Phenotypical characterization of regulatory T cells in humans and rodents

Summary

Regulatory T cells (T_regs) constitute a fascinating subpopulation of CD4⁺ T cells due to their ability to limit the immune response against self and non‐self antigens. Murine models and antibodies directed against surface and intracellular molecules have allowed elucidation of the mechanisms that govern their development and function. However, these markers used to their classification lack of specificity, as they can be expressed by activated T cells. Similarly, there are slight differences between animal models, in steady state and pathological conditions, anatomical localization and strategy of analysis by flow cytometry. Here, we revised the most common markers utilized for T_reg typification by flow cytometry such as CD25, forkhead box protein 3 (FoxP3) and CD127, along with our data obtained in different body compartments of humans, mice and rats. Furthermore, we revised and determined the expression of other molecules important for the phenotypical characterization of T_reg cells. We draw attention to the drawbacks of those markers used in chronic states of inflammation. However, until a specific marker for the identification of T_regs is discovered, the best combination of markers will depend upon the tissue or the degree of inflammation from which T_regs derive.

Introduction

Regulatory CD4⁺ T cells expressing the transcription factor forkhead box protein 3 (FoxP3) [regulatory T cells (T_regs)] constitute a fascinating subpopulation of T cells, capable of fine‐tuning the immune response raised against autoantigens and foreign antigens. In the 1980s, the introduction of a specific monoclonal antibody directed against the α‐chain of the interleukin (IL)‐2 receptor (CD25) opened a vast field in clinical and basic immunology that has led to the identification by flow cytometry of ‘suppressor’ cells critical for the maintenance of self‐tolerance 1. T_regs can be divided into two groups according to their site of origin: thymus‐derived T_regs (tT_regs) and peripherally derived T_regs (pT_regs) 2. tT_regs are selected positively in the thymus through the major histocompatibility complex (MHC) class II‐dependent T cell receptor (TCR) interactions resulting in a relatively high‐avidity selection 3. In contrast, pT_regs originate in the periphery from conventional T cells (T_cons) as a result of TCR stimulation in the presence of the transforming growth factor (TGF)‐β and IL‐2 3. Later, new markers were introduced, allowing a better identification of the T_regs population, such as FoxP3 and the low expression or absence of alpha chain of the IL‐7R (CD127); however, these markers do not allow accurate delineation of the T_reg population or distinguish between tT_regs and pT_regs. In addition, many subphenotypes of T_regs have been identified in human and rodent models based on the expression of different markers, identifying profiles associated with a major specific suppressive function, migration pattern and/or activation, suggestive of their heterogeneous nature.

Although other markers, such as Helios (a transcription factor of the Ikaros family) and Neuropilin‐1 (Nrp1, a surface molecule), have been introduced, their use is still controversial 4, 5. Fortunately, tT_regs could be distinguished from pT_regs through the identification of a non‐coding sequence within the FOXP3 locus named TSDR (T_reg‐specific demethylated region) that is completely demethylated on tT_regs, contrary to pT_regs or T_cons in which TSDR is methylated. As this strategy requires cell lysis to extract the DNA, it is not possible to use it to isolate viable cells 6. However, it would be desirable to assess the methylation status of FoxP3, in conjunction with characterization by flow cytometry.

The heterogeneous nature of T_regs highlights the different ways to turn off the immune response of T_cons and dendritic cells (DCs), such as cell‐to‐cell contact, metabolic disruption of the microenvironment and release of immunosuppressive cytokines 7. One of the most relevant contact mechanisms for T_regs is the expression of the cytotoxic T lymphocyte associated‐antigen 4 (CTLA‐4), a member of the CD28 superfamily of immunoreceptors that competes with CD28 for binding to B7 molecules (B7·1 or CD80 and B7·2 or CD86) on antigen‐presenting cells (APCs), that in turn inhibit DC‐T_cons aggregates, DC maturation and T cell activation at the immunological synapse 8, 9, 10. Programmed death‐1 (PD‐1) is another inhibitory molecule of the CD28 superfamily expressed on T_regs that interacts with its ligand PDL‐1 on CD4⁺ and CD8⁺ T cells, inhibiting their proliferation and release of proinflammatory cytokines 11. Moreover, the expression of PD‐1 on CD4⁺ T cells seems to mediate their conversion to T_regs 12. Interestingly, PD‐1‐expressing T_regs up‐regulate Tim‐3, a cell immunoglobulin and mucin domain‐containing protein (TIM), which induces selectively the cell death of fully differentiated T helper type 1 (Th1) cells through interaction with its ligand, galectin‐9, expressed in the latter cells 13. Inhibition of maturation and co‐stimulatory ability of DCs can also be mediated through contact of its MHC class II with the lymphocyte‐activation gene 3 (LAG‐3) expressed on T_regs 14. Although other molecules such as GITR [glucocorticoid‐induced tumour necrosis factor (TNF) receptor‐related protein] and OX‐40 (CD134) [TNF receptor superfamily, member 4 (TNF‐RSF4)] expressed on T_regs mediate contact with other immune cells, their role on T_reg function is more controversial; thus, it will be discussed in their corresponding section.

T_regs also turn off the immune response through anti‐inflammatory cytokine production, such as TGF‐β, IL‐10 and IL‐35, that are important for the control of the immune response in mucosal sites in response to non‐self antigens 15. Another interesting way for T_regs to suppress T_cons activation is by inducing metabolic alterations in the cellular microenvironment by consuming IL‐2 from the medium due their high expression of CD25, affecting T_cons survival and proliferation 16. Expression of ectoenzymes such as CD39 and CD73 catabolizes extracellular ATP generating adenosine that is bound to the A2A adenosine receptor on T cells or DCs, down‐regulating the activation and production of cytokines and co‐stimulatory molecule expression 17. Moreover, T_regs harbour high levels of cyclic adenosine monophosphate (cAMP), which could be transferred to T_cons via gap junctions, thereby inhibiting proliferation and IL‐2 production 18. Remarkably, homing markers that are expressed selectively for infiltrating T_regs from peripheral tissues, is required to exert total suppressive action; in particular, the expression of integrin α_E (CD103) allows the migration of T_regs from blood to inflamed epithelial tissues or gut by binding to E‐cadherin; moreover, its expression on T_regs has been associated with an effector/memory phenotype 19.

Currently, the availability of monoclonal antibodies has allowed a better delineation of T_reg populations. However, some drawbacks remain, such as the promiscuous nature of expression of several molecules used to identify T_regs, technical issues related to clones and fluorochrome conjugates of monoclonal antibodies, the staining and/or permeabilization protocols, the gating strategy for the analysis of flow cytometry to define positive events, the anatomical location of T_regs and the immunological status of the host. All these aspects could be the origin of some of the discrepancies observed in both humans and rodent species.

Altogether, these aspects render the characterization of T_regs plagued to a certain extent by variations. Hence, we will discuss pros and cons of the use of CD25, FoxP3 and CD127 in humans and rodents along with technical issues concerning to flow cytometry. We also summarize evidence regarding some markers we used to phenotype T_regs in humans and rodents, such as CTLA‐4, GITR, CD103, PD‐1 and OX40, and to compare this information with that obtained in other studies, along with representative gating strategies for T_reg identification by flow cytometry in lymphoid tissues from humans and rodents.

Expression of CD25, FoxP3 and CD127 to identify human T_regs

In the early 1990s, expression of CD25 was used to identify human T_regs on the total circulating CD4⁺ T cells with values oscillating between 10 and 30% 20, 21, comparable to the 15% that we have obtained in healthy individuals (Table 1).

Table 1.

Frequency and phenotype of T_regs in healthy humans, mice and rats determined by flow cytometry

	Humans*		Mice †		Rats ‡
Markers for Treg detection	PB°	RM*	LN	Spleen	PB	LN	Spleen
CD4+CD25+	14·5 ± 2·9	n.d.	14·3 ± 1·3	15·2 ± 0·4	8 ± 2	7·2 ± 1·5	8·5 ± 1·7
CD4+FoxP3+	6·4 ± 0·9	9·3 ± 4·1	11·2 ± 0·4	13 ± 2·6	7·6 ± 1·5	6·3 ± 0·7	7·2 ± 1·4
CD4+FoxP3+CD25+	4·9 ± 1·1	7·5 ± 2·4	9·6 ± 0·3	10·2 ± 1	4·3 ± 1·1	3·6 ± 0·5	4·1 ± 1·2
CD4+FoxP3+CD127low/–	6·1 ± 1·1	7 ± 2·2	n.d.	n.d.	n.d.	n.d.	n.d.
CD4+CD25+CD127low/–	6·0 ± 1·5	6·5 ± 3·2	n.d.	n.d.	n.d.	n.d.	n.d.
Phenotype of CD4+FoxP3+ Tregs
CD25	51·6 ± 9·1	45·4 ± 11·2	84·3 ± 2·5	82·7 ± 0·6	64·7 ± 5·8	62·2 ± 4·3	60·5 ± 5·1
CTLA‐4	23 ± 10·4	49·3 ± 13·4	n.d.	n.d.	23·9 ± 11·6	20·2 ± 6·1	18·4 ± 8·6
GITR	10·3 ± 5·5	n.d.	75 ± 14·1	76·7 ± 8·7	n.d.	n.d.	n.d.
CD103	9·3 ± 3·8	21·5 ± 10·7	n.d.	n.d.	39 ± 17·5	38 ± 17·5	30 ± 19·1
PD‐1	4·4 ± 3·7	16·8 ± 12	66·3 ± 14·6	65·5 ± 7·1	n.d.	n.d.	n.d.
OX‐40	n.d.	n.d.	62·8 ± 10·5	62·3 ± 3·8	n.d.	n.d.	n.d.

Data are expressed as mean ± standard deviation. *Data from humans were taken from our previous studies 29, 105. ^†BALB/c mice and ^‡Wistar rats data were taken from our original unpublished results. PB = peripheral blood; RM = rectal mucosae; LN = lymph nodes; n.d. = not determined; FoxP3 = forkhead box protein 3; T_regs = regulatory T cells; PD‐1 = programmed death‐1; GITR = glucocorticoid‐induced tumour necrosis factor receptor‐related protein; CTLA‐4 = cytotoxic T lymphocyte‐associated‐antigen 4.

High expression of CD25 in human CD4⁺ T cells has also been used to define T_regs, representing approximately 1–2% of the total CD4⁺ subset in peripheral blood and exhibiting the greatest suppressive function 22. However, the definition of ‘high’ using MFI (mean fluorescence intensity) as the criterion for its characterization 23 is inaccurate as it depends, among others, on the specific configuration of the flow cytometers and on the characteristics of the antibodies. Moreover, CD25 is expressed heterogeneously on T cells; thus, approximately 96% of CD25^high T cells express FoxP3, compared to only 35% FoxP3‐expressing CD4⁺CD25^int T cells 24. Also, the staining shows a smear 25 that hinders the establishing of the minimal level of CD25 expression required to define the T_reg population 26. Moreover, intermediate and high levels of CD25 expression are arbitrary. For this reason, identifying T_regs as T cells expressing high levels of CD25 should not be used. Several clones of anti‐CD25 antibodies have been reported in the literature; we have used the BC96 clone to identify CD25⁺ T cells with clear results (Fig. 1). In addition, the use of this antibody, conjugated with the phycoerythrin‐cyanin 5 (PE‐Cy5) fluorochrome, have allowed a correct discrimination when used alone or in combination with other antibodies constituting the T_reg panel, as other authors have reported 27.

Figure 1.

Representative plots of gating strategy for flow cytometry showing different ways of human T_reg characterization. Upper panel shows the forward‐scatter area (FSC‐A) versus forward‐scatter height (FSC‐H) plot for doublet exclusions, followed by a FSC‐A versus side‐scatter area (SSC‐A) plot to delimitate the lymphocytes region. Then, a CD3 versus CD4 plot was used to gate CD4⁺ T cells, and percentages of CD4⁺forkhead box protein 3 (FoxP3⁺), CD25⁺FoxP3⁺, CD127^–/lowFoxP3⁺ and CD127^–/low are shown on plots from rectal biopsies of healthy volunteers. Lower panel shows representative plots and percentages of FoxP3⁺cytotoxic T lymphocyte antigen‐4 (CTLA‐4⁺) FoxP3⁺CD103⁺ and FoxP3⁺programmed death‐1 (PD‐1⁺) cells on CD4⁺ gate. Numbers in brackets represent the % of CD4⁺FoxP3⁺ T cells positive for CTLA‐4, CD103 and PD‐1. Clone and fluorochrome conjugates from the monoclonal antibodies used are shown in Supporting information, Table S1. Clone of anti‐FoxP3 was PCH101.

The crucial role of FoxP3 as a master regulator in the development and function of T_regs was demonstrated by a frameshift mutation in the FOXP3 gene that leads to its impaired expression and to the development of a syndrome known as IPEX (immune dysregulation, polyendocrinopathy, enteropathy, X‐linked syndrome) in humans and scurfy in mice, resulting in a variety of autoimmune disorders that could be incompatible with life 3.

In humans, the percentage of CD4⁺FoxP3⁺ cells is variable, representing approximately 1–10% of the CD4⁺ population in peripheral blood 28, 29, 30; we have observed approximately 6% in blood (Table 1). Interestingly, lower percentages of CD4⁺FoxP3⁺ T_regs have been observed in frozen samples 30, which have been demonstrated previously to reduce the pool of CD4⁺CD25⁺FoxP3⁺, specifically in cells from HIV‐infected individuals 31. However, some authors have reported comparable values of T_regs in frozen samples and fresh samples, suggesting that these variations could be related to differences between cryopreservation and thawing protocols. To detect FoxP3 by flow cytometry, clone PCH101 (e‐Bioscience, San Diego, CA, USA) has been used widely with the highest detection of FoxP3; however, other clones such as 259D/C7, 236A/E7 and 206D have detected FoxP3 correctly, whereas clones 150d and 3G3 yielded the lowest values 32. In addition, detection of FoxP3 is linked to the fixation and permeabilization protocols used. In our experience, which is supported by other studies, the combination of clone PCH101 and the e‐Bioscience permeabilization buffer gave optimal results 32 in comparison to other clones and permeabilization buffers. We have also demonstrated good resolution between FoxP3⁺ and FoxP3^– events when anti‐FoxP3 is coupled to the fluorescein isothiocyanate (FITC) fluorochrome. However, other authors have reported that Alexa647 showed a better separation to detect FoxP3⁺ events than the R‐PE or FITC fluorochromes 32, and between PE and FITC, PE could be more effective in the detection of FoxP3⁺ events 33, 34.

We have observed the highest proportion of T_regs in samples from rectal mucosae compared to peripheral blood (approximately 9%, Table 1, Fig. 1 29), which could be related to a major local tolerance for the continuous exposition to exogenous antigens from the microbiota or diet 35. However, other authors have found similar percentages of T_regs in rectal samples and peripheral blood 36, which could be related to aggressive procedures used to obtain rectal mononuclear cells, affecting the expression of those markers.

Determination of FoxP3 and CD25 could share some disadvantages, as these molecules can be up‐regulated following in‐vitro activation in T_cons 21, 37, 38, 39; however, it has also been suggested that up‐regulation of FoxP3 in T_cons under these conditions is due to TGF‐β present in the serum of culture media, without conferring a suppressive function 40. In addition, expression of FoxP3 can be down‐regulated in inflammatory microenvironments, where T_regs are called ‘ex‐T_regs’ 41. Moreover, as we have reported, CD25 can also be down‐regulated during choric immune activation such as HIV infection 29, where a reduction in IL‐2 levels is observed 42. Another drawback of the use of FoxP3 is related to the necessity of fixation/permeabilization protocols preventing the isolation of viable cells for functional experiments.

In the early 2000s, the addition of FoxP3 to CD25 efficiently improved the identification of T_regs, and this strategy has been used widely by us (Fig. 1) and others 28. However, such a combination of CD25/FoxP3 is not perfect, as the expression of these markers can change depending on the inflammatory status, as mentioned above. In July 2006, two groups introduced a new marker for the identification of T_regs 43, 44; they found that low or lack of expression of CD127 in CD25⁺ T cells identify human T_regs with a potent suppressive function 45. In fact, this strategy has been used commonly for isolation of T_regs to assess their functional capacity. However, this strategy may not correspond exactly with CD4⁺CD25⁺FoxP3⁺ T_regs, as approximately 15% of CD25⁺CD127^low/– do not express FoxP3 43, 44 and < 10% of CD25⁺FoxP3⁺ cells retained high expression levels of CD127 43. Thus, as other markers to identify T_reg cells, CD127 expression has been show to be variable even in healthy individuals 46, and it can be also down‐regulated in T_cons during chronic activation (HIV infection) or after in‐vitro activation 47, 48.

Similar to other studies 44, 49, 50, we have shown that in peripheral blood of healthy individuals, the percentage of CD4⁺CD25⁺CD127^–/low cells varies from 6 to 8% of CD4⁺ T cells (Table 1). However, other authors have found higher values ranging from 7·2 to 30·4% (mean 17·6%) 46; such a high frequency of CD25⁺CD127^low/– T_regs observed in these healthy donors could be due to decreased expression of CD127 molecules after cryopreservation 24. In agreement with other researchers 43, we have found that peripheral CD4⁺ FoxP3⁺ T_regs in healthy individuals are highly comparable to those CD25⁺ CD127^–/low T_regs, as a high correlation between both strategies to analyse T_reg populations has been observed (r = 0·75, P = 0·0008). However, when the CD25⁺FoxP3⁺ combination was used, the strength of the correlation with CD4⁺FoxP3⁺ T_regs decreased (r = 0·51, P = 0.03). We have reported that in chronic inflammation, as in HIV infection, there is a high correlation between the CD4⁺FoxP3⁺ and CD127^–/lowFoxP3⁺ protocols for T_reg identification 51. Conversely, the CD25⁺FoxP3⁺ combination correlates strongly with CD4⁺FoxP3⁺ in rectal mucosae of healthy humans (r = 0·70, P = 0·02), rather than with the CD127^–/lowFoxP3⁺ combination, probably reflecting the fact that in mucosal tissues there is a particular regulation of CD127 expression on T_regs.

Considering the issues regarding the activation status that can modulate the expression of these markers, the use of a T_reg basic panel could be constituted by CD3, CD4, CD25, CD127 and FoxP3 which, in conjunction, may facilitate T_reg identification.

Expression of CD25, FoxP3 and CD127 to identify T_regs in mice

In mice, the identification of T_regs as CD4⁺CD25⁺ T cells has been used widely, representing approximately 5–10% of CD4⁺ T cells in the periphery 28. We have observed higher percentages of T_regs in lymphoid tissues, such as spleen (15·2%) and lymph nodes (14·3%) of healthy BALB/c mice (Table 1, Fig. 2), which corresponds with values reported previously in other studies 52. Slight differences in the levels of these cells could be due to the microorganism‐free environment in which mice are maintained, to their age or strain 53, 54. Similarly, gating strategy excluding cells expressing low levels of CD25 could be excluding T_regs receiving deficient co‐stimulator signals (CD28 or IL‐2) 55, 56. Although the use of CD25 as surrogate marker of functional T_regs was suggested previously, CD25 can also be expressed in T_cons after activation 16, as it can be lost in T_regs from aged mice even though they maintain their suppressive function 57. With regard to the anti‐CD25 clone, we have used PC61, which is used frequently.

Figure 2.

Representative plots of gating strategy for flow cytometry showing different ways of characterizing mouse T_regs. Upper panel shows the forward‐scatter height (FSC‐H) versus forward‐scatter area (FSC‐A) plot for doublet exclusions, followed by a FSC‐A versus side‐scatter area (SSC‐A) plot to delimitate the lymphocyte region. Then, a CD3 versus CD4 plot was used to gate CD4⁺ T cells, and percentages of CD4⁺CD25⁺, CD4⁺forkhead box protein 3 (FoxP3⁺) and CD25⁺FoxP3 cells are shown on plots from lymph nodes of healthy BALB/c mice. Lower panel shows representative plots and percentages of FoxP3⁺programmed death‐1 (PD‐1⁺), FoxP3⁺glucocorticoid‐induced tumour necrosis factor (TNF) receptor‐related protein (GITR) and FoxP3⁺OX‐40⁺ cells on CD4⁺ gate. Numbers in brackets represent the % of CD4⁺FoxP3⁺ T cells positive for PD‐1, GITR and OX‐40. Clone and fluorochrome conjugates from the monoclonal antibodies used are shown in Supporting information, Table S1. Clone of anti‐FoxP3 was FJK‐16s.

Regarding FoxP3 expression, we have found 13 and 11% CD4⁺ T cells expressing FoxP3 in spleen and lymph nodes, respectively (Table 1, Fig. 2); similar values have been reported by others 58. As well as in humans, the co‐expression of CD25 and FoxP3 for the characterization of T_regs in rodents is accepted widely. Analyses using both markers have shown a decreased frequency of this subpopulation when compared to expression of total FoxP3 alone (Table 1). This could be explained by the transient loss of CD25 expression in CD4⁺FoxP3⁺ T cells after chronic activation by internalization of the IL‐2/IL‐2R complex 59, or pT_regs with an effector memory phenotype that in steady‐state conditions has a low level of CD25 60. Nevertheless, after homeostatic proliferation or activation, CD25 expression could be restored, constituting a T_reg reservoir 61. We have found a high correlation between the CD25⁺FoxP3⁺ and CD4⁺FoxP3⁺ protocols for T_reg identification in spleen (r = 0·82, P = 0·03) of healthy mice. In addition, the use of the anti‐Foxp3 FJK‐16s clone conjugated with PE, in combination with permeabilization buffer from e‐Bioscience, allows a clearer separation of FoxP3⁺ cells compared to the MF23 clone 62.

The use of CD127 to identify mice T_regs has also been suggested, because their expression levels are lower than in CD4⁺CD25^– cells 63, and the depletion of T_regs through anti‐CD25 antibodies leads to a 90% reduction of CD4⁺CD25^highCD127^low cells 64. Similar to what occurs in humans, the use of this marker could also be controversial, as there is a subpopulation of activated CD4⁺FoxP3⁺ T_regs with high expression of CD127; in particular, T_regs expressing CD103 (Integrin α_E) and the inducible co‐stimulatory molecule (ICOS) 65. It is thought that the variation of CD127 expression in T_regs is mediated by their anatomical localization rather than by representing an intrinsic characteristic of T_regs 65. Consequently, the identification of mice T_regs requires at least the use of CD3, CD4, CD25 and FoxP3.

Expression of CD25, FoxP3 and CD127 to identify T_regs in rats

Characterization of T_regs in rats is less common, due mainly to lower availability of genetically modified animals and rat‐specific monoclonal antibodies. In general, the frequency of CD4⁺CD25⁺ T cells in Wistar rats is approximately 8% in peripheral and lymphoid tissue (Table 1, Fig. 3) and is supported by other studies in CD‐IGS and Dark Agouti rats 66, 67. In addition, we have found approximately 7·6% and 6·3% of CD4⁺FoxP3⁺ T_regs in peripheral blood and lymph nodes, respectively, which seems to be similar to previously reported percentages 68. Similar to humans and mice, not all CD4⁺FoxP3⁺ T_regs express CD25; only approximately 60% are CD25⁺ (Table 1). In addition, we have found approximately 4% of CD4⁺ T cells co‐expressing CD25 and FoxP3 in the periphery and lymphoid tissues of Wistar rats (Table 1, Fig. 3), similar to reports in Lewis rats 69 and Sprague–Dawley rats 70.

Figure 3.

Representative plots of gating strategy for flow cytometry showing different ways for rat regulatory T cell (T_reg) characterization. Upper panel shows the forward‐scatter height (FSC‐H) versus forward‐scatter area (FSC‐A) plot for doublets exclusions followed by a FSC‐A versus side‐scatter area (SSC‐A) plot to delimitate the lymphocyte region. Then, a CD3 versus CD4 plot was used to gate CD4⁺ T cells, and percentages of CD4⁺CD25⁺, CD4⁺forkhead box protein 3 (FoxP3⁺) and CD25⁺FoxP3 cells are shown on plots from lymph nodes of healthy Wistar rats. Lower panel shows representative plots and percentages of FoxP3⁺ cells, cytotoxic T lymphocyte antigen‐4 (CTLA‐4⁺) and FoxP3⁺CD103⁺ cells on the CD4⁺ gate. Numbers in brackets represent the % of CD4⁺FoxP3⁺ T cells positive for CTLA‐4 and CD103. Clone and fluorochrome conjugates from the monoclonal antibodies used are shown in Supporting information, Table S1. Clone of anti‐FoxP3 was FJK‐16s.

In addition, we have observed that the levels of CD4⁺FoxP3⁺CD25⁺ T_regs in rats are comparable with those in humans (Table 1), as also shown by Jacobo et al. 71. Because the CD25⁺FoxP3⁺ combination correlates strongly with the CD4⁺FoxP3⁺ protocol in spleen (r = 0·85, P < 0·0001), peripheral blood (r = 0·72, P = 0·003) and lymph nodes (r = 0·78, P = 0·0008) of rats, and to a lesser extent of CD127 to characterize rat T_regs, CD25⁺FoxP3⁺ is the best way to identify them.

Phenotypical characterization of T_regs in humans, mice and rats

Although several phenotypical markers have been used to characterize T_regs, we have focused on some classical molecules for which there is a large body of evidence in the literature concerning the expression of these molecules in T_regs, along with our own experience using them in T_regs in humans and rodents.

CTLA‐4

CTLA‐4 expression in T_regs is not only important for their suppressive function, but also to regulate their homeostasis and frequency in steady‐state conditions 72.

In humans, several authors have reported that CD4⁺CD25^+/high T cells express intracellular CTLA‐4 constitutively 21, 22, which correlates with a high suppressive activity, even in the CD4⁺CD25⁺FoxP3^– subset 73. However, the percentages of T_regs expressing intracellular CTLA‐4 are variable, and seem to depend upon the tissues recovered and on interindividual variations. For instance, some researchers have reported that between 50 and 90% of the peripheral FoxP3⁺ T_regs express CTLA‐4 in healthy human adults and neonates 74, 75; however, in our population we have found approximately 23% T_regs expressing CTLA‐4 in peripheral blood (Table 1), similar to that reported by Toussirot et al. 76. In our experience in rectal mucosae, approximately 50% FoxP3⁺ T_regs express CTLA‐4 (Table 1, Fig. 1).

In mice, the expression of CTLA‐4 by T_reg cells can vary depending on the tissue evaluated; in lymphoid tissue approximately 50% of T_regs express CTLA‐4 77, whereas in peripheral blood it is approximately 25% 78; moreover, these values can increase with age 77.

In rats, T_reg phenotypes have been less studied; we have observed that approximately 20% of CD4⁺FoxP3⁺ T_regs from blood, spleen and lymph nodes express CTLA‐4 (Table 1, Fig. 3); also, in purified splenic CD4⁺CD25⁺ T_regs from Wistar rats the expression has been reported to be approximately 40% 79, whereas in Sprague–Dawley rats a lower frequency (approximately 1%) of splenic CTLA‐4⁺ T_regs was observed 80.

Integrin α_E (CD103)

It has been reported that CD4⁺CD25⁺CD103⁺ T_regs produce high amounts of IL‐10 and exhibit the strongest suppressor capacity 81. In humans it was reported that CD103 is expressed by fewer than 5% of T_regs in the blood 75, 82. We have also found approximately 9% of CD103⁺ T_regs in blood and approximately 20% in biopsies from rectal mucosae (Table 1, Fig. 1). The high frequency of these cells in rectal mucosae could be associated with their memory phenotype and with a TGF‐β‐rich microenvironment provided by CD103⁺ DCs 83.

Most of the evidence for the role of CD103 in T_regs has been reported in mice; in steady state approximately 24% of lymphoid (non‐mesenteric lymph nodes) and splenic T_regs can express this molecule 78, 84, 85; however, in gut tissue the proportions of T_regs expressing CD103 are increased, especially in the lamina propria, with percentages of approximately 80% 85.

We have found that approximately 39% of the peripheral blood and lymph nodes FoxP3⁺ T_regs express CD103 in Wistar rats, whereas in spleen approximately 30% express CD103 (Table 1, Fig. 3). In Sprague–Dawley rats it has been observed that approximately 10% of splenic Foxp3⁺ T_reg express CD103 80.

Programmed death‐1 (PD‐1)

In healthy individuals, we have observed approximately 4 and 17% of human CD4⁺FoxP3⁺ T_regs expressing PD‐1 in peripheral blood and rectal mucosae, respectively (Table 1, Fig. 1 74). However, under certain conditions, such as chronic viral infection, their expression could be up‐regulated 11. Conversely, compartmentalization of PD‐1 expression has been reported, as approximately 90% of the CD4⁺FoxP3⁺ in healthy mice express PD‐1 intracellularly in lymph nodes and spleen 86, while we have found that more than 50% of mouse CD4⁺FoxP3⁺ T_regs in lymphoid tissue express PD‐1 on the surface (Table 1, Fig. 2), which might correspond to a effector/memory phenotype. So far, there are no reports about PD‐1 expression in rats.

GITR

GITR (CD357) is a member of the TNF receptor superfamily (TNF‐RSF) expressed constitutively in T_regs and up‐regulated in T_cons after activation 78. It has been demonstrated to be crucial for their thymic development, enhancing sensitivity to IL‐2 87. The role of GITR on T_regs is controversial in the periphery, as GITR engagement favours the expansion of competent functional T_regs preferentially 88; however, other reports have suggested that signalling mediated by GITR in T_regs leads to down‐regulation of different pathways that control their suppressive function 89. Moreover, it was proposed recently that GITR can inhibit FoxP3 cell induction, and in turn induce polarization to Th9⁺ cells mediated by epigenetic changes 90.

In humans, we have observed approximately 10% of CD4⁺FoxP3⁺ or CD4⁺CD25⁺FoxP3⁺ expressing GITR in peripheral blood (Table 1), in agreement with the results of Li et al., who found approximately 15% of CD4⁺CD25⁺GITR⁺ cells in healthy donors, whereas only 5·2% of the CD4⁺CD25^– cells expressed this marker 91. In contrast, in lamina propria of colonic mucosae approximately 40% of CD4⁺CD25⁺ cells can express GITR 92.

In mice, high expression levels of GITR have been observed in different tissues. We observed that more than 75% of T_regs in lymphoid tissues from BALB/c mice express this molecule (Table 1, Fig. 2). Previous studies in BALB/c and C57BL/6 mice have also reported that more than 80% of FoxP3⁺ T_regs express GITR in the spleen, lymphoid nodes, peripheral blood and thymus 52, 77. The high expression of GITR in T_regs from lymphoid tissues could be due the high proportion of memory T_regs found in these compartments compared to blood 93.

So far, there are no reports about GITR expression in rats.

OX40 (CD134)

OX40 is a co‐stimulatory molecule that belongs to the TNF‐RSF family. It is expressed constitutively in T_regs, although it is also up‐regulated in T cells following activation 78. As well as GITR, OX40 triggering on T_cons and CD8⁺ T cells promotes their proliferation and survival 94, but their function on T_regs is controversial. Some researchers have demonstrated that the OX40 signal promotes proliferation and survival of CD4⁺FoxP3⁺ T_regs in vitro and in vivo 95, 96; however, some evidence suggests that it did not appear to affect T_reg homeostasis 97, 98. Moreover, OX40 can also down‐regulate FoxP3 expression and impair the suppressive capacity of T_regs 98, 99. Probably, a dual effect as observed in T cell subpopulations could be attributed to the ligation strength of OX40, as high concentrations of the anti‐hOX40 antibody decreases T_reg suppressive function and cell numbers preferentially but maintains the proliferation of naive and memory T cells that become resistant to suppression by T_regs 100. In addition, ligation of OX40 from T_regs with OX40L on T_cons partially boosts the T_reg survival capacity in low inflammation conditions, where other factors such as TNF and plasmacytoid DCs could act to help T_reg activation in such conditions 101. Furthermore, in highly inflammatory conditions OX40 seems to mediate T_reg accumulation in the colon and peripheral lymphoid organs 102.

In humans, it has been reported that OX40 expression in CD25⁺CD127^–/lowFoxP3⁺ T_regs is ∼10%; on the contrary, in young and aged BALB/c mice its expression is approximately 60% in splenic CD4⁺FoxP3⁺ T_regs and values of approximately 80% in colonic T_regs has been observed 102, 103. We have observed similar percentages of OX40 in BALB/c mice in CD4⁺FoxP3⁺ T_regs from secondary lymphoid tissues (Table 1, Fig. 2). Conversely, the percentages of OX40 in splenic CD4⁺CD25⁺ T_regs are approximately 17% in Lewis rats 104.

Conclusions

The establishment of different experimental scenarios generated through polyclonal or antigen‐specific activation or through in‐vivo models mimicking steady state or low‐/high‐grade inflammation conditions have allowed delineation of the biology of T_regs. Several antibodies have been used to identify and characterize the T_reg populations, using the differential expression of markers, which could be mediating functional advantages. However, as these molecules can be expressed by other cells, it is necessary to select the combination of markers with the lowest variability, being particularly cautious when inflammatory conditions could be present. Similarly, the frequency and phenotype of T_regs can differ according to evaluated tissue or species. Finally, among the key points regarding T_reg typification that need to be considered are the following:

• Use viability stain, especially when handling cryopreserved samples.

• A proper selection of the antibody clones and fluorochromes. Information about the antibodies that we have used is shown in Supporting information, Table S1.

• Titration of antibodies and, where possible, using the same batch of antibodies.

• To define positive events it is recommended to use fluorescence minus one (FMO) rather than isotype controls. In some cases, negative populations that do not express the specific marker can also help to identify positive populations.

• Avoid discrimination of the T_reg population using high or intermediate expression levels of markers composing the T_reg panel.

• The minimum panel to characterize T_reg cells from all species should be CD3, CD4, CD25, and FoxP3.

• On flow cytometry, collect at least 100 000 gated lymphocytes for each sample.

Disclosure

The authors declare that there are no disclosures regarding the publication of this paper.

Supporting information

Additional Supporting information may be found in the online version of this article at the publisher's web‐site:

Table S1. Antibodies used for flow cytometry