Inhibition of hypoxia‐induced Mucin 1 alters the proteomic composition of human osteoblast‐produced extracellular matrix, leading to reduced osteogenic and angiogenic potential

Abstract The bone microenvironment is one of the most hypoxic regions of the human body and in experimental models; hypoxia inhibits osteogenic differentiation of mesenchymal stromal cells (MSCs). Our previous work revealed that Mucin 1 (MUC1) was dynamically expressed during osteogenic differentiation of human MSCs and upregulated by hypoxia. Upon stimulation, its C‐terminus (MUC1‐CT) is proteolytically cleaved, translocases to the nucleus, and binds to promoters of target genes. Therefore, we assessed the MUC1‐mediated effect of hypoxia on the proteomic composition of human osteoblast‐derived extracellular matrices (ECMs) and characterized their osteogenic and angiogenic potentials in the produced ECMs. We generated ECMs from osteogenically differentiated human MSC cultured in vitro under 20% or 2% oxygen with or without GO‐201, a MUC1‐CT inhibitor. Hypoxia upregulated MUC1, vascular endothelial growth factor, and connective tissue growth factor independent of MUC1 inhibition, whereas GO‐201 stabilized hypoxia‐inducible factor 1‐alpha. Hypoxia and/or MUC1‐CT inhibition reduced osteogenic differentiation of human MSC by AMP‐activated protein kinase/mTORC1/S6K pathway and dampened their matrix mineralization. Hypoxia modulated ECMs by transforming growth factor‐beta/Smad and phosphorylation of NFκB and upregulated COL1A1, COL5A1, and COL5A3. The ECMs of hypoxic osteoblasts reduced MSC proliferation and accelerated their osteogenic differentiation, whereas MUC1‐CT‐inhibited ECMs counteracted these effects. In addition, ECMs generated under MUC1‐CT inhibition reduced the angiogenic potential independent of oxygen concentration. We claim here that MUC1 is critical for hypoxia‐mediated changes during osteoblastogenesis, which not only alters the proteomic landscape of the ECM but thereby also modulates its osteogenic and angiogenic potentials.


| INTRODUCTION
The extracellular matrix (ECM), a complex of self-assembled macromolecules, is composed predominantly of collagen, glycoproteins, proteoglycans, and hyaluronan (Trapani et al., 2017). It not only serves as a 3D scaffold for the cells and facilitates their attachment, migration, proliferation, and differentiation but also acts as a reservoir for growth factors and cytokines. In bone, ECM comprises 40% organic (of which 90% constitutes collagen type I, and 10% represents noncollagenous proteins) and 60% inorganic compounds.
Moreover, its exact proteomic composition and mineralization vary based on bone type, age, sex, and health conditions (Lin et al., 2020).
Similarly, the proteomic composition and mineralization of osteoblast-secreted ECM can be modulated by altered culture conditions in vitro.
ECM formation with eventual mineralization serves as a hallmark of osteogenic differentiation of mesenchymal stromal cells (MSCs), following deposition and accumulation of proteins to form the ECM, mineralization proceeds, which marks the final phase of the osteoblast phenotypic development (Trapani et al., 2017). The bone ECM coordinates the interaction between osteogenic and angiogenic processes and drives bone growth and tissue restoration during bone repair. Angiogenesis is vital for tissue growth, maturation and repair by facilitating transport of cytokines and growth factors necessary for cell viability, proliferation, and interaction.
Hypoxia is a relevant situation for bone as the tissue is permanently hypoxic (Rankin et al., 2011) and as a consequence, the behavior of bone cells is subject to changes in ambient oxygen concentrations. MSCs are self-renewable and can differentiate into osteoblasts, fibroblasts, myocytes, and adipocytes (Ankrum et al., 2014). Previous studies including our own have shown that hypoxia inhibits osteogenic differentiation of MSCs and thus osteoblast-mediated ECM mineralization (Hsu et al., 2013;Nicolaije et al., 2013). Therefore, it is of primary interest to consider oxygen concentration as an important factor in tissue growth and repair for ECM mineralization and neovascularization. Hypoxia-inducible factors (HIFs) are the primary signaling molecules sensing oxygen concentration, which upon activation trigger several signaling cascades involving vascular endothelial growth factor (VEGF), transforming growth factor-beta (TGFβ), AMP-activated protein kinase (AMPK), and nuclear factor-κB (NF-κB) (Guo et al., 2016;Wang et al., 2007;D'Ignazio & Rocha, 2016).
Mucin 1 (MUC1) is a transmembrane heterodimeric glycoprotein expressed on the surface of secretory epithelial, mucosal, and hematopoietic cells (Nath & Mukherjee, 2014). We previously found that MUC1 is upregulated in differentiating osteoblasts derived from human MSC and that Muc1 plays an age-dependent role in murine bone development (Brum et al., 2018). The extracellular domain of MUC1 (N-terminal) is composed of a variable number of 20 amino acid tandem repeats with extensive O-glycans (Dhar & McAuley, 2019). The cytoplasmic tail of MUC1 (MUC1-CT) consists of a 58 amino acid extracellular domain, a 28 amino acid transmembrane domain, and a 72 amino acid cytoplasmic domain (Nath & Mukherjee, 2014). Upon stimulation, a part of the C-terminus is proteolytically cleaved, forms a homodimer, and translocases to the nucleus by a mechanism involving importin-β and Nup62 (Leng et al., 2007). MUC1 is overexpressed in a number of adenocarcinomas during malignant transformation and it is associated with a protective barrier function at the epithelial cell-cell surface (Kufe, 2009). Furthermore, it was suggested that MUC1 expression is directly correlated with hypoxia-induced tumor progression and metastatic urothelial carcinoma in bone (Kaira et al., 2012;Zanetti et al., 2011). Emerging evidence suggests that HIF-1α directly regulates MUC1 expression under hypoxia in lung and renal carcinoma (Aubert et al., 2009;Mikami et al., 2009;).
Previously, we discovered MUC1 among one of the top differentially modulated genes by hypoxia in a transcriptomic dataset of human osteoblasts (Nicolaije et al., 2013).
Despite the previously generated knowledge regarding the role of MUC1 in bone in vivo (Brum et al., 2018), there is a large knowledge gap as to how and what physiological changes in the ECM are modulated by MUC1 in human osteoblasts? Considering the upregulation of MUC1 under hypoxia and the oxygen concentration is a major modulating factor for ECM and its mineralization, we assessed the role of MUC1 and hypoxia on the proteomic composition of human osteoblast-produced ECM and its consequences for osteogenesis and angiogenesis.

| Cell culture
Human bone marrow-derived MSC were obtained from Lonza (PT-2501) from a healthy male donor. MSC were cultured at a density of 5000 cells/cm 2 in a growth medium formulated with Dulbecco's modified Eagle medium (DMEM) (GIBCO) supplemented with 10% heat-inactivated fetal bovine serum (FBS) (GIBCO), 100 U/ml penicillin and 100 μg/ml streptomycin (Sigma), pH 7.5, for 2 days at 5% carbon dioxide (CO 2 ) and 37°C. Subsequently, MSC were osteogenically differentiated by adding 10 nM dexamethasone and 10 mM β-glycerophosphate (Sigma) to the growth medium. Hypoxic conditions for MSC cultures were created using a multi-gas incubator (Bionex Solutions Inc.) containing a gas mixture composed of 93% nitrogen, 5% CO 2, and 2% oxygen. In some of the experiments, MSC were treated with 5 µM GO-201 (G7923; Sigma), a specific MUC1 inhibitor, during the course of osteoblast differentiation in both normoxic and hypoxic conditions. Cells were refreshed twice a week.  (Table S1). The Virapower system consists of three different plasmids encoding essential viral proteins to form a replication-deficient lentiviral particle.
Twenty-four hours later, the medium was refreshed with 8 ml 10% DMEM medium containing 20 mM HEPES, and the next day, the virus-containing medium was collected and filtered through a 0.45 µm filter and directly used for transduction. MSC cultured in 12-well plates were transduced with MUC1 shRNA-expressing lentivirus in a serum-free growth medium and incubated overnight. MSC cultured for 10 days posttransduction were harvested for messenger RNA (mRNA) isolation and MUC1 expression was assessed using RT-PCR. In another experiment, ECM mineralization of MSC cultured under 20% (normoxic) or 2% O 2 (hypoxic) for 17 days was assessed by measuring calcium content.

| Protein extraction and western blotting
Confluent monolayers of osteogenically differentiating MSC for 11 days were scraped in ice-cold phosphate buffered saline (PBS), pelleted, and lysed in RIPA buffer (R0278; Sigma) with proteases and phosphatase inhibitors. Proteins in the supernatant were quantified with the Pierce TM BCA assay (Thermo Fisher Scientific). For western blotting, equal amounts of proteins of each sample were boiled with 6X sample buffer (240 mM Tris pH 6.8, 40% glycerol, 8% SDS, 0.002% bromophenol blue, 0.002% β-mercaptoethanol) for 5 min.
The proteins were separated by 12% sodium dodecyl sulfatepolyacrylamide gel electrophoresis, transferred to nitrocellulose membranes, blocked for 1 h with 5% nonfat milk, and incubated overnight at 4°C with primary antibodies (Table S2). Then, the membranes were incubated with peroxidase-conjugated donkey antirabbit immunoglobulin G (IgG) (dilution 1:25,000; Thermo Fisher Scientific) or mouse IgG kappa binding protein (dilution 1:1000, Santa Cruz Biotechnology). The signals were enhanced with chemiluminescence reagents (Amersham ECL Prime; GE Healthcare), and quantified with a Fusion camera and its Capt Fx Software (Vilber-Lourmat).

| Devitalization
Devitalization is a process of removal of cytoplasm and nuclear material from cell cultures to obtain an ECM that was previously laid down by the cells. MSC cultured for 11 days under normoxic or hypoxic conditions ± GO-201 were devitalized as described before (Baroncelli et al., 2018). In brief, the culture medium was removed from osteogenically differentiating MSC and cells were washed twice with phosphate buffered saline (PBS) (Gibco BRL). Cells were subjected to 3X freeze-thaw cycles followed by DNase treatment for 30 min at 37°C (10 U/ml). Finally, DNase solution was removed and cells were washed extensively using PBS. The matrices were air-dried and stored at −20°C until further use.
2.5 | Alkaline phosphatase (ALP) activity and ECM mineralization assays ALP activity was measured at 6, 11, and 19 days of culture, and matrix mineralization at 6, 11, and 19 days of culture, as previously described (Bruedigam et al., 2011). A protein assay was performed on the same samples to correct for cell numbers as described before (Bruedigam et al., 2011). 2.6 | Proteomic profiling of ECM using mass spectrometry (MS) The proteomic composition of the ECM was determined using a label-free quantification (LFQ) method using MS. ECM samples were collected in PBS and processed for in-solution trypsin digestion.
Then, samples were processed for MS and analyzed as previously described (Baroncelli et al., 2018). Raw MS data were analyzed by using MaxQuant Software (version 1.5.6.0) and the Andromeda search engine and annotated against the human proteome as provided by the Uniprot database (taxonomy: Homo sapiens, release HUMAN_2016_10.0). Samples were run in duplicate and then averaged for analysis.
Proteins uniquely present in a group or fold change ≥2 in the comparison between two groups were used as input. Fisher's exact test with Bonferroni correction (p value ≤0.05) was used to identify statistically significant pathways (Bland & Altman, 1995;Fisher, 1992).

| Quantitative measurement of gene expression
Total RNA was extracted at Days 1, 6, 11, and 19 of MSC cultures or at Day 5 of HUVEC cultures using TRIzol ® reagent (Invitrogen) according to manufacturer's instructions. RNA was isolated as described before (Bruedigam et al., 2011) and quantified by a NanoDrop ND-1000 spectrophotometer (NanoDrop Technologies, Inc.). The obtained messenger RNA (mRNA) was reverse-transcribed to complementary DNA (cDNA) using a cDNA synthesis kit according to the manufacturer's instructions (Applied Biosystems). Gene expression was evaluated by RT-PCR on a QuantStudio 7 Flex Real-Time PCR System (Applied Biosystems), using SYBR Green PCR master mix reagents (VWR International). All primers sets overspanning at least one intron-exon boundary are provided in supplementary table 3 (Table S3) After incubation, cells were washed twice with PBS containing 5% FBS and resuspended in nuclear stain solution (55 µl PBS/5% FBS + 0.5 µl propidium iodide). Cells were sorted using the Accuri C6 flow cytometer and analyzed with the associated software (BD Biosciences).

| Statistical analysis
The data are shown as representative of multiple independent experiments. Quantitative graphs are shown as mean ± SD.
One-way analysis of variance was performed, followed by the Bonferroni post hoc test to calculate statistical significance unless specified otherwise. A p value ≤0.05 was considered significantly different.

| Hypoxia stabilizes HIF1α and upregulates MUC1 and VEGFA in human osteoblasts
To assess the effect of hypoxia, we measured HIF1α on Days 1 and 6 and found 3-and 2.5-fold higher protein levels, respectively, in hypoxic cells compared to cells under normoxic conditions. In addition, MUC1-CT inhibition under 2% oxygen reduced HIF1α levels on Day 1 compared to untreated cells (Figure 1a). Furthermore, we explored the effect of hypoxia on MUC1 expression and found that hypoxia significantly increased MUC1 during osteogenic differentiation of MSCs ( Figure 1b). This was confirmed by increased MUC1-CT protein expression at Day 11 ( Figure 1c). Hypoxia also increased VEGF expression during the course of MSC differentiation towards osteoblasts, which is independent of MUC1-CT inhibition (Figure 1d). In addition, CTGF expression was increased in hypoxic osteoblast cultures on Day 11 but there was no effect on PDGF-A and PDGF-B expression ( Figure S1).

| MUC1-CT inhibition affects phospho-AMPK/ mTORC1/phospho-S6 pathway in MSCs
We looked at the expression and phosphorylation of several proteins involved in the AMPK/mTORC1/S6 pathway as it is involved in the osteogenic differentiation of MSC (Pantovic et al., 2013). Here, we observed that MUC1-CT inhibition under hypoxia significantly increased phosphorylation of AMPK after 24 h of treatment compared to normoxic cells that were treated with GO-201 but not compared to hypoxia alone (Figure 4a,b). In addition, compared to normoxia and independent of GO-201, hypoxia significantly increased phosphorylation of mTOR while phosphorylation of raptor was decreased. GO-201 reduced hypoxia-induced phosphorylation of S6 (Figure 4a,b). MUC1-CT inhibition under normoxia led to a

| Collagenous changes in the ECM following hypoxia and/or MUC1-CT inhibition are mediated through activation of TGFβ-smad and NFκB pathways
As hypoxia promotes collagen formation in dermal fibroblasts by TGF-β1/Smad signaling (Mingyuan et al., 2018), we investigated this in human osteoblasts. We found that hypoxia and/or MUC1-CT inhibition did not change TGFβ1 (Figure 5a). However, TGFβ2 was (sprouting) involves three sequential cells types, that is, pharynx cells, stalk cells, and tip cells (leading cell) (Gerhardt, 2013), the latter of which are characterized by an increased ratio of TIE1/ TIE2 (Savant et al., 2015). We observed that ECMs of MUC1-CTinhibited normoxic or hypoxic osteoblasts reduced TIE1 expression (Figure 7c), whereas TIE2 was unaffected (Figure 7d). We also checked for expression of angiopoietins since angiopoietin 1 (ANGPT1) along with TIE2 promotes vascular stabilization and maturation and therefore reduces "tip-ness" of endothelial cells (Brindle et al., 2006). We found significantly increased expression of ANGPT1 and ANGPT2 by ECMs of MUC1-CT-inhibited normoxic or hypoxic osteoblasts (Figure 7e

| DISCUSSION
In this study, we explored the effects of hypoxia and/or MUC1 inhibition on osteogenic differentiation of human MSCs and their ECM.
We claim that hypoxia upregulates MUC1 in osteoblasts. Hypoxia or MUC1-CT inhibition delays the osteogenic differentiation of MSCs, alters the proteomic composition of their ECMs, and leads to differential effects on subsequent osteogenesis and angiogenesis ( Figure 8).
We found that hypoxia reduced osteogenic differentiation and ECM mineralization of MSCs, which corroborated with findings of Utting et al. (2006)  HIF1α in osteoblasts reduced bone volume (Shomento et al., 2010). On the basis of our findings that hypoxia upregulated COL1A1, COL5A1, and COL5A3 in osteoblasts among which, collagen type I contributes to approximately 90% of the proteinpart of bone matrix, this may indicate that HIF1α helps to maintain bone volume by increasing collagen production.
MUC1 is upregulated by hypoxia and inhibition of its signaling mimics the effect of hypoxia on differentiation of osteoblasts and ECM mineralization. Inhibition of MUC1-CT destabilized HIF1α indicating that MUC1 might interfere in the hydroxylation or proteasomal degradation of HIF1α (Chaika et al., 2012). This also suggests that the C-terminal part of MUC1 could be a key transcription factor to control the "stemness" of hMSCs in the hypoxic niche of bone marrow as shown in colorectal cancer (Li et al., 2020).
Hypoxia activates AMPK signaling (Dengler, 2020). We noticed changes in the AMPK/mTORC1/S6K pathway as an early wave during osteogenic differentiation. We found that that hypoxia alone did not change the phosphorylation of AMPK. However, in combination with MUC1-CT inhibition phosphorylation of AMPK at threonine (position 172) was activated. We also confirmed that hy- Previous studies from our group have successfully applied devitalized ECM to study in vitro and in vivo osteogenic differentiation and bone formation (Baroncelli et al., 2018). Baroncelli et al. (2018) reported that in vitro cell-deposited ECM is osteopromotive and shared more than 50% homology with human bone proteome. In this study, we have identified the proteins in the ECM of normoxic or hypoxic osteoblasts with or without MUC1-CT inhibition. Hypoxia upregulated proteins are involved primarily in angiogenesis, cell migration, and TGFβ1 signaling. Surprisingly, hypoxic ECM shared similarities of pathways with the MUC1-CT inhibited normoxic ECM particularly in up-regulated proteins. Collagens were mostly upregulated in hypoxic ECM, which were in contrast with the fact that they are considered as an indicator of osteogenesis (Tsai et al., 2010).
However, mineralization of the matrix involves cross-linking of collagens (Knott & Bailey, 1998). Although the collagens were upregu- In conclusion, our study claims that MUC1 is critical for hypoxiamediated changes during osteoblastogenesis that not only alters the proteomic landscape of the ECM but thereby also modulate its osteogenic and angiogenic potentials. The study emphasizes that it is pivotal to understand the role of hypoxia and MUC1 in developing customized ECM scaffolds to induce osteogenesis.

DATA AVAILABILITY STATEMENT
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.