Portal de Periódicos da CAPES

An optimized method to measure human FOXP3 + regulatory T cells from multiple tissue types using mass cytometry

2018; Wiley; Volume: 48; Issue: 8 Linguagem: Inglês

10.1002/eji.201747407

1521-4141

Nicholas A.J. Dawson, Avery J. Lam, Laura Cook, Romy E. Hoeppli, Raewyn Broady, Anne M. Pesenacker, Megan K. Levings,

Single-cell and spatial transcriptomics

We optimized a method to detect FOXP3 by mass cytometry and compared the resulting data to conventional flow cytometry. We also demonstrated the utility of the protocol to profile antigen-specific Tregs from whole blood, or Tregs from tissues such as cord blood, thymus and synovial fluid. High-dimensional phenotyping with mass cytometry has revealed new cell populations and functions as well as immunological networks 1 but its use to measure transcription factors, which are difficult to detect due to intranuclear localization and DNA association, has been limited. High and stable expression of the transcription factor FOXP3 defines regulatory T cells (Tregs) 2 so robust methodology to detect FOXP3 is essential for mass cytometry-based analysis of Tregs in health, disease, and/or in response to therapy. Several studies have analyzed human Tregs using mass cytometry. Mason et al. analyzed presorted CD4+CD25+CD127– Tregs but without FOXP3 staining in their mass cytometry panel 3. In unfractionated peripheral blood, changes in FOXP3 expression following low-dose IL-2 therapy for graft-versus-host disease 4, in aplastic anemia 5, or with cytotoxic protein coexpression patterns 6 have been reported. In tissue, mass cytometry was used to phenotype circulating versus tumor-infiltrating FOXP3+ Tregs in glioblastoma multiforme 7 or hepatocellular carcinoma 8. However, none of these studies reported validation of the FOXP3 staining protocol or compared results to "gold-standard" flow cytometry data. Using flow cytometry as a benchmark, we sought to optimize mass cytometry FOXP3 staining for detection of polyclonal and antigen-specific human Tregs from peripheral blood and other tissues. To develop an optimal protocol to detect FOXP3 by mass cytometry, we aimed for clear resolution of FOXP3+ Tregs, minimal staining of conventional T cells (Tconvs), and uncompromised detection of cell surface proteins necessary to identify Tregs/or and other cell types. We tested three buffer sets: the FOXP3 mass cytometry staining kit from Fluidigm; the commonly used eBioscience FOXP3 fix/perm buffer set developed for flow cytometry 9, and a custom buffer system using paraformaldehyde (PFA)-based fixation and saponin-based membrane permeabilization (termed PFA/saponin; see Supporting Information). For all samples and buffers, FOXP3 gates were set on the basis of staining in live CD4neg cells (Supporting Information Fig. S1A). We found that all three buffer systems resulted in similar proportions of FOXP3+ cells within the CD4+CD25+CD127– gate, which is the widely accepted combination of Treg-defining cell surface molecules in humans (Fig. 1A, Supporting Information Fig. S1A) 2, 10, 11. These data were confirmed by determining the proportion of CD25+CD127– cells within the CD4+FOXP3+ gate (Supporting Information Fig. 1B). In the majority of samples, a proportion of Tconvs (CD4+CD25–CD127+) displayed background FOXP3 staining using the eBioscience, but not Fluidigm or PFA/saponin, protocol (Fig. 1A). Since fixation methods can destroy or alter epitopes 12, 13, we determined the effect of each FOXP3 staining protocol on detection of a range of T-cell surface markers. The Fluidigm buffers had the greatest negative effect on several Treg-defining cell surface markers including CD3, CD4, and CD25, resulting in poor resolution of CD3+CD4+ cells (Supporting Information Fig. S1A) as well as CD25+CD127– cells (Fig. 1B). In contrast, PFA/saponin and eBioscience buffers showed equivalent and optimal detection of CD3, CD4, and CD25, supporting their use for detection of CD4+CD25+CD127– Tregs. Evaluation of other Treg-associated cell surface markers revealed that no single fixation method was optimal for detection of all markers tested. Although each method had its respective limitations (Supporting Information Fig. S2A and B), the PFA/saponin protocol was optimal for detection of Tregs by mass cytometry because it neither caused background staining of FOXP3 in Tconvs nor diminished detection of Treg-defining cell surface molecules. These data highlight the importance of determining the impact of fixation methods used to measure nuclear proteins on detection of other cytoplasmic or cell surface markers of interest. Having identified an optimal method for staining FOXP3 via mass cytometry, we next compared mass cytometry data obtained using the PFA/saponin protocol to fluorescence flow cytometry. Since human immune cells are often analyzed after cryopreservation, we compared ex vivo or cryopreserved cells on both platforms. We found that cell surface marker detection was similar on both platforms for most antigens tested (Supporting Information Fig. S3A–E), with the notable exception of significantly lower proportions of CCR4-expressing CD3+CD4+ T cells in mass cytometry. This difference was likely due to differential CCR4 clone sensitivity (Supporting Information Fig. S3F), highlighting the need for careful antibody clone selection to achieve optimal resolution in mass (and flow) cytometry. For both ex vivo and cryopreserved samples, the proportion of FOXP3-expressing cells (PCH101 clone) detected via mass cytometry was significantly lower than via flow cytometry (Fig. 1C, Supporting Information Fig. S4). However, within each platform, there were no significant differences in the proportion of FOXP3+ cells detected in ex vivo versus cryopreserved samples (Fig. 1D). To determine if reduced detection of FOXP3 in mass cytometry was a limitation of the PCH101 clone, the experiment was repeated with the FOXP3 clone 236A/E7, with similar results (Fig. 1E). This reduced sensitivity may be due to chemical properties of fluorophores versus metals, possibly resulting in a differential ability of fluorochrome versus metal-conjugated antibodies to pass through the nuclear membrane. Additionally, even the most sensitive metals (e.g. 162Dy) are less sensitive than the brightest fluorophores (e.g. PE) 14, 15. It is often desirable to detect antigen-specific Tregs, so we assessed the ability of the PFA/saponin protocol to enumerate FOXP3+ Tregs within a population of antigen-specific CD4+ T cells. Blood was stimulated with the indicated antigen for 44 hours and antigen-specific CD4+ T cells were detected by induced coexpression of CD25 and OX40 16 by flow or mass cytometry using optimal methods (see Supporting Information and Supporting Information Fig. S5). The proportion of antigen-specific CD4+ responder cells detected by mass cytometry was similar to flow cytometry and, since Tregs comprise a substantial proportion of recall responses 17, both mass cytometry and flow cytometry detected a clear population of antigen-specific FOXP3+ cells. Notably, in contrast to ex vivo cells, the proportion of FOXP3+ cells detected by mass cytometry was similar to that detected with flow cytometry. We next tested the applicability of PFA/saponin-based FOXP3 staining to detect CD4+CD25+CD127– Tregs in samples other than peripheral blood via mass cytometry. Mononuclear cells from umbilical cord blood, CD8-depleted thymocytes, or synovial fluid from juvenile idiopathic arthritis (JIA) patients were stained with a panel of T cell and Treg-related markers, including FOXP3. Data were analyzed with ACCENSE (automatic classification of cellular expression by nonlinear stochastic embedding) 18, which compares cells on a two-dimensional plot while maintaining single cell resolution and complexity, and then further identifies statistically significant subpopulations (Fig. 2A). The relative expression of each marker within the 25 ACCENSE-defined populations, as well as their relative abundance within each tissue, was determined (Fig. 2B). Strikingly, Tregs preferentially clustered into ACCENSE populations by tissue source. For example, Treg populations in cord blood were uniquely identified by high CD45RA, whereas Treg populations in JIA synovial fluid were defined by high expression of multiple activation/effector molecules and enriched for high PD-1. Despite strong tissue-specific segregation, a few populations were shared between tissues. For instance, a subset of CD45RA+ Tregs (population 4) was found in peripheral blood and cord blood, and populations 3 (CD127+) and 15 (low for all markers) were present in cord blood and thymus. Overall, these results support a growing body of evidence indicating that Tregs acquire unique tissue-specific phenotypes and that phenotypes in peripheral blood may not reflect those of tissue-resident cells 19, 20. In conclusion, we developed an optimal protocol to detect FOXP3 expression by mass cytometry, tested its suitability for ex vivo and cryopreserved samples, and showed its utility across a broad range of immune cell sources. We have further shown that antigen-specific FOXP3+ Tregs can be detected by mass cytometry in whole blood. An important consideration is that, at least in ex vivo peripheral blood samples, mass cytometry is significantly less sensitive than fluorescence flow cytometry at detecting FOXP3 expression. Overall, our optimized PFA/saponin protocol is the best-validated method described to date to detect FOXP3 expression by mass cytometry without compromising detection of cell surface markers. This method will enhance high-dimensional studies of Treg phenotype and function in the context of complex cellular networks. This work was supported by a grant from the Leukemia Lymphoma Society of Canada (to R.B. and M.K.L.). N.A.J.D. and A.J.L. hold CIHR Doctoral Research Awards; L.C. holds a BC Children's Hospital Research Institute postdoctoral fellowship; A.M.P. holds a JDRF postdoctoral fellowship; R.E.H. and M.K.L. hold salary awards from BC Children's Hospital Research Institute. We thank Mike Williams at the University of British Columbia Antibody Lab (Ablab.ca), Andy Johnson and David Ko at the University of British Columbia and BC Cancer Agency Flow Cores, respectively, and Fluidigm for helpful discussions, technical support, and custom antibody conjugation. The authors thank volunteers, patients, and their parents for contribution of samples, as well as BC Children's Hospital staff who made this study possible. We specifically thank Drs. Kim Morishita, Lori Tucker, Andrew Campbell, and Pascal Lavoie. N.A.J.D. designed and conducted experiments, analyzed data, and wrote the manuscript. A.J.L. designed and conducted experiments, analyzed data, and critically reviewed the manuscript. L.C. designed and conducted experiments, critically reviewed the manuscript, and provided experimental design guidance. R.E.H. and A.M.P. contributed to experiment design, reviewed results, and critically reviewed the manuscript. R.B. secured funding and contributed to experimental design. M.K.L. secured funding, provided overall guidance for experimental design and interpretation, and wrote the manuscript. The authors declare no commercial or financial conflict of interest. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.