Morbidly Obese Human Subjects Have Increased Peripheral Blood CD4⁺ T Cells With Skewing Toward a Treg- and Th2-Dominated Phenotype

A total of 13 morbidly obese (BMI >40 kg/m²) and 25 lean (BMI <25 kg/m²) healthy control subjects were included in this study. Subjects with overt type 2 diabetes mellitus or liver enzyme test abnormalities were excluded. The presence of concomitant medical illness was excluded by medical history assessment in morbidly obese and lean subjects. All subjects gave their written informed consent. The study was approved by the medical ethical committee of Erasmus University Medical Center.

Blood was obtained using vacuette sodium heparin–containing tubes (Greiner Bio-one, Alphen a/d Rijn, the Netherlands) and further processed within 1 h after collection. Plasma was isolated by centrifugation and frozen for further analyses. Peripheral blood mononuclear cells (PBMCs) were isolated using Ficoll density separation and viably frozen for further analyses.

Total leukocyte count was measured in freshly collected blood using a Coulter Counter (Beckman Coulter B.V., Woerden, the Netherlands), and leukocyte subpopulations were determined by flow cytometry based on CD45 expression and sideward scatter. For immunophenotyping of T-cell subpopulations, viably frozen PBMCs were used. Antibodies used for flow cytometric analyses and sorting experiments are summarized in Supplementary Table 1. T-cell subpopulations were defined as naive (CD45RA⁺ and CD27⁺), central memory (CD45RO⁺ and CD27⁺), effector memory (CD45RO⁺ and CD27⁻; although it is known that a small population of effector memory cells does still express CD27) (22), and terminally differentiated (CD45RA⁺ and CD27⁻). Natural Treg cells were defined as CD4⁺CD25⁺FoxP3⁺. For intracellular cytokine detection, PBMCs were stimulated with phorbol-12-myristate-13-acetate (PMA; 50 ng/mL; Sigma-Aldrich, St. Louis, MO) and ionomycin (500 ng/mL; Invitrogen Ltd., Paisley, UK) for 4 h in the presence of GolgiStop (BD Biosciences, San Jose, CA). Thereafter, cells were stained for extracellular markers, fixed with 2% paraformaldehyde, and permeabilized with 0.5% saponin, followed by intracellular staining for IFN-γ, IL-4, and IL-17A. Stained cells were measured using a FACS LSR-II (BD Biosciences), and data were analyzed with FlowJo software (TreeStar, Ashland, OR). Th1 cells were defined as CD4⁺IFN-γ⁺, Th2 cells as CD4⁺IL-4⁺, and Th17 cells as CD4⁺IL-17A⁺.

The following T-cell subpopulations were sorted with a purity of >90% in all samples using a FACSAria (BD Biosciences): CD4⁺ naive (CD3⁺TCR-γδ⁻CD4⁺CD45RO⁻CD27⁺), CD4⁺ memory (CD3⁺TCR-γδ⁻CD4⁺CD45RO⁺), CD8⁺ naive (CD3⁺TCR-γδ⁻CD8⁺CD45RO⁻CD27⁺), and CD8⁺ memory (CD3⁺TCR-γδ⁻CD8⁺CD45RO⁺) T cells. Total αβ-T cells were isolated with a purity of >90% in all samples using AutoMACS with allophycocyanin (APC)–labeled anti–TCR-αβ antibodies and anti-APC cell beads (Miltenyi Biotec GmbH, Bergisch Gladbach, Germany).

Signal joint TREC (sjTREC) analysis was used to evaluate thymic output and peripheral T-cell proliferation. The sjTREC is a stable circular DNA structure that is formed during TCR-α rearrangements in developing thymocytes (Supplementary Fig. 1A) (23). Because the sjTREC, in contrast to genomic DNA, is not duplicated during cell proliferation, it will dilute out during consecutive cell divisions, making it a useful marker to determine proliferation history in αβ-T cells (Supplementary Fig. 1B) (24).

To determine sjTREC dilution, DNA was extracted from different T-cell subpopulations using the GeneElute Mammalian Genomic DNA miniprep kit (Sigma-Aldrich). The sjTREC was detected by real-time quantitative PCR using an ABI Prism 7900 machine (Applied Biosystems, Foster City, CA) and the following primer and probe combination: forward primer 5′-CCATGCTGACACCTGGTT-3′, reverse primer 5′-TCGTGAGAACGGTGAATGAAG-3′, and probe 5′-CACGGTGATGCATAGGCACCTGC-3′ as described previously (24). To correct for total DNA input, the amount of albumin gene was determined (forward primer 5′-TGAACAGGCGACCATGCTT-3′, reverse primer 5′-CTCTCCTTCTCAGAAAGTGTGCATAT-3′, and probe 5′-TGCTGAAACATTCACCTTCCATGCAGA-3′) (25). Cycle threshold values for sjTREC were normalized to albumin cycle threshold values (25). For copy number quantification, serial dilutions of a pGEM-T Easy vector (Promega Benelux BV, Leiden, the Netherlands), in which the sjTREC sequence was cloned, were used (24).

To determine TCRB diversity, we performed GeneScan analysis for Vβ–Jβ gene rearrangements on DNA of sorted T-cell subpopulations with multiplex PCR as previously described (26).

Plasma levels of IL-2 (sensitivity >16.4 pg/mL), IL-4 (sensitivity >20.8 pg/mL), IL-6 (sensitivity >1.2 pg/mL), IL-10 (sensitivity >1.9 pg/mL), IL-12p70 (sensitivity >1.5 pg/mL), IL-17A (sensitivity >2.5 pg/mL), IFN-γ (sensitivity >1.6 pg/mL), CCL5 (sensitivity >25 pg/mL), and TNF-α (sensitivity >3.2 pg/mL) were measured simultaneously using bead-based FlowCytomix simplex kits (Bender Medsystems GmbH, Vienna, Austria). Plasma levels of IL-7 were measured by ELISA (Invitrogen Ltd).

Fasting blood glucose levels were measured with a Hitachi 917 Chemistry Analyzer (Roche Diagnostics, Almere, the Netherlands). Fasting insulin levels were measured using a chemiluminescent immunometric assay (Immulite 2000; Siemens Medical Solutions Diagnostics, Los Angeles, CA).

Subject characteristics are described as mean ± SD. The exact Mann-Whitney U test was used for statistical comparisons between morbidly obese and lean subjects. Statistics are displayed as median (range). All statistical analyses were performed with SPSS software version 15.0 (SPSS Inc., Chicago, IL). P < 0.05 (two-tailed) was considered statistically significant. Box-and-whisker plots display the 2.5 to 97.5 percentiles. Error bars are expressed as the SEM.

An initial general examination of the blood samples did not reveal clear differences in the total leukocyte number or the numbers of distinct leukocyte subpopulations between morbidly obese and lean subjects (characteristics of the subjects are summarized in Table 1). However, a trend toward increased lymphocyte numbers was present in morbidly obese subjects (Fig. 1A). Detailed flow cytometric analyses on isolated PBMCs revealed that NK- and B-cell numbers did not differ between the morbidly obese and lean subjects (Fig. 1B). T-cell numbers, however, were significantly (P < 0.01) increased in morbidly obese subjects. This was mainly due to a twofold increase in CD4⁺ T cells (P < 0.01), whereas CD8⁺ T-cell numbers remained normal (P = 0.35) (Fig. 1C). This resulted in an increased CD4-to-CD8 ratio (morbidly obese 2.82 [1.62–6.17] vs. lean 1.54 [1.29–5.23], P = 0.03, data not shown).

Next, we performed extensive flow cytometric analyses to determine whether distinct T-cell subpopulations are affected in morbidly obese subjects. Within the CD8⁺ T-cell compartment, cell numbers within the different subpopulations were similar in morbidly obese and lean subjects (Fig. 2A). Within the CD4⁺ T-cell compartment, increased numbers of naive (P = 0.04), central memory (P < 0.01), and terminally differentiated (P = 0.03) T cells were found in morbidly obese subjects, whereas no differences were found in the effector memory subpopulation between morbidly obese and lean subjects (Fig. 2A). Also, absolute counts of natural CD4⁺CD25⁺FoxP3⁺ Treg cells were increased (P < 0.01) (Fig. 2B). Effector CD4⁺ T-cell subpopulations were determined by measuring the intracellular cytokine profile after stimulation with PMA and ionomycin. The numbers of IFN-γ–producing T cells (Th1) and IL-17A–producing T cells (Th17) were similar in morbidly obese and lean subjects, whereas the number of IL-4–producing T cells (Th2) was increased in morbidly obese subjects (P = 0.03) (Fig. 2B).

Additional correlation analyses demonstrated a significant correlation between BMI and the number of total T cells as well as the numbers of CD4⁺ T cells, naive CD4⁺ T cells, terminally differentiated CD4⁺ T cells, central memory CD4⁺ T cells, natural CD4⁺CD25⁺FoxP3⁺ Treg cells, and Th2 cells. For age, only a significant correlation was observed with the number of Th2 cells (Supplementary Table 2).

Taken together, these data demonstrate that morbid obesity is associated with increased naive and memory CD4⁺ T cells and with increased numbers of anti-inflammatory natural CD4⁺CD25⁺FoxP3⁺ Treg cells and Th2 cells.

Several mechanisms, including increased thymic output, increased peripheral proliferation, decreased apoptosis, or altered redistribution, can account for the observed increased CD4⁺ T-cell numbers found in morbidly obese subjects. To distinguish between increased thymic output and increased peripheral proliferation or survival, the sjTREC content in peripheral blood T-cell subpopulations was determined (Supplementary Fig. 1C). A significantly lower sjTREC content, which together with increased cell numbers resembles increased proliferation (Supplementary Fig. 1C), was found in total αβ-T cells (P < 0.01) and CD4⁺ naive (P = 0.03), CD4⁺ memory (P = 0.02), and CD8⁺ naive (P = 0.02) T-cell subpopulations of morbidly obese subjects (Fig. 3A). Moreover, a significant negative correlation (P < 0.01) was found between αβ-T cell sjTREC content and BMI (Fig. 3B, left).

A negative correlation between sjTREC content and age has been reported previously (25), but although the morbidly obese group was significantly older than the lean control group (morbidly obese aged 45 years [28–62] vs. lean aged 31 years [25–51]; P = 0.02), the sjTREC content in αβ-T cells did not correlate significantly with age (Fig. 3B, right), although we cannot exclude that this might be due to the relatively limited number of subjects studied. Moreover, in multiple regression analysis, the BMI was the only variable significantly associated with the TREC content (R = 0.8, R² = 0.58, P_total = 0.001, P_age = 0.28, P_BMI = 0.002), demonstrating that the decreased sjTREC content in morbid obesity is mainly determined by obesity and not by age.

Overall, the decreased sjTREC content together with the increased T-cell numbers in morbidly obese subjects is indicative of increased proliferation within the T-cell compartment of these subjects.

Several studies describe a reduced diversity within the TCRB repertoire of T cells isolated from adipose tissue of obese mice, suggesting a local antigen-driven immune response toward the main antigens present within adipose tissue (13,27). We determined TCRB diversity in peripheral blood T-cell subpopulations. We observed a diverse TCRB repertoire in CD4⁺ and CD8⁺ naive and memory T cells from both morbidly obese and lean subjects (Fig. 4A and B). Minor alterations in the (normally Gaussian distributed) TCRB repertoire were found to a limited extent in the CD4⁺ memory T cells and CD8⁺ naive and memory T cells of morbidly obese subjects, suggesting the existence of a slightly skewed TCRB repertoire in morbidly obese subjects (Fig. 4A and B). In addition, we obtained fat tissue that was removed during surgery from five other morbidly obese subjects. No peripheral blood T cells were available from these patients because only adipose tissue and plasma samples were stored. Nevertheless, it gave us an opportunity to investigate the TCR repertoire of adipose tissue T cells. Hence, we performed GeneScan analysis on the T cells present in the adipose tissue. We observed a polyclonal TCRB repertoire (Supplementary Fig. 2), indicating that there was no strong skewing toward particular T-cell clones. Instead a rather broad TCR repertoire was present in the adipose tissue T cells.

Because obesity is characterized by abnormal production of proinflammatory cytokines (28), we hypothesized that cytokines involved in T-cell proliferation, survival, and recruitment might also be produced in excess in morbidly obese subjects. Therefore, we determined a broad panel of cytokines in plasma from morbidly obese and lean subjects.

Plasma levels of the proinflammatory cytokines IL-6 and TNF-α did not differ between morbidly obese and lean subjects (Fig. 5A). Also, plasma levels of IFN-γ, IL-4, and IL-17A, cytokines respectively associated with Th1, Th2, or Th17 subpopulations, were similar in morbidly obese and lean subjects (Fig. 5B).

The cytokines CCL5, IL-2, and IL-7 enhance T-cell proliferation, survival, and recruitment (29–31). Plasma levels of CCL5 (P < 0.01) and IL-7 (P < 0.01) were significantly elevated in morbidly obese subjects (Fig. 5C) and correlated positively with BMI (Supplementary Table 3). IL-2 plasma levels were similar in morbidly obese and lean subjects (Fig. 5C).

As expected, IL-7 and CCL5 plasma levels positively correlated with total CD4⁺ T-cell numbers but not with total CD8⁺ T-cell numbers (Fig. 5D and E). In the CD4⁺ T-cell compartment, a positive correlation was found between IL-7 plasma levels and the number of naive CD4⁺ T cells, terminally differentiated CD4⁺ T cells, central memory CD4⁺ T cells, and natural CD4⁺CD25⁺FoxP3⁺ Treg cells; CCL5 plasma levels correlated positively with the number of terminally differentiated CD4⁺ T cells, central memory CD4⁺ T cells, and natural CD4⁺CD25⁺FoxP3⁺ Treg cells (Supplementary Table 4).

Although the morbidly obese subjects did not have type 2 diabetes mellitus, we investigated the correlations between the increased CD4⁺ T-cell numbers and metabolic measures. Fasting glucose and insulin levels were determined in the morbidly obese group only (32). A significant correlation was found between fasting insulin levels and CD4⁺ T-cell numbers (Fig. 6A). Moreover, the glucose-to-insulin ratio was calculated as a measure of insulin sensitivity. This ratio also correlated with CD4⁺ T-cell numbers (Fig. 6B). Fasting insulin levels and insulin sensitivity did not correlate with CD8⁺ T-cell numbers (Fig. 6C and D). No significant correlations were found between fasting blood glucose levels and T-cell subpopulations (data not shown).

This study is the first to comprehensively investigate the peripheral blood T-cell compartment of morbidly obese subjects. Our main finding was a selective increase in CD4⁺ T-cell numbers within the peripheral blood T-cell compartment of morbidly obese subjects. Peripheral blood CD8⁺ T-cell numbers were normal in morbidly obese subjects. This latter observation is in contrast with the increased numbers of local effector and memory CD8⁺ T cells described in adipose tissue of obese subjects (12,13).

In mice, diet-induced obesity results in reduced sjTREC content in splenic CD4⁺ T cells (33). This is accompanied by a reduction in naive T cells and a more restricted TCRB repertoire, suggesting that in this mouse model, the decrease in sjTREC content is mainly the result of reduced thymic output (33). Because ageing also is associated with a reduction in thymic output, resulting in a reduced TREC content and a reduction in naive T-cell numbers, it was suggested that obesity is related to accelerated ageing of the T-cell compartment (33,34).

In our study, we found a decreased sjTREC content in peripheral blood T-cell subpopulations of morbidly obese subjects. However, in contrast to observations in ageing studies (35,36), the decreased sjTREC content was accompanied by increased numbers of naive as well as memory CD4⁺ T cells and only an insignificant skewing of the TCRB repertoire. Therefore, despite the limitation of the significant age difference between morbidly obese and lean subjects in our study, we conclude from the increased naive T-cell numbers that the decrease in sjTREC content in morbidly obese subjects predominantly results from increased proliferation rather than accelerated ageing and decreased thymic output. This notion is further supported by the decreased telomere length observed in leukocytes of obese subjects (37).

The increased proliferation within the peripheral blood T-cell compartment is more likely of homeostatic nature rather than driven by dominant antigens because the latter would result in increased memory and effector T-cell subpopulations with prominent skewing of the TCRB repertoire, whereas the naive T-cell compartment would remain unaffected. In our cohort, we do not see such changes in the peripheral blood T-cell compartment. Moreover, in an additional analysis, a rather polyclonal TCR repertoire was observed in adipose tissue T cells of morbidly obese subjects. It therefore seems likely that there is no vast change in TCR repertoire in the adipose tissue T cells in our cohort. However, we formally cannot exclude the possibility of some skewing of the TCRB repertoire within adipose tissue T cells.

Several T-cell mitogenic factors, such as adipokines, fatty acids, or bacterial products, can be elevated in plasma of morbidly obese subjects (38–40). Also, increased levels of IL-7 and CCL5, cytokines capable of stimulating homeostatic T-cell proliferation, survival, and recruitment (29–31), have been found in adipose tissue of obese mice and men (41–43). We also found highly elevated plasma levels of IL-7 and CCL5 in morbidly obese subjects in this study, which positively correlated with peripheral blood CD4⁺ T-cell numbers. On the basis of these data, we hypothesize that IL-7 and CCL5, as well as the other T-cell mitogenic factors, might contribute to the increased homeostatic CD4⁺ T-cell proliferation in morbidly obese subjects.

Despite the selective increase in CD4⁺ T-cell numbers in peripheral blood, CD8⁺ T cells also displayed decreased sjTREC content, indicating that CD8⁺ T cells also undergo increased homeostatic proliferation due to the increased IL-7 and CCL5 cytokine levels in morbidly obese subjects. However, peripheral blood CD8⁺ T-cell numbers were not increased, suggesting a selective redistribution of CD8⁺ T cells into adipose tissue, which is in line with the described preferential accumulation of CD8⁺ T cells in obese adipose tissue (12,13). Also, CCL5 is a more potent chemoattractant for CD4⁺ T cells than for CD8⁺ T cells (44), and in obesity, systemic levels of CCL5 are ∼100-fold higher than those locally produced within adipose tissue (41). Therefore, the elevated CCL5 plasma levels that we observed may contribute to the selective retention of CD4⁺ T cells in peripheral blood of morbidly obese subjects.

With regard to the increased numbers of peripheral blood CD4⁺ T cells, we observed that this was accompanied by a selective increase in natural CD4⁺CD25⁺FoxP3⁺ Treg cell numbers. In addition, stimulation of PBMCs with PMA/ionomycin specifically induced a Th2 phenotype within the CD4⁺ T-cell compartment of morbidly obese subjects. This indicates that the numerically elevated peripheral blood CD4⁺ T-cell compartment of morbidly obese subjects is skewed toward a Treg- and Th2-dominated phenotype, suggesting a more anti-inflammatory set point. Despite this clear skewing, plasma levels of cytokines associated with the Th2 phenotype were mostly undetectable in plasma from both morbidly obese and lean subjects, as was the case for cytokines associated with the Th1 and Th17 phenotypes.

Natural CD4⁺CD25⁺FoxP3⁺ Treg cells and Th2 cells are capable of polarizing monocytes/macrophages toward an anti-inflammatory M2 phenotype, which is characterized by the production of anti-inflammatory mediators such as IL-1 receptor antagonist, IL-10, and transforming growth factor-β (45,46). We hypothesize that the preferential skewing of the CD4⁺ T-cell compartment toward a Treg- and Th2-dominated phenotype can be considered as a mechanism to counterregulate the proinflammatory activity that exists systemically and locally within the monocyte/macrophage compartment (47–50) in obesity. The absence of increased IL-6 and TNF-α plasma levels in our cohort of morbidly obese subjects supports this notion.

We demonstrated this anti-inflammatory T-cell set point in a morbidly obese cohort that was selected on the basis of being nondiabetic and, thus, relatively healthy and free of comorbidities (although we are not informed on the atherosclerotic state of our patients). To date, it is unknown whether changes in this set point away from the anti-inflammatory phenotype are associated with the development of obesity-related comorbidities.

Atherosclerosis, which frequently occurs during obesity, is characterized by accumulation of Th1 CD4⁺ T cells within the plaques (51), whereas CD4⁺ T-cell depletion reduces the development of atherosclerosis in mice (52). Also, the development of type 2 diabetes mellitus is delayed in mice with diet-induced obesity when T cells are depleted (13,15). In addition, we also demonstrated that CD4⁺ T-cell numbers positively correlated with fasting insulin levels.

On the basis of these literature data and our own data presented herein, it is thus tempting to speculate that changes away from the Treg- and Th2-dominated phenotype toward a more proinflammatory Th1- or Th17-dominated set point may prove an important indicator, or even mediator, for the development of atherosclerosis or diabetes in morbidly obese subjects. Longitudinal studies in morbidly obese subjects will be important to further address these issues.

In conclusion, the peripheral blood T-cell compartment of morbidly obese subjects is characterized by an increased homeostatic proliferation of both CD4⁺ and CD8⁺ T cells to which cytokines such as IL-7 and CCL5 probably contribute. This increased homeostatic proliferation is associated with an increase in peripheral blood CD4⁺ T-cell numbers, with a skewing toward a Treg- and Th2-dominated phenotype, suggesting an anti-inflammatory set point of the peripheral blood CD4⁺ T-cell compartment.

This work was supported by internal grants from the Departments of Internal Medicine and Immunology of the Erasmus Medical Center (MC). F.J.T.S. is supported in part by KiKa (Stichting Kinderen Kankervrij—Children Cancer-Free), Netherlands Organisation for Health Research and Development (ZonMW), and Association for International Cancer Research.

No potential conflicts of interest relevant to this article were reported.

K.v.d.W. researched data and wrote the manuscript. W.A.D. contributed to discussion and wrote, reviewed, and edited the manuscript. B.S. and D.H.S. researched data. A.W.L. researched data, contributed to discussion, and reviewed and edited the manuscript. H.A.D. contributed to discussion and wrote, reviewed, and edited the manuscript. R.M.K., M.O.v.A., and A.v.H. researched data. J.J.M.v.D. and A.-J.v.d.L. contributed to discussion and reviewed and edited the manuscript. F.J.T.S. and P.M.v.H. contributed to discussion and wrote, reviewed, and edited the manuscript. P.M.v.H. is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

The authors thank Sjanneke Heuvelmans, Erasmus MC, for collecting materials; Edwin de Haas and Benjamin Bartol, Erasmus MC, for assistance with cell sorting; Henk Wind, Jeroen te Marvelde, and Dennis Tielemans, Erasmus MC, for their assistance with flow cytometry; Joyce Vermeulen and Ingrid Wolvers, Erasmus MC, for performing GeneScan analyses; Sandra de Bruin–Versteeg, Erasmus MC, for assistance with the figures; Jon Laman, Erasmus MC, for participation in discussions; and all other members of the laboratories of W.A.D. and F.J.T.S. for their technical assistance.