Comment on “Gut-associated lymphoid tissue contains the molecular machinery to support T-cell-dependent and T-cell-independent class switch recombination”

To the editor: Although recently confirmed in an activation-induced cytidine deaminase (AID) reporter mouse model,¹ AID expression and immunoglobulin A (IgA) class switch recombination (CSR) in the intestinal lamina propria (LP) remain a matter of debate.² In an article published in the last issue of Mucosal Immunology, Barone et al.³ revisited AID expression in the human intestine. Barone et al. observed AID expression in intestinal follicles but not LP.³ In addition, Barone et al.³ detected a proliferation-inducing ligand (APRIL) and its receptors TACI and BCMA in both intestinal follicles and LP.³ As previously published data show that human APRIL elicits AID expression and IgA CSR in the absence of T-cell help to B cells by CD40 ligand,^{4, 5} Barone et al. conclude that APRIL and its receptors are part of a molecular mechanism that promotes IgA CSR in intestinal follicles by both T-cell-dependent (TD) and T-cell-independent (TI) pathways.³ Given the additional involvement of APRIL in plasma cell survival,⁶ Barone et al. further suggest that APRIL promotes plasma cell survival in the LP.³

Although plausible these conclusions are not demonstrated by the data. Indeed, Barone et al.³ solely document intestinal expression of APRIL, but provide neither functional nor molecular evidence of APRIL involvement in follicular IgA CSR or LP plasma cell survival. Furthermore, Barone et al.³ provide neither functional nor molecular evidence to demonstrate the involvement of APRIL, TACI, and BCMA in TD or TI routes of B-cell activation within intestinal follicles. We also question the approach used by Barone et al.³ to study the expression of AID, a hallmark of ongoing CSR,⁵ in the human LP.

Barone et al.³ stained intestinal tissues for AID and APRIL through immunohistochemistry (IHC) and compared the results of these stainings with those obtained by our group through immunofluorescence analysis (IFA).⁵ This comparison is inaccurate, because IHC and IFA are fundamentally different methodologies that deal with paraffin-embedded and frozen tissues, respectively. Different tissue processing modalities can affect sensitivity. Indeed, the sensitivity of IHC can be attenuated by antigen retrieval procedures, which modify the antigenic structure of the protein under study due to its exposure to high temperatures. Limited sensitivity explains the detection by Barone et al.³ of AID in follicular B cells, but not in LP B cells or tonsillar subepithelial B cells. A sensitivity issue may also explain the finding of APRIL in crypt but not lumen-facing epithelial cells.³ Although we agree with the former observation,⁵ the lack of APRIL in lumen-facing epithelial cells differs not only from our results,⁵ but also from the results of Shang et al.⁷ and data available in the Swedish Human Proteome resource program (http://www.proteinatlas.org/show_image.php?image_id=1635790). Because of these reasons and the glaring lack of supporting functional data, the lack of APRIL expression by lumen-facing epithelial cells should not be used as an argument to question the role of bacterial Toll-like receptor ligands in APRIL release by intestinal epithelial cells.⁵ In this regard, Barone et al.⁵ fail to recognize that the flagellin receptor Toll-like receptor-5 promotes APRIL secretion rather than APRIL expression by epithelial cells.³

Antigen retrieval procedures can affect not only the sensitivity, but also the specificity of IHC. Indeed, Barone et al.³ detected AID in CD68⁺ LP macrophages, and attribute this finding to the poor specificity of an EK2-5G9 antibody to AID (anti-AID) used in some of our studies.⁵ We reject this claim, because we used IFA instead of IHC to visualize AID. IFA specifically detected AID in LP-activated B cells, plasmablasts, and plasma cells co-expressing IgA, BSAP (Pax5), IRF4, Blimp-1 and/or CD138,⁵ but not CD68 (B. He and A. Cerutti, unpublished data). Importantly, we validated and further extended our IFA-based results by means of fluorescence in situ hybridization (FISH) in both CD40-sufficient and CD40-deficient individuals,⁵ whereas Barone et al.³ did not. Moreover, the specificity of EK2-5G9 for both follicular and extrafollicular B cells, including tonsillar subepithelial B cells, is consistent not only with results by Cattoretti et al.,⁸ but also with additional data from our group generated with anti-AID antibodies different from EK2-5G9 (B. He and A. Cerutti, unpublished data).

Barone et al.³ back up their claim that there is no AID in the LP by comparing the presence of AID transcripts in microdissected tonsillar germinal centers and LP tissue through quantitative PCR. The sequence of the primers used to amplify AID mRNA is not disclosed. This issue aside, we contend that in this specific setting quantitative PCR is misleading, because based on the false assumption that germinal centers and LP have comparable B-cell densities and AID expression levels. Instead, germinal centers have a much higher B-cell density than the LP, which consists of a mixed population of B cells and non-B cells. Furthermore, the vast majority of germinal center B cells express AID, whereas only a fraction of LP B cells does.^{5, 9} Moreover, individual germinal center B cells are likely to express more AID than individual LP B cells. Thus, the observation that germinal centers contain more AID RNA than the LP is completely expected and should not allow one to conclude that the LP lacks AID. Similarly, naive B cells exposed to CD40 ligand and interleukin-4 contain negligible AID RNA compared with germinal center B cells, because only a fraction of naive B cells induce AID and undergo CSR upon in vitro stimulation (B. He and A. Cerutti, unpublished data). A more appropriate method to detect AID in scattered LP B cells is FISH (Figure 1a). Alternatively, one can PCR amplify AID mRNA from purified LP B cells.⁶ At any rate, even the data provided by Barone et al.³ seem to be consistent with the presence of AID in the LP, because CD20⁺ B-cell-enriched contain more AID mRNA than CD20⁺ B-cell-poor LP samples.³ This difference would be visually more evident in a graphic lacking germinal center data.

Figure 1

Some LP B cells express AID and the switch DNA region-targeting nuclear protein γ-H2AX. (a) Human LP from the small intestine of a patient with CD40 deficiency stained for IgA (green), AID mRNA (red) by IFA. DAPI (blue) counterstains nuclei. Original magnification, × 10. Orange- or yellow-appearing LP cells co-express IgA and AID mRNA. (b) Human colon LP from a healthy subject stained for AID (green), nuclear γ-H2AX (red), and IgA (blue). Cyan-appearing LP cells co-express IgA and AID protein. Bottom panel shows colocalization of nuclear γ-H2AX with DAPI-stained nuclear DNA (cyan). 1, AID^– γ-H2AX⁺IgA1⁺ LP B cell; 2, AID⁺ γ-H2AX⁺IgA1⁺ LP B cell; 3, AID⁺ γ-H2AX⁺IgA1^– LP B cell; 4, AID⁺ γ-H2AX⁺IgA1⁺ LP B cell showing AID and γ-H2AX colocalization in the nucleus; 5, AID⁺ γ-H2AX^–IgA1⁺ LP B cell; 6, AID^– γ-H2AX⁺IgA1⁺ crypt epithelial cells. Cycling crypt epithelial cells contain γ-H2AX because γ-H2AX is generally involved in DNA repair during cell proliferation. Original magnification, × 40. One of several experiments yielding similar results.

PowerPoint slide

An additional problem relates to the inaccurate discussion of the literature on the inductive function of the LP. For example, Barone et al.³ do not discuss recent studies showing LP AID expression and both TD and T1 LP IgA CSR events in an AID-reporter mouse model.¹ In addition, Barone et al.³ do not mention the fact that APRIL triggers IgA1 CSR in addition to IgA2 CSR,⁵ which may explain the presence of APRIL not only in IgA2-rich areas such as the colon LP,⁵ but also in IgA1-rich lymphoid areas such as mucosal follicles.^{3, 5, 10} Finally, Barone et al.³ do not discuss earlier evidence showing that APRIL cooperates with antigen, cytokines, and microbial molecular patterns to induce not only direct IgM-to-IgA1 CSR, but also sequential IgA1-to-IgA2 CSR and B-cell proliferation.^{5, 11, 12, 13, 14} This evidence is consistent with the local presence of clonally expanded B cells in the LP¹⁵ and with the expression of growth-associated nuclear proteins such as Ki-67 (http://www.proteinatlas.org/show_image.php?image_id=282924) and γ-H2AX (Figure 1b) in some LP B cells. Of note, the histone γ-H2AX targets AID-induced double-strand switch DNA breaks in B cells transiting through the G1 phase of the cell cycle.¹⁶

In summary, we contend that Barone et al.³ provide neither functional nor molecular evidence to demonstrate that APRIL signals IgA CSR in B cells from intestinal follicles through TD and TD pathways involving TACI and BCMA. We also question the conclusion that the human intestinal LP lacks AID, because the experimental approaches used by Barone et al.³ neither allow an accurate detection of LP AID nor can be compared with approaches successfully used in earlier works.^{1, 5, 17, 18}

Disclosure

The authors declared no conflict of interest.