Authors’ Response to Letter to the Editor on “Unidentified Variables May Account for Variability in Multiplexing Results”

Dear Editor,

This letter is in response to the Letter to the Editor by Bolognesi et al. ¹ that raises questions about some aspects of the virtual quadruple staining procedure we described in the February 2020 issue of the Journal of Histochemistry and Cytochemistry (JHC). ² This multiple staining method, which was developed to identify innate lymphoid cells (ILCs), comprises iterative single stainings on the same section with digital scanning and subsequent immunoglobulin and chromogen stripping after each staining round. Our protocol was based on publications by Kim et al. ³ and van den Brand et al. ⁴ We have used this immunohistochemical multiple staining protocol in several earlier peer-reviewed publications.^{5 -9}

In our recent article in JHC, ² we observed that four of nine antibodies showed reduced staining activity after the elution procedure. Bolognesi et al. suggested that this epitope loss-related reduction of staining intensity may be caused by flaws in our method. In their experience, only 3 of 314 antibodies used in their laboratory showed reduced staining activity after one staining round, while all remaining antibodies were equally potent after at least 30 staining rounds. This difference in outcome compared to ours was attributed by Bolognesi et al. to our use of tap water during the procedure, to the lack of disaccharides in our antibody diluent and buffers, and to our mounting of slides with gelatin/glycerin. Finally, they also question the use of NovaRed as chromogen for destaining slides.

Previous work of Bolognesi et al. on iterative staining/destaining involved immunofluorescence rather than chromogen staining. Their Letter to the Editor is the only support for their claim that multiplexing with chromogens can be performed for multiple rounds without any loss of quality. They show sections that were treated with stripping buffer during an extended incubation period assumed equivalent to 10 elution cycles. ¹ However, this surrogate multiple stripping is deceptive as the specimen was treated just once with stripping buffer, whereas in a real stain/destain experiment, sections are exposed to repeated strong changes in pH and repeated exposure to reagents that possibly influence antigenicity of the epitopes (such as H₂O₂, hematoxylin, chromogen and staining buffers). Furthermore, they do not show data of single sections that have been subjected to multiple stripping but instead show serial sections.

We do not disapprove the opinion of Bolognesi et al. that the use of tap water should be avoided as the calcium in the water may represent a possible culprit of epitope loss. However, the use of tap water in our protocol is crucial, since we need it for the optimal counterstaining with hematoxylin. This counterstaining is not only important for visual recognition of the tissue architecture, it is also crucial for specimen detection and focusing of the slide scanner, which fails if a sample does not contain enough contrast. It may also be possible that hematoxylin, a basophilic dye that strongly binds to DNA, has affected the antigenicity of epitopes in the nuclei. This latter possibility could also explain why three out of four antibodies of which we observed decreased reactivity, were directed against transcription factors (T-bet, GATA3 and RORγt) that are located in the nucleus.

In line with our observation, Bolognesi et al. ¹ confirmed the existence of stripping-sensitive epitopes, which underlines the importance of testing epitope survival upon stripping and prioritizing of the staining sequence, as we did for the protocol described in our JHC paper. ² They remark that a relatively high number of antibodies (we presume that they meant epitopes) did not survive the first cycle of antibody/dye removal in our study, as compared to their experience. Concerning this point, indeed four of nine antibodies appeared to detect stripping-sensitive epitopes, which is probably a matter of coincidence. We have used many more antibodies previously and have observed that most of them provided successful staining after multiple rounds of stripping without loss of staining intensity.^{5 -9} Hence, the presumptive epitope loss observed in our study is not as dramatic or unusual as suggested by Bolognesi et al. and seems instead to be related to the nature of the particular epitopes, and possibly choice of antibody and immuno-staining method applied. Regarding Bolognesi et al. questioning the effectiveness of NovaRed pigment removal in our protocols, we point out that destaining slides with NovaRed has been used by many groups in a similar setting. In addition, it is known to dissolve in aqueous solutions, which can be appreciated from the images we provided in our JHC publication. ²

In conclusion, epitope survival after antibody stripping is probably dependent on the method used: chromogen versus fluorescence staining. Our virtual quadruple staining is based on established protocols and includes all necessary controls to validate every step of the process. Therefore, we are convinced that our protocol is correct and suitable to detect ILCs.