Breast Tumor Cell-Stimulated Bone Marrow-Derived Mesenchymal Stem Cells Promote the Sprouting Capacity of Endothelial Cells by Promoting VEGF Expression, Mediated in Part through HIF-1α Increase

Simple Summary ROS and JAK/Stat3 cooperatively upregulate the expression of HIF-1α in bone marrow-derived mesenchymal stem cells under normoxic conditions in response to breast tumor cells. The upregulation of HIF-1α contributes in part to the increase in VEGF expression in the bone marrow-derived mesenchymal stem cells. Bone marrow-derived mesenchymal stem cells improve the angiogenic sprouting capacity of mature endothelial cells in a VEGF-dependent manner. Abstract Breast tumor cells recruit bone marrow-derived mesenchymal stem cells (BM-MSCs) and alter their cellular characteristics to establish a tumor microenvironment. BM-MSCs enhance tumor angiogenesis through various mechanisms. We investigated the mechanisms by which BM-MSCs promote angiogenesis in response to breast tumor. Conditioned media from MDA-MB-231 (MDA CM) and MCF7 (MCF7 CM) breast tumor cells were used to mimic breast tumor conditions. An in vitro spheroid sprouting assay using human umbilical vein endothelial cells (HUVECs) was conducted to assess the angiogenesis-stimulating potential of BM-MSCs in response to breast tumors. The ROS inhibitor N-acetylcysteine (NAC) and JAK inhibitor ruxolitinib attenuated increased HIF-1α in BM-MSCs in response to MDA CM and MCF7 CM. HIF-1α knockdown or HIF-1β only partially downregulated VEGF expression and, therefore, the sprouting capacity of HUVECs in response to conditioned media from BM-MSCs treated with MDA CM or MCF7 CM. Inactivation of the VEGF receptor using sorafenib completely inhibited the HUVECs’ sprouting. Our results suggest that increased HIF-1α expression under normoxia in BM-MSCs in response to breast tumor cells is mediated by ROS and JAK/Stat3, and that both HIF-1α-dependent and -independent mechanisms increase VEGF expression in BM-MSCs to promote the angiogenic sprouting capacity of endothelial cells in a VEGF-dependent manner.

MDA-MB-231 and MCF7 cells were cultured using their culture media until confluence. The cells were then washed with phosphate-buffered saline (PBS) and incubated for 3 days with DMEM/HG supplemented with 100 U/mL P/S or DMEM/F-12 (1:1) supplemented with 2 mM L-glutamine and 100 U/mL P/S for MDA-MB-231 or MCF7 cells, respectively. To prepare conditioned medium (CM) for the control, DMEM/HG supplemented with 100 U/mL P/S or DMEM/F-12 (1:1) supplemented with 2 mM L-glutamine and 100 U/mL P/S was incubated for 3 days under cell-free conditions. The CM from MDA-MB-231 (MDA CM), MCF7 (MCF7 CM), or the control (CON CM) was filtered through a 0.2 µm filter (Corning, Cornyn, NY, USA), aliquoted, and stored at −80 • C until use.

Conditioned Media from BM-MSCs Primed with MDA CM or MCF7 CM
BM-MSCs were cultured in 6-well plates until 80-90% confluence and incubated for 24 h with DMEM/low glucose (DMEM/LG) supplemented with 2 mM L-glutamine and 100 U/mL P/S. Then, the BM-MSCs were rinsed with PBS and incubated with MDA CM, MCF7 CM, or CON CM for 48 h. The CM from the cells was collected, filtered through a 0.2 µm filter, aliquoted, and stored at −80 • C until use. For the control, MDA CM, MCF7 CM, or CON CM was incubated under cell-free conditions for 48 h. For knockdown experiments, BM-MSCs were transfected with siRNA targeting HIF1A or ARNT and incubated for 24 h prior to adding MDA CM or CON CM.

Quantitative Real-Time Polymerase Chain Reaction (qRT-PCR)
Total RNA was isolated using the TRIzol reagent (Invitrogen, Waltham, MA, USA), and cDNA was synthesized using SuperScript III Reverse Transcriptase (Invitrogen), according to the manufacturers' protocols. Quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) was performed using the SYBR Green reagent (Invitrogen). The human ribosomal protein S9 gene (RPS9) was used as an endogenous control. The primer sequences were as follows:

In Vitro Spheroid Sprouting Assay Using HUVEC Spheroids
HUVECs were suspended at a density of 1.33 × 10 4 cells/mL in Medium 199 (Sigma-Aldrich) supplemented with 10% (v/v) FBS and 0.2% (v/v) methylcellulose (Sigma-Aldrich). The cell suspension (30 µL) was seeded into the non-adherent lid of a Petri dish. The lid was turned upside-down and incubated for 24 h to assemble single spheroids (~400 cells/spheroid). The HUVEC spheroids were collected in PBS and centrifuged at 200× g for 5 min. The spheroids were then resuspended in a methylcellulose mixture containing 20% (v/v) FBS and 0.95% (v/v) methylcellulose in Medium 199. For preparing the type I collagen gel mixture, 3 mg/mL type I-A collagen (Nitta Gelatin) was mixed with 10× Medium 199 (Sigma-Aldrich) and 10× reconstitution buffer (0.05 N NaOH, 0.261 M NaHCO 3 , and 0.2 M HEPES) at a ratio of 8:1:1 on ice. The collagen mixture was then added to the methylcellulose mixture containing spheroids at a 1:1 ratio on ice. This mixture (0.7 mL) was transferred into a 24-well plate and immediately incubated at 37 • C and 5% CO 2 for 30 min to polymerize the gel. Then, 100 µL of Medium 199 containing 20 ng/mL VEGF 165 (PeproTech, Rocky Hill, NJ, USA) or CM from BM-MSCs primed with MDA CM, MCF7 CM, or CON CM was added on top of the collagen gel containing HUVEC spheroids, and the mixture was incubated for 24 h to induce spheroid sprouting. If necessary, 5 µM sorafenib (Sigma-Aldrich) was pretreated for 30 min prior to the treatment with VEGF 165 or the CM. At least 10 spheroids were analyzed per group, and both the number of sprouts and the sprout length of all sprouts per spheroid (i.e., cumulative sprout length) were measured using the ImageJ software (NIH).

Statistical Analysis
Quantitative data are expressed as the mean ± SD. Statistical analysis was performed using Student's t-test in GraphPad version 9.4.0 (GraphPad Software Inc.). A value of p < 0.05 was considered significant.  The activation of JAK/Stat signaling is associated with the establishment of the tumor microenvironment [44,45], and Stats play important roles in regulating the stability of HIF-1α proteins [36,46]. We found that the level of phosphorylated Stat3 increased in BM-MSCs incubated with MDA CM. Stat3 phosphorylation increased at 15 min after MDA CM treatment and declined thereafter ( Figure 2A). Ruxolitinib, a selective inhibitor of JAK [47], suppressed the increase in Stat3 phosphorylation in BM-MSCs treated with MDA CM ( Figure 2B). ROS induces the activation of JAK/Stat signaling [48] and has also been found to regulate the activity and stability of the HIF-1α protein [26,49,50]. N-acetylcysteine (NAC), an ROS inhibitor, suppressed the phosphorylation of Stat3 in BM-MSCs treated with MDA CM ( Figure 2C). Ruxolitinib and NAC suppressed the increase in HIF-1α expression in BM-MSCs treated with MDA CM ( Figure 2D,E). The elevated levels of HIF-1α protein in BM-MSCs incubated with MDA CM were more significantly suppressed with ruxolitinib and NAC co-treatment than with ruxolitinib or NAC treatment alone ( Figure 2F). BM-MSCs incubated with MCF7 CM showed increased JAK-mediated phosphorylation of Stat3, and NAC inhibited Stat3 phosphorylation in BM-MSCs ( Figure 2G,H). The MCF7 CMmediated increase in HIF-1α was suppressed in ruxolitinib-or NAC-pretreated BM-MSCs ( Figure 2I,J), and co-treatment with ruxolitinib and NAC showed an addictive effect and greatly suppressed MCF7 CM-mediated HIF-1α expression ( Figure 2K). Stat3 and ROS signaling have been shown to mediate the increase in HIF-1α transcription in both hypoxic and normoxic conditions [27,28,51]. Neither ruxolitinib nor NAC treatment was able to suppress the increase in HIF-1α mRNA levels in BM-MSCs treated with MDA CM ( Figure 2L,M).

BM-MSCs Regulate VEGF Expression in Response to Breast Tumor-Mimicking Conditions in Both an HIF-1α-Dependent and HIF-1α-Independent Manner
We previously reported that BM-MSCs show increased VEGF expression in response to MDA CM [52]. In this study, we found that BM-MSCs showed increased VEGF expression in response to both MCF7 CM and MDA CM ( Figure 3A). We assessed whether this increase in VEGF transcription was induced by HIF-1α. HIF-1α knockdown by transfection of HIF1A-targeted siRNA partially suppressed the increase in the mRNA expression of both HIF-1α and VEGF in BM-MSCs treated with MDA CM, compared to the control siRNA transfection ( Figure 3B,C). To exclude the possibility that the partial inhibition of VEGF was due to partial knockdown of HIF-1α, we observed the knockdown effect of HIF-1α on other genes that are upregulated by HIF-1α. Glucose transporter 1 (SLC2A1) is one of the major genes upregulated by HIF-1α. SLC2A1 expression was upregulated in BM-MSCs by MDA CM treatment. The expression of SLC2A1 decreased in HIF-1α-knockdown cells to almost the basal levels found in the control group ( Figure 3D), indicating successful knockdown. HIF-1α knockdown was confirmed by performing Western blot analysis ( Figure S3). HIF-1β is required for HIF-1α-dependent transcriptional regulation of HIF-1α target genes such as VEGF. HIF-1β knockdown partially suppressed the increase in VEGF expression but almost completely suppressed the increase in SLC2A1 in BM-MSCs treated with MDA CM (Figure 3E-G). These results suggest that ROS and the activity of JAK/Stat3 mediate the increase in HIF-1α of BM-MSCs in response to breast tumor-mimicking conditions ( Figure 2). Therefore, we assessed the effects of NAC and the inhibition of JAK/Stat3 signaling on VEGF expression in BM-MSCs treated with MDA CM. Ruxolitinib and NAC suppressed the MDA CM-induced increase in VEGF expression in a synergistic manner, but not completely ( Figure 3H). Ruxolitinib and NAC co-treatment partially suppressed VEGF expression, as did HIF1A siRNA ( Figure 3H).

BM-MSCs Regulate VEGF Expression in Response to Breast Tumor-Mimicking Conditions in Both an HIF-1α-Dependent and HIF-1α-Independent Manner
We previously reported that BM-MSCs show increased VEGF expression in response to MDA CM [52]. In this study, we found that BM-MSCs showed increased VEGF expression in response to both MCF7 CM and MDA CM ( Figure 3A). We assessed whether this increase in VEGF transcription was induced by HIF-1α. HIF-1α knockdown by transfection of HIF1A-targeted siRNA partially suppressed the increase in the mRNA suggest that ROS and the activity of JAK/Stat3 mediate the increase in HIF-1α of BM-MSCs in response to breast tumor-mimicking conditions ( Figure 2). Therefore, we assessed the effects of NAC and the inhibition of JAK/Stat3 signaling on VEGF expression in BM-MSCs treated with MDA CM. Ruxolitinib and NAC suppressed the MDA CMinduced increase in VEGF expression in a synergistic manner, but not completely (Figure 3H). Ruxolitinib and NAC co-treatment partially suppressed VEGF expression, as did HIF1A siRNA ( Figure 3H).

BM-MSCs Primed with Breast Tumor-Mimicking Conditions Enhance In Vitro Angiogenic Sprouting of HUVECs through VEGF Signaling
We analyzed the angiogenic sprouting ability of HUVEC spheroids in response to BM-MSCs primed with MDA CM, MCF7 CM, or CON CM ( Figure 4A). The number of sprouts per HUVEC spheroid and the cumulative sprout length of all sprouts per spheroid increased in response to BM-MSCs primed with MDA CM and MCF7 CM compared to CON CM ( Figure 4B-G). These results suggest that BM-MSCs treated with MDA CM had increased VEGF expression compared to those treated with CON CM in a partially HIF-1α-dependent manner (Figure 3). Consequently, HIF-1α or HIF-1β knockdown only partially impaired the sprouting capacity of HUVECs in response to BM-MSCs primed with MDA CM (Figure 4H-J). BM-MSCs treated with MDA CM increased VEGF expression by approximately 30-fold compared to BM-MSCs treated with CON CM ( Figure S4). We examined whether HUVEC sprouting was mediated by VEGF in BM-MSCs. VEGF treatment promoted angiogenic sprouting in HUVEC spheroids while sorafenib-an inhibitor of various receptor tyrosine kinases, including VEGFR [53,54]-suppressed it ( Figure 5A-C).

Discussion
Tumor angiogenesis is one of the most important tumor-tropic mechanisms. BM-MSCs have been found to increase tumor angiogenesis through various mechanisms, such as secretion of the angiogenic factor VEGF [8,14,16]. In this study, we demonstrated that BM-MSCs enhance the in vitro angiogenic sprouting capacity of HUVECs under normoxic conditions in response to stimulation by breast tumor cells in a VEGFdependent manner. ROS and JAK/Stat3 cooperatively increased HIF-1α expression un-

Discussion
Tumor angiogenesis is one of the most important tumor-tropic mechanisms. BM-MSCs have been found to increase tumor angiogenesis through various mechanisms, such as secretion of the angiogenic factor VEGF [8,14,16]. In this study, we demonstrated that BM-MSCs enhance the in vitro angiogenic sprouting capacity of HUVECs under normoxic conditions in response to stimulation by breast tumor cells in a VEGF-dependent manner. ROS and JAK/Stat3 cooperatively increased HIF-1α expression under normoxia in BM-MSCs in response to factors secreted from breast tumor cells. This, in turn, increased HIF-1α, and HIF-1α-independent mechanisms increased VEGF expression in BM-MSCs.
The hyperactivation of Stats that occurs in tumor cells and non-transformed cells inside tumors is associated with metastasis, immunosuppression, proliferation, and angiogenesis [44,45]. Stats regulate HIF-1α stability [36,46]. ROS that activate JAK/Stat signaling [48] enhance the stability of HIF-1α through various mechanisms, such as by modulating PHD enzyme activity [49]. The ROS inhibitor NAC and the JAK inhibitor ruxolitinib suppressed the increase in HIF-1α protein in BM-MSCs in response to breast tumor-mimicking conditions. NAC and ruxolitinib suppressed the phosphorylation of Stat3 in BM-MSCs treated with MDA CM or MCF7 CM. These results suggest that ROS function upstream of Stat3 to enhance HIF-1α expression in BM-MSCs in response to breast tumor-mimicking conditions. However, the mechanism by which ROS activates Stat3 in BM-MSCs in response to breast tumor-mimicking conditions and the mechanism by which ROS and JAK/Stat3 increase HIF-1α expression remain unclear and must be further investigated. The levels of both HIF-1α mRNA and HIF-1α protein increased in BM-MSCs in response to breast tumor-mimicking conditions. It has been reported that ROS activate the HIF1A promoter via nuclear factor kappa B (NF-κB) or NRF2 [51,55]. However, neither ROS nor JAK/Stat3 was associated with the increase in HIF-1α mRNA in BM-MSCs treated with MDA CM. Whether the increase in HIF-1α mRNA is due to increased transcription of HIF-1α in BM-MSCs in response to breast tumor-mimicking conditions or due to HIF-1α mRNA stabilization is not yet known. Furthermore, it is unclear whether the increase in HIF-1α mRNA contributes to the increase in HIF-1α protein in BM-MSCs faced with breast tumors. Therefore, further experiments are required to clarify the mechanisms that increase HIF-1α mRNA levels in BM-MSCs and to determine whether the increase in HIF-1α mRNA contributes to the increase in HIF-1α protein levels.
Increased VEGF expression through HIF-1α-independent mechanisms has been reported. For example, the RAS oncogene has been shown to be able to induce VEGF expression in human cancer cells where HIF-1α is unable to bind to the VEGF promoter due to mutations in the HIF-1α binding sites of the promoter [34,56]. NF-κB has also been shown to regulate VEGF expression in response to cytokines such as interleukin 8 (IL-8), independent of HIF-1α [34]. VEGF is induced in response to TGF-β, regardless of hypoxia [57,58]. Breast tumor cells, including MDA-MB-231 and MCF7 cells, express TGF-β and various cytokines such as IL-8. Further experiments are required to determine the mechanism by which BM-MSCs increase VEGF expression independent of HIF-1α in response to breast tumor cells. BM-MSCs, in response to breast tumor-mimicking conditions, showed increased VEGF expression in vitro under normoxic conditions. However, hypoxia is a characteristic of the tumor microenvironment [59]. Therefore, the mechanisms that mediate the increase in HIF-1α levels in BM-MSCs under hypoxic conditions mimicking the tumor microenvironment, along with the mechanisms by which BM-MSCs regulate VEGF expression, must be investigated.

Conclusions
BM-MSCs play essential roles in inducing tumor tropism in the tumor microenvironment. As tumors grow, angiogenesis is required for tumor cells' proliferation and survival. We demonstrated that ROS and JAK/Stat3 synergistically lead to an increase in HIF-1α in BM-MSCs in vitro under normoxic conditions mimicking the breast tumor microenvironment. Furthermore, HIF-1α induced by in vitro normoxic conditions mimicking the breast tumor microenvironment promoted VEGF expression in BM-MSCs which, in turn, enhanced the angiogenic sprouting capacity of HUVECs. However, HIF-1α only partially induced VEGF expression, and BM-MSCs also showed HIF-1α-independent VEGF expression. A further understanding of the mechanisms underlying HIF-1α-independent induction of VEGF may provide therapeutic strategies for controlling angiogenesis enhanced by factors of the tumor microenvironment, such as BM-MSC secretions.

Data Availability Statement:
The data that support the findings of this study are available from the corresponding author upon reasonable request.