The Impact of KLF2 Modulation on the Transcriptional Program and Function of CD8 T Cells

Ethics Statement

All mice used were bred and maintained under specific pathogen-free conditions in the Biological Resource Unit at the University of Dundee. The procedures used were approved by the University Ethical Review Committee and authorised by a project licence under the UK Home Office Animals (Scientific Procedures) Act 1986.

Mice and cell culture

Mice used were P14 TCR transgenic mice with a TCR recognising the gp33-41 peptide of LCMV (KAVYNFATM) and OT-1 transgenic mice, which express a TCR recognising the ovalbumin derived peptide SIINFEKL. Naïve CD8 T cell were negatively selected from lymph nodes using AutoMacs (Miltenyi Biotechnology). 1ng/milliliter of SIINFEKL or 10ng/ml of the altered peptide ligands (SIIQFEKL, SIIQFEHL, SIIQFERL), were used for OT-1 TCR triggering. P14 CTL were produced by activating splenocytes for 2 days with 100ng/ml gp33-41 peptide, then washed and cultured with IL-2 (20ng/ml, Novartis) for a further 3-5 days unless otherwise stated. Where indicated, rapamycin (Calbiochem) was used at 20nM.

Retroviral transduction

The GFP control plasmid construct was made by PCR amplification of EGFP from a pEGFP construct (Clontech) and cloning into the pBMN-LZRS vector (Addgene) as a HindIII/NotI fragment, replacing the lacZ gene. KLF2 cDNA was PCR amplified and cloned into pEGFP-C1 as a HindIII/BamHI fragment. The GFP-KLF2 construct was made by PCR amplification of EGFP-KLF2 and cloning into pBMN-LZRS as an EcoRI fragment, replacing the lacZ gene. The GFP-FoxOAAA has been previously described[25]. Phoenix ecotropic packaging cells[26] were transfected with plasmid using calcium phosphate transfection. Virus was harvested and used to transduce T cells as previously described[22]. Transduced activated CD8 T cells were generated by activation with gp33-41 peptide, retroviral transduction at 18 hours post activation, washing at 48 hours post activation and followed by 2 days culture with IL-2 in all experiments.

Flow Cytometry and Cell Sorting

Cell counts were performed with Caltag counting beads (Invitrogen) according to the manufacturer’s instructions. The following antibodies were used for staining: CD8-FITC, CD62L-APC (BD Pharmingen) and CXCR3-PerCPCy5.5 (eBiosciences). Cellular DNA content was measured using Hoechst 33342 (Molecular Probes). DNA synthesis was measured using the Clickit-EdU kit (Invitrogen) according to the manufacturer’s instructions. CTL were set up at 5x10⁵/ml and incubated with 10µM EdU for 30 minutes before fixation and staining. Data were collected on FACS Calibur and LSR Fortessa machines (Beckton Dickinson) and analyzed using FlowJo software (Treestar). Fluorescence Activated Cell Sorting (FACS) was performed on a FACS Vantage Cell Sorter (Beckton Dickinson).

Real-Time PCR

RNA was extracted from CTL or naïve cells using the RNeasy Minikit (Qiagen) and used to make cDNA using the cDNA synthesis kit (Quanta Biosciences). Quantitative real time PCR was performed on an iQ5 (Bio-Rad) using SYBR Green Fastmix (Quanta Biosciences). Results were normalised to expression of HPRT or CD8. For KLF2 quantification in naive and retrovirally transduced cells a standard cDNA of known concentration was used to generate a standard curve against which samples were measured. Primers used were as follows: KLF2 forward 5’-TGTGAGAAATGCCTTTGAGTTTACTG-3’, reverse 5’-CCCTTATAGAAATACAATCGGTCATAGTC-3’, KLF2 cDNA forward 5’-GCGCTCTGACGAGCTTACC-3’, reverse 5’-ACATGTGTCGCTTCATGTGC-3’, CD62L forward 5’-ACGGGCCCCAGTGTCAGTATGTG-3’, reverse 5’-TGAGAAATGCCAGCCCCGAGAA-3’, S1P1 forward 5’-GTGTAGACCCAGAGTCCTGCG-3’, reverse 5’-AGCTTTTCCTTGGCTGGAGAG-3’, CXCR3 forward 5’-GCCAAGCCATGTACCTTGAG-3’, reverse 5’-GTCAGAGAAGTCGCTCTCG-3’, Il6ra forward 5’-GTCACGGGCACTCCTTGGATAG-3’, reverse 5’-AGGAATGTGGGCAGGGACATGG-3’, Serpinb9 forward 5’-CAGATGAGGGTGTGGACCTCAG-3’, reverse 5’-TCCCAAGCGCTGAAACAGAGAC-3’, T-bet forward 5’-GATCATCACTAAGCAAGGACG-3’, reverse 5’-GGTCCACCAAGACCACATC-3’, Perforin forward 5’-CGTCTTGGTGGGACTTCAG-3’, reverse 5’-GCATTCTGACCGAGTGGCAG-3’, Interferon gamma forward 5’-TTACTGCCACGGCACAGTC-3’, reverse 5’-AGATAATCTGGCTCTGCAGG-3’, c-myc forward 5’-CCACCAGCAGCGACTCTG-3’, reverse 5’-GAGATGAGCCCGACTCCG-3’, HPRT forward 5’-TGATCAGTCAACGGGGGACA-3’, reverse 5’-TTCGAGAGGTCCTTTTCACCA-3’, CD8 forward 5’-GATATAAATCTCCTGTCTGCCCATC-3’, reverse 5’-ATTCATACCACTTGCTTCCTTGC-3’

Chromatin Immunoprecipitation

Chromatin Immunoprecipitation was performed as previously described[25] using anti-Pol II (N-20, Santa-Cruz Biotechnology) antibody or normal rabbit IgG (Cell Signalling). The immunocomplexes were collected using chip Grade Protein G Magnetic beads (Cell Signalling) in presence of 0.2mg/ml BSA. Immunoprecipitated chromatin was amplified by Real-Time PCR as above. The primers used were as follows:

KLF2 upstream promoter (-1500bp upstream TSS) forward 5’-TCAGCCTCACATTCTGCCTC-3’, reverse 5’- GCAGAAACATTGGCGAACTAC-3’

Microarray samples and analysis

RNA was prepared as above from cells purified by FACS and samples were forwarded to the Finnish Microarray and Sequencing Center (Turku) for hybridization on Affymetrix 3’ IVT expression chips (mouse genome 430 2.0). The data were normalised and expression measures computed using the Robust Multiarray Average (RMA) method[27]. Hierarchical clustering of the normalised intensity probe-sets confirmed that the biological replicates accurately reflected distinctly identifiable data groups. Simple inter-replicate expression plots, where probe intensities for a given replicate are plotted against the probe intensities for the other replicates in the group, were used to ensure that no large scale systematic effects are present in the data. A linear model was fitted to the normalised intensity data with fixed constants for each replicate group using the R package ‘limma’.[28] P-values for the fold-changes calculated from the linear model are calculated by shrinkage of the empirical Bayes moderated t-statistic for each probe and are then adjusted for multiple hypothesis testing by controlling the False Discovery Rate (using the correction detailed by Benjamini and Hochberg[29]) to produce q-values for linear model analysis. Probe-sets with q-values <0.046% (2σ) were considered to be significantly regulated. Regulation of at least 2-fold was used as a threshold cutoff.

Data are available at: http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE26568

Transwell chemotaxis assay

CTL transduced with GFP-KLF2 retrovirus were competitively assayed with co-cultured untransduced CTL for their ability to migrate towards a CXCL10 gradient. Briefly, transwell plates (Corning) were coated with Fibronectin (Sigma) and CXCL10 (R&D Systems) was diluted in media to the relevant concentration. CTL were then added and input controls taken. After 4 hours incubation cell count and GFP percentage were analyzed by flow cytometry and compared to input controls.

Western Blotting

Cells were lysed at 3x10⁷/ml in 10mM Tris pH7.05, 50mM NaCl, 50mM NaF, 5μM ZnCl₂, 1mM DTT, 10% glycerol and 0.5% Triton X100 with complete protease inhibitors (Roche). The resultant lysates were separated by SDS-PAGE, transferred onto nitrocellulose membrane and detected by Western blotting using standard techniques. Antibodies used recognised Spi6 (Hycult), GSK3α/β, pT308 PKB, pan PKB, pT24/32FoxO1/3A (Cell Signaling Technology) and pan FoxO1 (Dundee Signal Transduction & Therapeutics).

Statistical Analyses

Data sets were analysed using SigmaPlot v11.0 (Systat Software Inc., USA). Comparisons between two groups were made using Student’s t test or a non-parametric Wilcoxon Rank-Sum test where appropriate. Comparisons between multiple groups were made using one way analysis of variance (ANOVA) test. Levels of significance are denoted as follows: * p<0.05, ** p<0.01, *** p<0.001. Non-significant results are either not marked or indicated ns.

KLF2 expression is determined by the strength of T cell antigen receptor and cytokine signalling

Previous studies have shown that KLF2 is highly expressed in naive T cells but rapidly downregulated at the protein and mRNA level in response to triggering of the T cell antigen receptor[6–8]. The kinetics of this response are shown in Figure 1A. In these experiments, P14-TCR transgenic CD8 T cells were triggered with a peptide of the lymphocytic choriomeningitis virus (LCMV), glycoprotein gp33-41, presented by the MHC class I molecule H-2D^b. These data show that KLF2 mRNA loss can be detected within 1 hour of triggering the TCR complex and is complete within 6 hours (Figure 1A). The gp33-41 LCMV peptide is a high affinity TCR ligand for the P14 TCR yet during a polyclonal immune response there will be a pool of T cells with diverse affinities for the same antigen. To study how TCR affinity for antigen impacts on KLF2 expression we compared KLF2 levels in OT-1 TCR transgenic CD8 T cells triggered with four altered peptide ligands (APL) derived from the original OT-1 ligand SIINFEKL (N4)[30]. These APL bind equally well to MHC as N4 but differ in their affinity for the OT-1 TCR[30]. Figure 1B shows that the extent of KLF2 downregulation is determined by the potency of the TCR ligand; high affinity ligands strongly reduce KLF2 levels whereas weak TCR ligands are much less effective. The expression of mRNA of two well described KLF2 target genes CD62L and S1P1 is correspondingly decreased by TCR activation with the OT-1 APLs (Figure 1C).

The expression of KLF2 is also modulated by the strength and duration of cytokine signalling. It has been shown previously that IL-2, which promotes CTL differentiation[10,11], maintains low levels of KLF2 in antigen primed T cells[7]. Importantly, IL-2 mediated loss of KLF2 is dependent on continual IL-2 signalling: if CTL are deprived of IL-2 they will re-express KLF2 mRNA within 2 hours and KLF2 mRNA is further increased by 24 hours (Figure 1D). Moreover, the level of KLF2 expression in CTL is also determined by the strength of IL-2 signalling with high concentrations of IL-2 having a stronger suppressive effect on KLF2 expression than low concentrations (Figure 1E).

The loss of KLF2 mRNA in activated T cells could be due to reduced transcription of KLF2 mRNA and/or reflect that TCR triggering decreases KLF2 mRNA stability. To explore these options we compared the impact of TCR engagement on the amount of KLF2 spliced and unspliced RNA. Non-spliced RNA was detected using primers directed to the second intron of the KLF2 transcript whereas spliced RNA was measured using primers spanning the second intron (Figure 2A). Using the OT-1 TCR model we show that both unspliced and spliced KLF2 RNA decreases approximately equally, proportional to baseline, in T cells activated with the different TCR ligands (Figure 2B). The reduction in expression of mature spliced KLF2 mRNA in TCR activated cells could reflect that there is less unspliced KLF2 RNA to be processed in TCR activated T cells compared to naïve T cells, indicating a transcriptional control mechanism. To confirm this finding, we examined the recruitment of RNA polymerase II (Pol II) to the KLF2 gene. We looked at two regions, the transcriptional start site (TSS) and a downstream region within the 3’ untranslated region of the KLF2 gene. Primers directed against a sequence 1500 base pairs upstream of the TSS were used as a negative control. These data (Figure 2C) show that the recruitment of Pol II to the KLF2 locus is strikingly reduced in TCR activated T cells compared to control cells. These data indicate that TCR engagement reduces transcription of the KLF2 gene by dramatically reducing the recruitment of Pol II to the KLF2 locus.

KLF2 expression in naïve T cells is dependent on expression of FoxO1. Hence FoxO1 null naïve T cells do not express KLF2 and have a corresponding loss in expression of CD62L and corresponding defects in cell trafficking[19]. We have shown previously that the ability of IL-2 to downregulate KLF2 expression in CTL is reversed by PI3K or PKB inhibitors[25]. We therefore considered that TCR activation of PKB might mediate TCR induced loss of KLF2. To test this hypothesis purified naïve cells from P14 mice were activated with gp33-41 peptide for 4 hours in the presence or absence of either the PKB inhibitor Akt VIII or the PI3K inhibitor IC87114. In these experiments, IC87114 and Akt VIII inhibited TCR induced phosphorylation of PKB and FoxO1 (Figure 2D). However, surprisingly, both inhibitors failed to prevent the TCR-induced downregulation of KLF2 mRNA (Figure 2E). These results argue that the TCR downregulates expression of KLF2 independently of its ability to control FoxO1 phosphorylation. To directly test this hypothesis we transduced CD8 T cells with the non-phosphorylatable FoxOAAA mutant. Here, all of the PKB sites responsible for FoxO 14-3-3 protein binding and hence the nuclear exclusion seen on TCR activation are mutated to alanine. Thus, the FoxOAAA protein will remain nuclear despite activation of PKB through TCR stimulation. As predicted, expression of FoxOAAA increases expression of KLF2 and its target genes CD62L and S1P1 (Figure 2F). However, the salient observation is that the activation of the TCR still results in a rapid and dramatic downregulation of KLF2 and its target genes. The TCR can thus downregulate expression of KLF2 independently of its ability to activate PKB and induce FoxO phosphorylation.

In Figure 1B we showed that the magnitude of KLF2 downregulation is graded according to the avidity of the different OT-1 TCR peptide ligands. The magnitude of KLF2 downregulation triggered by the different OT1 ligands correlates with the ability of these APLs to activate the p42/p44 MAP kinase (ERK1/2) pathway downstream of the TCR[30]. We therefore explored the role of the MAPK pathway in downregulating KLF2 expression. PD184352 is a well characterised allosteric inhibitor that prevents activation of MEK1, the kinase which phosphorylates and activates ERK1/2[31]. The inhibition of ERK1/2 activation with PD184352 partially blocked TCR mediated downregulation of total KLF2 mRNA and is sufficient to entirely block unspliced KLF2 mRNA downregulation (Figure 2G). Intriguingly, the combination of both MEK and PKB inhibition is required to completely block TCR mediated downregulation of KLF2 and its target gene CD62L (Figure 2G).

The importance of downregulating KLF2 in T cell activation

KLF2 drives expression of CD62L, an adhesion receptor essential for T cell transmigration from the blood into secondary lymphoid tissues, and S1P1, a lysosphingolipid receptor that controls T cell egress from thymic and peripheral lymphoid tissue[14–16]. KLF2 downregulation in response to triggering of antigen and cytokine receptors thus explains why immune activated effector cells lose expression of these key trafficking molecules. However, is KLF2 downregulation required for any other key processes of CD8 T cell activation? Note the impact of KLF2 loss in naïve T cells has been assessed previously[3,4,17] but no experiment has fully explored the importance of the loss of KLF2 expression that accompanies immune activation. The experimental strategy to address this question requires the genetic manipulation of T cells to prevent them downregulating expression of KLF2 in response to immune activation. Accordingly, we decided to use Affymetrix microarray analysis to examine the transcriptional profile of immune activated T cells that were retrovirally transduced to maintain expression of KLF2 mRNA to the levels seen in quiescent naïve T cells (Figure 3A, 3B). In these experiments we compared control activated CD8 T cells versus activated CD8 T cells transduced with a retrovirus encoding GFP-KLF2 fusion protein (Figure 3A). We could analyse four different populations from these experiments; cells transduced with either the GFP-KLF2 construct or with an empty vector GFP construct or those negative cells from the same cultures. This experiment allows an assessment of the importance of KLF2 downregulation for the immune activation of CD8 T cells. For clarity, we refer to these activated CD8 T cell populations throughout the text as GFP-KLF2^pos, GFP^pos, GFP-KLF2^neg and GFP^neg respectively (Figure 3A).

Using this system we compared by microarray activated CD8 T cells with normal downregulation of KLF2 with activated CD8 T cells where KLF2 expression was maintained. Approximately 10000 annotated genes were expressed in activated CD8 T cells and the impact of KLF2 re-expression was to upregulate 492 genes and downregulate 164 genes by at least twofold. The full list of genes regulated by KLF2 is detailed in Table S1. It should be noted that KLF2 is not detected as a regulated gene since the GFP-KLF2 construct lacks the 3’ UTR of KLF2, which is the sequence detected by probes in the microarray. The microarray data corroborated previous data showing that maintaining KLF2 expression prevents activated T cells downregulating expression of CD62L and S1P1 (Figure 3C). To verify these data we analysed GFP-KLF2^pos cells for expression of CD62L and S1P1 mRNA (Figure S1A) and CD62L surface protein expression (Figure S1B). The results show that a failure to downregulate KLF2 sustained expression of CD62L and S1P1 in activated CD8 T cells. CD62L expression is frequently used to distinguish terminally differentiated effector CTL from naïve or memory T cells. We therefore examined the impact of maintaining KLF2 expression on other key genes that determine CTL effector function (Figure 3D-F). These analyses found that KLF2 expression did not prevent immune activated T cells acquiring expression of cytolytic effector molecules such as the granzymes (Figure 3D), nor did it prevent expression of key transcription factors that control CTL differentiation such as STAT5 or eomesodermin (Figure 3F). On the contrary, the expression of mRNA encoding IFNγ and perforin were increased by maintaining KLF2 expression (Figure 3D).

Effector T cells can be distinguished by their cytokine receptor profiles. Notably effector T cells downregulate expression of receptors for IL-6 and IL-7 and are CD27 low. They do however express high levels of pro-inflammatory cytokine receptors, notably IL-12 receptors. In this context, maintaining KLF2 expression in activated CD8 T cells had no effect on expression of CD27 or IL-7 and IL-12 receptors but strikingly increased expression of mRNA encoding the IL-6 receptor alpha subunit (Figure 3E). This latter result was verified by qRT-PCR and flow cytometric analysis (Figure S1C, S1D). One other observation was that immune activated T cells that maintained KLF2 expression expressed much higher levels of mRNA encoding the Serpin family of serine protease inhibitors, notably Serpinb9 (Figure S1E). This is a serine protease inhibitor of granzyme B that controls cytotoxic granule integrity in effector CTL and is expressed at high levels in memory T cells[32,33]. KLF2 upregulation of Serpinb9 (Spi6) protein was verified by western blot analysis (Figure S1F). Moreover, qRT-PCR analysis confirmed a cell autonomous effect of KLF2 on Serpinb9 expression (Figure S1E). Collectively, these data show that KLF2 controls expression of a selective portion of the CD8 T cell transcriptional program but persistent KLF2 expression post activation is not sufficient to force the CD8 T cell transcriptional program to remain that of a naïve T cell.

KLF2 regulates CXCR3 expression in CD8 T cells

From the microarray analysis, we noted that a failure to downregulate KLF2 prevented upregulation of the inflammatory chemokine receptor CXCR3 in immune activated CD8 T cells (Figure 3C). Previous studies have shown that KLF2 loss can cause thymocytes to express CXCR3 but this is an indirect consequence of unrestrained cytokine production by KLF2 null thymocytes[17]. It was thus striking that KLF2 control of CXCR3 in activated CD8 T cells was cell autonomous. Hence CXCR3 mRNA expression is low in GFP-KLF2^pos activated CD8 T cells but high in co-cultured GFP-KLF2^neg activated CD8 T cells (Figure 4A). The cell autonomy of KLF2 control of CXCR3 was also confirmed by flow cytometry analysis (Figure 4B) that showed lower surface levels of CXCR3 on GFP-KLF2^pos activated CD8 T cells compared to the co-cultured GFP-KLF2^neg cells. Importantly, the downregulation of CXCR3 expression caused by KLF2 had a cell autonomous impact on the chemotaxis of activated CD8 T cells in response to the inflammatory chemokine CXCL10. GFP-KLF2^neg activated CD8 T cells thus show a good chemotactic response to CXCL10 but this response is lost in GFP-KLF2^pos activated CD8 T cells (Figure 4C). These data show that KLF2 downregulation is essential for induction of CXCR3 expression in activated CD8 T cells. Consistent with this idea, the OT-1 APLs induce CXCR3 expression in a manner inversely correlating with KLF2 expression (Figure 4D). Furthermore, the use of a MEK inhibitor, which partially blocks KLF2 downregulation (Figure 2E), is sufficient to reduce TCR mediated induction of CXCR3 (Figure 4E).

KLF2 downregulation is required for proliferation of activated CD8 T cell proliferation

Ectopic expression of KLF2 can drive T cell quiescence[12,13,16] but it is unclear whether KLF2 expression levels achieved in previous experiments were in the physiological range[8]. In this context, it has been shown that the loss of KLF2 is not sufficient for T cell proliferation nor is KLF2 essential to initiate T cell quiescence[8]. In the current experiments we maintained KLF2 mRNA expression in immune activated CD8 T cells to levels equivalent to that found in naive T cells (Figure 3B) and found no major differences in the cellular DNA content of GFP-KLF2^pos activated CD8 T cells versus GFP-KLF2^neg activated CD8 T cells: both populations contained high frequencies of cells in the proliferative S and G2 phases of the cell cycle (Figure 5A). In particular there was no evidence that sustained KLF2 expression caused cell cycle arrest in activated T cells. Previous studies have indicated that KLF2 induces T cell quiescence because KLF2 downregulates expression of c-myc and upregulates the cell cycle inhibitor p21 ^cip1[12,13]. However, the current microarray analysis did not detect any KLF2 regulation of either of these genes (Table S1). Analysis by qRT-PCR (Figure 5B) shows that c-myc mRNA levels are comparable in control and GFP-KLF2^pos activated CD8 T cells.

The cell cycle analysis (Figure 5A) shows that immune activated T cells that do not downregulate KLF2 can enter the cell cycle. However, a quantitative assessment of rates of DNA synthesis (Figure 5C) revealed that GFP-KLF2^pos activated CD8 T cells have a lower rate of DNA synthesis compared to control activated CD8 T cells that have terminated KLF2 expression. These data predict that activated CD8 T cells with high levels of KLF2 will take longer to complete a cell cycle. To explore this hypothesis we compared the rate of cell doubling of GFP-KLF2^pos versus GFP^pos activated CD8 T cells. These data revealed that T cells that maintained KLF2 expression had a reduced rate of T cell proliferative expansion (Figure 5D). Further interrogation of the microarray results revealed that maintaining KLF2 expression leads to an increase in expression of the cyclin dependent kinase inhibitors Cdkn2a, Cdkn2b and Cdkn2d, which could underpin this reduction in clonal expansion. Collectively, the results indicate that KLF2 is not sufficient to drive cells into a quiescent G0/G1 state but KLF2 expression can restrain T cell proliferative expansion.

KLF2 has threshold dependent functions

The current results show that maintaining KLF2 expression can restrain CD8 T cell proliferation. However, there are discrepancies in the literature and indeed in the present data that question an obligatory link between the downregulation of KLF2 expression and the induction of T cell proliferation. For example, weak TCR ligands such as the OT-1 peptide Q4H7 do not effectively terminate expression of KLF2 (Figure 1B) yet are able to initiate T cell proliferation[34]. Furthermore, the mTORC1 inhibitor and immunosuppressant rapamycin has been shown to induce KLF2 re-expression and cause re-expression of CD62L and S1P1 in activated CD8 T cells but rapamycin does not inhibit proliferation[7,35]. However, as we have shown, expression of KLF2 in T cells is not simply on or off but can be graded (Figure 1). It has thus been hypothesised that the magnitude of KLF2 expression might dictate the functional outcome in terms of T cell proliferation[8]. In this context although it has been shown that mTOR inhibitors can increase KLF2 expression in activated CD8 T cells it has never been explored whether rapamycin treatment restores KLF2 expression to the levels seen in naïve T cells. The experiment depicted in Figure 6A was designed to explore this issue and compares KLF2 expression in naïve CD8 T cells to activated CD8 T cells cultured in IL-2 in the presence or absence of rapamycin. The data show that rapamycin increases the expression of KLF2 in activated CD8 T cells but to levels far below that seen in naïve cells (Figure 6A). Importantly, mTOR inhibition with rapamycin is sufficient to induce expression of the KLF2 target gene CD62L (Figure 6B) and to repress CXCR3 expression both at the mRNA (Figure 6C) and protein levels (Figure 6D). However, mTOR inhibition with rapamycin does not suppress DNA synthesis in activated CD8 T cells (Figure 6E).

These results generated a hypothesis that retaining low levels of KLF2 in activated T cells would be sufficient to modulate the CD8 T cell trafficking program yet only high levels of KLF2 similar to those found in naïve T cells would inhibit proliferation. To test this hypothesis we initially examined CD62L expression and DNA synthesis in immune activated CD8 T cells transduced with low, intermediate or high levels of GFP-KLF2 (Figure 7A). Strikingly, the data show that low levels of KLF2 were sufficient to induce maximal expression of CD62L (Figure 7B, 7C) but only very weakly suppressed DNA synthesis (Figure 7D). We also noted that low levels of KLF2 were sufficient to maximally repress expression of CXCR3 (Figure 7E). This argues that KLF2 has different functional consequences depending on its level of expression, with low levels being sufficient to redirect homing and high levels required to inhibit proliferation.