ORIGINAL PAPER
All-trans retinoic acid is an effective inhibitor of hyaluronatesynthesis in a human dermal equivalent
Madhura Deshpande • Suzanne Papp •
Lana Schaffer • Tara Pouyani
Received: 10 October 2013 / Revised: 3 February 2014 / Accepted: 18 February 2014
� Springer-Verlag Berlin Heidelberg 2014
Abstract All-trans retinoic acid (ATRA) is known to
have beneficial effects on skin. It has been used extensively
for the treatment of photodamaged skin. To assess the
effects of all-trans retinoic acid on the dermis, specifically
its effect on hyaluronate (hyaluronic acid, HA) synthesis
and inhibition, tissue-engineered human dermal equiva-
lents were prepared in the presence of varying concentra-
tions of ATRA using the method of ‘‘self-assembly’’. A
substantial extracellular matrix was formed at the end of
the culture period. The extracellular matrices of these
dermal constructs were characterized and compared to the
construct prepared in the absence of ATRA (Normal).
Inhibition of hyaluronate was observed in constructs pre-
pared in the presence of varying concentrations of all-trans
retinoic acid. Chondroitin sulfate synthesis was unaffected
up to 1 lM ATRA. Collagen synthesis was enhanced at
lower concentrations of ATRA (250 and 500 nM) and
inhibited at higher concentrations of ATRA. Differential
gene array experiments were performed comparing the
construct grown in the presence of 500 nM ATRA to one
grown in the absence of ATRA, to obtain preliminary
information regarding the gene(s) involved in HA inhibi-
tion using a GLYCOv4 gene chip. These preliminary
experiments demonstrated the differential expression of
127 genes and suggest that down-regulation of five key
enzymes in the HA biosynthetic pathway may be involved
in this inhibitory process.
Keywords Dermis � All-trans retinoic acid � Hyaluronic
acid � Tissue engineering � Gene profiling
Introduction
The effects of all-trans retinoic acid (ATRA) on skin have
been the subject of considerable interest because of its
reported ability to reverse the aging process [15]. In
addition to the normal process of aging, sun exposure of
skin also adds to the structural and functional changes that
the dermis and epidermis undergo over time [3]. The
effects of ultraviolet light (UV) on the dermal and epi-
dermal portions of skin are well documented in the litera-
ture [14, 31]. ATRA has been shown to have beneficial
effects on photodamaged skin as well, and can to some
degree reverse the deleterious effects of UV light on skin
structures [9, 37]. In vitro studies on skin organ cultures
have supported these observations. Organ cultures treated
with ATRA showed a propensity for long-term survival
unlike their untreated counterparts. Moreover, the dermis
in these ATRA-treated cultures was histologically well
preserved and protein and ECM synthesis was enhanced as
compared to controls [35]. Organ cultures from sun-
exposed skin, which showed features of photodamage,
characterized by destruction of normal connective tissue
elements, treated with ATRA showed signs of reversal
[36]. A prominent zone of healthy connective tissue below
the dermal/epidermal junction was seen in ATRA-treated
cultures as compared to controls. These results support the
hypothesis that ATRA treatment can potentially enhance
the repair of sun-damaged skin.
M. Deshpande � T. Pouyani (&)
Department of Pharmaceutical Sciences, Bouve College
of Health Sciences, Northeastern University, Boston,
MA 02115, USA
e-mail: [email protected]
S. Papp � L. Schaffer
DNA Array Core Facility, The Scripps Research Institute,
10550 N Torrey Pines Rd, La Jolla, CA 92037, USA
123
Arch Dermatol Res
DOI 10.1007/s00403-014-1460-z
Studies on the effects of ATRA on monolayer cultures
of human dermal fibroblasts have also provided valuable,
though in some cases, contradictory information. Some
reports have shown fibroblast proliferation in the pre-
sence of ATRA and others inhibition [17, 24]. A similar
situation exists in regard to collagen synthesis in the
presence or absence of ATRA [10, 27]. A report has
concluded that the ability of ATRA to stimulate fibro-
blast proliferation depends largely on the proliferative
state of the cells at the time of treatment with ATRA
[34]. The same report concludes that stimulation of ECM
production by ATRA occurs over a very narrow range of
concentrations. Hyaluronate inhibition has also been
reported in confluent human dermal fibroblast monolayer
cultures treated with ATRA; however, the exact mech-
anism of HA inhibition by ATRA was unclear. Pre-
liminary experiments suggested that HA synthesis was
disrupted, and that degradation/catabolism of HA was
not responsible for the diminished quantities observed
[32].
To assess the effects of ATRA on the dermis, human
dermal matrices were prepared in the presence of varying
concentrations of ATRA under completely serum-free
conditions and in the absence of a three-dimensional
scaffold using the technique of ‘‘self-assembly’’. Human
dermal matrix is a self-assembled dermal equivalent with
a well-characterized extracellular matrix that bears strong
resemblance to the human dermis and contains high levels
of the glycosaminoglycan (GAG), hyaluronate [28, 29].
The lack of serum in the growth medium during matu-
ration allows for the addition of exogenous compounds to
directly assess their effects on ECM assembly and com-
position [30]. In this study, we chose the following con-
centrations of ATRA, 250 nM, 500 nM, 1 lM, 5 lM and
10 lM. Human dermal matrices were prepared in the
presence of these concentrations of ATRA and compared
to the base construct (Normal), prepared in medium that
does not contain ATRA. Characterization of the human
dermal matrices prepared in the presence of varying
concentrations of ATRA demonstrated the formation of a
substantial extracellular matrix after 21 days of culture.
Inhibition of hyaluronate was observed at all concentra-
tions of ATRA, whereas chondroitin sulfate synthesis was
unaffected at the lower and intermediate ATRA concen-
trations. To shed light on the genes involved in HA
inhibition, we chose the human dermal matrix cultured in
the presence of 500 nM ATRA and conducted differential
gene array experiments, in which it was compared to the
Normal construct. In these preliminary experiments, dif-
ferential expression of 127 genes was observed; of these,
49 genes were down-regulated and 78 genes were up-
regulated.
Materials and methods
Production of human dermal equivalents
Neonatal human dermal fibroblasts (ATCC) (Manassas,
VA) (passage 7) were harvested at confluence and seeded
at high density (3 million cells/4.52 cm2) on a porous
polycarbonate membrane in a transwell format (Corning,
Big Flats, NY) and periodically fed for 21 days with serum-
free medium. Specifically, the formulation contains the
following ingredients: a base 3:1 mixture of Dulbecco’s
Modified Eagle’s Medium (high-glucose formulation,
without L-glutamine) and Hams F-12 (Sigma-aldrich, St.
Louis, MI) medium supplemented with 4 mM L-glutamax
(Invitrogen, Carlsbad, CA), 5 ng/mL epidermal growth
factor (Millipore, Billerica, MA), 0.4 lg/mL hydrocorti-
sone (Sigma-aldrich, St. Louis, MI), 1 9 10-4 M etha-
nolamine (Sigma-aldrich, St. Louis, MI), 1 9 10-4 M o-
phosphoryl-ethanolamine (Sigma-aldrich, St. Louis, MI),
5 lg/mL insulin (Sigma-aldrich, St. Louis, MI), 5 lg/mL
transferrin (Sigma-aldrich, St. Louis, MI), 20 pM triiodo-
thyronine, 6.78 ng/mL selenium, 50 lg/mL L-ascorbic
acid-2-phosphate (Wako Chemicals, Richmond, VA),
0.2 lg/mL L-proline and 0.1 lg/mL glycine. The lower
chamber was fed 3 mL and the upper chamber 2 mL of
medium throughout the 21-day culture period with medium
changes every 2–3 days. ATRA in varying concentrations
was added to the medium using 95 % ethanol as a co-
solvent.
Cell count procedure
Human dermal matrices were digested with collagenase IV
(Sigma-aldrich, St. Louis, MI) for 1 h at 37 �C in colla-
genase premix (120 mL of PBS, 14 mL of 2.5 % trypsin
and 1 mL 0.45 % glucose). Serum containing media was
added to the digest to stop further enzymatic activity. Cells
were diluted 1:1 with trypan blue (Sigma-aldrich, St. Louis,
MI) and counted using a hemocytometer.
Determination of total collagen
Total collagen was quantified using the Hydroxyproline
Assay. In brief, the constructs were placed in 6 N hydro-
chloric acid (HCl) in a sealed glass tube and hydrolyzed for
18 h at 110 �C. At this point, all the collagen and proteins
present were degraded to the component amino acids. After
completion of hydrolysis, the hydrolysates were partially
neutralized with 5 N NaOH and centrifuged to sediment
particulate debris. The supernatants were transferred to
fresh tubes for further analysis. The standards, blanks and
unknowns (with appropriate dilutions) were pipetted in
Arch Dermatol Res
123
triplicate in 50 lL aliquots onto a 96-well plate. Oxidizing
solution (100 lL) containing chloramine T was added to
each well except those designated as blanks. To those
wells, oxidizing buffer solution w/o chloramine T was
added. The plate was gently shaken (on a laboratory rota-
tor/shaker) for 5 min at room temperature. Ehrlich’s
reagent (100 lL) was added to each well with thorough
mixing. The plate was covered with an adhesive plate seal
and incubated in a water bath or oven at 60 �C for 45 min.
The absorbance was then read at 570 nm. The hydroxy-
proline values (expressed in lg/mL) were determined
directly from the standard curve. To convert the hydroxy-
proline values to the total amount of collagen present in the
constructs, a conversion factor of 7.46 was used. The final
data are expressed as lg collagen/mg wet weight of tissue
or lg collagen/construct.
Hyaluronic acid inhibition ELISA assay
Human dermal matrix samples were digested with 0.5 mg/
mL Proteinase K (Invitrogen, Carlsbad, CA) overnight at
60 �C. The digested mixture was boiled for 10 min to
inactivate the protease, spun down and the supernatant
transferred to a fresh tube. ELISA plates (Immulon4) were
prepared by coating overnight with a 100 lg/mL solution
of sodium hyaluronate (Genzyme Corporation, Cambridge,
MA) in sodium bicarbonate buffer, pH 9.2. The plate was
then washed with 0.1 % BSA in PBS and blocked with 5 %
BSA for 6 h. Samples to be tested were appropriately
diluted with 1 % BSA and 100 lL of each were added to
individual wells. A solution of biotinylated HA binding
protein (Associates of Cape Cod, East Falmouth, MA)
(1:1,000 dilution) was added to the samples and allowed to
equilibrate at 4 �C overnight. The wells were washed with
0.05 % Tween/PBS and avidin horseradish peroxidase
(Vector Laboratories, Burlingame, CA) (1:5,000) was
added to each well and incubated at 37 �C for 1 h. The
plates were washed and the substrate ABTS (Kirkegaard
and Perry Laboratories, Gaithersburg, MD) was added to
each well and the color was allowed to develop for 10 min.
The reaction was stopped by the addition of 1 % SDS to
each well. The plates were read in a microtitre plate reader
at 410 nm.
GAG disaccharide analysis by FACE
Proteinase K digested samples of human dermal matrices
were initially digested with 20 U/mL of hyaluronidase SD
(Associates of Cape Cod, East Falmouth, MA) in 100 mM
ammonium acetate, pH 7.00 for 2 h at 37 �C followed by
chondroitinase ABC (Associates of Cape Cod, East Fal-
mouth, MA) digestion for 3 h. The digests and disaccha-
ride standards (Associates of Cape Cod, East Falmouth,
MA) were subjected to reductive amination and conjuga-
tion with AMAC (2-aminoacridone) (Invitrogen, Carlsbad,
CA) to produce fluorescent GAG disaccharides. The
labeled GAGs were subjected to polyacrylamide gel
electrophoresis and imaged with a ChemiDoc XRS imager
(BioRad, Hercules, CA) to obtain the demonstrated pro-
files. For quantification, AMAC-labeled glucose (Sigma-
aldrich, St. Louis, MI) was used as a standard, and
Quantity One software was used to determine various
GAG values.
Histological processing
Samples to be analyzed were fixed overnight in 10 %
phosphate-buffered formalin (Thermofisher Scientific,
Pittsburgh, PA). They were subsequently transferred to
vials containing 70 % ethanol prior to embedding and
processing. Samples were embedded into paraffin blocks,
sectioned into six micrometer sections and stained with
hematoxylin and eosin (H&E), imaged and photographed
at 4009 using a Westover Micromaster I microscope
equipped with a digital camera.
Transmission electron microscopy (TEM)
Human dermal matrices (4.52 cm2) were fixed overnight at
4 �C in 2.5 % glutaraldehyde in 0.1 M sodium cacodylate
buffer pH 7.4 (Electron Microscopy Sciences, Hatfield,
PA), washed in 0.1 M cacodylate buffer and postfixed with
1 % osmium tetroxide (OsO4)/1.5 % potassium ferrocya-
nide K4[Fe(CN)6] for 1 h, washed in water three times and
incubated in 1 % aqueous uranyl acetate for 1 h followed
by two washes in water and subsequent dehydration in
grades of alcohol (10 min each; 50, 70, 90 %, 2 9 10 min
100 %). The samples were then put in propylene oxide for
1 h and infiltrated overnight in a 1:1 mixture of propylene
oxide and TAAB Epon (Marivac Canada Inc. St. Laurent,
Canada). The following day, the samples were embedded
in TAAB Epon and polymerized at 60 �C for 48 h.
Ultrathin sections (about 60 nm) were cut on a Reichert
Ultracut-S microtome, picked up on to copper grids stained
with lead citrate and examined in a JEOL 1200EX
Transmission electron microscope or a TecnaiG2 Spirit
BioTWIN. Images were recorded with an AMT 2k CCD
camera.
Measurement of tissue thickness
Thickness of H&E stained cross sections of the human
dermal matrices were measured by digital image analysis
using Micron imaging software. The reported thickness is a
mean of 10 measurements ± SEM.
Arch Dermatol Res
123
Statistical analyses
Statistical analysis of the data was done using a one-way
ANOVA with equal variances assumed. Multiple compar-
isons were done by the Bonferroni method. Statistical
Package for Social Sciences software was used for con-
ducting the one-way ANOVA significance tests. The sig-
nificance level (a) used was 0.05. Probability values of
p \ 0.05 were considered statistically significant. Experi-
ments were repeated three times, and data are expressed as
mean ± standard error (SEM) unless otherwise indicated.
Total RNA extraction
Total RNA was extracted from human dermal matrix using
the Qiagen RNeasy� Fibrous Tissue Midi Kit (Qiagen,
Valencia, CA). Dermal matrices were placed in a 15-mL
centrifuge tube with 3 mL RLT buffer (provided in the kit)
and 10 lL b-mercaptoethanol (Sigma-aldrich, St. Louis,
MI) and homogenized using a TissueRuptor homogenizer
(Qiagen, Valencia, CA) for 1 min. The homogenized tissue
was treated with Proteinase K for 15 min at 55 �C and
subjected to centrifugation at 3,5009g for 5 min at
20–25 �C. The supernatant was transferred to a new 15-mL
centrifuge tube and 0.5 volume of ethanol (Sigma-aldrich,
St. Louis, MI) was added to the lysate. The sample was
then transferred to an RNeasy Midi spin column placed in a
15-mL collection tube, 3 mL at a time, and each time
centrifuged for 5 min at 3,5009g. The flow through was
discarded. The column membrane was washed with RW1
wash buffer and centrifuged. DNAse digestion was carried
out by addition of 160 lL of DNase I incubation mix
(20 lL DNase I stock solution in 140 lL RDD buffer) to
the spin column membrane for 18 min at room tempera-
ture. The column was washed with RW1 buffer and cen-
trifuged for 5 min at 3,5009g. Two other washes were also
carried out with RPE buffer. The RNA was eluted by
adding 150 lL RNAse-free water twice directly to the spin
column membrane and centrifuged each time for 3 min at
3,5009g. The RNA samples were monitored with the 2100
Agilent Bioanalyzer, consistently demonstrating high-
quality RNA as evidenced by the 28S/18S ratio of
approximately 2.
Gene array analysis
The GlycoV4 oligonucleotide array is a custom Affymetrix
GeneChip (Affymetrix, Santa Clara, CA) designed for the
consortium for Functional Glycomics (http://www.functio
nalglycomics.org/static/consortium/consortium.shtml). A
complete description and annotation for the GlycoV4 array
is available at http://www.functionalglycomics.org/static/
consortium/resources.shtml
The GlycoV4 focused array includes probes for *1,260
human probe-ids and *1,200 mouse probe-ids related to
glycogenes. This array does not contain mismatched
probes.
Data normalization was performed using RMA Express
1.0 with quantile normalization, median polish and back-
ground adjustment [4, 18].
The Limma package in the R software was used to find
transcripts with differential expression [2, 33]. The fold
changes and SEMs were estimated by fitting a linear
model for each gene and empirical Bayes smoothing was
applied to the SEMs. Results are presented between two
or more experimental conditions as a fold change in
expression level, the moderated t-statistic, the p value and
the adjusted p value. The adjusted p value is the p value
adjusted for multiple testing using Benjamini and Hoch-
berg’s method. The transcripts identified as differentially
expressed were those with adjusted p value \0.15 and
fold change [1.4.
The heatmap was generated with dChip program (http://
www.dChip.org). Red indicates increased and blue indi-
cates decreased expression relative to the mean transcript
expression value.
Results
Preparation of human dermal matrices and histological
analysis
Neonatal human dermal fibroblasts were seeded on a
polycarbonate membrane in a transwell format in a state
of ‘‘super-confluence’’. After 21 days of periodic feeding
with serum-free media containing varying concentrations
of ATRA (250 nM, 500 nM, 1 lM, 5 lM, 10 lM), a
substantial extracellular matrix formed that was peeled
off the membrane for analysis, and compared to the
construct grown in the absence of ATRA (Normal). The
human dermal matrices were fixed in formalin, embedded
in paraffin and subjected to H&E staining. Analysis of
these H&E sections demonstrated a consistent pattern of
‘‘thinning’’ with increasing ATRA concentrations (Fig. 1).
The Normal construct grown in the absence of ATRA
had a thickness of 283 ± 7.92 lm. The presence of
250 nM ATRA gave rise to a dermal construct with a
thickness of 237 ± 3.77 lm. Increasing the ATRA con-
centration to 500 nM resulted in a thinner construct with
a measured thickness of 196 ± 4.38 lm. The human
dermal matrix grown in the presence of 1 lM ATRA had
a thickness of 134 ± 4.83 lm. The final concentrations
of 5 and 10 lM yielded even thinner human dermal
matrices with a thickness of 60 ± 4.78 and
31 ± 1.93 lm, respectively.
Arch Dermatol Res
123
Hyaluronic acid, collagen and cell count analysis
The human dermal matrices were next subjected to ECM
analysis. Measurement of the hyaluronate content of the
human dermal matrices was accomplished by first
digesting the tissue constructs with proteinase K to lib-
erate the GAGs. An inhibition ELISA was used to mea-
sure the HA content of the proteinase K digests [13]. The
results of these experiments demonstrate a successive
reduction in the amount of hyaluronate with increasing
ATRA concentration (Fig. 2a). Thus, the Normal con-
struct was observed to have an HA content of
233 ± 13.6 lg and the 250 nM ATRA construct an HA
content of 106 ± 3.79 lg. A higher level of HA inhibi-
tion was observed with the 500 nM ATRA construct with
an HA content of 70.3 ± 4.26 lg. Successively higher
concentrations of ATRA produced human dermal matri-
ces with even lower HA content; the 1 lM ATRA
Fig. 1 H&E stained sections of
human dermal matrices grown
in the presence of varying
concentrations of ATRA. The
Normal construct is grown in
the absence of ATRA; other
concentrations are shown on the
left of the sections. The
thickness of the constructs is
shown on the right of the H&E
sections. *p value \0.05 for
comparison with Normal
construct thickness using one-
way ANOVA at a = 0.05
Arch Dermatol Res
123
construct contained 28.3 ± 2.91 lg HA and the 5 lM
ATRA construct 6.27 ± 2.13 lg HA. The 10 lM ATRA
human dermal matrix showed the highest level of HA
inhibition with a measured content of 3.26 ± 1.16 lg.
Hyaluronate content of the human dermal matrices was
also expressed per cell count to take into account the
reduction of cell numbers with increasing concentrations
of ATRA (Fig. 2b).
The total collagen content of the human dermal matrices
was measured using the hydroxyproline assay [6]. The
Normal construct contained 1.44 ± 0.033 mg collagen. An
enhancement of collagen production was observed for the
human dermal matrices cultured in the presence of 250 and
500 nM ATRA (Fig. 2c). These constructs had a total
collagen content of 1.72 ± 0.025 and 1.67 ± 0.038 mg,
respectively. Higher concentrations of ATRA in the culture
medium of the human dermal matrices resulted in inhibi-
tion of collagen synthesis as compared to the Normal
construct. Human dermal matrix cultured in the presence of
1 lM ATRA had a total measured collagen content of
1.18 ± 0.04 mg. Higher levels of inhibition of collagen
synthesis was observed in the human dermal matrices
cultured in the presence of 5 and 10 lM ATRA. The total
collagen content for these constructs was measured at
0
50
100
150
200
250
300
ug
HA
/4.5
2 cm
2
Hyaluronic Acid
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
0
1
2
3
4
5
6
7
8
Mill
ion
cel
ls/4
.52
cm2
Cell Count
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
0
0.5
1
1.5
2
mg
co
llag
en/4
.52
cm2
Collagen
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
0
5
10
15
20
25
30
35
40
µg
HA
/Cel
l Co
un
t
Hyaluronic Acid
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
mg
Co
llag
en/C
ell C
ou
nt
Collagen
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
*
*
A B
C D
E
**
** *
*
*
**
**
**
**
** *
Fig. 2 Profile of collagen content, hyaluronic acid content and cell
replication for the self-assembled human dermal matrices. a All
concentrations of ATRA tested inhibit HA synthesis during the
21-day culture period. b Hyaluronate content of human dermal
matrices cultured in the presence of varying amounts of ATRA
normalized to cell count. c Human dermal matrices (250 and 500 nM
ATRA) produce enhanced amount of total collagen as compared to
the Normal construct (no ATRA). At higher concentrations of ATRA
inhibition of collagen synthesis is observed. d Collagen content of
human dermal matrices normalized to cell count. e Human dermal
matrices (Normal and 250 nM ATRA) undergo approximately one
population doubling over the 3-week culture period. Inhibition of cell
replication is observed beginning with the 500 nM construct and was
enhanced in the presence of higher concentrations of ATRA (1, 5 and
10 lM). In the 10 lM ATRA human dermal matrix, approximately
equal numbers of cells were observed at the end of the 21-day culture
compared to the initial seeding density. All measurements are an
average of three independent measurements ± SEM, n = 3. Each
error bar represents one SEM. *p value \0.05 compared to Normal
construct using one-way ANOVA at a = 0.05
Arch Dermatol Res
123
0.799 ± 0.01 and 0.487 ± 0.096 mg, respectively. To
account for reduction in cell number with increasing
ATRA concentration, collagen content of the human der-
mal matrices was also expressed per cell count (Fig. 2d).
The neonatal dermal fibroblasts in the Normal human
dermal matrix typically undergo one population doubling
during the 21-day culture period to give a final cell count of
6.58 ± 0.116 cells per construct [28]. A slight increase in
cell replication was observed in the construct grown in the
presence of 250 nM ATRA at 6.82 ± 0.038 cells per
construct (Fig. 2e). At higher concentrations of ATRA,
inhibition of cell replication was observed; the 500 nM
human dermal matrix had a final cell count of 5.63 ± 0.53
cells per construct. Increasing the concentration of ATRA
to 1 lM resulted in a human dermal matrix with a final cell
count of 4.91 ± 0.116 cells per construct. The two human
dermal matrices cultured in the presence of 5 and 10 lM
ATRA showed even higher levels of cell replication inhi-
bition at 3.93 ± 0.02 cells per construct and 2.94 ± 0.219
cells per construct, respectively.
Ultrastructural analysis of human dermal matrices
Ultrastructural analysis of the human dermal matrices was
accomplished by TEM. Analysis of the collagen fibril
diameters indicates that culturing the human dermal
matrices in the presence of ATRA had no effect on colla-
gen fibril diameter (Fig. 3a). The Normal construct was
observed to have a fibril diameter of 48.3 ± 6.26 nm.
Comparison of the human dermal matrices cultured in the
presence of varying concentrations of ATRA to the Normal
construct showed no significant difference in collagen fibril
diameter (Fig. 3a). A study of the statistical distribution of
the collagen fibril diameters showed a relatively tight dis-
tribution for all human dermal matrices cultured in the
presence of ATRA, when compared to the Normal con-
struct (Fig. 3b). This observation suggests that collagen
fibrillogenesis was normal. The longitudinal collagen fibers
were analyzed at high magnification (Fig. 3c). This ana-
lysis demonstrated that all human dermal matrices cultured
in the presence of varying concentrations of ATRA showed
the appropriate quarter staggered 67-nm periodicity char-
acteristic of collagen fibers in human skin [12]. Electron
micrographs of human dermal matrices grown in the pre-
sence of ATRA at lower magnification showed the pre-
sence of fibroblasts dispersed in a dense collagenous matrix
analogous to that seen in the Normal construct (Fig. 3d).
GAG profiling by fluorophore-assisted carbohydrate
electrophoresis (FACE)
To assess the effects of ATRA on GAG synthesis the
human dermal matrices grown in the presence of ATRA
were subjected to disaccharide analysis by fluorophore-
activated carbohydrate electrophoresis (FACE) [7, 8].
Human dermal matrices were digested with proteinase K to
liberate the GAGs; the digests were then further treated
with hyaluronidase and chondroitinase ABC (CABC) to
Fibril Diameter: 48.3 ± 6.26 nmRange: 36 – 64 nm, n=200
dicAcioniteRTAMn052lamroN
500 nM AT Retinoic Acid 1 µM AT Retinoic Acid
5 µM AT Retinoic Acid 10 µM AT Retinoic Acid
Fibril Diameter: 49.1 ± 5.68 nmRange: 39 – 68 nm, n=200
Fibril Diameter: 48.5 ± 5.10 nmRange: 37 – 62 nm, n=200
Fibril Diameter: 48.2 ± 6.75 nmRange: 34 – 73 nm, n=200
Fibril Diameter: 47.9 ± 5.67 nmRange: 33 – 60 nm, n=200
Fibril Diameter: 47.9 ± 6.24 nmRange: 31 – 62 nm, n=200
mn001mn001
mn001mn001
mn001mn001
A
Fig. 3 Ultrastructural analysis of human dermal matrices grown in
the presence of varying concentrations of ATRA. a TEM images of
cross sections of collagen fibrils of human dermal matrices. The fibril
diameters of human dermal matrices prepared in the presence of the
indicated amount of ATRA are similar to fibril diameters observed in
the Normal construct cultured in the absence of ATRA. Fibril
diameters are reported as mean ± SD, n = 200. b Histogram dem-
onstrating frequency and distribution of fibril diameters in the human
dermal matrices. A relatively narrow distribution suggests normal
collagen fibrillogenesis. A similar narrow distribution was observed
for all human dermal matrices. c TEM images demonstrating the
organization of collagen fibers in human dermal matrices. Banded
collagen fibers are seen in human dermal matrices demonstrating the
67-nm periodicity characteristic of fibers observed in human skin.
d TEMs of human dermal matrices at low magnification showing
fibroblasts dispersed in a dense collagenous matrix
Arch Dermatol Res
123
obtain the component disaccharides. The disaccharides
were then subjected to reductive amination, labeled with
2-aminoacridone (AMAC) and run on a high percentage
polyacrylamide gel (Fig. 4a). Imaging of the gel allowed
for the quantification of the disaccharides using glucose as
a standard (Fig. 4b). The fundamental GAG profile of the
Normal 250 nM AT Retinoic Acid
500 nM AT Retinoic Acid 1 µM AT Retinoic Acid
5 µM AT Retinoic Acid 10 µM AT Retinoic Acid
100 nm 100 nm
100 nm 100 nm
100 nm 100 nm
Normal 250 nM AT Retinoic Acid
500 nM AT Retinoic Acid 1 µM AT Retinoic Acid
5 µM AT Retinoic Acid 10 µM AT Retinoic Acid
500 nm 500 nm
500 nm 500 nm
500 nm 500 nm
B
DC
0
5
10
15
20
25
30
28 32 36 40 44 48 52 56 60 64 68 72
Fre
qu
ency
Fibril Diameter (nm)
Normal
250 nM ATRA
500 nM ATRA
1 uM ATRA
5 uM ATRA
10 uM ATRA
Fig. 3 continued
Arch Dermatol Res
123
human dermal matrices did not change with increasing
concentrations of ATRA. The DDi-HA (unsaturated hyal-
uro disaccharide) content showed a systematic decrease
with increasing ATRA concentrations similar to that seen
with total HA content. The DDi-4S, DDi-6S and DDi-4,6S
(unsaturated chondro disaccharides) showed no significant
change with increasing concentrations of ATRA up to
1 lM as compared to the Normal construct. At higher
concentrations of ATRA (5 and 10 lM) in the culture
medium, this specificity was lost and a decrease in these
disaccharides was observed as well (Fig. 4b).
Differential gene array profiling of human dermal
matrices
Having established that ATRA was an effective inhibitor of
HA we sought to design preliminary experiments to iden-
tify the mechanism by which this inhibition occurred. To
accomplish this objective, we chose to conduct a differ-
ential gene array analysis of the human dermal matrix
cultured in the presence of 500 nM ATRA and the Normal
construct, in which no ATRA was present in the culture
medium. To accomplish this, we used cDNA microarray
containing approximately 1,260 human sequences (Gly-
coV4 oligonucleotide array). The results from these
experiments demonstrated the differential expression of
127 genes (Fig. 5); of these, 49 genes were down-regulated
and 78 genes were up-regulated (Table 1). Analysis of the
down-regulated genes showed the presence of five genes
that are known participants in the HA biosynthetic pathway
(Fig. 6) [16]. These genes were identified as glucose
phosphate isomerase (GPI), glutamine-fructose-6-phos-
phate transaminase 1 (GFAT, GFPT1), UDP-N-acetyl-
glucosamine pyrophosphorylase (UAP1), UDP-glucose
dehydrogenase (UDP-GlcDH) and hyaluronate synthase 2
(HAS2).
Discussion
Neonatal human dermal fibroblasts cultured in a serum-free
medium in the presence of varying concentrations of
Fig. 4 GAG profiling of
various human dermal matrices
by FACE. a Image of gel
demonstrating GAG profile of
human dermal matrices grown
in the presence of varying
amounts of ATRA compared
with available disaccharide
GAG standards. b Tabulation of
disaccharide content in human
dermal matrices. DDi-HA
denotes an unsaturated hyaluro
disaccharide; the following are
unsaturated chondro
disaccharides, the numbers
indicate the position of the
sulfate group on the N-acetyl
galactosamine moiety, DDi-OS
(unsulfated), DDi 6S, DDi-4S,
DDi2S, DDi-4,6S, DDi-2,6S,
DDi-2,4S. *p value \0.05 for
comparison with GAG
disaccharide levels of Normal
construct using one-way
ANOVA at a = 0.05
Arch Dermatol Res
123
ATRA formed a substantial extracellular matrix and were
harvested at the end of the 21-day culture period for ana-
lysis. A study of the H&E stained section of the various
human dermal matrices showed that with higher concen-
trations of ATRA thinner constructs were formed. This
pattern of dermal thinning shows an interesting correlation
with hyaluronic acid content. Thus, higher concentrations
of ATRA which induce higher levels of hyaluronate inhi-
bition also appear to result in constructs with reduced
thickness. This phenomenon is not unexpected, as hyalu-
ronate is known to be a space-filling molecule that can
absorb large amounts of water and is responsible for tissue
hydration [20]. Dermal thinning with loss of HA has been
previously observed when self-assembled human dermal
matrices were grown in the presence of increasing con-
centrations of triiodothyronine (T3) [30]. Skin fragility has
been observed in patients treated with oral and systemic
retinoids and may be a result of altered dermal connective
tissue [5, 38]. Although it has not been proven that loss of
HA is responsible for this observed phenomenon, we
speculate that it may contribute to skin fragility and pos-
sible alteration of dermal integrity. Total collagen content
of the human dermal matrices was significantly increased
in constructs grown in the presence of 250 and 500 nM
ATRA. The 500 nM construct showed the highest level of
collagen synthesis per cell count. This is consistent with
previous reports in the literature suggesting that ATRA is
able to enhance dermal collagen synthesis but only within a
very narrow range of concentrations [34]. The observed
enhanced collagen synthesis may be responsible for the
repair of photodamaged skin observed with ATRA. Higher
concentrations of ATRA (1, 5 and 10 lM) resulted in
successively higher levels of inhibition of total collagen
content. When the collagen content was expressed per cell
count, inhibition was observed only in the 5 and 10 lM
ATRA human dermal matrices. This inhibition of total
collagen synthesis may to some degree contribute to
reduction of human dermal matrix thickness with increas-
ing concentrations of ATRA. We observed no significant
enhancement of cell replication even at the lowest con-
centration of 250 nM ATRA. Higher concentrations of
ATRA resulted in successively higher levels of inhibition
of cell replication. In the human dermal matrix grown in
the presence of 10 lM ATRA we observed similar number
of cells at the end of the 21-day culture period as compared
to the initial seeding density of three million cells.
Ultrastructural analysis of the human dermal matrices
cultured in the presence of varying amounts of all-trans
retinoic acid by TEM showed no appreciable difference
when compared to the Normal construct cultured in the
absence of ATRA. Even at high concentrations of ATRA
collagen fibril diameters remained constant. The relatively
Fig. 5 A heatmap
representation of genes
differentially expressed in
human dermal matrices (Normal
and 500 nM ATRA). Red
indicates increased and blue
indicates decreased expression
relative to the mean transcript
expression value
Arch Dermatol Res
123
Table 1 Regulated genes in the 500 nM ATRA human dermal matrix compared to the Normal construct (no ATRA)
Gene symbol Gene name Fold
change
p value Adj. p value
SFRP1 Secreted frizzled-related protein 1 13.03 3.54E-09 5.85E-07
COLEC12 Scavenger receptor with CTLD 10.60 1.92E-10 1.18E-07
CCL11 CCL11 9.65 1.05E-06 0.000066
CXCL14 CXCL14 8.61 5.63E-07 0.0000415
BMP4 Bone morphogenetic protein 4 preproprotein 8.16 5.31E-11 6.69E-08
CXCL12 Stromal cell-derived factor 1 7.13 2.45E-08 0.00000309
SULF2 Sulfatase 2 6.28 1.25E-06 0.0000713
CCL2 Small inducible cytokine A2 5.75 0.000108 0.00315682
GALNT12 ppGalNAc T12; UDP-N-acetyl-a-D-galactosamine: polypeptide 5.45 3.72E-09 5.85E-07
FGF13 Fibroblast growth factor 13 5.43 2.26E-09 4.75E-07
SULF1 Sulfatase 1 5.13 3E-07 0.0000236
PDGFD Platelet-derived growth factor D 5.04 6.28E-10 1.98E-07
CCL13 CCL13 4.12 2.22E-05 0.00084576
LIFR Leukemia inhibitory factor receptor precursor 3.82 1.28E-08 0.00000179
WNT11 Wingless-type MMTV integration site 11 3.13 7.86E-08 0.00000825
GFRA1 GDNF family receptor a 1 3.07 1.90E-09 4.75E-07
OLR1 Oxidized low-density lipoprotein (lectin-like) 3.02 2.06E-05 0.00083491
CXCL6 Chemokine (C-X-C motif) ligand 6 (granulocyte chemotactic protein 2) 2.78 9.94E-06 0.00043149
FGF10 Fibroblast growth factor 10 precursor 2.74 0.000631 0.01118139
CCL8 CCL8 2.72 8.78E-06 0.00039479
CXCL13 B cell chemoattractant 2.70 0.010349 0.08863915
SRGN Serglycin 2.64 0.002536 0.0316107
FST Follistatin isoform FST317 precursor (probe set may cross-hyb between isoforms) 2.52 1.39E-07 0.0000134
MCOLN2 Mucolipin 2 2.51 5.93E-05 0.00191507
IGFBP2 Insulin-like growth factor binding protein 2 2.49 0.000729 0.0125682
FUCA1 Fucosidase, a-L-1, tissue 2.40 7.28E-05 0.00218278
FZD4 Frizzled 4 2.40 0.000609 0.01102701
WNT2 Wingless-type MMTV integration site 2 2.38 7.51E-08 0.00000825
IL8 Interleukin 8 2.34 0.003207 0.03605245
CD302 CD302 2.33 1.94E-05 0.00081242
KITLG KIT ligand 2.27 4.46E-06 0.00020818
INHBB Inhibin b B subunit preproprotein 2.11 0.005914 0.05817112
IFNAR1 Interferon-a receptor 1 2.08 0.00251 0.0316023
FZD2 Frizzled 2 2.06 0.000478 0.00970436
ST6GALNAC3 ST6 (a-N-acetyl-neuraminyl-2,3-b-galactosyl-1,3)-N-acetylgalactosaminide a-2,6-sialyltransferase 3 1.99 1.17E-06 0.0000703
TGFB3 Transforming growth factor b 3 1.97 0.007217 0.06780472
PLA2R1 Phospholipase A2 receptor 1 1.95 0.000555 0.01075564
CXCL1 Melanoma growth stimulating activity, a 1.94 0.000126 0.00344537
TNFSF13B Tumor necrosis factor (ligand) superfamily 1.91 0.003932 0.04304522
PDGFRA Platelet-derived growth factor receptor a 1.91 0.004745 0.04896477
CD248 Endosialin 1.90 0.001589 0.02223014
LY75 LY75 1.89 5.93E-07 0.0000415
CHST14 Carbohydrate (N-acetylgalactosamine 4-0) sulfotransferase 14 1.87 0.000434 0.00895874
GPC3 Glypican 3 1.86 0.000613 0.01102701
FGFR1 Fibroblast growth factor receptor 1 1.83 6.18E-05 0.00194639
EXT1 HS copolymerase GAG Enzyme 1.83 2.37E-05 0.00087732
IL11RA Interleukin 11 receptor, a 1.82 0.000126 0.00344537
TPST1 Tyrosylprotein sulfotransferase 1 1.76 6.58E-05 0.00202157
EMCN Endomucin 1.76 3.89E-06 0.00019612
PEG3 Paternally expressed 3 1.75 1.85E-06 0.0000973
HES1 Hairy and enhancer of split 1 1.74 0.022027 0.14436611
B3GALNT1 b-1,3-N-acetylgalactosaminyltransferase 1 (globoside blood group) 1.74 0.000528 0.01039416
Arch Dermatol Res
123
Table 1 continued
Gene symbol Gene name Fold
change
p value Adj. p value
B3GALT6 UDP-Gal:bGal b 1,3-galactosyltransferase polypeptide 6 1.68 0.019228 0.13432054
WNT2B Wingless-type MMTV integration site 2B 1.67 0.000423 0.008881
HSPC159 HSPC-159 1.67 0.001745 0.02362093
TGFBR2 Transforming growth factor b receptor II 1.67 0.013137 0.10467801
LEPROT Leptin receptor gene-related protein 1.66 0.000286 0.00692579
GALNAC4S-
6ST
B cell RAG-associated protein 1.65 0.015132 0.11684757
LARGE LARGE 1.64 0.001506 0.02179076
SMO Smoothened 1.62 0.00266 0.03250833
FGF11 Fibroblast growth factor 11 1.61 0.00128 0.01918196
ICAM2 Intercellular adhesion molecule 2 precursor 1.60 0.000244 0.00614
ASAHL N-acylsphingosine amidohydrolase-like protein 1.58 0.001427 0.02110118
ACVR1B Activin A type IB receptor isoform c precursor 1.58 0.003546 0.03951214
PIGC Phosphatidylinositol glycan anchor biosynthesis, class C 1.57 0.000318 0.00754822
GPC4 Glypican 4 1.57 0.000375 0.00843647
GALNT3 Polypeptide N-acetylgalactosaminyltransferase 3 1.57 1.85E-06 0.0000973
FMOD Fibromodulin 1.56 0.011091 0.09278106
FPGT Fucose-1-phosphate guanyltransferase 1.55 0.017417 0.12823155
PDGFA Platelet-derived growth factor a 1.54 0.000991 0.01540526
GBA Glucosidase, b; acid 1.53 0.008247 0.07416403
GAL3ST4 Galactose-3-O-sulfotransferase 4 1.52 0.001652 0.02286248
CSGALNACT1 CSGalNAcT1/ChGalNAcT1; chondroitin b1 4 1.49 0.022475 0.14436611
SLC35E3 solute carrier family 35 member E2 1.44 0.006228 0.06078049
GLA Galactosidase, a 1.44 0.018195 0.13090067
MRC2 MRC2 (Endo 180) 1.43 0.000584 0.01102701
IFNAR2 Interferon-a/b receptor 2 1.41 0.002832 0.03335813
ALG1 ALG1 1.41 0.010457 0.08895611
SLC35A2 Solute carrier family 35 (UDP-galactose transporter), member A2 0.71 0.00035 0.00801546
GALNTL4 GALNT18 (UDP-N-acetyl-a-D-galactosamine: polypeptide N-acetylgalactosaminyltransferase 18
(GalNAc-T18)
0.70 0.003008 0.03506707
XYLT2 Xylosyltransferase II 0.70 0.009617 0.08292899
MUC1 Mucin 1, cell surface associated 0.69 0.007797 0.07113512
CXCR4 CXCR4 0.69 0.021251 0.1430779
CHST1 Carbohydrate (keratan sulfate Gal-6) sulfotransferase 1 0.67 0.003043 0.03515313
FGF5 Fibroblast growth factor 5 precursor 0.66 0.018875 0.13350129
ICAM1 Intercellular adhesion molecule 1 precursor 0.66 0.018788 0.13350129
IDS Iduronate 2-sulfatase (Hunter syndrome) 0.66 0.003199 0.03605245
FZD3 Frizzled 3 0.66 0.003657 0.04039235
GCNT1 Glucosaminyl (N-acetyl) transferase 1, core 2 (b-1,6-N-acetylglucosaminyltransferase) 0.66 0.0004 0.00866328
WNT6 Wingless-type MMTV integration site 6 0.65 0.008216 0.07416403
WBSCR17 UDP-N-acetyl-a-D-galactosamine:polypeptide N-acetylgalactosaminyltransferase 20 (GalNAc-T20) 0.65 0.01245 0.10207091
CX3CL1 CX3CL1 0.64 0.002835 0.03335813
SLC35E1 Solute carrier family 35 member E1 0.64 0.022949 0.14666549
VEGFA Vascular endothelial growth factor A 0.64 0.019085 0.13423296
MGEA5 Meningioma expressed antigen 5 (hyaluronidase) 0.64 0.012485 0.10207091
GALNT5 UDP-N-acetyl-a-D-galactosamine:polypeptide 0.63 0.002216 0.02872372
C1ORF103 Collectin K1 0.62 0.000599 0.01102701
XYLT1 Xylosyltransferase I 0.62 0.001231 0.01867336
ATRN Attractin 0.61 0.007771 0.07113512
JAG1 Jagged 1 precursor 0.60 0.001796 0.02405236
COG6 Component of oligomeric golgi complex 6 0.59 0.0009 0.01452486
UGDH UDP-glucose dehydrogenase 0.59 0.004885 0.05000109
Arch Dermatol Res
123
Table 1 continued
Gene symbol Gene name Fold
change
p value Adj. p value
GALNT14 ppGalNAc T14; UDP-N-acetyl-a-D-galactosamine:polypeptide 0.58 0.000782 0.013131
HGF Hepatocyte growth factor isoform 1 or 3 0.58 0.000181 0.00464006
PDGFC Platelet-derived growth factor C precursor 0.57 0.000113 0.00322765
PKD1 Polycystin 1 0.57 0.00858 0.07661476
CHPF Chondroitin polymerizing factor 0.56 0.002236 0.02872372
JUNB Jun B proto-oncogene 0.56 0.000391 0.00864426
RPL27A Ribosomal protein L27a 0.55 0.019872 0.13746717
GPI Glucose phosphate isomerase 0.55 0.000972 0.01530317
GFPT1 Glutamine-fructose-6-phosphate transaminase 1 0.55 0.004581 0.04813459
GALNT9 Polypeptide N-acetylgalactosaminyltransferase 9 0.54 3.96E-05 0.00132872
CSPG4 NG2, chondroitin sulfate proteoglycan 4 0.53 3.83E-05 0.00132872
CSGALNACT2 GALNACT-2 (CSGalNAcT2/ChGalNAcT2;chondroitin b1 4) 0.53 0.000598 0.01102701
HAS2 Hyaluronate Synthase 2 0.53 0.015673 0.11815513
B4GALNT1 b-1 4-N-acetyl-galactosaminyl transferase 1 0.53 0.001998 0.02620662
FLNA Filamin A a 0.53 0.004531 0.04813459
TPST2 Tyrosylprotein sulfotransferase 2 0.52 0.000921 0.01468342
LEPR Leptin receptor 0.51 0.00676 0.06497193
LAMP3 Lysosomal-associated membrane protein 3 0.49 0.021372 0.14312278
UAP1 UDP-N-acteylglucosamine pyrophosphorylase 1 0.47 0.0011 0.01688544
B3GNT5 UDP-GlcNAc:bGal b-1,3-N-acetylglucosaminyltransferase 5 0.45 0.00076 0.01292345
LGALS7 Galectin 7 0.39 0.001684 0.02304475
WNT5B Wingless-type MMTV integration site 0.37 0.000139 0.00364597
HS6ST2 Heparan sulfate 6-O-sulfotransferase 2 0.23 0.005174 0.05172416
AREG Amphiregulin (schwannoma-derived GF) 0.15 0.000528 0.01039416
PRG4 Lubricin, proteoglycan 4 0.11 2.81E-10 1.18E-07
Fig. 6 Biosynthesis of
hyaluronate [16]. Results
obtained from the differential
gene array analysis by
comparing the 500 nM ATRA
construct to the Normal
construct demonstrates the
down-regulation of five genes
involved in hyaluronate
biosynthesis. The red arrows
indicate the position of these
enzymes in the HA biosynthetic
pathway
Arch Dermatol Res
123
tight distribution of collagen fibril diameters of human der-
mal matrices showed normal collagen fibrillogenesis as
evidenced by the histogram in Fig. 3b. The constructs ana-
lyzed at higher magnification all showed the appropriate
collagen banding pattern. Similarly, at lower magnification
all the human dermal matrices showed the presence of
fibroblasts dispersed in a collagenous matrix. These data
collectively suggest that ATRA at different concentrations
had no deleterious effect on collagen fibril assembly in the
human dermal matrices.
Glycosaminoglycan analysis by FACE demonstrated the
same fundamental GAG disaccharide profile for the human
dermal matrices grown in the presence of varying con-
centrations of ATRA. This observation suggests that the
fundamental fibroblast phenotype does not change, even in
the presence of high concentrations of ATRA. A decrease
in HA disaccharide content was observed with increasing
concentrations of ATRA. As previously observed with
other hormones (T3 and hydrocortisone) [11] no change in
other GAG disaccharide content was observed up to an
ATRA concentration of 1 lM ATRA. However, at higher
concentrations of ATRA (5 and 10 lM) this specificity was
lost and inhibition of other GAG disaccharide content was
also observed.
To obtain preliminary evidence to shed light on the
genes involved in hyaluronate inhibition observed in the
human dermal matrices cultured in the presence of varying
concentrations of ATRA, a differential gene array analysis
was conducted, in which the 500 nM ATRA human dermal
matrix was compared to the Normal construct (no ATRA).
Of the 127 differentially expressed genes, we specifically
analyzed the 49 down-regulated genes for clues as to the
participants in the inhibition of HA. Among these 49 genes
we observed 5 genes that are known participants in hyal-
uronate biosynthesis. GPI is responsible for converting
glucose-6-phosphate to fructose-6-phosphate in the early
stages of the glycolytic pathway [23]. Glutamine-fructose-
6-phosphate transaminase 1 (GFPT1, GFAT) is a rate
limiting enzyme of the hexosamine pathway. It catalyzes
the formation of glucosamine-6-phosphate and thus con-
trols the flux of glucose into the hexosamine pathway [25].
UAP1 catalyzes the synthesis of UDP-GlcNAc from Glc-
NAc-1-P and uridine-50-triphosphate [26]. Collectively,
these three enzymes control the formation of UDP-N-
Acetylglucosamine (UDP-GlcNAc), one of the two main
substrates for hyaluronate synthesis. Down-regulation of
these enzymes may therefore result in reduction of the
cellular content of UDP-GlcNAc. UDP-GlcDH is an oxi-
doreductase that catalyzes the conversion of UDP-glucose
to UDP-glucuronate (UDP-GlcUA) [1]. UDP-glucuronate
is the second substrate used for hyaluronate synthesis.
Down-regulation of UDP-GlcDH may result in reducing
the cellular content of UDP-glucuronate. In addition to the
potential reduction of the cellular pool of UDP precursors
to hyaluronate synthesis, HAS2, the enzyme responsible
for catalyzing the synthesis of HA from UDP-GlcNAc and
UDP-GlcUA in dermal fibroblasts was also observed to be
down-regulated [19]. Combined together, down-regulation
of these five enzymes may explain the sharp decrease in
hyaluronate content observed in the 500 nM ATRA human
dermal matrix. In spite of the reduction of the UDP-sugars,
no significant inhibition of other GAG (chondroitin sul-
fates) synthesis was observed in the 500 nM ATRA human
dermal matrix. This phenomenon can be attributed to the
fact that the UDP-sugars hyaluronate uses for its synthesis
reside in the cytosol, whereas the chondroitin sulfates are
synthesized in the Golgi apparatus. Since the concentra-
tions of the UDP-sugars necessary for HA and other GAG
synthesis are much higher in the Golgi as compared to the
cytosol, depletion of the cellular pool of UDP-sugars
affects HA synthesis rapidly, whereas the other GAGs
synthesized in the Golgi remain protected from this sub-
strate deficiency [11, 21, 22]. No up-regulation of any of
the hyaluronidases (HYAL 1-3) responsible for HA deg-
radation was observed, suggesting that degradation/catab-
olism of HA was most likely not responsible for the
diminished quantities observed in the 500 nM ATRA
construct. Future studies will focus on unraveling in detail
the contribution of each enzyme to the HA inhibition
process.
In conclusion we have shown in this study that addition
of ATRA to the culture medium in varying concentrations
produced self-assembled human dermal matrices with dif-
fering extracellular matrix characteristics. These novel
dermal equivalents can be potentially used to probe the role
of HA in the wound healing process. They can also serve as
model systems useful for deciphering the inhibitory effects
of various molecules on biosynthetic pathways as demon-
strated in this study for ATRA/HA biosynthesis.
Acknowledgments We thank Maria Ericsson and Louise Trakimas
(Harvard Medical School) for their valuable assistance with TEM. We
are grateful to Gilberto Hernandez (Scripps Research Institute) for his
assistance with RNA analysis. This research was supported by a grant
from the Charles H. Hood Foundation to TP.
References
1. Axelrod J, Kalckar HM, Maxwell ES, Strominger JL (1957)
Enzymatic formation of uridine diphosphoglucuronic acid. J Biol
Chem 224:79–90
2. Benjamini Y, Hochberg Y (1995) Controlling the false discovery
rate: a practical and powerful approach to multiple testing. J R
Stat Soc B 57:289–300
3. Berneburg M, Plettenberg H, Krutmann J (2000) Photoaging of
human skin. Photodermatol Photoimmunol Photomed 16:239–244
4. Bolstad BM, Irizarry RA, Astrand M, Speed TP (2003) A com-
parison of normalization methods for high density
Arch Dermatol Res
123
oligonucleotide array data based on bias and variance. Bioinfor-
matics 19:185–193
5. Bowman PH, Hogan DJ (2000) Ulcerated atrophic striae from
etretinate. Cutis 65:327–328
6. Brown S, Worsfold M, Sharp C (2001) Microplate assay for the
measurement of hydroxyproline in acid-hydrolyzed tissue sam-
ples. Biotechniques 30:38–42
7. Calabro A, Benavides M, Tammi M, Hascall VC, Midura RJ
(2000) Microanalysis of enzyme digests of hyaluronan and
chondroitin/dermatan sulfate by fluorophore assisted carbohy-
drate electrophoresis (FACE). Glycobiology 10:273–281
8. Calabro A, Hascall VC, Midura RJ (2000) Adaptation of FACE
methodology for microanalysis of total hyaluronan in chondroitin
sulfate composition from cartilage. Glycobiology 10:283–293
9. Darlenski R, Surber R, Fluhr JW (2010) Topical retinoids in the
management of photodamaged skin: from theory to evidence-
based practical approach. Br J Dermatol 163:1157–1165
10. Demetriou AA, Levenson SM, Rettura G, Seifter E (1985)
Vitamin A and retinoic acid: induced fibroblast differentiation
in vitro. Surgery 98:931–934
11. Deshpande M, Papp S, Schaffer L, Pouyani T (2013) Hydrocor-
tisone effect on hyaluronate synthesis in a self assembled human
dermal equivalent. J Tissue Eng Regen Med. doi:10.1002/term.
1809
12. Fleischmajer R, Perslish JS, Burgeson RE, Shaikh-Bahai F,
Timpl R (1990) Type I and type III collagen interactions during
fibrillogenesis. Ann N Y Acad Sci 580:161–175
13. Fosang AJ, Hey NJ, Carney SL, Hardingham TE (1990) An
ELISA plate-based assay for hyaluronan using biotinylated pro-
teoglycan G1 domain. Matrix 10:306–313
14. Gilchrest BA (1979) Relationship between actinic damage and
chronological aging in keratinocyte cultures of human skin.
J Invest Dermatol 72:219–223
15. Gilchrest BA (1994) Turning back the clock: retinoic acid
modifies intrinsic aging changes. J Clin Invest 94:1711–1712
16. Graves MV, Burbank DE, Roth R, Heuser J, DeAngelis PL, Van
Etten JL (1999) Hyaluronan synthesis in virus PBCV-1-infected
chlorella-like green algae. Virology 257:15–23
17. Hein R, Mensing H, Muller PK, Braun-Falco O, Krieg T (1984)
Effect of vitamin A and its derivatives on collagen production
and chemotactic response of fibroblasts. Br J Dermatol 111:37–44
18. Irizarry RA, Bolstad BM, Collin F, Cope LM, Hobbs B, Speed TP
(2003) Summaries of Affymetrix GeneChip probe level data.
Nucl Acids Res 31:e15
19. Itano N, Kimata K (2002) Mammalian hyaluronan synthases.
IUBMB Life 54:195–199
20. Itano N (2008) Simple primary structure, complex turnover reg-
ulation and multiple roles of hyaluronan. J Biochem 144:131–137
21. Jokela TA, Jauhiainen M, Auriola S, Kauhanen M, Tiihonen R,
Tammi MI, Tammi RH (2008) Mannose inhibits hyaluronan
synthesis by down-regulation of the cellular pool of UDP-N-
acetylhexosamines. J Biol Chem 283:7666–7673
22. Kakizaki I, Kojima K, Takagaki K, Endo M, Kannagi M, Ito M,
Maruo Y, Sato H, Yasuda T, Mita S, Kimata K, Itano N (2004) A
novel mechanism for the inhibition of hyaluronan biosynthesis by
4-methylumbelliferone. J Biol Chem 279:33281–33289
23. Kanno H, Fujii H, Miwa S (1998) Expression and enzymatic
characterization of human glucose phosphate isomerase (GPI)
variants accounting for GPI deficiency. Blood Cells Mol Dis
24:54–61
24. Lacroix A, Anderson GD, Lippman ME (1980) Retinoids and
cultured human fibroblasts. Effects on cell growth and presence
of cellular retinoic acid-binding protein. Exp Cell Res
130:339–344
25. McKnight GL, Mudri SL, Mathewes SL, Traxinger RR, Marshall
S, Sheppard PO, O’Hara PJ (1992) Molecular cloning, cDNA
sequence, and bacterial expression of human glutamine: fructose-
6-phosphate amidotransferase. J Biol Chem 267:25208–25212
26. Mio T, Yabe T, Arisawa M, Yamada-Okabe H (1998) The
eukaryotic UDP-N-acetylglucosamine pyrophosphorylases. Gene
cloning, protein expression, and catalytic mechanism. J Biol
Chem 273:14392–14397
27. Nelson DL, Balian G (1984) The effect of retinoic acid on col-
lagen synthesis by human dermal fibroblasts. Coll Relat Res
4:119–128
28. Pouyani T, Ronfard V, Scott PG, Dodd CM, Ahmed A, Gallo R,
Parenteau N (2009) De novo synthesis of human dermis in vitro
in the absence of a three-dimensional scaffold. In Vitro Cell Dev
Biol Anim 45:430–441
29. Pouyani T, Schaffer L, Papp S (2012) Tissue engineered fetal
dermal matrices. In Vitro Cell Dev Biol Anim 8:493–506
30. Pouyani T, Sadaka BH, Schaffer L, Papp S (2013) Triiodothy-
ronine inhibits hyaluronate synthesis in a human dermal equiv-
alent by down regulation of HAS2. In Vitro Cell Dev Biol Anim
49(8):178–188
31. Smith JG Jr, Davidson EA, Sams WM Jr, Clark RD (1962)
Alterations in human dermal connective tissue with age and
chronic sun damage. J Invest Dermatol 39:347–353
32. Smith TJ (1990) Retinoic acid inhibition of hyaluronate synthesis
in cultured human skin fibroblasts. J Clin Endocrinol Metab
70:660–665
33. Smyth GK (2004) Linear models and empirical Bayes methods
for assessing differential expression in microarray experiments.
Stat Appl Genet Mol Biol 3(1) (Article 3)
34. Varani J, Mitra RS, Gibbs D, Phan SH, Dixit VM, Mitra R Jr,
Wang T, Siebert KJ, Nickoloff BJ, Voorhees JJ (1990) All-trans
retinoic acid stimulates growth and extracellular matrix produc-
tion in growth-inhibited cultured human skin fibroblasts. J Invest
Dermatol 94:717–723
35. Varani J, Fligiel SEG, Schuger L, Perone P, Inman D, Griffiths
CEM, Voorhees JJ (1993) Effects of all-trans retinoic acid and
Ca?? on human skin in organ culture. Am J Pathol 142:189–198
36. Varani J, Perone P, Griffiths CEM, Inman DR, Fligiel SEG,
Voorhees JJ (1994) All-trans retinoic acid (RA) stimulates events
in organ-cultured human skin that underlie repair. J Clin Invest
94:1747–1756
37. Weiss JS, Ellis CN, Headington JT, Tincoff T, Hamilton TA,
Voorhees JJ (1988) Topical tretinoin improves photoaged skin: a
double blind, vehicle-controlled study. J Am Med Assoc
259:527–532
38. Williams ML, Elias PM (1981) Nature of skin fragility in patients
receiving retinoids for systemic effect. Arch Dermatol
117:611–619
Arch Dermatol Res
123