MP CD4 T cells from normal mice (i.e., mice housed in specific pathogen-free facilities) proliferate quite rapidly. BrdU labeling reveals that ~10% of these cells from lymph nodes take up BrdU in a single day and more than 30% of MP cells are Ki-67+; on the basis of our estimate that the great majority of Ki-67+ cells have divided within the past 3 d, this implies that more than 30% of lymph node MP cells divide in 3 d, a result that is confirmed by more extended BrdU labeling. The frequency of Ki-67bright MP cells is somewhat less in the spleen. By contrast, NP cells take up BrdU at ~0.1% per day and very few are Ki-67+.
What drives the rapid proliferation of the MP cells has been a matter of uncertainty. Some have concluded that their proliferation is driven by TCR engagement on the basis of transfer to lymphopenic hosts, where it is observed that by 7 d after transfer the majority of the surviving cells have divided seven times or more. Zamoyska and colleagues 
and Leignadier et al. 
have used a tetracycline/off system to delete TCR from mature T cells. In both instances, deleting TCR resulted in a substantial diminution in the proportion of CD44bright
CD4 T cells that had gone through multiple divisions when transferred to lymphopenic recipients. Similarly, anti-MHC class II antibody blocked expansion of CD4 T cells introduced into neonatal recipients 
, and we showed here that the rapid proliferation of MP CD4 T cells introduced into Rag2−/−
recipients was completely inhibited by the anti-MHC class II antibody Y3P. Furthermore, Surh and colleagues reported that the rapid proliferation of CD4 T cells that occurred when these cells were introduced into scid
mice was largely lost if the scid
recipients were GF 
. As a group, these observations clearly indicate that in severely lymphopenic settings, expansion of MP cells depends on TCR recognition of peptide/MHC complexes.
The results we present here indicate that only a portion of the transferred MP cells undergo this striking proliferation. When we sequenced CDR3 segments from the Vβ2/Jβ1.1 TCRs 3 d after transfer of MP cells into Rag2−/−
recipients, we found only three sequences among the rapidly dividing cells, whereas there were 37 sequences among those that had not divided, implying that the rapid proliferation was a property of a limited set of cells among the transferred MP population. This result is consistent with the observation that naïve CD4 T cells from most TCR transgenic donors fail to rapidly proliferate on transfer to lymphopenic recipients 
and on our immunoscope analysis of TCR Vβ complexity in Rag2−/−
recipients of numbers of CD4 T cells varying from 10 million to 10,000 
suggesting that only ~3% of the transferred cells undergo rapid expansion. It is interesting that two of the sequences represented frequently in the dividing cells were also found in the nondividing population, implying that not all cells of the same specificity are stimulated in a lymphopenic environment.
However, the results obtained by the study of transfer to lymphopenic environments do not appear to be a valid representation of the mechanisms underlying the rapid proliferation of MP cells in situ. Indeed, survival of MP cells in lymphocyte-sufficient settings has been reported to not require expression of TCRs. Furthermore, there is a large literature demonstrating that survival of antigen-specific memory cells arising during immunization does not require TCR engagement, but rather depends upon the availability of cytokines, particularly IL-7 and IL-15 
. However, we wish to point out that the analysis of antigen-specific memory cells emerging from intentional immunization may not necessarily represent what governs the proliferative behavior of MP cells.
Indeed, both the work presented here and recent studies analyzing antigen-specific CD4 memory T cells at varying times after priming show that antigen-specific memory cells emerging from intentional immunization divide relatively slowly compared to MP cells. Lenz at al. 
infected mice with LCMV. 50 d later, a 7-d exposure to BrdU resulted in only 15% of BrdU+
spleen cells among those capable of producing interferon gamma (IFNγ) in response to challenge with two different LCMV peptides. Purton et al. 
transferred TCR transgenic SMARTA CD4 T cells, specific for an LCMV epitope, into C57BL/6 mice that were then infected with LCMV. 72 d after infection, 12% of the transgenic cells took up BrdU during a 5-d labeling period. Jenkins and colleagues 
infected mice with Listeria monocyogenes
expressing an ovalbumin peptide (LM2W1S). 40 d after infection, the mice received BrdU for 14 d; among spleen and lymph node CD4 T cells capable of binding an ovalbumin tetramer, only 11.5% were BrdU+
Our results are consistent with these reports. We infected C57BL/6 mice with LCMV. Fifteen and 60 d later, the frequency of Ki-67+
cells among tetramer+
cells was measured. At 15 d, 8% of the CD44bright
cells were Ki-67+
; at 60 d, ~7%. Collectively, these studies indicate that after the expansion phase following immunization is complete, antigen-specific CD44bright
CD4 T cells divide at a rate of <1% to ~2.5% per day. The possibility that MP cells and authentic memory cells might represent distinct cell types, or rather that the MP pool contains both authentic memory cells and another cell population, was also suggested by our prior study in SW GF mice that showed that their MP CD4 T cells proliferated at a rate similar to MP cells from conventional donors 
. We have examined this point in greater detail here in GF and conventional C57BL/6 mice and confirm that the proportion and absolute number of non-Treg CD44bright
CD4 T cells from peripheral and mesenteric lymph nodes of GF mice are similar to those from conventional mice as are the proportion and number of proliferating MP cells. It should also be pointed out that prior studies of GF mice maintained on elemental diets (i.e., antigen-free mice) had shown the presence of substantial numbers of activated CD4 T cells, equivalent in frequency to those in conventional mice 
. While these studies were carried out before the availability of the reagents now used to classify MP cells, they strongly suggest that antigen-free mice have similar numbers of MP CD4 T cells as do conventional mice and thus support the concept that foreign (including commensal) antigens are not critical to the emergence of the majority of MP cells.
Here we have shown that the in situ proliferation of MP cells is not inhibited by anti-MHC class II antibody, using a reagent that strikingly inhibits the proliferation of antigen-specific cells in response to antigen challenge and that blocks the rapid proliferation of MP cells transferred to lymphopenic recipients. Rather, we observe that anti–IL-7Rα antibody diminishes, but does not abolish, proliferation of MP cells, implying that IL-7 or TSLP plays a role in this proliferation.
An alternative way to examine the issue of whether the rapid proliferation of these cells represents an antigen-driven response, during which one would anticipate that limited numbers of precursors give rise to bursts consisting of multiple divisions, is to examine the TCR sequence diversity of proliferating MP cells and to compare that to the sequence diversity of quiescent MP cells. If MP proliferation was primarily due to burst-like clonal expansion stimulated by exposure to antigen, it would be expected that the sequence diversity of the proliferating cells would be substantially less than that of the quiescent cells. Indeed, when we studied proliferating and nonproliferating MP cells in lymphopenic recipients, this is precisely what we observed.
We examined CDR3 sequences from Vβ2/Jβ1.1 and Vβ4/Jβ1.1 MP CD4 T cells that were Ki-67bright or Ki-67negative. We chose to limit our study to these two TCR Vβ sets so that we could sample a larger proportion of these defined subrepertoires than we could with the same number of sequences of total CD44bright Ki-67bright cells. As an estimate of the number of cells under study, we used the following considerations. The total number of lymph node CD4 T cells is ~8 million. ~15% of these cells are CD44bright, or ~1.2 million. Of these, ~half are CD25+, so that MP cells constitute ~600,000; ~5% express Vβ2 or Vβ 4, or ~30,000. Of the Vβ2 or Vβ 4 expressing cells, ~10% are Jβ1.1, or ~3,000. The Ki67bright cells are ~10% of the MP cells so that the total number of Ki-67bright Vβ2/Jβ1.1 or Vβ4/Jβ1.1 MP cells in all the lymph nodes of the animal is ~300. Thus, the maximum number of unique sequences would be 300 in this cell population; this would be substantially less if repetitive sequences existing among these cells, which would be anticipated on the basis of the likelihood that the generation of MP cells from naïve precursors involved clonal expansion. Thus, in our initial analysis, involving >200 sequences from the CD44bright Ki-67bright Vβ2/Jβ1.1 cells of mouse 1, our sample, while not complete, is quite substantial. Even the samples of 70 to 80 sequences in the other cases are sufficient to provide useful information about complexity, as judged by our observations of multiply repeated sequences.
A further point is our reliance on CDR3 sequences from the β chain of the TCR as a clonal marker. It is possible that we have overestimated the frequency of repeats since there may be occasions in which the same Vβ is used with different Vα's, but we suspect that in the vast majority of cases the CDR3 sequence of the TCR β chain is indeed a clonal marker.
Our results indicate that the distribution of sequences in the Ki-67bright and Ki-67negative populations was not markedly different. If we made the assumption that any sequences that occurred many times among the Ki-67bright cells represented cells that had recently been in a burst-like expansion, then ~25% of the Ki-67bright cells would be judged to be in such bursts. To obtain this estimate we arbitrarily assigned any sequence that occurred four or more times among the Ki-67bright cells to the set that occurred “many” times. However, this could easily be an overestimate depending on how one interprets those instances in which a similarly high frequency of the same sequence was found among Ki-67negative cells. On the one hand, this might reflect a large clone in which cell division occurred on a stochastic basis so that the frequency of the clone was similar among the dividing and nondividing cells. If we make this assumption, then the proportion of Ki-67bright cells that were in bursts becomes ~7%. Alternatively, instances in which a sequence found in the Ki-67bright cells was equally (or over-represented) among the Ki-67negative cells may represent the late-phase of a burst episode in which a portion of the cells had already stopped dividing and lost Ki-67 expression but others continued to divide. Overall, we conclude that the sequence data are consistent with a minority of the Ki-67bright cells being part of burst; whether that minority is small or considerable cannot easily be determined. However, when taken together with the failure of anti-class II antibody to block proliferation of MP cells, it is reasonable to conclude that the proportion of Ki-67bright cells that are part of burst-like expansion is quite small.
There may be circumstances in which clonal expansion/burst-like antigen-driven proliferation plays a much great role than is found among MP cells from normal mice. It has been argued that the dynamics of MP cells from SIV-infected macaques, in which proliferative rates are far higher than proliferative rates of MP cells from noninfected macaques, is best explained by multiple overlapping burst episodes of several cell divisions occurring within a brief period of time 
. These cells, which exist in a highly inflammatory setting, may well show enhanced sensitivity to their cognate antigens or to self-peptide/MHC complexes. A detailed analysis of their sequence complexity and of the complexity of MP cells derived from chronically infected individuals or individuals that were in a state of chronic inflammation would help to clarify this point.
As discussed above, we found that GF and conventional C57BL/6 mice have comparable numbers of MP CD4 T cells in their peripheral and mesenteric lymph nodes and these cells display similar proliferative rates. We sequenced CDR3 gene segments from Vβ2/Jβ1.1 Ki67bright and Ki-67negative MP cells of two GF mice and found that they, like the comparable cells from conventional mice, were very similar in their sequence diversity. One could argue that GF mice are not antigen free, although there are reports that antigen-free mice show normal numbers of “activated” CD4 T cells. Nonetheless, there can be little doubt that GF mice have a much reduced antigenic experience. If the MP cells of GF mice represent clonal expansions because of an extremely limited set of naïve cells responding to a comparably limited set of antigenic stimuli, it would be anticipated that the MP cells from GF mice would have a much more restricted repertoire than MP cells from conventional mice. While our limited number of sequences may not be sufficient to reach a definitive conclusion on this point, there does not appear to be any major difference in the degree of diversity of the Ki-67bright or Ki-67negative MP cells from the two sets of donors.
On the basis of these observations, we propose that MP cells may be a more diverse population than had been considered and that only some of these cells may have emerged by exposure to conventional foreign antigens. The maintenance of most MP cells and their rapid proliferative rate appear to be largely dependent on cytokines. The origin of the infrequently occurring proliferation bursts remains to be clarified, but an obvious possibility is through recognition and response to self-peptides on competent antigen-presenting cells. It should be pointed out that diversity in CD8 T cells has also been described, with one population being designated “bystanders” and that such cells take on a memory phenotype in mice deficient in the transcription factor KLF2, the signaling kinase itk, or the histone acetyltransferase CBP 
. Whether such “bystander” CD8 cells bear a relationship to the rapidly dividing CD4 MP cells discussed here remains to be determined.
Overall, one may ask what is the function of the large set of MP cells in normal mice? We have proposed 
that they represent a pool of cells capable of making a rapid effector response to cross-reactive antigens of pathogens during a period in which the naïve cells proliferate and differentiate. MP cells might play an even more important role in instances in which naïve cells are limiting and no “authentic” memory cells are specific for an introduced pathogen, such as might be the case in aged individuals. Devising models in which these cells are absent will be essential to testing their function. Finally, why proliferative rates of authentic memory and MP cells are different is not clear.