|
|
||||||||
Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104
| Abstract |
|---|
|
|
|---|
| Introduction |
|---|
|
|
|---|
are short-lived and do not develop into memory cells (8), while memory CD8 T cells arise from IFN-
-producing effectors (9). The initial response of Ag-specific T cells is difficult to analyze because the precursor frequency of specific cells is too low to visualize by conventional methods. Thus, most studies have examined the primary responses of CD4 and CD8 T cells in vitro by stimulation with anti-CD3 mAb or in vivo by adoptive transfer of CFSE-labeled TCR transgenic cells (10). These studies have shown that the CD4 T cell response is heterogeneous in nature with responding cells dividing and differentiating to various extents (11, 12, 13, 14, 15, 16, 17, 18, 19), while the CD8 T cell response is more homogeneous with all responding cells dividing and differentiating extensively (1, 2, 3). The reason for these differences between CD4 and CD8 T cells is not known, and it remains to be seen whether these differences also occur during the response to an infection in normal hosts. In vitro stimulation with mAbs to CD3 and CD28 in the presence of IL-2 may not effectively simulate the complex inflammatory environment that occurs during infection (20). For example, important cytokines and costimulatory molecules that are available in vivo may not be provided appropriately in in vitro settings (21). This is particularly important to consider when comparing the responses of CD4 and CD8 T cells because CD4 T cells have been shown to be more dependent on costimulation than CD8 T cells (22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35). Similarly, it is not known whether the response of a large number of adoptively transferred monoclonal TCR transgenic T cells accurately reflects the response of Ag-specific T cells at a physiological precursor frequency. Thus, it remains to be determined whether the reported differences in CD4 and CD8 T cell division and differentiation reflect the true nature of CD4 and CD8 T cell responses.
To investigate the differences in CD4 and CD8 T cell proliferation and differentiation in response to a natural infection, we developed an adoptive transfer system that enabled us to study polyclonal CD4 and CD8 T cells in vivo during the course of a primary Listeria monocytogenes (LM)4 infection. Surprisingly, we observed that adoptively transferred polyclonal CD4 and CD8 T cells proliferated extensively in response to LM infection. These findings contrast strikingly with previous results that TCR transgenic CD4 T cells divide a limited number of times in response to infection, while TCR transgenic CD8 T cells divide extensively under the same conditions. We hypothesized that the extent of monoclonal TCR transgenic CD4 T cell division was limited due to clonal competition between CD4 T cells of the same specificity imposed by an abnormally high precursor frequency. Indeed, we found that reducing the number of adoptively transferred TCR transgenic CD4 T cells increased the number of divisions completed by these cells and the percentage of differentiated effectors produced by these cells. As we approached physiological precursor frequencies, TCR transgenic CD4 T cells divided as extensively as polyclonal CD4 T cells. In addition, the protection provided by these cells increased as the number of transferred cells decreased. Our results suggest that the adoptive transfer of TCR transgenic CD4 T cells may significantly underestimate the extent of proliferation and differentiation of CD4 T cells in response to infection.
| Materials and Methods |
|---|
|
|
|---|
OT-I, OT-II, and BALB/c-TgN(DO11.10)Loh mice were previously described (36, 37, 38). OT-I and OT-II mice were bred onto the B6.PL-Thy-1.1 background. C57BL/6 and B6.Ly-5.2/Cr (Ly-5.1) mice were obtained from the National Cancer Institute (Bethesda, MD). B6.PL-Thy-1.1 mice were obtained from The Jackson Laboratory, and BALB/c-Thy-1.1 mice were obtained from C. Surh (The Scripps Institute, La Jolla, CA). Construction of rLM-OVA was described previously (11, 39). The LD50 of rLM-OVA in BALB/c mice is
5 x 104, and the LD50 of rLM-OVA in C57BL/6 mice is
5 x 105; experimental mice were infected with 0.1 LD50 of rLM-OVA.
Adoptive transfer
Splenocytes from OT-I, OT-II, DO11.10, or Ly-5.1 mice were labeled with CFSE, as described (40). For OT-I/OT-II combined transfers, T cells were enriched by depleting splenocytes with B220 and MHC class II MicroBeads by MACS (Miltenyi Biotec) and CFSE labeled, and the indicated number of each cell type was cotransferred per mouse. For the Ly-5.1 and individual OT-II and DO11.10 transfers, total splenocytes were labeled with CFSE and the indicated number of splenocytes was transferred. Experimental groups included two to five mice per group, and similar results were observed in at least two independent experiments.
Flow cytometry
Proliferation of transferred cells was visualized by incremental loss of CFSE fluorescence. Transferred OT-I and OT-II cells were identified by staining with mAb to Thy-1.1, V
2, and CD8 or CD4, respectively. DO11.0 cells were identified by staining with mAbs to Thy-1.2, CD4, and KJ1-26. For the OT-II and DO11.10 titration experiment, 3-fold more events were acquired for each dilution, starting with 1 million events for 4.5 million transferred cells. For the unlabeled competitors experiment, the same number of events was acquired for each dilution of competitors (10 million events) because the number of CFSE-labeled cells remained the same for each dilution. Ly-5.1 cells were identified by staining with mAbs to Ly-5.1 and CD4 or CD8; cells that stained positive for Ly-5.2 were excluded. Three million events/mouse were acquired to visualize responding cells. All mAbs were obtained from BD Pharmingen, except KJ1-26, which was obtained from Caltag Laboratories. Intracellular IFN-
staining was performed with the Cytofix/Cytoperm Plus kit (BD Pharmingen) after 5 h of in vitro stimulation with 1 µM OVA257264, 3 µM OVA323339, or 3 µM listeriolysin O (LLO)190201 peptides.
Bacterial loads
Three titrations of splenocytes from OT-II mice containing 1.5 x 106, 1.5 x 105, and 1.5 x 104 OVA-specific CD4 T cells were adoptively transferred into C57BL/6 mice. One day later, the recipients were infected with 1 x 105 CFU rLM-OVA, i.v. Bacterial loads were determined 3 days later by plating 10-fold serial dilutions of spleen and liver homogenates in 1% Triton X-100/PBS on brain heart infusion agar plates (Difco). Bacterial numbers were log transformed before statistical analyses were conducted by t tests (two tailed, 95% confidence intervals) using GraphPad Prism (GraphPad).
| Results |
|---|
|
|
|---|
To compare the proliferation of CD4 and CD8 T cells in the context of a natural infection, we developed an adoptive transfer system to visualize primary polyclonal T cell responses. CFSE-labeled CD4 and CD8 T cells from C57BL/6 mice were adoptively transferred into Thy-1.2 or Ly-5.2 congenic mice. For comparison, we also adoptively cotransferred CFSE-labeled monoclonal TCR transgenic CD4 and CD8 T cells from OT-II and OT-I mice (2 x 106 specific cells each), respectively. Recipient mice were subsequently infected with rLM expressing OVA (rLM-OVA). As seen before, OT-II T cells divided to a limited extent with a spectrum of responding cells in each division, while OT-I T cells divided extensively with most of the responding cells becoming CSFE negative, indicating seven or more divisions (Fig. 1A). However, much to our surprise, polyclonal CD4 T cells divided to the same extent as polyclonal CD8 T cells with most of the responding cells dividing seven or more times (Fig. 1B). Thus, we observed a difference in the extent of division between adoptively transferred monoclonal CD4 T cells and polyclonal CD4 T cells. Monoclonal CD4 T cells divided a limited number of times in response to infection, while all responding polyclonal CD4 T cells divided extensively. In contrast, both monoclonal and polyclonal CD8 T cells divided extensively in response to infection with the vast majority of responding CD8 T cells becoming CFSE negative (Fig. 1, A and B).
|
The CFSE-negative polyclonal CD4 and CD8 T cells had increased expression of the activation marker CD44 (Fig. 1D), CD25, and CD132 (data not shown) in comparison with the nonresponding CFSE-positive cells, consistent with them having responded to antigenic stimulation. We also compared the activation of host CD4 and CD8 T cells over the course of infection (Fig. 1E). Naive mice contained nearly equal percentages of activated/memory (CD44high) CD4 and CD8 T cells, 30 and 34%, respectively. By day 5 postinfection, the CD8 T cell subset had more activated cells (
65% CD44high) than the CD4 T cell subset (
45% CD44high). However, by the peak of the response on day 8, both CD4 and CD8 T cells contained
70% activated cells. These results are consistent with the above CFSE data showing that more donor CD8 T cells had accumulated by day 5, yet similar numbers of CFSE-negative CD4 and CD8 T cells were present by day 8 (Fig. 1C). Together, these results strongly indicate that, while more activated CD8 T cells are present early during the course of the immune response, the total CD4 T cell response is as robust as the CD8 T response.
In summary, the adoptive transfer of CFSE-labeled splenocytes enabled the simultaneous visualization of polyclonal CD4 and CD8 T cell responses during a primary LM infection. Our results show that polyclonal CD4 T cells, like CD8 T cells, divide extensively in response to LM infection. These results also reveal a discrepancy between the proliferative responses of adoptively transferred monoclonal and polyclonal CD4 T cells.
Adoptively transferred polyclonal T cells become activated and differentiate to the same extent as host T cells
We reasoned that the response of adoptively transferred polyclonal T cells was more likely to mimic the endogenous response of host T cells than the response of adoptively transferred monoclonal T cells. However, it is still possible that adoptively transferred T cells may become activated during ex vivo manipulation or may not engraft properly. We therefore examined whether the response of the adoptively transferred polyclonal T cells accurately reflected the response of the endogenous host T cells by comparing the activation and differentiation of the donor and host polyclonal T cells. Donor CD4 and CD8 T cells were activated to the same extent as recipient CD4 and CD8 T cells in naive mice and on days 5 and 8 postinfection, as measured by increased surface expression of CD44 (CD44high) (Fig. 2A). For example, at the peak of the response on day 8 postinfection, 73.4 and 70.0% of donor and host CD4 T cells were activated, respectively, and 72.1 and 70.4% of donor and host CD8 T cells were activated, respectively. No statistical differences occurred between donor and host T cells at any time point examined. These results demonstrate that the adoptively transferred T cells followed the same activation kinetics as endogenous T cells.
|
staining on days 5 and 8 postinfection (Fig. 2B). CD4 T cells mount a broad response to many different LM epitopes; however, we examined the response to LLO190201 because it is clearly immunodominant (41). The donor and host CD4 T cell responses to LLO190201 were nearly the same for both time points. Approximately 2% of both donor and host CD4 T cells were LLO190 specific on day 5. This increased to 9% at the peak of the response on day 8. Likewise, both the donor and host CD8 T cell responses to OVA257264 were almost identical. Approximately 1% of both donor and host CD8 T cells were OVA257 specific on day 5, increasing to
4% on day 8. No statistical differences were observed between donor and recipient T cells for either T cell subset on the two time points examined. Thus, the same levels of Ag-specific CD4 and CD8 T cells were generated from the donor and host T cell populations. The activation and Ag-specific response data show that adoptively transferred polyclonal T cells behave like the endogenous T cells in response to LM infection. Thus, CD4 T cells responding to infection most likely undergo extensive proliferation, as seen with adoptively transferred polyclonal CD4 T cells, rather than limited proliferation, as seen with TCR transgenic CD4 T cells.
Reducing the precursor frequency of monoclonal CD4 T cells results in a proliferative profile similar to polyclonal CD4 T cells
We considered the possibility that the limited division of the monoclonal CD4 T cells was due to clonal competition imposed by a high precursor frequency of cells with the same specificity. Therefore, we transferred 3-fold serial dilutions of OT-II TCR transgenic CD4 T cells into recipient mice and examined the extent of division of these cells in response to infection with rLM-OVA. Graded levels of transferred cells were present in the spleens of uninfected control mice, as expected. In mice infected with rLM-OVA, we found that the transfer of decreasing amounts of TCR transgenic CD4 T cells resulted in more extensive division of these cells (Fig. 3A). As seen before, the OT-II cells were nearly equally distributed in divisions 1 through 7 in mice that received the largest number of transferred cells (4.5 x 106 OVA-specific CD4 T cells) on day 8 postinfection. This trend was apparent as early as day 3 postinfection. In contrast, the majority of OT-II cells responding to infection in the mice that received the lowest number of transferred cells (1.5 x 105 OVA-specific CD4 T cells) were CFSE negative on day 8 postinfection, resembling the proliferative pattern of the polyclonal CD4 T cells seen in Fig. 1B. Similar results were also seen with DO11.10 cells, which are on the BALB/c background and are specific for the same CD4 T cell epitope of OVA as OT-II cells (Fig. 3). We have shown previously that the DO11.10 cells that proliferate in response to rLM-OVA infection do so in an Ag-specific manner because there is no proliferation of the TCR transgenic cells in response to infection with wild-type LM (11). Furthermore, use of the KJ1-26 clonotypic Ab in this model ensures that the CFSE-negative cells are indeed Ag specific.
|
High precursor frequencies of TCR transgenic CD4 T cells do not rapidly clear bacteria from the spleens and livers of mice
The limited proliferation of transferred monoclonal CD4 T cells could also be the result of less antigenic stimulation due to the rapid clearance of bacteria by the large number of Ag-specific clonal T cells present. Therefore, we examined the level of rLM-OVA recovered from the spleens and livers of mice receiving 1.5 x 106, 1.5 x 105, and 1.5 x 104 OVA-specific OT-II TCR transgenic CD4 T cells on day 3 postinfection. Similar numbers of bacteria were present in the spleens of mice receiving 1.5 x 106, 1.5 x 105, and 1.5 x 104 OT-II T cells, while bacterial loads were decreased by
5-fold in the livers of the mice that received 1.5 x 105 and 1.5 x 104 OT-II cells in comparison with those that received 1.5 x 106 OT-II cells (Fig. 4). This is consistent with the finding that CD8 T cells are more important than CD4 T cells for clearance of LM in the spleen, but that CD4 T cells are as protective as CD8 T cells in the liver (42). These results also indicate that the transfer of fewer numbers of TCR transgenic CD4 T cells and the corresponding greater proliferation of these cells lead to enhanced protection in the liver. More importantly, mice that received the higher number (1.5 x 106) of OT-II cells had the same number of bacteria in the spleen and more bacteria in the liver than mice that received fewer (1.5 x 105 OT-II cells). These results indicate that the decreased proliferation of TCR transgenic CD4 T cells with increasing numbers of cells transferred cannot be due to less antigenic stimulation.
|
Because the differentiation of naive T cells into cytokine-producing effectors has been linked to the procession through multiple rounds of division (13, 14, 15), and our results have shown that CD4 T cell division can be greatly limited by unusually high precursor frequencies, we examined whether high precursor frequencies limit the extent of CD4 T cell differentiation. We transferred 1.5 x 105 and 1.5 x 106 CFSE-labeled OVA-specific cells from OT-II or OT-I mice into congenic recipients and infected them with rLM-OVA. On day 8 postinfection, we analyzed the proliferation and IFN-
production of the TCR transgenic cells. We have seen previously that no IL-4+ or IL-2 single+ (IL-2+/IFN-
+ double-positive cells can be detected) CD4 T cells exist in the spleen following i.v. infection with LM (our unpublished results). In mice that received 1.5 x 105 OT-II cells, 26.8% of the responding cells (those that have divided at least once) differentiated into IFN-
-producing cells compared with only 10.7% in mice that received 1.5 x 106 OT-II cells (Fig. 5A). As expected, 98.1 and 97.2% of responding OT-I cells produced IFN-
in mice that received 1.5 x 105 or 1.5 x 106 OT-I cells, respectively (Fig. 5B). Therefore, nearly all of the CD8 T cells that were recruited into division differentiated into cytokine-producing effectors regardless of initial precursor frequency. The absolute number of Ag-specific donor CD4 T cells capable of secreting IFN-
was
4.1 x 104 and 2.1 x 104 in mice that received 1.5 x 105 and 1.5 x 106 OT-II cells, respectively. Thus, twice as many differentiated CD4 T cells were generated when the initial precursor frequency was 10 times lower. In contrast, the absolute number of Ag-specific donor CD8 T cells capable of secreting IFN-
was 2-fold less (2.2 x 106 vs 4.4 x 106) when the initial precursor frequency was 10-fold lower. These data show that the extent of differentiation of CD4 T cells in this system is inhibited by abnormally high precursor frequencies, and that the differentiation of CD4 T cells is inversely related to the initial precursor frequency.
|
| Discussion |
|---|
|
|
|---|
, tetramer staining, or tracking T cell clones (45). However, these techniques may underestimate the CD4 T cell response because the percentage of responding CD4 T cells that differentiate into cytokine-producing effectors may be lower than that of responding CD8 T cells (13, 16), the availability of MHC class II tetramers is limited, and the response to a few isolated epitopes may not reflect the CD4 T cell response as a whole. For example, for many years, the Ag-specific CD4 T cell response to LM infection was considered to be very weak because the frequency of CD4 T cells responding to all known LM epitopes was almost too low to study. However, a recently identified CD4 T cell epitope, LLO190201, reacts with more T cells than any LM-specific CD8 T cell epitope identified to date (41). In this study, we adoptively transferred naive polyclonal T cells and compared the CD4 and CD8 T cell responses during a primary LM infection. We found that both CD4 and CD8 T cells divide extensively following infection, with nearly all responding cells in both subsets becoming CFSE negative (dividing seven or more times). In addition, the absolute number of responding CD4 and CD8 T cells was similar at the peak of response on day 8 after infection. Combined, these data indicate that the CD4 T cell response to primary LM infection is as robust as the CD8 T cell response in terms of the total number of responding cells generated.
The extent of polyclonal CD4 T cell division reported in this study differed from previous findings that adoptively transferred TCR transgenic CD4 T cells undergo significantly limited division in the spleen (11, 12, 46) in comparison with adoptively transferred TCR transgenic CD8 T cells in response to infection (1, 2, 3, 11). What accounts for the difference in the extent of monoclonal and polyclonal CD4 T cell division? Our results show that the proliferation of monoclonal CD4 T cells is limited due to an unusually high precursor frequency. We titrated the number of adoptively transferred TCR transgenic CD4 T cells and found that the extent of proliferation increased as the precursor number decreased. Our results are consistent with the findings that recruitment of CD4 T cells into division and burst size are inversely related to precursor frequency (47). Furthermore, we demonstrated that when the number of precursors was decreased to
1 x 104 cells (2 x 105 transferred cells with
5% engraftment), which is still 100 times greater than the estimated precursor frequency for naive Ag-specific T cells (48), the majority of cells divided seven or more times in response to infection. Combined with the results of our polyclonal T cell experiments, we believe that this extensive division of CD4 T cells reflects the natural response of CD4 T cells to LM infection. Although we dont know whether all of the CFSE-negative cells are Ag specific, several lines of evidence suggest that they are. First, we have shown previously that the DO11.10 cells that proliferate in response to rLM-OVA infection do so in an Ag-specific manner because there is no proliferation of the TCR transgenic cells in response to infection with wild-type LM, thus ruling out bystander proliferation. Furthermore, OT-1 cells and P14 TCR transgenic CD8 T cells dont proliferate in response to infection with Listeria strains that dont contain the peptides for which the transgenic T cells are specific (49).
Both CD4 and CD8 T cells have been shown previously to compete for access to Ag on APCs in several different experimental systems (47, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60). In this study, we examined the effects of high precursor frequencies on the proliferation and differentiation of CD4 T cells. When the precursor frequency of Ag-specific CD4 T cells is high, fewer CD4 T cells undergo extensive division and a smaller percentage of responding CD4 T cells produce IFN-
than when the precursor frequency of these cells is low. A previous report demonstrated that clonal competition serves as a mechanism to control the overall size of the CD4 T cell response so that the same number of Ag-specific CD4 T cells is generated despite large differences in initial precursor frequencies (47). However, we show that clonal competition due to high precursor frequencies causes fewer differentiated CD4 T cells to be generated. Thus, while the same number of responding CD4 T cells may be generated in the face of clonal competition, fewer of these cells will be differentiated effectors. A number of reports have proposed that procession through multiple rounds of division is necessary for differentiation (13, 14, 61). However, a study by Laouar and Crispe (62) suggests that the relationship between proliferation and differentiation is correlative and not mechanistically linked. Our results show that during an infection, most, if not all, of the IFN-
production comes from the population of CD4 T cells that have divided multiple times. Moreover, we see a greater percentage of differentiated CD4 T cells when these cells progress through more divisions. CD4 T cells at a high clonal precursor frequency have been shown to undergo more rounds of division in the presence of more Ag (47), and numerous reports have suggested that T cell competition occurs at the level of Ag-APC (47, 51, 52). Thus, we propose that adoptively transferred TCR transgenic CD4 T cells have less access to Ag, and therefore have inhibited proliferation and differentiation, due to clonal competition.
In summary, we showed that the natural proliferative response of CD4 T cells to infection is more extensive than previously described, with the majority of responding CD4 T cells dividing seven or more times. In addition, we demonstrated that high precursor frequencies inhibit the proliferation and differentiation of CD4 T cells due to clonal competition. Thus, the adoptive transfer of TCR transgenic CD4 T cells to study the response of CD4 T cells may significantly underestimate the extent of division and differentiation of the natural response to the same Ag. The effects of clonal competition on CD4 T cell proliferation and differentiation may need to be considered in designing rational vaccine strategies, particularly ones that incorporate multiple boosts in which pools of memory cells may compete with each other.
| Acknowledgments |
|---|
| Disclosures |
|---|
|
|
|---|
| Footnotes |
|---|
1 This work was supported by National Institutes of Health Grant AI-45025 (to H.S.). ![]()
2 Current address: Cellular Immunology Section, Vaccine Research Center, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20892. ![]()
3 Address correspondence and reprint requests to Dr. Hao Shen, Department of Microbiology, University of Pennsylvania School of Medicine, Philadelphia, PA 19104. E-mail address: hshen{at}mail.med.upenn.edu ![]()
4 Abbreviations used in this paper: LM, Listeria monocytogenes; LLO, listeriolysin O. ![]()
Received for publication May 13, 2004. Accepted for publication December 23, 2005.
| References |
|---|
|
|
|---|
- and
-chain genes under the control of heterologous regulatory elements. Immunol. Cell Biol. 76: 34-40. [Medline]This article has been cited by other articles:
![]() |
C. A. Shaw and M. N. Starnbach Both CD4+ and CD8+ T Cells Respond to Antigens Fused to Anthrax Lethal Toxin Infect. Immun., June 1, 2008; 76(6): 2603 - 2611. [Abstract] [Full Text] [PDF] |
||||
![]() |
J. K. Whitmire, N. Benning, B. Eam, and J. L. Whitton Increasing the CD4+ T Cell Precursor Frequency Leads to Competition for IFN-{gamma} Thereby Degrading Memory Cell Quantity and Quality J. Immunol., May 15, 2008; 180(10): 6777 - 6785. [Abstract] [Full Text] [PDF] |
||||
![]() |
A. J. Wolf, L. Desvignes, B. Linas, N. Banaiee, T. Tamura, K. Takatsu, and J. D. Ernst Initiation of the adaptive immune response to Mycobacterium tuberculosis depends on antigen production in the local lymph node, not the lungs J. Exp. Med., January 21, 2008; 205(1): 105 - 115. [Abstract] [Full Text] [PDF] |
||||
![]() |
D. A. Blair and L. Lefrancois Increased competition for antigen during priming negatively impacts the generation of memory CD4 T cells PNAS, September 18, 2007; 104(38): 15045 - 15050. [Abstract] [Full Text] [PDF] |
||||
![]() |
P. Marrack Conference Summary Proceedings of the ATS, August 15, 2007; 4(5): 469 - 471. [Full Text] [PDF] |
||||
![]() |
B. Alabyev, Z. S. M. Rahman, and T. Manser Quantitatively Reduced Participation of Anti-Nuclear Antigen B Cells That Down-Regulate B Cell Receptor during Primary Development in the Germinal Center/Memory B Cell Response to Foreign Antigen J. Immunol., May 1, 2007; 178(9): 5623 - 5634. [Abstract] [Full Text] [PDF] |
||||
![]() |
Z. Garcia, E. Pradelli, S. Celli, H. Beuneu, A. Simon, and P. Bousso Competition for antigen determines the stability of T cell-dendritic cell interactions during clonal expansion PNAS, March 13, 2007; 104(11): 4553 - 4558. [Abstract] [Full Text] [PDF] |
||||
![]() |
M. L. Ford, B. H. Koehn, M. E. Wagener, W. Jiang, S. Gangappa, T. C. Pearson, and C. P. Larsen Antigen-specific precursor frequency impacts T cell proliferation, differentiation, and requirement for costimulation J. Exp. Med., February 19, 2007; 204(2): 299 - 309. [Abstract] [Full Text] [PDF] |
||||
| ||||||||||||||||||||||||||||||||||||||||||||||