Low-Frequency Intermittent Hypoxia Promotes Subcutaneous Adipogenic Differentiation

Obstructive sleep apnea (OSA), characterized by intermittent hypoxia (IH), is associated with obesity and metabolic disorders. The mass and function of adipose tissue are largely dependent on adipogenesis. The impact of low-frequency IH on adipogenesis is unknown. Sprague-Dawley rats were subjected to IH (4 min for 10% O2 and 2 min for 21% O2) or intermittent normoxia (IN) for 6 weeks. The degree of adipogenic differentiation was evaluated by adipogenic transcriptional factors, adipocyte-specific proteins, and oily droplet production in both subcutaneous adipose tissue (SAT) and visceral adipose tissue (VAT). Upregulation of proadipogenic markers (CEBPα, PPARγ, and FABP4) and downregulation of antiadipogenic markers CHOP in line with smaller size of adipocytes were found in IH-exposed SAT. In vitro experiments using human preadipocytes (HPAs) of subcutaneous lineage during differentiation phase, subjected to IH (1% O2 for 10 min and 21% O2 for 5 min; 5% CO2) or IN treatment, were done to investigate the insulin-like growth factor 1 receptor (IGF-1R)/Akt pathway in adipogenesis. IH promoted the accumulation of oily droplets and adipogenesis-associated markers. IGF-1R kinase inhibitor NVP-AEW541 attenuated the proadipogenic role in IH-exposed HPAs. In summary, relatively low frequency of IH may enhance adipogenesis preferentially in SAT.


Introduction
Intermittent hypoxia (IH), a characteristic feature of obstructive sleep apnea (OSA), is regarded as a pathogenic factor in OSA-related morbidities, including cardiovascular and metabolic disorders. Obesity, in particular visceral obesity, is a well-established factor contributing to the development of metabolic syndrome or various cardiometabolic diseases such as type 2 diabetes mellitus (DM) or hypertension [1][2][3]. Increasing evidence suggests that OSA may contribute towards obesity-induced metabolic disorders [4,5].
Adipose tissue not only serves as an organ to control energy balance by storing and mobilizing triglycerides but also actively exerts profound impact on glucose metabolism, immunologic response, inflammatory response, and angiogenesis [6]. Various fat depots in the body are known to have different metabolism, and their excessive accumulation would predispose to different detrimental effects [7]. Adipose tissue accumulation is caused by enlargement of existing adipocytes (hypertrophy), generation of new adipocytes (hyperplasia), or both. Adipogenesis (adipogenic differentiation) is the process resulting in adipocyte hyperplasia. Adipocytes have a finite storage capacity, and once existing cells have reached that limit, new adipocyte formation is required to prevent lipid deposition in the liver, muscle, or other inappropriate locations. Most studies support a reduction of adipogenesis in the obese state [8].
Intermittent hypoxia, in various experimental models, can trigger both detrimental and beneficial effects on multiple body systems [9]. The features of IH profile (nadir hypoxia level and hypoxia episodes per hour) determine the outcomes, which would also differ in various organs/tissues. While detrimental effects of severe IH have been consistently demonstrated, the impact of lesser degrees of IH is not well characterized. Experimental data has demonstrated that IH of relatively low cycle numbers (5-15 episodes/hour) and less profound hypoxia (nadir of 9-16% inspired O 2 ) may lead to beneficial rather than detrimental effects on the cardiovascular and metabolic systems [10][11][12][13]. The adverse effect of severe IH profile on adipose tissue and adipocyte dysfunction has also been investigated extensively [14][15][16], but relatively little is known about the impact of IH, especially of mild or modest degree, on adipogenesis.
We hypothesize that low frequency of IH may alter adipogenic differentiation in a depot-specific manner, through modulation of adipogenic transcriptional factors and/or adipogenic extracellular factors. Using the in vivo rat model of IH exposure, the effect of low-frequency IH on adipocyte differentiation status in VAT and SAT and the associated mechanisms were investigated. The extracellular signaling pathways and intracellular transcriptional factors for adipogenesis were further investigated in an in vitro IH-exposed subcutaneous adipocyte model.

In Vivo Intermittent Hypoxia-(IH-) Exposed Rat Model.
Twelve healthy male Sprague-Dawley (SD) rats (~200 g; sixweek old) were randomly divided into intermittent normoxia (IN) and IH groups. Rats were fed with standard chow diet. One rat from the IH group died during the course of study period. Treatments of IN and IH were simultaneously performed in 2 identical chambers (Oxycycler model A84; BioSpherix, Redfield, NY, USA) for 6 hours daily, during 9:00 am to 3:00 pm, for 6 weeks. The O 2 concentration in the chamber was continuously measured by an O 2 analyzer during the exposure (see Supplementary Figure S1). The profile of IH was designed as approximately 240 seconds (for 10% O 2 ) and 120 seconds (for 21% O 2 ). For IN, the period of hypoxic (10% O 2 ) gas supply was replaced by air (21% O 2 ) while keeping other chamber conditions the same. After 6-weeks IN or IH exposure, rats were sacrificed with overdose of sodium pentobarbital anesthesia (100 mg/kg, i.p.). Epididymal adipose tissue and inguinal adipose tissue were isolated to represent visceral (VAT) and subcutaneous (SAT) adipose tissues, respectively [17,18]. Isolated tissues were snap-frozen and stored at −70°C for the measurements of mRNA and protein expressions for adipogenic transcriptional factors and adipogenic extracellular factors. Arterial blood was obtained via cardiac puncture. Rat serum was prepared by centrifugation at 2200 ×g for 10 minutes for measurement of metabolic parameters such as triglyceride, total cholesterol, glucose, and free fatty acid levels. All animal procedures conformed to the guidelines from Directive 2010/ 63/EU of the European Parliament. The experiment was approved by the Committee on the Use of Live Animal in Teaching and Research (CULATR number 2371-11) of The University of Hong Kong.
2.2. Adipose Tissue Morphometry. Epididymal adipose tissue (VAT) and inguinal adipose tissue (SAT) were collected in 10% buffered formalin, fixed overnight, and embedded in paraffin. Hematoxylin and eosin staining was used for adipocyte morphometry. Images were taken using Nikon Eclipse Ni-U microscope ×20 objective (Nikon Instruments Inc., Melville, NY, USA).
To investigate the role of IGF-1R/Akt signaling in IH-regulated adipogenesis, HPAs and differentiated HPAs were subjected to IH exposure for 96 cycles with the stimulation of 100 μg/ml insulin (Sigma) in either growth medium (for HPAs) or maintenance medium (for differentiated HPAs). In addition, HPAs were incubated with a selective IGF-1R kinase inhibitor NVP-AEW541 [19] (0.1 μmol/l, Cayman Chemicals, MI, USA) for 30 min before undergoing differentiation for 12 consecutive days.

Real-Time Reverse Transcription Polymerase Chain
Reaction (RT-PCR) Analysis. For quantitative PCR (qPCR), total RNA was isolated from VAT, SAT, and HPAs using Trizol (Invitrogen, Life Technologies). Reverse transcription was performed using 1 μg of total RNA with Qiagen RT kit (Qiagen, Life Technologies) according to the manufacturer's instructions. Primers for FABP4, GLUT4, PPARγ, CEBPα, CEBPδ, and GAPDH were designed by Primer 3 (BioTools, University of Massachusetts Medical School, USA) and synthesized by IDT (Integrated DNA Technologies, Singapore). The sequences for primers are listed in Table 1. Amplification of target was carried out with Power SYBR Green PCR Mix (Applied Biosystems, Life Technologies Inc., California, USA) using StepOnePlus Real-Time PCR System (Applied Biosystems, Life Technologies Inc., California, USA). The cycle threshold value for amplification was normalized using glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as the reference gene, and the data were analyzed using the ΔCt method. Data represent the mean ± SEM of independent experiments. 2.6. Measurement of Adiponectin from Conditioned Media of Adipose Tissue. Approximately 400 mg fresh SAT and VAT was isolated and chopped into pieces and then incubated with 1 ml medium (DMEM/F12 plus 10% FBS) for 24 hours. After incubation, conditioned media were centrifuged at 500g for 5 min and stored at −70°C. The levels of adiponectin being released in conditioned media were analyzed using commercial ELISA kits (eBioscience, San Diego, CA, USA) according to the manufacturer's instructions.

Releases of Inflammation-Associated Mediators from
Cultured HPAs. Conditioned media were collected from HPAs at the end of each maintenance period of adipogenic differentiation. The releases of IL-6 and MCP-1 in conditioned media were analyzed using commercial ELISA kits, respectively (eBioscience, San Diego, CA, USA) according to the manufacturer's instructions.
2.9. Statistical Analysis. Data were expressed as mean ± SEM from at least 4 independent cell culture experiments or experimental animals. Data from the IH and IN groups were compared by Student's t-test.
The variance equality was analyzed for all comparisons. If variance equality is not met, the unpaired t-test with Welch's correction was used. All statistical analyses were performed via GraphPad Prism 5.0 (GraphPad Software Inc., San Diego, CA, USA). A P value (two-sided) <0.05 was considered statistically significant.

Discussion
This study provided evidence that daily IH exposure of relatively low frequency (10 events/hour) for 6 hours over 6 weeks accelerated adipogenesis of SAT but not of VAT in a lean rat model. In line with the in vivo findings, IH induced In recent years, the potential role of OSA in the pathogenesis of metabolic abnormalities has been extensively investigated both in clinical studies and in animal models [21,22]. As adipose tissue exerts important endocrine functions in metabolic regulation, the complex relationship between OSA, adipose tissue dysfunction, and other metabolic derangements is noted. Intermittent hypoxia, a hallmark pathophysiologic feature in OSA, has been reported to aggravate adipose tissue dysfunction and metabolic disorders [14,15], while data on the impact of IH on adipogenesis is limited.
In comparison to subcutaneous obesity, visceral obesity tends to be associated with OSA-associated insulin resistance, dyslipidemia, and glucose intolerance [23]. Increased deposition of fat in the visceral compartment may occur as an unfavorable outcome consequent upon saturation of SAT to store fat [24]. Unlike adiposity accumulation induced by over nutrition, our in vivo data showed that IH at this relatively low frequency upregulated adipogenesis in SAT but not in VAT, in agreement with their respective adipocyte morphometry. These findings implied that adipogenic progenitor cells from different depots responded differently to the same IH stimulus. In line with the in vivo upregulation of FABP4 and GLUT4 expression in SAT, IH also stimulated upregulation of FABP4 and GLUT4 in HPAs in vitro, supporting the regulation of adipogenesis by IH in subcutaneous fat depot.
The two fat depots, VAT and SAT, have different cellular characteristics and origins, [25,26], which may contribute to the differential effect of IH. Compared to VAT, SAT contains higher capacity for preadipocyte proliferation and differentiation [27][28][29], implying that adipogenic ability may be a dominant response to stress stimuli in SAT. In this study, the morphological changes induced by IH were consistent with enhanced adipogenesis with smaller adipocytes in SAT but not in VAT. Furthermore, cellular response elicited by IH would likely be determined by the severity of the stimulus. Our in vivo model exposed mice to IH 10 cycles/hour, while a previous study which showed visceral adipose tissue dysmetabolism used an IH regimen of 20 cycles/hour [30]. It has been reported that low abdominal subcutaneous preadipocyte adipogenesis is associated with visceral obesity and a dysmetabolic state [31]. Hence, it is tempting to speculate that subcutaneous adipogenesis seen in the current study indicates a mechanism by which the subcutaneous fat depot is acting as a metabolic buffer in the face of IH challenge.
There is a recent hypothesis that mild OSA may have a "protective" effect on metabolic hemostasis, probably due to preconditioning to hypoxia [32][33][34]. Subcutaneous fat is the largest adipose depot in the human body for storage of excess lipids, and if SAT undergoes hyperplasia with generation of smaller and more functional adipocytes, this added functional capacity may attenuate or reduce the risk of metabolic diseases [35]. Recent animal studies proposed a therapeutic role of transplantation of subcutaneous fat in alleviating type 2 diabetes mellitus, obesity, and insulin resistance [36][37][38]. In addition, the removal of visceral fat and replacement with subcutaneous fat could alleviate or prevent metabolic dysregulation [39]. In obesity, subcutaneous adipocyte hyperplasia precedes adipocyte hypotrophy [40]. In this IH-exposed rat model, there was a significant elevation of serum lipid markers such as triglyceride (TG) and free fatty acid (FFA) (see Supplementary Figure S3), similar to a previous study [41]. The enhanced adipogenic capacity of subcutaneous preadipocytes after IH exposure may be a defensive response to produce more mature adipocytes for the safe storage of lipids, reducing the chance of ectopic deposition of lipids which triggers dysmetabolism. Notwithstanding the enhanced adipogenesis seen, serum triglyceride remained elevated, suggesting still inadequate capacity. Differentiation ability of adipocytes is regulated by various transcription factors [42]. The expression of PPARγ is required to induce 3T3L1 differentiation [43], and we demonstrated upregulation of PPARγ during adipogenic differentiation with or without IH exposure. C/EBPα induces many adipocyte genes directly, and in vivo studies have indicated an important role for this factor in the development of adipose tissue [44]. In our in vivo model, upregulation of C/EBPα was demonstrated in SAT but not in VAT. In addition, the SAT of IH-exposed rats showed a significant reduction of CHOP, an inducible inhibitor of adipocyte differentiation [20], in comparison with that of IN-exposed rat, providing another possible mechanism responsible for IHinduced acceleration of subcutaneous adipogenesis. Besides PPARγ and CEBPα, CEBPδ is crucial in the early stage of adipogenesis as it could activate mitotic clonal expansion (MCE) which is an indispensable factor for the induction of preadipocytes into differentiated programming [45]. In agreement with previous studies, a remarkable reduction of CEBPδ was observed in the control (IN) differentiable HPAs after initiation of adipogenesis. IH exposure significantly mitigated the reduction of CEBPδ, suggesting that IH potentiated more preadipocytes into differentiation and further extended the initial stage of adipogenesis.
Apart from various transcriptional factors, IGF-1R/Akt signaling also functions for the transduction of differential signals. IGF-1 and insulin are essential factors for the initiation of adipocyte differentiation [46,47]. The effect of insulin on adipogenic differentiation is due to its binding with IGF-1 receptors (IGF-1Rs), which are more abundant in preadipocytes compared to mature adipocytes [48,49]. Our in vitro results indicated that IH significantly upregulated the expression of IGF-1Rs along with increased phosphorylation of Akt in HPAs (preadipocytes), suggesting the involvement of the IGF-1R/Akt pathway in IH-driven adipogenesis. In support, the selective inhibitor of IGF-1R significantly suppressed IH-facilitated HPA differentiation. However, it is unexpected that a similar effect did not exist in IN-exposed HPAs at the same concentration of IGF-1R inhibitor, which may be attributed to robustly compensatory adipogenic pathways in normal preadipocytes. Besides, the expressions of IGF-1R pathway were significantly suppressed in both IH-exposed differentiated HPAs (mature adipocytes) and adipose tissue (SAT and VAT). The possible reason is that IGF-1R is associated with insulin resistance in mature adipocytes, which accounted for approximately 80% in adipose tissue [48,50]. As IH has been shown to induce insulin resistance [47], in agreement with an elevation of serum glucose level in IH-exposed rats (Supplementary Figure S3), the findings that downregulated expressions of IGF-1R pathway in both IH-exposed differentiated HPAs (mature adipocytes) and adipose tissue (SAT and VAT) provide novel mechanistic insight into the relationship between IH and insulin resistance of mature adipocytes.
In addition to extracellular signals and the transcriptional cascades, proinflammatory signaling in the adipocyte has recently been identified as an essential step for adipose tissue expansion [51]. Our in vitro study showed that as HPA cells matured in normoxic condition, less IL-6 and MCP were secreted compared to the earlier stages of cell proliferation and maturation, suggesting that proinflammatory capacity might be more prominent at the beginning of physiologic adipogenic differentiation. However, when the cells reached maturity, IH exposure compared to IN led to an increase in the inflammation profile, consistent with our previous in vivo data using the same IH profile [52].
There are several limitations in the current study. Firstly, our findings, being a snapshot at a specific time point of low-frequency IH exposure in a lean rat model, cannot distinguish if the enhanced subcutaneous adipogenesis is acting as a "metabolic sink" or it is an early phase of detrimental adipose tissue accumulation nor can the findings be generalized to that in the obese state. Secondly, the in vitro IH model as used in the current study, or in most other in-vitro IH studies in the literature, has inherent limitations. Oxygenation sensed by the cells would be derived via oxygen diffusion from the gas in the culture chamber into the culture medium, a process that highly depends on the rate of diffusion and equilibrium between gas/liquid, which is far less efficient compared to the biologic system in the animal model. Thus, the real IH pattern experienced by the cells in the in vitro system would not be comparable to that seen in the in vivo model and the cellular events cannot be precisely correlated with the oxygen concentration or pattern to which the cells were exposed [53]. Finally, our study focused on adipocytes, while adipose tissues comprise of both adipocytes and macrophages which are of vital importance in fat tissue metabolism. In obese or hypoxia state, adipose tissue inflammation is accompanied by an imbalance in the ratio of M1/M2 macrophages, with the enhancement of M1 proinflammatory macrophages and the downregulation of M2 anti-inflammatory macrophages [54]. It has been reported that IH exposure (20 cycles/hour, nadir of FiO2: 6.4% for 8 weeks) promoted M1/M2 macrophage polarization in VAT and AT surrounding tumors [55]. It is still unknown whether macrophage infiltration plays any role in the adipogenesis of IH-exposed preadipocytes in lean state, and further studies would be needed to investigate this specific aspect.
In summary, this study demonstrated that IH promoted subcutaneous adipogenesis in vivo in the lean rat model and this was confirmed in an in vitro model of human subcutaneous preadipocytes. The IH-induced subcutaneous adipogenesis involved transcriptional factors including CEBPα, CEBPδ, PPARγ, and IGF-1R/Akt signaling pathway ( Figure 8). As subcutaneous fat is the lesser metabolically harmful adipose depot, our findings of enhanced subcutaneous adipogenesis is further hypothesized to be part of a homeostatic response to IH challenges, in particular when the IH is of mild degree. In the face of the growing epidemic of obesity and associated OSA which comprise of a wide range of severities, it is of tremendous clinical relevance to understand the role of IH on evolution of adipose tissue expansion and distribution. Our current finding is novel and warrants further research, which may shed light on whether the body's intrinsic responses may be harnessed to mitigate adverse sequelae.

Additional Points
Statement of Significance. Obstructive sleep apnea is strongly associated with metabolic disorders, and intermittent hypoxia from recurrent obstructed breathing may be a key pathogenetic factor in metabolic dysfunction. Adipose tissue plays a pivotal role in metabolic regulation, and its expansion is largely dependent on adipocyte differentiation (adipogenesis). We demonstrated that relatively low-frequency intermittent hypoxia in a lean mouse model promoted adipogenesis of subcutaneous but not visceral fat tissue, and the finding was corroborated by enhanced adipogenesis in human subcutaneous preadipocyte cultures. The novel findings suggest a differential response of subcutaneous and visceral fat depot towards this relatively milder degree of IH challenge. A deeper understanding of the underlying mechanisms by which OSA modulates adipose tissue metabolism would be critical in improving management strategy in sleep-disordered breathing which has a wide range of severities.  In addition, the enhanced inflammation elicited by IH exposure could be another possible mechanism responsible for intensified adipogenesis of subcutaneous adiposity. The accelerative subcutaneous adiposity may be a compensatory response to IHstimulated metabolic disorders.

Supplementary Materials
Supplementary Figure S1: