In a number of studies it has been observed that mice with genetic mutations causing enhanced BCR signaling have substantially reduced numbers of follicular B cells. Examples of this phenomenon include mutations that ablate, Lyn [27
] or SHP1 [38
], and a mutation that enhances CD45 activity [39
]. The first two of these molecules act in a feedback inhibitory pathway in which Lyn phosphorylates a tyrosine residue in the cytoplasmic domain of CD22, leading to recruitment of the protein tyrosine phosphatase SHP1, which inhibits BCR signaling at a relatively upstream step. A major function of CD45 is to maintain Src-family tyrosine kinases in a primed state by dephosphorylating a C-terminal negative regulatory tyrosine phosphorylation, and in this way it is thought to regulate the magnitude of signaling by the BCR. The reason for increased BCR signaling caused by these mutations leading to a substantial decrease in the size of the follicular B cell compartment is not known. In the present study, we used in vivo
BrdU labeling and flow cytometry to measure the numbers and turnover rates of various immature and mature B cell subpopulations in wild type and Lyn−/−
mice. We then applied our most recent mathematical model of B cells development and maturation to these data sets to gain insight into how Lyn deficiency affects B cell maturation and survival. The results obtained from the model indicate that the B cells in Lyn-deficient mice exhibit perturbations at several maturation and survival steps, with an especially dramatic increase in the death rate of follicular B cells. These alterations likely underlie the large decrease in the number of follicular B cells in these mice.
The mathematical model of B cell maturation and survival dynamics that we had previously derived from analysis of wild type mice was found to apply to the data sets developed with Lyn−/−
mice, with a few minor alterations. First, we found that it was necessary to assume that the BrdU labeling protocol used in these experiments did not achieve 100% labeling of proliferating B cell precursors, which was consistent with the experimental observations. Second, it was clear that the follicular B cells in both wild type and Lyn−/−
mice were comprised of cells that were not uniform in their half-lives, but rather consisted of a minority of cells that were turning over relatively rapidly and a majority of cells that had a turnover that was slow enough it could not be estimated from these experiments. The long-lived follicular B cells included about 85% of follicular B cells in Lyn−/−
mice and about 60% of the much greater number of follicular B cells in wild type mice. For the purposes of modeling these data, the very slowly turning over B cells were considered to be "static" cells, or non-dividing, non-dying cells. Please note that we only refer to these cells as "static” because they were not labeled in the 8 days of the experiment; estimating how long they actually live would require longer experiments. Interestingly, we found that such static cells are necessary to explain incomplete cell labeling in some of our other studies [36
] and also in our current work in progress (unpublished data). Moreover, after carefully examining our results in the study with the Allman lab ( in [34
] and in [35
]), we see that even there, none of the populations has reached 100% labeling by day 7. Thus, this behaviour seems to be the rule rather than the exception.
When the above two characteristics were incorporated into our mathematical equations of B cell maturation and turnover, the range of maturation rate and death rate parameters needed to model the new wild type data set were similar to the corresponding parameters calculated from earlier data sets [35
]. Moreover, it was also possible to obtain compatible sets of maturation rate and death rate parameters with the data from the Lyn−/−
mice. Some of those parameters were similar to the wild type parameters, but several of them were strikingly different. Thus, we were able to model successfully the population dynamics of B cell populations from Lyn−/−
mice and assess how deficiency of Lyn affects B cell maturation and survival in the spleen.
The earliest B cell maturation state in the spleen, the T1 transitional B cells, labeled with BrdU with very similar kinetics in wild type and Lyn−/−
mice, and our mathematical model found that the rate of maturation from bone marrow immature B cells to splenic T1 cells of these two mouse strains was identical (parameter δBM1
in ). These results are consistent with data from mixed bone marrow chimeras, which found that there was not a statistically significant competitive disadvantage of Lyn-deficient B cell precursors accessing the T1 population (Gross et al. manuscript submitted). Those mixed bone marrow chimera experiments did detect a competitive disadvantage of Lyn-deficient B cells in accessing the next maturation state, the T2 stage in the spleen, and moreover the steady state size of this population was decreased by approximately 3-fold in Lyn−/−
mice. Our mathematical modeling previously concluded that wild type T2 cells arise both from splenic T1 cells and from bone marrow immature B cells that have reached the T2 phenotype in that location before exiting to the spleen [34
]. In contrast, we found here that Lyn-deficient T2 cells only arose from splenic T1 cells with minimal contribution from bone marrow T2-like cells. This difference may in part explain the decreased size of the T2 population in Lyn−/−
mice. This finding may be tested by following the kinetics of reconstitution of bone marrow chimeras, where B cells – whether wild type or Lyn−/−
– are allowed to develop in irradiated hosts. Following the appearance of the immature populations in a relatively homogeneous wave, as done by Allman et al
], will show how Lyn−/−
and wild type B cells compare in terms of reconstitution kinetics, vs. the steady state analysis on which our model is based. The model's predicted kinetics of growth of the modeled B cell populations, simulated with the best fit parameter values for wild type and Lyn−/−
cells, are presented in . These kinetics clearly show a slower growth of the T2 compartment in Lyn−/−
mice compared with its growth in the wild type mice, due to the direct differentiation of immature BM B cells into the T2 compartment, which is almost absent in the Lyn−/−
The predicted kinetics of growth of the modeled B-cell populations
Once Lyn-deficient cells reached the T2 stage, they were able to mature to the follicular population at a similar but somewhat higher rate in Lyn−/−
mice than in wild type mice (δ2m
parameter in ). Whereas the rate of maturation of T2 cells was altered by Lyn deficiency to a modest degree, there was a much greater impact on alternative fates of this cell type: in Lyn−/−
mice, the T2 cells that didn’t mature all died, and none accessed the T3 anergic population [40
]. In contrast, wild type T2 cells that did not mature were most likely to become anergic T3 cells, and only a minor fraction died directly from the T2 population (δ23
parameters in ). In order to validate the different developmental fates from T2 compartment predicted by our model, one may isolate T2 cells from wild type vs. Lyn−/−
mice, transfer them into wild type recipient mice (distinguished with the Ly5 allelic difference), and measure what fraction of the transferred cells turn into Fo vs. T3 B cells. A more complicated experiment may be possible if a way is found to conditionally delete the Lyn
gene at various developmental stages, as then we could follow the developmental fates of the progeny of the Lyn-deleted cells.
The most dramatic change in B cell population dynamics as indicated by our mathematical model was in the fate of those follicular mature B cells that were turning over moderately fast. In wild type mice, these less stable follicular B cells shifted to the anergic T3 phenotype and then eventually died from this population. In contrast in Lyn−/− mice, the less stable follicular B cells died directly. Interestingly, the numbers of follicular B cells in Lyn−/− mice are increased by approximately 10-fold by expression of a B cell-specific Bcl-2 transgene or by knockout of the pro-apoptotic BH3-only factor Bim (Gross et al, manuscript submitted), which is consistent with a high death rate contributing to the decreased numbers of follicular B cells in Lyn−/− mice.
We previously found that the elevated BCR signaling in B cells in Lyn−/−
mice is most profound in T3 and follicular B cell populations, and is elevated to a lesser degree in T1 and T2 transitional B cells [30
]. Thus, an attractive hypothesis to explain the large increase in the death rate of follicular B cells resulting from Lyn deficiency is that the dying cells are those cells with a low degree of self-reactivity that, combined with strongly elevated BCR signaling, leads to clonal deletion, whereas in wild type mice, this degree of self-reactivity instead leads to clonal anergy and acquisition of the T3 phenotype. This hypothesis may be validated by using Ig transgenic mice with a weak self-reactivity, crossed with Lyn-deficient mice, and examining how the developmental kinetics would differ between Lyn−/−
and wild type mice with this BCR transgene.
According to this hypothesis, T2 cells with a moderate degree of self-reactivity die directly and those with a lower degree of self-reactivity mature to the follicular B cell type. In wild type B cells, maturation to the follicular stage is accompanied by multiple developmental changes in signaling that balance enhanced positive signaling with increased activity of the inhibitory Lyn-CD22-SHP1 pathway, resulting in a threshold behavior in which low levels of BCR signaling are dampened out, but the B cell can still respond to foreign antigen in sufficient amounts [30
]. Thus, wild type follicular B cells can either adapt to a low level of self-reactivity or can enter the anergic population, depending on the degree of self-reactivity. In Lyn−/−
B cells, however, the developmental changes increasing BCR signaling are unopposed by SHP1 and so a low degree of self-reactivity induces excessive BCR signaling leading to clonal deletion. A related hypothesis may explain the low access of Lyn-deficient T2 cells to the T3 anergic population, that is, as the cells begin to transition to the T3 stage, their signaling becomes magnified and this leads to rapid death before the T3 phenotype is fully realized. According to this view, access to the T3 population directly from the T1 population (δ13
in ), rather than from the T2 or Fo subsets, may be possible in Lyn−/−
mice because elevated BCR signaling is only manifest in later stages, giving more time for other negative regulatory mechanisms to balance off BCR signaling induced by a low degree of self-reactivity.
It should be noted that Lyn−/−
mice have elevated levels of BAFF [41
], a TNF family cytokine which promotes survival of T2, T3 and follicular B cells and may promote maturation of transitional B cells to mature B cells [42
]. It may be that the increased death rates of Lyn−/−
T2 and follicular B cells would be further increased if BAFF levels were not elevated in these mice. Elevated BAFF levels have previously been shown to support survival of anergic B cells [43
] and thus the increased BAFF levels in Lyn−/−
mice may explain the decreased death rate we observed in the T3 population (parameter μ3
Interestingly, the percent of "static" cells – cells that did not become labelled during the 8 days of the experiments – was higher in Lyn-deficient mice. The low degree of self-reactivity that is sufficient for Lyn-deficient cells to mature may thus not be sufficient to maintain a normal homeostatic division rate in this population. In summary, our previously derived mathematical model of B cell population dynamics was applied to analysis of the alterations in B cell maturation and survival in Lyn-deficient mice and was found to apply well to this substantial perturbation of B cells and to provide insights into the nature of those perturbations. The maturation rate and death rate parameters determined by the model were generally in accord with experimental observations regarding changes in the importance of negative regulation by Lyn as B cells mature and suggested reasonable hypotheses regarding how such changes might impact mechanisms for tolerizing self-reactive B cells in the periphery. These results provide insights into how alterations in BCR signaling affect B cell population dynamics and tolerance induction, although how tolerance breaks down in Lyn−/− mice to give spontaneous lupus-like autoimmunity remains to be elucidated.