Delayed Re-epithelialization in Ppm1a Gene-deficient Mice Is Mediated by Enhanced Activation of Smad2*

Background: The in vivo function of Ppm1a in mammals remains unknown. Results: Mice lacking Ppm1a developed normally but showed delayed re-epithelialization with retarded keratinocyte migration caused by overactivation of Smad2 during cutaneous wound healing. Conclusion: Ppm1a, through suppressing Smad2-mediated signaling, plays a critical role in re-epithelialization. Significance: We provided the first direct and critical genetic evidence of the in vivo role of Ppm1a. Protein phosphatase magnesium-dependent 1A (PPM1A), a protein serine/threonine phosphatase, controls several signal pathways through cleavage of phosphate from its substrates. However, the in vivo function of Ppm1a in mammals remains unknown. Here we reported that mice lacking Ppm1a developed normally but were impaired in re-epithelialization process during cutaneous wound healing. Specifically, complete or keratinocyte-specific deletion of Ppm1a led to delayed re-epithelialization with reduced keratinocyte migration upon wounding. We showed that this effect was the result of an increase in Smad2/3 phosphorylation in keratinocytes. Keratinocyte-specific Smad2 deficient mice displayed accelerated re-epithelialization with enhanced keratinocyte migration. Importantly, Smad2 and Ppm1a double mutant mice also exhibited accelerated re-epithelialization, demonstrating that the effect of Ppm1a on promoting re-epithelialization is mediated by Smad2 signaling. Furthermore, the decreased expression of specific integrins and matrix metalloproteinases (MMPs) may contribute to the retarded re-epithelialization in Ppm1a mutant mice. These data indicate that Ppm1a, through suppressing Smad2 signaling, plays a critical role in re-epithelialization during wound healing.

Cutaneous wound healing comprises three well-defined temporally overlapping stages: inflammation, new tissue formation with synthesis of new connective tissue and epithelial wound closure (re-epithelialization), and remodeling (1). Re-epithelialization, which is responsible for the rapid restoration of an intact epidermal barrier with neo-epidermis, necessitates keratinocyte migration, proliferation, and differentiation (2). Re-epithelialization is influenced by a combination of growth factors, including hepatocyte growth factor (HGF), fibroblast growth factors (FGFs), epidermal growth factor (EGF), heparin-binding epidermal growth factor (HB-EGF), as well as transforming growth factor-␤ (TGF-␤) in a temporospatial manner (2).
TGF-␤ binds to its receptors at the cell surface, facilitating phosphorylation of the type I receptor (T␤RI) by the type II receptor (T␤RII). The activated T␤RI then phophorylates the receptor-activated Smads (R-Smads), Smad2 and Smad3, at two serines in their C-terminal SXS motifs. The phosphorylated Smad2/3 complex associates with the common Smad (Co-Smad), Smad4, which translocates to the nucleus and regulates expression of specific target genes, leading to a particular biological response (3,4). Studies from TGF-␤1 or Smad4 knockout mice and activin A transgenic mice have indicated that TGF-␤ signaling pathway promotes re-epithelialization during cutaneous wound healing (5)(6)(7). However, overwhelming evidence from genetic alteration of other TGF-␤/Smad signaling components, such as overexpression of TGF-␤1, BMP6, Smad2, or dominant negative type II TGF-␤ receptor, and loss of Smad3, supports the inhibitory effects of TGF-␤ signaling on re-epithelialization (8 -15). Recently, Dpr2, ␣3-integrin and ␤3-integrin have been shown to regulate the rate of re-epithelialization by modulating TGF-␤ signaling (16 -18). Therefore, identifying new regulators of TGF-␤/Smad pathway involved in re-epithelialization will be helpful for better understanding of the sophisticated roles of TGF-␤ signaling in wound healing.
In the present study, we generated complete Ppm1a gene knock-out mice and found that Ppm1a was dispensable for normal embryonic and postnatal development. We, therefore, analyzed the wound healing phenotypes of two in vivo models of Ppm1a deficiency and demonstrated that Ppm1a promoted reepithelialization by suppressing Smad2-mediated signaling.

EXPERIMENTAL PROCEDURES
Generation of Ppm1a Mutant Mice-A 13 kb genomic fragment, comprising the 4 kb 5Ј-region, the LoxP-floxed exon 2 and exon 3, and the 3 kb 3Ј-region, were cloned into the Pfrt1 vector containing a FRT-floxed neomycin resistance cassette and a TK cassette. The neomycin cassette was inserted 300 bp upstream of exon 2, and the other LoxP was inserted 3.5 kb downstream of exon 3. NotI-linearized targeting vector was electroporated into TC1 mouse embryonic stem (ES) cells (27). Correct targeted ES cells were confirmed by Southern blot analysis. The chimeric mice were generated by microinjection of targeted ES cells into C57BL/6J blastocysts.
In Vivo Wound-healing Experiments-Littermates at 6 -8 weeks of the same sex were used. Four 4-mm full-thickness cutaneous biopsy punch wounds were generated in the back skin of the gene-deficient mice and control littermates (16). The mice were killed at different time points (3, 5, 7 days) after injury and wounds were excised including 2 mm of the epidermal margins for RNA or protein isolation. Wounded tissue was removed and fixed overnight in cold 4% PFA for paraffin or snapfrozen in OCT (Thermo Lifesciences). The wound width was quantitated by measuring the distance between epithelial edges across the wound bed. The width of wound sites minus the wound width was neo-epithelial length.
Histological Analysis, Immunohistochemistry, and Immunofluorescence-Mouse tissues were embedded in paraffin, sectioned at 6 m, and stained with hematoxylin and eosin (H&E) or analyzed by immunological methods. For BrdU labeling, mice were injected intraperitoneally with 100 g/g body weight of BrdU (Sigma) 2 h before sacrifice. For immunohistochemistry and immunofluorescence, the slides were incubated with K14 antibody (1:1000, Covance) and BrdU antibody (1:300, Abcam), followed by incubation with FITC-conjugated antirabbit IgG and TRITC-conjugated anti-rat IgG, or biotin-conjugated goat anti-rabbit IgG (all 1:100, Zhongshan, Beijing). Signals were detected by DAB (Zhongshan, Beijing) staining or directly by fluorescence microscopy. Slides were observed under a Nikon E600 microscope (Tokyo, Japan) with a digital camera.
Preparation and in Vitro Culture of Keratinocytes-Primary mouse keratinocytes were isolated from skin of newborn mice. Full thickness skin was treated with 5 mg/ml Dispase (Invitrogen) solution overnight (14ϳ16 h) at 4°C. The epidermis was gently separated from the dermis and transferred onto the surface of the drop of TrypLE Select (Invitrogen) with the basal layer downward. The epidermis was incubated for 20 -30 min at room temperature, followed by suspending in Keratinocyte-SFM medium with supplements (Invitrogen). Cells were separated from the epidermis and incubated in dishes coated with coating solution (10 g/ml fibronectin, 1% v/v vitrogen 100 collagen, 100 g/ml BSA, 10 mM HEPES) at 34°C in 5% CO 2 . Attached cells were further cultured in fresh Keratinocyte-SFM medium with supplements and 5% fetal bovine serum (Hyclone), which was replaced once every 2 days.
In Vitro Scratch Closure Assay-The 80 -90% confluent keratinocytes in 6-well plates were treated with 10 g/ml mitomycin C (Sigma) for 2 h to remove the influence of cell proliferation. Cells were then scratched with a pipette tip. Cell migration was monitored by microscopy at 0 and 48 h after wounding. The images acquired for each sample can be further analyzed quantitatively. For each image, distances between one side of scratch and the other can be measured. By comparing the images from time 0 to 48 h, we obtain the distance of each scratch closure on the basis of the distances that are measured by software.
Statistical Analysis-All values were expressed as mean Ϯ S.D. Statistical analysis of the data were performed by twotailed Student's t test. *, p Ͻ 0.05; **, p Ͻ 0.01.

RESULTS
Normal Embryonic Development in Complete Ppm1a-deficient Mice-The Cre-LoxP site-specific recombination system was used to target Ppm1a for complete and conditional deletion in mouse. To generate the Ppm1a targeting construct, a neomycin (neo) resistance cassette flanked by two FRT sites and a single LoxP site is inserted into the upstream of exon 2 of Ppm1a gene, and the second LoxP site was introduced into intron 3 (supplemental Fig. S1A). This targeting construct was electroporated into mouse ES cells. The homologous recombinant ES clones were determined by Southern blotting analysis (data not shown), and were injected into blastocysts to generate chimeric mice. The chimeric mice exhibited germline transmission of the LoxP-floxed Ppm1a allele (Ppm1a fl/ϩ ). Ppm1a fl/ϩ mice were bred with EIIa-Cre mice (28) to generate complete Ppm1a-deficient mice (Ppm1a Ϫ/Ϫ ) (supplemental Fig. S1B).
Murine Ppm1a was expressed in a ubiquitous manner, including in various adult tissues and embryos (supplemental Fig. S1C). To evaluate the efficiency of Ppm1a excision in Ppm1a Ϫ/Ϫ mice, we detected the RNA from a number of tissues by RT-PCR analysis, and verified that no detectable Ppm1a transcripts were found in liver, stomach, and skin tissues from Ppm1a Ϫ/Ϫ mice (supplemental Fig. S1D).
Ppm1a Ϫ/Ϫ mice were born with Mendelian frequencies, and presented no genotype dependent differences in gross morphological abnormalities, or fertility. These findings indicated that Ppm1a was not essential for normal embryonic development.
Deletion of Ppm1a Led to Delayed Re-epithelialization during Wound Healing via Decreased Keratinocyte Migration-To investigate the role of Ppm1a during cutaneous wound healing, full-thickness 4-mm punch biopsies were made in the back skin of wild-type mice. Interestingly, we found that, after injury, the amount of Ppm1a at wound edge was decreased by 25 and 80% at days 1 and 3, respectively (Fig. 1A). This raises a question of whether or not Ppm1a is required for normal wound healing. Thus, we explored the consequence of Ppm1a deletion during wound healing. As expected, Ppm1a protein was detectable in whole skin tissue or in the isolated primary keratinocytes from control heterozygous mice (Ppma ϩ/Ϫ ), but was undetectable from Ppm1a Ϫ/Ϫ mice (Fig. 1B). The healing process was analyzed at days 3, 5, and 7 days after injury (n ϭ 5 for each time point and genotype). Wounds were immunostained with Keratin 14 antibody for examination of re-epithelialization (measured by wound widths and lengths of neo-epithelium). Because the re-epithelialization rate in heterozygous mice (Ppm1a ϩ/Ϫ ) was comparable with that in wild-type mice (data not shown), we used the Ppm1a ϩ/Ϫ mice as controls in this study. We observed that the re-epithelialization in Ppm1a Ϫ/Ϫ mice was markedly delayed compared with that of control littermates (Fig. 1C). The wound widths were larger and neo-epithelial lengths were shorter at days 3, 5, 7 in Ppm1a Ϫ/Ϫ healing skin than controls (Fig. 1, D and E; p Ͻ 0.05). In addition, the granulation tissue formation in both Ppm1a-null and control mice occurred comparably at day 5 (data not shown).
Re-epithelialization is mainly dependent on keratinocyte migration and keratinocyte proliferation (2). There was a comparable number of cells proliferating at wound edge from Ppm1a Ϫ/Ϫ and control mice, as determined by bromodeoxyuridine (BrdU) labeling (Fig. 2, A and B). To assess whether loss of Ppm1a affected keratinocyte migration, primary keratinocytes were isolated from Ppm1a Ϫ/Ϫ and control newborn mice, and an in vitro scratch assay, in which a "wound" is introduced in cultured monolayer keratinocytes, was performed. The rate of scratch closure from Ppm1a Ϫ/Ϫ keratinocytes was notably slower by 37% than that from controls at 48 h after wounding (Fig. 2, C and D; p Ͻ 0.01), suggesting that delayed re-epithelialization observed in Ppm1a mutant mice might be attributable to reduced keratinocyte migration.
To clarify whether or not Ppm1a-deficient keratinocytes, but not other cell types, mainly accounted for the delayed re-epithelialization observed in Ppm1a Ϫ/Ϫ mice, we generated keratinocyte-specific Ppm1a mutant (K5-Cre;Ppm1a fl/fl ) mice by breeding Ppm1a fl/fl mice with K5-Cre transgenic mice (29,30). Dramatic reduction of Ppm1a was demonstrated in the isolated Ppm1a mutant primary keratinocytes (supplemental Fig. S2A). Similar to complete Ppm1a-null mice, K5-Cre;Ppm1a fl/fl mice also exhibited impaired re-epithelialization and retarded keratinocyte migration (supplemental Fig. S2, B-F), suggesting that Ppm1a in keratinocytes was responsible for maintaining the normal rate of re-epithelialization during wound healing.
Overactivation of Smad2 in Ppm1a-deficient Mice-Previous studies have shown that Ppm1a could modulate Smad and PI3K/Akt signaling pathways, both of which are involved in wound healing process and keratinocyte migration (9,12,19,21,31). So we checked the expression levels of p-Akt and p-Smad2/3 at wound edge from Ppm1a-null and control mice by Western blot. p-Akt expression remained unchanged in Ppm1a-null mice compared with control mice post-wounding (supplemental Fig. S3A). Both p-Smad2 and p-Smad3 were upregulated in control mice 3 days after injury, and were profoundly increased in Ppm1a-null mice compared with that in controls before and after wounding (Fig. 3A). The up-regulation of p-Smad2 was also detected at wound edge of keratinocyte-specific Ppm1a mutant (K5-Cre;Ppm1a fl/fl ) mice (Fig. 3B). Consistently, the increased expression of p-Smad2/3 was observed in the isolated primary keratinocytes of K5-Cre; Ppm1a fl/fl mice (Fig. 3C). In addition, Smad2 expression was not changed after Ppm1a deletion (supplemental Fig. S3B). These data suggested that enhanced activation of Smad2/3 signaling in keratinocytes possibly contributed to the delayed reepithelialization in Ppm1a-null mice.
Keratinocyte-specific Smad2 Deletion Resulted in Accelerated Re-epithelialization-To determine if keratinocytic Smad2 plays a critical role during wound healing, we generated keratinocyte-specific Smad2 mutant (K5-Cre;Smad2 fl/fl ) mice by breeding Smad2 fl/fl mice (32) with K5-Cre transgenic mice. Although mice with keratinocyte-specific Smad2 deletion exhibit accelerated formation and malignant progression of chemically induced skin tumors (33), the consequence of keratinocyte-specific Smad2 deletion in wound healing remains unknown. Western blot confirmed that Smad2 expression was significantly down-regulated in primary keratinocytes of K5-Cre;Smad2 fl/fl mice (Fig. 4A). In contrast to Ppm1a-null mice, the rate of re-epithelialization was markedly accelerated in K5-Cre;Smad2 fl/fl mice compared with controls (Fig. 4, B-D). The wound widths were smaller and neo-epithelial lengths were longer at days 3, 5, 7 in K5-Cre;Smad2 fl/fl healing skin compared with controls (Fig. 4, C and D; p Ͻ 0.05, n ϭ 5). Similar to that in Ppm1a-null mice, the rate of proliferative keratinocytes at wound edge was equivalent in both K5-Cre; Smad2 fl/fl and control mice (Fig. 4, E and F). Moreover, migration of Smad2 deficient keratinocytes was significantly enhanced by 41% compared with control cells at 48 h in an in vitro scratch assay (Fig. 4, G and H; p Ͻ 0.01), indicating that accelerated re-epithelialization observed in Smad2-deficient mice was associated with enhanced keratinocyte migration.
Simultaneous Deletion of Smad2 and Ppm1a Accelerated Reepithelialization Rate-To further confirm that whether or not Smad2 is required for Ppm1a-regulated re-epithelialization, we generated Smad2 and Ppm1a double mutant mice. Western blot confirmed that the expression of both Smad2 and Ppm1a was significantly downregulated in primary keratinocytes of K5-Cre;Smad2 fl/fl ;Ppm1a Ϫ/Ϫ mice (Fig. 5A). As expected, the rate of re-epithelialization was dramatically accelerated in K5-Cre;Smad2 fl/fl ;Ppm1a Ϫ/Ϫ mice compared with controls (Fig. 5, B-D; p Ͻ 0.05). Furthermore, the rate of scratch closure from double mutant keratinocytes was notably faster by 45% than that from controls at 48 h after wounding (Fig. 5, E and F; p Ͻ 0.01). The above data demonstrated that the effect of Ppm1a on promoting re-epithelialization is mediated by Smad2 signaling, at least during cutaneous wound healing.

Altered Expression of Integrins and MMPs in Ppm1a Mutant
Mice-During the re-epithelialization process, keratinocytes should be able to detach from the underlying basal lamina and migrate through the fibrin and extracellular matrix (ECM) meshwork of the wound. Two classes of molecules, the integrins and matrix metalloproteinases (MMPs), have been shown to facilitate keratinocyte migration during wound healing (34,35). We examined integrins and MMPs expression in Ppm1a single mutant, Smad2 single mutant or Ppm1a and Smad2 double mutant mice by real-time PCR assay. At day 3 post-injury, the expression levels of integrin ␤1, integrin ␣5, integrin ␣v, MMP13, and MMP2 were apparently down-regulated at K5-Cre; Ppm1a fl/fl wound edge compared with that at control wound edge (Fig. 6; p Ͻ 0.01). Conversely, the expression of the above molecules were up-regulated in Smad2 single deficient mice or Ppm1a and Smad2 double mutant mice compared with corresponding controls (Fig. 6; p Ͻ 0.05). In addition, similar changes of the above integrins and MMPs were found in complete Ppm1a gene knock-out mice (supplemental Fig. S4). These expression differences of key effectors of keratinocyte movement likely contributed to the delayed wound closure in Ppm1a-deficient mice.

DISCUSSION
There are three novel findings in the present study. First, deletion of Ppm1a is dispensable for embryonic development and adult homeostasis; second, deletion of Ppm1a leads to delayed re-epithelialization during wound healing  DECEMBER 9, 2011 • VOLUME 286 • NUMBER 49 via decreased keratinocyte migration; third, Ppm1a can repress overactivation of Smad2 signaling, which is inhibitory for epidermal cell migration in wound healing. Taken together, we identified, for the first time, the pathophysiological role of Ppm1a in re-epithelialization process during wound healing.

Loss of Ppm1a Causes Delayed Wound Healing
We provided the first and critical in vivo genetic evidence showing a role of Ppm1a in promoting keratinocyte migration during wound healing. Ppm1a has been shown to regulate cell proliferation and cell differentiation in vitro (19,21). A recent study has revealed that PPM1A is involved in human cytrophoblast cell migration (36). Although PPM1A decreases the growth-inhibitory effect of TGF-␤ via dephosphorylating Smad2/3 in HaCaT cell (19), an immortalized human keratinocyte cell line, the role of PPM1A in keratinocyte migration is largely unknown. In this study, skin development and homeo-stasis were normal in both complete and conditional keratinocyte specific Ppm1a mutant mice, which suggested functional redundancy of other members of Ppm1 family in skin formation. However, under conditions of the cutaneous stress response, the function of Ppm1a during re-epithelialization process is indispensable. Deletion of Ppm1a led to delayed reepithelialization due to the retarded keratinocyte migration, but not due to altered keratinocyte proliferation ( Figs. 1 and 2). More importantly, we found that the phenotypes were identical between complete and epiderm-specific Ppm1a mutant mice, suggesting that Ppm1a in keratinocyte, but not in other cell types, is required for normal re-epithelialization process. Consistent with the notion, Ppm1a is expressed at high levels in epidermis, but at very low levels in dermal fibroblasts, vessels, and microvasculature (37).
Growing evidence has shown that Ppm1a could dephosphorylate its distinct targets to control particular cellular activities (19,21). Here, we first exemplified that Ppm1a could suppress Smad2 signaling, thereby promoting keratinocyte migration and re-epithelialization during wound healing. Deletion of Ppm1a elevated the expression level of p-Smad2 at wound edge post-wounding (Fig. 3, A and B) and in primary isolated keratinocytes (Fig. 3C). Consistently, a previous study has indicated that overexpression of Smad2 is inhibitory for re-epithelialization (12). Delayed wound healing because of retarded keratinocyte migration has been found in the Smad2 transgenic mice (12), which is quite similar to that observed in Ppm1a mutant mice. To support this notion, in this study, we showed that keratinocyte-specific Smad2 deletion led to accelerated re-epithelialization caused by the enhanced keratinocyte migration. Similar to the phenotypes found in Smad2 single mutant mice,  Smad2 and Ppm1a double mutant mice exhibited accelerated re-epithelialization rate, strongly suggesting that Smad2 is required for Ppm1a-regulated re-epithelialization. Moreover, the inverse expression patterns of specific integrins and MMPs in Ppm1a-null and Smad2-null mice or Ppm1a and Smad2 double mutant mice suggested that Ppm1a might modulate expression of the above effectors possibly through its Smad2-antagonizing activity. During wound healing, Ppm1a expression at wound edge was down-regulated (Fig. 1A), and accordingly, p-Smad2/3 expression was gradually increased post-wounding (Fig. 3A). This raises an interesting question of what mechanism by which Ppm1a was regulated, thereby influencing Smad2 signaling. Further study involved in regulation of Ppm1a would provide new insights into the role of Ppm1a in fine tuning diverse signals during wound healing process.
Our present study provided new evidence supporting the notion that TGF-␤/Smad signaling plays an inhibitory role in re-epithelialization during skin wound healing, at least in the full thickness wound scenario. We found that mice deficient in Ppm1a, which exhibited overactivation of Smad2, displayed significantly decreased re-epithelialization, demonstrating that epidermal TGF-␤/Smad signaling repressed re-epithelialization. Smad2 mutant mice showed the opposite phenotypes, further supporting the point of view. Consistently, many previous in vivo studies have indicated that TGF-␤/Smad signaling suppresses wound re-epithelialization. The re-epithelialization of full thickness wounds is delayed in TGF-␤1, BMP6, or Smad2 overexpressing transgenic mice (10,(12)(13)(14), and accelerated in mice expressing a dominant-negative type II TGF-␤ receptor (8) or Smad3-null mice (9). In contrast, the results from the mice with deletion of TGF-␤1 or Smad4, and mice overexpressing activin A, suggest a positive role of TGF-␤/Smad signaling in re-epithelialization (5-7). These obvious discrepancies might largely due to various effects of TGF-␤ signaling mediated by different cell types involved in wound healing. Future studies using various cell-and stage-specific genetically modified mice will be crucial to better define the effect of TGF-␤/ Smad signaling on wound re-epithelialization.