Impaired HDL function in obese adolescents: Impact of lifestyle intervention and bariatric surgery

Authors


  • Disclosure: The authors declared no conflict of interest.

  • Funding agencies: Proteome studies in the Department of Functional Genomics (N.J., U.V.) were supported by the BMBF within the GANI Med program. The study was supported by the Federal Ministry of Education and Research (BMBF) (project funding reference number: 01GI1120A) and is integrated in the Competence Network Obesity (CNO).

Abstract

Objective

HDL regulates endothelial function via stimulation of nitric oxide production. It is documented that endothelial function is impaired in obese adolescents, and improved by lifestyle interventions (LI).

Design and Methods

HDL function in obese adolescents and the impact of LI or Roux-en-Y gastric bypass surgery (RYGB) was assessed. HDL was isolated from 14 adolescents with normal body mass index (HDLcontrol), 10 obese (HDLobese) before and after 6 month LI, and five severe obese adolescents before and one year after RYGB. HDL-mediated phosphorylation of endothelial nitric oxide synthase (eNOS)-Ser1177, eNOS-Thr495, and PKC-ßII was evaluated. In addition the HDL proteome was analyzed.

Results

HDLobese-mediated eNOS-Ser1177 phosphorylation was reduced, whereas eNOS-Thr495 phosphorylation increased significantly when compared to HDLcontrol. No impact of obesity was observed on PKC-ßII phosphorylation. LI and RYGB had no impact on HDL-mediated phosphorylation of eNOS and PKC-ßII. A principle component plot analysis of the HDL particle separated controls and severe obese, whereas the interventions did not trigger sufficient differences to the HDL proteome to permit distinction.

Conclusion

These results demonstrated that HDL-function is impaired in obese adolescents, and that LI or RYGB did not correct this dysfunction. This might be an argument for developing earlier prevention strategies in obese adolescents to avoid HDL dysfunction.

Introduction

The prevalence of pediatric obesity, defined as a body mass index (BMI) ≥95th percentile corrected for age and sex is rising at an alarming rate over the past three decades. Currently, 13% to 33% adolescents are obese worldwide [1]. Obesity is closely related with multiple diseases such as type two diabetes mellitus, hypertension, dyslipidemia, sleep disordered breathing, and fatty liver [2]. Furthermore, childhood obesity is a predictor of an increased risk of cardiovascular disease (CVD) and its mortality [3-5]. Endothelial dysfunction is an early sign of atherosclerosis and the indicator of a worsened prognosis [6]. An abnormality in endothelium-dependent vasodilatation is one of the most important changes in the early subclinical stage of atherosclerotic disease [reviewed in [7]]. An important factor responsible for endothelium-dependent vasodilatation is the bioavailability of nitric oxide (NO).

High-density lipoprotein (HDL) levels above 40-60 mg/dl have been proposed as strong independent predictor of lower CVD risk [reviewed in [8]]. Besides promotion of reverse cholesterol transport, HDL has been found to exert important anti-atherogenic effects by stimulation of endothelial cell NO production and endothelial repair as well as anti-inflammatory and anti-oxidant effects [9-13]. In recent studies it became evident that the functional properties of HDL with respect to stimulation of NO production are significantly impaired in patients with diabetes [14], CVD [15], and primary antiphospholipid syndrome [16]. Mechanistically it is proposed that endothelial nitric oxide synthase (eNOS)-dependent NO production is activated by phosphorylation of eNOS at position Ser1177. Besides, this pathway is deactivated by phosphorylation of eNOS at position Thr495 and activation of PKC-ßII [17]. Based on these findings on the functional importance of HDL in influencing endothelial function, the therapeutic approach targeting HDL is beginning to shift towards improving HDL function rather than just increasing its concentration [reviewed in [17]].

To critically assess the relation between pediatric obesity and HDL function, and the impact of different treatment modalities, we isolated HDL from normal body weight (BW) and obese adolescents before and after intervention [lifestyle intervention (LI) or Roux-en-Y gastric bypass (RYGB) surgery], and determined its ability to stimulate eNOS activation. In addition a proteome approach was used to analyze the composition of the HDL particles in the different groups.

Methods

Patient population and blood sampling

A total of 29 adolescents (age of 12-19) were included into this study. A healthy (no visible signs of any illness and no history of chronic disease) control group (n = 14) with normal BW(, BMI <89th percentile) for age and gender, was randomly selected as control group from our previously described cohort (University of Leipzig) [18].

Fifteen severely obese adolescents (BMI ≥99th percentile) were either treated by a LI program (n = 10, Obesity Center INSULA, Germany) or by RYGB treatment (n = 5 Cincinnati Children's Hospital Medical Center, Cincinnati, OH, USA) (Figure 1). Measurements of the intervention groups were performed at baseline and after 6 month (LI) or one year (RYGB group) follow-up. Written informed consent was obtained from all subjects and/or from a legal representative before initiation of any study-related activities.

Figure 1.

Flow chart of the study protocol. LI, lifestyle intervention; RYGB, Roux-en-Y gastric bypass.

Lifestyle intervention program

Ten severely obese adolescents enrolled in a weight reduction program conducted by the Obesity Rehabilitation Center INSULA (Berchtesgaden, Germany). The INSULA center is a specialized clinic to treat obese children and adolescents with a focus on solution-orientated psychotherapy with behavior therapy, exercise training (ET), dietetic treatment as well as an extensive clinical diagnosis and therapy of obesity associated diseases. Special goals are long-term modification of physical activity, leisure time, and dietary behavior. In addition an adventure educational approach was used during which patients engaged in group activities such as water rafting or other physical activities in order to increase self-confidence. Usually patients stayed for 6 months in the INSULA center. During the stay at the clinic approximately 1700 kcal were delivered daily. Besides background information about healthy foods conveyed to the patients in educational sessions, patients participated in practical courses in the in-house teaching kitchen. The ET consists of regular strength and endurance training (at least four times per week) for the whole therapy duration. Discussion with parents was held with staffs of all therapeutic areas to prevent relapse into old behavior pattern.

Laparoscopic RYGB

Five adolescents underwent a laparoscopic RYGB for severe obesity at the Division of Pediatric General/Thoracic Surgery—Weight loss program for Teens, Cincinnati Children's Hospital Medical Center (Cincinnati, OH, USA). All patients met established criteria for candidacy for bariatric surgery [19]. These five subjects were enrolled in the National Institutes of Health sponsored “Adolescent Gastric Bypass and Diabetic Precursors” study (NIH R03DK068228; protocol # 05-05-14 approved by CCHMC Institutional Review Board [IRB]).

Anthropometry and laboratory testing

Anthropometric assessments and laboratory testing procedures are described in detail in the supplementary material.

Isolation of HDL

HDL was isolated from serum obtained from normal BW controls (HDLcontrol), obese adolescents before and after LI (HDLLI-pre, HDLLI-post) and obese adolescents before and after the RYGB (HDLRYGB-pre, HDLRYGB-post) by sequential density ultracentrifugation (d = 1.006-1.21 g/ml) as recently described in detail [14, 15]. Quality of isolated HDL was evaluated by polyacrylamide gel electrophoresis followed by coomassie brilliant blue staining.

Cell culture and incubation with isolated HDL

Human aortic endothelial cells (HAEC, Cell Systems Biotechnology, Troisdorf, Germany) were cultured in EGM-2 cell culture medium (Lonza, Walkersville MD, USA) until 80-90% confluence. HAECs were incubated for 5, 10, 15, 30, or 60 min with 50 µg/ml isolated HDL. Thereafter, HAECs were harvested with ice cold lysis buffer (50 mmol/l Tris-HCl pH 7.4, 1% NP-40, 0.25% Na-deoxycholate, 150 mmol/l NaCl, 1 mmol/l EDTA, 0.1% Triton X-100, 0.2% SDS) containing protease inhibitor mix M (Serva, Heidelberg, Germany) as well as phosphatase inhibitors (Phosphatase inhibitor mix II, Serva, Heidelberg, Germany). Protein concentration was determined using BSA as standard (BCA method, Pierce, Rockford IL, USA).

Western blot analysis

Ten microgram of total protein was separated on a denaturing polyacrylamide gel and transferred to a polyvinylidene difluoride (PVDF) membrane. To detect specific proteins the following antibodies were applied: antitotal eNOS (Santa Cruz, Heidelberg, Germany, 1:200 dilution, 2 h at room temperature), anti-phospho-eNOS-Ser1177, anti-phospho-eNOS-Thr495 (both BD Bioscience, Heidelberg, Germany; 1:2000 dilution, 4°C over night), anti-apolipoprotein A1 (Abcam, Cambridge, UK, 1:10,000 dilution, 1 h at room temperature), antitotal PKC-ßII, anti-phospho PKC-ßII-Ser660, anti-angiotensinogen (all Santa Cruz, Heidelberg, Germany, 1:200 dilution, 4°C over night). After incubation with a horseradish peroxidase-conjugated secondary antibody (Santa Cruz, Heidelberg, Germany, 1:200 dilution, 1 h at room temperature), specific bands were visualized by enzymatic chemiluminescence (Super Signal West Pico, Pierce, Bonn, Germany). Band intensity was quantified by densitometry using a 1D scan software package (Scanalytics, Rockville, USA). All samples were analyzed in triplicate. For the evaluation of HDL-induced phosphorylation of eNOS the maximal stimulation of eNOS irrespective of the time point of recording was used.

Proteome analysis of isolated HDL particles

The proteome of HDL particles was investigated by shotgun LC-MS/MS (LTQ Orbitrap Velos mass spectrometry) in five randomly selected adolescents from each group—control, severe obese (RYGB group), LI pre and post, RYGB pre and post. For detailed description see supplementary material.

Statistical analysis

SPSS version 16.0 (SPSS, Chicago, Ill) was used for all statistical analyses. Data are expressed as mean ± SEM. Comparisons among groups were tested with ANOVA. When data were not normally distributed or the variance was not equal, the Kruskal-Wallis nonparametric test was used. A value of P < 0.05 was considered statistically significant. All measurements were made by investigators blinded for the treatment group.

Results

Patient characteristics and follow-up after interventions

The baseline values for all individuals included into the study are depicted in Table 1. The adolescents in the LI program stayed for an average of 6 month (range 0.48 month to 7.2 month) in the INSULA center. As expected the control group significantly differed from obese with respect to BW, BMI, and blood pressure. A lipid profile measurement revealed no significant difference among the three groups. LI led to a decrease of BW, BMI, diastolic blood pressure, fasting plasma glucose, and HOMA-IR index (Table 1). The RYGB group showed a significant decrease in BW, BMI, and systolic blood pressure. In addition a decrease of the HOMA-IR index and a significant increase of the HDL concentration was evident (Table 1).

Table 1. Anthropometric and metabolic characteristics of the study cohort at begin and after the lifestyle and surgical intervention
InterventionLean controlLifestyle interventionRYGB
PrePostPrePost
  1. Data are presented as means and range.

  2. a

    P < 0.05 vs. lean control

  3. b

    P < 0.05 vs. pre

  4. BW, body weight; BMI, body mass index; BMI-SDS, Body-Mass-Index- Standard-Deviation-Score; BP, blood pressure; FPG, fasting plasma glucose; FPI, fasting plasma insulin; HDL, high density lipoprotein; LDL, low density lipoprotein; TG, triglyceride;

No.14105
Sex (female/male)11/35/52/3
Age (years)14.313.213.616.517.5
 (12.6-15.9)(12.0-14.9)(12.4-15.4)(14.4-18.5)(15.4-19.4)
BW (kg)53,592.7a79.8a, b180.3a105.6a, b
 (40.0-68.4)(71.3-127.4)(62.5-110.9)(127.7-233)(70.2-141)
Height (cm)158.5162.8164.4174.7175.3
 (140-180)(157-169)(159-169)(163.7-182.2)(163.3-184.5)
BMI (kg/m2)19.134.9a29.4a, b59,2a34,8a, b
 (17.5-20.8)(28.5-43.4)(24.1-39.8)(48,1-70,2)(26,0-42,2)
BMI-SDS0.162,89a2.25a, b4.2a2.6a, b
 (−0.31-0.55)(2.30-3.76)(1.52-3.34)(3.4-5.0)(1.7-3.5)
BMI percentile (%)56.399.697.4>99>99
 (38.0-70.8)(99.0-100.0)(93.0-100)  
Systolic BP (mmHg)102120a115a131a122a, b
 (80-130)(105-143)(103-126)(124-138)(117-126)
Diastolic BP (mmHg)6178a67b6876a
 (40-80)(73-85)(56-80)(62-74)(69-83)
Glucose metabolism
FPG (mmol/l)4.24.34.2b4.84.4
 (3.6-4.9)(3.9-5.0)(3.7-4.6)(4.4-5.2)(4.3-4.6)
FPI (mU/l)17.626.7a15.823.712.3
 (12.1-31.6)(15.7-36.0)(10.6-22.8)(19.1-29.3)(10.8-14.1)
HOMA-IR3.35.0a2.9b5.0a2.4b
 (2.0-5.8)(3.3-6.9)(2.0-4.1)(4.1-6.1)(2.1-2.8)
Lipid metabolism
HDL (mmol/l)1.31.31.21.11.4b
 (0.9-1.9)(0.8-1.9)(0.8-1.5)(1.0-1.3)(1.2-1.5)
LDL (mmol/l)2.52.32.02.51.9
 (1.3-3.4)(1.1-3.5)(0.6-3.1)(2.1-2.9)(1.5-2.3)
TG (mmol/l)1.211.081.121.340.86
 (0.47-2.23)(0.16-2.07)(0.46-1.99)(1.08-1.59)(0.66-1.05)

HDL-mediated eNOS phosphorylation and PKC-ßII phosphorylation in obese adolescents

Incubating HAEC with HDLcontrol the phosphorylation of eNOS at position Ser1177 was increased 2.92 ± 0.45 fold vs. unstimulated cells. This stimulation of eNOS-Ser1177 phosphorylation was significantly lower with HDLobese (1.91 ± 0.23 fold vs. unstimulated cells; P = 0.046 vs. HDLcontrol) (Figure 2A). Analyzing the eNOS phosphorylation at the inhibitory position Thr495, HDLobese stimulated the phosphorylation at this position to a significantly greater extent than HDLcontrol (obese: 2.47 ± 0.76 fold vs. unstimulated cells; controls: 1.37 ± 0.09 fold vs. unstimulated cells; P = 0.046). (Figure 2B).

Figure 2.

Human aortic endothelial cells were incubated with HDL isolated from controls (n = 14) and obese adolescents (n = 15) and the phosphorylation of eNOS at position Ser1177 (A), and Thr495 (b) as well as the phosphorylation of PKC-ßII (C) was quantified. For quantitative analysis the x-fold increase in HDL-induced phosphorylation (maximal stimulation irrespective of time point) vs. untreated cells was determined. Values are expressed as mean ± SEM. Representative western blots are shown on the right side. The incubation time in minutes (min) of the cells with the isolated HDL is depicted on top of the blots.

Incubation of HAEC with HDLcontrol resulted in a 2.26 ± 0.28 fold increase in PKC-ßII phosphorylation at position Ser660 when compared to unstimulated cells. This ability to stimulate PKC-ßII phosphorylation was not significantly different with HDLobese (obese: 2.36 ± 0.32 fold vs. unstimulated cells; P = NS vs. control) (Figure 2C).

HDL-mediated eNOS phosphorylation and PKC-ßII phosphorylation—impact of LI and RYGB

Comparing the functional properties of HDL before and after LI a trend towards an improvement in phosphorylation of eNOS at position Ser1177 was evident (pre: 2.09 ± 0.38 fold vs. unstimulated cells; post: 2.70 ± 0.44 fold vs. unstimulated cells; P = 0.08) (Figure 3A). No impact of the LI was observed with respect to the HDL-mediated phosphorylation of eNOS at position Thr495 (pre: 1.52 ± 0.13fold vs. unstimulated cells; post: 1.53 ± 0.12 fold vs. unstimulated cells; P = NS) (Figure 3B) and the phosphorylation of PKC-ßII (pre: 1.76 ± 0.32 vs. post: 1.35 ± 0.16 fold vs. unstimulated cells; P = NS) (Figure 3C).

Figure 3.

Human aortic endothelial cells were incubated with HDL isolated from adolescents before (pre) and after (post) the lifestyle intervention (LI) as well as before and after RYGB surgery. Phosphorylation of eNOS at position Ser1177 (A,D) and Thr495 (B,E) as well as phosphorylation of PKC-ßII (C,F) was evaluated. For quantitative analysis the x-fold increase in HDL-induced phosphorylation vs. untreated cells was determined. Values are expressed as mean ± SEM.

In the RYGB group, the functional properties of HDL did not significantly improve after the surgery with respect to the phosphorylation of eNOS at position Ser1177 (pre: 2.81 ± 0.46 fold vs. unstimulated cells; post: 3.93 ± 1.27 fold vs. unstimulated cells; P = NS), Thr495 (pre: 4.62 ± 2.08 fold vs. unstimulated cells; post: 3.64 ± 2.39 fold vs. unstimulated cells; P = NS) and PKC-ßII phosphorylation (pre: 3.55 ± 0.38 fold vs. unstimulated cells; post: 2.99 ± 0.42 fold vs. unstimulated cells; P = NS).

Proteome analysis of isolated HDL particles

Proteome composition of isolated HDL was analyzed by LC-MS/MS analysis. In total, 123 proteins from 452 unique peptides were identified using the Mascot algorithm; 98 proteins with at least one peptide per protein and 34 proteins with at least two peptides per protein were identified. Label-free quantification of 34 proteins was applied using peptide intensities as indicator for protein abundance and variability. LC-MS/MS analysis reproducibly identified HDL-associated proteins with known functions as described elsewhere (Supporting Information Table 1) [20, 21]. In addition, we found protein groups not previously known to reside in HDL particle, such as Indian hedgehog protein (IHH) and fibroblast growth factor-binding protein 2 (FGFP2) A principal component analysis plot separated controls and severe obese adolescents (adolescents from the RYGB group) (Figure 4A), whereas the interventions did not trigger sufficient differences to the HDL proteome to permit distinction of proteome patterns before (baseline) and after LI or RYGB (Figure 4B). Proteome analysis revealed that complement component C9 (p-value <0.016) was significantly decreased in level, whereas angiotensinogen was present at significant higher levels in obese adolescents. Apolipoprotein F, angiotensinogen, clusterin, prenylcysteine oxidase 1, phospholipid transfer protein, and aminopeptidase N displayed statistically relevant increase in level following LI or RYGB (Supporting Information Table 1). As expected, weight loss following RYGB showed a stronger influence on protein levels within HDL particles (apolipoprotein F, clusterin, prenylcysteine oxidase 1, phospholipid transfer protein, Aminopeptidase N) than LI (angiotensinogen, apolipoprotein A-II; Supporting Information Table 1).

Figure 4.

A principal component analysis plot to separate lean controls (black circle) and severe obese adolescents (grey circle) (A) as well as pre and postlifestyle intervention (black (pre) and grey (post) squares) and RYGB surgery (black (pre) and grey (post) circles) (B).

Angiotensinogen protein bound to HDL isolated from obese adolescents

Based on the results obtained from the proteome analysis where angiotensinogen was found to be significantly incresaed in HDLobese (P = 0.009, Supporting Information table 1), we used western blot analysis to confirm this result. As shown in Figure 5, a significant greater expression of angiotensinogen by 193% was evident in the obese group, when compared to lean controls (lean control: 0.082 ± 0.007 vs. obese: 0.24 ± 0.05 arb. Units; P < 0.05).

Figure 5.

Quantification of angiotensinogen attached to HDL isolated either from lean controls (control) or obese adolescents (obese) Values (angiotensionogen / apolipoprotein A1) are expressed as mean ± SEM. Representative western blots are depicted.

Discussion

In recent years it became more evident that not only the quantity but also the functional capacity of HDL is important for influencing the risk of CVD, and therefore, strategies have been developed to increase HDL quantity and function [17]. With respect to functional properties of HDL in obese adolescents and the impact of a LI or RYGB, several findings emerge from the present study.

First, the functional capacity of HDL to phosphorylate eNOS at the activating position Ser1177 in endothelial cells (ECs) is impaired, whereas phosphorylation of eNOS at the inhibitory position Thr495 in ECs is increased in obese adolescents. Second, LI and RYGB in obese adolescents did not restore the functional capacity of HDL following treatment. Third, HDL particles isolated from obese adolescents can be distinguished from those of controls by proteome analysis, whereas LI and RYGB do not alter the composition of the HDL particle tremendously. Together, these findings suggest that obesity has a negative effect on HDL function with respect to eNOS activation, which cannot be improved at least in severely obese adolescents by lifestyle modification or RYGB within 6 month to one year. Based on the results one might envision an “obesity threshold” above which HDL function is impaired and an intervention strategy has to get below this threshold to achieve improvement on HDL function. The existence of such a hypothetical threshold might be an argument to consider an earlier resolute prevention/intervention for pediatric obesity with respect to CVD prevention.

Obesity and the ability of HDL to phosphorylate eNOS

Numerous experimental and clinical studies have suggested that the endothelial production of NO is critical for the regulation of vascular tone and structure [22]. Therefore, eNOS-derived NO is a crucial determinant of vascular homeostasis, and reduced bioavailability of NO plays an important role in the development of endothelial dysfunction, a primary sign of the early stages of atherosclerosis [23]. In recent years it became more evident that obese adolescents exhibit endothelial dysfunction, similar to patients with CVD [24, 25]. Different mechanisms linking obesity and endothelial dysfunction like low-grade inflammation and increased oxidative stress [reviewed in [26]] or increased adipokines [27] have been postulated. In the present study we focused for the first time on HDL functional capacity to modulate NO production via eNOS phosphorylation and documented that the capacity of HDL isolated from obese adolescents to activate eNOS is significantly impaired (increased phosphorylation at position Thr495 and reduced phosphorylation at position Ser1177). This result is in good agreement with the findings in other atherosclerotic disorders like primary antiphospholipid syndrome [16], diabetes mellitus [14], and CVD [15] in adults. In summary, despite no significant change in HDL concentration (probably because of the small sample size), a significant impact of obesity on HDL-mediated activation of eNOS via phosphorylation of the enzyme could be noted.

Impact of LI and RYGB on HDL function in obese adolescents

LI including ET is an accepted strategy in pediatric obesity [28]. However, the success rate of LI on weight loss is insufficient. Only a limited number of individuals succeeded in maintaining substantial weight loss over time [29]. Pediatric obesity specialists agree that adolescents with comorbidities of obesity, who have been unsuccessful in attaining a healthy BW by those conservative strategies, should be considered for bariatric surgery [19, 30], the most definitive intervention of weight loss treatment by far [31]. With respect to an improvement of endothelial function after a LI or RYGB only limited data are available. In a study by Roberts and colleagues [32], a LI program consisting of 2-2.5 hours ET per day plus high-fiber low-fat diet for two weeks, a significant improvement of endothelial function because of amelioration of several risk factors was noted. The impact of RYGB on endothelial function is only investigated in adults [33], showing an improvement in flow mediated dilatation. Using cultured human ECs and HDLobese isolated before and after each intervention (LI or RYGB), we were not able to detect a significant improvement in HDL function, measured as the capacity to phosphorylate eNOS at Ser1177 and Thr495. As explanations for this negative effect on HDL function at least two possible reasons have to be considered. First, the training intensity performed by the obese adolescents during the LI was not high enough to achieve a beneficial effect on HDL function. This assumption is supported by the observation that the applied LI did not affect the HDL concentration either, which is influenced in many studies by ET [34, 35]. In addition Roberts and colleagues documented an improvement in HDL function with respect its anti-inflammatory function in obese men performing a LI including a stringent exercise program (daily treadmill walking for 45-60 min at 70-85% maximal heart rate) [36]. Second, LI as well as RYGB let to a significant weight loss over time, but even after one year the adolescents are still classified as obese as evident by the BMI percentile. Therefore, a possible working hypothesis may be that a certain threshold in BW has to be reached (below the 95Th BMI-percentile) before a significant effect on HDL function can be achieved. Nevertheless, this threshold hypothesis has to be verified by further experiments.

Impact of obesity and LI/RYGB on HDL composition

To determine the effect of obesity and subsequent intervention strategies on the composition of the HDL particle a shotgun proteome analysis was performed. The main outcome of such an analysis can be summarized as follows. First, the overall protein signature of HDLcontrol differs from HDLobese displayed by principal component analysis (Figure 4A). This difference may contribute to the difference in HDL functional capacity between control and obese adolescents. However, at this stage it is not clear which proteins are responsible for this effect. Second, both interventional strategies elicited different small changes in HDL protein composition, demonstrating that the LI and RYGB affect the HDL particle in different ways. In addition the intervention did not result in a change in the overall signature of HDL thereby supporting the negative effect on HDL functional capacity to modulate eNOS. Third, at least two formerly unknown proteins (Indian hedgehog protein, fibroblast growth factor-binding protein 2) were detected in HDL particles by the proteome analysis. It is known from the current literature that they may have an impact on endothelial function. For example, the Indian hedgehog protein activates pro-angiogenic responses in ECs [37] and activates hematopoiesis and vasculogenesis [38]. Only a few publications describe the pathophysiological function of fibroblast growth factor-binding protein 2 related to tumor angiogenesis [39]. Unfortunately both proteins were not altered in level in HDLobese vs. the control group and interventions did not render the level of the two proteins. Therefore the relevance of these two new HDL-associated proteins remains unclear at the moment.

Study limitations

Some limitations of the present study should be mentioned. First, only small numbers of adolescents were included into each group. Specially, the RYGB group consisted only of five individuals. Until now, the RYGB is still a relative rarity in adolescents, and therefore it is difficult to collect a larger group of samples. Second, the impact of altered eNOS phosphorylation on NO generation was not investigated in the present study. However, earlier studies clearly documented that the phosphorylation elicited by HDL on Ser1177 and Thr495 residues also regulates eNOS activity and subsequent NO production [14, 15]. Third, the results presented in the study are obtained in cell culture experiments using HDL isolated from different patient cohorts. Therefore it has to be answered if this modulation of HDL also has an impact on endothelial function in vivo. Preliminary and unpublished results from our laboratory documented that at least a correlation between HDL-mediated eNOS phosphorylation and endothelial function exists in adults with CVD. Therefore, the in vitro measured HDL functional capacity may be an important parameter for regulation of endothelial function.

Conclusion

In summary, obese adolescents exhibit reduced ability of HDL to regulate eNOS activity via phosphorylation. Of note not only a LI but also RYGB does not correct this HDL dysfunction. One possible explanation may be the threshold hypothesis, meaning that the weight loss has to reach a certain level, before an intervention is effective in influencing the altered HDL function. Despite its definitive and long lasting effectiveness, RYGB is a strategy reserved only for severe obese adolescents with secondary disease at the present time. This result might be an argument for developing better prevention strategies to avoid severe obesity in adolescents.

Acknowledgments

We would like to thank Dr. Andreas Schubert, Fraunhofer Institute, Leipzig, Germany for providing the access to an ultracentrifuge for HDL isolation. YM, AO, and VA conceived and carried out molecular analysis of HDL function, NC and UV conceived and carried out proteome analysis of isolated HDL particles. MW, AM, and WS performed the lifestyle intervention, TI and HT were responsible for the biatric surgery and UM and GS collected the specimens from the healthy adolescents. AL performed the confirmatory western blot analysis for angiotensionogen, NK and MK performed the statistical analysis. All authors were involved in writing the manuscript and had final approval of the submitted and published version.

Ancillary