Cytochrome P450, CYP26AI, is expressed at low levels in human epidermal keratinocytes and is not...
Click here to load reader
Transcript of Cytochrome P450, CYP26AI, is expressed at low levels in human epidermal keratinocytes and is not...
Cytochrome P450, CYP26AI, is expressed at low levels in humanepidermal keratinocytes and is not retinoic acid-inducible
C.POPA, A.J.DICKER, A.L.DAHLER AND N.A.SAUNDERS
Epithelial Pathobiology Group, Centre for Immunology and Cancer Research, University of Queensland Department of Medicine,
Princess Alexandra Hospital, Brisbane, Queensland 4102, Australia
Accepted for publication 29 April 1999
Summary Retinoids, and their synthetic analogues, are well-established regulators of the squamous differ-
entiation programme both in vivo and in vitro. Despite this, very few studies have focused on themechanism by which retinoid action is terminated, e.g. metabolism. Recently, a new cytochrome
P450 family member (CYP26AI) was cloned. CYP26AI was reported to have substrate speci®city for
retinoids and to be retinoid-inducible. In this study, we have examined the expression and retinoicacid (RA) inducibility of CYP26AI in human epidermis and cultured keratinocytes. We found very
low levels of CYP26AI mRNA expression in both epidermis and keratinocytes. Furthermore, we
found no evidence for RA inducibility of CYP26 mRNA expression. This lack of RA inducibility wasnot due to inactivity of the retinoids, as we show that transglutaminase was still repressed by RA in
the same cultures. Despite the low levels of CYP26AI expression in the keratinocytes, the
keratinocytes were still capable of signi®cant RA metabolism. In conclusion, our study reports, forthe ®rst time, that CYP26AI is unlikely to contribute to RA metabolism in keratinocytes. These
studies also indicate that as yet unknown isoforms of cytochrome P450 may be involved in RA
metabolism in keratinocytes.
Key words: cytochrome P450, CYP26AI, human keratinocytes, retinoic acid
Squamous differentiation of keratinocytes is a tightlyregulated multistep process. Both in vivo and in vitro,
this process is initiated by the irreversible growth arrest
of proliferating cells and the subsequent induction ofsquamous differentiation-speci®c activities. The initia-
tion of growth arrest is characterized by inhibition of
expression of proliferation-associated genes such asE2F1 or cdk1.1±3 The induction of differentiation is
characterized by the expression of various squamous-speci®c genes such as transglutaminase type 1 (TG 1) or
corni®n.3±6 To a large extent, the tightly co-ordinated
changes in gene expression that accompany differentia-tion are transcriptional in mechanism.4,7 Disruption to
the control of this differentiation programme is
frequently associated with the development of variouspathologies.
It is well established that natural and synthetic
retinoids are potent regulators of the squamous differ-entiation programme.8 The actions of these retinoids
are mediated by a family of ligand-dependent nuclear
retinoic acid (RA) receptors (RARa, RARb and RARg)and retinoid X receptors (RXRa, RXRb and RXRg). Of
relevance to epidermal physiology are the reports that
keratinocytes selectively express RARg (90%), RARa
(10%) and RXRa.9,10 As retinoids are such potent
regulators of the differentiation programme, they have
extensive use in the treatment of various derma-topathologies such as neoplasms, psoriasis and acne.11
Although the action of retinoids is well understood,the mechanism by which their actions can be termin-
ated in the skin remain unclear. In this regard, the
recent cloning of a novel member of the cytochromeP450 family, CYP26AI, is of considerable interest.
CYP26AI was originally cloned by differential display
of RNA from RA-treated zebra®sh ®ns.12 This wasfollowed closely by the cloning of the mouse and
human homologues.13,14 CYP26AI was found to be
able to metabolize RA when transfected into Cos-1cells.15 Furthermore, the metabolism of RA could be
induced by pretreatment of F9 cells with RA.15 Meta-
bolite analysis indicated that the RA was degraded byoxidative metabolism consistent with that of the
cytochrome P450 family.12±14,16 Furthermore, these
British Journal of Dermatology 1999; 141: 460±468.
460 q 1999 British Association of Dermatologists
Correspondence: Nicholas Saunders. E-mail:NSaunders@medicine. pa.uq.edu.au
metabolites were consistent with RA metabolitesobserved in vivo.
Several observations suggest that CYP26AI may have
a part in RA metabolism in skin. First, it has beenreported that CYP26AI is expressed in a keratinocyte-
derived cell line.15,16 Secondly, it has been shown that
CYP26AI is selectively induced in the wounded epi-thelium of RA-treated zebra®sh ®ns.12 Thirdly, studies
of the RA inducibility of CYP26AI showed it to be
mediated by isoforms of receptors that are the majorforms also found in skin, namely RARg and RXRa.9,10
These ®ndings, coupled with the observation that the
epidermis is a major target of retinoid action, promptedus to examine the expression of CYP26AI in human
epidermal keratinocytes.
Materials and methods
Cell culture and treatment
Human epidermal keratinocytes (HEKs) and human
dermal ®broblasts were isolated from neonatal foreskinsand cultured as described.1,2 The culture of the kerati-
nocyte-derived squamous carcinoma cell line SCC25,
and the spontaneously immortalized keratinocyte cellline HaCaT, have also been described.17,18
The effects of the retinoid all-trans-RA, or its isomers9-cis-RA and 13-cis-RA, on CYP26AI and TG 1 mRNA
expression were investigated by incubating proliferative
and con¯uent keratinocytes with 1 mmol/L retinoid(Sigma, Sydney, Australia) in the presence or absence of
the generic P450 inhibitor, ketoconazole19 (10 mmol/L;
ICN Biomedicals, Melbourne, Australia) for 48 h. Insome instances, cells were pretreated with the generic
P450 inhibitor proadifen20 (Sigma) for 12 h prior to
incubation with [11,12±3H(N)-]RA ([3H]RA). All treat-ments involving RA were performed in reduced light
conditions to minimize isomerization to the 9-cis-RA or
13-cis-RA forms.
RNA isolation and reverse transcription±polymerase chain
reaction analysis
Total RNA was extracted from cells using Trizol reagent
(Life Technologies, Sydney, Australia), as described.1
Single-stranded cDNA was then synthesized from 2 mg
total RNA using SuperScript II RNaseHÿ reverse tran-
scriptase, as described.1 The expression of CYP26AI,actin and TG 1 mRNA were determined by polymerase
chain reaction (PCR) ampli®cation of 1 mL of the reverse
transcription (RT) reaction. All PCR ampli®cations were
performed under conditions of linearity with respect tocycle number.1 Actin primer sequences were: forward
primer, 50 GAA ATC GTG CGT GAC ATT AAG 30 and
reverse primer, 50 CTA GAA GCA TTT GCG GTG GACGAT GGA GGG GCC 30. These primers spanned exons 3±
5 and generated an mRNA-speci®c PCR product of
507 bp.21 TG 1 primer sequences were: forwardprimer, 50 TGG AGG CAC AGG ACG CGG TGA 30 and
reverse primer, 50 GAG AGC TGT GGG CTG TCC AAG 30,
which generated a PCR product of 580 bp. TheCYP26AI primer sequences were: forward primer, 50
AGG CAC TAA AGC AAT CTT CAA 30 and reverse
primer, 50 CAT GGA AAT GGG TGA ATC TTG 30,which generated a PCR product of 621 bp. PCR ampli-
®cations were carried out at a denaturation tempera-
ture of 94 8C for 40 s, an annealing temperature of 55 8C(actin and CYP26AI) or 65 8C (TG 1) for 40 s and an
extension temperature of 72 8C for 45 s. The PCR pro-
ducts were fractionated by agarose gel electrophoresisand then blotted on to a nylon membrane (Schleicher
and Schuell Inc., Keene, NH, U.S.A.), as previously
described.22 The DNA was ultraviolet cross-linked onto the membrane and probed for 24 h with the respec-
tive [g-32P]dCTP (Bresatec, Adelaide, Australia) cDNA
probes, and washed at high stringency, as described.23
The image was then analysed on an SF phosphorimager
(Molecular Dynamics, Melbourne, Australia) andinequalities in RT and PCR ampli®cation were normal-
ized to the actin expression of the same sample. The
range of linearity for determining mRNA levels for eachgene product was then determined by performing tem-
plate titration analysis.
Sequencing
To con®rm that the CYP26AI PCR product was genuineCYP26AI, the PCR product was puri®ed and subjected
to automated sequencing on an ABI automated sequen-
cer (Dr Brenda Cheung, Grif®th University, Australia).
Metabolism studies
The ability of proliferative keratinocytes to metabolize
high RA concentrations (1 mmol/L RA spiked with
33´33 nmol/L [3H]RA; New England Nuclear, Boston,MA, U.S.A.) and low RA concentrations (33´33 nmol/L
[3H]RA) over a 6-h period was examined. In other
experiments, cells were studied following either no RApretreatment, pretreatment with 1 mmol/L RA for 48 h
or pretreatment with 1 mmol/L RA plus 10 mmol/L
ketoconazole for 48 h. As a control in these experiments,
CYP26AI IN HUMAN EPIDERMAL KERATINOCYTES 461
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
media were spiked with 33´33 nmol/L [3H]RA, added tocells and removed immediately.
Following the incubations, media and keratinocytes
were harvested separately. The media were collectedand the cells trypsinized and resuspended in 500 mL
phosphate-buffered saline. Lipid-soluble retinoids were
separated from water-soluble retinoids by a modi®edBligh and Dyer method of total lipid extraction.24 All
organic solvents for this extraction contained the anti-
oxidant butylated hydroxytoluene (0´05%; Bioscienti®cPty Ltd, Brisbane, Australia). All tubes and tips were
coated with SurfaSil siliconizing ¯uid (Pierce, U.S.A.) to
prevent binding of lipids to surfaces; the extractionef®ciency by this method was 87 6 2% and 99 6 2%
when high and low RA concentrations were used,
respectively. Transformation of [3H]RA to water-solubleproducts was measured by liquid scintillation spectrom-
etry of aliquots of the aqueous phase of both media and
cell extracts. In some experiments, lipid-soluble extractswere evaporated to dryness and resuspended in mobile
phase (80 : 20 methanol/H2O v/v with 10 mmol/L
ammonium acetate) for high-performance liquidchromatography (HPLC) analysis.
Organic phase-soluble retinoids were separated by
HPLC on a Waters 3´9 ´ 150 mm C18 NovaPak analy-tical column using an isocratic mobile phase consisting
of 80 : 20 methanol/H2O with 10 mmol/L ammoniumacetate at a ¯ow rate of 0´5 mL/min. The HPLC system
consisted of a Waters Model M 45 solvent delivery
system, Waters Model 710B Wisp, Waters 490Programmable Multiwavelength detector set at
350 nm and Waters System Interface Module. The
computer package used for processing of acquireddata was the MilleniumTM 2010 Chromatography
Manager. All HPLC components were from Millipore
Corporation Waters Association (Milford, MA, U.S.A.).Column ef¯uent was monitored for 3H by scintillation
counting of collected fractions.
Results
CYP26AI expression in human epidermal keratinocytesand other tissues and cell lines
Several human cell types and tissues were screened forthe expression of CYP26AI mRNA by RT±PCR. HaCaT
cells and human liver were shown to express high levels
of CYP26AI mRNA, whereas the epidermis, SCC25cells, proliferating HEKs, differentiated HEKs or HEKs
treated for 48 h with RA showed very little CYP26AI
mRNA expression (approximately 50-fold less than the
liver; Fig. 1). The levels of CYP26AI expression observed
in the epidermis and HEKs could only be detected byprobing blots of PCR products (i.e. no PCR product
visible on an ethidium bromide-stained gel after 40
ampli®cation cycles). The low level of CYP26AI mRNAexpression was not due to contaminating DNA, as the
product was present in DNase I-treated RNA samples.
These data suggested that it is unlikely that the con-stitutive CYP26AI expression in the epidermis or cul-
tured keratinocytes is catalytically active. These data
also indicated that CYP26AI mRNA expression was notdependent upon the differentiation status of the HEKs,
as CYP26AI mRNA expression in proliferative HEKs was
not signi®cantly different from that observed in differ-entiated HEKs. Con®rmation of the differentiation status
of the HEKs was shown by a fourfold induction in TG 1
mRNA expression in these cultures (not shown).Furthermore, treatment of either proliferative or con-
¯uent HEKs with RA did not signi®cantly alter the
CYP26AI mRNA expression, even after a 5±8 day treat-ment (Fig. 2A,B). The bioactivity of the added RA was
con®rmed by its ability to downregulate the expressionof TG 1 mRNA in the same cultures of differentiated
HEKs (Fig. 2B). This demonstrated that the inability of
RA to induce CYP26AI mRNA was not due to theinactivity of the RA used. Combined, these data suggest
that the expression of CYP26AI mRNA in cultured
HEKs and in the epidermis is much lower than that
462 C.POPA et al.
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
Figure 1. Expression of CYP26AI in human tissue and cells. Total
RNA was extracted from cells or tissue and 2 mg used in reverse
transcription±polymerase chain reaction analysis. CYP26AI mRNA
levels were estimated by phosphorimage analysis as a percentage ofliver expression. Data expressed as mean 6 SEM of no less than three
experiments. All samples normalized to actin expression. Control, no
DNA; HaCaT, spontaneously derived keratinocyte cell line; SCC25,keratinocyte-derived squamous cell carcinoma cell line; Epid, epidermis;
Prol, proliferating keratinocytes; Diff, differentiated keratinocytes, RA,
proliferating keratinocytes treated with 1 mmol/L retinoic acid for 48 h.
observed in the liver and that CYP26AI mRNA is not RA
responsive in cultured HEKs.
Retinoic acid metabolism
Relatively low levels of CYP26AI mRNA were detected
in cultured HEKs by RT-PCR. Paradoxically, cytochrome
P450-dependent RA metabolism has previously beenreported in both cultured keratinocytes and skin.19,25,26
These previous studies demonstrated that the uptake of
RA, its metabolism and the release of RA metabolitesback into the media, can occur within 6 h. To examine
whether the low level of CYP26AI mRNA expression
observed in HEKs was associated with an inability tometabolize RA, proliferative keratinocytes were incu-
bated with either high (1 mmol/L) or low (33´33 nmol/L)
RA concentrations for 6 h.Treatment of HEKs with [3H]RA for 6 h resulted in a
large increase in the water-soluble radioactivity
recovered in the media (mean 6 SEM 28´8 6 0´86%)compared with the control cells (5´7 6 0´39%; Fig. 3A).
This was re¯ected in the loss of radioactivity recovered
in the lipid phase (97´2 6 0´18% to 62´3 6 3´08%;Fig. 3B). HEKs that had been pretreated with 1 mmol/L
cold RA for 48 h prior to being exposed to [3H]RA for 6 h
also showed an increase (16´7 6 0´67%) in the produc-tion of water-soluble radioactivity recovered in the
media relative to control cells (5´7 6 0´39%), but this
increase was lower than that observed in cells whichhad not been pretreated (Fig. 3A,B). Again, the increase
in water-soluble radioactivity correlated well with the
loss of radioactivity in the lipid phase (Fig. 3A,B).Furthermore, the production of water-soluble retinoid
derivatives was signi®cantly reduced in HEKs treated
with ketoconazole irrespective of whether the cells hadbeen pretreated with RA or not (Fig. 3A,B). Ketoconazole
has previously been shown to inhibit cytochrome P450-
dependent metabolism.26±28 These data indicated thatthe conversion of RA to water-soluble derivatives in
HEKs was inhibited by ketoconazole and, as such, wasconsistent with a cytochrome P450-dependent event.
However, other ketoconazole-sensitive pathways could
not be excluded. To address this issue, we preincubatedproliferating HEKs with the generic P450 inhibitor
proadifen (5 mmol/L) for 12 h, followed by incubation
with 33´33 nmol/L [3H]RA for 6 h. This resulted in areduction of water-soluble RA from 34´6 6 1´7% in the
untreated cells to 21´7 6 0´5% in the proadifen-treated
cells. The water-soluble RA in the spiked controls was10´4 6 0´5%. These data strongly suggest that P450-
dependent catalysis is involved in RA metabolism in
human keratinocytes.It is possible that RA-inducible RA conversion was
not detected in HEKs because the high concentration of
cold RA (1 mmol/L) used in the pretreatment (that mayhave remained after extensive washing) may have
saturated the converting enzymes or reduced the spe-
ci®c activity of the added [3H]RA. To address thisquestion, the previous experiments were repeated
using a high (1 mmol/L) RA concentration. Similar to
the studies conducted using low RA concentrations, theHEKs converted RA to water-soluble derivatives (Fig. 3C).
This conversion was inhibited by ketoconazole, and
there was no evidence for RA-induced RA conversion
CYP26AI IN HUMAN EPIDERMAL KERATINOCYTES 463
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
Figure 2. Time-dependent alterations in CYP26AI or transglutamin-
ase type 1 (TG 1) mRNA expression following retinoic acid (RA)treatment. Proliferative (A) or differentiated (B) human epidermal
keratinocytes (HEKs) were treated with 1 mmol/L RA for varying
lengths of time and then assayed for CYP26AI or TG 1 mRNA
expression. Total RNA was extracted and 2 mg used in the reversetranscription±polymerase chain reaction. mRNA levels were then
quanti®ed, following blotting and probing, by phosphorimage analy-
sis. Data presented as a percentage of untreated HEKs (day 0) and as
mean 6 SEM from three experiments. All samples normalized to actinexpression. Large SEMs were associated with estimating CYP26AI
mRNA levels, possibly due to its low expression in HEKs and the
sensitivity limits of the assay.
(Fig. 3C,D). This suggests that RA metabolism in culturedHEKs is cytochrome P450-dependent but not RA indu-
cible. Similar results were obtained using cell extracts ofthese same cultures (Fig. 4A±D). Ninety per cent of the
recoverable 3H retinoids were associated with the media.
The previous experiments established that culturedHEKs could metabolize RA despite the low expression of
the RA-speci®c cytochrome P450, CYP26AI. We next
examined the lipid extracts from HEKs reported inFigures 3 and 4 by HPLC analysis. Under the conditions
used in this laboratory, the retention time of RA was
approximately 10´2 min (corresponding to peak 4 inFig. 5) and for the isomers 9-cis-RA (peak 3, Fig. 5) and
13-cis-RA (peak 2, Fig. 5) were 9´02 min and 8´2 min,
respectively. Despite carrying out all experiments involving
RA in reduced light conditions, RA isomerization wasdif®cult to prevent (Fig. 5B). These data indicate that
isomerization of RA occurs readily in the cultures ofHEKs or during the extraction. This is consistent with
reports of an isomerase being present in epidermis.29
Despite the presence of the 9-cis and 13-cis isomers ofRA, we could detect no metabolites of RA by HPLC. The
lack of RA metabolites may be due either to their
transient nature or due to `masking' under peak 1which was predominantly non-enzymatic in origin.
This applied to samples from all the conditions shown
in Figures 3 and 4. The peak shown in Figure 5 (peak 1)was present in all samples, including the control, and
was not reduced following ketaconazole treatment (not
shown). This suggested that peak 1 was the product of
464 C.POPA et al.
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
Figure 3. Aqueous and lipid-soluble
derivatives of [3H]retinoic acid (RA) in
media following a 6-h incubation. Human
epidermal keratinocytes (HEKs) weretreated with either 33´33 nmol/L RA (A,B)
or 1 mmol/L RA (C,D) for 6 h. Media were
then extracted and lipid (B,D) or aqueous(A,C) soluble RA derivatives counted by
liquid scintillation spectrometry. Control,
media spiked with either a low
(33´33 nmol/L [3H]RA) or high (1 mmol/Lcold RA�33´33 nmol/L [3H]RA) RA
concentration; 6 h, HEKs treated with RA
for 6 h; keto, HEKs treated with RA plus
10 mmol/L ketoconazole for 6 h; pretreated,HEKs pretreated with 1 mmol/L RA for 48 h
then either a low or high RA concentration
([3H]) for 6 h, pre� keto, HEKs pretreated
with 1 mmol/L RA and 10 mmol/Lketoconazole for 48 h followed by a 6-h
treatment with either a low or high [3H]RA
concentration� ketoconazole. Datapresented as mean 6 SEM of triplicate
determinations from two experiments. All
data presented as a percentage of total
recoverable radioactivity.
non-enzymatic breakdown. These data indicated that:(i) RA is readily isomerized; (ii) there is a possibility that
the suppression of TG 1 mRNA by RA in HEKs is mediatedby 9-cis-RA or 13-cis-RA; (iii) qualitatively, RA intermedi-
ates could not be de®nitively identi®ed, as differences in
peaks between treatment groups and control cells couldnot be shown; (iv) ketoconazole did not alter the HPLC
pro®le signi®cantly; and (v) the RA derivative in peak 1
may be a non-catalytic conversion product of RA, as itwas present in the control sample as well.
Biological activity of retinoic acid derivatives
The previous ®ndings raised the possibility that the
retinoid effects observed in this study are mediated by
the parent compound, all-trans-RA, or by RA isomers.To address this question, the ability of the 9-cis-RA and
13-cis-RA isomers to suppress TG 1 mRNA expressionin con¯uent cultures of HEKs was tested. Figure 6 shows
that the mRNA expression of TG 1 markedly decreased
following a 48-h exposure to 1 mmol/L 9-cis-RA, 13-cis-RA or all-trans-RA. These data indicate that all-trans-
RA, 9-cis-RA or 13-cis-RA have similar abilities to
suppress TG 1 mRNA expression in differentiated HEKs.
Discussion
The present study shows, for the ®rst time, that the
newly cloned CYP26AI enzyme, thought to contribute
to RA metabolism, is poorly expressed and not inducible
CYP26AI IN HUMAN EPIDERMAL KERATINOCYTES 465
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
Figure 4. Aqueous and lipid-solublederivatives of [3H]retinoic acid (RA) in
human epidermal keratinocyte (HEK)
extracts following a 6-h incubation. HEKs
were treated with either 33´33 nmol/L RA(A,B) or 1 mmol/L RA (C,D) for 6 h. HEKs
were harvested and then extracted. Lipid
(B,D) or aqueous (A,C) soluble RAderivatives were then counted by liquid
scintillation spectrometry. Control, media
spiked with either a low (33´33 nmol/L
[3H]RA) or high (1 mmol/L coldRA�33´33 nmol/L [3H]RA) RA
concentration; 6 h, HEKs treated with RA
for 6 h; keto, HEKs treated with RA plus
10 mmol/L ketoconazole for 6 h; pretreated,HEKs pretreated with 1 mmol/L RA for 48 h
then either a low or high RA concentration
([3H]) for 6 h; pre� keto, HEKs pretreated
with 1 mmol/L RA and 10 mmol/Lketoconazole for 48 h followed by a 6-h
treatment with either a low or high [3H]RA
concentration� ketoconazole. Datapresented as mean 6 SEM of triplicate
determinations from two experiments. All
data presented as a percentage of total
recoverable radioactivity.
in HEKs. This is despite the observation that keratino-cytes possess signi®cant RA-metabolizing activity. This
®nding has signi®cant implications for our understand-
ing of normal keratinocyte biology as well as retinoid
therapeutics. For instance, it is established that kera-tinocytes are a major target for retinoids. However, our
study shows that the termination of retinoid action in
the keratinocyte may be mediated by isoform(s) ofcytochrome P450 other than CYP26AI. Our studies
also show that if modulation of retinoid levels is to be
achieved by modulating RA-catabolizing enzymes, thenenzymes other than CYP26AI should be considered.
In this study, we demonstrated that CYP26AI is
unlikely to be a major contributor to endogenousmetabolism of retinoids in the skin. In the ®rst instance,
we found that CYP26AI mRNA levels in both epidermis
and cultured keratinocytes were approximately 50-foldless than those in the liver. In contrast, the sponta-
neously derived immortal keratinocyte cell line, HaCaT,
had similar levels of expression to that of the liver. Thiswould explain the earlier ®nding that a keratinocyte cell
line could express CYP26AI.16 Furthermore, the expres-
sion of CYP26AI is not a characteristic of all establishedkeratinocyte cell lines, as the SCC25 cells reported in
this study showed little CYP26AI expression. The levels
of CYP26AI mRNA expression in keratinocytes werenot visible on an ethidium bromide-stained gel and
would appear to be expressed at a lower level than
`rare' mRNA species (e.g. E2F1) that we have previously
466 C.POPA et al.
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
Figure 5. High-performance liquid chromatography (HPLC) pro®les ofmedia derived from [3H]retinoic acid (RA)-treated human epidermal
keratinocytes (HEKs). (A) An HPLC pro®le of [3H]RA. Proliferating
HEKs were treated with [3H]RA for 6 h in the absence (B) or presence
(C) of 10 mmol/L ketoconazole. Following incubation, lipid extractswere taken and the sample injected on an HPLC. Eluted fractions were
collected and subjected to liquid scintillation spectrometry. Data
presented as disintegrations per minute eluted vs. the time of thefraction. The ®gures are representative of at least two independent
experiments. (1) Putative non-enzymatic degradation product; (2)
peak comigrating with authentic 13-cis RA; (3) peak comigrating
with authentic 9-cis RA; (4) peak comigrating with authentic RA.
Figure 6. Activity of retinoids in suppressing transglutaminase type 1
(TGase1) mRNA expression. Con¯uent (CONFL) human epidermal
keratinocytes (HEKs) were treated with 1 mmol/L all-trans, 9-cis or 13-
cis retinoic acid (RA) for 48 h. Total RNA was extracted and 2 mg usedin reverse transcription±polymerase chain reaction (PCR) analy-sis.
The mRNA levels for TGase1 were estimated by phosphorimage
analysis of Southern blots from the TGase1 PCR product. Datapresented as a percentage of the con¯uent HEKs and represent mean -
6 SEM of four experiments. Proliferating HEKs (PROL) are provided to
con®rm the differentiated status of the con¯uent cells. All samples
normalized to actin expression.
reported.1 These data suggest that the low levels ofCYP26AI mRNA expression reported in this study are
unlikely to be associated with catalytically active
CYP26AI. Further support for this comes from theobservation that the CYP26AI mRNA expression in
HEKs was not modulated by pretreatment with RA.
Previous studies have established that cells in whichCYP26AI appears to be catalytically active are also
subject to RA induction.16,30 The HEKs used in this
study were responsive to RA, as characterized by thesuppression of TG 1 mRNA and the induction of CRAB-
PII mRNA (not shown), but not with respect to
CYP26AI. Combined, these data strongly suggest thatCYP26AI is likely to be catalytically inactive in HEKs.
However, it may be possible that functional levels of the
CYP26AI protein are present despite little mRNAexpression.
Despite the low levels of CYP26AI mRNA expression,
HEKs maintained the capacity to metabolize RA. It hasbeen previously reported that microsomes isolated from
both skin and cultured keratinocytes are able to meta-
bolize RA extensively.25,29,31,32 The major metabolitesreported in these studies were 4-OH and 4-oxo RA.25
There have also been reports of 5,6-epoxy RA,25 but its
formation remains contentious. Formation of these meta-bolites appeared to be cytochrome P450-dependent, as
their production was inhibited by both ketoconazoleand liarozole.25,31 Although metabolites of RA have
been reported, there still is no clear evidence as to
what isoform of cytochrome P450 may be responsible.In fact, there is more evidence excluding isoforms than
there is evidence identifying isoforms. For instance,
there is now evidence to exclude the following isoforms:CYP26AI, CYP1AI, CYP3f and CYP1A2´32 These data
indicate the need for an exhaustive examination of skin-
associated cytochrome P450s. Only by this means willwe be able to identify potential cytochrome P450 iso-
forms involved in RA metabolism. This is an important
question to address given the extensive use of therapeu-tic retinoids, the importance of retinoids in normal skin
homeostasis and the continuing evidence implicating
retinoid metabolism in the development of retinoidinsensitivity in various tumours.
Acknowledgments
The authors extend their gratitude to Paul Taylor, Paul
Salm and Drs Paul Masci and Mark Bowles from theDepartment of Clinical Pharmacology for their expert
assistance with the HPLC. The authors also thank
Dr Julie Joyner from the Department of Surgery for the
generous gift of human liver RNA. This work wassupported by a grant from the Princess Alexandra
Hospital Foundation. Dr Dicker is supported by an
Australian NH & MRC Predoctoral Fellowship andMs C.Popa was supported by a CICR Honours Research
Scholarship.
References
1 Jones SJ, Dicker AJ, Dahler AL, Saunders NA. E2F as a regulator of
keratinocyte proliferation: implications for skin tumour develop-ment. J Invest Dermatol 1997; 109: 187±93.
2 Dahler AL, Jones SJ, Dicker AJ, Saunders NA. Keratinocyte growth
arrest is associated with activation of a transcriptional repressor
element in the human cdk1 promoter. J Cell Physiol 1998; 177:474±82.
3 Saunders NA, Jetten AM. Control of growth regulatory and
differentiation-speci®c genes in human epidermal keratinocytes
by interferon-g. Antagonism by retinoic acid and transforminggrowth factor b1. J Biol Chem 1994; 269: 2016±22.
4 Saunders NA, Bernacki SH, Vollberg TM, Jetten AM. Regulation of
transglutaminase type I expression in squamous differentiatingrabbit tracheal epithelial cells and human epidermal keratino-
cytes: effects of retinoic acid and phorbol esters. Mol Endocrinol
1993; 7: 387±98.
5 Marvin KW, George MD, Fujimoto W et al. Corni®n: a new cross-linked envelope precursor in keratinocytes. Down regulation by
retinoids. Proc Natl Acad Sci USA 1992; 89: 11026±30.
6 Medvedev A, Saunders NA, Chistokhina A et al. Regulation of
transglutaminase type I: identi®cation of DNA elements involvedin transcriptional control. J Biol Chem 1999; 274: 3887±96.
7 Eckert RL, Welter JF. Transcription factor regulation of epidermal
keratinocyte gene expression. Mol Biol Rep 1996; 23: 59±70.8 Jetten AM, Fujimoto W, Zhang LX et al. Regulation of gene
expression during squamous cell differentiation by multiple
retinoic acid signalling pathways. In: Retinoids from Basic Science
to Clinical Applications (Livrea MA, Vidali G, eds). Basel: BirkhaÈuserVerlag, 1994: 243±52.
9 Fisher GJ, Tawar HS, Xiao JH et al. Immunological identi®cation of
functional quantitation of retinoic acid and retinoid X receptor
protein in human skin. J Biol Chem 1994; 269: 20629±35.10 Xiao JH, Durand B, Chambon P, Voorhees JJ. Endogenous retinoic acid
receptor (RAR) -retinoid X receptor (RXR) heterodimers are the major
functional forms regulating retinoid-responsive elements in adult
human keratinocytes. J Biol Chem 1995; 270: 3001±11.11 Fujimoto W, Marvin KW, George MD et al. Expression of corni®n in
squamous differentiating epithelial tissues, including psoriatic and
RA treated skin. J Invest Dermatol 1993; 107: 268±74.12 White JA, Guo YD, Baetz K et al. Identi®cation of the retinoic acid-
inducible all-trans-retinoic acid 4-hydroxylase. J Biol Chem 1996;
271: 29922±7.
13 Ray WJ, Bain G, Yao M, Gottlieb DL. CYP26, a novel mammaliancytochrome P450, is induced by retinoic acid and de®nes a new
family. J Biol Chem 1997; 272: 18702±8.
14 Fujii H, Sato T, Kaneko S et al. Metabolic inactivation of retinoic
acid by a novel P450 differentially expressed in developing mouseembryos. EMBO J 1997; 16: 4163±73.
15 Abu-Bed SS, Beckett BR, Chiba H et al. Mouse P450RAI (CYP26)
expression and retinoic acid-inducible retinoic acid metabolism inF9 cells are regulated by retinoic acid receptor g and retinoid X
receptor a. J Biol Chem 1998; 273: 2409±15.
CYP26AI IN HUMAN EPIDERMAL KERATINOCYTES 467
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468
16 White JA, Beckett-Jones B, Guo YD et al. cDNA cloning of human
retinoic acid-metabolizing enzyme (hP450RAI) identi®es a novel
family of cytochromes P450 (CYP26). J Biol Chem 1997; 272:18538±41.
17 Saunders NA, Dicker AJ, Popa C et al. Histone deacetylase inhibi-
tors as potential anti-skin cancer agents. Cancer Res 1999; 59:
399±404.18 Boukamp P, Petrussevska RT, Breitkreutz D et al. Normal kera-
tinization in a spontaneously immortalized aneuploid human
keratinocyte cell line. J Cell Biol 1988; 106: 761±71.
19 Randolph RK, Simon M. Metabolism of all-trans-retinoic acid bycultured human epidermal keratinocytes. J Lipid Res 1997; 38:
1374±83.
20 Bort R, Ponsoda X, Jover R et al. Diclofenac toxicity to hepatocytes:a role for drug metabolism in cell toxicity. J Pharmacol Exp Ther
1999; 288: 65±72.
21 Nakajima-Iijima S, Hamada H, Reddy P, Kakunaga T. Molecular
structure of the human cytoplasmic beta-actin gene: interspecieshomology of sequences in the introns. Proc Natl Acad Sci USA
1985; 82: 6133±7.
22 Sambrook J, Fritsch EF, Maniatis T. Molecular Cloning: a Laboratory
Manual, 2nd edn. Cold Spring Harbor, NY: Cold Spring HarborLaboratory Press, 1989: 9.39±9.55.
23 Saunders NA, Dicker AJ, Jones SJ, Dahler AL. E2F1 mRNA is
destabilised in response to growth inhibitors in normal humankeratinocytes but not in a squamous carcinoma cell line. Cancer
Res 1998; 58: 1646±9.
24 Bligh EG, Dyer WJ. A rapid method of total lipid extraction and
puri®cation. Can J Biochem Physiol 1959; 37: 911±17.25 Duell EA, Astrom A, Grif®ths EM et al. Human skin levels of
retinoic acid and cytochrome P450-derived 4-hydroxy retinoic
acid after topical application of retinoic acid in vivo compared to
concentrations required to stimulate retinoic acid receptor-mediated transcription in vitro. J Clin Invest 1992; 90: 1269±74.
26 Hodam JR, Creek KE. Uptake and metabolism of 3H-retinoic acid
delivered to human foreskin keratinocytes either bound to serum
albumin or added directly to the culture medium. Biochim BiophysActa 1996; 1311: 102±10.
27 Roberts AM, Nichols MD, Newton DL, Sporn MB. In vitro metabo-
lism of retinoic acid in hamster intestine and liver. J Biol Chem
1979; 254: 6296±302.28 Howell SR, Shirley MA, Ulm EH. Effects of retinoid treatment of
rats on hepatic microsomal metabolism and cytochrome P450.
Correlation between retinoic acid receptor/retinoid X receptorselectivity and effects on metabolic enzymes. Drug Metab Dispos
1998; 26: 234±9.
29 Duell EA, Kang S, Voorhees JJ. Retinoic acid isomers applied to
human skin in vivo each induce a 4-hydroxylase that inactivatesonly trans retinoic acid. J Invest Dermatol 1996; 106: 316±20.
30 Marikar Y, Wang Z, Duell EA et al. Retinoic acid receptors regulate
expression of retinoic acid 4-hydroxylase that speci®cally inactiv-
ates all-trans retinoic acid in human keratinocyte HaCaT cells.J Invest Dermatol 1998; 111: 434±9.
31 Kang S, Duell EA, Kim KJ, Voorhees JJ. Liarozole inhibits human
epidermal retinoic acid 4-hydroxylase activity and differentiallyaugments human skin responses to retinoic acid and retinol in
vivo. J Invest Dermatol 1996; 107: 183±7.
32 Li XY, Astrom A, Duell EA et al. Retinoic acid antagonizes basal as
well as coal tar and glucocorticoid-induced cytochrome P4501A1expression in human skin. Carcinogenesis 1995; 16: 519±24.
468 C.POPA et al.
q 1999 British Association of Dermatologists, British Journal of Dermatology, 141, 460±468