Foxp3+ cells and CTLA-4 have been ascribed major roles in downregulating immune responses. To address the relationship between CTLA-4 expression and Foxp3+ cells, we generated littermate CTLA-4–sufficient (Ctla4+/+), CTLA-4–haploinsufficient (Ctla4+/−), and CTLA-4–deficient (Ctla4−/−) Foxp3-gfp knock-in C57BL/6 mice, permitting us to characterize the phenotype of Foxp3+ cells and to test their ex vivo T regulatory (Treg) suppressor activity. CD3+, CD4+, and CD8+ cells, but not CD19+ cells, were markedly expanded in Ctla4−/− mice compared with Ctla4+/+ or Ctla4+/− mice. In Ctla4−/− mice, the relative expansion of the Foxp3+ population was greater than that of the CD3+, CD4+, or CD8+ populations because of increased survival of Foxp3+ cells. Foxp3+ Treg cells from Ctla4−/− mice and Foxp3+ Treg cells from Ctla4+/+ mice exerted identical ex vivo suppressor function. This may be related to differential expression of GITR, CD73, and CD39 on Foxp3+ Treg cells from Ctla4−/− mice versus that on corresponding cells from littermate Ctla4+/+ or Ctla4+/− mice, with GITR and CD39 being upregulated and CD73 being downregulated on Foxp3+ Treg cells from Ctla4−/− mice. Moreover, CTLA-4 expression in Ctla4+/+, Ctla4+/−, and Ctla4−/− mice correlated with their percentages of Foxp3+ cells, suggesting an important role for CTLA-4 expression in Treg cell homeostasis. This may have vital ramifications for the treatment of patients for whom augmentation of suppressor function would be beneficial (e.g., patients with autoimmune diseases) and for whom diminution of suppressor function would be beneficial (e.g., patients with cancer).
Expression of Foxp3 is indispensable to immune homeostasis. This is evidenced by the rapid and lethal autoimmune T cell proliferative disease observed in both Foxp3−/− mice and Scurfy mice (which bear a nonfunctional mutated Foxp3 gene) and by the ability of adoptively transferred T regulatory (Treg) cells from a Foxp3-sufficient host to rescue disease development in Foxp3-deficient recipients (1–3).
Foxp3, being a transcription factor, is expressed intracellularly, but not on the cell surface. Accordingly, determination of whether a cell is Foxp3+ requires fixation and permeabilization of that cell to permit entry of staining agents that interrogate the intracellular contents. Although this approach is effective in identifying Foxp3+ cells, it kills the cells, thereby rendering it impossible to study the function of the Foxp3+ cells.
To circumvent this limitation, Foxp3-gfp knock-in mice have been generated in which in the native Foxp3 gene has been deleted and replaced with a Foxp3-gfp construct (4). In such mice, Foxp3+ cells express GFP, whereas Foxp3− cells do not. This specific and restricted expression of GFP in Foxp3+ cells permits not only identification of Foxp3+ cells by standard flow cytometry procedures but permits isolation of intact living Foxp3+ cells by FACS for functional analyses.
Foxp3 transcriptionally upregulates the Ctla4 gene (5, 6), and CTLA-4 is expressed constitutively on CD4+Foxp3+ Treg cells (7, 8). Ctla4−/− mice develop a rapid and lethal T cell proliferative disease, like that observed in Foxp3−/− or Scurfy mice (9, 10). Thymic positive and negative selection are each normal in Ctla4−/− mice (11, 12), indicating that the physiologic defect in these mice lies in the control of peripheral T cell activation rather than in central T cell development, and the diverse and unbiased TCR repertoire among the activated T cells in Ctla4−/− mice (13) indicates that no individual self or environmental Ag uniquely drives the pathologic response.
Under certain conditions, subtotal neutralization of CTLA-4 can promote and/or exacerbate autoimmunity. As examples, treatment of human CTLA4 knock-in mice with an anti–CTLA-4 mAb promoted development of circulating anti-dsDNA Abs (14), partial blockade of CTLA-4 promoted development of juvenile-onset diabetes in mice bearing a type 1 diabetes-permissive MHC locus (H2g7) (15), and administration of anti–CTLA-4 mAb accelerated and exacerbated onset and severity of experimental autoimmune encephalomyelitis (EAE) and autoimmune diabetes (16, 17).
These examples notwithstanding, reduction in CTLA-4 expression not only does not immutably promote autoimmunity or autoimmune disease but may have no effect on autoimmune disease (18) or may actually protect from autoimmune disease (19–21). Taken together, the divergent results suggest a complex relationship between CTLA-4 expression and Foxp3+ cells.
To address this issue, we generated littermate CTLA-4–sufficient (Ctla4+/+), CTLA-4–haploinsufficient (Ctla4+/−), and CTLA-4–deficient (Ctla4−/−) mice that each bore a Foxp3-gfp knock-in construct, thereby permitting us not only to characterize the phenotype of CD4+Foxp3+ Treg cells but to isolate them and test their suppressor activity in an ex vivo assay. Our experiments demonstrated that CD3+, CD4+, and CD8+ cells, but not CD19+ cells, were markedly expanded in Ctla4−/− mice compared with either Ctla4+/+ or Ctla4+/− littermate mice. In these Ctla4−/− mice, the relative expansion of the Foxp3+ population was greater than that of the CD3+, CD4+, or CD8+ populations. Foxp3+ Treg cells from Ctla4−/− mice and Foxp3+ Treg cells from Ctla4+/+ mice possessed identical ex vivo suppressor function, perhaps related to differential expression of GITR, CD73, and CD39 on Foxp3+ Treg cells from Ctla4−/− mice versus that on Foxp3+ Treg cells from littermate Ctla4+/+ or Ctla4+/− mice. Moreover, the percentages of Foxp3+ cells in Ctla4+/+, Ctla4+/−, and Ctla4−/− mice correlated with the degree of CTLA-4 expression, suggesting that the level of CTLA-4 expression plays an important role in Treg cell homeostasis.
Materials and Methods
All reported studies were approved by the University of Southern California Institutional Animal Care and Use Committee.
All mice used in this study bore the identical Foxp3-gfp knock-in. To achieve this, we obtained C57BL/6 (B6) mice bearing a Foxp3-gfp knock-in from Dr. Song Guo Zheng. These Foxp3-gfp knock-in mice were intercrossed with B6.Ctla4+/− mice (22) to yield B6.Ctla4+/− mice bearing the Foxp3-gfp knock-in. Because the Foxp3 gene is located on the X chromosome, male mice were hemizygous for the Foxp3-gfp knock-in, whereas females were bred to homozygosity for the Foxp3-gfp knock-in.
To generate B6.Ctla4−/− mice, we mated B6.Ctla4+/− male and female mice to yield littermate B6.Ctla4+/+, B6.Ctla4+/−, and B6.Ctla4−/− mice. Because B6.Ctla4−/− mice die by 3–6 wk of age (9, 10), experiments with these mice (and their littermates) were conducted at 3 wk of age.
All mice were housed in the same specific pathogen-free room.
Cell surface staining
Spleen mononuclear cells were stained singly or with combinations of fluorochrome-conjugated mAb specific for CD3, CD4, CD8, CD19, CD25, GITR, CD73, and CD39 (BioLegend or BD Pharmingen) and analyzed by FACS using FlowJo V.7.6.1 software (Treestar).
Intracellular staining for CTLA-4
Spleen mononuclear cells were surface stained for CD4 and CD25, fixed and permeabilized, and stained with PE-conjugated anti–CTLA-4 mAb. Control samples were identically treated, substituting PE-conjugated Armenian hamster IgG isotype control mAb for the anti–CTLA-4 mAb (BioLegend). Cells were gated on the CD4+CD25+Foxp3+ (Treg cell) population, and data were analyzed for CTLA-4 expression.
Normalized mean fluorescence intensity
For expression of CTLA-4, GITR, CD73, and CD39 in each individual experiment (which included all littermates of a given litter), the normalized mean fluorescence intensity (MFI) was calculated as ([sample MFI] − [mean MFI])/SD. Results from every experiment, by virtue of normalization to a SD of 1.0, could be pooled and were reported as the number of SDs above or below the mean. Only those experiments in which there was at least one mouse from each Ctla4 genotype (+/+, +/−, −/−) were included.
In vivo proliferation of Foxp3+ cells
Spleen mononuclear cells were surface stained for CD4, fixed and permeabilized, and stained with allophycocyanin-conjugated anti–Ki-67 mAb. Control samples were identically treated, substituting Armenian hamster IgG isotype control mAb for the anti–Ki-67 mAb (BioLegend). Cells were gated on either the CD4+Foxp3+ or CD4+Foxp3− populations and were analyzed for Ki-67 expression.
In vivo survival of Foxp3+ cells
Spleen mononuclear cells were stained for CD4 and stained with the APC Annexin V Apoptosis Detection Kit with 7-AAD (7-aminoactinomycin D) (BioLegend) according to the manufacturer’s instructions. Cells were gated on either the CD4+Foxp3+ or CD4+Foxp3− populations and were analyzed for Annexin V expression and 7-AAD inclusion.
Ex vivo Treg cell function
T responder (Tresp; CD4+Foxp3−) cells from wild-type mice and Treg (CD4+Foxp3+) cells from B6.Ctla4+/+, B6.Ctla4+/−, and B6.Ctla4−/− mice were isolated by FACS spleen CD4+ T cells. The Tresp cells were stained with proliferation dye eFluor 670 (eBioscience) and were stimulated with anti-CD3 mAb (0.025 mg/ml; BioLegend) and irradiated B6.Ctla4+/+ non-T cells (30 Gy, 1:1 ratio) in the presence of graded ratios of Treg cells. At day 3, cells were harvested, and the division index (DI; total number of Tresp cell divisions/original number of Tresp cells), the proliferation index (PI; total number of Tresp cell divisions/number of Tresp cells undergoing at least one division), and percentage of dividing cells (number of Tresp cells undergoing at least one division/original number of Tresp cells) were evaluated by FACSCalibur (BD Biosciences). Results are expressed as % suppression, defined as ([the value obtained in cultures with Tresp cells only] − [the value obtained in cultures with Tresp cells + Treg cells])/(the value obtained in cultures with Tresp cells only).
All analyses were performed using SigmaPlot software. Parametric testing between two groups was performed by the unpaired t test. When the data were not normally distributed or the equal variance test was not satisfied, nonparametric testing was performed by the Mann–Whitney rank sum test between two groups. A p value ≤0.05 was considered significant.
Expansion of T cells in B6.Ctla4−/− mice
Multiple matings between male and female B6.Ctla4+/− mice yielded 34 female and 27 male B6.Ctla4+/+ mice, 49 female and 52 male B6.Ctla4+/− mice, and 22 female and 22 male B6.Ctla4−/− mice. At 3 wk of age, no differences were appreciated in spleen total mononuclear cells, CD19+ cells, CD3+ cells, CD4+ cells, or CD8+ cells between male and female mice of a given Ctla4 genotype (p ≥ 0.062 for any comparison), so results for male and female mice of a given genotype were pooled.
Spleen total mononuclear cells were modestly expanded in B6.Ctla4−/− mice relative to those in B6.Ctla4+/+ or B6.Ctla4+/− mice (41.1% and 33.8% increases, respectively), whereas no difference was appreciated between B6.Ctla4+/+ and B6.Ctla4+/− mice (Fig. 1A). In agreement with our previous results (22), this mononuclear cell expansion was driven by T cells. Whereas median numbers of CD19+ cells were similar in all three cohorts, median numbers of CD3+, CD4+, and CD8+ cells were considerably greater in B6.Ctla4−/− mice than in B6.Ctla4+/+ (322%, 351%, and 331% increases, respectively) or B6.Ctla4+/− mice (322%, 365%, and 331% increases, respectively). No differences in CD3+, CD4+, and CD8+ cells were appreciated between B6.Ctla4+/+ and B6.Ctla4+/− mice (Fig. 1B–E).
Preferential expansion of Foxp3+ cells in B6 mice as a function of Ctla4 genotype
To determine whether the lethal T cell proliferative disease that develops in Ctla4−/− mice (9, 10) could be explained by a relative deficiency in Foxp3+ Treg cells, we determined the percentages and numbers of Foxp3+ cells in littermate B6.Ctla4+/+, B6.Ctla4+/−, and B6.Ctla4−/− mice (Fig. 2). Not only were Foxp3+ cells not reduced in B6.Ctla4−/− mice, but they were markedly greater in B6.Ctla4−/− mice (median 2.58 × 106) than in B6.Ctla4+/+ (median 0.39 × 106) or B6.Ctla4+/− (median 0.46 × 106) mice (reflecting 562% and 461% increases, respectively; (Fig. 2E). Of note, Foxp3+ cells were modestly, but significantly, greater in B6.Ctla4+/− mice than in B6.Ctla4+/+ mice; that is, the numbers of Foxp3+ cells associated with the Ctla4 genotype being the least in B6.Ctla4+/+ mice, intermediate in B6.Ctla4+/− mice, and greatest in B6.Ctla4−/− mice.
The relative expansion of Foxp3+ cells in B6.Ctla4−/− mice was considerably greater than the relative expansions of CD3+, CD4+, or CD8+ cells in these mice, with Foxp3+ cells as percentages of CD3+ cells or CD4+ cells being greater in B6.Ctla4−/− mice than in B6.Ctla4+/+ (43.4% and 39.5% increases, respectively; (Fig. 2F) or B6.Ctla4+/− mice (22.4% and 16.7% increases, respectively; (Fig. 2G). As with numbers of Foxp3+ cells, percentages of Foxp3+ cells were greater in B6.Ctla4+/− mice than in B6.Ctla4+/+ mice (Fig. 2F, 2G), collectively reinforcing an inverse relationship between CTLA-4 expression and expansion of Foxp3+ cells.
Because CD4+CD25−Foxp3+ cells may display in vivo plasticity and convert to proinflammatory Th17 cells (23), we also assessed percentages of CD4+CD25+Foxp3+ cells. As with the entire Foxp3+ population, the relative expansion of CD4+CD25+Foxp3+ cells was greater in B6.Ctla4−/− mice than in B6.Ctla4+/+ or B6.Ctla4+/− mice (52.3% and 31.2% increases, respectively), with the percentage again being greater in B6.Ctla4+/− mice than in B6.Ctla4+/+ mice (Fig. 2H). Indeed, regression line analysis demonstrated that the percentages of Foxp3+ cells were greatest in B6.Ctla4−/− mice, followed by B6.Ctla4+/− mice, followed by B6. Ctla4+/+ mice for any given CD19+ cell count (Fig. 2I). A similar regression line pattern was also observed for CD4+CD25+Foxp3+ cells (Fig. 2J).
In vivo proliferation of Foxp3+ cells in B6 mice as a function of Ctla4 genotype
To determine whether the preferential expansion of Foxp3+ cells in B6.Ctla4−/− mice reflected preferential proliferation of these cells, we assessed expression of Ki-67, a marker of recent cell division, in both CD4+Foxp3+ and CD4+Foxp3− cells. Compared with proliferation of CD4+Foxp3+ cells in B6.Ctla4+/+ or B6.Ctla4+/− mice, proliferation of these cells in B6.Ctla4−/− mice was modestly greater (16.1% and 11.5% increases, respectively; (Fig. 3). Of note, the respective relative increases in proliferation of CD4+Foxp3− cells in B6.Ctla4−/− mice were much greater (200% and 143%, respectively), so preferential expansion of Foxp3+ cells in B6.Ctla4−/− mice cannot be explained by preferential proliferation.
In vivo survival of Foxp3+ cells in B6 mice as a function of Ctla4 genotype
To determine whether the preferential expansion of Foxp3+ cells in B6.Ctla4−/− mice reflected preferential survival of these cells, we assessed the absence of Annexin V expression by, and the exclusion of 7-AAD from, CD4+Foxp3+ and CD4+Foxp3− cells. Compared with survival of CD4+Foxp3+ cells in B6.Ctla4+/+ or B6.Ctla4+/− mice, survival of these cells in B6.Ctla4−/− mice was substantially greater (47.7% and 65.4% reductions in death, respectively; p = 0.019 and p = 0.002, respectively; (Fig. 4). In contrast, survival of CD4+Foxp3− cells in B6.Ctla4−/− mice was no different from that in either B6.Ctla4+/+ or B6.Ctla4+/− mice, so preferential expansion of Foxp3+ cells in B6.Ctla4−/− mice likely is attributable, at least in part, to preferential survival of these cells.
Ex vivo suppressor activity of Foxp3+ Treg cells from B6.Ctla4+/+ and B6.Ctla4−/− mice
Given that B6.Ctla4−/− mice develop a lethal T cell proliferative disease despite the increased numbers and percentages of CD4+Foxp3+ cells in these mice, we asked whether the suppressor function of Foxp3+ Treg cells in Ctla4−/− mice was reduced. To that end, we used Tresp cells from B6.Ctla4+/+ or B6.Ctla4−/− mice as the targets and tested the ability of CD4+Foxp3+ Treg cells from B6.Ctla4+/+ or B6.Ctla4−/− mice to suppress proliferation. Ex vivo suppressor activity of Foxp3+ Treg cells from B6.Ctla4−/− mice was not different from that of Foxp3+ Treg cells from littermate B6.Ctla4+/+ mice in three independent head-to-head experiments using Tresp cells from B6.Ctla4+/+ mice, regardless of whether suppression was assessed by DI, PI, or percentages of cells that underwent division (Fig. 5A, 5B). This also remained the case in two additional head-to-head experiments using Tresp cells from B6.Ctla4−/− mice as the targets (Fig. 5C).
Expression of CTLA-4 and Treg cell–associated markers in B6.Ctla4+/+, B6.Ctla4+/−, and B6.Ctla4−/− mice
Given that suppressor activity of Foxp3+ Treg cells from B6.Ctla4−/− mice is as great as that of Foxp3+ Treg cells from B6.Ctla4+/+ mice, we checked expression of CTLA-4 and confirmed that the CTLA-4 phenotype corresponded to the Ctla4 genotype, with CTLA-4 not being detectable in B6.Ctla4−/− mice and CTLA-4 being expressed by B6.Ctla4+/− mice at a level intermediate to that expressed by B6.Ctla4+/+ mice (Fig. 6A).
In parallel, we also assessed surface expression of GITR, CD73, and CD39 markers with known effects on Treg cell function (reviewed in Refs. 24, 25). Expression of each of these markers on Foxp3+ Treg cells from B6.Ctla4−/− mice diverged from that on Foxp3+ Treg cells from B6.Ctla4+/+ or B6.Ctla4+/− mice, with GITR and CD39 being significantly upregulated and CD73 being significantly downregulated on Foxp3+ Treg cells from B6.Ctla4−/− mice (Fig. 6B–D). No difference in expression of GITR was noted between Foxp3+ Treg cells from B6.Ctla4+/+ mice and Foxp3+ Treg cells from B6.Ctla4+/− mice, whereas expression of CD73 and CD39 on Foxp3+ Treg cells from B6.Ctla4+/− mice was modestly greater than that on Foxp3+ Treg cells from B6.Ctla4+/+ mice.
CTLA-4 has long been known and universally accepted as a key regulator of T cell activation, and multiple mechanisms of action likely underpin its suppressor activity. Some studies have pointed to a cell-intrinsic mechanism leading to inhibition of signaling downstream from the TCR (26, 27), and other studies have pointed to the preferential binding of CD80/CD86 to CTLA-4 rather than to CD28 (28, 29), thereby blocking T cell costimulation.
Regardless of whether the biologic effect of CTLA-4 is cell intrinsic or cell extrinsic, CTLA-4 has had a great impact in multiple medical disciplines. In the rheumatologic arena, abatacept, a fusion protein between CTLA-4 and IgG Fc, is FDA approved for the treatment of rheumatoid arthritis, juvenile idiopathic arthritis, and psoriatic arthritis. In the organ transplant arena, belatacept, also a fusion protein between CTLA-4 and IgG Fc, is FDA approved for the prophylaxis of organ rejection in adult patients receiving a kidney transplant. The clinical benefits of both abatacept and belatacept come from their functional antagonism of the costimulatory interactions between CD28 on T cells with CD80/CD86 on APCs. In the oncologic arena, ipilimumab, an anti-CTLA-4 mAb, is FDA approved for treatment of melanoma, renal cell carcinoma, colorectal cancer, hepatocellular carcinoma, non-small cell lung cancer, and malignant pleural mesothelioma. The clinical benefit of ipilimumab comes from its blocking the antagonistic effects of CTLA-4, thereby permitting augmented activating interactions between CD28 on T cells with CD80/CD86 on APCs.
Complete absence of CTLA-4 in Ctla4−/− mice promotes unopposed CD28-driven costimulation of T cells, resulting in marked T cell expansion, infiltration of vital organs, and death (9, 10). Conversely, Ctla4−/− mice deficient in either CD80/CD86 or CD28 are protected from such lethality (30, 31), and blocking engagement of CD28 to CD80/CD86 with CTLA-4–Ig in such Ctla4−/− mice prevents development of lymphoproliferation as long as CTLA-4–Ig continues to be administered (32).
By studying B6.Ctla4−/− mice at 3 wk of age (i.e., a time just before the onset of mortality in such mice) and their B6.Ctla4+/+ or B6.Ctla4+/− littermates, we confirmed the T cell expansion in non–autoimmune-prone mice, consistent with previous results (9, 10, 22). Of note, Foxp3+ cells expanded to a greater extent than did Foxp3− cells. Whereas proliferation of Foxp3− cells was greater than proliferation of Foxp3+ cells in B6.Ctla4−/− mice, survival in these mice of Foxp3+ cells was greater than survival of Foxp3− cells, leading to a net preferential expansion of Foxp3+ cells.
Relative expansion of Foxp3+ cells was not limited to B6.Ctla4−/− mice but was also appreciated to a lesser degree in B6.Ctla4+/− mice. For any given CD19+ cell count, percentages of Foxp3+ cells were greatest in B6.Ctla4−/− mice, followed by B6.Ctla4+/− mice, followed by B6.Ctla4+/+ mice. That is, the extent of Foxp3+ cell expansion was related to the degree of CTLA-4 expression. Analysis focused solely on CD4+CD25+Foxp3+ cells gave the same result, with their percentages being greatest in B6.Ctla4−/− mice, intermediate in B6.Ctla4+/− mice, and lowest in B6.Ctla4+/+ mice. The molecular mechanisms that underpin these results require further investigation.
Because our mice harbored a Foxp3-gfp knock-in, we were able to assess not only numbers and percentages of Foxp3+ cells, but we were also able to directly isolate Foxp3+ cells and assess their Treg cell (suppressor) function ex vivo. Our results documenting preferential expansion of the Foxp3+ population in B6.Ctla4−/− mice coupled with intact ex vivo suppressor activity of these Foxp3+ cells as compared with their Ctla4+/+ or Ctla4+/− littermates validates the findings of previous studies in which direct isolation of Foxp3+ cells was not possible (33, 34).
Although the preferential expansion of Foxp3+ Treg cells in B6.Ctla4−/− mice does not protect these mice from early lethality, expansion of Foxp3+ Treg cells in a CTLA-4–deficient host can be clinically important in multiple contexts. Indeed, the expanded Foxp3+ Treg cell population in Ctla4−/− mice plays a role in protecting against EAE (20), and Treg cells from Ctla4−/− mice can be protective in an adoptive transfer model of colitis (19). Moreover, conditional deletion of CTLA-4 in adulthood results in expansion of (CTLA-4−) Foxp3+ Treg cells and confers protection from EAE (21). Whether this observation is recapitulated in other murine autoimmune diseases and/or in human autoimmune diseases needs to be experimentally investigated.
Whereas Ab-mediated blockade of CTLA-4 can inhibit suppressor function of CTLA-4+ cells (7, 8), the molecular underpinnings of suppressor function mediated by CTLA-4–deficient Treg cells remain incompletely defined. We herein demonstrated that GITR and CD39 are greatly upregulated on Foxp3+ Treg cells from B6.Ctla4−/− mice, whereas CD73 is greatly downregulated on these cells. Whether these dramatic differences in expression of GITR, CD73, and CD39 between Foxp3+ Treg cells from B6.Ctla4−/− mice on the one hand and Foxp3+ Treg cells from B6.Ctla4+/+ or B6.Ctla4+/− mice on the other hand are directly related to the intact suppressor activity of CTLA-4−deficient Treg cells remains to be experimentally determined. Regardless, these observations, coupled to those of others that Lag3 and PD-1 are upregulated on CTLA-4–deficient cells (21), indicate that consequent to changes in CTLA-4 expression, changes occur in surface expression of multiple surface molecules known to contribute to suppressor function, thereby strongly suggesting that Treg cells employ homeostatic mechanisms to preserve their function. This may have vital ramifications for the treatment of patients for whom augmentation of suppressor function would be beneficial (e.g., patients with autoimmune diseases) and for whom diminution of suppressor function would be beneficial (e.g., patients with cancer).
This work was supported by a grant from the Selena Gomez Fund.
Abbreviations used in this article
The authors have no financial conflicts of interest.