The JI
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     
 

This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Related articles in The JI
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Chappell, C.
Right arrow Articles by Jacob, J.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Chappell, C.
Right arrow Articles by Jacob, J.
The Journal of Immunology, 2006, 176: 4562-4572.
Copyright © 2006 by The American Association of Immunologists

DNA Methylation by DNA Methyltransferase 1 Is Critical for Effector CD8 T Cell Expansion1

Craig Chappell*, Caroline Beard{dagger}, John Altman*, Rudolph Jaenisch{dagger},{ddagger} and Joshy Jacob2,*

* Emory Vaccine Center, Department of Microbiology and Immunology, Emory University, Atlanta, GA 30329; and {dagger} Whitehead Institute for Biomedical Research and {ddagger} Biology Department, Massachusetts Institute of Technology, Cambridge, MA 02139


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
Transcriptional silencing mediated by DNA methylation is a critical component of epigenetic regulation during early embryonic development in animals. However, the requirement for DNA methylation during activation and differentiation of mature CD8+ T cells into effector and memory cells is not clear. Using cre-mediated deletion of DNA methyltransferase 1 (Dnmt1) at the time of CD8+ T cell activation, we investigated the obligation for maintaining patterns of DNA methylation during the generation of Ag-specific effector and memory CD8+ T cells in response to acute viral infection of mice with lymphocytic choriomeningitis virus. Dnmt1–/– CD8+ T cells failed to undergo the massive CD8+ T cell expansion characteristic of lymphocytic choriomeningitis virus infection, leading to >80% reductions in Ag-specific effector CD8+ T cells at the height of the response. Despite this, Dnmt1–/– CD8+ T cells efficiently controlled the viral infection. Interestingly, the number of Ag-specific Dnmt1–/– memory CD8+ T cells was moderately reduced compared with the reductions seen at day 8 postinfection. Our data suggest that ablation of Dnmt1 and subsequent DNA methylation affect the finite proliferative potential of Ag-specific CD8+ T cells with moderate effects on their differentiation to effector and memory CD8+ T cells.


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 Disclosures
 References
 
Exposure of the immune system to pathogenic infection often induces the establishment of long-lived CD4+ and CD8+ memory T lymphocytes capable of responding to secondary immunogens with enhanced kinetics and magnitude compared with their naive counterparts (1, 2, 3, 4, 5). CD8+ T cell differentiation during both thymic development and Ag-specific immune responses is mediated through a series of sequential cell fate decisions that include at each stage of differentiation increases in cell type-specific gene expression and concomitant silencing of genes no longer required for cellular identity. DNA methylation of CpG dinucleotides is a key component of transcriptional silencing and plays a critical role during both development and cellular differentiation.

In mammals, DNA methyltransferase 1 (Dnmt1)3 is responsible for replicating the pattern of CpG methylation from the parental DNA strand to the daughter strand during DNA synthesis (6, 7), thereby maintaining DNA methylation during cell replication. DNA methylation mediates its effects directly by blocking access of key transcription factors to their DNA binding sites, or indirectly via epigenetic control of their access to DNA through alterations in chromatin structure (8, 9). The current model linking DNA methylation to chromatin remodeling occurs through recognition of methyl-CpG by DNA methyl-binding proteins (i.e., MBD2 and MeCP2) that in turn recruit histone deacetylase activity and associated nucleosome remodeling proteins to form chromatin remodeling complexes (10, 11, 12, 13). These complexes directly influence the degree of DNA coiling, and act to reinforce the pattern of gene expression set by DNA methylation through establishment of transcriptionally silent heterochromatin.

Even though multiple studies have shown direct correlations between promotor methylation status and gene expression in the immune system (14, 15, 16, 17, 18), the extent to which Dnmt1 and DNA methylation are required to coordinate the events of Ag-specific CD8+ T cell differentiation (i.e., naive-effector-memory) is unknown. The embryonic-lethal phenotype of mice lacking DNA methyltransferases has thwarted the ability to study DNA methylation during the latter stages of cellular differentiation in vivo (6, 19). Therefore, using cre-mediated conditional ablation of Dnmt1 at the time of mature CD8+ T cell activation, we investigated the requirement of maintaining DNA methylation patterns for proper effector and memory CD8+ T cell differentiation during a live viral infection with lymphocytic choriomeningitis virus (LCMV). We found that following viral infection, expansion of Ag-specific CD8+ T cells was severely impaired in the absence of Dnmt1. Despite this, the generation of memory CD8+ T cells was disproportionately high, suggesting a preferential development of memory CD8+ T cells. Our data suggest that ablation of Dnmt1 and subsequent DNA methylation affect the finite proliferative potential of Ag-specific CD8+ T cells with moderate effects on their differentiation to effector and memory CD8+ T cells.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 Disclosures
 References
 
Animals, cells, and virus

All animals were housed in an American Association of Laboratory Animal Care-accredited facility under specific pathogen-free conditions at the Emory University Vaccine Research Center. Granzyme B-cre (GBC) and Dnmt12lox/2lox mice have each been previously described (20, 21, 22). All mice were bred and maintained at Emory Vaccine Center following guidelines outlined by the Institutional Animal Care and Use Committee. GBCxDnmt12lox/2lox mice were genotyped by PCR with tail DNA using primers described previously (20, 22). Littermate mice negative for cre recombinase (GBC–/–) or heterozygous for Dnmt1 (Dnmt1wt/2lox) were used as negative controls. LCMV Armstrong was triple-plaque purified on Vero cell monolayers and amplified once on BHK-21 cells (American Type Culture Collection), as described (23). LCMV Armstrong infections consisted of one injection of 2 x 105 PFU given i.p. Viral titers were determined by plaque assay on Vero cell (American Type Culture Collection) monolayers, as previously described (23). LCMV Armstrong was a gift from R. Ahmed (Emory University, Atlanta, GA).

Cell isolation, Abs, and flow cytometry

Spleen and lymph nodes were harvested from mice euthanized via cervical dislocation. We made single cell suspensions and lysed RBC using RBC lysis buffer (Sigma-Aldrich). For detection of cell surface Ags, ~2 x 106 splenocytes were incubated for 20–30 min on ice in 50 µl of FACS buffer (1x PBS; 0.01% sodium azide; 2% FBS; 2 mM EDTA) containing optimal concentrations of fluorochrome-conjugated Abs to CD4, CD8, B220, CD27, CD43, CD44, CD62L, CD69, IFN-{gamma}, or IL-2 (all from BD Biosciences). Allophycocyanin-conjugated tetramers of peptide/MHC class I molecules, when used, were included with Abs during the primary stain. Cells were then washed three times with FACS buffer and kept on ice until analysis. For detection of apoptotic cells, annexin V-biotin was used according to manufacturer’s instructions (BD Biosciences). Propidium iodide (PI) (5 µg/ml in 1x PBS) was used to exclude dead cells by incubation with cells for 10 min on ice. Streptavidin-allophycocyanin (5 µg/ml) (eBioscience) was used for secondary staining when appropriate. Intracellular cytokine staining was performed with reagents from BD Biosciences, according to manufacturer’s instructions. Flow cytometry data were acquired using a FACSCalibur (BD Biosciences) flow cytometer running CellQuest software. Data were analyzed with FlowJo software (TreeStar).

Cell sorting

RBC-lysed, splenocyte suspensions from LCMV-infected mice were stained with anti-CD8 PE, anti-CD44 FITC, and allophycocyanin-conjugated LCMV Db/GP33–41 or Db/nucleoprotein (NP)396–404 MHC class I tetramers. CD8+ T cells were electronically sorted using either a MoFlo (DakoCytomation) or FACSVantage (BD Biosciences) cell sorting apparatus. CD8+ T cells from naive mice were stained with anti-CD8 FITC and isolated using anti-FITC microbeads, according to manufacturer’s instructions (Miltenyi Biotec). Cell purities were ≥88% for all sorted populations.

Bisulfite modification, PCR amplification, and sequencing of H19 5' untranslated region (UTR)

We followed the previously described procedure optimized for bisulfite treatment of DNA derived from small sample sizes (24, 25). We conducted nested PCR using bisulfite-modified DNA as template using mH1-mH4 primers exactly as described (26) to amplify the 5' UTR of the paternally imprinted H19 promoter region. PCR products were cloned into the TOPO-TA cloning vector from Invitrogen Life Technologies, and individual colonies were chosen and used as template for colony PCR using M13 forward and reverse primers. PCR cycling conditions included an initial denaturation step for 8 min at 95°C, followed by 45 cycles of 95°C for 30 s, annealing at 50°C for 30 s, and extension for 1 min at 72°C. Samples remained at 4°C until purification. Colony PCR products were purified using the Multiscreen 96-well Filtration system (Millipore), according to manufacturer’s instructions. Approximately 50 ng of purified PCR product was used as template for sequencing reactions using M13 forward primers with the DYEnamic ET Dye Terminator sequencing kit from Amersham Biosciences. DNA sequences were acquired using a MEGABace 1000 capillary sequencing apparatus (Amersham Biosciences) and analyzed with EditView software (Applied Biosystems).

In vitro T cell stimulations

For cell division analysis, 107-108 RBC-lysed splenocytes from naive or LCMV-immune GBCxDnmt12lox/2lox and control mice were incubated in 1x PBS containing 5.0 µM CFSE (Molecular Probes) for 10 min at room temperature, followed by 5 min on ice. Loading was quenched by flooding with 100% FBS, followed by two washes with 1% FBS, 2 mM EDTA in 1x PBS. Cells were cultured in complete RPMI 1640 (10% FBS, 2 mM L-glutamine, 1000 U of penicillin G, 1000 U of streptomycin, 1 mM sodium pyruvate, 10 mM HEPES (all from Mediatech)) at 37°C in a humidified chamber with 5% ambient CO2. For cultures containing naive cells, stimulations consisted of PHA (10 µg/ml) (Sigma-Aldrich) plus human rIL-2 (40 U/ml) (Sigma-Aldrich), plate-bound anti-CD3 (5 µg/ml) (clone 145-2C1; BD Biosciences) and anti-CD28 (2 µg/ml) (clone 37.51; BD Biosciences), PMA (10 ng/ml) and ionomycin (250 ng/ml) (each from Sigma-Aldrich), or growth-supplemented medium, according to manufacturer’s recommendations (Igen International). Cultures were conducted in 96-well plates (2.5 x 106 cells/ml for whole splenocytes; 1.25 x 105/ml for purified T cells) for a total of 3.5 or 6 days. For cultures using immune splenocytes, 2 x 106 splenocytes from LCMV-immune mice were cultured in 96-well plates in complete RPMI 1640 alone, or medium supplemented with 0.5 µg/ml each GP33–41 and NP396–404, peptides. Cultures were incubated for 3–4 days before harvesting for flow cytometry analysis. Stimulations for intracellular cytokine staining were conducted by culturing 1 x 106 splenocytes with 1 µg/ml either GP33–41 or NP396–404 peptide, or medium alone for 5 h in the presence of human rIL-2 (25 U/ml) (Sigma-Aldrich) and GolgiPlug (BD Biosciences), according to manufacturer’s recommendations.

CTL assay

CD8+ T cells were enriched before the assay by removing CD19+ and CD4+ cells by cell sorting. Based on the number of CD8+ T cells, 50:1 E:T ratios using 3 x 103 targets were established, and the number of effector cells was titrated using 1/5 dilutions. MC57G targets were loaded in complete RPMI 1640 supplemented with 1 µg/ml NP396–404 peptide together with 100 µCi of 51Cr per 1 x 106 cells for 90 min at 37°C. Following washes, target cells were combined with effectors in triplicate, and cultures were incubated at 37°C for 5 h. A total of 100 µl of supernatent from each well was transferred to Lumaplates (PerkinElmer) and dried overnight. Radioactivity was detected using a Wallac Trilux liquid scintillation counter. Specific lysis was calculated using the formula: ((E – S)/(M – S)) x 100, in which E = experimental release, S = spontaneous release, and M = maximum release. The absolute number of NP396-specific CD8+ T cells for each animal was determined by tetramer staining and used to calculate the E:T ratio required to achieve 15% specific lysis of target cells.

Statistical methods

Student’s t test was used to generate all statistical values stated. Total numbers of cells reported were calculated from frequencies obtained by flow cytometry and cell counts using a standard hemocytometer. For statistical indications: *, indicates p ≤ 0.05; **, indicates p ≤ 0.005; and ***, indicates p ≤ 0.001.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 Disclosures
 References
 
Generation and characterization of GBCxDnmt12lox/2lox mice

To investigate the requirement of maintaining DNA methylation patterns during the generation and maintenance of Ag-specific CD8+ T cell responses, we devised a mouse model in which the DNA maintenance methyltransferase, Dnmt1, is selectively inactivated in mature CD8+ T cells following their activation in the periphery using the cre-loxP recombination strategy. We crossed mice transgenic for cre recombinase under the control of the truncated granzyme B promotor (GzmB-cre; GBC) (20, 21) with Dnmt12lox/2lox mice (22) to obtain doubly transgenic offspring (GBCxDnmt12lox/2lox). Our laboratory has previously used GzmB-cre mice crossed to a cre-reporter strain (ROSA26R) to successively track activated CD8+ T cells (21). The truncated granzyme B promotor is silent in developing thymocytes and naive CD8+ T cells. In the GBCxDnmt12lox/2lox mouse model used in this study, cre-mediated recombination of the inserted Dnmt12lox allele occurs only within T cells that have exited the thymus and received activation signals, yielding dysfunctional Dnmt1 protein. In all experiments in this study, littermate mice either lacking the cre recombinase transgene (GBC–/–) or heterozygous for Dnmt1 (GBC+Dnmt1wt/2lox) were used as negative controls, as Dnmt1wt/2lox mice maintain normal DNA methylation following recombination. Flow cytometry analysis of splenocytes from naive GBCxDnmt12lox/2lox offspring 3–5 wk of age showed these mice contained average numbers of CD4+, CD8+, and B220+ splenocytes compared with littermate controls (Fig. 1a). In addition, the number of CD8+ T cells with a CD44high memory phenotype in 3- to 5-wk-old GBCDnmt12lox/2lox mice did not significantly differ from littermate controls (Fig. 1b). These results show GBCxDnmt12lox/2lox mice support normal B and T lymphocyte development.


Figure 1
View larger version (15K):
[in this window]
[in a new window]
  		FIGURE 1. Normal lymphocyte development in GBCxDnmt12lox/2lox mice. Cohorts of 4- to 6-wk-old naive GBCxDnmt12lox/2lox ({blacksquare}) and control ({square}) mice were sacrificed, and the total number of CD4+, CD8+, or B220+ splenocytes was calculated from frequencies obtained by flow cytometry analysis. a, Shown are total number of splenocytes and lymphocyte subsets per animal. b, Plotted are the total number of CD8+ and CD4+ T cells per spleen with a CD44high phenotype. Error bars represent ±SEM. The data were derived from three independent experiments using two to three mice per group.

 
Reduced expansion of Ag-specific Dnmt1–/– CD8+ T cells during LCMV infection

Gene array and other studies have demonstrated that multiple changes in gene expression occur during CD8+ T cell differentiation to effector and memory T cells (27, 28, 29, 30). Although many studies have correlated demethylation of multiple genes with differentiation of CD8+ T cells, the extent to which DNA methylation is required to coordinate these processes is unknown. Therefore, we followed the differentiation of Ag-specific CD8+ T cells during the course of an antiviral response to the Armstrong strain of LCMV, an acute virus that produces strong CD8+ T cell responses in mice. To analyze the generation of effector CD8+ T cells, we infected GBCxDnmt12lox/2lox mice and littermate controls with LCMV, sacrificed the mice at the peak of the immune response (day 8), and determined the frequency and number of Ag-specific CD8+ T cells in the spleen using MHC class I tetramers and their ability to produce IFN-{gamma} following peptide stimulation.

Following infection of control mice, CD8+ T cells expanded to account for 22.12 ± 1.77% (SEM) of splenocytes by day 8 postinfection (p.i.) (Fig. 2a). In contrast, Dnmt1–/– CD8+ T cells accounted for only 5.88 ± 0.67% of splenocytes at this time (p < 0.001). Accordingly, total cell recovery from spleens of GBCxDnmt12lox/2lox mice was less than controls (data not shown). The frequency of CD8+ T cells in GBCxDnmt12lox/2lox mice expressing high levels of the activation marker CD44 was elevated, but did not achieve the same level as seen in littermate controls (65.07 ± 2.79% vs 80.78 ± 6.58%; p = 0.034). However, owing to a reduced frequency and total number of CD8+ T cells, the absolute number of activated CD44high Dnmt1–/– CD8+ T cells was significantly less than controls (p = 0.0006) (Fig. 2b). Next, we analyzed LCMV-specific CD8+ T cells in GBCxDnmt1–/– and control mice using tetramer staining and intracellular cytokine staining (ICS). Analysis of CD8+ T cells specific for the Db/GP33–41 (gp33) or Db/NP396–404 (NP396) epitope using MHC class I tetramers revealed significant reductions in their frequencies compared with littermate controls (1.47 ± 0.87% vs 5.16 ± 1.64% for gp33 (p = 0.0005); 5.93 ± 2.45% vs 12.81 ± 5.77% for NP396 (p = 0.0171)) (Fig. 2c). Similar to the CD8+CD44high T cell population, enumeration of gp33- and NP396-specific CD8+ T cells revealed GBCxDnmt12lox/2lox mice contained ≥84% reductions in Ag-specific CD8+ T cell numbers compared with littermate controls (p = 0.0002 for gp33; p = 0.002 for NP396) (Fig. 2d). Ag-specific CD8+ T cells from infected GBCxDnmt12lox/2lox and control mice displayed typical activation phenotypes in that they expressed high levels of CD44 (Fig. 2c), CD43, and granzyme B; expressed low levels of CD62L (L-selectin); and were bimodal for CD27 (Fig. 2e). These results demonstrate a normal activation phenotype together with a significant impairment of effector CD8+ T cells to accumulate in GBCxDnmt12lox/2lox mice.


Figure 2
View larger version (36K):
[in this window]
[in a new window]
  		FIGURE 2. Reduced effector CD8+ T cell accumulation 8 days following LCMV infection. Cohorts of GBCxDnmt2lox/2lox ({blacksquare}) and littermate control mice ({square}) were infected with LCMV and sacrificed 8 days later for analysis of Ag-specific T cell responses. Splenocytes were costained with anti-CD8{alpha}, anti-CD44, and Db/GP33–41 or Db/NP396–404 MHC class I tetramers. a, Flow cytometry plots show the frequency of activated (CD44high) CD8+ T cells within the live gate population from two representative mice within each group. Numbers within plots indicate frequency of total splenocytes. b, Total number (mean ± SEM) of CD8+CD44+ T cells per spleen. c, Representative contour plots gated on CD8+ T cells show CD44 expression among tetramer+ cells. Numbers within plots represent the frequency within total splenocytes (bottom) or CD8+ T cells (top). d, Total number (mean ± SEM) of CD8+CD44+ Ag-specific T cells per spleen. e, Activation marker expression among gated Ag-specific CD8+ T cells from GBCxDnmt12lox/2lox (dotted line) or control (solid line) mice. Bar graphs depict the combined results from two independent experiments containing three to five mice per group.

 
Because GBCxDnmt12lox/2lox mice exhibited diminished CD8+ effector T cell accumulation, we wished to determine whether these mice were able to clear the viral infection. Viral titers in wild-type mice infected with LCMV Armstrong generally peak 48–72 h following infection and are absent from the serum and tissues by day 8 p.i. (31, 32). Analysis of serum, kidney, and spleen samples on days 3, 5, and 9 p.i. revealed that GBCxDnmt12lox/2lox contained higher viral titers (1–2 logs) in both spleen and kidneys on days 3 and 5 p.i. compared with controls (Fig. 3, a and b). By day 9 p.i., both groups of mice cleared the infection in the kidneys. In the spleen, 100% of control mice and 57% of GBCxDnmt12lox/2lox mice (four of seven) cleared virus by day 9 p.i.; the remaining GBCxDnmt12lox/2lox mice contained low titers of LCMV in their spleens (~103 PFU/g in three of seven animals), which demonstrated a slightly reduced ability to control the infection in this compartment. Finally, we were unable to detect LCMV in serum from either group at any time point. This suggested that Dnmt1–/– CD8+ T cell effector functions were intact and sufficient to contain the viral infection. To investigate this, we assayed CD8+ T cells 8 days p.i. using ICS for their ability to produce IFN-{gamma}, TNF-{alpha}, or IL-2 in response to stimulation with either GP33–41 or NP396–404 peptide. Upon stimulation in vitro, CD8+ T cells from both GBCxDnmt12lox/2lox and control mice showed increased CD69 expression and produced IFN-{gamma} (Fig. 3c), although the frequency of IFN-{gamma}+ Dnmt1–/– CD8+ T cells was significantly reduced (2.54% ± 0.40 vs 9.77% ± 0.19 for gp33 (p = 0.0009); 5.92% ± 0.54 vs 19.15% ± 1.75 for NP396 (p = 0.0029)). Similar to the tetramer staining shown in Fig. 2d, the total number of IFN-{gamma}+ CD8+ T cells per spleen in GBCxDnmt12lox/2lox mice was significantly reduced compared with littermate controls (p = 0.0006 for gp33; p = 0.0042 for NP396) (Fig. 3d). TNF-{alpha} production was not impaired in GBCxDnmt12lox/2lox mice, as similar frequencies of IFN-{gamma}-secreting CD8+ T cells were capable of TNF-{alpha} production when compared with controls (Fig. 3e). These responses were Ag specific, as incubation with medium alone did not induce activation or cytokine production (data not shown). LCMV-specific CD8+ T cells found 8 days following LCMV infection are not capable of IL-2 production; rather, they gradually obtain this capability as CD8+ T cell memory develops (32, 33). Analysis of Ag-specific IL-2 production by ICS showed that neither control nor Dnmt1–/– CD8+ T cells produced IL-2 8 days following infection (data not shown). The ability to up-regulate IFN-{gamma} and TNF-{alpha}, but not IL-2, together with the normal expression of activation markers and viral clearance in vivo, demonstrated that Dnmt1–/– CD8+ T cells were capable of effector cell differentiation, albeit at reduced numbers. Finally, we assayed Dnmt1–/– and control CD8+ T cells for their ability to lyse peptide-loaded MC57G target cells via 51Cr release assay. We found that fewer (~2.5-fold) NP396-specific CD8+ T cells from GBCxDnmt2lox/2lox mice were required to lyse 15% of peptide-loaded targets compared with their control counterparts (expressed as lytic units; Fig. 3f). The increased ability to lyse peptide-loaded target cells may result from increased perforin expression in response to demethylation of the perforin 5' enhancer region and is consistent with the ability of these mice to control the viral infection despite significantly decreased numbers of Ag-specific CD8+ T cells.


Figure 3
View larger version (34K):
[in this window]
[in a new window]
  		FIGURE 3. Effector functions of Dnmt1–/– CD8+ T cells are comparable to control mice. Cohorts of GBCxDnmt12lox/2lox and control mice were infected with LCMV and analyzed for viral titers, cytokine production, and killing activity on various days p.i. Viral titers (mean ± SEM) in the spleen (a) and kidney (b) are shown for control and GBCxDnmt12lox/2lox mice on days 3, 5, and 9 p.i. Limit of detection = 50 PFU. n = 3–4 mice/group/day. c, Intracellular cytokine assay showing IFN-{gamma} production and CD69 expression by CD8+ T cells stimulated with 1 µg/ml GP33–41 or NP396–404 peptide for 5 h. Plots are gated on CD8+ T cells. Numbers within plots represent the frequency within total splenocytes (bottom) or CD8+ T cells (top). d, Total number (mean ± SEM) of IFN-{gamma}+CD8+ T cells per spleen is plotted. n = 7–8 mice per group. e, The frequency (mean ± SEM) of CD8+IFN-{gamma}+ T cells that also secreted TNF-{alpha} is graphed for control and GBCxDnmt12lox/2lox mice. f, 51Cr release assay comparing lysis of NP396-loaded MC57G target cells by control and Dnmt1–/– CD8+ T cells. Lytic units are graphed (mean ± SEM) and defined as the E:T ratio required to achieve 15% lysis of targets. n = 3–5 mice/group.

 
Next, we determined the extent to which loss of Dnmt1 affected the degree of genomic methylation in Ag-specific effector CD8+ T cells from GBCxDnmt12lox/2lox mice by analyzing individual CpG dinucleotides within the H19 promotor region. Due to paternal imprinting mechanisms, the 5' UTR of the H19 locus in wild-type mice is highly methylated in adult cells, including differentiated CD8+ T cells (E. Kersh, personal communication), and therefore was used as a surrogate marker for genomic methylation (34). We performed bisulfite sequence analysis on genomic DNA isolated from FACS-sorted CD8+CD44low, CD8+CD44highgp33+, and CD8+CD44highNP396+ T cells taken from two individual GBCxDnmt12lox/2lox mice infected 8 days earlier with LCMV. Fig. 4a depicts the methylation status of 11 CpG dinucleotides from individual clones derived from each population. The analyses showed that CD8+CD44low (i.e., naive) T cells from LCMV-infected mice contained heavily methylated loci; 123 of 132 (93.2%) total CpG analyzed were methylated. In contrast, Ag-specific effector CD8+ T cells showed extensive demethylation of the H19 locus. Among the four Ag-specific populations analyzed (two from each animal), two populations (gp33+ from 6.R1 and NP396+ from 6.RL) displayed 100% demethylation within the H19 5' UTR. The remaining two populations showed 45 and 56% CpG demethylation. We next performed PCR on the isolated genomic DNA to determine the extent of cre-mediated recombination within the floxed Dnmt1 allele. Fig. 4b shows that CD8+CD44low T cells from each animal contained very few copies of the recombined 1lox allele (8 and 15%). In contrast, three of four Ag-specific populations contained 84–98% of the recombined 1lox allele, while the remaining population (gp33-specific CD8+ T cells from 6.RL) contained ~61% 1lox allele. These results demonstrate that the majority of Ag-specific clones at this time have indeed deleted Dnmt1. These results also indicate that the degree of cre-mediated recombination in GBCxDnmt12lox/2lox mice was variable, a result confirmed by the presence of several partial to fully methylated clones seen in Fig. 4a. In sum, 76% (392 of 517) of the total CpG analyzed from the four Ag-specific populations contained nonmethylated CpG compared with 7% (9 of 132) from nonactivated CD8+ T cells. Collectively, these results demonstrate that Ag-specific CD8+ T cells from these animals are highly reduced in CpG methylation compared with their nonactivated counterparts.


Figure 4
View larger version (39K):
[in this window]
[in a new window]
  		FIGURE 4. Decreased methylation of H19 5' UTR in Dnmt1–/– effector CD8+ T cells. CD8+CD44low, CD8+CD44highgp33+, and CD8+CD44highNP396+ T cells were sorted from two individual GBCxDnmt12lox/2lox mice (6.RL and 6.R1) infected 8 days previously with LCMV. Genomic DNA was isolated from each population, modified with sodium bisulfite, and used as template for PCR amplification and DNA sequencing. a, The pattern of CpG methylation for each clone obtained from individual mice is shown. {blacksquare}, Indicate methylated CpG. {square}, Indicate nonmethylated CpG. b, PCR results showing the degree of recombination at the Dnmt1 locus for each of the sorted CD8+ T cell populations.

 
Generation of dysfunctional memory CD8+ T cells in GBCxDnmt2lox/2lox mice

Considering that the reduced numbers of Ag-specific CD8+ T cells in GBCxDnmt12lox/2lox mice produced an effector response that controlled the viral infection, we next asked whether the generation of CD8+ T cell memory occurred in the absence of Dnmt1. We allowed LCMV-infected GBCxDnmt12lox/2lox mice and littermate controls to rest until 30, 60, or 180 days p.i. and then harvested spleens for analysis of Ag-specific memory CD8+ T cell responses (Fig. 5). Surprisingly, we detected both gp33- and NP396-specific CD8+ T cells in the spleens of GBC x Dnmt12lox/2lox mice at all memory time points examined following infection. However, gp33- and NP396-specific CD8+ T cells were present in reduced frequency (Fig. 5a) and total number (Fig. 5b) in GBCxDnmt12lox/2lox mice compared with controls. ICS following stimulation with specific peptide demonstrated that Dnmt1–/– memory CD8+ T cells were capable of IFN-{gamma} (Fig. 5c) and IL-2 (data not shown) production, although the amount of cytokine produced on a per cell basis was significantly reduced (see below). Enumeration of IFN-{gamma}+ CD8+ T cells showed similar reductions in Ag-specific cells as the tetramer staining (Fig. 5d). These results demonstrate a reduced development (2- to 4-fold at days 60 and 180 p.i.) of Ag-specific memory CD8+ T cells in GBCxDnmt12lox/2lox mice.


Figure 5
View larger version (44K):
[in this window]
[in a new window]
  		FIGURE 5. Ag-specific memory CD8+ T cell development in GBCxDnmt2lox/2lox mice. Cohorts of GBCxDnmt2lox/2lox and control mice were infected with LCMV; allowed to rest for 30, 60, or 180 days p.i.; and analyzed for the presence of gp33- and NP396-specific memory CD8+ T cells using MHC class I tetramers and ICS for IFN-{gamma} as in Fig. 1. a, Representative contour plots showing the frequency of gp33-specific (top row) and NP396-specific (bottom row) CD8+ T cells on the indicated days p.i. b, The total number of tetramer+ CD8+ T cells (mean ± SEM) is graphed for the indicated days p.i. c, Representative contour plots showing IFN-{gamma} production and CD69 expression in response to 5-h stimulation with gp33 (top row) and NP396 (bottom row) peptides. d, The total number of CD8+IFN-{gamma}+ T cells (mean ± SEM) is graphed for the indicated days p.i. Numbers within contour plots represent the frequency of either tetramer+ (a) or IFN-{gamma}+ (c) cells among CD8+ T cells (top) or total splenocytes (bottom). Bar graphs represent the combined data from two independent experiments using two to four mice per group.

 
Following CD8+ effector T cell expansion, a contraction phase ensues that consistently results in a pool of memory cells that represent 5–10% of the size found at the peak (day 8) of the LCMV response (32, 35, 36). Interestingly, we noted that in contrast to the large reduction (83–91%) of Ag-specific cell numbers seen in GBCxDnmt12lox/2lox mice compared with controls 8 days following infection, the reductions were more modest during the immune phase of the response. The gp33- and NP396-specific memory CD8+ T cells were only reduced 68 and 56%, respectively, at day 60 p.i. compared with control CD8+ T cell numbers, as determined by tetramer staining. By day 180 p.i., Dnmt1–/– Ag-specific CD8+ T cells were reduced 67–71% compared with control mice. These results implied CD8+ T cells from GBCxDnmt12lox/2lox mice underwent less contraction compared with controls. Analysis of the Ag-specific CD8+ T cell contraction from days 8 to 60 p.i. in control mice revealed that gp33- and NP396-specific CD8+ T cells contracted to 10 and 7%, respectively, of their peak numbers seen at day 8. However, 60 days following infection of GBCxDnmt12lox/2lox mice, gp33-specific cells only contracted to 37% of their peak numbers at day 8 p.i., while NP396-specific CD8+ T cells showed higher contraction rates, displaying 18% of their population size seen at day 8 p.i. By day 180 p.i., GBCxDnmt12lox/2lox mice continued to display a decreased contraction that did not equal the reductions seen in control mice (20 vs 5.5% for gp33; 9.3 vs 5.5% for NP396). These results show a decreased contraction phase among Ag-specific CD8+ T cells in GBCxDnmt12lox/2lox mice following LCMV infection.

In addition to rapid cytokine production following peptide stimulation, one hallmark of memory CD8+ T cells is the ability to undergo proliferation following encounter with cognate Ag. To test the ability of memory Dnmt1–/– CD8+ T cells to proliferate in vitro, we stimulated CFSE-labeled whole splenocytes from LCMV-immune GBCxDnmt12lox/2lox and control mice with a combination of gp33 and NP396 peptides in vitro. Following 3–4 days in culture, cells were harvested and the loss of CFSE fluorescence was determined by flow cytometry. Fig. 6a shows that 34–65% of CD8+ T cells from three individual control mice displayed significant loss of CFSE (three to five divisions). In contrast, cultures established from GBCxDnmt12lox/2lox mice displayed little to no division. In the experiment shown, ~17% of CD8+ T cells from one mouse showed loss of CFSE, while <5% of CD8+ T cells from the remaining two mice underwent division. Additionally, we detected no increases in apoptosis or cell death (assessed by annexin V and 7-aminoactinomycin D staining) during the culture period compared with controls, suggesting that Dnmt1–/– memory CD8+ T cells did not divide significantly and die before day 4 (data not shown). We obtained similar results in a separate experiment in which the number of splenocytes plated from each animal was adjusted to normalize the number of Ag-specific CD8+ T cells per well. These results demonstrate a significant defect among Dnmt1–/– memory CD8+ T cells to undergo division following peptide stimulation in vitro.


Figure 6
View larger version (17K):
[in this window]
[in a new window]
  		FIGURE 6. Dnmt1–/– memory CD8+ T cells have defective memory cell properties. CFSE-labeled whole splenocytes from LCMV-immune (day 60 p.i.) GBCxDnmt12lox/2lox and control mice were stimulated in vitro with a combination of GP33–41 and NP396–404 peptides for 72 h and analyzed for loss of CFSE by flow cytometry. a, Histograms show log10 CFSE fluorescence among gated CD8+ T cells in which each plot represents an individual animal. Results are representative of three independent experiments using two to four mice per group. b, Bar graphs depict the geometric mean fluorescence intensity (GMFI ± SEM) of IFN-{gamma} produced by CD8+ T cells taken from mice 8, 30, 60, and 180 days p.i. following 5-h stimulation with either gp33 or NP396 peptide. Results represent three to seven mice per group at each time point. c, ICS analysis showing the frequency of memory CD4+ cells (day 60 p.i.) that produced IFN-{gamma} or IL-2 in response to stimulation with a combination of GP61–80 and NP309–324 peptides. Results represent three to four mice per group.

 
Although Ag-specific CD8+ T cells from GBC x Dnmt12lox/2lox mice produced IFN-{gamma} following peptide stimulation, we noted that the quantity of cytokine produced on a per cell basis was consistently less than that elaborated by control CD8+ T cells. Fig. 6b shows summarized data of geometric mean fluorescence intensities of IFN-{gamma} produced by Ag-specific CD8+ T cells from GBCxDnmt12lox/2lox and control mice throughout the course of LCMV infection. This decrease in IFN-{gamma} production is not due to a decrease in the number of Ag-specific CD4+ Th cells during the response because similar frequencies of these cells capable of both IFN-{gamma} and IL-2 production were detected by ICS 60 days following infection in both groups of mice (Fig. 6c).

Genomic demethylation in memory CD8+ T cells from GBCxDnmt12lox/2lox mice

We next considered the possibility that the Ag-specific CD8+ T cells detected in the memory phase of the response in GBC x Dnmt12lox/2lox mice were the result of an outgrowth of small numbers of nonrecombined effector CD8+ T cells. Therefore, we analyzed the methylation status of the H19 5' UTR from FACS-sorted CD44high gp33-specific (day 50 p.i.), CD44high NP396-specific (day 60 p.i.), and CD44low CD8+ T cells (days 50 and 60 p.i.) from LCMV-immune GBCxDnmt12lox/2lox and cre-negative control mice. Fig. 7a shows the pattern of CpG methylation of individual clones from each sample analyzed. The results showed, as expected, heavily methylated loci in both naive CD8+CD44low and Ag-specific memory CD8+CD44highgp33+ and CD8+CD44highNP396+ T cells from control LCMV-immune mice. Collectively, 85% of the 473 CpG sites analyzed from control mice were methylated. Similarly, 91% (37 of 429) CpG sites analyzed from Dnmt1–/– naive CD8+CD44low T cells isolated from LCMV-immune GBCxDnmt12lox/2lox mice contained methyl groups. In contrast, CD8+CD44highgp33+ memory T cells from these mice showed extensive demethylation of the H19 5' UTR. Of the 308 total CpG sites analyzed, only 38% were methylated in gp33-specific memory CD8+ T cells from GBCxDnmt12lox/2lox mice. The extent of H19 demethylation among the two animals was variable, and this result was confirmed by PCR for Dnmt1 recombination. The PCR results (Fig. 7b) showed extensive deletion (72%) of the 2lox allele in CD8+ T cells from animal 4.L, whereas CD8+ T cells from animal 4.R displayed 26% deletion of the 2lox allele. Analysis of NP396-specific CD8+ T cells from immune GBCxDnmt12lox/2lox mice showed that, again, the degree of demethylation was variable among three individual mice (Fig. 7c). Demethylated CpG within the H19 5' UTR from three individual mice accounted for 23, 45, and 91%, respectively, of the total CpG analyzed. PCR for recombination revealed 60–70% deletion of the 2lox allele among the three samples (Fig. 7d). Although fully demethylated clones were rare, 57% of Ag-specific clones analyzed showed >70% demethylation, demonstrating that Ag-specific memory CD8+ T cells survived up to 60 days following infection in the absence of significant genomic methylation.


Figure 7
View larger version (57K):
[in this window]
[in a new window]
  		FIGURE 7. Decreased methylation of H19 5' UTR in Dnmt1–/– memory CD8+ T cells. CD8+ T cells from spleens of GBCxDnmt2lox/2lox and control mice infected 50 or 60 days previously with LCMV were sorted into CD8+CD44low and CD8+CD44highgp33+ (day 50) or CD8+CD44low and CD8+CD44highNP396+ (day 60) fractions. Genomic DNA was isolated from each population, modified with sodium bisulfite, and used as template for PCR amplification and DNA sequencing. The methylation pattern of each clone obtained from gp33-specific (a) or NP396-specific (c) CD8+ T cells from individual mice (depicted as 4.L and 4.R) is shown. {blacksquare}, Indicate methylated CpG. {square}, Indicate nonmethylated CpG. b and d, Display PCR results showing the degree of recombination at the Dnmt1 locus within gp33-specific (b) and NP396-specifc (d) CD8+ T cells from GBCxDnmt12lox/2lox and control mice.

 
Naive Dnmt1–/– CD8+ T cells are capable of multiple cell divisions in vitro

During LCMV infection, the expansion of Ag-specific CD8+ T cell clones approaches ≥2000-fold (37). Although the number of effector CD8+ T cells in GBCxDnmt12lox/2lox mice showed great reductions, to reach the limit of detection by conventional methods implies these cells have undergone a significant number of divisions. To determine whether naive Dnmt1–/– CD8+ T cells could proliferate in vitro, we cultured CFSE-labeled splenocytes in the presence of plate-bound anti-CD3/anti-CD28, PHA plus IL-2, PMA and ionomycin, or growth factor containing medium (Igen International) for either 3.5 or 6 days. As shown in Fig. 8, a similar frequency of CD8+ T cells from both GBCxDnmt12lox/2lox and control mice underwent up to seven divisions by day 3.5 in response to both PHA plus IL-2 or growth-supplemented medium (Fig. 8a). By day 6, all stimulation conditions displayed significant growth of CD8+ T cells with comparable cell loss in each group (Fig. 8b).


Figure 8
View larger version (31K):
[in this window]
[in a new window]
  		FIGURE 8. Dnmt1–/– CD8+ T cells have normal in vitro proliferation capability. Naive GBCxDnmt12lox/2lox (lower panels) and control (upper panels) splenocytes were labeled with CFSE and cultured for either 3.5 (a) or 6 (b) days in the presence of the indicated stimuli. The plots show log10 CFSE fluorescence among gated CD8+ T cells. The results shown are representative of four independent experiments using two to four mice per group. c, PCR showing the degree of recombination at the Dnmt1 locus within genomic DNA isolated from in vitro cultures. Lanes 1 and 5, Anti-CD3/Anti-CD28; lanes 2 and 6, PMA and ionomycin; lanes 3 and 7, PHA plus IL-2; lanes 4 and 8, growth-supplemented medium.

 
In a separate experiment, purified CD8+ T cells from GBCxDnmt12lox/2lox and control mice were stimulated as above, in vitro, and genomic DNA was harvested following 3.5 and 6 days. PCR assays for cre-mediated recombination within these samples demonstrated the normal division profiles displayed by Dnmt1–/– CD8 T cells were not due to cre-escape mutants, as all stimuli showed a similar degree of recombination (62–75%) (Fig. 8c). Approximately 15–30% of T cells in each stimulation condition were not induced to divide, and therefore would not be expected to undergo recombination of the Dnmt12lox allele. Although the degree of recombination was not complete among the various stimuli, cell recovery was similar between control and GBCxDnmt12lox/2lox cultures, which would not be observed if a large fraction of the cells died during culture. In addition, we detected no increases in CD8+ T cell death during early LCMV infection (see below). We conclude from these results that Dnmt1–/– CD8+ T cells have normal, short-term proliferative capabilities in vitro.

Apoptosis and development of Ag-specific T cell responses during early LCMV infection

The proliferation experiments above suggested that Dnmt1–/– CD8+ T cells are capable of at least seven cellular divisions. These data seemingly contrast the highly reduced number of CD8+ effector T cells in GBCxDnmt12lox/2lox mice following infection. To better understand why an effector cell response fails to fully mature in GBCxDnmt12lox/2lox mice, we considered two possible scenarios. First, Dnmt1–/– CD8+ T cells may divide at an equivalent rate to wild-type cells, but reach a limit of division and undergo apoptosis en masse. Alternatively, the proliferation or recruitment of Dnmt1–/– CD8+ T cells may have been delayed or decreased compared with controls. To address these issues, we first analyzed CD8+ T cells from LCMV-infected GBC x Dnmt12lox/2lox mice for increased apoptosis on days 5 through 9 of the response using annexin V and PI to detect apoptotic and viable cells, respectively.

Following infection, we found that the kinetics of CD8+ T cell death were similar between GBCxDnmt12lox/2lox mice and littermate controls (Fig. 9a). We did not detect significant CD8+ T cell apoptosis until day 8 of the response, at which time 54.03% ± 4.71 of the control CD8+ T cells and 32.12% ± 2.57 of Dnmt1–/– CD8+ T cells showed annexin V reactivity (p = 0.004). CD8+ T cell apoptosis at day 9 p.i. was less in both groups compared with day 8; however, GBCxDnmt12lox/2lox mice maintained higher frequencies of surviving cells than controls (88.57% ± 1.37 vs 74.63% ± 6.26; p = 0.095). Furthermore, the number of PI+ cells was similar between Dnmt1–/– and control CD8+ T cells at each time point (data not shown). These data demonstrate that increased cell death among Dnmt1–/– CD8+ T cells is not responsible for the decreased burst size of effector cells following LCMV infection.


Figure 9
View larger version (12K):
[in this window]
[in a new window]
  		FIGURE 9. Dnmt1–/– CD8+ T cells display decreased apoptosis and normal growth kinetics in vivo. We sacrificed cohorts of LCMV-infected mice on days 5–9 following the response and analyzed their splenocytes for CD8+ T cell apoptosis and development of Ag-specific CD8+ T cell responses. a, Bar graphs (mean ± SEM) depict the frequencies of PI– annexin V+ CD8+ T cells in GBCxDnmt12lox/2lox ({blacksquare}) and control mice ({square}) on the days indicated. The total number (mean ± SEM) of gp33-specific (b) and NP396-specific (c) CD8+ T cells per spleen is plotted on the indicated days and shows a slower development of Ag-specific CD8+ T cell numbers in GBCxDnmt12lox/2lox compared with control mice. n = 3–7 mice/group for each time point.

 
Although the in vitro proliferation experiments in Fig. 8 suggested Dnmt1–/– CD8+ T cells divided at rates equivalent to controls, we next considered a slower proliferation of LCMV-specific CD8+ T cells in GBCxDnmt12lox/2lox mice by following the development of Ag-specific CD8+ T cells during the early phase of LCMV infection using tetramers. The data in Fig. 9, b and c, show that LCMV gp33- and NP396-specific CD8+ T cells from both groups of mice were present by day 5; however, GBCxDnmt12lox/2lox mice already had fewer numbers of Ag-specific cells at this time. By day 8 of the response, both control and Dnmt1–/– CD8+ T cells had increased in number and showed similar rates of division. Because the frequency of Ag-specific CD8+ T cells is too low for detection by flow cytometry at time points earlier than day 5, we investigated the recruitment, proliferation, and apoptosis of CD8+ T cells from GBCxDnmt12lox/2lox and control mice following adoptive transfer to a lymphopenic environment in vivo. We transferred equal numbers of CFSE-labeled naive CD8+ T cells from control and GBCxDnmt12lox/2lox mice to sublethally irradiated Ly-5.1 hosts and harvested their splenocytes for analysis on days 2, 3, 5, 15, and 60 posttransfer. Representative histograms showing CFSE loss from each time point are shown in Fig. 10a. Similar to early LCMV infection, we detected no increase in Dnmt1–/– CD8+ T cell apoptosis over controls at any of the time points examined following transfer (data not shown). However, 48 h following transfer, 51.1% ± 6.8 of CD8+ T cells from GBCxDnmt12lox/2lox mice remained undivided compared with 26.7% ± 5.6 of control CD8+ T cells (p < 0.001) (Fig. 10b). One day later, the frequency of undivided CD8+ from GBCxDnmt12lox/2lox mice remained significantly higher than control CD8+ T cells (23.9% ± 0.98 vs 34.1% ± 4.3; p < 0.01), which resulted in fewer CD8+ T cells from GBCxDnmt12lox/2lox mice at subsequent generations (Fig. 10c). By day 5 posttransfer, the overall recruitment of CD8+ T cells was similar between the two groups; however, fewer Dnmt1–/– CD8+ T cells had divided ≥5 times compared with controls (14.8% ± 3.5 vs 25.5% ± 6.5; p < 0.05) (Fig. 10d). This trend was still evident 15 days following transfer, in which a small population of Dnmt1–/– CD8+ T cells was detected that retained low levels of CFSE compared with control CD8+ T cells that had lost CFSE fluorescence by this time. By day 60 following transfer, CD8+ T cells from each group were CFSE negative (Fig. 10e). Together, these results suggest that while similar frequencies of Dnmt1–/– and control CD8+ T cells were recruited into division, the recruitment was significantly delayed in Dnmt1–/– vs control CD8+ T cells.


Figure 10
View larger version (21K):
[in this window]
[in a new window]
  		FIGURE 10. Dnmt1–/– CD8+ T cells exhibit delayed recruitment during homeostatic proliferation. Equal numbers of CD8+ T cells from GBCxDnmt12lox/2lox and control mice were transferred to sublethally irradiated (650 rad) Ly-5.1 hosts. On various days following transfer, splenocytes were harvested and analyzed for apoptosis and CFSE fluorescence by flow cytometry. a, Representative histograms showing log10 CFSE fluorescence among transferred Dnmt1–/– and control Ly-5.2+CD8+ T cells on various days posttransfer. b–d, Bar charts indicate the mean frequency (± SEM) of transferred CD8+ T cells having undergone the indicated number of divisions on days 2, 3, and 5 posttransfer, respectively. e, Graphed are the mean frequencies (± SEM) of transferred CD8+ T cells retaining CFSE fluorescence on days 15 and 60 posttransfer. n = 3–5 mice/group/day.

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
Disclosures
 References
 
The present results show that DNA methylation by Dnmt1 is not critical for the activation and differentiation of effector CD8+ T cells during viral infection, but is required for accumulation of robust effector CD8+ T cell responses. The ability of CD8+ T cells from GBCxDnmt12lox/2lox mice to maintain a type 1 cytotoxic CD8+ T cell effector phenotype following activation in vivo is consistent with recent studies by Makar and colleagues (17, 38) using a similar model in which conditional ablation of Dnmt1 occurs in CD4+CD8+ double-positive thymocytes (CD4creDnmt12lox mice). They reported that although Th2 cytokine expression by Dnmt1–/– CD8+ T cells was induced under nonpolarizing stimulation conditions, Dnmt1–/– CD8+ T cells in their system did not significantly up-regulate these genes under Th1 polarizing conditions, an environment that would prevail during viral infection. Although the investigation of Th2 cytokine regulation was beyond the scope of this report, the ability of CD8+ T cells from GBCxDnmt12lox/2lox mice to respond to cognate signals and maintain proper effector function is consistent with the findings by this group.

Our proliferation studies showed that CFSE-labeled Dnmt1–/– CD8+ T cells achieved significant division following polyclonal stimulation in vitro. This is in contrast to reports showing that loss of Dnmt1 significantly decreased the capacity for cellular proliferation in other experimental systems. In Dnmt1–/– p53–/– primary embryonic fibroblasts (22), as well as naive Dnmt1–/– CD8+ T cells (system discussed above (38)), cell growth was limited to ≤5 population doublings. Our in vitro studies, however, were consistent with the development of detectable numbers of Ag-specific CD8+ T cells following LCMV infection of GBCxDnmt12lox/2lox mice. Although we found that short-term proliferative capacities in vitro were normal in Dnmt1–/– CD8+ T cells, extensive proliferation by these cells was limited in vivo following LCMV infection. The expansion of Ag-specific CD8+ T cells during LCMV infection approaches 2000-fold; Blattman et al. (37) have estimated the number of LCMV-specific CD8+ precursors in naive mice to be 1–2 in 105 CD8+ T cells for any given epitope (100–200/mouse). Assuming an average number of 150 precursors per mouse, we estimate that gp33- or NP396-specific Dnmt1–/– CD8+ T cells in our system underwent ~12 or 13 cell divisions, respectively, by day 8 p.i. In contrast, control CD8+ T cells underwent ~16–17 divisions. These calculations assume no significant cell death among memory precursors occurred during the expansion phase. Indeed, we found no increases in CD8+ T cell apoptosis during LCMV infection, in vitro proliferation, or homeostatic proliferation. Instead, beginning as early as day 5 p.i., we consistently detected fewer Ag-specific CD8+ T cells in GBCxDnmt12lox/2lox mice, which continued to expand in number through day 8 p.i. Together with the normal proliferation seen following polyclonal stimulation in vitro, this suggests that decreased proliferation rates were not the reason for decreased accumulation of effector cells following infection. Rather, it appears that a delayed recruitment of CD8+ T cells resulted in decreased accumulation, as evidenced by the homeostatic proliferation experiments shown in Fig. 10.

The successful differentiation of small numbers of CD8+ effector T cells raises the question of how CD8+ T cells are able to differentiate in the absence of significant DNA methylation. One possible explanation is that gene expression patterns required for differentiation are established rapidly following TCR engagement, before significant demethylation of the genome. Although we did not assay for levels of Dnmt1 protein, it is reasonable to assume that following deletion of the Dnmt1 allele, one to two rounds of cell division are required to deplete the cell of Dnmt1 enzyme. This would be followed by demethylation of the genome, an event that also requires several rounds of DNA replication and cell division. This progression of events may leave a small window of time in which Dnmt1 production has ceased, yet DNA methylation is present, which would allow not only for the cell division reported in this study, but also for establishment of proper gene expression patterns. Although (the absence of) DNA methylation is conventionally thought to facilitate the formation of euchromatin, the decreased IFN-{gamma} production (Fig. 6b) and regulated secretion of IL-2 by Dnmt1–/– CD8+ T cells in our system suggest chromatin is not uniformly in an open configuration. In the absence of DNA methylation, these gene expression patterns may be maintained by more central events such as chromatin remodeling that, once established, may not rely as heavily on DNA methylation for its preservation. Although technically difficult, direct analysis of chromatin structure at these key loci and others (CD8, CD4, etc.) would be valuable in understanding how gene expression patterns are maintained in Ag-specific CD8+ T cells from GBCxDnmt12lox/2lox mice.

Despite 80–90% reductions in Ag-specific Dnmt1–/– effector CD8+ T cells 8 days following infection, the number of memory cells that developed in GBCxDnmt12lox/2lox mice was disproportionately high, due to a decreased contraction of effector CD8+ T cells. One obvious explanation for the decreased contraction is a result from a defective program of activation-induced cellular apoptosis in Dnmt1–/– effector CD8+ T cells. Indeed, we found Dnmt1–/– CD8+ T cells underwent less apoptosis on days 8 and 9 p.i. when compared with controls (Fig. 9a). Alternatively, the decreased contraction may reflect, at least in part, separate effector and memory cell lineages, in which a preferential survival of memory precursor, but not effector, CD8+ T cells occurred. Although analysis of IL-7R{alpha} expression revealed no significant differences among Ag-specific Dnmt1–/– CD8+ T cells 8 days p.i. when compared with controls (data not shown), the reduced numbers of Ag-specific CD8+ T cells seen during the first week of the infection in GBCxDnmt12lox/2lox mice suggest that memory precursors may have been selected from cells with a shorter proliferative history. Rigorous testing of this hypothesis, in which a separate lineage of memory precursor cells develops from less-divided progenitors, however, will require techniques capable of tracking cell division beyond the current limitations afforded by CFSE.

In summary, we have shown that ablating Dnmt1 at the time of naive T cell activation led to a reduction in the burst size of Ag-specific CD8+ T cells. Surprisingly, Dnmt1–/– CD8+ T cells retained their cellular identity and regulated expression of activation markers and effector cytokines. Despite fewer numbers and decreased IFN-{gamma} expression per cell, GBCxDnmt12lox/2lox mice were only slightly delayed in their ability to control viral infection with LCMV in the absence of significant DNA methylation. These results suggest that additional epigenetic mechanisms are operating to maintain cellular identity in the absence of DNA methylation.


   Acknowledgments
 
We gratefully acknowledge Dr. Christopher Wilson and Joseph Dauner for critical reading of the manuscript. We also thank Karin Smith for mouse colony management.


   Disclosures
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
References
 
The authors have no financial conflict of interest.


   Footnotes
 
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1 These studies were supported by funding from the National Institutes of Health. Back

2 Address correspondence and reprint requests to Dr. Joshy Jacob, 954 Gatewood Road NE, Room 2042, Atlanta, GA 30329. E-mail address: jjacob3{at}emory.edu Back

3 Abbreviations used in this paper: Dnmt1, DNA methyltransferase 1; GBC, granzyme B-cre; ICS, intracellular cytokine staining; LCMV, lymphocytic choriomeningitis virus; NP, nucleoprotein; p.i., postinfection; PI, propidium iodide; UTR, untranslated region. Back

Received for publication July 6, 2005. Accepted for publication January 20, 2006.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 Disclosures
 References
 

  1. Zinkernagel, R. M., M. F. Bachmann, T. M. Kundig, S. Oehen, H. Pirchet, H. Hengartner. 1996. On immunological memory. Annu. Rev. Immunol. 14: 333-367. [Medline]
  2. Ahmed, R., D. Gray. 1996. Immunological memory and protective immunity: understanding their relation. Science 272: 54-60. [Abstract]
  3. Kaech, S. M., E. J. Wherry, R. Ahmed. 2002. Effector and memory T-cell differentiation: implications for vaccine development. Nat. Rev. Immunol. 2: 251-262. [Medline]
  4. Sprent, J., C. D. Surh. 2002. T cell memory. Annu. Rev. Immunol. 20: 551-579. [Medline]
  5. Seder, R. A., R. Ahmed. 2003. Similarities and differences in CD4+ and CD8+ effector and memory T cell generation. Nat. Immunol. 4: 835-842. [Medline]
  6. Li, E., T. H. Bestor, R. Jaenisch. 1992. Targeted mutation of the DNA methyltransferase gene results in embryonic lethality. Cell 69: 915-926. [Medline]
  7. Robert, M. F., S. Morin, N. Beaulieu, F. Gauthier, I. C. Chute, A. Barsalou, A. R. MacLeod. 2003. DNMT1 is required to maintain CpG methylation and aberrant gene silencing in human cancer cells. Nat. Genet. 33: 61-65. [Medline]
  8. Singal, R., R. Ferris, J. A. Little, S. Z. Wang, G. D. Ginder. 1997. Methylation of the minimal promoter of an embryonic globin gene silences transcription in primary erythroid cells. Proc. Natl. Acad. Sci. USA 94: 13724-13729. [Abstract/Free Full Text]
  9. Bednarik, D. P., C. Duckett, S. U. Kim, V. L. Perez, K. Griffis, P. C. Guenthner, T. M. Folks. 1991. DNA CpG methylation inhibits binding of NF-{kappa}B proteins to the HIV-1 long terminal repeat cognate DNA motifs. New Biol. 3: 969-976. [Medline]
  10. Jones, P. L., G. J. Veenstra, P. A. Wade, D. Vermaak, S. U. Kass, N. Landsberger, J. Strouboulis, A. P. Wolffe. 1998. Methylated DNA and MeCP2 recruit histone deacetylase to repress transcription. Nat. Genet. 19: 187-191. [Medline]
  11. Zhang, Y., H. H. Ng, H. Erdjument-Bromage, P. Tempst, A. Bird, D. Reinberg. 1999. Analysis of the NuRD subunits reveals a histone deacetylase core complex and a connection with DNA methylation. Genes Dev. 13: 1924-1935. [Abstract/Free Full Text]
  12. Feng, Q., Y. Zhang. 2001. The MeCP1 complex represses transcription through preferential binding, remodeling, and deacetylating methylated nucleosomes. Genes Dev. 15: 827-832. [Abstract/Free Full Text]
  13. Fuks, F., P. J. Hurd, D. Wolf, X. Nan, A. P. Bird, T. Kouzarides. 2003. The methyl-CpG-binding protein MeCP2 links DNA methylation to histone methylation. J. Biol. Chem. 278: 4035-4040. [Abstract/Free Full Text]
  14. Agarwal, S., A. Rao. 1998. Modulation of chromatin structure regulates cytokine gene expression during T cell differentiation. Immunity 9: 765-775. [Medline]
  15. Fitzpatrick, D. R., K. M. Shirley, L. E. McDonald, H. Bielefeldt-Ohmann, G. F. Kay, A. Kelso. 1998. Distinct methylation of the interferon {gamma} (IFN-{gamma}) and interleukin 3 (IL-3) genes in newly activated primary CD8+ T lymphocytes: regional IFN-{gamma} promoter demethylation and mRNA expression are heritable in CD44highCD8+ T cells. J. Exp. Med. 188: 103-117. [Abstract/Free Full Text]
  16. Lee, D. U., S. Agarwal, A. Rao. 2002. Th2 lineage commitment and efficient IL-4 production involves extended demethylation of the IL-4 gene. Immunity 16: 649-660. [Medline]
  17. Makar, K. W., C. B. Wilson. 2004. DNA methylation is a nonredundant repressor of the Th2 effector program. J. Immunol. 173: 4402-4406. [Abstract/Free Full Text]
  18. Lu, Q., A. Wu, D. Ray, C. Deng, J. Attwood, S. Hanash, M. Pipkin, M. Lichtenheld, B. Richardson. 2003. DNA methylation and chromatin structure regulate T cell perforin gene expression. J. Immunol. 170: 5124-5132. [Abstract/Free Full Text]
  19. Okano, M., D. W. Bell, D. A. Haber, E. Li. 1999. DNA methyltransferases Dnmt3a and Dnmt3b are essential for de novo methylation and mammalian development. Cell 99: 247-257. [Medline]
  20. Jacob, J., D. Baltimore. 1999. Modelling T-cell memory by genetic marking of memory T cells in vivo. Nature 399: 593-597. [Medline]
  21. Maris, C. H., J. D. Miller, J. D. Altman, J. Jacob. 2003. A transgenic mouse model genetically tags all activated CD8 T cells. J. Immunol. 171: 2393-2401. [Abstract/Free Full Text]
  22. Jackson-Grusby, L., C. Beard, R. Possemato, M. Tudor, D. Fambrough, G. Csankovszki, J. Dausman, P. Lee, C. Wilson, E. Lander, R. Jaenisch. 2001. Loss of genomic methylation causes p53-dependent apoptosis and epigenetic deregulation. Nat. Genet. 27: 31-39. [Medline]
  23. Ahmed, R., A. Salmi, L. D. Butler, J. M. Chiller, M. B. Oldstone. 1984. Selection of genetic variants of lymphocytic choriomeningitis virus in spleens of persistently infected mice: role in suppression of cytotoxic T lymphocyte response and viral persistence. J. Exp. Med. 160: 521-540. [Abstract/Free Full Text]
  24. McDonald, L. E., G. F. Kay. 1997. Methylation analysis using bisulfite genomic sequencing: application to small numbers of intact cells. BioTechniques 22: 272-274. [Medline]
  25. Grunau, C., S. J. Clark, A. Rosenthal. 2001. Bisulfite genomic sequencing: systematic investigation of critical experimental parameters. Nucleic Acids Res. 29: E65-65.
  26. Warnecke, P. M., D. Biniszkiewicz, R. Jaenisch, M. Frommer, S. J. Clark. 1998. Sequence-specific methylation of the mouse H19 gene in embryonic cells deficient in the Dnmt-1 gene. Dev. Genet. 22: 111-121. [Medline]
  27. Kaech, S. M., S. Hemby, E. Kersh, R. Ahmed. 2002. Molecular and functional profiling of memory CD8 T cell differentiation. Cell 111: 837-851. [Medline]
  28. Latner, D. R., S. M. Kaech, R. Ahmed. 2004. Enhanced expression of cell cycle regulatory genes in virus-specific memory CD8+ T cells. J. Virol. 78: 10953-10959. [Abstract/Free Full Text]
  29. Wan, J., D. Martinvalet, X. Ji, C. Lois, S. M. Kaech, U. H. Von Andrian, J. Lieberman, R. Ahmed, N. Manjunath. 2003. The Bcl-2 family pro-apoptotic molecule, BNIP3 regulates activation-induced cell death of effector cytotoxic T lymphocytes. Immunology 110: 10-17. [Medline]
  30. Hathcock, K. S., S. M. Kaech, R. Ahmed, R. J. Hodes. 2003. Induction of telomerase activity and maintenance of telomere length in virus-specific effector and memory CD8+ T cells. J. Immunol. 170: 147-152.