IRX1 is expressed in the basal stem cell layer of murine and human gingiva and a potential mesenchymal stroma cell niche. To analyze IRX1 expression in the oral epithelial basal layer in mice, we first performed whole-mount X-gal staining of mouse mandibular tissue from our recently generated heterozygous (Irx1+/–; HET) mouse strain, in which exons 1–4 (the entire coding sequence) on 1 allele of the endogenous Irx1 gene were replaced by a LacZ reporter gene and neomycin resistance gene (10). We found that IRX1 is expressed in the gingiva around the molar rows (Figure 1A). Sections from the X-gal–stained tissue showed that IRX1 was expressed in the basal and immediate suprabasal layer of the GE, as well as in the outer enamel epithelium (OEE) (Figure 1B). Immunofluorescence (IF) staining of adult mice showed that IRX1 was expressed in the junctional epithelium (JE) and the basal layer and immediate suprabasal layer of the gingiva in front of the first molar. IRX1 antibody staining overlapped with X-gal staining in the Irx1+/– samples, indicating the specificity of the IRX1 antibody. The expression level of IRX1 was reduced in the Irx1+/– Het gingiva under homeostasis due to the deletion of 1 allele, but the expression pattern did not change compared with WT (Figure 1C).

IRX1 is expressed in the basal layer of mouse gingiva. (A) Whole mount X-gal staining (blue, IRX1 expression) in mouse mandible in Irx1+/– adult mice at 2 or 4 weeks old (wo). Scale bar: 500 μm. (B) Eosin staining of sections from X-gal–stained adult mandible at 2 wo. Scale bar: 100 μm. (C) Representative IF staining of IRX1 and LacZ in Irx1+/+ and Irx1+/– 3-month-old gingiva area in front of the first molar. Blue staining represents nuclei. Scale bar: 100 μm. M, molar; OEE, outer enamel epithelium; T, tooth; GE, gingival epithelium; JE, junctional epithelium.

In addition, we analyzed 2 deidentified human gingiva samples for IRX1 expression and found that IRX1 was expressed in the basal and immediate suprabasal layer of the epithelium as well as in a stroma mesenchymal cell niche (Supplemental Figure 1, A–F; supplemental material available online with this article; https://doi.org/10.1172/jci.insight.179815DS1). As a control, H&E-stained human gingival samples are shown depicting normal histology (Supplemental Figure 2). To initially determine if these mesenchymal cells were contained in a stem cell niche, we costained with STRO-1, a marker for mesenchymal stem cells (20, 21). Both IRX1 and STRO-1 were coexpressed in these cells, suggesting a possible stem cell niche in the mesenchymal stromal layer (Supplemental Figure 1G).

Because GE is a keratinized stratified epithelium similar to the skin epidermis (1), we asked if IRX1 had a similar expression pattern in the skin epithelium layer. Interestingly, we found that IRX1 was not expressed in the epidermis but was instead expressed in the hair follicles by performing X-gal and eosin staining of the skin tissue on the back of Irx1+/– mice at different ages (Supplemental Figure 3, A–D). In addition, IRX1 is expressed in the bulge and dermal papilla regions of the mature hair follicles, where stem cells reside contributing to skin homeostasis and the healing of cutaneous wounds (22–24). JE and the basal layer of GE serve as pools of gingival epithelial stem/progenitor cells (25–28). Considering that IRX1 is highly expressed in these regions, it suggests a potential role for IRX1 in stem/progenitor cell maintenance and tissue healing.

Irx1 haploinsufficiency delays gingival wound closure and reepithelization. We assessed the morphology of homeostatic tissue of Irx1+/+ and Irx1+/– mice before examining the injured tissue. The morphology, thickness, and stratification of the adult gingiva in front of the first molar in Irx1+/– mice is thinner and less developed compared with Irx1+/+ mice (Figure 2). These results suggest defects in oral epithelial stratification.

Irx1+/– mice show defective stratification of the mouse oral epithelium. (A) Schematics of the structure of the gingiva in front of the first molar. (B) Representative H&E staining of the homeostatic gingiva of a 3-month-old (mo) Irx1+/+ mouse. Scale bar: 100 μm. (C) H&E staining of the homeostatic gingiva of 3 mo Irx1+/– mouse. Scale bar: 100 μm. T, tooth; GE, gingival epithelium; JE, junctional epithelium; SC, stratum corneum; KC, keratinocyte; GM, gingival mesenchyme.

Oral epithelium, including GE, is generally recognized as primed with a set of genes, whose expression facilitates the fast and scarless wound healing process compared with cutaneous wounds (7, 29–31). The expression pattern of IRX1 indicated it is one of the genes that contribute to gingival wound healing. To test this hypothesis, we established a murine gingival wound healing model in adult (10–16 weeks old [wo]) Irx1+/+ WT and Irx1+/– (Het) mice, since the Irx1–/– null mice are early neonatal lethal (10). We extracted RNA from the unwounded gingival tissue in front of the first molar and performed quantitative PCR (qPCR). The homeostatic gingival tissue of Irx1+/– mice showed significantly lower expression of Irx1 compared with Irx1+/+ animals (n = 6) (Figure 3A). The gingival tissue in front of the first molar was removed with a biopsy punch (ø0.75 mm) (Figure 3B). Every wound was carefully confirmed under the dissection scope to ensure (a) the wound site was consistently in front of the right first molar and (b) the JE was removed (Figure 3C). To confirm the consistency of the operation, we sacrificed mice from both genotypes immediately after surgery and dissected the mandible for imaging under a dissection scope (Figure 3D). H&E staining of sagittal sections of the tissue further validated that JE and GE were both removed (Figure 3E). The wound size measurements indicated that a relatively consistent wound area (~0.45 mm2) was created in the mice (Figure 3F).

Establishment of a mouse gingival injury model. (A) The expression level of Irx1 mRNA in Irx1+/+ and Irx1+/– gingival tissues in front of the first molar assessed by qPCR. n = 6. (B and C) Operation technique for the mouse gingival injury model. Mice aged 10–16 weeks old were subjected to injury. (B) After anesthesia, the mouth was opened using a retractor and the wound was made with a biopsy punch (ø0.75 mm). (C) Picture taken from the dissection scope showing that the wound was made in front of the first molar on the right side. (D and E) The gingiva, including the junctional epithelium, was removed after the surgery. Scale bar: 500 μm. (F) Quantitation of the wound area. n = 4. M, molar.

Next, we assessed oral epithelial regeneration activity in Irx1+/+ and Irx1+/– mice at 1 dpi and 3 dpi. At 1 dpi, wounds in Irx1+/+ samples were almost closed, with an average remaining wound size of 0.08 mm2 (n = 3) (Figure 4, A, B, and G). H&E images show that the reepithelization was complete. However, the RE was not fully stratified and was absent of rete pegs, with immature formation of the JE (Figure 4C). These features indicate that the RE in Irx1+/+ mice was in the process of tissue remodeling at 1 dpi. In Irx1+/– mice at 1 dpi, however, the wounds were largely open, with an average size of 0.31 mm2 (n = 5) (Figure 4, D, E, and G). The RE was absent in the 500 μm region proximal to the first molar in H&E staining, indicating that reepithelization was incomplete (Figure 4F).

Irx1+/– mice exhibit delayed wound closure and reepithelization. Mouse mandible tissues were collected at 1 day postinjury (dpi) and 3 dpi. (A–F) Mandible tissue harvested at 1 dpi. (A–C) Representative mandible images and H&E staining show the healing of Irx1+/+ mice at 1 dpi. The wound was closed, and a thin layer of epithelium was formed. (D–F) Representative mandible images and H&E staining show the healing of Irx1+/– mice at 1 dpi. The wound is largely exposed. (G) Quantitation of the wound size in Irx1+/+ and Irx1+/– samples at 1 dpi. n = 3-5. (H–M) Mandible tissue harvested at 3 dpi. (H–J) Representative mandible images and H&E staining show the healing of Irx1+/+ mice at 3 dpi. The wound was fully closed, and epithelium layer was stratified. (K–M) Representative mandible images and H&E staining show the healing of Irx1+/– mice at 3 dpi. The wound was still open, and reepithelization was not complete. (N) Quantitation of the wound size in Irx1+/+ and Irx1+/– samples at 1 dpi. n = 4–5. Black dashed lines demarcate the gingival epithelium and mesenchyme. epi, epithelium; G, gingiva. Scale bar: 500 μm.

Immune homeostasis is essential for tissue regeneration, and immune effector cells such as macrophages are mediators of tissue regeneration (32, 33). Furthermore, exosomes derived from mesenchymal cells affect wound healing, and both macrophages and exosomes are characterized by CD63 and CD11b expression (32, 33). The initial phase of wound healing involves the migration of macrophages and exosomes to the wound site to illicit the wound healing response. To determine if Irx1 played a role in the immune response, we analyzed the wound response for CD63 and CD11b expression at 1 dpi in both Irx1+/+ and Irx1+/– mice. The expression of CD63 was unchanged in the wound area and mesenchyme of both Irx1+/+ and Irx1+/– mice at 1 dpi (Supplemental Figure 4). The expression of CD11b was unchanged in both injured mice at 1 dpi (Supplemental Figure 5). These data suggest that Irx1 does not regulate the immune response in oral wound healing.

At 3 dpi, wounds in the Irx1+/+ samples exhibited partially restored gingiva (Figure 4, H and I). H&E staining showed the formation of stratified epithelium and rete pegs, but without the JE (Figure 4J). The average remaining wound size was 0.01 mm2 (n = 5) (Figure 4N). In the Irx1+/– mice, the wounds were open and reepithelization remained incomplete (Figure 4, K and L). The regenerative front (RF) was ~600 μm to the first molar (Figure 4M). The average remaining wound size was 0.19 mm2 at 3 dpi in Irx1+/– mice (n = 4) (Figure 4N). At 7dpi, the Irx1+/+ mice had completely restored gingiva with fully developed GE and JE. However, the Irx1+/– mice had a remaining wound size of 0.02 mm2, and the development of the GE and JE was incomplete (n = 4) (Figure 4N). Together, these data show a delayed gingival wound closure and morphological recovery due to haploinsufficiency of Irx1+/‑ after injury compared with WT mice. These data demonstrate a major role for Irx1 in regulating gingival wound healing.

KC differentiation in the RE is defective in Irx1+/– HET mice at 3 dpi. During wound healing, successful reepithelization requires the rigid orchestration of KC activities, including migration, proliferation and differentiation. The delay of reepithelization made us ask if KC activities were affected in the RE.

Under homeostasis, epithelial stem/progenitor cells residing in the basal layer committed to the differentiation trajectory move upward and differentiate into KCs in suprabasal layers. As KCs undergo terminal differentiation, they become corneocytes, a group of anucleate cells that eventually form the stratum corneum layer (Figure 5A) (2, 34, 35). Although the exact KC activities during wound healing in the gingiva are still largely unknown, several studies have demonstrated how KC activities are regulated during cutaneous wound healing. The differentiation process during reepithelization of basal cells transitioning into suprabasal cells (spinous and granular KCs) occurs in concert with migration, in a wound time–specific and proximity-specific manner (36–39). Whereas the formation of the stratum corneum is suspended in the wound-activated KCs, it is restored when reepithelization is complete after injury (34, 38–40). By performing H&E staining, we observed at 3 dpi that the stratum corneum layer was formed in the RE of Irx1+/+ mice, including the region of the RF (Figure 5B). In the RF and the wound periphery of Irx1+/– mice, however, the stratum corneum layer was not formed, as evidenced by the presence of nucleated cells on the surface of the RE (Figure 5C). The lack of stratum corneum indicates an incomplete KC terminal differentiation in the Irx1+/– mouse RE. To evaluate the terminal differentiation in both genotypes, we costained for the epithelial marker E-cadherin and terminal differentiation marker Loricrin at 3 dpi (Figure 5, D–G). In uninjured control Irx1+/+ mouse gingiva, E-cadherin was expressed mainly in the basal and spinous layer, and Loricrin marked the differentiated cell types in the suprabasal layer (Figure 5D). In the RE, E-cadherin extended its expression in all cells of the RF. Although the thickness of the RE was not recovered to homeostatic level, Loricrin expression was already observed in the differentiated KC types from the wound periphery through the RF in Irx1+/+ mice (Figure 5E). In the homeostatic epithelium of Irx1+/– uninjured mice, a similar expression level and pattern of E-cadherin and Loricrin was observed compared with Irx1+/+ WT mice (Figure 5F). Nevertheless, the RF in Irx1+/– mice did not express Loricrin, as its expression can only be seen in the distal part of the wound periphery (Figure 5G). These data indicate that KC terminal differentiation was defective at 3 dpi in RE of Irx1+/– mice.

Irx1+/– mice show delayed keratinocyte differentiation in the RE at 3 dpi. (A) Schematics showing the structure of stratified epithelium in the gingiva. Keratinocytes originating from the basal layer differentiate and move vertically to become mature keratinocytes to form spinous layer and granular layer. The stratum corneum layer consists of dead keratinocytes, indicating a successful differentiation process. (B and C) Representative H&E staining images showing the regenerative front (RF) during reepithelization in Irx1+/+ and Irx1+/– mice at 3 dpi. Scale bar: 100 μm. (B) The regenerated epithelium of Irx1+/+ at 3 dpi. Arrows indicate the fully formed stratum corneum. (C) The regenerated epithelium of Irx1+/– at 3 dpi. Arrowheads indicate the nucleus-containing cells in the superficial layer of the regenerated epithelium. The regenerative front is ~600 μm to the first molar. The stratum corneum is absent at this stage in Irx1+/–. (D–G) Representative IF staining images of Loricrin and E-cadherin in control and injured gingival tissue of Irx1+/+ and Irx1+/– mice at 3 dpi. Blue staining represents nuclei. E-cadherin (red) marks the epithelial layer. (D) Loricrin expression was examined in the Irx1+/+ control gingiva. Loricrin marks the spinous and granular keratinocytes. (E) Loricrin expression was examined in the regenerated epithelium in Irx1+/+ at 3 dpi. (F) Loricrin expression was examined in the Irx1+/– control gingiva. Loricrin marks the spinous and granular keratinocytes. (G) Loricrin expression was examined in the regenerated epithelium in Irx1+/– at 3 dpi. Arrowheads indicate Loricrin expression in the distal region of the regenerated epithelium. The regenerative front is ~600 μm to the first molar. Image area covers the region of injury. Black or white dashed lines demarcate the gingival epithelium and mesenchyme. Scale bar: 100 μm. RF, regenerative front; T, tooth.

We next wanted to determine the mechanism of defective KC differentiation in Irx1+/– mice . The oral mucosa is primed with genes that are not normally expressed in the epidermis, including Sox2 and Pitx1 (7). Epidermal overexpression (OE) of Sox2 results in a less differentiated but more wound-activated phenotype, which accelerates cutaneous wound healing by activating proliferation and migration and by suppressing differentiation (8). We performed IF staining of SOX2 to assess its expression in the RE at 3 dpi (Figure 6, A–D). SOX2 protein was expressed in the basal and spinous layers in uninjured Irx1+/+ mice (Figure 6A). At 3 dpi, the expression of SOX2 in the wound periphery was similar to uninjured controls and the layers of the RF expressed SOX2 (Figure 6B). The uninjured Irx1+/– mouse oral epithelia showed decreased SOX2 expression especially near the JE compared with Irx1+/+epithelia (Figure 6C). Interestingly, the Irx1+/– Het injured mice demonstrated ectopic expression of SOX2 in all layers of the RE (Figure 6D). Keratin 6A (Krt6a) is a direct downstream target of SOX2, which activates the Krt6a promoter in oral epithelial cells (41). KRT6A is a wound-activated keratin that is extensively upregulated in the RE during reepithelization upon wounding and returns to normal levels in the wound periphery when reepithelization is complete (42). In uninjured control epithelia, KRT6A is expressed in the spinous and granular layers and in the JE in Irx1+/+ mice (Figure 6E). However, the expression of KRT6A is decreased in these layers as well as the JE of the Irx1+/– mice (Figure 6G). Interestingly, KRT6A expression was only observed in the RF of injured Irx1+/+ mice; however, it was found in all layers of the injured Irx1+/– mice RE, even in the wound periphery at 3 dpi (Figure 6, F and H). This ectopic expression pattern of SOX2 and KRT6A can still be observed in the 7 dpi Irx1+/– mice. KRT6A is a hallmark of psoriasis, a disease resulting from ectopic proliferation and dysregulated differentiation of KCs (42–44). The epidermal differentiation markers, including Loricrin, are reduced in psoriatic lesions, which also exhibit nucleated corneocytes (44). The prolonged KRT6A expression and reduced Loricrin expression in the RE of injured Irx1+/– mice demonstrate a psoriatic phenotype of gingival wound healing in these mice. Taken together, these data suggest defective and delayed KC differentiation in the RE of injured Irx1+/– mice, which may result from the prolonged SOX2 and KRT6A expression in the wound periphery.

The regenerated epithelium of Irx1+/– mice showed prolonged high SOX2 and KRT6 expression at 3 dpi. (A–D) Representative IF staining images of SOX2 in control and injured gingival tissue of Irx1+/+ and Irx1+/– mice at 3 dpi. (A) SOX2 expression was assessed in the Irx1+/+ control gingiva. SOX2 marked the basal and spinous layer. (B) SOX2 expression was assessed in the regenerated epithelium in Irx1+/+ at 3 dpi. (C) SOX2 expression was assessed in the Irx1+/– control gingiva. SOX2 marked the basal and spinous layer. (D) SOX2 expression was assessed in the regenerated epithelium in Irx1+/– at 3 dpi. The regenerative front is ~600 μm to the first molar. (E–H) Representative IF staining images of SOX2 and Keratin 6A in control and injured gingival tissue of Irx1+/+ and Irx1+/– mice at 3 dpi. (E) KRT6A expression was assessed in the Irx1+/+ control gingiva and costained with SOX2. KRT6A marks the JE and spinous and granular keratinocytes. (F) KRT6A expression was assessed in the regenerated epithelium in Irx1+/+ at 3 dpi and costained with SOX2. (G) KRT6A expression was assessed in the Irx1+/– control gingiva and costained with SOX2. KRT6A marks the JE and spinous and granular keratinocytes. (H) KRT6A expression was assessed in the regenerated epithelium in Irx1+/– at 3 dpi and costained with SOX2. The regenerative front is ~600 μm to the first molar. Blue staining represents nuclei. White dashed lines demarcate the gingival epithelium and mesenchyme. Image area covers the region of injury. Scale bar: 100 μm. RF, regenerative front; T, tooth; JE, junctional epithelium.

Impaired KC migration in Irx1+/– HET mice. Two different models have been proposed during cutaneous wound reepithelization for KC migration: (a) migrating Keratin 5+ (KRT5+) basal KCs move into the wound bed and transform into suprabasal KCs in a unidirectional manner, and (b) epidermal cells in the wound periphery migrate into the wound bed by crawling or leaping over one another and then dedifferentiate to form basal cells (45, 46). The former model was supported by lineage tracing the activity of suprabasal KC, showing that dedifferentiation of suprabasal to basal KCs did not occur during cutaneous wound healing (36, 37, 47).

To examine if KC migration was affected during reepithelization in Irx1+/– mice, we first stained for KRT5, which labels the basal KCs in uninjured Irx1+/+ WT mouse GE and JE tissues (Figure 7A). At 3 dpi, KRT5 was expressed in the RF, as well as the basal layers at the wound periphery of injured Irx1+/+ mice (Figure 7B). Homeostatic GE and JE showed reduced KRT5 expression in Irx1+/– mice (Figure 7C). At 3 dpi, we observed reduced basal cell expression of KRT5 (Figure 7D). In addition, there were fewer KRT5+ cells in the RF, which may suggest an altered KC proliferation and migration during gingival wound healing in Irx1+/– mice. KRT5 marks the cells in the RF upon wound healing, which is necessary for KC proliferation and migration (48, 49).

Irx1+/– mice have reduced regenerated epithelium and less KRT5+ cells in the RF at 3 dpi. (A) Representative IF staining of Irx1+/+ homeostatic gingiva in front of the first molar. KRT5 marks the basal layer of gingival epithelium as well as the junctional epithelium. (B) Representative IF staining of the regenerated epithelium in Irx1+/+. KRT5 marks the cells in the basal layer at the distal area and the cells in the regenerative front. (C) Representative IF staining of Irx1+/– homeostatic gingival in front of the first molar. KRT5 marks the basal layer of gingival epithelium as well as the junctional epithelium. (D) Representative IF staining of the regenerated epithelium in Irx1+/–. KRT5 marks the cells in the basal layer at the distal area and the cells in the regenerative front. The regenerative front is ~600 μm to the first molar. Image area covers the region of injury. Blue staining represents nuclei. White dashed lines demarcate the gingival epithelium and mesenchyme. Scale bar: 100 μm.

We determined the effect of IRX1 on cell proliferation in LS-8 oral epithelial cells that stably overexpress Irx1. In these cells, CCND2 expression was not increased; however, proliferating cell nuclear antigen (PCNA) expression was increased over 2-fold compared with LS-8 cells without IRX1 OE (Supplemental Figure 6A). PCNA recruits factors for DNA replication, repair, and chromatin remodeling and is associated with cell proliferation (50). OE of IRX1 in LS-8 cells was confirmed by Western blots (Supplemental Figure 6B) and resulted in increased cell proliferation (Supplemental Figure 6C). We identified decreased expression of proliferation markers CCNA2 and MKI-67 in Irx1+/– 3 dpi RE tissues; however, CCND2 expression was increased (Supplemental Figure 6D). These data suggest that IRX1 indirectly regulates cell proliferation through the activation of other transcription factors.

To trace the basal cell activity upon wounding in the oral epithelium, we performed lineage tracing using Krt14CreERT;Rosa26mTmG;Irx1+/+ and Krt14CreERT;Rosa26mTmG;Irx1+/– mice (Figure 8). We first determined if KRT14 expression was affected in the Irx1+/– mice. We stained for KRT14 in Irx1+/+ and Irx1+/– mice and found no change in KRT14 expression in the Irx1+/– mice compared with Irx1+/+ mice (Supplemental Figure 7). Mice were injected with Tamoxifen for 2 consecutive days, and gingival injury was performed 2 days after the last dose. Tissues were then collected at 0, 3, and 7 dpi for analysis (Figure 8A). The oral epithelia basal cell layer expressing Krt14Cre was induced with Tamoxifen causing the expression of GFP from the Rosa26mTmG reporter. At 0 dpi (after the surgery), the GFP signal was mostly concentrated in the gingiva around the molar row. As a control, we show that the gingiva in front of the first molar was successfully removed after the surgery (Figure 8B). At 3 dpi, increased GFP signal at the injury site indicated the reepithelization process was near completion in Irx1+/+ WT mice (Figure 8, C and D). At 7 dpi, the GFP signal was diluted in the wound bed, indicating the turnover of the RE in Irx1+/+ mice (Figure 8, E and F). In Irx1+/– mice at 3 dpi, the GFP signal was absent at the wound area, and the GFP pattern was still mainly concentrated in the molar row, which was largely similar to 0 dpi (Figure 8, G and H). At 7 dpi, the GFP signal was expanded and could be seen at the wound bed; thus, GFP retaining cells migrated to close the wound, indicating oral tissue regeneration (Figure 8, I and J). These data demonstrate that Irx1 regulates basal cell proliferation and stratification of the layers of the gingiva during wound healing.

Lineage tracing of KRT14+ basal cells upon wounding in the oral epithelia of Irx1+/+ and Irx1+/– mice. (A) Timeline of the lineage tracing experiments for both Krt14CreERT;Rosa26mTmG;Irx1+/+ and Krt14CreERT;Rosa26mTmG;Irx1+/– groups. Mice from both groups were induced with Tamoxifen for 2 consecutive days at 12–16 weeks old and injured 2 days after the second dose. Mandibular tissues were harvested at 0, 3, and 7 dpi for imaging and further analysis. (B) A representative image showing the successful induction of Cre expression (GFP signal) and tissue removal in front of the first molar. (C and D) Representative images of the basal cell activity at 3 dpi in Irx1+/+. (E and F) Representative images of the basal cell activity at 7 dpi in Irx1+/+. (G and H) Representative images of the basal cell activity at 3 dpi in Irx1+/–. (I and J) Representative images of the basal cell activity at 7 dpi in Irx1+/–. Scale bar: 1 mm. Arrows point to the wound site.

The IF signal of GFP to assess the basal cell activities during gingival wound healing showed that cells at the wound edge migrated toward the tooth to form the RE. During healing, the RE can be separated into 3 zones according to the previous research: the RF that is highly migrative, intermediate RE (IRE) that is more proliferative than migrative, and distal RE (DRE) that has a return-to-normal gene expression signature (36, 37, 47). At 3 dpi, the basal layer and spinous layer cells were labeled with GFP in the RF of Irx1+/+ mice (Figure 9A). In the IRE, there are several GFP+ cell clusters in the basal/spinous layers. There were decreased cell clusters observed in the DRE zone (Figure 9A). Furthermore, in Irx1+/+ mice at 3 dpi, the KRT5+ cells, which represented the division of the basal cells and progenitor cell production to provide a cell source for reepithelization, were observed in the basal and spinous layers and RF (Figure 9C). In the RF of Irx1+/– mice at 3 dpi, there were less KRT5+ cells, indicating a smaller, less active RF structure compared with Irx1+/+ mice (Figure 9D). We could not distinguish the IRE and DRE in the Irx1+/– mice at this stage, demonstrating defective regenerated epithelia. Moreover, we did not observe GFP+ cells in the upper suprabasal layer in the RE of Irx1+/– mice, indicating a defect in basal cell stratification, consistent with previous results (Figure 9B). In summary, at 3 dpi, Irx1+/– mice exhibited a smaller migrative zone and less proliferative cell clusters, revealing a defect in proliferation of basal cells and stratification of differentiating cells during reepithelization.

Delayed KC proliferation and migration in Irx1+/– mice at 3 dpi. (A) Representative IF image staining for GFP in Irx1+/+ WT 3 dpi mice. Yellow arrows indicate the expanded basal/spinous population. (B) Representative IF image staining for GFP in Irx1+/– Het 3 dpi mice. Yellow arrowheads indicate the smaller basal/spinous population. The RF is 650 μm to the tooth. (C) Representative IF image for GFP and KRT5 costaining in Irx1+/+ mice. (D) Representative IF image for GFP and KRT5 costaining in Irx1+/– mice. The RF is 650 μm to the tooth. Blue staining represents nuclei. RF, regenerative front; IRE, intermediate regenerated epithelium; DRE, distal regenerated epithelium; T, tooth. White dashed lines demarcate the gingival epithelium and mesenchyme. Scale bar: 200 μm.

RNA-Seq reveals a wound-activated transcriptome in Irx1+/+ mice, which is diminished in Irx1+/– mice upon injury. To understand the molecular mechanisms of regeneration and reepithelization of oral gingival epithelia, we performed total RNA-Seq of Irx1+/+ mice at 3 dpi. We first assessed the RNA expression profile between Irx1+/+ injured and uninjured oral epithelia (Figure 10A). There were 1,117 upregulated genes and 704 downregulated genes in the injured gingiva compared with uninjured (P < 0.05; |log2 FC| > 1) (Figure 10B). By performing gene ontology (GO) analysis on the 1,117 upregulated genes, we identified several significantly regulated GO terms that related to epithelial wound healing (Figure 10C). Several small proline-rich proteins (SPRR), which belong to the epidermal differentiation complex (EDC) (51), were significantly upregulated in the injured tissue over the control in Irx1+/+ mice, except Sprr1a, which was downregulated (Figure 10D). We also assessed keratin mRNA levels from both groups. Basal cell–specific KRT5 and KRT14 as well as wound-activated KRT6, KRT16, and KRT17 were significantly upregulated upon wounding in the Irx1+/+ mice compared with controls. Krt15, which has been reported to be suppressed in the wound-activated skin epithelium, was downregulated in the injured Irx1+/+ group (Figure 10E) (52). Cyclin genes were also identified as being differentially expressed, with most of them found to be upregulated in the injured tissue, with the exception of Ccnd2 (Figure 10F). Interestingly, the Sox9 gene, a known transcription factor that regulates stem/progenitor cells in the skin, was also upregulated in the injured gingiva tissue in the Irx1+/+ mice at 3 dpi (53, 54). Previously, we reported that SOX9 expression was decreased in the Irx1–/– null embryo (10). This mRNA expression profile revealed a wound-activated gene expression feature in the injured tissue of Irx1+/+ mice compared with the uninjured control at 3 dpi. The regenerated tissue exhibited enhanced proliferation, migration, and differentiation at 3 dpi.

Upregulation of the wound-activated transcriptome in the RE of Irx1+/+ mice at 3 dpi. (A) Schematics showing the left (control) and right (injured) mandibles of Irx1+/+ mice and the tissue analyzed by RNA-Seq. n = 3. (B) Volcano plot showing differentially expressed genes between the control and injured gingiva at 3 dpi in Irx1+/+ mice. (C) GO analysis of the 1,117 significantly upregulated genes in the injured tissue over the control. Padj < 0.05. (D) The heatmap showing the relative mRNA expression levels of small proline-rich protein (SPRR) family members in the control and the injured gingiva in Irx1+/+ mice. (E) Heatmap showing the relative mRNA expression levels of keratins in the control and the injured gingiva in Irx1+/+ mice. (F) The heatmap showing the relative mRNA expression levels of genes related to proliferation in the control and the injured gingiva in Irx1+/+ mice. RE, regenerated epithelium.

In support of IRX1 regulation of SOX9 expression, we performed IF staining for SOX9 in Irx1+/+and Irx1+/–mouse oral epithelium. SOX9 is expressed in the basal cell layer in Irx1+/+ mice and decreased in Irx1+/– mice (Figure 11, A and B). To demonstrate direct IRX1 regulation of the Sox9 promoter, the Sox9 promoter containing the IRX1 binding site (5’ACAnnTGT3’) was cloned into a luciferase plasmid. As a control, the Irx1 sequence was mutated in the Sox9 promoter, and both the WT sequence and mutant sequence were tested for luciferase activity with and without IRX1 OE in LS-8 oral epithelial cells (Figure 11C). IRX1 activated the Sox9 promoter but not the Sox9 promoter with a mutated IRX1 binding site (Figure 11C). We next used ChIP experiments in OCCM30 cells, which express IRX1 and SOX9, to determine in vivo binding of IRX1 to the Sox9 promoter (Figure 11D). Primers flanking a negative control region and the IRX1 binding site were designed and used to amplify input chromatin, and used to amplify chromatin resulting from ChIP reactions using nonspecific IgG and a monoclonal IRX1 antibody. Both sets of primers amplified a product from input chromatin samples (Figure 11E), and the primers flanking the IRX1 binding site only amplified a product when the specific antibody was used to perform the ChIP. The negative control primers did not amplify a product when either antibody was used. Using qPCR, the enrichment of the chromatin containing the IRX1 binding site precipitated with the specific antibody was confirmed (Figure 11F). There was no enrichment with the negative control primer set using either antibody to precipitate chromatin (Figure 11G).

IRX1 regulates SOX9 expression. (A) Representative IF staining image showing the SOX9 expression in the homeostatic gingiva of Irx1+/+ mice. (B) Representative IF staining image showing the SOX9 expression in the homeostatic gingiva of Irx1+/– mice. (C) Irx1 plasmid or a control empty vector were overexpressed along with the WT Sox9 promoter fused to luciferase reporter or with a promoter containing a mutated canonical Irx1 binding site. n = 4. (D) A schematic of the Sox9 promoter, highlighting the IRX1 binding site and the negative control regions that were amplified by the given primer sets in the ChIP assay. (E) Gel-electrophoresis of PCR amplified products from the input and ChIP reactions using primers flanking the IRX1 binding site and the negative control region. The specific band amplified is labeled with an asterisk. (F) qPCR demonstrating an enrichment using primers flanking the IRX1 binding site when the specific antibody was used compared with the nonspecific control. n = 3. (G) There is no difference in the samples pulled down with IgG and the specific IRX1 antibody when the primer set flanking a control region lacking an IRX1 binding site is used. n = 3.

To determine the molecular mechanisms of defective reepithelization in Irx1+/– mice at 3 dpi, we observed the mRNA expression profile of the RE in Irx1+/+ mice compared with Irx1+/– mice (Figure 12A). The analysis identified 451 upregulated genes and 421 downregulated genes significantly differentially expressed in the Irx1+/– injured tissues (P < 0.05; |log2 FC| > 1) (Figure 12, B and C). We then analyzed the genes with more strict parameters (-Log10P > 3 and log2 FC < –2), and identified the top genes that were downregulated in Irx1+/– Het injured tissue compared with control injured tissue (Figure 12, C and D). Genes upregulated upon wound healing in the epithelial tissue or positively regulating wound healing were significantly downregulated in the injured Irx1+/– Het mice. Several genes, including Epgn, a ligand of EGF signaling pathway regulating KC migration and proliferation (5, 6, 55, 56); Mmp13, a matrix metalloproteinase (MMP) that regulates KC migration; and Postn, an essential gene for KC proliferation during cutaneous wound healing (57), were all decreased in Irx1+/– injured mice (Figure 12D). SPRR2E and SPRR2I regulate KC differentiation (51), DNASE2B controls DNA degradation during KC cornification and regulates their terminal differentiation (58), and IL-17C and IL-24 — encoding 2 epithelial ILs, which are highly upregulated during cutaneous wound healing in the KCs and regulate their migration and proliferation (59–61) — are all decreased in Irx1+/– injured mice (Figure 12D). PTGS2 is upregulated during skin wound healing (62, 63) and is decreased in Irx1+/– injured mice (Figure 12D).

RNA-Seq analysis showed downregulation of wound healing–related genes in the injured gingiva of Irx1+/– Het mice. (A) Schematics showing the Irx1+/+ WT and Irx1+/– Het injured mandibles at 3 dpi and the tissue collected for RNA-Seq. n = 3. (B) Heatmap showing the significantly up- and downregulated genes in the injured Irx1+/– Het mouse tissue compared with Irx1+/+ WT mouse injured tissue. (C) Volcano plot displaying differentially expressed genes between the control and injured gingiva at 3 dpi in Irx1+/+ WT mice. The right panel showed the hits with –log10 P > 3 and Log2 FC < –2. Some of the most significantly downregulated genes in the Irx1+/– mice were highlighted. (D) The heatmap showing the most downregulated genes in the Irx1+/– Het mice. (E) GO analysis of the 421 significantly downregulated genes in the injured tissue of Irx1+/– Het mice over the Irx1+/+ WT mice. Padj < 0.05. RE, regenerated epithelium.

GO analysis of the 421 downregulated genes revealed significant GO terms related to wound healing and revealed positive regulation of KC migration and KC differentiation (Figure 12E). The RNA-Seq data identify genes and pathways that are regulated by IRX1 and that are positively related to wound healing, regeneration, and reepithelization. These data demonstrate the role of IRX1 in positively regulating the wound healing process.

EPGN expression is activated during oral epithelial regeneration while its expression is decreased and delayed in Irx1+/– HET injured mice. We identified 10 genes whose activation were impaired due to Irx1 haploinsufficiency upon gingival wounding (Figure 13A). Among these genes, epithelial mitogen (Epgn) has been identified as one of the most significantly differentially expressed and was induced only in the Irx1+/+mouse injured tissue. It is a ligand of the EGF signaling pathway, which is essential to control wound healing by regulating KC migration, proliferation, differentiation, and epidermal-mesenchymal transition (5, 6, 55, 56). EPGN is induced upon injury in the cutaneous wound (47, 64). We performed qPCR of Epgn mRNA levels in 1, 3, and 7 dpi samples from both genotypes, confirming that the induction of Epgn was reduced in the Irx1+/– injured tissue (Figure 13B). EPGN protein was not expressed in the homeostatic (uninjured) GE or JE in either genotype (Figure 13, C and D). At 3 dpi, EPGN was activated in the leading edge of the RE in the Irx1+/+ mice but was not observed in the RE of Irx1+/– mice (Figure 13, E and F). At 7 dpi, EPGN expression returned to normal levels in the RE of Irx1+/+ mice, while some induction of EPGN was observed in the RF of Irx1+/– mice (Figure 13, G and H). The delayed onset of EPGN expression upon wounding in Irx1+/– mice suggests a role of EPGN in mediating Irx1 regulation of oral epithelial regeneration and reepithelization.

Decreased EPGN underlies the delayed reepithelization in Irx1+/– mice. (A) Heatmap showing the relative mRNA expression levels of a subset of the significant genes upregulated in the injured tissue of Irx1+/+ mice over uninjured Irx1+/+ mice as well as the injured tissue of Irx1+/– mice. (B) qPCR detected the mRNA level of Epgn at 1, 3, and 7 dpi. n = 3 (C and D) Representative IF staining of EPGN protein in the unwounded gingiva of Irx1+/+ and Irx1+/– mice. (E and F) Representative IF staining of EPGN protein in the injured gingiva of Irx1+/+ and Irx1+/– mice at 3 dpi. (G and H) Representative IF staining of EPGN protein in the injured gingiva of Irx1+/+ and Irx1+/– mice at 7 dpi. Blue staining represents nuclei. Scale bar: 100 μm.

To demonstrate a direct role for EPGN in cell migration, we performed a scratch assay to determine if the addition of EPGN to LS-8 oral epithelial cells would increase cell migration and scratch closure. Oral epithelial cells (LS-8) were transfected with either Irx1, sh-control (con), or sh-Irx1 (inhibits IRX1 expression) plasmids (1 μg) and a uniform scratch was made to remove cells from the plate area. Scratch closure was monitored at 0 hours (T0), T12, and T24. Inhibition of endogenous IRX1 (sh-Irx1; Supplemental Figure 6B) reduced cell migration compared with LS-8, LS-8 OE IRX1, and LS-8 sh-con cells (Supplemental Figure 8A). However, the addition of EPGN (0.5 ng/μL) to cell cultures prior to scratch assay rescued cell migration when IRX1 expression was inhibited by sh-Irx1 (Supplemental Figure 8B). Quantitation of the scratch assays are show