Adenoviral Vector Vaccination Induces a Conserved Program of CD8+ T Cell Memory Differentiation in Mouse and Man

Summary Following exposure to vaccines, antigen-specific CD8+ T cell responses develop as long-term memory pools. Vaccine strategies based on adenoviral vectors, e.g., those developed for HCV, are able to induce and sustain substantial CD8+ T cell populations. How such populations evolve following vaccination remains to be defined at a transcriptional level. We addressed the transcriptional regulation of divergent CD8+ T cell memory pools induced by an adenovector encoding a model antigen (beta-galactosidase). We observe transcriptional profiles that mimic those following infection with persistent pathogens, murine and human cytomegalovirus (CMV). Key transcriptional hallmarks include upregulation of homing receptors and anti-apoptotic pathways, driven by conserved networks of transcription factors, including T-bet. In humans, an adenovirus vaccine induced similar CMV-like phenotypes and transcription factor regulation. These data clarify the core features of CD8+ T cell memory following vaccination with adenovectors and indicate a conserved pathway for memory development shared with persistent herpesviruses.


Correspondence beatrice.bolinger@unibas.ch
In Brief Bolinger et al. define the transcriptional program associated with sustained CD8 + T cell memory induced by adenoviral vaccination. They relate these to memory ''inflation'' induced following infection by persistent CMVs. Core features are found to be shared by mouse and man, including a prominent role for TBX21.

INTRODUCTION
After viral infection, naive antigen-specific CD8 + T cells clonally expand and differentiate. Massive proliferation and differentiation to effector T cells is coupled to changes in homing, function, and gene expression, leading to CD8 + T cell memory. Defining mechanisms that drive effective CD8 + T cell memory pools is critical for the design of vaccines.
Broadly, two subsets of memory CD8 + T cells are described: central memory CD8 + T cells (T CM ) and the effector memory CD8 + T cells (T EM ) (Sallusto et al., 1999). ''Central'' memory pools are typically contracted memory populations that express lymph-node-homing markers (CD62L and CCR7). ''Effector memory'' subsets lack these and are found distributed in tissues, e.g., lung and liver. These pools are linked to infection with persistent viruses, the best examples being human and murine cytomegaloviruses (HCMV and MCMV).
A characteristic of the CMV immunobiology is the induction of an expanded, sustained effector-memory T cell response to specific epitopes, a phenomenon termed CD8 + T cell ''memory inflation'' (Karrer et al., 2003). In parallel, classic ''non-inflating,'' central memory responses develop against many epitopes. Molecular profiling of CMV-specific CD8 + T cells in humans revealed that the development of CMV-specific CD8 + T cells is a dynamic process, with key features of the HCMV-specific CD8 + T cell phenotype installed early after infection (Hertoghs et al., 2010).
CD8 + T cell memory induced by vaccines could provide protection against complex pathogens. Whereas CMV-based vectors show promise in studies of SIV infection (Hansen et al., 2011), such viruses are complex. One technology that has shown potency in generation of antiviral T cell pools in clinical studies is based on replication-deficient adenoviral vectors. Several trials have indicated such vectors are safe and can induce substantial immune responses against pathogens such as HCV Colloca et al., 2012).
Studies of vaccine-induced T cell responses in a murine model using a recombinant replication-deficient HuAd5 vector expressing lacZ (Ad-lacZ) (Bolinger et al., 2013) revealed two distinct pathways for memory-an inflationary response to one epitope and a typical contracting response to a second epitope. The sustained response showed phenotypic features typical of effector memory and was enriched in tissues, whereas the reverse was true for the contracting response. Figure 1. Frequency, Function, and Gene Expression Signature of MCMV-Specific CD8 + T Cells C57BL/6 mice were infected intravenously (i.v.) with 1 3 10 6 pfus MCMV. (A) Tetramer staining for M38-and M45-specific CD8 + T cells on 7 and 50 days postinfection in spleen. Mean percentages of live tetramer + CD8 + T cells are indicated (n = 8; mean ± SEM). (B) Time course for M38-(red) and M45-(blue) specific CD8 + T cells. Splenocytes from 0, 7, 21, 50, and 100 days postinfection mice were stained with tetramers and analyzed by flow cytometry. Mean percentages of live tetramer + CD8 + lymphocytes are indicated (n = 8; mean ± SEM).
Because these peptide epitopes are both derived from the same expressed transgene, this model provides a controlled system for analysis of two divergent vaccine-induced memory pools.
Here, we define the transcriptional changes in MCMV infection and Ad-LacZ vaccination and addressed to what extent parallel changes can be observed in human memory pools induced by CMV and adenoviral vectors. Our data clearly show that a subset of stable memory CD8 + T cell responses, whether induced by vaccine vectors or natural virus, in mouse and man, display a common molecular profile divergent from that of acute effector, central memory, and exhausted CD8 + T cells.

Two Functional and Transcriptionally Distinct Memory
Patterns of MCMV-Specific CD8 + T Cells To establish a data set for gene expression in CD8 + T cell memory pools in the setting of a persistent infection, we analyzed the wellcharacterized model of MCMV infection. This has the advantage of a parallel human data set for comparison (Hertoghs et al., 2010). Infection of C57BL/6 mice with MCMV resulted in two distinct CD8 + T cell responses, the conventional (non-inflationary) and the expanded (inflationary) CD8 + T cell response in blood, spleen, and organs (liver or lung; Figures 1A, 1B; Figure S1A). An analogous profile was observed when M45-and M38-specific CD8 + T cells were analyzed for IFNg and TNFa production ( Figures  1C and 1D). Furthermore, the expression of chemokines XCL1, CCL3, CCL4, CCL5, and CCL9 ( Figure 1G) and LAMP1 followed the same pattern. These results confirm previous data that, after MCMV infection, two distinct types of CD8 + T cell responses are induced and demonstrate that both types of CD8 + T cells retained polyfunctionality over time (Munks et al., 2006;Sierro et al., 2005).
To generate gene-expression profiles of MCMV-specific memory and effector CD8 + T cells, M38-and M45-specific CD8 + T cells from days 7 and 50 postinfection were tetramer sorted. As naive controls, sorted CD62L hi CD44 lo CD8 + T cells from naive C57BL/6 mice were used ( Figure S1B). We compared gene expression in M38-and M45-specific CD8 + T cells from day 7 and day 50 to naive CD8 + T cells. Considering genes which were >2-fold differentially expressed over naive CD8 + T cells (adjusted p < 0.01) revealed a total of 627 upregulated and 246 downregulated genes for the M45-specific population on day 7 postinfection and 400 up-and 168 downregulated genes on day 50 postinfection. For the early M38-specific population, a total of 964 genes were upregulated and 667 genes downregulated, rising to 1,088 up-and 553 downregulated genes for the late, inflating M38-specific CD8 + T cells ( Figure 1E; GEO: GSE73314).
Comparison showed that the expression profiles of the four populations differed markedly ( Figure 1F). M45-specific CD8 + T cells downregulated a large number of genes from the acute phase as they entered the central memory phase and most closely resembled naive CD8 + T cells. In contrast, M38-specific CD8 + T cells maintained expression of many of their acute phase genes (1,324 shared genes), also acquiring inflation-specific genes, totaling overall the highest number of differentially expressed genes ( Figure 1F). These data confirm that MCMV, on the transcriptional level, clearly induced two distinct CD8 + T cell memory responses maintained in parallel.

Inflationary CD8 + T Cell Memory Induced by a Nonreplicating Adenovirus Vector
We next addressed the qualities of the vaccine-induced responses following administration of a replication-deficient bgalrecombinant adenovirus vector (Ad-LacZ) (Bolinger et al., 2013). Intravenous (i.v.) immunization of C57BL/6 mice with Ad-LacZ induced two types of CD8 + T cell responses to bgal. bgal 96 -and bgal 497 -specific CD8 + T cells displayed similar expansion on day 21 postimmunization, but at day 50, bgal 96 -specific CD8 + T cells had further increased to 13% of total CD8 + T cells, whereas bgal 497 -specific CD8 + T cells contracted (Figures 2A and 2B). Consistent with previous studies (Bolinger et al., 2013), these responses maintained functionality and were enriched in tissues such as lung and liver (data not shown).
To examine further the function and stability of vaccineinduced responses, we used an adoptive transfer approach. Cells from CD45.1 + mice were transferred at different time points after vaccination to CD45.2 + -vaccinated or naive mice and tracked using tetramers. The extended CD8 + T cell responses were sustained over 8 weeks in the recipients, even in the absence of antigen, maintaining their phenotype (CD44 hi , CD62L lo , CD27 lo , and CD127 lo ) and distributing to tissues such as liver and lung (data not shown). Overall, these features are consistent with those of a stable memory T cell pool.
To assess the T cell transcriptome of bgal-specific CD8 + T cells after Ad-LacZ immunization, a gene expression array analysis of tetramer-sorted bgal-specific CD8 + T cells from day 21 and day 100 and of naive CD8 + T cells (CD62L hi and CD44 lo ) was (C) Representative flow cytometry plots of splenocytes producing IFNg or TNFa stimulated with either M38 or M45 peptide in mice 7 and 50 days postinfection, gated on live lymphocytes. Numbers indicate the percentage of IFNg-and TNFa-positive CD8 + T cells (n = 8; mean ± SEM). (D) Time course plotting the percentage of M38-(red) or M45-(blue) specific CD8 + T cells from the spleen producing IFNg and TNFa after stimulation with either the M38 or M45 peptide. Mean percentage of IFNg and TNFa producing cells within the CD8 + T cell compartment is indicated (n = 8; mean ± SEM). (E) Number of genes upregulated (red) and downregulated (blue) in M38-and M45-specific CD8 + T cells 7 and 50 days postinfection, compared to naive CD8 + T cells. Filter criteria of at least 2-fold change with p % 0.05 compared to naive CD8 + T cells are shown. (F) Venn diagram showing the number of differentially expressed genes between M38-specific CD8 + T cells 7 (yellow) and 50 (red) days postinfection and M45specific CD8 + T cells 7 (orange) and 50 (blue) days postinfection. Filter criteria of at least 2-fold change with p % 0.05 compared to naive CD8 + T cells are shown. Upregulated genes are indicated in red and downregulated genes in blue. (G) Representative flow cytometry plots of splenocytes expressing LAMP1 or producing effector molecules stimulated with either M38 or M45 peptide in mice days 7 and 50 postinfection, gated on live lymphocytes (n = 8; ±SEM). Longitudinal flow cytometry analysis shows the percentage of M38-(red) or M45-(blue) specific CD8 + T cells from the spleen producing effector molecules after stimulation with either the M38 or M45 peptide. Mean percentages of LAMP1-, XCL1-, CCL3-, CCL4-, CCL5-, and CCL9-positive cells within the CD8 + T cell compartment are indicated (n = 8; ±SEM). See also Figure S1. performed. By only considering genes that were at least 2-fold differentially expressed (adjusted p < 0.01) compared to naive CD8 + T cells, we found a total of 937 up-and 403 downregulated genes for bgal 96 -specific (expanded) CD8 + T cells ( Figure 2C). By directly comparing the bgal 96 -specific CD8 + T cell population with the bgal 497 -specific memory population, it is evident that memory bgal 96 -specific CD8 + T cells showed a distinct expression profile-neither identical to effector CD8 + T cells nor memory bgal 497 -specific CD8 + T cells ( Figure 2C). These data confirm previous analyses indicating that two distinct memory pools are induced in parallel following Ad-lacZ immunization (Bolinger et al., 2013;GEO: GSE73314).

Combined Analysis Reveals a Conserved Expression Profile and Phenotype
Gene set enrichment analysis (GSEA) confirmed enrichment of the significantly up-and downregulated genes of the inflating bgal 96and M38-specific CD8 + T cells ( Figure S2A). Overall, these comparisons revealed that M38-and bgal 96 -specific inflating CD8 + T cells shared 663 upregulated and 290 downregulated genes (compared to naive CD8 + T cells; Figure S2B). We analyzed the transcriptional changes in the MCMV-induced and vaccineinduced memory pools, focusing on the cell cycle, apoptosis, TFs, and receptors for cytokines and chemokines, similar to an analysis of human CMV infection (Hertoghs et al., 2010).
For cell-cycle-related genes, the highly upregulated genes in the expanded populations in both models included MKI67 (Ki-67), a marker of proliferation. Its expression was confirmed by FACS, and both models showed a similar profile. By day 100, the levels of proliferation in the expanded population had dropped to a similar level to those in the contracted memory ( Figures  3A and 3B).
For cell survival, whereas expanded T cells maintained expression of genes seen in acute infection, contracted memory exhibited a similar expression profile to that of naive CD8 + T cells ( Figure 3C). For example, FACS analysis of the antiapoptotic protein Bcl-2 demonstrated downregulation in the acute phase in all responses but recovered at later time points only in the classical memory pools ( Figure 3D).
''Tuning'' of function through inhibitory and activation receptors occurs on T cells. We analyzed CD27 and CD357 (Watts, 2005) and confirmed reduced expression of these stimulatory receptors on expanded CD8 + T cells (data not shown). Similarly, for both models, we extended previous data (Bolinger et al., 2013;Snyder et al., 2008) and showed overexpression of Klrc1 (NKG2A), Klrg1 (KLRG1), Klre1, Klra1 (Ly-49c), and Klrk1 (Nkg2d) in the expanded responses and unique expression of Ly49C mRNA, confirmed by flow cytometry analysis or qPCR (data not shown). Further analysis of inhibitory molecules associated with exhaustion (Blackburn et al., 2009;Wherry et al., 2007) revealed that the genes for PD-1, TIM3, CD160, LAG3, and BTLA were expressed after infection and immunization (data not shown). However, FACS analysis of PD1, TIM3, and BTLA showed high expression of these markers on days 7/21 followed by a gradual decrease in the memory phase to levels similar to naive CD8 + T cells (Figure 3E). Consistent with retained functionality, the expression of granzymes, such as Gzma, Gzmb, Gzmm, and Gzmk were upregulated in both models, compared to naive CD8 + T cells (data not shown).
Next, we further focused on the differential expression of mRNAs of TFs ( Figure 4C). T-bet and Eomes, critical for T cell memory formation, were upregulated on both specificities of CD8 + T cells in the acute phase: on M38-and bgal 96 -specific CD8 + T cells, Eomes then decreased to levels similar to that of naive CD8 + T cells ( Figure 4D). T-bet expression, however, showed divergence in the memory phase, being low on contracting CD8 + T cells and remaining high on expanded CD8 + T cells; after adenovector vaccination, Tbet remained high on 60% of the expanded cell population at day 200 versus <10% on contracted memory at that time point ( Figure S3A). Relative expression of other TFs such as Rog (repressor of GATA), Blimp-1 (Prdm1), Bcl-6, Id2, and Id3 were also consistent with this ( Figure S3B; data not shown).
To define distinctions between memory populations furtherand the role of TFs in driving these-we used principal-component analysis (PCA) ( Figure S4). These data demonstrated that all antigen-experienced cell types cluster separately from naive cells and, compared to the ''acute'' sample, the classic memory pools (M45/bgal 497 ) regress toward the naive cell position whereas the expanded (M38/bgal 96 ) populations further extended the distance. The PCA analysis was confirmed when only using significantly modulated TFs (post-ANOVA test; Figure S4), indicating that the principal components separating these memory populations derived from changes in a limited number of TFs.
Finally, many of these features of MCMV-induced and adenovector-induced memory cells closely resembled data from a study of HCMV infection (Hertoghs et al., 2010). Similar GSEA profiles were seen using expression sets based on MCMV infection (confirming a recent analysis; Quinn et al., 2015) and, strikingly, adenovirus vaccination ( Figure S4). Additionally, from GSEA analyses on the TF subset, we confirmed that TFs upregulated in HCMV (Hertoghs et al., 2010) are enriched in upregulated TFs post-MCMV or -adenovector, notably Tbet (ranked 1; Figure S4).

Preserved Phenotype, Function, and TF Expression in Human-Adenoviral-Vaccine-Induced Responses
To assess in depth whether such changes in T cell phenotype and in TF expression could be observed in human-adenovector-vaccine-induced responses, we analyzed further CD8 + T cells primed using a chimpanzee adenovirus strategy (ChAd3), specific for dominant HCV epitopes from NS3 (NS3 95 , HLA-A2 restricted and NS3 103 HLA-A1 restricted). We compared them to CMV-specific responses from the same individuals. We addressed (1) whether a CMV-like effector memory phenotype could be observed and (2) whether this was associated with a Tbet-high sustained memory subset.
First, using a CYToF data set (Swadling et al., 2014), we analyzed the distribution of effector-memory-associated phenotypic markers on CMV-and HCV-specific (i.e. vaccine-induced) cells in parallel. This demonstrated a shared phenotype and one distinct from T cell responses to the non-persistent pathogen influenza ( Figure 5A), most clearly seen in analysis of CD57. Overlapping populations of CD57 + CD127 low CD8 + T cell pools can be seen in both CMV-specific and adenovirus-induced responses. PCA revealed overlapping distribution of CMV-specific and vaccine-induced CD8 + T cell responses and distinct from those induced by influenza infection ( Figure 5B).
Next, we analyzed the expression of two TFs-Tbet and Eomes. In bulk CD8 + T cell populations, these showed associations with distinct memory pools, with highest Tbet levels in Tem and Temra pools ( Figure 5C). In antigen-specific cells, we identified similar patterns of Tbet and Eomes regulation in CMV-and vaccine-induced cells, with enrichment for Tbet + Eomes À cells in the HCV-specific vaccine populations (Figure 5D). This distribution was maintained when the virus-specific cells were analyzed following subdivision using CCR7/CD45RA expression ( Figure S5A) or within a single individual ( Figure S5B) or comparing populations that had received only adenoviral vaccination and those where there had been boosting using an MVA construct ( Figure S5C).
The data confirmed that the adenovirus-vaccine-induced responses in humans are dominated by effector memory populations. These are high in Tbet and relatively low in Eomes, similar to CMV-specific responses from the same individuals.

Memory T cells are considered long-lived cells that require little if
any antigen re-encounter for survival. However, continued lowlevel stimulation through persistence of antigen is common in vivo. We hypothesized the latter could underlie the ability of adenoviral vectors to promote T cell responses in humans. Here, by profiling CMV-specific and adenovirus-vaccine-specific memory CD8 + T cells, we identified patterns of CD8 + T cell evolution that are (1) conserved between infection and vaccination and (2) consistent between mouse and man.
Memory inflation has been shown to be independent of initial immunodominance in MCMV (Karrer et al., 2004), and inflationary epitopes are immunoproteasome independent-suggesting they are not presented on DCs (Hutchinson et al., 2011). The current model is that repetitive antigen presentation occurs on a non-classical antigen-presenting cell (APC), potentially in the vascular endothelium or a non-hematopoietic (nh) lymph node cell (Smith et al., 2014;Torti et al., 2011). A separate study with a recombinant adenovirus vector concluded that nhAPC might play a role in memory (Bassett et al., 2011).
The expanded CD8 + T cell pools downregulated the antiapoptotic marker Bcl-2 compared to conventional memory CD8 + T cells. However, the anti-apoptotic Bcl-X L was uniquely upregulated. Bag3, another anti-apoptotic molecule, was highly expressed, and Bim, a pro-apoptotic marker, was downregulated ( Figure 3C). Bcl-X L expression can be induced by CX3CR1 (Boehme et al., 2000), which was highly upregulated in inflation ( Figure 4B). Co-stimulation via 4-1BB has been shown to be critical for memory inflation (Humphreys et al., 2010); because 4-1BB also induces Bcl-X L (Stä rck et al., 2005), we hypothesize that both CX3CR1 and 4-1BB contribute to survival of CD8 + T cells.
CD8 + effector memory pools express high levels of T-bet and to a lower extent Eomes. T-bet expression is induced by TCR signaling and amplified by IL-12-and mTOR-mediated signals (Intlekofer et al., 2008). This program seems to be present in HCV-specific CD8 + T cells induced using ChAd3 in volunteers. Such pools show sustained effector-memory phenotype, analogous to hCMV-specific responses, and a T-bet hi /Eomes lo status, consistent with the mouse data. This is particularly resonant for HCV-specific responses, because this is the transcriptional pattern linked with clearance of virus (Paley et al., 2012). The frequencies of responses induced by vaccination in humans do not reach those seen in ''inflationary'' models; however, they do reflect frequencies seen with intramuscular as opposed to (optimized) i.v. inoculation, with the same phenotypic/transcriptional changes noted (Bolinger et al., 2013).
Overall, we have identified gene sets modulated in CD8 + T cells post-adenoviral vaccine compared to acute effector CD8 + T cells and to conventional memory CD8 + T cells. By comparing these to the ''inflating'' CD8 + T cells from MCMV-infected animals, we show the CD8 + T cell gene profile is distinct from that seen in exhaustion and conventional memory and sustains a memory development program likely linked with low-level antigen persistence. Such CD8 + T cells possess a conserved phenotypic and functional pattern, underpinned by a core set of TFs. We show that this program of memory induction is reproduced in chimpanzee-adenovirus clinical vaccine trials, with implications for immunity induced by such regimens in humans.

Ethics Statement
Mouse experiments were performed according to UK Home Office regulations (project license number PPL 30/2235 and 30/2744) and after review and approval by the local ethical review board at the University of Oxford.

Viruses
MCMV strain (Strain Smith; ATCC: VR194) was used and kindly provided by Professor U.H. Koszinoswki, Department of Virology, Max von Pettenkofer Institute. MCMV was propagated and titrated on NIH 3T3 cells (ECACC), stored at À80 C, and injected i.v. at a dose of 2 3 10 6 pfus.

Adenoviral Vector
Recombinant adenovirus expressed the bgal protein under the control of the human CMV promoter (Ad-LacZ; Krebs et al., 2005). Ad-LacZ was propagated on HER-911 cells and purified with the Vivapure AdenoPack 20 (Sartorius; Stedim Biotech). Virus titer was determined in a cytopathic effect assay (Krebs et al., 2005). Ad-LacZ was stored at À80 C in PBS and injected i.v. at a dose of 2 3 10 9 pfus.
Mice C57BL/6 mice were obtained from Harlan, kept under conventional conditions in individually ventilated cages, and fed with normal chow diet.

Flow Cytometry
Single-cell suspensions were generated from the indicated organs, and 1 3 10 6 cells were incubated with the indicated mAb at 4 C for 20 min. For PBL samples, erythrocytes were lysed with FACS Lysing Solution (BD PharMingen). Cells were analyzed by flow cytometry using a BD LSR II flow cytometer and FlowJo (Treestar), gated on viable leukocytes using the live/dead fixable near-IR dead cell stain kit from Invitrogen.

Intracellular Staining and Peptide Stimulation
Following surface staining, cells were fixed and permeabilized using the FOXP3 Fixation/Permeabilisation Kit from eBioscience. Cells were resuspended in permeabilization buffer containing the appropriate amount of antibody and incubated for 30 min at 4 C. For peptide stimulation, see Supplemental Experimental Procedures.

Construction of Tetrameric MHC Class I Peptide Complexes
MHC class I monomers complexed with M38 (H-2Kb), M45 (H-2Db), and bgal (H-2Kb) were produced as previously (Altman et al., 1996) and tetramerized by addition of streptavidin-PE (BD Bioscience) or streptavidin-APC (Invitrogen). Aliquots of 1 3 10 6 cells or 100 ml of whole blood were stained using 50 ml of a solution containing tetrameric class I peptide complexes at 37 C for 20 min followed by staining with mAbs.
Sorting Cells by MACS CD8 + T cells were isolated from spleens and single-cell suspensions prepared, resuspended in MACS buffer at a concentration of 10 7 cells/90 ml, and purified by negative selection using the CD8 + T cell selection kit/AutoMACS (Miltenyi Biotec).

Cell Sorting by MoFLO
For cell sorting, at least four spleens were pooled per sort using MoFlo (Beckman Coulter Genomics). Splenocyte samples depleted of erythrocytes and enriched for CD8 + T cells using MACS beads were stained with saturating concentrations of antibody and tetramer, incubated for 1 hr on a rolling platform at 4 C, washed, filtered, and resuspended. Sorted cells were pelleted at 9,000 rpm for 10 min and snap frozen using methanol and dry ice. Reanalysis of sorted cells indicated a purity of over 95%. , and BTLA on M38-(red) and M45-specific CD8 + T cells (blue; n = 6-8; ±SEM) and on bgal 96 -(red) and bgal 497 -specific (blue) CD8 + T cells in the spleen (n = 6-8; ±SEM). Histograms show expression of PD-1, TIM-3, and BTLA and are gated on live naive (green) M38/bgal 96 -specific (red) and M45/bgal 497 -specific CD8 + T cells (blue).

RNA Extraction and Microarray Analyses
Total RNA was isolated from sorted cells using the mirVana Kit (Ambion). RNA integrity was checked using Agilent 2100 Bioanalyzer, and samples with a RIN value of 8 or more were used for amplification. cDNA was synthesized and cRNA made using Illumina Total Prep RNA amplification Kit (Ambion). Samples were hybridized to Illumina MouseWG-6 v2.0. Filter criteria of at least 2-fold changes with p % 0.05, compared to naive CD8 + T cells, are shown. (D) Flow cytometry analysis showing the MFI of cells expressing the TFs EOMES and Tbet on M38-(red) and M45-specific CD8 + T cells (blue; n = 6-8; ±SEM) and on bgal 96 -(red) and bgal 497 -specific (blue) CD8 + T cells in the spleen (n = 5-9; ±SEM). Histograms show expression of EOMES and Tbet and are gated on live naive (green), M38/bgal 96specific CD8 + T cells (red), and M45/bgal 497 -specific CD8 + T cells (blue). See also Figures S3 and S4.
For microarray data analysis, see the Supplemental Experimental Procedures.

Statistical Analysis of Real-Time and Flow Cytometry Data
Unpaired two-tailed Student's test was used. Statistical data analysis was performed using Graph-Pad Prism version 5.0a for MACs (GraphPad Software).

Human Pentamer Staining and Flow Cytometry
Pentamer staining was as previously (Swadling et al., 2014). For staining details, see the Supplemental Experimental Procedures.

Mass Cytometry Using CyTOF
Detailed methodology and validation of the mass cytometry used here are described elsewhere (Newell et al., 2012). T cell phenotype and function were determined by mass cytometry analysis on frozen PBMCs from two volunteers receiving ChAd3-NSmut and MVA-NSmut. Samples were taken 14 weeks after MVA-NSmut boost vaccination. Two healthy controls each for CMV and FLU, previously shown to have a detectable T cell response by pentamer FACS staining, were also included. For PCA details, see Supplemental Experimental Procedures.