 4929 transcripts changed in NFATc1 and NFATc2 KO T cells, respectively.Among these transcripts, 2704 were affected in both NFATc1 and NFATc2 KO T cells (Fig 4G).NFAT DKO cells showed only 266 unique transcripts that were not affected in the single KO CD8 + T cells.In the heatmap, unsupervised clustering of samples, using top 500 genes that were differentially expressed (p<0.05) in NFAT DKO, resulted in distinct clades for each genotype (Fig 4F).NFAT KO cells showed reduced abundance of Gzma, Cx3cr1, Klrg1, and Prf1.Some transcripts such as E2f2, Zeb2, Id2 were only reduced in NFATc1 and NFAT DKO cells (Figs 4F and S6D).Similary, genes involved in memory phenotype like Id3, Eomes, Il7r, Cd27 and Ccr7 were selectively upregulated in NFATc1 and NFAT DKO cells.Similar pattern was observed for genes involved in cell cycle regulation, where only T cells lacking NFATc1 showed downregulated gene expression for cell cycle genes, such as Cdk1 and Cdk2.Gene set enrichment analyses confirmed that expression of effector genes [51] was curtailed only in NFATc1 KO, but not in NFATc2 KO cells in a statistically significant manner (Fig 4H).Taken together, the data showed that NFAT signaling drives transcriptional changes associated with CD8 + T cell differentiation and effector function, and NFATc1 and NFATc2 molecules regulate an only partly overlapping set of genes.

Intrinsic NFAT signaling dictates CD8 + T cells trafficking to lymphoid and non-lymphoid organs in early infection
NFAT signaling controlled the CD8 + T cell transcription and their differentiation into effector T cells.The expression of chemokine receptors was also affected in CD8 + T cells that lack NFAT molecules (S6C Fig) .We observed that NFATc2 and NFAT DKO CD8 + T cells showed increased Il7r gene expression, and loss of NFATc1 or NFATc2 affected Il15 and other cytokine receptor gene expression.This prompted us to investigate if changes in chemokine receptors translate to changes in CD8 + T cell migration to different lymphoid and non-lymphoid organs.We investigated the role of NFAT signaling in acute infection of BMC animals, and observed that antigen-specific responses from NFATc1-deficient or NFAT DKO CD8 + T cells were impaired in lungs, but not in lymph nodes or spleen (Fig 5A).Akin to NFATc2 KO OTI To test if the accumulation of antigen-specific CD8 + T cells observed in LN in the BMC model is related to CD8 + T cell intrinsic signaling, we performed a similar analysis using NFAT deficient OTI T cells.We observed an accumulation of NFAT DKO T cells in mesenteric LN but not in the lungs (Fig 5F).We analyzed CD8 + T cell responses in peripheral and non-draining LNs, such as brachial or inguinal LN, and observed an increased total number of NFAT DKO OTI cells.Furthermore, NFAT DKO OTI cells showed a CM phenotype in all lymphoid and non-lymphoid organs (S7D Fig) .RNA-seq profile of WT and NFAT DKO OTI cells from LN was compared to the transcriptional profile of cells isolated from the spleen.WT cells isolated from LN and spleen clustered in distinct clades (Fig 5G ), where cells isolated from spleen showed enhanced expression of genes involved in effectors function, such as Tbx21, Klrg1 and Gzma.NFAT DKO OTI cells from spleen and LN showed much less differences from each other.In both cases, they failed to upregulate genes involved in effector functions, and showed an increased expression of genes associated with memory phenotype.Interestingly, we observed that NFAT DKO CD8 + T cells showed altered transcriptional profile of genes involved in cell cycling, such as Mki67 (Fig 5G).Ki-67 labelling showed a modest reduction in Ki-67 + population among CM NFAT DKO CD8 + T cells, while the Ki-67 + population frequency among total CD8 + T cells did not change significantly (Fig 5H).Furthermore, NFAT DKO cells showed altered expression of genes involved in CD8 + T cells migration, such as S1pr1, S1pr5 and chemokine receptors (S7E Fig) .Since memory cells preferably home to lymphoid organs [52,53], it is likely that the accumulation of NFAT DKO cells in LNs is due to their memory phenotype.

Virus control by NFAT deficient CD8 + T cells
NFATc1 has been shown to control CD8 + T cell cytotoxicity in vitro [54].Flow cytometric analysis of in vitro SIINFEKL re-stimulated splenocytes obtained from MCMV-ie2-SIINFEKL infected mice showed that combined loss of NFATc1 and NFATc2 results in lower IFN-γ production and reduced frequency of IFN-γ + TNF-α + T cells (Fig 6A).We have shown previously that IFN-γ and TNF-α producing CD8 + T cells control MCMV virus replication in co-culture settings [38].While CD8 + T cells play an important role in virus control, they are not necessary for control of acute virus infection due to redundant immune effectors [55,56] antiviral capacity of NFAT-deficient CD8 + T cells, we used a model where CD8 + T cells were required for virus control.Hence, we adoptively transferred WT or NFAT-deficient OTI cells to RAG2gc KO animals, lacking T and B cells and harboring only non-functional NK cells, and infected them with an MCMV expressing the target SIINFEKL epitope.The animals receiving OTI cells did not show any significant weight loss upon infection, whereas the control group without OTI cells showed weight loss and some animals succumbed to virus infection (Fig 6B).Interestingly, NFATc1 and NFATc2 deficient OTI cells were able to control the virus infection as efficiently as WT OTI cells (Fig 5C).On the other hand, NFAT DKO cells controlled the virus with a similar efficacy only in the spleen, but virus titers were significantly elevated in SG, liver and lungs, consistent with the reduced NFAT DKO T cell migration to non-lymphoid organs (Fig 5E).We observed a modest reduction in virus titer in the salivary glands of animals that received WT CD8 + T cells, whereas the NFAT DKO T cell group did not show this reduction.Altogether, these data show that NFATc1 and NFATc2-deficient CD8 T cells are cytotoxic and may control virus replication in vivo, and only a combined loss of NFATc1 and NFATc2 leads to modest reduction in virus control.

Discussion
It has been proposed that the persistent response of CD8 + T cells to MCMV called memory inflation is continuously renewed and maintained by sporadic antigenic stimulation from the latent virus, where latent virus infection and persistent T cell responses constitute a feedback loop.As the latent virus sporadically replicates and stimulates, the T cells limit virus replication and keep MCMV in its latent form.This model predicts that memory inflation crucially depends on TCR activation and downstream signaling.
In our presented study on the role of NFAT signaling in T cell responses to a latently persisting herpesvirus, we observed a crucial role of NFAT family members in the induction of inflationary memory responses, in their maturation and in cell trafficking.NFAT-family members are core regulators of T cell development and maturation [28,30,35], activation [26,45] and functionality [31,33,54], with partly redundant and partly distinct roles.We found that NFATc1 is crucial for maintaining persistent CD8 + T cell responses during chronic virus infection.The NFATc2 KO T cells, on the other hand, showed overshooting responses suggesting that NFATc1 and NFATc2 serve a non-redundant, and somewhat opposite function during chronic infection.It was previously documented that heightened levels of NFATc1/αA, an isoform which is transcriptionally upregulated in a positive feedback loop [32], correlates with T cell exhaustion during chronic LCMV infection [33,57], now thought to be an essential early player in T cell exhaustion [58].Somehow opposing, we report that NFATc1 maintains the functional persistent responses during the MCMV latency.Surprisingly, the number of virusspecific T cells in the LNs was not affected during chronic infection.It has been suggested that antigen presentation on non-hematopoietic cells in LNs drives the persistent memory responses [20].The unchanged CD8 + T cell frequencies in the LNs in conjunction with the differentiation defects during the chronic phase would suggest that the lack of NFATc1 leads to insufficient TCR signaling for further differentiation including upregulation of homing receptors like CX3CR1 and / or impaired signal transduction for maintenance in peripheral organs.We analyzed the expression of cytokine receptors responsible for promoting CD8 + T cell survival, as we have previously shown that IL-15 is crucial for CMV-specific inflationary memory T cell maintenance [59].We observed that combined loss of NFATc1 and c2 resulted in an upregulation of Il7r and Il21r gene expression, while Il2ra and Il15ra expression did not change, which argues that the altered cytokine receptor profile is unlikely to be the primary cause for the observed lack of CD8 + T cells persistence.Contrary to the chronic phase, even the combined loss of NFATc1 and NFATc2 did not impair CD8 + T cell responses to acute MCMV infection, suggesting that other signaling cascades induced by strong inflammatory conditions triggered by MCMV infection can compensate the loss of most NFAT signaling.In line, suppression of CD28/B7 or CD27/CD70 signaling cripples primary CD8 + T cell responses to MCMV [22], while CD28/B7 and CD27/CD70 signaling is not important, but also not available during chronic MCMV infection [21], possibly explaining the higher dependence on TCR/NFAT signal transduction for the inflated memory.Taken together, these and previous observations advocate that distinct molecular pathways regulate the early and the late CD8 + T cell responses to MCMV.NFAT signaling via TCR ligation is crucial for CD8 + T cell persistence during the chronic phase, but weaker TCR signaling due to less NFAT can be compensated by the strong costimulation in the context of acute MCMV infection.
Xu et al. demonstrated that NFATc1 and NFATc2 differentially regulate T cell differentiation upon acute LCMV infection [30].In contrast to MCMV acute infection, the loss of NFAT impairs the virus-specific responses during LCMV infection.These contrasting outcomes may be explained by differences in priming requirements for these two viruses, since priming of MCMV-specific cells is facilitated by 'signal 2' pathways, rather than 'signal 3' cytokine co-stimulation that dominates responses to LCMV [22].As co-stimulation via CD28, CD27 or similar (signal 2) in MCMV priming activates NFAT-independent pathways like NF-κB and MAPK, it appears reasonable to assume that the initial response to MCMV-antigens may proceed in less NFAT-dependent manners than the LCMV responses.Notably, the gene expression profile associated with cell cycling was uniquely altered in the absence of NFATc1, and this transcriptional shift was evident even during the acute infection phase.These early alterations in the transcriptional landscape exclusively in NFATc1 knockout cells may explain the subsequent NFATc1-dependent loss of memory inflation during the later stages of virus infection.Interestingly, NFATc1 and NFATc2 played distinct roles in persistence of CD8 + T cell responses.The stronger differentiation defect observed in absence of NFATc1 than in the ablation of NFATc2, as well as the strongest effects due to combined loss of NFATc1 and NFATc2, suggest a hierarchy of roles that these transcription factors play in regulating differentiation.Namely, our data suggest that NFATc1 appears to be a key regulator of differentiation towards an effector phenotype, while NFATc2 contributes to it, but was not sufficient for cell differentiation in NFATc1 absence.This is in line with a transient, but robust upregulation of NFATc1/ αA in effector CD8 + T cells [54].Hence, the DKO cells, in which NFATc2 could no longer partly compensate the loss of NFATc1, completely failed to upregulate effector genes and receptors, whereas chemokine receptors and cell adhesion molecules typically present on memory cells were upregulated.Notably, DKO cells with a central memory phenotype that is associated with homing to secondary lymphoid organs accumulated in both draining and non-draining LNs, showing that this is unlikely due to an egress defect, because a pronounced accumulation was not observed only in draining LNs.Furthermore, the transcriptomic profiles of DKO cells isolated from LNs and spleen were remarkably similar, although this was not the case in WT controls, which further argues that DKO cells from spleen preferentially migrated to LNs, instead of lungs, due to their memory phenotype.
Based on in vitro experiments, NFATc1 has been suggested as a core modulator of CD8 + T cell cytotoxicity [54].However, we observed that selective ablation of NFAT molecules did not impair the ability of CD8 + T cells to control the virus in adoptive transfer settings in vivo, where only a combined loss of NFATc1 and NFATc2 showed a modest reduction in virus control.This discrepancy may be a reflection of redundant signaling cascades, especially T cell costimulation by other cells, which again could compensate for the loss of NFATc1 in the in vivo situation.Interestingly, we observed a modest CD8 + T cell-mediated virus control in the salivary glands (SG), an organ traditionally considered refractory to CD8 + T cell control due to virus-mediated MHC-I downregulation.We suspect that this modest virus titer reduction in the salivary gland likely reflects the robust control in other visceral organs by CD8 + T cells rather than direct control in SG.Consequently, lower virus levels migrate to SG, resulting in a subtle reduction in titer.Our previous study evidenced that selective ablation of individual NFAT molecules in BM-co-transplanted CD3 + T cells ameliorates graft-versus-host disease (GvHD) while maintaining anti-tumor control [34].In this setting, CD8 + T cell responses were intact.Here, we show that selective loss of NFAT members in T cells does not impair their anti-viral control, which provides an important avenue for CMV disease management.CMV remains a major cause of morbidity and mortality in organ [12] and allogenic stem cell transplantation (allo-HSCT) recipients [11].Organ transplant recipients are often treated with an allo-transplant of CMV-specific T cells for controlling the virus [60].However, broad-spectrum immune suppression including calcineurin inhibition, required to suppress organ rejection or GvHD, also affects the anti-viral function of adoptively transferred T cells.Our study suggests that an advanced therapy with selective NFAT repression may lower the risk of organ rejection while keeping the anti-CMV CD8 + T cell compartment functional.Using selective NFAT inhibitors [61] instead of broad-spectrum immune suppression might therefore improve virus control while ensuring survival of transplanted organs.For the setting of allo-HSCT, we found that CRISPR/Cas9-mediated single NFAT member ablation in BM-co-transplanted T cells is sufficient to ameliorate GvHD [62].When we now extrapolate our new findings we predict that single NFAT ablation will not only preserve the anti-tumor effect [34], but also CMV reactivation.
Fig 1B).The virus-specific CD8 + T cell response against the inflationary M38 epitope was significantly reduced in the spleen of animals lacking NFATc1 (Fig 1C).Following acute virus infection, the total CD8 + and CD4 + T cell response remained unaffected by the absence of NFATc1 or NFATc2 in the spleen or lymph nodes at 7 dpi (S2A and S2B Fig).The combined loss of NFATc1 and NFATc2 resulted in a modest increase in the frequency of CD44 + CD62L + CD4 + T cells at 7 dpi in spleen, though not statistically significant (S2C Fig).Additionally, we observed a slight reduction in total CD8 + T cell response in the lungs of intranasally infected Nfatc1 fl/fl .Cd4cre animals at 30 dpi, with a tenfold decrease in the number of M38-specific CD8 + T cells (S2D Fig), and a significant decline in the frequency of resident memory cells among M38-specific CD8 + T cells (S2E Fig).

Fig 1 .
Fig 1. NFAT KO mice fail to mount inflationary CD8 + T cell response following chronic MCMV infection.Animals lacking either NFATc1, NFATc2 or both were infected with 10 6 PFU of MCMV intraperitoneally.(A) Total CD8 + T cells and tetramer specific response kinetics in peripheral blood were tracked.Mean ± SEM are plotted for data pooled from 3 independent experiments (n�12).Total CD4 + T cells, CD8 + T cells (B) and M38 tetramer + T cell responses (C) in spleen at 90 dpi.(D) mice were sacrificed at 5 dpi to titrate the virus replication in spleen and lungs.(E) Infected mice were sacrificed 6 months post infection to quantify MCMV latent virus load, which is presented as virus copies per 10 5 host cells in spleen and salivary glands.Data in panel D and E are pooled from two experiments and each dot represents one mouse.Statistically significant differences are highlighted; *, p < 0.05; **, p < 0.01; ***, p < 0.001; ****, p < 0.0001; (Mann-Whitney U Test). https://doi.org/10.1371/journal.ppat.1012025.g001

Fig 3 .
Fig 3. NFAT signaling regulates effector CD8 + T cell differentiation during chronic infection.Mixed BMC animals were infected with 10 6 PFU of MCMV and analyzed at 90 dpi.(A) Representative flow-cytometric plots showing KLRG1 + CD27 -and KLRG1 -CD27 + populations from BMC mice with WT and NFATc1c2 DKO BM.The plots are pre-gated on primed (CD44 + CD11a + ) CD45.2 + CD8 + T cells (live CD3 + CD8 + ).(B) Pairwise comparison of KLRG1 + CD27 -and KLRG1 -CD27 + CD8 + T cell frequencies in blood, spleen and lungs of individual mice during chronic MCMV infection.Lines connect data from individual animals (C) Flow-cytometric plots showing representative central memory (CM) populations among primed blood CD8 + T cells of chronically infected mice (left).Kinetics of these CM populations in blood are shown on the right Lines connect group means at indicated time points, error bars are SEM.(D) Percentage of CM CD8 + T cells in spleen and lungs at 90 dpi.Bar plots represent the group average, error bars are SEM and each dot represents mouse.(E) Percentage of CM cells among M45 and M38 tetramer specific CD8 + T cells from spleen at 90 dpi.Bar plots represent mean ± SEM and each dot is a mouse.(F) Representative flow-cytometric plots of blood CD8 + T cells showing CXCR3 + population among primed (CD44 + CD11a + ) cells (left).Mean CXCR3 + CD8 + T cells population from blood, spleen and mesenteric LN at 90 dpi are shown as mean ± SEM, each dot is a mouse.(G) Representative flow-cytometric plots of blood CD8 + T cells showing CX3CR1 + population among primed

Fig 4 .
Fig 4. NFAT molecules controls CD8 + T cell differentiation following acute infection and promote distinct transcriptional signature.10 4 naïve OTI T cells were transferred to congenic C57BL/6 mice and activated by acute MCMV infection.CD8 + T cell responses were analyzed at 7 dpi.(A) Absolute number of OTI T cells in blood.(B) Absolute and relative size of CM OTI T cell in blood of acutely infected animals.(C) Frequency of KLRG1 + CD27 -, KLRG1 -CD27 + and KLRG1 + CD27 + populations among blood OTI T cells are plotted.(D-H) Transcriptional analysis (RNA sequencing) was performed on OTI T cells isolated from spleen of acutely infected animals (7 dpi).(D) Principal component analysis of all RNA sequencing samples.Replicates of the same group are

Fig 6 .
Fig 6.Virus control by CD8 + T cells that lack NFAT signaling.(A) 10 4 naïve OTI T cells were transferred to congenic wildtype animals and activated by MCMV infection.Representative plots show IFNγ and TNFα staining of OTI T cells (FACS plots) and bar plots show IFNγ + or IFNγ + TNFα + OTI T cells isolated from spleen at 7 dpi.Error bars show mean ± SEM, data pooled from 2 experiments.(B-C) 10 5 naïve OTI T cells were adoptively transferred to RAG2gc KO animals one day prior to MCMV infection with 10 5 PFU.(B) Animal body weight was monitored following MCMV infection and percent of initial body weight is plotted as mean ± SEM.Animals from control group that lost more than 20% of initial body weight were sacrificed, these are denoted by † sign.(C) Animals were sacrificed 14 dpi and different organs were collected for virus titration.Virus was titrated from spleen, liver, lungs and salivary glands on mouse embryo fibroblast cells.For control group the values correspond to virus titer at 14 dpi or at the time of death.Bars show median value and each dot represents one mouse.Data pooled from 2 experiments.Statistically significant differences are highlighted; *, p < 0.05; **, p < 0.01; (Mann-Whitney U Test). https://doi.org/10.1371/journal.ppat.1012025.g006 The responsible state office (Lower Saxony State Office of Consumer Protection and Food Safety) approved all animal experiments performed in Helmholtz-Zentrum fu ¨r Infektionsforschung, Braunschweig under permit no.33.19-42502-04-16/2342 and 33.19-42502-04-18/2897.