Attack on the clones? Human FOXP3 detection by PCH101, 236A/E7, 206D, and 259D reveals 259D as the outlier with lower sensitivity

To the editor:

We read with great interest the report by Tran and colleagues in which the authors studied the induction of FOXP3 in naive human T cells.¹  The authors state that, in the absence of exogenous TGF-β, only a minor subset of activated human T cells express FOXP3. In contrast, several prior reports including ours^2,–4  have shown FOXP3 expression on a substantial proportion of activated T cells (reviewed in Pillai and Karandikar⁵ ). The authors attribute these findings to the “nonspecific” staining of human T cells by clone PCH101 (eBioscience, San Diego, CA), used in multiple prior studies. The authors rely on clone 259D (Biolegend, San Diego, CA), claiming that this was similar to other clones such as 206D (Biolegend) and 236A/E7 (eBioscience), distinct from PCH101 (data mentioned but not demonstrated in the publication). If true, these observations would call into question many prior conclusions.

However, we believe that the authors' conclusions are based on somewhat erroneous interpretation of flow cytometric data. In Figure 3 of their publication,¹  where PCH101 staining is demonstrated alongside 259D, the authors' “negative gates” were set solely on the basis of isotypes. When evaluating intracellular staining on activated cells (or sometimes even simple surface staining), isotypic antibodies may be insufficient and misleading controls.^6,7  Experience shows that multiple flow cytometric antibodies can cause negative populations to significantly shift beyond isotypic cutoffs. Rather than interpreting this as “nonspecific staining,” it is important to include other controls during the validation of staining, such as antibody titrations and cells known to be negative for the marker. Use of unstimulated, FOXP3⁻ cells would have undoubtedly resulted in a higher cutoff and their “2 peaks” of FOXP3 expression would have been correctly interpreted as negative and positive peaks.²  In fact, the authors' siRNA inhibition data demonstrated the disappearance of the positive FOXP3 peak, confirming the specificity of the “second peak.” The first peak remained intact because it was not “FOXP3⁺” to begin with.

In addition, basing conclusions exclusively on 259D staining may also be problematic. We have performed parallel staining for FOXP3 with multiple clones of antibodies (Figures 1, 2). In these experiments, inclusion of carboxyfluorescein diacetate succinimidyl ester (CFSE) staining allowed us to clearly differentiate stimulated and proliferating cells from those that did not enter cell cycle. Of note, even with “pan T-cell stimulation” such as anti-CD3, not all T cells undergo complete activation, nor do all cells express FOXP3 (providing excellent internal controls). However, almost all of the fully activated cells truly express FOXP3, based on carefully picked cutoffs and titrated antibodies. This is also corroborated by prior PCR analysis of sorted populations.² 

Clones PCH101, 206D, and 236A/E7 provide essentially similar results, distinct from those of 259D. In fact, clone 259D would miss FOXP3 expression in some cases (Figures 1L, 2C), when all other clones robustly detect such expression (Figures 1I, K, 2B). The differences may arise from the specific epitopes recognized by the antibodies. Regardless, it can be conclusively stated that 259D is the outlier among all these clones and a less sensitive detector of total FOXP3.

Our data reaffirm the validity of previous results using PCH101, provided the correct cutoffs were used. Clearly, CD4⁺CD25⁻FOXP3⁻ T cells that undergo full activation express robust amounts of FOXP3. Importantly, these data expose an important pitfall of relying solely on isotypic controls in flow cytometric evaluation, which should help other investigators in their evaluation of this and other markers.