Bmp2 ^lacZ mice were generated in the Rosen Lab and model characteristics and genotyping are previously described (Salazar et al., 2019). Experiments were performed in compliance with the Guide for the Care and Use of Laboratory Animals and were approved by the Harvard Medical Area Institutional Animal Care and Use Committee (IACUC). Mice were housed and cared for in accordance with IACUC standards in an AAALAC-accredited facility. Mice were euthanized via CO₂ inhalation and cervical dislocation as secondary confirmation in accordance with IACUC standards.

Femurs were dissected from two male and two female postnatal day 14 Bmp2 ^lacZ mice (see Figures 1A–D). Most of the muscle and connective tissue were removed before placing femurs in PBS on ice. Under a dissecting microscope, a scalpel was used to bisect the growth plates (perpendicular to the longitudinal axis of the femur) to remove most of the epiphyseal ends and associated connective tissue without exposing the bone marrow. Fine-tipped Dumont forceps (Fine Science Tools, 11203-23) were used to separate as much muscle and connective tissue from the femur as possible without disrupting the periosteum. These tissues were gently removed using a scalpel with a #10 blade in a cutting motion parallel to the periosteal surface to avoid pulling off the periosteum. At each epiphyseal end, the scalpel was used to make a single cut around the circumference of the femur 1 mm beneath the growth plate to avoid capturing perichondrium. The scalpel was then used to make a single cut in the periosteum from one epiphyseal end to the other and the periosteum was peeled from the femur using Dumont forceps. The resulting periosteal tissue was incubated in 1 mg/mL Collagenase type I (Millipore, scr103) in sterile PBS in a cell culture incubator (37°C, 5% CO₂) for 1 h. The digestion solution was then filtered through a 70 μm filter (Corning, 431751) and resulting cells were cultured in MEMα (Gibco, A10490-01) supplemented with 10% fetal bovine serum (FBS) and 1% PenStrep (Gibco, 15140122). Primary cells did not survive if cloned immediately after isolation, so extensive passaging was performed first to eliminate populations incapable of being immortalized. The cells isolated from all four mice were pooled together and passaged twenty times. Cell density never exceeded 80% confluence to avoid potential osteogenic differentiation. Passage 21 (P-21) cells were diluted and seeded at a density of one cell per well in one 96 well plate to establish individual clones. The plate was examined under a microscope to confirm no more than one cell was present in each well. After 2–3 weeks, twelve clones that were viable and proliferated to achieve a confluence greater than 50% were further cultured for two additional passages (P-22, P-23) to observe expansion and viability on a larger scale. Nearly a dozen other wells contained cells, but these populations were excluded based on evidence of cell death and a lack of proliferation. Four of the twelve selected clones proliferated at a comparatively staggering rate and exhibited extensive cell death, so these clones were excluded. The eight remaining candidate clones were reserved in stock vials and examined for mRNA markers of muscle, tendon, and periosteum (Supplementary Figure S1). Stock vials consist of approximately 1 million cells in culture media supplemented with 10% DMSO (Sigma-Aldrich, D2650) and are stored in liquid nitrogen.

Experiments were conducted using both early (P-24–P-26) and later passages (P-32–P-34) to examine consistency. Osteogenic differentiation: media consisted of MEMα supplemented with 20% FBS, 1% PenStrep, 10 mM β-glycerophosphate (Sigma-Aldrich, G9422), and 100 μM Ascorbic acid (Sigma-Aldrich, A4544). Media was changed every 2–3 days and mineral deposition was detected after 3–4 weeks of treatment. Cells were fixed in 10% neutral-buffered formalin (Millipore, HT501128) for 20 min at ambient temperature and incubated in 1.5% Alizarin Red S (Sigma-Aldrich, A5533) solution to detect matrix deposition. Chondrogenic differentiation: periosteal cells were seeded at a density of 100,000 cells per 15 μL to create micromasses. Micromasses were differentiated using the Gibco StemPro Chondrogenesis Differentiation Kit (Gibco, A1007101). Media was changed every 2–3 days and matrix deposition was detected after 1–2 weeks of treatment. Cells were fixed in 4% paraformaldehyde (Electron Microscopy Sciences, 15710) for 15 min at ambient temperature and incubated in 1% Alcian Blue (Electron Microscopy Sciences, 10350) in 0.1 N HCl to detect cartilaginous matrix deposition. Adipogenic differentiation: periosteal cells were treated with DMEM High Glucose (Sigma-Aldrich, D5796) supplemented with 10% FBS, 1% PenStrep, 62.5 mM IBMX (Sigma-Aldrich, I5879), 1 mM Dexamethasone (Sigma-Aldrich, D4902), 20 mM Rosiglitazone (Cayman Chemical, 71742-10), and 2 mM Insulin (Sigma-Aldrich, I6634) for 4 days, followed by treatment with DMEM High Glucose supplemented with 10% FBS, 1% PenStrep, 20 mM Rosiglitazone, and 2 mM Insulin for 3 days. Media was changed every 2 days and lipid formation was detected after 1 week of treatment. Cells were fixed in 10% neutral-buffered formalin for 45 min at ambient temperature and incubated in 0.35% Oil O Red (Sigma-Aldrich, O0625) in isopropanol solution to detect lipid accumulation. Cells were counterstained with 0.1% Crystal Violet (Sigma-Aldrich, C6158). Images were collected at 4X or ×10 magnification using a Keyence BZ-X710 microscope and stitched together using its associated software.

Experiments were conducted using both early (P-24–P-26) and later passages (P-32–P-34) to examine consistency. To detect changes in mRNA transcription: periosteal cells were incubated in MEMα alone for 2 h to synchronize cell activity. Cells were then incubated in MEMα containing 100 ng/mL BMP2 (Genetics Institute) or 1 ng/mL TGFβ-1 (Peprotech, 100-21C) and lysed after 4 h. For Western blotting: periosteal cells were incubated in MEMα alone for 6 h to synchronize cell signaling. Cells were then incubated in MEMα containing 100 ng/mL BMP2 or 1 ng/mL TGFβ-1 and lysed after 45 min of treatment.

Experiments were conducted using both early (P-24–P-26) and later passages (P-32–P-34) to examine consistency. Periosteal cells were seeded on 75 × 38 × 1 mm glass slides (Corning, CLS294775X38) and exposed to fluid flow upon exceeding 80% confluence. Fluid shear was applied using a previously described parallel-plate oscillatory fluid flow apparatus (Moore et al., 2018b). Briefly, slides were inserted into chambers containing regular culture media, placed in a cell culture incubator for 30 min to acclimate, and exposed to 1 h of oscillatory fluid flow at 1 Hz with a peak shear stress of 10 dyn/cm². Static controls were similarly placed into chambers and incubated alongside fluid shear samples. Slides were removed from the chambers, rinsed with PBS, and lysed for RTqPCR or Western blotting at the conclusion of flow.

RNA was isolated using TRIzol reagent (ThermoFisher, 15596018) and a RNeasy Kit (Qiagen, 74106). RNA was converted to cDNA using the High-Capacity cDNA Reverse Transcription Kit (ThermoFisher, 4368814). qPCR was performed using Faststart universal SYBR Green (Sigma-Aldrich, 4913850001) and a StepOnePlus Real-Time PCR System (Applied Biosciences). mRNA values were normalized to GAPDH or β-actin–housekeeping genes constitutively expressed at high levels–to account for general variability in mRNA expression between samples. Genes that were within 12 cycles of the cycle at which GAPDH reached the threshold for expression were considered expressed in the PDC line. Experimental groups are expressed as a fold change in relation to controls normalized to a value of “1”. Primer sequences are available upon request.

Protein was isolated using RIPA Buffer (Cell Signaling Technology, 9806S) supplemented with protease and phosphatase inhibitors (ThermoFisher, 78440). Samples were examined by immunoblotting after SDS-PAGE using a 10% Bis-Tris gel (Invitrogen, WG1202BX10). Following transfer, membranes were blocked in TBST containing 5% non-fat dry milk and 5% bovine serum albumin and incubated overnight at 4°C in the following primary antibodies diluted in blocking buffer (1:1000): pSmad1 (Cell Signaling Technology, 13820S), Smad1 (Cell Signaling Technology, 9743S), Smad2/3 (Cell Signaling Technology, 8685S), pSmad2 (Cell Signaling Technology, 3108L), and β-actin (Cell Signaling Technology, 4967S). Membranes were then incubated for 1 h at ambient temperature in an anti-Rabbit IgG HRP-linked secondary antibody (Cell Signaling Technology, 7074S) diluted in blocking buffer (1:2000). Blots were developed using SuperSignal West Femto Maximum Sensitivity Substrate (ThermoFisher, 34095). Images were acquired with a PXi4 Chemiluminescent and Fluorescent Imaging System (Syngene, Bangalore, India) and quantification was performed using ImageJ software (National Institutes of Health).

Dissected femurs were fixed in 10% formalin overnight, decalcified in 0.5 M EDTA (VWR, 75800-470) for 1 week, dehydrated, embedded in paraffin, and sectioned in 5–10 μm slices. Slides were rehydrated and stained using Weigert’s Iron Hematoxylin A (VWR, 26044-05), Weigert’s Iron Hematoxylin B (VWR, 26044-15) and Van Gieson’s Solution (VWR, cat #26046-05). Images were collected at ×20 magnification using a Keyence BZ-X710 microscope and its associated software.

The clonal cell lines used for this experiment were determined to be negative for mycoplasma using the MycoAlert™ Mycoplasma Detection Kit (Lonza, LT07-118). Clones were tested the first passage after thawing a stock vial (passage 24, P-24), at P-28 or P-29 during experimentation, and at the conclusion of experiments at P-34.

PDCs at passage 28 (P-28) and P-33 were seeded at two different densities on 25 cm² flasks and shipped at ambient temperature overnight to KaryoLogic, Inc. in Durham, NC for karyotyping. Cytogenetic analysis was performed on twenty G-banded metaphase spreads and we consulted with a senior analyst to interpret the findings.

Researchers were blinded to all data analysis. Differences between control and experimental groups were determined using a two-tailed Student’s t-test. Values are reported as mean ± SEM, with p < 0.05 considered statistically significant. The sample size was selected to achieve a power of at least 80%. Statistical analysis was conducted using GraphPad Prism (San Diego, CA).

Primary periosteal cells were isolated from postnatal day 14 (P14) Bmp2 ^LacZ mice to maximize cell yield and to utilize X-gal staining to detect BMP2. Briefly, periosteal tissue from femurs was pooled and digested to extract primary cells (Figure 1). These cells were passaged 20 times to select for cells that exhibited high-passage potential and then seeded at a density of 1 cell per well to generate clones. Clones that survived and proliferated were examined for mRNA expression of published markers for periosteal (αSMA, Gli1, Pdgfrα, Ctsk, Prx1, Pstn, and Sca-1), skeletal muscle (Mef2c, Myh2, Myod1, and Myog), and connective tissue (Scx, Tnmd) cells (Colnot et al., 2012; Grcevic et al., 2012; Debnath et al., 2018; Duchamp De Lageneste et al., 2018; Jo et al., 2019; Matthews et al., 2021; Piasecka et al., 2021). We first examined these markers in freshly isolated periosteum from P14 Bmp2 ^LacZ mice. All periosteal markers except Gli1 were detected and the muscle and connective tissue markers were not detected. Two candidate clones expressed all periosteal markers, and muscle and connective tissue markers were not detected (Supplementary Figure S1). In fact, none of the candidate clones examined expressed Scx or Tnmd. This mRNA expression profile remained consistent for 10 passages, suggesting the clones are stable with further passaging (Figure 1E). The following experiments were conducted using cells within the 10 passages examined for mRNA expression and findings at different passages were consistent. For simplicity, we highlight data for one clone to describe a periosteum-derived cell (PDC) line in this manuscript, but results were consistent for both clones (Supplementary Figure S2; Figure 1F).

Periosteal stem and progenitor cell populations are known to differentiate into osteoblasts, chondrocytes, and adipocytes in vitro, so we examined the lineage potential of our line (De Bari et al., 2006; Arnsdorf et al., 2009a; Debnath et al., 2018; Perrin et al., 2021). PDCs incubated in osteogenic differentiation media deposited mineral in 3–4 weeks (Figure 2A). We detected cartilage matrix deposition in PDC micromasses incubated in chondrogenic differentiation media for 1–2 weeks (Figure 2B). Lipid accumulation was observed in PDCs treated with adipogenic differentiation media after 1 week (Figure 2C). mRNA markers for osteogenesis, chondrogenesis, and adipogenesis were significantly upregulated in differentiated PDCs (Figure 2D). These studies indicate the PDC line is multipotent and differentiates into the lineages expected for a periosteal stem cell in vitro.

Recent work indicates the importance of BMP and/or TGFβ signaling for normal periosteal activity during skeletal development, fracture repair, and load-induced bone formation (Tsuji et al., 2006; Chen et al., 2012; Salazar et al., 2016; Wu et al., 2016; Salazar et al., 2019; Xia et al., 2020). We therefore determined whether our PDC line could be utilized to study these signaling pathways in vitro. First, we examined mRNA expression of components in the TGFβ/BMP superfamily in freshly isolated periosteum and early and later passages of the PDC line (Figure 3A). Nearly all the associated components were expressed in the PDC line except for the ligands Bmp2, Bmp3, Gdf8, and Nog, and Type I receptors Alk1 and Alk7. This expression profile was consistent between the early and later PDC passages, but we noted some inconsistencies with whole periosteum. Specifically, we detected Bmp2, Bmp3, and Gdf8 expression in whole periosteum, which was absent in PDCs. We then examined BMP and TGFβ signaling in the PDC line by treating with ligands known to activate these pathways in periosteal cells. PDCs incubated in media containing recombinant BMP2 exhibited an increased ratio of phosphorylated Smad1 (pSmad1) to total Smad1 compared to vehicle controls, which is indicative of activated canonical BMP signaling (Figure 3B). The ratio of pSmad2/Smad2/3 was unchanged with BMP2 treatment compared to vehicle controls. PDCs incubated in media containing recombinant TGFβ-1 exhibited increased levels of pSmad2/Smad2/3, which is typically associated with activated canonical TGFβ signaling. Interestingly, pSmad1/Smad1 levels were also slightly elevated with TGFβ-1 treatment. When signaling is activated, pSmads bind Smad4 and translocate to the nucleus to trigger transcription of genes associated with canonical BMP (Id1, Id3) and TGFβ (Serpine1) signaling. We therefore examined changes in mRNA expression of these target genes with treatment (Figure 3C). Indeed, Id1 and Id3 were uniquely upregulated with BMP2 treatment and Serpine1 was upregulated only by TGFβ-1 treatment.

Periosteal cells are known to be mechanosensitive and are important for load-induced bone formation (Arnsdorf et al., 2009a; Moore et al., 2018b; Moore et al., 2019; Xiao et al., 2020). We examined whether our PDC line was mechanosensitive using a custom fluid shear chamber device previously used to study primary periosteal cell mechanotransduction in vitro (Moore et al., 2018b). Indeed, PDCs exposed to fluid flow exhibited increases in Cox2 and Opn, genes that are upregulated in bone cells in response to physical stimulation (Figure 4A) (Wadhwa et al., 2002; Ponik and Pavalko, 2004; Arnsdorf et al., 2009b; Hoey et al., 2012; Moore et al., 2018b). TGFβ signaling is believed to be important in periosteal cell mechanotransduction so we examined whether this pathway was upregulated in response to fluid flow (Figures 4B, C) (Raab-Cullen et al., 1994; Klein-Nulend et al., 1995; Nguyen et al., 2020). As expected, pSmad2/Smad2/3 levels increased with fluid flow and mRNA expression of the target genes Serpine1 and Tgfβ1 also increased. The role of BMP signaling in load-induced bone formation and periosteal cell mechanotransduction is less clear (Kopf et al., 2012; McBride-Gagyi et al., 2015). In our system, we found that pSmad1/Smad1 levels decreased in PDCs exposed to fluid flow and Id1 and Id3 expression was unchanged compared to static controls. However, Bmp2 expression was significantly elevated with application of. The decreased pSmad1/Smad1 levels seen in PDCs exposed to fluid flow are in part due to a significant increase in total Smad1. Smad2/3 levels were comparable between static and fluid flow groups.

Standard G-banded karyotyping was performed on PDCs at P-28 and P-33 to determine what chromosomal abnormalities, which are expected for immortalized cell lines, were present (Stepanenko and Dmitrenko, 2015). Abnormal karyotypes were present in both passages, with several consistent observations between the varying spreads. At both passages multiple polysomies, or additional copies of chromosomes, were observed and only one copy of Chromosome 14 was present. P-28 cells exhibited a modal chromosome number of 72, ranging from 63 to 75 across the spreads. P-33 cells had a modal chromosome number of 72 ranging from 68 to 74 across spreads. Markers, or structurally abnormal chromosomes that cannot be unambiguously identified by conventional banding cytogenetics, were also detected. P-28 cells exhibited 2-6 markers among spreads and this range increased to 4-10 in P-33 cells. A single dicentric chromosome was observed in two spreads of P-33 cells. As dicentrics can have difficulty passing through mitosis, it is recommended that the PDC line be used up to P-34, the highest passage we validated. Further validation is encouraged when using cells beyond P-34. X and Y sex chromosomes were present, indicating the PDC line was derived from a male mouse.