microRNA-31/factor-inhibiting hypoxia-inducible factor 1 nexus regulates keratinocyte differentiation

FIH-1 Expression Is Elevated in Conditions of Abnormal Epithelial Differentiation.

Skin diseases such as psoriasis and basal cell and squamous cell carcinomas have been linked to abnormal Notch signaling (17, 18); however, the mechanism(s) through which Notch participates in the etiology of these diseases is not well understood. Given the association of FIH-1 with Notch (10, 11) and previous work demonstrating that Notch plays a positive role in regulating epithelial differentiation (for reviews see refs. 2, 3), we investigated whether FIH-1 expression in psoriasis differed from that of normal epidermis in a manner that was inversely associated with Notch activity. When surveying tissue from patients with psoriasis, we observed: (i) a marked increase in FIH-1 expression (Fig. 1B) and (ii) a marked decrease in Hey 2-stained cells (Figs. 1 D and I) compared with normal skin (Fig. 1 A, C, and I). Because the corneal epithelium is frequently the target of patients with diabetic complications (19), we surveyed the corneas from patients with diabetic keratopathies and observed a similar reciprocal relationship between FIH-1 and Hey 2 expression, where FIH-1 was increased (Fig. 1F) and Hey 2 expression was decreased (Fig. 1 H and J) in the corneal epithelium from these patients compared with normal corneal epithelium (Fig. 1 E, G, and J). Furthermore, keratin 3 (K3) expression, a marker of corneal epithelial differentiation (20), was also decreased in these patients (Fig. 1H). Collectively, these observations indicate that FIH-1 expression is correlated with Notch activity and associated with impaired epidermal and corneal epithelial differentiation.

Fig. 1.

FIH-1 expression is elevated in conditions of abnormal epithelial differentiation. (A–H) Paraffin sections of human epidermis (A–D) and corneal epithelium (E–H) immunostained with an antibody that recognizes FIH-1 (A, B, E, and F), Hey 2 (C, D, G, and H), and K3 (G and H). (I) Quantification of Hey 2 staining in psoriatic epidermis. (J) Diabetic corneal epithelium. Values are means ± SD *P < 0.01.

FIH-1 Impairs Keratinocyte Differentiation, in Part, by Attenuating Notch Signaling.

To investigate whether FIH-1 can impair epidermal differentiation, we sustained expression of FIH-1 by retroviral delivery of a FIH-1 ORF (FIH-1-coding DNA sequence; FIH-1–cds) construct to submerged cultures of HEKs. Such cells were switched to high calcium, which is a well-established means of inducing stratification and differentiation in human keratinocytes (21). FIH-1–expressing HEKs elaborated minimal amounts of differentiation markers (K10, desmogelin 1, and loricrin) (22) compared with the control transduced HEKs (Fig. 2A). To investigate this FIH-1 attenuation in HEK differentiation and increase in proliferation in a more physiologically relevant manner, we used a three-dimensional organotypic raft model of human epidermis (23). FIH-1 was detected in the nucleus and cytoplasm of cells in all but the uppermost cornified layer of raft cultures generated from FIH-1–transduced HEKs (Fig. S1 A and B). The FIH-1–transduced HEKs gave rise to rafts that were morphologically and biochemically distinct (Fig. 2 B, D, F, and H and Fig. S1F) from control-transduced HEKs (Fig. 2 B, C, E, and G and Fig. S1F). For example, rafts generated from the FIH-1–transduced HEKs, harvested at day 3 (Fig. 2B), were biochemically similar to submerged cultures (Fig. 2A) with little expression of K10, desmoglein 1, and loricrin. Rafts harvested at days 7 and 12 (Fig. 2B) from the FIH-1–transduced HEKs were less differentiated although not as dramatically as the FIH-1–transduced HEKs seen at day 3 (Fig. 2B) and in submerged HEKs (Fig. 2A). This is most likely due to the complexity of the raft system and more closely reflects diseased states where K10/desmoglein 1 are reduced but still abundantly expressed (24). Morphologically, basal cells in the rafts derived from the FIH-1–transduced HEKs (Fig. 2D) contained larger nuclei with multiple nucleoli, and numerous nucleolar organizer regions (NORs) (Fig. 2D, Inset) compared with basal cells in control rafts (Fig. 2C). A greater frequency of mitotic figures and BrdU ^{stained cells were observed in the FIH-1–transduced HEK-derived rafts at days 3, 7 (}Fig. S1 C and D), and 12 (Fig. 2 D and F and Fig. S1 C and D) compared with control rafts (Fig. 2 C and E and Fig. S1 C and D). This more rapidly proliferating phenotype is consistent with the less differentiated state of the raft culture. Interestingly, submerged cultures of FIH-1–expressing HCEKs more rapidly proliferated compared with control transduced HCEKs (Fig. S2). Recently, FIH was shown to enhance in vitro proliferation of LS174 colon adenocarcinoma and A375 melanoma cells (25). Our work shows a unique function for FIH-1 in the positive regulation of keratinocyte proliferation.

Fig. 2.

FIH-1 impairs keratinocyte differentiation, in part, by attenuating Notch signaling. (A) Western blot analysis of K10, desmoglein 1 (DSG1), loricrin, and glyceraldeyhyde-3-phosphate dehydrogenase (GAPDH) in submerged cultures of primary human epidermal keratinocytes (HEKs) that were transduced with an empty vector (EV; control) or FIH-1–cds. HEKs were maintained in high calcium to induce differentiation. (B) Western blot analysis of K10, DSG1, loricrin, and GAPDH in 3D organotypic raft cultures derived from HEKs retrovirally transduced with an empty vector (EV; control) or FIH-1–cds. Raft cultures were harvested at days 3, 7, and 12. (C and D) H&amp;E and silver-stained nucleolar organizing region (Inset) analyses of 12 d, 3D organotypic raft cultures derived from HEKs transduced with an empty vector (control; C) or FIH-1–cds (D). Arrows point to mitotic figures present in FIH-1–cds transduced rafts (D). (E and F) Bromodeoxyuridine (BrdU) analysis of 12 d, 3D organotypic raft cultures derived from HEKs transduced with an empty vector (control; E) or FIH-1–cds (F) showing cells (red stain) in the S phase of DNA synthesis. DAPI (blue) is used as a counterstain. (G) Notch intracellular domain (ICD) immunostaining of 12 d, 3D organotypic raft cultures derived from HEKs transduced with an empty vector (control; G) or FIH-1–cds (H). (I) Real time qPCR analysis of Hey 2 levels in HEKs that were retrovirally transduced by either FIH-1–cds or empty vector LZRS in low calcium. Values are means ± SD of three independent experiments. *P < 0.05. (J) Luciferase assay of Notch activity reporter-transduced HEKs, ectopically expressing either FIH-1 or empty vector. HEKs were maintained in low calcium. Values are means ± SD of three independent experiments. *P < 0.05. (K). Western blot analysis of FIH-1, Notch ICD, keratin 10 (K10), and GAPDH in submerged cultures of HEKs transduced with an empty vector (control) or FIH-1–cds. HEKs were subsequently transfected with an adeno-Notch ICD in high calcium. (L) Western blot analysis of FIH-1, Notch ICD, keratin 3 (K3), plasminogen activator inhibitor-type 2 (PAI-2), involucrin (IVL), and GAPDH in submerged cultures of HCEKs transduced with an empty vector (EV; control) or FIH-1–cds. HCEKs were subsequently transfected with an adeno-Notch ICD and switched to high calcium. (M) Western blot analysis of FIH-1, Notch 1, K3, Muc1, and GAPDH in submerged cultures of HCEKs maintained in low calcium and transduced with an empty vector control or a shRNA–FIH-1 (shFIH-1) construct. Numbers below the panels represent the normalized expression signal of proteins.

Immunohistochemical analysis of the developing rafts (Fig. S1F) showed close agreement with the immuoblotting data (Fig. 2B). A marked decreased in immunostaining for keratin 10, desmoglein 1, and loricrin were noted in the FIH-1–transduced HEK-derived rafts throughout their development (Fig. S1F). However, the spatial expression patterns of these products were not altered compared with the control rafts.

In agreement with the in vivo data on the psoriatic epidermis and the diabetic corneal epithelium (Fig. 1), a correlation between FIH-1 expression and Notch activity was observed in the raft cultures with a decrease in Notch ICD-stained nuclei observed in FIH-1–overexpressing rafts (Fig. 2H) compared with control rafts (Fig. 2G and Fig. S1E). Taken together these data support the idea that our organotypic raft setting reflects the in vivo situation and that FIH-1 limits keratinocyte differentiation.

To address further whether the FIH-1 restriction in the differentiation potential of keratinocytes involved Notch signaling, we assessed Notch downstream target genes and the ability of the Notch ICD to rescue keratinocyte differentiation. In these studies, we used submerged cultures of HEKs and HCEKs, which elaborate a set of differentiation products unique to the corneal epithelium. Keratinocytes were transduced with the FIH-1–cds. We observed decreases in the expression of Hey 2 (Fig. 2I) and Notch transcriptional activity (Fig. 2J). In concert with this decrease in Notch activity, FIH-1–expressing HEKs failed to increase the expression of differentiation markers K10 and desmoglein 1 in response to a calcium switch (Fig. S3A). FIH-1–expressing HCEKs (Fig. S3B) did not express K3 and mucin 1 (Muc1), which are markers of corneal epithelial differentiation (20, 26). To confirm that FIH-1–induced impairment of differentiation involved Notch signaling, we transduced the FIH-1–expressing HEKs and HCEKs with an adenovirus encoding Notch 1 ICD. Such treatment likely enabled the excess Notch ICD to escape hydroxylation by FIH-1, which allowed translocation of Notch ICD to the nucleus. This restored the ability of the FIH-1–expressing cells to differentiate in response to a calcium switch as evidenced by the increased expression of differentiation markers (Fig. 2 K and L and Fig. S3 C–F). Having observed attenuation in keratinocyte differentiation resulting from increased levels of FIH-1, we reasoned that silencing FIH-1 in keratinocytes grown in low calcium should mimic a calcium switch as evidenced by the expression of differentiation markers. HCEKs were transduced with a shFIH-1 for 4 d and a significant knockdown of FIH-1 was achieved along with an increase in K3 and Muc1 proteins compared with a control shRNA (Fig. 2M and Fig. S3 G–I). Collectively, these findings establish that FIH-1 impairs keratinocyte differentiation, in part, by attenuating Notch signaling.

To extend these in vitro findings, we investigated the phenotype of mice with a null mutation in the FIH gene (27), focusing on the limbal epithelium. The limbal epithelium, which is the preferential site of the corneal epithelial stem cells (20, 28), interestingly has high levels of FIH-1 expression compared with the corneal epithelium (12). In concert with a lack of FIH-1 staining in the limbal epithelium (Fig. 3), we observed: (i) an increase in Hes 1 staining, (ii) an increase in Hey 2 mRNA (Fig. S4A), (iii) a decrease in p63 expression, (iv) a decrease in K15 staining, and (v) an increase in plasminogen inhibitor-type 2 (PAI-2) staining. These in vivo results confirm our in vitro findings that FIH-1 attenuates Notch signaling and negatively regulates keratinocyte differentiation.

Fig. 3.

FIH-1 alters limbal epithelial Notch signaling and differentiation in vivo. Paraffin sections of limbal epithelium from wild-type (WT) and FIH-1 null (FIH-1^{) mice were immunostained with antibodies against FIH-1, Hes 1, keratin 15 (K15), and PAI-2. Frozen sections of limbal epithelium from WT and FIH-1 null (FIH-1) mice were immunostained with an antibody against p63.}

miR-31 Targets FIH-1 to Positively Regulate Differentiation, in Part, via Notch Signaling.

We have recently demonstrated that miR-31 regulates FIH-1 in corneal epithelial keratinocytes to maintain glycogen metabolism (12). To investigate the relevance of this miR-31/FIH-1 relationship to keratinocyte differentiation, we altered miR-31 levels in submerged cultures of HCEKs (Fig. 4A). HCEKs were cultured under low and high calcium conditions and treated with either Antago-31 or an irrelevant antagomir. The expression of FIH-1 and markers of corneal epithelial differentiation were monitored. Antago-31 treatment markedly increased FIH-1 protein in HCEKs (Fig. 4A). This resulted in the inability of these cells to express K3 and Muc1 following a calcium switch (Fig. 4A). Treating HCEKs with Antago-31 to increase FIH-1 levels also decreased Notch transcriptional activity (Fig. 4B) and Hey 2 transcription (Fig. 4C). We did not see a decrease in Hes 1 (Fig. S4B). Collectively these data clearly establish a miR-31/FIH-1 nexus that regulates corneal keratinocyte differentiation, in part, via Notch signaling.

Fig. 4.

miR-31 targets FIH-1 to positively regulate differentiation, in part, via Notch signaling. (A) Western blot analysis of FIH-1, K3, mucin 1 (Muc1), and GAPDH in submerged cultures of HCEKs grown to confluence in low calcium and treated with an irrelevant antagomir (Ir-antago) or an antagomir to miR-31 (Antago-31) for 48 h and then switched to high calcium. Numbers below the panels represent the normalized expression signal of proteins. (B) Luciferase assay of Notch activity reporter-transduced HCEKs, treated with either antago-31 or an Ir-antago. HCEKs were maintained in low calcium. Values are means ± SD of three independent experiments. *P < 0.05. (C) Real-time qPCR analysis of Hey 2 levels in HCEKs grown to confluence in low calcium and treated with an Ir-antago or an Antago-31 for 48 h. Values are means ± SD of three independent experiments. *P < 0.01.

miR-31 Down-Regulates FIH-1 to Enhance Differentiation.

Having demonstrated that FIH-1 overexpression can attenuate stratified epithelial differentiation (Fig. 2 B–H), we reasoned that overexpression of miR-31 should enhance differentiation. To show this, we took advantage of a telomerase immortalized human corneal epithelial cell line (hTCEpi) (29) that constitutively expressed lower levels of miR-31 (Fig. 5A) and had higher levels of FIH-1 than primary cultures of HCEKs (Fig. 5B). Consequently, raising calcium levels in this immortalized cell line did not markedly enhance differentiation (Fig. 5C), as evidenced by the lack of expression of Muc1 and PAI-2. As predicted, lentiviral transduction of hTCEpi cells with miR-31 (hTCEpi–miR-31) reduced the expression of FIH-1 and increased the expression of Muc1 and PAI-2, 24 h after a switch to high calcium (Fig. 5C). To see whether we could recapitulate a corneal epithelium, we created three-dimensional organotypic rafts from the hTCEpi–miR-31 cells. Not surprisingly, such rafts expressed lower levels of FIH-1 (Fig. 5H) and were more differentiated than controls (hTCEpi-C; Fig. 5 D, F, and H) as evidenced by increased expression of involucrin (IVL) and PAI-2 (Fig. 5H) as well as greater numbers of K3- and PAI-2–stained keratinocytes (Fig. 5 E and G). These results clearly indicate that modulation of miR-31/FIH-1 in this cell line, in essence created a corneal epithelium. To further establish the relationship between miR-31/FIH-1 and differentiation, we overexpressed FIH-1 in the miR-31–transduced hTCEpi cells (Fig. S5). In these cultures, FIH-1 was sufficient to inhibit differentiation induced by miR-31, supporting the idea that FIH-1’s effect on differentiation is downstream of miR-31. To confirm the Notch–miR-31 connection in corneal keratinocyte differentiation, we conducted a Notch loss-of-function experiment using DAPT to inhibit Notch signaling (30). miR-31 treatment increased expression of PAI-2, involucrin, and Muc1, whereas DAPT treatment in the miR-31–transduced cells inhibited their expression (Fig. 5I). This is strong evidence that miR-31’s prodifferentiation effect is dependent on Notch activity.

Fig. 5.

miR-31 down-regulates FIH-1 to enhance differentiation. (A) Northern blot analysis of miR-31 and U6 in submerged cultures of a human limbal-derived cell line (hTCEpi) and primary human corneal epithelial keratinocytes (HCEKs). (B) Western blot analysis of FIH-1 and α-tubulin in submerged cultures of hTCEpi cells and HCEKs. (C) Western blot analysis of FIH-1, PAI-2, Muc1, and GAPDH in submerged cultures of hTCEpi cells transduced with a lenti–miR-31 or empty vector in low (0.15 mM) calcium. Cells were plated to confluence and then switched to high (1.6 mM) calcium. (D and E) Immunostaining for K3 (red-stained cells) in three-dimensional organotypic raft cultures derived from hTCEpi cells transduced with a lenti–miR-31 (E) or empty vector (D). (F and G) Immunostaining for PAI-2 (red-stained cells) in three-dimensional organotypic raft cultures derived from hTCEpi cells transduced with a lenti–miR-31 (G) or empty vector (F). DAPI (blue) is used as a counterstain to mark the nuclei. (H) Western blot analysis of FIH-1, PAI-2, involucrin, and GAPDH in 3D organotypic raft cultures derived from hTCEpi cells transduced with a lenti–miR-31 or empty vector. (I) Western blot analysis of FIH-1, PAI-2, IVL, Muc1, and GAPDH in the empty vector or miR-31–transduced hTCEpi cells maintained in calcium (1.6 mM) for 48 h and treated with or without DAPT (20 μM). Numbers below the panels represent the normalized expression signal of proteins and RNAs.

miR-31/FIH-1 Nexus Contributes to Corneal Epithelial Morphogenesis.

This study reveals an unrecognized role for FIH-1 in regulating keratinocyte fate decisions. In the epidermis, miR-31 is primarily expressed in the growing (anagen) hair follicles in mice (31). Interestingly, expression profiling indicated that with respect to the ocular anterior segmental epithelium, miR-31 is a corneal epithelial-preferred miRNA (32). As described previously, FIH-1 protein is abundantly detected in human limbal epithelium but not in human corneal epithelium (12). We propose that low levels of miR-31 and correspondingly high levels of FIH-1, which attenuates Notch signaling, may be one mechanism that maintains the relatively undifferentiated phenotype characteristic of limbal epithelial basal cells (20). Accordingly, the preferential expression of miR-31 in corneal epithelium may, in part, function to regulate the transition from the less differentiated limbal basal cells to the more differentiated corneal epithelial basal cells. It should be noted that other mechanisms may also be involved in mediating the potent roles that FIH-1 plays in keratinocyte proliferation and differentiation.

Tissues.

Blocks of paraffin-embedded specimens from seven (7) individuals with psoriasis and five (5) individuals with diabetic keratopathies were obtained from the Dermatopathology and Ophthalmic Pathology archival tissue repositories of deidentified patients seen at Northwestern University under approval of Northwestern University Institutional Review Board (IRB) and in accordance with the Declaration of Helsinki principles. Tissue sections were diagnosed by dermatopathologists and ophthalmic pathologists. Normal human skin was obtained prospectively from three (3) volunteers under approval of the IRB and in accordance with the Declaration of Helsinki principles. All volunteers gave informed consent for use of their tissues. Normal human corneas (n = 50) were obtained from the Midwest Eye Banks (Ann Arbor, MI).

FIH-1 null mice.

The FIH-1 null mice were generated by breeding FIH ^{mouse with the Ella-Cre transgenic mouse (}27). Animal experiments were approved by the Northwestern University Animal Care and Use Committee (NUACUC).

Constructs and oligonucleotides.

A cDNA encoding miR-31–resistant FIH-1 (FIH-1–cds) was ligated between BamHI and XhoI sites of the retroviral expression plasmid LZRS. The following constructs were used in this study: Human premicroRNA Expression Construct Lenti-miR-31 (System Biosciences I) and FIH-1 shRNA expression lentiviral vector pGIPZ–shRNA–FIH-1 (Open Biosystems). Antagomirs were synthesized by Dharmacon. Sequences were 5′-mA_smG_smC_smUmAmUmGmCmCmAmGmCmAmUmCmUmUmG_smC_smC_smU-Chol 3′ (Antago-31), 5′-mG_smG_smC_smAmUmUmCmAmCmCmGmCmGmUmGmCmC_smU_smU_smA-Chol 3′ (irrelevant-antago). “mN” represents 2′-O-methyl-modified oligonucleotide, the subscript “s” represents a phosphorothioate linkage, and “Chol” represents linked cholesterol.

Three-dimensional organotypic raft cultures.

Raft cultures of HEKs were produced by the Northwestern University Skin Disease Research Center’s keratinocyte core facility. Briefly, HEKs were grown to confluence in E-medium [DMEM/F12, 5% (vol/vol) FBS] with EGF (5 ng/mL) and switched to an air-medium interface for 3, 7, and 12 d (23). The raft cultures of hTCEpi cells were derived as previously described (33). The hTCEpi cells were grown in a duplex feeder system to confluence in corneal raft medium [DMEM/F12, 10% (vol/vol) FBS, 10 ng/mL EGF (Millipore), 5 mg/mL human recombinant insulin, 500 ng/mL hydrocortisone, 5 mg/mL human Apo-transferrin, 250 ng/mL isoproterenol hydrochloride (Sigma), and antibiotics] and then switched to an air-medium interface for 12 d. Duplicate 3D organotypic raft cultures were prepared for histological and biochemical analysis. All raft experiments were replicated at least three times.

Supplementary Material