Differences in Gene Expression and Cytokine Release
Profiles Highlight the Heterogeneity of Distinct Subsets
of Adipose Tissue-Derived Stem Cells in the
Subcutaneous and Visceral Adipose Tissue in Humans
Sebastio Perrini1, Romina Ficarella1, Ernesto Picardi3, Angelo Cignarelli1, Maria Barbaro1,
Pasquale Nigro1, Alessandro Peschechera1, Orazio Palumbo5, Massimo Carella5, Michele De Fazio2,
Annalisa Natalicchio1, Luigi Laviola1, Graziano Pesole3,4, Francesco Giorgino1*
1Department of Emergency and Organ Transplantation, Section on Internal Medicine, Endocrinology, Andrology and Metabolic Diseases, University of Bari Aldo Moro,
Bari, Italy, 2 Section on General Surgery and Liver Transplantation, University of Bari Aldo Moro, Bari, Italy, 3Department of Biosciences, Biotechnology and
Pharmacological Sciences, University of Bari Aldo Moro, Bari, Italy, 4 Institute of Biomembranes and Bioenergetics, National Research Council, Bari, Italy, 5 Istituto di
Ricovero e Cura a Carattere Scientifico Casa Sollievo della Sofferenza, San Giovanni Rotondo (FG), Italy
Abstract
Differences in the inherent properties of adipose tissue-derived stem cells (ASC) may contribute to the biological specificity
of the subcutaneous (Sc) and visceral (V) adipose tissue depots. In this study, three distinct subpopulations of ASC, i.e.
ASCSVF, ASCBottom, and ASCCeiling, were isolated from Sc and V fat biopsies of non-obese subjects, and their gene expression
and functional characteristics were investigated. Genome-wide mRNA expression profiles of ASCSVF, ASCBottom and ASCCeiling
from Sc fat were significantly different as compared to their homologous subsets of V-ASCs. Furthermore, ASCSVF, ASCCeiling
and ASCBottom from the same fat depot were also distinct from each other. In this respect, both principal component
analysis and hierarchical clusters analysis showed that ASCCeiling and ASCSVF shared a similar pattern of closely related genes,
which was highly different when compared to that of ASCBottom. However, larger variations in gene expression were found
in inter-depot than in intra-depot comparisons. The analysis of connectivity of genes differently expressed in each ASC
subset demonstrated that, although there was some overlap, there was also a clear distinction between each Sc-ASC and
their corresponding V-ASC subsets, and among ASCSVF, ASCBottom, and ASCCeiling of Sc or V fat depots in regard to networks
associated with regulation of cell cycle, cell organization and development, inflammation and metabolic responses. Finally,
the release of several cytokines and growth factors in the ASC cultured medium also showed both inter- and intra-depot
differences. Thus, ASCCeiling and ASCBottom can be identified as two genetically and functionally heterogeneous ASC
populations in addition to the ASCSVF, with ASCBottom showing the highest degree of unmatched gene expression. On the
other hand, inter-depot seem to prevail over intra-depot differences in the ASC gene expression assets and network
functions, contributing to the high degree of specificity of Sc and V adipose tissue in humans.
Citation: Perrini S, Ficarella R, Picardi E, Cignarelli A, Barbaro M, et al. (2013) Differences in Gene Expression and Cytokine Release Profiles Highlight the
Heterogeneity of Distinct Subsets of Adipose Tissue-Derived Stem Cells in the Subcutaneous and Visceral Adipose Tissue in Humans. PLoS ONE 8(3): e57892.
doi:10.1371/journal.pone.0057892
Editor: Andrea Vergani, Children’s Hospital Boston, United States of America
Received October 14, 2012; Accepted January 28, 2013; Published March 5, 2013
Copyright: � 2013 Perrini et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits
unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Funding: This work was supported in part by grants from the Ministero dell’Universita` e Ricerca (Italy), the COST Action BM0602, the Fondazione della Societa`
Italiana di Diabetologia per la Ricerca in Diabetologia e Malattie Metaboliche (Fo.Ri.SID), and NovoNordisk to F. Giorgino. The funders had no role in study design,
data collection and analysis, decision to publish, or preparation of the manuscript.
Competing Interests: This study was partly funded by NovoNordisk. Francesco Giorgino is a PLOS ONE Editorial Board member. However, there are no patents,
products in development or marketed products to declare. This does not alter the adherence of the authors to all the PLOS ONE policies on sharing data and
materials, as detailed online in the guide for authors.
* E-mail: francesco.giorgino@uniba.it
Introduction
The biological diversity of adipose tissue depots has become
a fundamental issue in recent years, in light of its potential impact
on human health [1]; [2]. It is known that visceral adipose tissue is
morphologically and functionally different from subcutaneous
adipose tissue [3]; [4]. Depot-related variations have long been
described for a variety of biological endpoints, such as signaling
reactions [5], glucose metabolism [6]; [7] and cytokine secretion
[8]. It has been also proposed that extrinsic factors, including
depot-specific blood flow, cell density, cell heterogeneity and/or
innervation [9], could contribute to distinct gene expression
patterns and metabolic profiles of adipocytes from different
anatomical regions. Alternatively, variations in the inherent
properties of undifferentiated fat cell progenitors may dictate the
biological specificity between the two fat depots bringing about the
innate characteristics of Sc and V adipose cells. This concept has
been supported by the demonstration that pre-adipocytes from
distinct fat depots retain specific dynamic characteristics and gene
expression patterns even after 40 cell doublings [10]. More
recently, we have shown that depot-related differences in gene
expression, adiponectin secretion, and insulin signaling and action
PLOS ONE | www.plosone.org 1 March 2013 | Volume 8 | Issue 3 | e57892
were still evident when precursor cells isolated from bioptic
adipose tissue fragments were differentiated in vitro [11].
Altogether, these findings support the concept that there could
be an early commitment of fat precursor stromal cells able to
influence the biological responses of the resulting adipocytes,
independently of extrinsic influences deriving from tissue micro-
environment. However, little is known about the identity,
localization, or specific characteristics of endogenous adipocyte
progenitors.
Adipocyte progenitors recognizably reside in the adipose
stromal-vascular fraction (SVF), a heterogeneous mixture of cells
operationally defined by enzymatic dissociation of fat tissue
followed by density separation from adipocytes [12]. Initially,
the stromal-derived cells were designated as «pre-adipocytes», but
since 2004 the International Fat Applied Technology Society
adopted the term «adipose-derived stem cells» (ASC) to define the
plastic-adherent cells isolated from SVF of adipose tissue with self-
renewal properties and able to regenerate adipocytes (ASCSVF)
[12]; [13]. Further complexity to this scenario, however, is
provided by the observation that, during the procedure of fat tissue
enzymatic digestion and centrifugation, two additional ASC
populations can be isolated from the «fat cake» at the top of the
supernatant [13]. One ASC population is obtained from a pre-
adipocyte fraction in the fat cake not previously collected together
with the ASCSVF; these cells can be grown at the bottom surface of
the culture flask (ASCBottom). Another ASC population develops
from mature adipocytes of the fat cake adherent to the ceiling
surface of the culture flask filled with growth medium (ASCCeiling).
It has been suggested that ASCCeiling may derive from mature
adipocytes through an asymmetric mitosis [14]; [15]. In culture,
ASCCeiling and ASCBottom display both cell-surface markers that
are similar to those expressed by ASCSVF, including CD105, SH3,
Stro-1, CD49d and CD44, and show similar potential for
unlimited self-renewing proliferation and differentiation along
the mesenchymal lineage to produce adipocytes, osteoblasts, and
chondrocytes [15].
In addition to their ability to differentiate into multiple
mesenchymal cell lineages, ASC actively produce paracrine
factors, hormones and metabolic signals, in a distinct manner
from that of differentiated fat cells. Indeed, non-fat cells and ASC
are considered one of the main sources of pro-inflammatory
adipokine released by the adipose tissue [16], with great potential
for repercussions on distant target tissues and metabolic and
cardiovascular regulation [17]. Several studies have reported that
ASC and adipocytes contribute roughly similarly to the overall
secretion/expression of adipokines, except for adiponectin and
leptin [16]; [18]; [19]. Thus, the different production of
adipokines, such as IL-6 or MCP-1, between subcutaneous and
visceral adipose tissues may reflect either intrinsic properties of the
resident ASC or a different ASC proportion within each fat depot
[20]; [21], underlying the contribution of ASC to the biological
diversity of specific adipose tissue depots.
In this study, ASCSVF, ASCBottom and ASCCeiling were isolated
from abdominal subcutaneous (Sc) and visceral omental (V)
adipose tissue biopsies obtained from non-obese subjects and
studied through a genome-wide differential gene expression
analysis followed by an in-depth bioinformatics examination. In
addition, 27 cytokines were measured in the culture medium
collected from each of the 6 ASC populations. By performing
comparative intra-depot (i.e., ASCSVF vs. ASCCeiling vs. ASCBot-
tom) and inter-depot (i.e., Sc-ASC vs. V-ASC) analyses, differences
in transcripts data and cytokines output emerged, which allowed
us to identify six ASC subsets in the Sc and V fat depots that differ
among them for gene expression profiles e cell functions.
Materials and Methods
Materials
All tissue culture reagents were purchased from Life Technol-
ogies (Carlsbad, CA, US). Unless otherwise stated, all chemicals
used were obtained from Sigma-Aldrich (St. Louis, MO, US).
Recombinant human insulin was purchased from Roche Diag-
nostics (Mannheim, Germany). The thiazolinedione (TZD)
compound rosiglitazone was kindly provided by GlaxoSmithKlein
(Middlesex, UK). Recombinant human fibroblast growth factor 2
(FGF2) and EGF were obtained from Miltenyi Biotec (Bergisch
Gladbach, Germany).
Human Donors and Adipose Tissue Biopsies
Paired Sc and V adipose tissue biopsies were obtained from 15
non-obese subjects undergoing elective open-abdominal surgery (9
men, 6 women; age 6867 yrs; BMI 27.061.5 kg/m2; fasting
plasma glucose 85611 mg/dl). None of the patients had diabetes
or severe systemic illness, and none were taking medications
known to affect adipose tissue mass or metabolism. The protocol
was approved by the Independent Ethical Committee at the
Azienda Ospedaliero-Universitaria Policlinico Consorziale, Bari,
Italy. All patients gave their written informed consent.
ASC Isolation
Human ASC were isolated as previously described [11]; [14],
with minor modifications. Briefly, fat tissue fragments were minced
and digested in medium containing 1 mg/ml collagenase, type I,
with gentle shaking at 37uC for 1 h. Resulting material was filtered
through 250 mm mesh, and adipocytes and free oil were separated
from stromovascular components by centrifugation at 1,200 rpm
for 5 min at room temperature. The floating fraction consisting of
pure isolated adipocytes was placed in 25-cm2 culture flasks
completely filled with DMEM/Ham’s F12 1:1 supplemented with
20% fetal bovine serum, and cells were incubated at 37uC in 5%
CO2. The primary ASC grown at top and bottom of flask,
respectively, were cultured for 7 days until they reached
confluence (defined as passage 0), and were then split into 60-
mm plates. The stromovascular pellet was resuspended in
erythrocyte lysis buffer, consisting of 154 mM NH4Cl, 10 mM
KHCO3, and 0.1 mM EDTA, for 5 min. The cell suspension was
centrifuged at 1,200 rpm for 5 min and then resuspended in
a culture medium consisting of DMEM/Ham’s F12, 10% FCS
and antibiotics. This cell suspension was filtered through a 25-mm
sterile nylon mesh before being plated. After a 16-h incubation for
cell attachment, cells were cultured in ASC growth medium
(DMEM/Ham’s F12 1:1 supplemented with 100 units/ml
penicillin, 0.1 mg/ml streptomycin, 2.5% FCS, 1 ng/ml FGF2,
and 10 ng/ml EGF). Except when indicated, all cells were used in
the experimental procedures at passage 4.
Mesenchymal Differentiation
For adipogenic differentiation, confluent ASC at passage 4 were
differentiated in a chemically defined serum-free medium contain-
ing antibiotics, 2 nM triiodotyronine (T3), 100 nM human insulin,
100 nM dexamethasone, and 1 mM rosiglitazone, as previously
reported [11]; for the first 4 days of the differentiation period,
0.5 mM of methyl-isobutylxanthine was also added. Osteogenic
differentiation was induced as previously described [22]. Differ-
entiation into chondrocytes was induced by StemPro Chondro-
genesis Differentiation Medium, according to the manufacturer’s
instructions.
Heterogeneity of Human Adipose Tissue Precursors
PLOS ONE | www.plosone.org 2 March 2013 | Volume 8 | Issue 3 | e57892
Histochemical Staining of Differentiated Cells
Oil-Red-O, Alizarin Red, and Alcian Blue staining, respective-
ly, were performed as described previously [11]; [22]; [23], and
pictures were taken on wide-field microscopes (Nikon) with a color
CCD camera.
RNA Isolation and Quantitative RT-PCR (qRT-PCR)
RNA from the ASC populations was isolated by using an
RNeasy kit (Qiagen, Hamburg, Germany). 250 ng of RNA were
reverse-transcribed with standard reagents (Applied Biosystems).
One microliter of each reverse-transcription reaction was ampli-
fied by using SYBR Green PCR master mix from Applied
Biosystems, using the ABI 7500 real-time PCR system. For each
gene, mRNA expression was calculated relative to 18S rRNA.
Amplification of specific transcripts was confirmed by melting
curve profiles at the end of each PCR. Primer sequences for each
gene are given in Table S1.
Microarray Analysis
Total RNA was isolated from the ASC of 5 lean subjects. The
quality and integrity of total RNA (RNA Integrity Number [RIN]
$8.0) was evaluated on an Agilent Bioanalyzer (Agilent Technol-
ogies, Waldbronn, Germany). RNA was then processed for
hybridization on Human Gene 1.0 ST Array chip (Affymetrix,
High Wycombe, UK), covering 28,869 well-annotated genes with
764,885 distinct probes, using standard Affymetrix protocols. Raw
signal intensities of gene expression data were processed, analyzed,
and linearly scaled using GeneChip Operating Software 1.1
(GCOS, Affymetrix) to a mean hybridization intensity of 500
units. For each array, GCOS output was imported as CEL files
into Partek Genomic Suite (Partek GS, Partek Inc., 2008, Revision
6.3) software, and data were normalized using the RMA (Robust
Multichip Averaging) algorithm. RNA hybridization intensity data
obtained from the array analysis were concordant for all
ASCBottom, and for 4 of the 5 ASCCeiling and ASCSVF from both
Sc and V fat depot. The remaining one ASC samples were
excluded from the array analysis, since their data were ambiguous
or incorrect, resulting in cell-mismatched design when compared
with the homologous Sc and V-ASC subsets. Similarities and
differences among gene expression profiles were assessed by
hierarchical clustering using Partek GS software. Statistical
significance was defined as being differentially expressed with an
adjusted p-value,0.01. The array data have been deposited at the
Gene Expression Omnibus archive, accession number GSE37324.
A further analysis to identify the biological mechanisms, pathways
and functions and the most relevant overexpressed genes was
performed using the Ingenuity Pathways Analysis (IPA) software
and database (Ingenuity Systems, http://www.ingenuity.com).
Briefly, the dataset containing gene identifiers and corresponding
fold changes was uploaded into the web-delivered application, and
each gene identifier was mapped to its corresponding gene object.
The threshold for a significant association was determined by the
p-value ,0.01, becoming significant any score above 3 [24]; [25].
For all analyses, Fisher’s exact test was used to calculate a p-value
and to determine a score for all networks that were ranked on the
probability that a collection of genes equal to or greater than the
number in a network could be achieved by chance alone.
Multiplex Analysis of Cytokine Production
ASC-derived conditioned medium (CM) was obtained by
incubating confluent cells in serum-free DMEM/Ham’s F12 for
16 h, after which time all CM was collected, pooled, cleared by
centrifugation, frozen in dry ice, and stored at 280uC. Concen-
trations of various cytokines in CM were analyzed using
a multiplex biometrix immunoassay from Bio-Rad, containing
fluorescent dyed microspheres conjugated with a monoclonal
antibody specific for the target proteins according to the
manufacturer’s instructions (Bio-Plex Human Cytokine Assay;
Bio-Rad, Munich, Germany). The following cytokines were
assayed: IL-1b, IL-1ra, IL-2, IL-4, IL-5, IL-6, IL-7, IL-8, IL-9,
IL-10, IL-12 (p70), IL-13, IL-15, IL-17, FGF basic, eotaxin, G-
CSF, GM-CSF, IFN-c, IP-10, MCP-1 (MCAF), MIP-1a, MIP-1b,
PDGF-BB, RANTES, TNF-a, and VEGF. The amount of protein
in the medium was normalized for 106 cells [26]. The inter- and
intra-assay coefficients of variation for all cytokines under
investigation were less than 10%.
Statistical Analysis
Data are presented as mean 6 SE. Comparisons between the
values were performed using a two-tailed Student’s t-test. For the
comparison of multiple groups, a one-way ANOVA test followed
by Fisher’s post-hoc test was applied. For all statistical analyses, the
level of significance was set at a probability of p,0.05. All
experiments were repeated at least 3 times. These analyses were
performed using SPSS 16 (SPSS, Chicago, IL).
Results
Phenotypic Analysis and Stem Cell Properties of ASC
Populations
ASCSVF, ASCCeiling and ASCBottom isolated from the Sc and V
fat depots grew as a characteristic cell monolayer in culture dishes
(Fig. 1A). To identify these cells as adipocyte progenitors and to
rule out potential contamination by other adipose tissue cell types,
analysis by qRT-PCR for a selection of known hematopoietic,
endothelial and stem cell-associated markers was carried out. The
expression of the key ASC markers CD105, CD49d, and CD44
was noticeably pronounced and expressed at similar magnitude in
ASCSVF, ASCCeiling and ASCBottom from both Sc and V fat depots
(Fig. 1B; Table S2). The initial Sc-ASCSVF and V-ASCSVF cultures
contained a subset of cells that were also positive for the
endothelial and hematopoietic lineage markers CD11b, CD45,
and CD31 (Fig. 2A). However, with successive passages, the
expression of CD11b, CD45, and CD31 declined significantly,
becoming negligible by passage 4 (p,0.001 vs. passage 0; Fig. 2A),
whereas expression levels of ASC markers remained constant
(p=0.860 vs. passage 0; Fig. 2B).
To confirm the multipotent mesenchymal stem cell properties of
the distinct ASC populations, a comparative in vitro tri-lineage
differentiation assay was carried out by exposing culture-expanded
ASC derived from Sc and V adipose tissue to osteogenic,
chondrogenic, or adipogenic differentiation media, respectively.
The qualitative histochemical evaluation of functional outcomes
indicated that ASCSVF, ASCCeiling and ASCBottom from both
adipose tissue depots underwent dif