Examination of ex-vivo viability of human adipose tissue slice culture

Obesity is associated with significantly higher mortality rates, and excess adipose tissue is involved in respective pathologies. Here we established a human adipose tissue slice cultures (HATSC) model ex vivo. HATSC match the in vivo cell composition of human adipose tissue with, among others, mature adipocytes, mesenchymal stem cells as well as stroma tissue and immune cells. This is a new method, optimized for live imaging, to study adipose tissue and cell-based mechanisms of obesity in particular. HATSC survival was tested by means of conventional and immunofluorescence histological techniques, functional analyses and live imaging. Surgery-derived tissue was cut with a tissue chopper in 500 μm sections and transferred onto membranes building an air-liquid interface. HATSC were cultured in six-well plates filled with Dulbecco’s Modified Eagle’s Medium (DMEM), insulin, transferrin, and selenium, both with and without serum. After 0, 1, 7 and 14 days in vitro, slices were fixated and analyzed by morphology and Perilipin A for tissue viability. Immunofluorescent staining against IBA1, CD68 and Ki67 was performed to determine macrophage survival and proliferation. These experiments showed preservation of adipose tissue as well as survival and proliferation of monocytes and stroma tissue for at least 14 days in vitro even in the absence of serum. The physiological capabilities of adipocytes were functionally tested by insulin stimulation and measurement of Phospho-Akt on day 7 and 14 in vitro. Viability was further confirmed by live imaging using Calcein-AM (viable cells) and propidium iodide (apoptosis/necrosis). In conclusion, HATSC have been successfully established by preserving the monovacuolar form of adipocytes and surrounding macrophages and connective tissue. This model allows further analysis of mature human adipose tissue biology ex vivo.


Introduction
The prevalence of overweight and obesity, commonly measured by a high body mass index (BMI), is notably increasing worldwide [1]. Meanwhile, current research has identified a high

Tissue samples
This study has been approved by the Ethical Committee at the Medical Faculty, Leipzig University (#290-13-07102013). All patients declared their informed consent in written form. Subcutaneous AT was obtained from the Department of Orthopedics, Trauma Surgery, and Plastic Surgery (University Hospital Leipzig, Germany). ATs were derived from abdomen, dorsum, mamma, pelvis and thigh ( Table 1). The samples were transported in sterile Hanks' Balanced Salt Solution (HBSS, Gibco, Life Technologies, Carlsbad, USA) or DMEM (Gibco) and were processed within one to six hours after dissection.

Live imaging
15 minutes prior to imaging the media were removed and replaced by basic culture media containing fluorescent dyes: Hoechst 33342 (Nuclei, 1:1000, Sigma Aldrich), propidium iodide (apoptosis/necrosis, 1:1000, PI, Calbiochem, Darmstadt, Germany) and Calcein-AM (unspecific metabolism 1:200, Life Technologies). Pictures for 3D reconstruction or videos were taken with an Olympus IX81 confocal microscope (FV1000, Olympus) equipped with a humidified incubator and a motorized stage. During the imaging procedure, inside temperature was adjusted to 35˚C, 5% CO 2 , and 60% humidity.

PLOS ONE
Viability of human adipose tissue slice culture

Tissue analysis
At 0, 1, 7 or 8, and 14 days in vitro (DIV), slices were fixed over-night in 4% paraformaldehyde (PFA) prior to paraffin embedding. Paraffin sections (10 μm) were cut, dewaxed in xylene, dehydrated in decreasing alcohol series, and stained with hematoxylin/eosin (H/E) for conventional histology. Pictures were taken using an optical microscope, Axioplan 2 (Carl Zeiss, Oberkochen, Germany). In order to establish immunofluorescence, sections were pretreated with citrate buffer (pH 6) in a microwave for 10 minutes and antibodies were incubated overnight at 4˚C in 1,5% Triton/PBS with 0,5% bovine serum albumin (BSA, Sigma Aldrich) and 10% normal goat serum or normal donkey serum (NGS or NDS, Jackson Immuno Research, West Grove, USA

Western blot
48 hours prior to the experiment, adipocyte tissue cultures were serum starved overnight by changing medium to serum and insulin free medium. To determine Akt phosphorylation, adipocytes were stimulated with insulin (10 nM), diluted in pre-warmed serum free medium for 15 min. Separation of membranes and cytosol was performed by a protocol modified from Nishiumi and Ashida [55]. Briefly, adipocyte cultures were collected in buffer A (50 mmol/l Tris, 0.5 mmol/l dithiothreitol, adjusted to pH 8.0 and 1% phenylmethylsulfonyl fluoride (PMSF), 10 mM sodium orthovanadate and 1% Sigma protease inhibitor cocktail freshly added) and stored at -80˚C until further analysis. Western blot analysis was performed as described earlier [56]. Blots were incubated with Phospho-Akt (1:1000, Cell Signaling) at 4˚C overnight. Immunoreactions were detected with the appropriate peroxidase-conjugated anti- rabbit IgG secondary antibody (1:5000 for phospho-specific antibodies; Vector Laboratories, Peterborough, UK) at room temperature for 2 h. Peroxidase activity was visualized with an enhanced chemiluminescence kit (Amersham, Pharmacia, Freiburg, Germany). In addition, blots were stripped and incubated with pan-Akt antibody (1:3000, Cell Signaling) followed by a secondary antibody (1:10000, anti-rabbit IgG, Vector Laboratories). In addition, blots were stripped and incubated with anti-glyceraldehyde-3-phosphate dehydrogenase (GAPDH) monoclonal antibody (diluted 1:100000, Research Diagnostics, Flanders, Netherlands) followed by the anti-mouse IgG secondary antibody (1:10000, Vector Laboratories). GAPDH antibodies were used as a loading control. Semiquantitative evaluation of arbitrary unit was performed with the ImageJ plugin for western blot analysis.

Statistical analysis
One-way-ANOVA with Bonferroni correction was performed using GraphPad Prism 6 (GraphPad Software, Inc., La Jolla, USA). P <0.05 was considered significant.

Results
Adipose tissues were derived from nine abdominal, four dorsal, three breast, three pelvic, and three limb surgeries (Table 1) and were kept in culture for up to 14 days. The tissue donors were 43,9 years (SD ±15,9) old on average, had a BMI of 31,0 kg/m 2 (SD ±8,0) on average, and 40,9% were female. In 14 cases, the patients had secondary diagnoses (Table 1). Tissue were transferred from surgery and cut on a tissue chopper between 1 and 6 hours after removal ( Fig  1). Pioneering experiments investigating the optimal thickness for cultivation showed that 500 μm were ideal for tissue preparation and handling. 350 μm thick slices often collapsed during the preparation process while 750 μm slices proved difficult to embed into paraffin for further histological analysis. 500 μm thick slices maintained approximately five to seven cell layers and all layers were well preserved during cultivation. Tissue integrity was macroscopically well preserved up to 14 DIV and adipocyte appearance did not change (Fig 2). Tissue slices cultivated with 10% FBS ( Table 2, III) showed minor slice shrinkage (Fig 2B, 2E, 2H and 2K). To induce distinct tissue damage, TNFα was supplemented to serum-free media provoking cell death (Figs 2C, 2F, 2I, 2L and 4).
Histological analysis of H/E staining's demonstrated the well-maintained characteristics of AT (Fig 3). The cross-section area of adipocytes increased under TNFα supplementation, while no measurable difference was observed between the two different culture media between 0 and 14 DIV (Fig 3). Slices cultivated without serum supplementation (I, Table 2) maintained their cellular composition up to 14 DIV. In medium supplemented with serum, stroma tissue appeared to expand, but no obvious discrepancy could be observed between conditions with a dose of 5% FBS (II, Table 2) as compared to 10% FBS (III, Table 2) (S1 Fig). Taking into consideration the high serum doses used in previous experiments (e.g. 15% FBS [57]), 10% FBS was used for our further experiments.
Viability of adipose cells was determined via immunofluorescence with antibodies against Perilipin A (Fig 4). No obvious differences between the endpoints of the standard conditions I and III (Table 2) could be observed. Homogeneous expression of the lipid droplet surface protein in both conditions proved the survival of adipocytes in slice cultures for 14 days. As a positive control for cell death, TNFα was added to the medium (IV, Table 2; Fig 4D). Thus, positive controls confirmed the predictive validity of Perilipin A. To further investigate function of adipocytes the phosphorylation of Akt, a key step in insulin signaling, was investigated after 15 min of insulin stimulation. After 7 and 14 DIV the phosphorylation of Akt remains low in the control condition, whereas the insulin condition shows an enhanced expression of phosphorylated Akt (Fig 5A & 5B). Using antibodies against IBA1 and CD68 revealed viability of macrophages. Some CD68-positive macrophages were co-localized with Ki67, proving proliferation processes on 14 th DIV (Fig 6, Circle).
In live imaging analyses medium with Calcein-AM was added 15 minutes prior to taking pictures. The non-fluorescent Calcein-AM diffused through cell membranes, intracellular

PLOS ONE
Viability of human adipose tissue slice culture esterases hydrolyzed the acetoxymethylester-group, and the fluorescent Calcein accumulated in cell plasma [58]. This activating process could be shown in vital adipocytes and stroma cells on day 7 and 14 in vitro (Fig 7, S1 Video). However, exposing the tissue to laser light provoked a positive control cellular death of both adipocytes and stroma cells after prolonged exposure (S2 Fig, S2 Video).

Discussion and conclusion
Organotypic slice cultures of human adipose tissue maintained their morphological characteristics and their metabolism for up to 14 days in culture. It can thereby be concluded that HATSC provides a platform to investigate human adipose tissue in a controlled ex vivo setting with little ethical constraints.
Morphological analysis of H/E staining's cannot discriminate viable adipocytes from dead adipocytes; The distinction between living and dead adipocytes (lipid droplets with or without

PLOS ONE
Viability of human adipose tissue slice culture cell nuclei) cannot be made on the basis of cell nuclear morphology. In histologic standard sections (approx. 10 μm) of normal-sized adipocytes (approx. 50-150 μm) their small nucleus is not necessarily part of the histological section. Therefore, the state of a given adipocyte cannot always be judged in the basis of its morphology (e.g. euchromatic, pyknotic, fragmented). Studies in the literature on changes in the cell volume of adipocytes as a distinguishing feature of vitality or apoptosis/necrosis of adipocytes do not exist. Jo et al. did not find an alteration in cell volume during apoptosis, using mathematical models [59]. They could show that under prolonged weight-loss conditions large adipocytes shrink, but at the same time the smaller adipocytes die first [60]. A faster cell death of the small adipocytes could lead to an increase in the average cell volume, even if the large adipocytes themselves shrink. In a human study, Verboven et al. could show that people with obesity, i.e. those with an increased inflammation in fatty tissue, have more large and very large adipocytes, although they have an increased basal lipolysis [61]. They attribute this to a decrease in the number of small adipocytes. Both processes, early cell death of small adipocytes and the slow shrinkage of large adipocytes could explain the total increase in adipocyte surface area of the TNFα condition that was observed in the present study (see Fig 3A).
The survival of adipocytes was demonstrated by immunofluorescence staining and the viability of adipose tissue by live imaging and functional experiments. Perilipins are lipid dropletassociated proteins and their phosphorylation is essential in lipolysis. Perilipin A is a known marker for viability of adipocytes and has been used in immunological research [62][63][64][65][66]. Using our own experiments, we were able to confirm the sensitivity of Perilipin A as a viability marker of adipocytes (see Fig 4A and 4D).
Pathological remodeling of adipose tissue includes hypertrophy, accumulation of immune cells such as macrophages, decreased capillary density, and fibroblast activation [67]. There is some evidence that adipose tissue is able to control local regulation and proliferation of macrophages independently of the influx of blood precursors, but no evidence currently exists supporting the role of local myelopoiesis in adipose tissue [68][69][70][71][72]. The preservation of macrophages in HATSC was displayed with immunofluorescence staining with IBA1 and CD68. Both antibodies were selected for their wide distribution in the study of macrophages in adipose tissue [73,74]. Even the local proliferation in the absence of blood could be demonstrated on DIV 14 in human adipose tissue, where CD68-positive macrophages were co-localized with Ki67 (Fig 6).
Live-imaging was used to further study the formation of crown-like structures consisting of proliferating macrophages around dying adipocytes in murine adipose tissue [75,76]. The results show that such experiments could also be performed in human tissues. In a rodent study by Weisberg et al. TNFα was shown to be distributed by macrophages and not by adipocytes as part of the stroma-vascular fraction during inflammation and diabetes [19,77]. Contrary to this, human adipocytes have the potential to secrete TNFα, thus signaling to immune cells [78][79][80]. Such potential species differences can now be worked out. Moreover, studying human adipose tissues from individuals with different BMI with and without type II diabetes might help to better understand how inflammation and oxidative stress drives insulin resistance, arteriosclerosis, angiogenesis, as well as cancer [81][82][83][84]. This can be concluded since the open access of the system allows for studies of the accumulation of (secreted) molecules in the medium. The large number of standardized samples which can be prepared from small probes

PLOS ONE
Viability of human adipose tissue slice culture further enables the investigation of therapeutics, targeting essential biochemical pathways, drug accumulation, and modern pharmaceutics for gene therapy [85].
In conclusion, slice cultures derived from human adipose tissue have been successfully established whereas the unique monovacuolar shape of the adipocytes as well as the complex organization of the tissue could be maintained. Hence, this method serves as a confirmation of the current findings about rodent adipose tissue while it allows to further dissect its biology in the human system.