An increasingly sophisticated understanding of the human CD27+ class switched Bc recall response is emerging from studies identifying subpopulations of activated human mBc. These responses are known to be heterogeneous4,11
, with antibody secreting cells, cytokine secreting cells45
, antigen presenting and co-stimulatory20
cells, and a mBc pool46
all emerging within the same memory recall response. Understanding the division of labor between the various activated mBc subsets, particularly when altered by adjuvant during vaccination, may allow us to bias the response to favor one set of functions. It is for this reason that we focused on TLR-9 activated CD27+
mBc, as CpG ODN adjuvants are being actively developed to augment vaccine responses47,48
Previous transcriptome analyses of Bc subsets have examined differential gene expression in mouse naïve and memory subsets from spleen, or human Bc from tonsil49,50
, identifying different numbers of genes or probes (50-450 human genes, ~3,000 murine genes, ~5,500 murine probes) differentially expressed between Bc subsets. Our results demonstrated ~2,000 differentially expressed genes between the pairs of the activated memory Bc subsets at a FDR = 0.03. There are several key differences in study methodology that account for this difference with the human studies. First, a number of these papers used a mathematical pattern recognition method which did not specify a false discovery or level of statistical significance, so the quantity of differentially expressed genes are not directly comparable. Also, these studies did not examine the differences between dividing, antibody-secreting, and non-dividing subpopulations of mBc. In contrast, we report transcriptome analysis on subpopulations of stimulated mBc based on generation and the expression of CD27. In addition, many of these reports used an older gene array with 12,000 probe sets, compared to our use of second generation gene arrays with ~54,000 probe sets. Indeed, many B cell and cell division specific genes lacked corresponding probes in the older arrays. Finally, our results suggest a novel functional division between the proliferating and undivided populations of CpG activated, class switched mBc, as well as several potential feedback mechanisms that may regulate the partitioning of CpG activated, class switched mBc into antibody secreting, receptor editing, and antigen-presenting phenotypes.
Several groups have described patterns of cytokine secretion by activated murine and human Bc in vivo
and in vitro3,51
. Stimulation by synthetic CpG-B class ODN, such as the CpG2006
used in our experiments, has been reported to induce up-regulation of activation markers, IL-12, and IL-6 by Bc in bulk culture52
. We found CpG-induced IL-6, IL-7, IL-15, and IL-24 transcripts at the highest levels in the activated but undivided Bc population. IL-24 belongs to the IL-10 family of cytokines, and has been reported to inhibit plasma cell differentiation of human germinal center Bc53
. It is notable that IL-6 and IL-7 were the other predominant cytokine genes highly expressed in the undivided sub-population. IL-7 expression and signaling is associated with recombination activating genes in germinal center Bc54
, and IL-6 acts as an autocrine growth and differentiation factor. However, in our experimental system, CpG-stimulation of class-switched mBc does not appear to induce strong, polarizing cytokine secretion associated with classical Breg functions of Th1 versus Th2 immune response deviation3
Although derived from an in vitro system, our finding of contemporaneous transcription of IL-6, IL-7, and IL-24 genes do suggest a possible mechanism for activated but undivided mBc to regulate CD27lo sub-populations within a germinal center or T cell independent lymphoid follicle. Active gene expression and secretion of these cytokines by undivided Bc could regulate Bc subsets expressing high levels of the IL-6R, IL-7R, and IL20/22R complexes, leading to suppression of plasma cell differentiation while supporting proliferation and receptor editing in a subset of the CpG activated Bc. Consistent with this hypothesis is our finding that CD27lo cells are a proliferating but non-antibody secreting population, with a gene expression pattern suggesting receptor editing and affinity maturation potential. The availability of such adaptive mechanisms after antigen-independent Bc activation also suggests a linkage between innate and adaptive Bc responses, suggesting further mechanisms for the actions of vaccine adjuvants, and targets for future in vivo studies.
Of note, we also found that the undivided cells were not inactive, but appear to have an antigen presentation/co-stimulation phenotype, as suggested by high transcript levels of class II HLA antigens and CD83. We show that these transcripts were generally downregulated by CpG+CK, with intermediate levels in the proliferating CD27lo
population. The exceptions are CD80 and CD86 whose transcripts are increased in CD27lo
cells. In addition to their role in T cell co-stimulation, these proteins are upregulated on CpG-activated mBc and, when engaged can increase the antibody production55
. This suggests that CD80 and CD86 may play a role in activating antibody production within the CD27lo
population. This might provide a pathway for CD4 potentiation of antibody secretion during the transition from T cell-independent CpG activation to a more T cell-dependent adaptive immune response.
Also of interest, we found that the CD27lo
Bc subpopulation expresses AICDA and precedes CD27hi
cells in Bc development. Transcriptome profiling by others found AICDA expression in bulk cultures of stimulated mBc56
. Our study points to the CD27lo
subpopulation as being the primary producers of AICDA in such systems. CpG is known to stimulate proliferation in CD27-
mBc population found in healthy human subjects and enriched in SLE patients57
. This Bc population was also reported to be FCRL4-
, while the proliferating CD27lo
population that we describe here had a higher FCRL4 expression compared to expanding CD27hi
cells. Like FCRL4+
Bc identified in other studies2
, the CD27lo
cells also expressed higher amounts of AICDA, SOX5, and ITGAX, than CD27hi
cells, while other markers of FCRL4+
cells (RUNX2, CCNB2, and TNFSF11) were not co-expressed. Thus, while there are similarities between the CD27lo
subpopulation in CpG+CK-stimulated mBc and the populations described by others, the CD27lo
transitional phenotype described here is unique.
Finally, our work suggests another potential point of interaction between co-cultured CpG-stimulated mBc subsets that could stimulate further studies. IL-6 receptors and IL6ST were upregulated in CD27hi
subpopulation, indicating a mechanism through which the activated but undivided Bc subpopulation, which produces IL-6 transcripts, might support plasmablast development. Others have proposed a Bc IL-6 signaling autocrine loop58
, although it is beyond the scope of this study to confirm it in this system.
The gene regulatory network model generated from our data implies a dependency of CD27lo
subpopulation development and AICDA expression upon NF-κB activation. Although receptor editing in early Bc has been associated with NF-κB activation36
, it is difficult to assess alterations of NF-κB signaling in mBc, as early Bc development can be severely compromised by NF-κB disruption59
. There is a recent case report of reduced NF-κB signaling in EBV-transformed Bc from two patients with mBc deficiencies60
. While our analyses suggest that NF-κB may play a role in development of the CD27lo
subpopulation, due to the redundancy of cellular signaling, other pathways may yet be found that provide that role as well.