AMP-Activated Kinase (AMPK) Activation by AICAR in Human White Adipocytes Derived from Pericardial White Adipose Tissue Stem Cells Induces a Partial Beige-Like Phenotype

Beige adipocytes are special cells situated in the white adipose tissue. Beige adipocytes, lacking thermogenic cues, morphologically look quite similar to regular white adipocytes, but with a markedly different response to adrenalin. White adipocytes respond to adrenergic stimuli by enhancing lipolysis, while in beige adipocytes adrenalin induces mitochondrial biogenesis too. A key step in the differentiation and function of beige adipocytes is the deacetylation of peroxisome proliferator-activated receptor (PPARγ) by SIRT1 and the consequent mitochondrial biogenesis. AMP-activated protein kinase (AMPK) is an upstream activator of SIRT1, therefore we set out to investigate the role of AMPK in beige adipocyte differentiation using human adipose-derived mesenchymal stem cells (hADMSCs) from pericardial adipose tissue. hADMSCs were differentiated to white and beige adipocytes and the differentiation medium of the white adipocytes was supplemented with 100 μM [(2R,3S,4R,5R)-5-(4-Carbamoyl-5-aminoimidazol-1-yl)-3,4-dihydroxyoxolan-2-yl]methyl dihydrogen phosphate (AICAR), a known activator of AMPK. The activation of AMPK with AICAR led to the appearance of beige-like morphological properties in differentiated white adipocytes. Namely, smaller lipid droplets appeared in AICAR-treated white adipocytes in a similar fashion as in beige cells. Moreover, in AICAR-treated white adipocytes the mitochondrial network was more fused than in white adipocytes; a fused mitochondrial system was characteristic to beige adipocytes. Despite the morphological similarities between AICAR-treated white adipocytes and beige cells, functionally AICAR-treated white adipocytes were similar to white adipocytes. We were unable to detect increases in basal or cAMP-induced oxygen consumption rate (a marker of mitochondrial biogenesis) when comparing control and AICAR-treated white adipocytes. Similarly, markers of beige adipocytes such as TBX1, UCP1, CIDEA, PRDM16 and TMEM26 remained the same when comparing control and AICAR-treated white adipocytes. Our data point out that in human pericardial hADMSCs the role of AMPK activation in controlling beige differentiation is restricted to morphological features, but not to actual metabolic changes.


Introduction
The energy balance of an organism depends on the net of energy intake and energy expenditure. The disequilibrium between energy uptake and energy expenditure has causative role in the pathogenesis of metabolic diseases [1][2][3][4][5]. Energy expenditure stems from the energy deliberated by biochemical processes, physical activity or the action of mitochondria-rich tissues such as skeletal muscle, brown adipose tissue or cardiac muscle [6]. Recently a novel cell type called beige, or brite (blended from brown and white) adipocytes were identified in white adipose tissue (WAT) that seems to play an indispensable role in energy expenditure [7,8].
Inactive beige cells morphologically appear as normal WAT cells, however upon adrenergic stimulation beige adipocytes do not only enhance lipolysis but upregulate mitochondrial biogenesis and mitochondrial oxidation and futile cycle of creatine phosphate generation at the same time [7,[9][10][11]. It is conceivable therefore that due to the proportions of WAT in the human body, beige adipocytes may have similar importance in energy expenditure comparable to skeletal muscle. Importantly, functional beige adipocytes were shown in humans too [7,12], moreover these cells are transplantable [13].
Activating stimuli in beige adipocytes induce SIRT1 that deacetylates and hence activates PPARγ declutching a set of mitotropic events involving peroxisome proliferator activated receptor cofactor-1α (PGC-1α) that lead to the enhanced mitochondrial biogenesis and oxidation [21] that is fine-tuned by a set of micro-RNAs [15,24]. AMP-activated protein kinase (AMPK) is an upstream activator of SIRT1 [25]. This protein kinase is activated by the shortage of cellular energy stores and upon activation AMPK initiates cellular programs that silence anabolic processes to save energy and induce catabolism to resolve the energy crisis [25]. Importantly, AMPK activation had been implicated in brown adipocyte differentiation and function [26]. The known involvement in brown adipocyte function and mitochondrial biogenesis made it very likely that AMPK could be involved in the function of beige adipocytes too. In the present study we set out to assess that possibility.

Methods Chemicals
Unless otherwise stated, all chemicals were from Sigma-Aldrich (St. Louis, MO, USA). Isolation, culture and differentiation of human adipose derived mesenchymal stem cells (hADMSCs) On the day of heart surgery the pericardial adipose tissue specimen were processed as described in [14]. Samples were digested in PBS with 120U/ml collagenase for 1 hour at 37°C and filtered through a sieve with pore size 100μm. Isolated hADMSCs were resuspended in DMEM-F12 medium containing 10% FBS (Gibco) and seeded to the appropriate vessels. After cell culture reached confluency, differentiation was initiated. White adipogenic differentiation was carried out using the protocol of Fischer-Posovszky and co-workers [27]. For brown adipogenic differentiation cells were treated according to Elabd and co-workers [28]. For differentiation FBSfree medium was used. To assess the result of AMPK activation, white adipocytes were differentiated in the presence of 100 μM [(2R,3S,4R,5R)-5-(4-Carbamoyl-5-aminoimidazol-1-yl)-3,4-dihydroxyoxolan-2-yl]methyl dihydrogen phosphate (AICAR) (Tocris, Bristol, UK) [29]. Cells were assayed after 14 days differentiation.

Image acquisition, recognition and texture analysis
Cellular texture analysis was performed using laser scanning cytometry on differentiated cells as described in [14,30]. hADMSCs were plated on Ibidi eight-well μ-slides and differentiated as previously described. On the day of measurement, cells were stained with Hoechst 33342 (50 μg/ml) and with Nile Red (NR, 25 μg/ml) for 20 minutes. Cells were washed once with phosphate buffered saline (PBS) and then kept in fresh medium.
Images were obtained by using iCys (CompuCyte) laser scanning cytometer (Thorlabs Imaging Systems, Sterling, VA, USA) according to the protocol of Doan-Xuan and co-workers [30]. Images were processed and analyzed by CellProfiler (The Broad Institute of MIT, MA). Hoechst-stained nuclei from both differentiated and undifferentiated cells were first identified and marked as primary objects. Differentiated adipocytes were later recognized by their lipid droplet specific Nile Red absorption signal. Cells above a preset threshold were considered as differentiated adipocytes. The threshold was calculated from the efficiency of NR staining in every measurement where intensity values were plotted against individual cell counts. Staining efficiency of undifferentiated preadipocytes were used as a baseline. Texture analysis per identified objects was done with built-in modules in CellProfiler. Parameter entropy measured the randomness of intensity distribution; sum entropy informed about the number of lipid droplets. Parameter variance measured the difference between intensity of the central pixel and its neighborhood; sum variance depicted the size of lipid droplets.

Analysis of mitochondrial fusion
The structure of the mitochondrial network in cells changes its shape as a function of cellular bioenergetics (e.g. fasting or feeding) and environmental stimuli [31,32]. The structure of the mitochondrial network changes between a fully fused (long, interconnected mitochondrial tubes) and fully fragmented state (smaller, individual mitochondria with dotted appearance) as hypothetical endpoints, where the fused state is associated with better mitochondrial oxidative activity [31,32].
Preadipocytes were seeded on Ibidi eight-well μ-slides and differentiated as previously described. On the day of analysis cells were stained with Mitotracker Red (Thermo Scientific, MA, USA) using a working concentration of 100 nM for 20 minutes at 37°C. That dye charges the mitochondria enabling the visualization of the mitochondrial network. Cells were washed once with PBS and then kept in fresh medium. Images were taken by a Leica TCS SP8 confocal microscope (Leica Microsystems, Wtzlar, Germany).
Differentiated cells were grouped into three categories according to the morphology of their mitochondrial network. When characterizing a sample, we analyzed and scored 100 cells and each cell was scored between 1 to 3 as a function of mitochondrial network morphology. Cells with dotted staining were scored 1 (Stage 1 on the figures) implying no or minimal fusion between mitochondria. In cells scored 2 (Stage 2 on the figures) the mitochondrion-specific staining draws a much more elongated mitochondria as a result of elevated fusion. Cells

Determination of cellular oxygen consumption
Oxygen consumption was measured using an XF96 oximeter (Seahorse Biosciences, North Billerica, MA, USA). Cells were seeded and differentiated in 96-well XF96 assay plates. On the day of measurement, after recording the baseline oxygen consumption, cells received a single bolus dose of dibutyril-cAMP (500 μM final concentration) simulating adrenergic stimulation. Then, stimulated oxygen consumption was recorded every 30 minutes. The final reading took place at 7 h post-treatment. As a last step, cells received a single bolus dose of antimycin A (10 μM) for baseline correction [14,34].

Reverse transcription-coupled quantitative PCR (RT-qPCR)
Total RNA preparation, reverse transcription, and RT-qPCR were performed as in [35]. Total cellular RNA was isolated using TRIzol Reagent (Molecular Research Center, OH, USA). Primers are summarized in Table 1. All reactions were run in a LightCycler 480 (Roche Diagnostics, Mannheim, Germany) instrument using SYBR green chemistry except for TBX1 where primers and probes were designed and supplied by Applied Biosystems (Taqman Hs00271949_m1, Applied Biosystems). Expression was normalized to the geometric mean of two control genes (G6PD-glucose-6-phosphate dehydrogenase, 36B4-ribosomal protein, large, P0). Gene expression values were calculated based on the ΔΔCt method, where white samples were designated as calibrator.

Statistical analysis
The appropriate test and the details of cohort are given in the figure legends.

AICAR-induced AMPK activation leads to beige-like morphological changes in hADMSCs-derived white adipocytes
Each hADMSCs cell line (= each individual) was differentiated in three directions, namely towards beige adipocytes, white adipocytes and AICAR-treated white adipocytes. First we checked AMPK activity in the three groups at the end of the differentiation. AMPK activity was higher in beige than in white adipocytes (Fig 2). Importantly, the treatment of white adipocytes with 100 μM AICAR enhanced AMPK activity almost to the same extent as in beige adipocytes (Fig 2) that suggested a role for AMPK in beige adipocyte differentiation and function. Next we evaluated morphological changes upon AICAR treatment. Previous studies have shown that beige adipocytes have smaller lipid droplets in larger numbers as white adipocytes [14,30]. Laser scanning cytometry (LSC) was proved to be an efficient tool to characterize adipocyte morphology [14,30], therefore we performed LSC analysis on our samples. We have reproduced the previously-described size and number differences of the lipid droplets between white and beige adipocytes [14,30] (Fig 3A-3C). The treatment of white adipocytes with AICAR reduced the average size of lipid droplets and concomitantly increased the total number of droplets (Fig 3A-3C).
The biological function of beige adipocytes depend on mitochondrial biogenesis and the upregulation of mitochondrial oxidation, therefore we continued our experiments by assessing mitochondrial function. First we assessed mitochondrial morphology that changes in accordance with the mitochondrial oxidative function [36][37][38][39]. The mitochondrial network was assessed by fluorescent microscopy. The mitochondrial network was more fused when cells were differentiated towards beige adipocytes as compared to white adipocytes (Fig 4). Upon the activation of AMPK by AICAR, the mitochondrial network of adipocytes became more similar to the one of beige adipocytes that was dominated by fused mitochondria suggesting enhanced mitochondrial activity (Fig 4).

AICAR treatment of hADMSCs-derived white adipocytes does not yield functional beige adipocytes
The experiments in the previous chapter suggest enhanced mitochondrial oxidation, therefore we measured mitochondrial oxygen consumption. In accordance with our previous report [14], beige adipocytes displayed higher basal and cAMP-stimulated oxygen consumption rate as white adipocytes, however AICAR-treatment of white adipocytes did not increase oxygen consumption (Fig 5A).
This surprising finding prompted us to assess validated markers of beige differentiation [40]. We measured the mRNA levels of uncoupling protein-1 (UCP1) that is a mitochondrial internal membrane protein that bypasses ATP synthase and creates heat out of the proton gradient, CIDEA, PRDM16 that are transcriptional co-activators, T-box protein 1 (TBX-1) that is a transcription factor and TMEM26, a transmembrane protein of unknown function. Out of these markers TBX-1 is known as a validated beige marker, the rest appear both in beige and brown adipocytes [40][41][42]. All markers were expressed at higher levels in beige cells as compared to white adipocytes, however we were unable to detect increases upon AICAR treatment as compared to white adipocytes (UCP1 was slightly induced) (Fig 5B).
Taken together, these data suggest that white adipocytes that are differentiated from pericardium-derived hADMSCs, upon treatment AICAR do not become functional beige cells. They Red and mitochondrial network was assessed by confocal microscopy; cells were scored as described in the Materials section. The proportions of the fragmented and fused mitochondria were plotted as pie charts (n = 4). To assess statistical significance chi square test was performed, where the distribution of the white adipocytes was the expected distribution. The distribution of the AICAR-treated white adipocytes and beige adipocytes were significantly different from the untreated white adipocytes (p<0.01), while there was no statistical difference between AICAR-treated white adipocytes and beige adipocytes. Representative images of the mitochondrial network is presented on the figure. A part of these images had been enlarged. display only certain morphological features, but apparently in terms of biochemical functions, AICAR-treated cells rather behave as white adipocytes.

Conclusions
Beige adipocytes represent a newly described pool of adipocytes with physiologically important capacity of oxidation that is implicated in the pathology of type II diabetes and probably a myriad of other metabolic diseases [7,43]. Qiang and colleagues have identified the SIRT1 -PPARγ pathway as the central regulator of beige adipocyte differentiation and function [21]. SIRT1, a protein deacetylase, is the member of the sirtuin protein family that was implicated in several human metabolic and degenerative diseases [2,44]. SIRT1 has central role in the adaptation to fasting and other nutritional or environmental stress. SIRT1 achieves its effects through deacetylating key transcription factors (PGC-1α, FOXO1, FOXO3, p53) that then initiate transcription programs culminating in the suppression of cellular synthetic processes and the induction of cellular catabolism and mitochondrial biogenesis [2,44]. SIRT1 works in collaboration with other cellular energy sensors such as mechanistic Target of Rapamycin (mTOR), hypoxia-inducible factors (HIFs), nuclear respiratory factors (NRFs) or AMPK [2,[44][45][46][47]. AMPK has very similar role to SIRT1, when cellular energy charge decreases marked by increase in AMP levels and the decrease of ATP, AMPK is activated and shuts down energy-consuming anabolic processes and turns on mitochondrial oxidation and catabolism to replenish cellular energy [25,48]. SIRT1 and AMPK not only display functional synergy, but their activity is cross-regulated [25,[49][50][51][52][53]. Several investigators have shown that AMPK and SIRT1 act on the same transcription factors (e.g. PGC-1α) that requires both AMPK-mediated phosphorylation and SIRT1-mediated deacetylation for activation and the subsequent induction of mitochondrial biogenesis [25]. Furthermore, AMPK and SIRT1 are also cross-connected through modulating NAMPT and hence cellular NAD + salvage [52]. Since the same target transcription factors and SIRT1 were implicated in the differentiation and function of beige cells (e.g. PGC-1α) [15,17,21,22,54,55] it was logical to assess the function of AMPK, which is situated upstream of SIRT1, in beige differentiation. This hypothesis was further supported by the fact that AMPK activation is important in brown adipose tissue differentiation, moreover induces differentiation of white adipocyte precursors to brown adipocytes [55][56][57][58][59][60][61].
We used AICAR for the activation of AMPK [29] and found typical morphological features of beige cells are formed upon AICAR treatment. However, the very necessary induction of mitochondrial biogenesis and the increased expression of the markers of beige differentiation (TBX-1, UCP1, TMEM26, PRDM16, CIDEA) did not take place. It seems that AMPK activation does not induce the formation of functional beige cells in pericardial hADMSCs, but AMPK activation is restricted to the induction of morphological changes. The actual cause of the restricted capability of AMPK in inducing beige differentiation remains elusive. Terminal differentiation of beige cells depend on SIRT1 activation [21], while AMPK is capable of activating SIRT1 [25]. The interconnection between AMPK and SIRT1 relies on the modulation of small molecule enzyme cofactors such as NAD + [25,48]. Characteristic differences in the metabolism of these cofactors or low AMPK expression in beige cells may explain blunted response to AMPK activation, although to verify these hypothesis require further investigations.
It is a question whether the restricted role of AMPK is true for adipose tissue-derived stem cells of any origin or is it specific for only (a) certain fat depot(s). Indeed, the expression of PRDM16 and TBX-1 in beige cells versus white cells is lower in our study than in another study using stem cells from subcutaneous adipose tissue [14]. Yet the question cannot be answered with certainty, however the known differences between adipose tissue depots [62,63] makes it likely that AMPK activation may have a more pronounced effect in other adipose tissue depots. It should be noted, however, that this study is the first to report that pericardial adipose tissue-derived stem cells can be differentiated to beige adipocytes.
Although several known external stimuli (hormones [7,[14][15][16][17][18], neuronal stimuli [19,20] or drugs [21,22]) may activate beige cells, the intracellular signal transduction pathways that translate these signals are still to be elucidated. It seems that β-adrenergic stimulus and signals induce beige adipocyte function [64,65]. However, serotonin (5HT) had been shown to blunt thermogenic activity in brown and beige cells in mice [66,67]. Our data nominate the energy and nutrient sensing system as possible actors in beige function and in signal transduction. Along the same line, there is a vast area of metabolic processes and drugs yet not investigated with regard to beige function, such as fasting and exercise mimetics, metabolic drugs and proteins, regulatory circuits of feeding and fasting or circadian rhythm-our data presented here clearly point out the vast potential of they may provide.