Differentiating Keratinocytes Express a Novel Cytochrome P450 Enzyme, CYP2B19, Having Arachidonate Monooxygenase Activity*

The novel cytochrome P450, CYP2B19, is a specific cellular marker of late differentiation in skin keratinocytes. CYP2B19 was discovered in fetal mouse skin where its onset of expression coincides spatially (upper cell layer) and temporally (day 15.5) with the appearance of loricrin-expressing keratinocytes during the stratification stage of fetal epidermis. CYP2B19 is also present postnatally in the differentiated keratinocytes of the epidermis, sebaceous glands, and hair follicles. CYP2B19 mRNA is tightly coupled to the differentiated (granular cell) keratinocyte phenotypein vivo and in vitro. In primary mouse epidermal keratinocytes, it is specifically up-regulated and correlated temporally with calcium-induced differentiation and expression of the late differentiation genes loricrin and profilaggrin. Recombinant CYP2B19 metabolizes arachidonic acid and generates 14,15- and 11,12-epoxyeicosatrienoic (EET) acids, and 11-, 12-, and 15-hydroxyeicosatetraenoic (HETE) acids (20, 35, 18, 7, and 7% of total metabolites, respectively). Arachidonic acid metabolism was stereoselective for 11S,12R- and 14S,15R-EET, and 11S-, 12R-, and 15R-HETE. The CYP2B19 metabolites 11,12- and 14,15-EET are endogenous constituents of murine epidermis and are present in similar proportions to that generated by the enzymein vitro, suggesting that CYP2B19 might be the primary enzymatic source of these EETs in murine epidermis.

Fetal epidermis comprises a single cell layer at day 14 (E14) 1 in the mouse. By E18, a stratified, squamous epithelium is formed, comprising five cell layers (1). The appearance of differentiating (suprabasal) keratinocytes in fetal epidermis is characterized by the sequential expression and post-translational processing of keratins and other differentiation-specific proteins, which form cornified cell envelopes in terminally differentiated keratinocytes. The biochemical mechanisms underlying these remarkable structural changes, the keratinization and differentiation of epidermal keratinocytes, are largely unknown.
We identified a cytochrome P450 (CYP gene superfamily) in fetal mouse skin, CYP2B19, that is a new and novel cellular marker for differentiated keratinocytes. Mammalian cytochrome P450 enzymes are monooxygenases that metabolize small, hydrophobic compounds including steroids, sterols, fatty acids, fat-soluble vitamins, drugs, and toxins (2). Recombinant CYP2B19 metabolizes arachidonic acid, a normal constituent of cellular membranes and the precursor of biologically active lipids such as epoxyeicosatrienoic (EET) acids, hydroxyeicosatetraenoic (HETE) acids, leukotrienes, thromboxanes, and prostaglandins. Normal fatty acid metabolism is critical to the permeability barrier function of the epidermis. Scaling epidermal lesions, abnormal keratinocyte differentiation, and chronic hyperproliferation are associated with abnormal levels of fatty acids including arachidonate; these skin phenotypes can be induced by dietary essential fatty acid deficiency (3,4) or skin disease, such as psoriasis (5,6).
Cytochrome P450 arachidonic acid monooxygenases are present in many mammalian tissues where they comprise the only known enzymatic sources of regioisomeric EETs. Murine epidermis appears to utilize a unique, keratinocyte-specific arachidonic acid monooxygenase (CYP2B19) that generates EETs and HETEs. CYP2B12, a homologous cytochrome P450 in rat skin, is also an arachidonic acid monooxygenase. Like CYP2B19 it is keratinocyte-specific but generates only EETs from arachidonate (7,8). Expression data indicate that the catalytic activities of these two enzymes are highly compartmentalized to the most differentiated, viable cell layers in the epidermis (granular cells), hair follicles (inner root sheath), or sebaceous glands (sebocytes).
The P450-mediated bioactivation of arachidonate generates eicosanoids that may function in vasomodulation (EETs and 19-and 20-HETE) in vascular tissues or as secretagogues (EETs) for neuroendocrine cells (9,10). One possible mechanism of action of EETs was demonstrated in endothelial cells where EETs activate ion channels and effect increases in cytosolic Ca 2ϩ concentrations (11). While EETs or their enzymatic source have not been studied previously in cutaneous tissues, a physiological role involving ion transport or Ca 2ϩ -dependent signaling in keratinocytes must be considered since Ca 2ϩ is a key regulator of keratinocyte differentiation in vitro (12). The Ca 2ϩ gradient maintained in the epidermis in situ also suggests a regulatory role for this cation in vivo, extracellular and intracellular free Ca 2ϩ concentrations are greatest in the granular cell layer where CYP2B19 is expressed and lowest in the basal cell layer (13).

EXPERIMENTAL PROCEDURES
RNA Isolation and Polymerase Chain Reaction-Total RNA was isolated as described (14). For reverse transcription-polymerase chain reaction (RT-PCR), total RNA was primed with oligo(dT) 16 , reversetranscribed and amplified using a GeneAmp RNA PCR kit (Perkin-Elmer). CYP2B subfamily-specific oligonucleotides 1 and 2 (Table I) were used to screen embryonic RNAs. The PCR products were sizefractionated on agarose gels, visualized with ethidium bromide, subcloned, and DNA sequenced. Oligonucleotides 3-6 were used for tissue distribution studies.
In Situ Hybridization and Northern Analysis-Mice (CD-1 outbred strain) obtained from Charles River and rats (Sprague-Dawley) obtained from Harlan Sprague Dawley were handled according to protocols approved by Vanderbilt University. Tissue preparation and in situ hybridization methods were described (14). Specific hybridization was distinguished by comparing silver grain development produced by sense and antisense cRNAs applied to consecutive sections on the same slide. For Northern analyses, total RNA was size-fractionated on formaldehyde-containing 1% agarose gels and transferred by capillarity to Ge-neScreen Plus membranes (DuPont). Membranes were baked at 80°C and stained with methylene blue to assess RNA integrity. The following murine cDNAs were used for in vitro transcription to generate [ 35 S]UTP-labeled cRNAs for in situ hybridization or were random labeled with [ 32 P]dCTP for Northern blot analyses: CYP2B19 (GenBank accession number AF047529, 558 -981 and 2381-2639 bp), keratin 10 (number V00830, 1727-1944 bp), keratin 14 (number M13806, 681-1063 bp), loricrin (number M34398, 1518 -1684 bp), profilaggrin (number J03458, 12-188 bp), and low density lipoprotein (LDL) receptor (number X64414, 2109 -2272 bp). For Northern data analyses, radioactivity on the washed membranes was quantitated using a PhosphorImager and the volume integration method of ImageQuant software (Molecular Dynamics). In every sample lane, the relative counts (average from two scans) corresponding to each radiolabeled cDNA were normalized for those corresponding to cyclophilin mRNAs (PstI-BamHI fragment of p1B15) (15). The data are reported as % of the 0 h control value for each experimental treatment.
Isolation of a CYP2B19 cDNA and Construction of pCW2B19 Expression Vector-One ϫ 10 6 plaques of a whole mouse skin (adult female C57B/6) cDNA library (Stratagene) were screened by plaque hybridization. A 595-bp fragment corresponding to the 5Ј-end (exonic) of CYP2B19 was generated by RT-PCR and randomly labeled with [ 32 P]dCTP for hybridization. Secondary and tertiary screening resulted in five positive plaques. The cDNA inserts were sequenced, including both strands of clone 2-1B-1A.
A full-length CYP2B19 cDNA (pBS2B19; clone 2-1B-1A) was modified as described (16,17) and subcloned into pCWoriϩ, generating pCW2B19. Briefly, the 3Ј-untranslated sequence of CYP2B19 was shortened by PCR mutagenesis using oligonucleotides 7 and 8 (Table I). This antisense oligo contains a KpnI site, for ligation back into pBS2B19 and an adjacent, internal XbaI site for subsequent ligation into pCWoriϩ. The PCR product (635 bp) was restricted with HindIII and KpnI, and a 200-bp fragment containing the stop codon was purified and ligated back into pBS2B19. Next, the first seven residues of native CYP2B19 were changed to MALLLAV by PCR mutagenesis using oligonucleotides 9 and 10. The PCR product (750 bp) was restricted with BamHI and NcoI, and a 140-bp fragment containing the modified N terminus was purified and ligated back into pBS2B19. Finally, pBS2B19 modified at both ends was restricted with NdeI and XbaI, producing a 1.5-kb fragment, which was ligated into pCWoriϩ. By DNA sequence analysis, pCW2B19 contained the entire open reading frame of CYP2B19, the N-terminal MALLLAV modification, and 25 bp of 3Ј-untranslated sequence.
Arachidonic Acid Metabolism Studies-Arachidonate monooxygenase activity was assayed as described (20). Partially purified CYP2B19 was reconstituted in the presence of [1-14 C]arachidonic acid (50 Ci/ mol; 50 M final concentration). Organic extracts of the reaction mixtures were resolved by reversed-phase HPLC with on-line liquid scintillation detection. Radioactive fractions containing metabolites were collected and analyzed on different chromatographic systems (reversed-phase, normal phase or chiral columns) to identify analytes (21,22). Endogenous EETs were measured in fresh frozen murine epidermis. Whole skins from 15 neonatal mice were frozen flat on solid CO 2 and scraped with a razor blade. An accurate wet weight measurement of the pooled, scraped epidermis was not obtained because of frozen water vapor in the sample. Tissue was extracted in the presence of triphenylphosphine, and the content of 8,9-, 11,12-, and 14,15-EET was measured by gas chromatography/mass spectrometry as described (21).
Primary Keratinocyte Cultures-Epidermal keratinocytes were isolated from 2-3-day-old mice as described (23). Cells were plated in complete keratinocyte growth medium (Medium 154XP, Cascade Biologics) containing 0.05 mM Ca 2ϩ and 1% Chelex-treated fetal bovine serum (1 pup equivalent/two 100-mm dishes). The cultures were fed on the third day and every other day thereafter with the same medium (no serum). In 6 -7 days when the cultures were 75-85% confluent, treatments were initiated with fresh growth medium (no serum) containing 0.125 mM Ca 2ϩ to induce differentiation. Some cultures were treated with higher Ca 2ϩ concentrations or other divalent cations in the presence of 0.05 mM Ca 2ϩ where indicated. Finally, some cultures (in 0.05 mM Ca 2ϩ ) were treated with platelet activating factor (1-O-hexadecyl-2-acetyl-glycerophosphocholine) (Sigma), phorbol 12-myristate 13-acetate (TPA) (Sigma), or vehicle (0.1% ethanol). Control cultures were maintained in 0.05 mM Ca 2ϩ . At timed intervals after initiating treatments (0 h), cells were washed with phosphate-buffered saline. Attached cells were lysed in 1.5 ml of guanidinium thiocyanate. Cells in the wash were collected by centrifugation, lysed with 0.5 ml guanidinium thiocyanate, pooled with the attached cell lysates, frozen in liquid nitrogen, and stored at Ϫ80°C.

Identification of a Novel CYP Monooxygenase in Fetal Mouse
Skin-Mouse embryo RNAs were screened by RT-PCR to identify CYP genes expressed during fetal development. Degener-       ate oligonucleotides specific for mammalian CYP2B genes amplified a fragment of a novel gene, Cyp2b19, with RNA isolated from developing limbs (E15.5). This nucleotide sequence was more closely related to two rat genes, CYP2B12 (7,8) and CYP2B15 (24), than to previously identified murine genes. Authenticity of this partial cDNA was confirmed by the specific hybridization of antisense CYP2B19 cRNA to fetal mouse epidermis (E15.5; Fig. 1A). In serially sectioned mouse embryos, specific hybridization was not observed with any other tissue or organ. Silver grains representing CYP2B19 mRNA localized to the uppermost suprabasal cells (Fig. 1, B and C), the most differentiated keratinocytes in this immature (E16.5) epithelium. Specific hybridization was not observed in basal cells or in many of the intermediate cells. Temporally, CYP2B19 mRNA was first detected at E15-E15.5, coincident with the appearance of suprabasal cells in fetal epidermis. Expression levels appeared to increase proportionately with increases in the number of differentiated keratinocytes, as the epidermis thickens between E15 and term (ϷE19).
Spatially, CYP2B19 is expressed in the same cellular layer as loricrin, a marker of late differentiation of epidermal keratinocytes (granular cell layer) (1). Loricrin mRNAs localized to the most differentiated (uppermost) suprabasal cells in fetal (E16.5) epidermis (Fig. 1, D and E). Considering its much higher expression level than CYP2B19, relatively few silver grains representing loricrin mRNA were present over intermediate cells. A different pattern was observed for keratin 10, a marker of newly differentiating keratinocytes (spinous cell layer) (1). Keratin 10 was expressed at moderately high and uniform levels in all suprabasal cells (Fig. 1, F and G). None of these mRNAs were detectable in basal cells. These data dem-onstrate that CYP2B19, like loricrin, is a specific marker for the most differentiated, nucleated keratinocytes in murine epidermis. The coincident onset of CYP2B19 expression and the appearance of loricrin-expressing suprabasal keratinocytes suggest a function for this cytochrome P450 enzyme in the stratification stage during development of fetal mouse epidermis.
Isolation and Characterization of a Full-length CYP2B19 cDNA-Sequence analyses of RT-PCR-generated CYP2B clones led to identification of low levels of CYP2B10 mRNA in mouse skin (not shown). CYP2B10 is a predominant CYP2B in mouse liver having 16-hydroxylase activity with androgens, like testosterone (25). CYP2B10 mRNAs were represented in the mouse skin cDNA library, so a CYP2B19 hybridization probe and stringency conditions were chosen to minimize crosshybridization with CYP2B10. Several CYP2B19 cDNA clones were isolated (Fig. 2) that contained the novel CYP2B sequence originally amplified from mouse embryo RNA. Except for a missing ATG start codon, clone 2-1B-1A is full-length (2.7 kb).
It contains an open reading frame encoding a 492-amino acid protein, 1.2 kb of 3Ј-untranslated sequence, and a polyadenylated tail. A start codon was likely missing because the second codon contains an EcoRI site, the same restriction site used to prepare the cDNA library. An ATG start codon was found in a 2-kb clone (5-2A-1A) containing the first 0.96 kb of the open reading frame, up to the putative exon 6/7 boundary. Exon/ intron boundaries are conserved within CYP gene subfamilies (26). The remainder of this clone might encode intron 6 because the next 50 bp are 90% identical to intron 6 of CYP2B15 (24). Only the first 50 bp of the 5Ј-end of this intron have been reported. Sequence data obtained beyond this point were un- Murine CYP2B19 is 87 and 86% identical to rat CYP2B12 and rat CYP2B15, respectively. The deduced amino acid sequences (Fig. 3) of these three homologous enzymes are typical of cytochromes P450. They are unremarkable except for residue 24, located near the end of the N-terminal, membrane-spanning region and preceding the conserved proline-rich region (27). This residue is present in CYP2B19, CYP2B15, and CYP2B12 but is absent (gapped) in all other CYP2Bs, in alignments generated by multiple sequence comparisons. While the functional significance of this "extra" residue is unknown, there is a strict correlation between its presence and keratinocytespecific expression. Hence, this residue may serve as a signature of keratinocyte-specific CYP2B monooxygenases.
CYP2B15 Is the Putative CYP2B19 Orthologue in Rat Skin-The striking sequence identities among CYP2B19, CYP2B15, and CYP2B12 led us to compare their expression patterns throughout development. All three genes were found to be uniquely expressed in differentiated, skin keratinocytes (Table  II). Data for rat CYP2B12 have been reported (8). Because the expression patterns of CYP2B15 are identical to those of CYP2B19, the results in Fig. 4 are equally applicable to CYP2B19 in mouse skin and CYP2B15 in rat skin. CYP2B19 and CYP2B15 are expressed not only in fetal epidermis, but throughout all stages of postnatal development, in both sexes. The only notable difference between these species is that epidermal CYP2B15 mRNA levels in rat skin appeared to diminish somewhat in adulthood. In adult mouse skin (Fig. 4, A and  B), CYP2B19 mRNAs localized to the differentiated keratinocytes of three cutaneous structures: suprabasal cells in the epidermis, sebocytes in sebaceous glands, and the inner root sheath of hair follicles. While epidermal CYP2B19 expression was first detected at E15-E15.5, follicular and sebaceous expression were first detected in neonatal skin. These skin appendages develop morphologically during the first week of post-natal development in rodents.
In rat skin (Fig. 4, C and D), CYP2B15 localized to suprabasal cells in the epidermis and sebocytes in the preputial gland, the large paired sebaceous glands beneath the genital skin of rodents. Sectioned close to the midline, sebaceous tissue expressing CYP2B15 is shown surrounding a central collecting duct that directs the flow of sebum to the skin surface. Sebocytes in other types of sebaceous glands also express CYP2B15: meibomian glands in the eyelid, anal glands in anogenital skin, and sebaceous glands associated with hair follicles (not shown). A serially sectioned rat vibrissae follicle (Fig. 4, E and F) shows a single layer of CYP2B15-expressing keratinocytes, which outlines the hair canal. An adjacent section cut obliquely (Fig. 4G) exposes a relatively large surface of inner root sheath expressing CYP2B15.
The tissue distribution of CYP2B19 and CYP2B15 was analyzed by RT-PCR. For CYP2B19 mRNAs in mouse tissues, products of expected size were obtained using RNAs isolated from all cutaneous tissues examined, but not from mouse adrenal gland, ovary, testis, vesicular gland, uterus, liver, lung, kidney, small intestine, colon, caecum, spleen, brain, heart, tongue, tibia, or femur (not shown). No products were observed in the absence of reverse transcriptase. Similar results were obtained for CYP2B15 mRNAs in rat tissues. Products of expected size were amplified with RNA isolated from all cutaneous tissues examined, but not from rat adrenal gland, ovary, uterus, heart, liver, kidney, or spleen (not shown). In situ hybridization studies also confirmed that CYP2B19 is not detectable in the keratinocytes of mucosal epithelia such as the rectum, colon, vagina, uterus, cornea, and conjunctiva (not shown). Hence, CYP2B19 and CYP2B15 appear truly specific for skin keratinocytes.
CYP2B19 and CYP2B15 are predicted to be orthologous because their expression patterns are indistinguishable and their 3Ј-untranslated sequences are more identical to each other than to that of CYP2B12. Importantly, CYP2B12 expression patterns differ (8) (Table II), being limited to sebocytes, a subset of differentiated skin keratinocytes. The inability to detect epidermal and follicular CYP2B12 expression by in situ hybridization further suggests that the orthologues will be CYP2B19 and CYP2B15.
CYP2B19 Is an Arachidonic Acid Monooxygenase-Arachidonic acid was demonstrated to be a substrate for recombinant CYP2B19. Turnover numbers estimated for CYP2B19 with [1-14 C]arachidonic acid averaged 3.5 min Ϫ1 (Fig. 5A). Metabolites generated by CYP2B19 were resolved by reversed-phase HPLC (Fig. 5B). Five fractions (P1-P5) were collected and analyzed further, and characterized structurally (Table III). All five P450 metabolites were produced enantioselectively, attesting to their enzymatic origin. Fraction P5 had the same retention time in reversed-phase HPLC as a mixture of authentic 11,12-and 8,9-EET. Further analysis confirmed that P5 contained only 11,12-EET and that the major enantiomer was 11S,12R-EET. Fraction P4 had the same retention time as authentic 14,15-EET and contained nearly exclusively 14S,15R-EET. Fraction P3 contained a mixture of 11-, 12-, and 15-HETE, in which the predominant enantiomers were 11S-, 12R-, and 15R-HETE. The radioactive material in P1 had a retention time similar to that of a mixture of dihydroxyeicosatrienoic acids and likely arose by hydration of the EETs. The radioactive material in P2 remains to be identified. Importantly,or -1 hydroxylated arachidonic acids were not detected. In summary, EETs accounted for approximately twothirds of the identified metabolites, while HETEs accounted for the remaining one-third. Significantly, CYP2B19 is the first cytochrome P450 identified in skin that generates 12R-HETE. Abnormally high concentrations of 12R-HETE are present in hyperproliferative epidermis, in psoriasis and related dermatoses (5,6,28,29), implicating a role for the enzymatic source of this eicosanoid in skin diseases.
Differentiation Induced by Calcium Up-regulates Cyp2b19 Gene Expression in Vitro-Northern analysis was used to investigate whether normal mouse epidermal keratinocytes express CYP2B19 in vitro and whether mRNA levels correlate temporally with those of two other gene markers of late differentiation, loricrin and profilaggrin. Proliferating keratinocyte cultures were induced to differentiate by increasing extracellular Ca 2ϩ from 0.05 mM (promotes basal cell phenotype) to 0.125 mM (induces granular cell phenotype) as described (12,23). Higher Ca 2ϩ concentrations were less effective in up-regulating cytokeratins 1 and 10 and filaggrin (12). Fig. 6 shows a 6.6-fold increase in profilaggrin mRNA in primary mouse epidermal keratinocytes at 36 h after inducing differentiation and a 9-fold increase at 48 h. CYP2B19 mRNA levels were upregulated in parallel with those of profilaggrin, achieving a 15-fold increase by 48 h, while little change was observed in LDL receptor mRNA levels. Some spontaneous differentiation occurred in keratinocytes maintained in low Ca 2ϩ (0.05 mM) medium (5-fold increase at 48 h for profilaggrin and CYP2B19 mRNAs). This is not surprising since the cultures become confluent over the time course of treatment, and this factor alone causes keratinocytes to differentiate (30). These studies demonstrate a close correlation between CYP2B19 and profilaggrin mRNA levels, two markers of late differentiation, whether differentiation is induced by Ca 2ϩ or occurs spontaneously.
Results in Fig. 6 are consistent with data from another study in which mRNA levels for five genes were measured after the same calcium induction protocol, over a 36-h time course (Table  IV). These data demonstrate the specificity of Ca 2ϩ (0.125 mM) to up-regulate genes specifically associated with late differentiation of epidermal keratinocytes. For example, at 36 h after increasing Ca 2ϩ , message levels for CYP2B19, profilaggrin, and loricrin ranged from 500 to 700% of the 0-h control values, while those of cytokeratin 14 (a basal cell cytokeratin marker) and LDL receptor varied only Ϯ20%. The effect of calcium was specific since two other stimuli that act on keratinocytes, platelet activating factor (31), and phorbol ester (32) either had little effect or even decreased message levels of the five genes.
Up-regulation of CYP2B19 during the Ca 2ϩ -induced differentiation protocol was investigated further by testing cation specificity and dose responsiveness to Ca 2ϩ , over a 72-h time course (Fig. 7). Maximal CYP2B19 mRNA levels were observed at 72 h (not shown). Previous studies have shown that Sr 2ϩ at relatively high concentrations (Ն2.5 mM) can mimic some of the effects of Ca 2ϩ in primary keratinocytes, while Mg 2ϩ and five other divalent cations had no effect (33) . Fig. 7, A and B, shows that 5 mM Mg 2ϩ and 0.125 mM Sr 2ϩ had no effect on relative mRNA levels for CYP2B19 or profilaggrin. At 5 mM, Sr 2ϩ had only a modest effect on CYP2B19 mRNA, suggesting the specificity of Ca 2ϩ to induce a program of differentiation events FIG. 3. Comparison of the deduced amino acid sequences of three homologous CYP2B monooxygenases unique to differentiated skin keratinocytes. The complete coding sequence of murine CYP2B19 is shown (top, 492 residues); only residues that differ are shown for rat CYP2B12 (87% identity; 492 resides) and rat CYP2B15 (86% identity, 495 residues). The dashed arrows correspond to the positions of oligonucleotides originally used to identify CYP2B19 in mouse embryo RNA. The underline indicates the cytochrome P450 signature sequence: FXXGXXXCXG. 2, residue 24 contained in these keratinocyte-specific CYP2Bs is missing (i.e. gapped in sequence alignments) in all other CYP2Bs. *, cysteine residue conserved in all CYP genes that coordinates the heme prosthetic group. †, stop codon. leading to expression of the Cyp2b19 gene. The concentration of extracellular Ca 2ϩ affected the time course and extent of up-regulation of CYP2B19 mRNA (Fig. 7), as demonstrated previously for the cytokeratin genes 1 and 10, and profilaggrin (12) . Fig. 7, C and D, shows that 0.125 mM Ca 2ϩ was optimal for up-regulation of both CYP2B19 and profilaggrin mRNA in normal mouse epidermal keratinocytes, during 72 h of treatment. The increase in message appears slightly later in these cultures compared with those represented in Fig. 6 and Table IV, but the data are otherwise consistent among different primary cell preparations.
Murine Epidermis Contains Endogenous EETs-To determine whether the EETs generated by recombinant CYP2B19 are present in mouse skin, we extracted epidermis from neonatal mice and analyzed its EET content by gas chromatography/mass spectrometry. Of 36 ng total EET recovered, 25% (9.7 ng) was 14,15-EET, 41% (14.6 ng) was 11,12-EET, and 32% (11.6 ng) was 8,9-EET. Similar ratios of 11,12-EET:14,15-EET were calculated for EETs generated by recombinant CYP2B19 (1.75:1) and endogenous EETs in mouse epidermis (1.5:1), suggesting that CYP2B19 may participate in the endogenous formation of these eicosanoids in epidermis. The endogenous 8,9-EET suggests the presence of an additional epoxygenase, but we were unable to identify another major CYP2B in mouse skin as a possible source of 8,9-EET, as we did in rat skin (i.e. CYP2B12 and CYP2B15) (8).

DISCUSSION
CYP2B19 is an arachidonate monooxygenase expressed exclusively in differentiated skin keratinocytes. It is a specific marker of the most differentiated, nucleated keratinocytes in cutaneous tissues. Other differentiation-specific keratinocyte markers are either intermediate filaments (cytokeratins) or enzymes involved in formation of cornified cell envelopes. CYP2B19 differs from previously identified keratinocyte markers because it catalyzes the formation of lipid mediators that are likely to participate in keratinocyte signaling pathways.
A role for CYP2B19 in lipid metabolism was suspected initially because differentiated keratinocytes in the epidermis and sebaceous glands are highly enriched in the lipid precursors that ultimately form the water permeability barrier in the epidermis and sebum in sebaceous glands (34). The arachidonic acid epoxygenase activity of recombinant CYP2B19 suggested to us that this enzyme potentially functions in an intracellular signaling pathway and unlikely has a direct role in the biosynthesis of structural lipids or sebum. CYP2B19 expression is always detected when differentiated keratinocytes are present, from the ontogeny of fetal epidermis throughout postnatal development. Hence, it will be important in future studies to prove whether CYP2B19 catalytic activity and signals generated by its metabolites are required for the terminally differentiated (granular cell) phenotype of keratinocytes in skin. Based on in situ hybridization, RT-PCR, and Northern analyses, expression of CYP2B19 and CYP2B15 is restricted to a single cell type, the differentiated keratinocytes in skin, as is the homologous cytochrome P450, rat CYP2B12 (8). The cell type specificity of these three skin-specific CYP2Bs is unparalleled among mammalian CYP genes. In the CYP2B15 gene (24), a sequence motif upstream (Ϫ827 bp) of the putative "TATA" box conforms to the consensus sequence (AAPuC-CAAA) found in the 5Ј-region of epidermal cytokeratin genes (35). CYP2B19, CYP2B15, and CYP2B12 are specific for skin keratinocytes, unlike previously described keratinocyte markers. For example, loricrin is also present in the granular cell layer of oral, esophageal, and fore-stomach mucosa (36).
It is not apparent why epidermis would require a keratinocyte-specific cytochrome P450 to epoxidize arachidonic acid. No other arachidonic acid epoxygenase has been shown to be specific for a single cell type. Perhaps differentiated keratinocytes have a unique requirement for the EET and HETE products of CYP2B19, or perhaps the Cyp2b19 gene contains regulatory elements required to target expression specifically to the most differentiated, nucleated keratinocytes in skin. The potential to form 12R-HETE in the skin by a CYP-dependent mechanism is novel and physiologically relevant. A cytochrome P450 enzyme has been suspected as the source of this eicosanoid in human skin since Wollard (28) identified 12R-HETE as the predominant stereoisomer in psoriatic lesions. This finding, which also proved true for related dermatoses (6,29), ruled out platelet type 12-lipoxygenase as the source because this enzyme generates only 12S-HETE. Isolated human epidermal cells form EETs and HETEs, including 12R-HETE, from arachidonic acid (37). These activities were stimulated by NADPH and inhibited    by CO, hallmarks of cytochrome P450 enzyme activities. It is likely that the arachidonic acid monooxygenase activities in human epidermis will be catalyzed by an orthologue of murine CYP2B19. Recently, a 12R-lipoxygenase was identified in human skin (38). It will be important to demonstrate whether this novel lipoxygenase is expressed in the same cutaneous cell types as CYP2B19, since it is very likely that orthologues corresponding to both enzymes exist in human and murine skin. The ultimate physiological and biochemical functions of HETEs and EETs in normal epidermal keratinocytes remain to be established. A critical question is whether any specific eicosanoid metabolite(s) resulting from CYP2B arachidonate monooxygenases activities are involved in epidermal hyperproliferation and scaling skin phenotypes in murine and human skin.
Murine epidermis contains endogenous 11,12-and 14,15-EET in similar proportion to that generated in vitro by recombinant CYP2B19. The endogenous epidermal EET profile supports the hypothesis that CYP2B19 may be responsible for endogenous 11,12-and 14,15-EET synthesis in mouse epidermis and that this biosynthesis is largely restricted to the upper, granular cell layer of the epidermis. This is because CYP2B19 is only expressed in epidermal keratinocytes that have committed to differentiate and passed through the spinous (newly differentiating) cell layer. The physiological reasons for such highly regulated and compartmentalized biosynthesis of epidermal EETs during the terminal stages of keratinocyte differentiation remain to be determined.
The general importance of fatty acid metabolism in keratinocyte differentiation was demonstrated by studies of essential fatty acid-deficient epidermal keratinocytes. Essential fatty acid supplementation induced differentiation of these keratinocyte cultures (39,40), and altered the metabolism and cellular levels of active retinoids (41), another key regulator of keratinocyte growth and differentiation. It is easy to imagine that CYP2B19-dependent arachidonic acid metabolites might function in vivo by participating in Ca 2ϩ -dependent intracellular signaling pathways, since EETs are well known to mediate the mobilization of Ca 2ϩ , Na ϩ , K ϩ , and H ϩ in renal and endothelial cells (9,10). Also, ongoing studies have shown that purified EET regioisomers induce intracellular Ca 2ϩ mobilization responses in keratinocyte-derived cell lines. Clearly changes in intracellular Ca 2ϩ concentrations are somehow critical to the later stages of keratinocyte differentiation in the skin, whether effected by changes in extracellular Ca 2ϩ concentrations in vitro (42,43) or by endogenous regulatory mechanisms in vivo. Although the in vivo function of CYP2B19 remains to be proven, it is conceivable that its arachidonic acid monooxygenase activities might serve in a signaling pathway that ultimately leads to the biosynthesis of barrier lipid precursors or programmed cell death. Since CYP2B19 expression can be induced by Ca 2ϩ in primary epidermal keratinocytes in a differentiation-specific manner, it is possible to investigate potential functions of this enzyme, and effects of its specific metabolites, in an in vitro system. Future studies will be aimed at proving whether CYP2B19 catalytic activities are required for the terminally differentiated phenotype of skin keratinocytes. FIG. 7. Northern analysis of effects of divalent cations on mRNA levels of CYP2B19 and profilaggrin, two markers of late differentiation, in normal murine epidermal keratinocytes. At 80% confluence, primary keratinocyte cultures were treated with media containing cations as indicated: A and B, Mg 2ϩ (5 mM) or Sr 2ϩ (0.125 mM and 5 mM) were added to media containing 0.05 mM Ca 2ϩ . C and D, Ca 2ϩ was added to media at 1.25, 0.5, and 0.125 mM final concentration (0.05 mM Ca 2ϩ for 0 h control cultures). Data are expressed as % of 0 h control values, normalized for the relative amount of cyclophilin mRNA.