Stat1 deletion rescues juvenile wasting in Sox2-Cre Pdgfrb^+/D849V mice

To determine whether the cause of death involves STAT1, we established a breeding strategy using Sox2-Cre for germline deletion of Stat1-floxed alleles and/or activation of PDGFRβ^D849V. For simplicity, mice with PDGFRβ^D849V are designated K, for the kinase domain mutation. We thus generated four offspring genotypes: Stat1^+/− (referred to here as S^+/−), Stat1^−/− (S^−/−), Pdgfrb^+/D849V Stat1^+/− (KS^+/−), and Pdgfrb^+/D849V Stat1^−/− (KS^−/−) (Fig. 1C). Importantly, while Stat1-deficient mice are nonresponsive to IFNs, they do not exhibit obvious growth or survival phenotypes unless challenged (Durbin et al. 1996; Meraz et al. 1996). Marginally increased bone formation has also been reported in S^−/− mice compared with S^+/+ mice (Sahni et al. 2001; Kim et al. 2003). As expected from previous studies, S^+/− and S^−/− mice did not have obvious phenotypes. Therefore, we focused subsequent analysis on KS^+/− and KS^−/− mice. KS^+/− mice were underweight after day 7 and died in the third week postnatal, but, interestingly, young KS^−/− mice were normal in overall appearance (Fig. 1D–F). Therefore, juvenile lethality depends on STAT1. Importantly, even though KS^−/− mice appeared normal at 3 wk of age, they succumbed to early death at ~8 wk (Fig. 1F).

PDGFRβ–STAT1 signaling drives autoinflammation

STAT1 was constitutively phosphorylated in VSMCs from Pdgfrb^+/D849V mice (He et al. 2015). To determine whether this occurs in other cell types, we used PDGF-treated primary dermal fibroblasts (DFs) from the four genotypes of mice described above. STAT1 was not detected in S^−/− and KS^−/− DFs. STAT1 became phosphorylated in S^+/− DFs only after exposure to PDGF-BB but was constitutively phosphorylated in KS^+/− DFs (Fig. 2A, lane 9). Many STAT1 target genes are ISGs, a broad class of IFN-inducible genes that includes chemokines (Ccl2, Cxcl9, Cxcl10, and Cxcl11), antigen processing and antigen presentation machinery (Psmb8 and B2m), antiviral response factors (Mx1 and Ifi27), and transcription factors (Irf1 and Stat1 itself). Unstimulated KS^+/− DFs overexpressed several ISGs (Cxcl10, Psmb8, Mx1, Irf1, and Ifi27) (Supplemental Fig. S1), but ISG expression in KS^−/− DFs was comparable with S^+/− DFs. This demonstrates that PDGFRβ^D849V activates STAT1 and thereby induces ISG expression in DFs.

Open in a separate window

Figure 2.

PDGFRβ drives autoinflammation by activating STAT1. (A) STAT1 expression and Y701 phosphorylation in primary DFs after 24 h in DMEM + 0.1% serum and stimulation with PDGF-BB. Data are representative of three experiments. The red box highlights constitutive STAT1 phosphorylation. Note that the weak top band in the total STAT1 blot is nonspecific. (B,C) Fold change in ISG mRNA levels in inguinal white adipose tissue (ingWAT) from 14-d-old mice by quantitative RT–PCR (qRT–PCR) with normalization to Gapdh. n = 3 mice per genotype. (D) Representative flow cytometry plots of CD45⁺ leukocytes in ingWAT from 14-d-old mice of the indicated genotypes. Numbers in the bottom right corner of each plot indicate the percentage of CD45⁺ cells among total cells analyzed. (E) Quantification of CD45⁺ leukocyte absolute number measured by flow cytometry. n = 3 mice per genotype. (F) Histological analysis of trichrome-stained ingWAT from 14-d-old mice. (*) P < 0.01 by Student's t-test. Data represent mean ± SD.

We also monitored ISG expression in inguinal white adipose tissue (ingWAT), a subcutaneous fat depot that atrophies in Sox2-Cre Pdgfrb^+/D849V (KS^+/+) mice (Olson and Soriano 2011). Stat1 was overexpressed in KS^+/− compared with S^+/− fat (Fig. 2B). B2m, Ccl2, Cxcl9, Cxcl10, and Cxcl11 were also overexpressed in KS^+/− fat, but their expression in KS^−/− was comparable with S^+/− and S^−/− fat (Fig. 2B,C). Next, we used flow cytometry to quantify the infiltration of CD45⁺ leukocytes into the tissue. KS^+/− fat contained more cells overall compared with other genotypes, and a greater percentage of these cells were CD45⁺ leukocytes. Mirroring ISG expression, full deletion of Stat1 in KS^−/− mice reduced leukocyte infiltration to the level of S^+/− and S^−/− mice (Fig. 2D,E). Histological changes in the fat pad also tracked ISG expression, as KS^+/− fat at day 14 had an atrophied appearance with small adipocytes and abundant stromal cells and leukocytes, but KS^−/− was indistinguishable from S^+/− or S^−/− fat (Fig. 2F). Taken together, these data demonstrate that STAT1 is required for PDGFRβ-driven autoinflammation characterized by ISG up-regulation and leukocyte infiltration into peripheral tissues.

STAT1 suppresses PDGFRβ expression and signaling in DFs

The rescue of lethal wasting by Stat1 knockout prompted us to investigate PDGFRβ expression and downstream signaling in cells from the affected mice. PDGFRβ mRNA and protein levels were significantly reduced in KS^+/− DFs compared with S^+/− and S^−/− DFs (Fig. 3A). This is consistent with previous microarray data showing a 50% reduction of Pdgfrb expression in Pdgfrb^+/D849V pericytes compared with wild type (GSE29284) (Olson and Soriano 2011). However, complete Stat1 deletion fully restored mRNA levels and partially restored protein levels in KS^−/− DFs (Fig. 3A,B). PDGFRβ is constitutively autophosphorylated in PDGFRβ^D849V-expressing embryonic fibroblasts and adult VSMCs (Olson and Soriano 2011; He et al. 2015). To investigate whether Stat1 deletion has an effect on PDGFRβ autophosphorylation, we serum-starved DFs and examined phosphorylation under basal conditions or after PDGF treatment. As expected, PDGFRβ phosphorylation was low in S^+/− and S^−/− DFs under serum-starved conditions. Basal phosphorylation was increased in KS^+/− and KS^−/− DFs compared with S^+/− and S^−/− DFs but was significantly stronger in KS^−/− compared with KS^+/− (P < 0.01) (Fig. 3C). PDGF induced strong PDGFRβ phosphorylation in all genotypes, with a trend toward the highest response in KS^−/− cells. These data indicate that STAT1 negatively regulates Pdgfrb gene expression and signaling.

Open in a separate window

Figure 3.

STAT1 suppresses PDGFRβ expression and signaling in DFs. (A) Fold change in Pdgfrb mRNA in DFs by qRT–PCR with normalization to Gapdh. S^+/− expression levels were set as onefold. (*) P < 0.05 S^−/− or KS^−/− versus S^+/−; (**) P < 0.01 KS^+/− versus S^+/−. n = 3 biological replicates per genotype. (B) Fold change in PDGFRβ protein in DFs normalized to bActin. S^+/− levels were set as onefold. (*) P < 0.05 KS^−/− versus S^−/− and KS^+/− versus KS^−/−; (**) P < 0.01 KS^+/− versus S^+/−. n = 8 blots; a representative blot is shown. (C–E) DFs were starved in DMEM + 0.1% serum for 24 h and then stimulated with PDGF-BB for 10 min. Unstimulated S^+/− DFs were set as onefold. n = 3 blots; a representative blot is shown. (C) Fold change in p-PDGFRβ phosphorylation normalized to total PDGFRβ. (**) P < 0.01 KS^+/− basal versus S^+/−; (***) P < 0.001 KS^−/− basal versus S^−/− basal basal and KS^+/− basal versus KS^−/− basal. (D) Fold change in PLCγ phosphorylation normalized to total PLCγ. (**) P < 0.01 KS^−/− basal versus KS^+/− basal; (***) P < 0.001 KS^+/− basal versus S^+/− basal and KS^+/− stimulated versus S^+/− stimulated. (E) Fold change in STAT3 phosphorylation normalized to total STAT3. (**) P < 0.01 KS^+/− basal versus S^+/− basal and KS^−/− basal versus S^−/− basal. (F) Fold change in STAT5 phosphorylation normalized to total STAT5. (*) P < 0.05 KS^−/− basal versus S^−/− basal and KS^+/− basal versus S^+/− basal. Statistical tests were by paired (B) or unpaired (A,C–F) Student's t-test. Data represent mean ± SEM.

We showed previously that PLCγ was constitutively phosphorylated in KS^+/+ mouse embryonic fibroblasts, while two other PDGF pathway effectors, Akt and Erk1/2, were phosphorylated only in response to PDGF (Olson and Soriano 2011). We asked whether Stat1 deletion alters these events. Once again, the phosphorylation of Akt and Erk1/2 was PDGF-dependent in all DF genotypes (Supplemental Fig. S2). Also, as shown previously, PLCγ was constitutively phosphorylated in KS^+/− DFs compared with S^+/− DFs (Fig. 3D, lane 5 vs. lane 1), but PDGF-induced phosphorylation was blunted in KS^+/− compared with other genotypes (Fig. 3D, lane 6 vs. lane 2). Furthermore, Stat1 deletion resulted in stronger basal PLCγ phosphorylation in KS^−/− DFs compared with KS^+/− DFs (Fig. 3D, lane 7 vs. lane 5). We also examined STAT3 and STAT5, two other PDGF-regulated STATs, and found them constitutively phosphorylated in serum-starved KS^+/− and KS^−/− DFs compared with S^+/− and S^−/− (Fig. 3E,F). Therefore, PLCγ phosphorylation by PDGFRβ^D849V is enhanced by Stat1 deletion, but there is no effect on constitutive STAT3/5 phosphorylation.

Taken together, these results demonstrate that STAT1 suppresses the gain-of-function signaling exerted by PDGFRβ^D849V. The mechanism of suppression might involve STAT1-dependent ISGs that exert negative feedback on upstream signaling. For example, the suppressor of cytokine signaling (SOCS) family of proteins consists of negative regulators of cytokine and some growth factor signaling cascades (Linossi et al. 2013; Kazi et al. 2014). We analyzed the expression of eight SOCS members (SOCS1–7 and CIS) in cultured DFs. Socs4–7 expression did not differ greatly between genotypes. However, Socs1 and Socs3 exhibited a trend of up-regulation only in KS^+/− DFs, suggestive of regulation like other ISGs (Supplemental Fig. S3). Therefore, if any SOCS proteins mediate negative feedback on PDGFRβ, SOCS1 and SOCS3 would be the best candidates. In contrast, Socs2 and Cis were up-regulated in KS^+/− DFs (approximately threefold) and further overexpressed in KS^−/− DFs (~40-fold and ~10-fold, respectively) (Supplemental Fig. S3). This suggests changes in STAT-regulated transcription, where STAT1 opposes the function of other pathways/effectors that stimulate Socs2 and Cis expression (potentially STAT3/5). The SOCS proteins exert negative feedback on cytokine receptors by inhibiting kinase activity or mediating receptor degradation. However, further studies are needed to determine whether and under what conditions PDGFRβ may be a target of SOCS negative feedback.

STAT1 controls skin and adipose tissue phenotypes linked to PDGFRβ gain of function

Increased PDGFRβ signaling in KS^−/− DFs suggested that KS^−/− mice might develop novel phenotypes related to increased signaling in the absence of STAT1. For instance, PDGF signaling stimulates fibroblast proliferation and extracellular matrix secretion, which can lead to organ fibrosis (Bonner 2004; Andrae et al. 2008). To identify new phenotypes in KS^+/− or KS^−/− mice, we first examined the skin, focusing on the collagen-rich upper dermis and the underlying dermal WAT (dWAT) (Driskell et al. 2014). The dorsal skins of S^+/− and S^−/− littermates were indistinguishable. At 3 wk of age, KS^+/− skin was very thin without obvious fat, but KS^−/− skin exhibited an opposite phenotype with a thickened upper dermis (Fig. 4A, double arrow). Furthermore, the KS^−/− dWAT exhibited areas of interstitial collagen accumulation and expansion of stromal cells between adipocytes (Fig. 4A, circle). To quantify fibrosis, we measured the upper dermal thickness from the epidermal–dermal junction to the dermal–fat junction. By this measure, the KS^+/− dermis was significantly thinner than that of S^+/−, and there was significant dermal thickening in 3-wk-old KS^−/− mice (Fig. 4B). Moreover, by 8 wk of age, the KS^−/− upper dermis had thickened so much that it replaced dWAT, demonstrating the progressive nature of the fibrosis (Supplemental Fig. S4).

Open in a separate window

Figure 4.

PDGFRβ signaling causes dermal atrophy or dermal fibrosis. (A) Trichrome-stained dorsal skin from 3-wk-old mice. Arrows indicate dermal thickness (distance from the epidermal–dermal junction to the dermal–fat junction). (B) Quantification of dermal thickness in tissue sections. Each data point represents the average of four measurements for one mouse. (***) P < 0.001 KS^+/− versus S^+/−; (****) P < 0.0001 KS^−/− versus S^+/− by Student's t-test. Data represent mean ± SD. (C–E) Immunofluorescence analysis of dWAT by staining for extracellular matrix (collagen 3), adipocyte cytoplasm (perilipin 1), or PDGFRβ with DAPI counterstain. Circled areas highlight clusters of interstitial cells. Bars: A, 100 µm; C–E, 20 µm.

We stained 3-wk-old skin for collagen 3 (Col3) to mark areas of fibrosis or for perilipin (Plin1) to mark adipocytes. Consistent with histological analysis, abundant Col3 protein was detected in the KS^−/− dermal fat, particularly in the expanded interstitial space between individual adipocytes (Fig. 4C, circle). Col3 staining in the S^+/− and S^−/− fat was markedly lower than KS^−/−. There was high Col3 staining throughout the KS^+/− dermis, probably due in part to compression of the atrophic tissue. Plin1 stains the cytoplasm around the central lipid droplet of each adipocyte, which highlights a ring-like morphology for each adipocyte in S^+/−, S^−/−, and KS^−/− skin. The atrophied KS^+/− skin contained very small Plin1⁺ cells with a fibroblast-like morphology and no obvious lipid droplet (Fig. 4D). Interestingly, highly fibrotic skin of 8-wk-old KS^−/− mice also contained very small Plin1⁺ cells still embedded in the lower dermis (Supplemental Fig. S5A). Therefore, in both atrophied KS^+/− skin and fibrotic KS^−/− skin, Plin1⁺ adipocytes persist without obvious lipid droplets. Since STAT1 suppressed PDGFRβ expression and signaling (Fig. 3), we examined PDGFRβ expression in skin. In all genotypes, PDGFRβ was highly expressed in the hair follicle dermal sheath and in mural cells surrounding blood vessels. However, exclusively in KS^−/− skin, there was strong PDGFRβ expression in the fibroblast-like interstitial cells located in dermal fat (Fig. 4E, circle).

Next, we examined other fat depots, which generally exist in the form of white fat specialized for lipid storage or brown adipose tissue (BAT) specialized for thermogenesis. Like the skin, all tissues from S^+/− and S^−/− mice had similar morphology. The remnant of KS^+/− ingWAT from 3-wk-old mice contained only stromal vascular cells without obvious adipocytes (Fig. 5A). The KS^−/− ingWAT at 3 wk appeared grossly normal, but some perivascular fibrosis was apparent by trichrome stain (Fig. 5A arrows). By 8 wk, KS^−/− ingWAT was highly fibrotic (Fig. 5B). Like the skin, Plin1⁺ fibroblast-like cells were still detectable in the atrophied remnant of KS^+/− ingWAT at 3 wk (Supplemental Fig. S5B) and also in the fibrotic remnant of KS^−/− ingWAT at 8 wk (Supplemental Fig. S5C). Perigonadal WAT (pWAT) displayed a similar trend of rapid atrophy in KS^+/− mice and slower fibrotic remodeling in KS^−/− mice (Fig. 5C,D). In contrast, BAT was resistant to atrophy in KS^+/− mice, and minimal perivascular fibrosis in KS^−/− BAT was not accompanied by overt tissue remodeling even by 8 wk (Fig. 5E,F). Therefore, WAT is highly sensitive to elevated PDGFRβ signaling, while BAT is resistant. Interestingly, WAT-selective remodeling was also reported in mice with elevated PDGFRα signaling (Sun et al. 2017). In summary, PDGFRβ–STAT1 signaling causes atrophy of the dermis and WAT, but, in the absence of STAT1, PDGFRβ^D849V causes fibroproliferative replacement of WAT with collagen-rich matrix (Fig. 5G).

Open in a separate window

Figure 5.

Atrophy or fibrotic remodeling of WAT but not BAT. (A–F) Trichrome-stained ingWAT, pWAT, and iBAT at 3 or 8 wk of age. Arrows in A indicate perivascular fibrosis. Bar, 200 µm. (G) Graphical summary of the skin/WAT phenotype, where KS^+/− WAT undergoes STAT1-dependent atrophy by 3 wk of age, and KS^−/− WAT undergoes STAT1-independent fibroproliferative remodeling by 8 wk of age.

VSMC hyperplasia in Pdgfrb^+/D849V aortas is STAT1-independent

PDGFRβ^D849V causes VSMC hyperplasia that leads to arterial stiffening and gradual outward remodeling of the aorta (Olson and Soriano 2011). To determine whether STAT1 is required for PDGFRβ^D849V–induced VSMC hyperplasia, we generated a cohort of KS mutants with VSMC-selective Sm22α-Cre. This Cre driver is advantageous because mice survive up to 1 yr when PDGFRβ^D849V expression is restricted to VSMCs. At 6 wk of age, the KS^+/− and KS^−/− aortas were similarly dilated and dense with hyperplastic VSMCs compared with S^+/− and S^−/− aortas (Supplemental Fig. S6A). KS^+/− aortas exhibited perivascular inflammation (Supplemental Fig. S6A, arrows), which was also shown previously in KS^+/+ aortas (He et al. 2015). As expected, inflammation was rescued from KS^−/− aortas but was replaced by fibrosis of the adventitia (Supplemental Fig. S6A, asterisks). We also stained for the contractile VSMC markers αSMA, Sm22α, and SMMHC. These markers were down-regulated in KS^+/− and KS^−/− aortas compared with S^+/− and S^−/− (Supplemental Fig. S6B), indicating a loss of contractile phenotype. These morphological and molecular changes show that PDGFRβ^D849V induces VSMC hyperplasia independently of STAT1.

STAT1 modulates PDGFRβ^D849V-driven skeletal phenotypes

Skeletal phenotypes have not been shown in Pdgfrb^+/D849V mice but occur in humans with PDGFRB gain-of-function mutations. To address this discrepancy, we first analyzed skeletons by whole-mount staining at birth. At this time, we did not find consistent defects in KS^+/− or KS^−/− skeletons compared with S^+/− and S^−/−, but KS^+/− ribcages were sometimes narrower (Supplemental Fig. S7). Instead, consistent differences were noted at 3 wk of age: By this time, S^+/− and S^−/− calvaria were mineralized and stained with Alizarin red, but KS^+/− calvaria were poorly mineralized and incompletely stained (Fig. 6A). Intriguingly, KS^−/− calvaria exhibited a granular staining pattern suggestive of internal changes (Fig. 6B). Upon sectioning the center of the calvaria, we found that the KS^−/− skulls had marrow-filled channels throughout the bone (Fig. 6C); similar channels were seen only near the sutures of 3-wk-old S^+/− and S^−/− calvaria (Supplemental Fig. S8). In marked contrast to KS^−/− calvaria, KS^+/− calvaria were very thin and lacked marrow-filled channels (Fig. 6C,D). S^−/− calvaria were slightly thicker than S^+/− calvaria, consistent with previous reports (Xiao et al. 2004). In all genotypes, PDGFRβ was expressed in the sagittal suture mesenchyme, where skeletal stem cells reside (Zhao et al. 2015), plus broad expression in the periosteum, dura, and endosteum, which are regions populated by more differentiated osteoblast lineage cells (Supplemental Fig. S8).

Open in a separate window

Figure 6.

STAT1 modulates PDGFRβ^D849V- driven skeletal defects. (A) Whole-mount analysis of 3-wk-old calvaria stained with Alizarin red/Alcian blue. (B) Boxed areas from A enlarged to show a granular staining pattern in KS^−/−. (C) Parietal bones at 3 wk showing a hematoxylin and eosin-stained cross-section near the center of the bone. Bar, 50 µm. (D) Parietal bone thickness. Data points represent individual animals. (*) P < 0.05 S^−/− versus S^+/−; (***) P < 0.001 KS^+/− versus S^+/−. (E) Whole-mount analysis of 3-wk-old fourth and fifth ribs at the costochondral junction stained with Alizarin red/Alcian blue. (c) Cartilage; (b) bone. Arrows indicate the ridge of nonmineralized tissue overlying the perichondrial groove in KS^−/− ribs. (F) Trichrome-stained cross-sections through the growth plate of the fourth or fifth rib. The perichondrial groove is boxed in the top panels and is shown at high magnification in the bottom panels. (c) Cartilage; (b) bone. Arrows indicate the axes measured in G and H. Bar, 50 µm. (G) Growth plate width measured at the level of the perichondrial groove. Data points represent the mean of two to four ribs per individual mouse (fourth and fifth ribs, right and left sides). (*) P < 0.05 S^−/− versus S^+/− and KS^+/− versus S^+/−; (***) P < 0.001 KS^−/− versus S^+/− by Student's t-test. (H) Growth plate height measured from the bone–cartilage border to the level of the groove. Data points represent the mean of two to four ribs per mouse (fourth and fifth ribs, right and left sides). (***) P < 0.001 KS^−/− versus S^+/−. All statistics by Student's t-test. Data represent mean ± SD.

The ribcage was also altered in 3-wk-old PDGFRβ^D849V-expressing mice. Each rib has a dorsal bony segment attached to the spine and a ventral cartilaginous segment attached to the sternum. A growth plate forms at the bone–cartilage interface in the middle of each rib. KS^+/− ribs were thin, and bone was poorly mineralized (Fig. 6E; Supplemental Fig. S9). However, in KS^−/− ribs, the cartilage and bone near the growth plate were slightly enlarged, forming a noticeable bulge (Fig. 6E; Supplemental Fig. S9). KS^−/− ribs also typically displayed ridges of nonmineralized tissue over the groove of Ranvier (Fig. 6E arrows). This groove in the perichondrium forms at the periphery of the growth plate of most long bones and runs circumferentially around the bone. It is believed to harbor progenitor cells that govern circumferential growth of cartilage and the formation of periosteum around the growth plate (Shapiro et al. 1977; Karlsson et al. 2009; Yang et al. 2013). Histological analysis of this region failed to show consistent differences between S^+/− and S^−/− bones, but the groove of KS^+/− ribs was typically shallow and hypocellular (Fig. 6F, boxed area enlarged below). In contrast, the groove of KS^−/− ribs was hyperplastic with an expansion of fibrous tissue superior to the groove (Fig. 6F), corresponding to the nonmineralized ridge of tissue noted by whole mount (Fig. 6E arrows). There was a slight but statistically significant increase in growth plate width in S^−/− ribs compared with S^+/−, without a difference in height (Fig. 6G,H). The KS^−/− growth plates were significantly enlarged. The KS^+/− growth plates appeared disorganized and lens shaped, with an average width intermediate between S^−/− and KS^−/−. Therefore, taking together the skull and rib defects, the pattern of bone phenotypes corresponds with the fat and skin phenotypes, where the KS^+/− tissues exhibit degenerative features, and the KS^−/− tissues exhibit overgrowth. Poor mineralization can result from inadequate levels of crucial metabolites, but we did not find significant differences in serum phosphate, calcium, or vitamin D among the four genotypes at 3 wk of age (data not shown).

In a cohort of S^+/−, S^−/−, and KS^−/− mice at 6–8 wk of age, ~75% of the KS^−/− mice were overgrown (defined as >25 g or at least two standard deviations above the average weight of controls) (Supplemental Fig. S10A,B). Since KS^−/− mice are not obese, most of the increase in weight should be due to overgrowth of the skin, connective tissue, and skeleton. KS^−/− mice also developed severe kyphosis and abnormally long ribs that caused the sternum to protrude (Supplemental Fig. S10A,C). Moreover, after removing the skin, the KS^−/− calvaria were red instead of the usual pale pink (Supplemental Fig. S11A). Sectioning revealed that the color change was due to a greatly expanded marrow area (Supplemental Fig. S11B), which extended the observations in three wk-old KS^−/− calvaria (Fig. 6C). Therefore, the subtle skull and rib defects in 3-wk-old KS^−/− mice presaged rapid skeletal overgrowth by 8 wk of age. In contrast, KS^+/− mice exhibited underdeveloped or atrophied skull and rib bones, which was already quite severe by 3 wk. It is clear from these data that PDGFRβ regulates skeleton growth, a process that is strongly modified by the STAT1 pathway.

PDGFRβ expression in NIH3T3 cells

Vectors were created to express wild-type mouse PDGFRβ or mutant isoforms with single amino acid change P583R, V664A, or D849V (corresponding to human P584R, V665A, or D850V). First, PCR-amplified wild-type or mutant cDNAs were inserted in pENTR (a gateway cloning-compatible construct; gift from Eric Campeau). Each construct was then mixed with pLenti CMV BLAST DEST (Eric Campeau) with Clonase II (Thermofisher) to yield the expression vectors here. A pLenti eGFP plasmid was used as a control. Two micrograms of plasmid was transfected into NIH3T3 cells with Lipofectamine 3000 (Thermofisher) in cells growing in DMEM with 0.1% FBS and no antibiotics. Cell lysates were collected 24 h later.

RNA isolation and quantitative RT–PCR (qRT–PCR)

Total RNA was isolated from mouse ingWAT tissues using RNeasy kit (Qiagen) or from cultured DFs using Trizol (ThermoFisher). cDNA reverse transcription was performed using random primers and SuperScript III RT (Invitrogen). qPCR was performed on a Bio-Rad iCycler with iQ SYBR Green master mix. Results were normalized to Gapdh. Primer sequences are listed in the Supplemental Material.

Flow cytometry

Fat pads (ingWAT) were minced and digested with 300 µg/mL collagenase type I (Life Technologies) in Hank's balanced salt solution (pH 7.4) for 60 min at 37 °C. After filtering (70-µm filter), cells were resuspended in PBS with 1 mM EDTA and 2% FBS and incubated with Fc-blocking antibody (eBioscience) for 15 min on ice before staining with FITC-CD45.2 antibody (1:100; Biolegend). Cells were also stained with propidium iodide to identify live/dead cells. After washing, immunofluorescence was detected by an LSR II (BD Biosciences). Data were analyzed by post-collection compensation using FlowJo software (Tree Star, Inc.).

Primary DFs and Western blotting

After removing hair with cream (Nair), mouse DFs were isolated from 2- to 3-wk-old mice by explanting 1-mm² pieces of skin, dermis side down, with DMEM plus 10% FBS, 25 U/mL penicillin/streptomycin, and 2 mM L-glutamine. After 2 wk, skin pieces were removed, and DFs were passaged. All assays were performed at passages 2–5. For Western blotting, cells were starved in medium containing 0.1% FBS for 24 h and then treated with 10 ng/mL PDGF-BB (R&D Systems) for 10 min or according to a time course. Cells were lysed in RIPA buffer (50 mM Tris at pH 7.4, 1% NP-40, 0.25% sodium deoxycholate, 150 mM NaCl, 0.1% sodium dodecyl sulfate) with 1 mM NaF, Na₃VO₄, PMSF, and 1× protease inhibitor cocktail (Complete, Roche), and protein concentration was determined by Pierce BCA assay (ThermoFisher). Five micrograms or 10 µg of protein was then separated by SDS-PAGE and transferred to nitrocellulose membranes, blocked with 5% BSA, and then subjected to Western blotting with primary antibodies listed in the Supplemental Material. Membranes were then probed with horseradish peroxidase-conjugated secondary antibodies at 1:5000 (Jackson ImmunoResearch) in 5% milk. Blots were developed with Pierce ECL Western blotting substrate (ThermoFisher) and autoradiography film (Santa Cruz Biotechnology).

Tissue and histology

Tissue was immersion-fixed overnight in 4% paraformaldehyde (PFA; Electron Microscopy Sciences) or Bouin's fixative (Sigma). Hair was removed with cream (Nair). Bones were decalcified in 10% EDTA solution for 1 wk. For histochemical staining, Bouin's-fixed tissues were embedded in paraffin and sectioned at 5 µm before staining with Masson's Trichrome (Electron Microscopy Sciences) or hematoxylin and eosin (Vector Laboratories). For immunohistochemistry, PFA-fixed samples were cryosectioned at 8 µm and stained with primary antibodies listed in the Supplemental Material. This was followed by the appropriate donkey secondary antibodies (Jackson ImmunoResearch) and DAPI (Sigma) as a nuclear stain. Immunofluorescence microscopy was performed with a Nikon Eclipse 80i microscope connected to a digital camera.

Whole-mount analysis of bones

Skin was removed, and pups were fixed in 100% EtOH for 3 d followed by Alcian blue stain (15 mg of Alcian blue, 80 mL of 95% EtOH, 20 mL of glacial acetic acid) for 12 h. Next, the skeletons were rinsed in 100% EtOH overnight, cleared in 1% KOH for 6 h, and then counterstained in 5 mg of Alizarin red in 100 mL of 1% KOH overnight. Skeletons were then cleared in 1% KOH, washed in a graded series of 1% KOH and glycerol, and finally stored in 100% glycerol. Adult skeletons were treated similarly except for overnight trypsin treatment (1% trypsin in 7:3 distilled H₂O:saturated sodium borate solution) before preclearing in 2% KOH, and all subsequent steps were performed in 2% KOH.

We thank Cameron Steele, Tony Tang, Linda Thompson, James Tomasek, Mary Beth Humphrey, and Yanick Crow for their assistance and helpful discussion. We also thank the Flow Cytometry Core Facility and Imaging Core Facility (associated with P20GM103636) and the Mouse Phenotyping Core Facility (associated with P30GM114731) of the Oklahoma Medical Research Foundation Centers of Biomedical Research Excellence. These studies were funded by National Institutes of Health grants (P20GM103636 and R01AR070235), a grant from the Oklahoma Center for Adult Stem Cell Research, and the Pew Charitable Trusts. C.S. and S.C.M. were supported by predoctoral fellowships from the American Heart Association.