Response: Extended analysis of microarray data does not contradict preplasmablast phenotype of human CD20⁺CD27⁺CD43⁺ cells

Li et al¹  used additional approaches to analyze the microarray data presented in our article describing a preplasmablast phenotype for the proposed human B1 cells.²  Their analysis positions “B1” cells closer to memory B cells (CD19⁺CD20⁺CD27⁺CD43⁻) than to plasmablasts (CD19⁺CD20⁻CD27^hiCD43^hi), which, according to them, contradicts our conclusions.²  On the contrary, we feel that the analysis by Li et al¹  supports and extends our findings. Both our analysis and theirs are compatible with the notion that the putative B1 cells have a phenotype intermediate between memory B cells and plasmablasts. In addition, the observation that surface expression of CD43 and CD27 is intermediate between that of memory B cells^3,4  and plasmablasts supports our conclusions. Importantly, an independent analysis of our data demonstrates that transcriptional features of mouse B1 cells are clearly distinct from those of the proposed human B1 cells, whereas mouse and human germinal center B cells share such features.⁵ 

Other concerns were raised by Li et al.¹  We agree that B1 cells can switch to (mostly) immunoglobulin A (IgA). However, in mice, the majority of B-1 is known to secrete IgM.⁶  In our study, we found that IgA was the predominant isotype produced and even found similar frequencies of cells producing IgM and IgG. Rather, our data demonstrate a parallel between plasmablasts and the proposed B-1 cells.

Li et al¹  suggested that tetanus toxin–specific cells in our sorted “B1” population result from a small inadvertent contamination. However, our enzyme-linked immunospot data demonstrate that if production of anti–tetanus toxin antibodies by B1 cells was due to contaminating plasmablasts, the contamination of B1 cells by plasmablasts would have accounted for up to 50%. Moreover, such contamination would have aligned “B1” cells much closer to plasmablasts in the gene expression analysis, which is inconsistent with Li et al’s interpretation of the gene expression analysis.¹  The absence of antibody production in memory B cells (both spontaneous and vaccination-induced) further underscores the purity of our sorted samples.^2,7 

Li et al¹  argued that differentiation of cells to plasmablasts and plasma cells does not discriminate between B-cell lineages, as we actually discussed previously.²  However, this assertion contradicts the Rothstein group’s previous publications in which the inability of human “B1” cells to differentiate to plasmablasts was twice claimed to distinguish B-1 cells from preplasmablasts.^8,9 

Last, Li et al¹  raised technical concerns about the absence of CD69 and CD70 expression in our stimulated B-cell cultures, pointing toward the lack of positive controls and that the 5-day timespan might have led to decreased expression. Although not mentioned in our publication, we performed parallel control experiments in which peripheral blood mononuclear cells were stimulated with R-848 and interleukin-2 and consistently observed expression of CD69 and CD70 in CD19⁻ lymphocytes. Before choosing day 5, we performed kinetic studies and observed a gradual increase of CD20⁺CD27⁺CD43⁺ cells from days 3 to 5 without expression of CD69 or CD70.

In conclusion, the microarray analysis performed by Li et al¹  and by Mabbot and Gray⁵  support the concept that the putative B1 cells have a phenotype intermediate between memory B cells and plasmablasts.