Background
Neovascularization is not only a prerequisite of tumor growth, but also initiates or enhances other malignant behaviors of cancer, such as invasion and metastasis [
1,
2]. While endothelial cells (ECs) play multidimensional roles in both physiological and pathological vascularization [
3,
4], vascular mural cells (vMCs), including vascular smooth muscle cells (vSMCs) and pericytes, are also essential for vessel development and functions [
5]. Under physiological conditions, vMCs are fundamental for maintaining vessel structure and regulating vessel contraction/relaxation and other functions [
6]. However, tumor vessels are characterized by reduced and/or abnormal vMCs, leading to destabilized tumor vasculature [
2]. Moreover, vMCs often lose their anatomical localization in tumors, and switch from a contractile into a secretory/proliferation phenotype, contributing to the cancer-associated fibroblasts (CAFs) repertoire [
7‐
9]. With the secretory/proliferation phenotype, vMCs produce cytokines and chemokines to facilitate proliferation, invasion, and metastasis of tumor cells and an immune-suppressive milieu to strengthen tumor malignancy [
7,
10]. Elucidating the precise functions and regulation of vMCs in tumors could provide novel strategies for efficient tumor therapy [
11].
Several signaling pathways and transcriptional factors, such as nuclear factor kappa B (NF-κB), have been implicated in vMCs in tumors [
10‐
12]. The Notch signaling pathway, which is composed of Notch ligands (Dll1, 3, and 4, and Jagged 1 and 2), Notch receptors 1–4, transcription factor recombination signal binding protein Jκ (RBPj), and downstream Hes family effectors, plays a critical role in cell fate determination in vascular development and homeostasis [
13,
14]. Notch signaling is initiated by γ-secretase-dependent cleavages of Notch receptors, liberating the Notch intracellular domain (NIC) that serves as a transcription factor to transactivate RBPj. The Notch pathway plays an essential role in the development of vSMCs because mutations in Notch-related molecules have been associated with several human genetic diseases involving vSMCs [
14,
15]. More recently, Notch signaling has been implicated in vSMC phenotype switch, which is involved in cardiovascular diseases [
15]. Blocking Notch signaling leads to CAF activation and promotes CAF and tumor cell expansion [
15‐
17]. However, the exact role of Notch signaling in vMCs in cancer remains unelucidated.
SM22α is a 22 kDa protein that physically associates with cytoskeletal actin filament bundles in contractile vSMCs [
18,
19]. Previous studies have shown that SM22α is abundantly expressed in vSMCs and myofibroblasts in tumors [
20]. In this study, we show that SM22α
+ cells are primarily distributed in the perivascular region of tumors and are essential for vessel stability. Moreover, SM22α
+ vMC (SM22-MC) phenotypes are modified by the tumor microenvironment (TME). We demonstrate that Notch signaling plays a critical role in regulating SM22-MC phenotypes, namely, Notch activation promotes contractile and represses secretory phenotypes in SM22-MCs. Our results provide an insight into the dual roles of SM22-MCs in tumors, and show that intervening with the phenotype of these cells could help cancer therapy.
Methods
Human samples
Samples of low differentiation human lung adenocarcinoma biopsies were purchased from Wuhan Servicebio Technology Co., Ltd. (Wuhan, China) (Additional file
2: Table S1).
Animals
Mice of C57BL/6 background with specific genetic modifications (Additional file
1: Figure S1A) were maintained under specific pathogen-free (SPF) condition. The SM22α-CreER
T2 mice, which express a tamoxifen-inducible Cre recombinase under the control of SM22 promoter, were kindly provided by Zhu MH [
19]. Rosa-Stop
floxed-NIC mice harbor a murine Notch1 NIC (amino acids 1749–2293) gene followed by Ires-GFP inserted at the Rosa26 locus, and were originally derived from Jackson Laboratory (stock #008159, Bar Harbor, ME). RBPj
floxed mice were previously reported [
21]. Cre-mediated recombination activates or blocks Notch signaling by ectopically overexpressing NIC or disrupting RBPj, respectively, in these mice [
21,
22]. Rosa-Stop
floxed-tdTomato (stock #007909) mice and Rosa-Stop
floxed-DTA (stock # 009669) mice, which label or kill specific cell types by Cre-activated expression of tdTomato or diphtheria toxin A chain (DTA), respectively, were also obtained from Jackson Laboratory. The mice were genotyped by PCR, using tail DNA as a template and primers listed in Additional file
2: Table S2. For induction of Cre-mediated recombination, 6- to 8-week-old male mice were intraperitoneally (i.p) injected every day with tamoxifen (50 mg/Kg, Sigma-Aldrich, St. Louis, MO) for 5 consecutive days.
For tumor models, LLC and B16-F10 melanoma cells were inoculated subcutaneously (s.c) on the right back of mice (1 × 106 cells/100 μl PBS) a day after the last tamoxifen injection. Tumors were removed 21 (LLC) or 16 (B16) days post-inoculation. Tumors were weighed and their sizes evaluated using a caliper (tumor size = π × [d2 × D]/6, with d = short diameter and D = long diameter). For metastasis, LLC cells were transfected with lentivirus overexpressing luciferase (GeneChem Technology, Inc., Shanghai, China), following the supplier’s protocol. Mice were inoculated with luciferase-expressing LLC cells. Tumors were surgically removed on day 14 under anesthesia. The mice were maintained 21 days further, and then injected with D-luciferin (150 mg/Kg, Yeasen Biotechnology, Shanghai, China), and sacrificed 8 min later. Lungs were collected and analyzed using a bioluminescence imaging system (IVIS) (Xenogen, Perkin-Elmer, Fremont, CA, USA). To detect CTCs, LLC cells were labeled with GFP using lentivirus (GeneChem). GFP+ LLC cells were inoculated as above. Blood was collected 21 days post-inoculation. After erythrolysis using a Red-lysis buffer (Cwbio, Beijing, China), GFP+ cells were counted under a fluorescence microscope (BX51, Olympus, Tokyo, Japan).
Histology
Mice were anesthetized and cardiac-perfused with PBS. Tumors were harvested, sectioned, and routinely stained with hematoxylin and eosin (H&E). For immunofluorescence, tissues were fixed in 4% paraformaldehyde (PFA) at 4 °C for 4 h and then transferred into 30% sucrose overnight until tissues sank. Samples were embedded in optimal cutting temperature (OCT) compound (Sakura Finetek, Inc., Torrance, CA), cryosectioned at 8-μm thickness, and air-dried for 2 h at room temperature. Sections were blocked and permeabilized with 1% bovine serum albumin (BSA) plus 0.5% Triton X-100 in PBS, and then incubated overnight at 4 °C with primary antibodies. After washing, sections were stained with matched secondary antibodies at 37 °C for 1 h. Images were captured with a confocal fluorescence microscope (A1R, Nikon Instruments, Inc., Shanghai, China). Antibodies are listed in Additional file
2: Table S3. Immunohistochemistry was performed in a similar way, except that the secondary antibodies were horseradish peroxidase (HRP)-labeled and the sections were developed using a 3, 3′-diaminobenzidine (DAB) substrate kit (Zhong Shan Jin Qiao Biotech, Beijing, China). After counter-staining with hematoxylin, images were acquired under a microscope. To assess hypoxia, tumor-bearing mice were injected with pimonidazole hydrochloride (PIMO, 60 mg/Kg) 1 h before tumor collection. Cryosections were then immunostained with a Hypoxyprobe-1-Mab1 kit (Hypoxyprobe, Inc., Burlington, MA), following the manufacturer’s instructions. In some experiments, hypoxia was evaluated by Glut1 immunofluorescence. To evaluate vascular perfusion, mice were injected intravenously (i.v) with 100 μl of FITC-conjugated Dextran-2MD (25 mg/ml) (Sigma-Aldrich). Mice were perfused with PBS 15 min later and tumors were dissected and analyzed using immunofluorescence.
Images were quantitatively analyzed with Image-Pro Plus 6.0 or Image J (NIH). In general, over four random high-power fields of each slide were captured, and independently quantified by two investigators in a blinded fashion. Necrosis, hypoxia and hemorrhage were quantified by calculating necrotic, hypoxic and hemorrhagic areas as the percentage of whole tissue areas in each field. Tumor cell proliferation was quantified by counting the number of Ki67+ cells as the percentage of total number of Hoechst+ cells in each field. Vessel density was assessed by calculating CD31+ area as the percentage of whole tissue areas in each field. Mural cell coverage and vessel perfusion were quantified by the ratio of signal density of SM22α, α-SMA, CNN1, SMMHC, or Dextran-2MD to that of CD31 determined by Image-Pro Plus 6.0. The average proximity of SM22α+ cells to vessels was measured by determining the radial distribution of SM22α+ cells to the CD31+ vessels using the Image J software (NIH).
Flow cytometry assay
For CD31+CD45− cells in peripheral blood detection, peripheral blood was collected and erythrocyte were lysed. Cells were then resuspended and stained with anti-CD31-FITC and anti-CD45-APC antibodies. The concentration of CD31+CD45− cells in peripheral blood were analyzed using a FACSCanto II flow cytometer (BD Biosciences, San Jose, CA).
For tdTomato+ cells isolation, tissues were harvested and digested in 1 mg/ml collagenase I and 10 μg/mL DNase I (Sigma-Aldrich) for 40 min at 37 °C. After passing through a 70-μm tissue strainer, cell suspensions were centrifuged for 3 min at 1300 rpm at 4 °C, followed by erythrolysis. The tdTomato+ cells were sorted using a FACSAriaII™ flow cytometer (BD Biosciences) and immediately used for RNA isolation.
Cell culture and transfection
LLC, B16-F10, and bEnd.3 murine EC cell lines were purchased from American Type Culture Collection (ATCC, Manassas, VA). Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM), supplemented with 10% fetal calf serum (FCS) and 2 mM L-glutamine (Invitrogen, Carlsbad, CA). The γ-secretase inhibitor (DAPT, Alexis Biochemicals, Lausen, Switzerland) was used at a concentration of 25 μM.
Adenovirus expressing human Notch1 ICD (AdNIC, 5261 ~ 7665 bp from the 1st coding region nucleotide, NM_017617.4) was purchased from Vigene Biosciences (Parklawn, Rockville, MD). To isolate primary vSMCs, the dorsal aorta of mice was minced mechanically and digested in 1 mg/ml collagenase I and 10 μg/ml of DNase I for 30 min at 37 °C. After passing through a 70-μm tissue strainer, cell suspension was centrifuged for 4 min at 1200 rpm at 4 °C and then resuspended in DMEM containing 10% FBS. Cells (vSMCs-DA) were then plated in a culture dish and transduced with AdNIC or control adenovirus (AdCtrl) at MOI = 300, according to the supplier’s procedures. Cells were harvested 24 or 48 h post-viral infection.
To prepare tumor cell-conditioned medium (TCM), tumor cells were plated (5 × 106 cells) and cultured overnight in 10-cm culture dishes, and the medium replaced with 10 ml serum-free medium (SFM) at about 70% confluence. The medium was collected 48 h after the medium change, filtered through a 0.22-μm filter, and centrifugated for 10 min at 12,000 rpm at 4 °C. vSMCs-DA (1 × 105 cells/ml) were treated with TCM for 24 or 48 h, with SFM as a control.
To prepare CM from vSMCs-DA, cells were seeded and cultured overnight in 6-well plates at a density of 5 × 105 cells/well, and then transduced with AdNIC or AdCtrl, or treated with DAPT and DMSO for 48 h. The medium was then replaced with 1.5 ml serum-free medium. The medium was collected 48 h later, and filtered through a 0.22-μm filter, and centrifugated for 10 min at 12,000 rpm at 4 °C before use.
In vitro cell proliferation, migration, and adhesion assays
Cell proliferation was evaluated by incubation with 50 μmol/L EdU (RiboBio Co., LTD, Guangzhou, China) in complete medium for 2 h and then fixed in 4% PFA at room temperature for 30 min. Cells were stained with Apollo 567 according to the standard procedures. Images were captured with a fluorescence microscope. Cell proliferation was evaluated by the number of EdU+ cells in total Hoechst+ cells in at least three fields of each stained sample.
For cell migration, AdNIC- or AdCtrl-transduced or DAPT- or DMSO-treated vSMCs-DA were dissociated and cultured in a Transwell chamber (Merck Millipore, Darmstadt, Germany) in complete medium for 24 h. Cells migrating to the lower side of the polycarbonate membrane were stained with crystal violet and evaluated under a microscope. Cell invasion was evaluated in a similar way using a matrigel invasion chamber (Corning, NY), according to the manufacturer’s instructions. Briefly, LLC or B16-F10 cells were seeded in the invasion chambers and CM from vSMCs-DA was added to the lower chambers. After culturing for 24 h, cells were fixed in 4% PFA for 20 min and stained with crystal violet as above. The number of crystal violet+ cells per field were counted using Image-Pro Plus 6.0.
To evaluate vSMCs-DA adhesion, cells were incubated with Vybrant® DiI (Thermo Fisher Scientific, Waltham, MA) for 15 min at 37 °C and then washed with serum-free medium for 3 times. The labeled cells were plated in 24-well plates, which were pre-coated with bEnd.3 cells. Non-adherent cells were discarded 2 h later by washing with serum-free medium for 3 times. The number of DiI+ adhesion cells per field were observed under a fluorescence microscope and counted using Image-Pro Plus 6.0.
Collagen gel contraction assay
Collagen gel contraction assay was performed as described [
23]. Briefly, vSMCs-DA were resuspended in serum-free medium at 1 × 10
7/ml, and 10 μl of the cell suspension was mixed with 90 μl of cold neutralized collagen gel solution (Solarbio, Beijing, China) and added to one well of low adhesion 96-well cell contraction plate (Corning). Gels were allowed to solidify for 20 min at 37 °C. After polymerization, 100 μl of medium containing 1% FBS was added to the well. Gels without cells were incubated as negative control. After 24 h, the gels were fixed in 4% PFA for 20 min and captured by a camera. The area of each collagen gel was determined by Image-Pro Plus6.0, and the contraction rate (%) was calculated as (x-n)/n (x = area of each group with cells, n = area of negative groups).
Enzyme-linked immunosorbent assay (ELISA)
TNFα and Cxcl10 were evaluated using ELISA kits (Thermo Fisher) according to the manufacturer’s instructions. The results were read at 450 nm using a microplate reader (Biotek, Winooski, Vermont).
RNA sequencing
RNA sequencing and data analyses were conducted by a commercial service (RiboBio). Briefly, vSMCs-DA transfected with AdNIC or AdCtrl for 48 h were subjected to RNA extraction, using the Trizol reagent (Thermo Fisher). The RNA integrity was evaluated using the Agilent 2200 Tape Station (Agilent Technologies, Santa Clara, CA). Ribosomal RNA (rRNA) was removed using the Epicentre Ribo-Zero rRNA Removal Kit (Illumina, San Diego, C). The remaining RNA was fragmented into approximately 200 bp fragments. Subsequently, samples were subjected to first- and second-strand cDNA synthesis, followed by adaptor ligation and enrichment with a low-cycle PCR using Tru Seq® RNA LT/HT Sample Prep Kit (Illumina). The purified library products were evaluated using the Agilent 2200 Tape Station and Qubit® 2.0 (Life Technologies, Foster City, CA) and then diluted to 10 pM for cluster generation in situ on the HiSeq3000 pair-end flow cell, followed by sequencing (2 × 150 bp) on HiSeq3000. Bioinformatic analyses were performed using the OmicShare tools at
www.omicshare.com/tools or TBtools at
github.com/CJ-Chen/TBtools/releases.
qRT-PCR
Total RNA was isolated and precipitated using TRIzol. cDNA was synthesized using a reverse transcription kit (Takara Biotechnology, Dalian, China). Real-time PCR was performed using a SYBR Premix Ex Taq Kit (Takara Biotechnology) and ABI PRISM7500 real-time PCR system (Life Technologies), with β-actin as an internal control. Primers are listed in Additional file
2: Table S2.
Western blotting
Cells were lysed in the RIPA buffer (Beyotime, Shanghai, China), containing 10 mM phenylmethanesulfonyl fluoride (PMSF), and the protein concentration determined using a BCA protein assay kit (Thermo Fisher). To detect the protein level of nuclear p65, the nuclear extraction was prepared using an extraction kit (Beyotime) according to the manufacturer’s instruction. Total cell lysates and nuclear proteins were then separated by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE), transferred onto polyvinylidene fluoride (PVDF) membranes, probed with primary antibodies, followed by HRP-conjugated secondary antibodies. β-actin was used as a loading control. HRP-based detection was done using an enhanced chemiluminescence (ECL) system (Clinx Science Instruments, Shanghai, China). Antibodies are listed in Additional file
2: Table S3.
Statistical analyses
Statistical analyses were performed with Image-Pro Plus 6.0, Image J (NIH) and GraphPad Prism 8.0 software. Data are expressed as means ± SD. Statistical significance was calculated using Student’s t-test. P < 0.05 was considered significant.
Discussion
The multi-dimensional roles of stromal cells in tumor malignancy have been appreciated for decades [
8,
10]. But, so far controversial results have been reported depending on tumor types and experimental settings. Depletion of CAFs or pericytes induces immunosuppression and hypoxia-associated epithelial-to-mesenchymal transition (EMT), respectively, leading to aggravated malignancy [
9,
32]. Xian et al. have also demonstrated that pericytes limit tumor cell metastasis [
33]. A more recent report by Tian et al. identified a group of genes associated with good prognosis (GPAGs) in human cancers, many of which are related to vMCs [
34]. These findings collectively suggest that stromal cells play a suppressive role in tumor progression. However, on the other hand, many studies have also indicated that tumor stromal cells support tumor growth and exacerbate malignancy [
35]. The discrepant functions of stromal cells in tumors could at least partially be attributed to the different subsets of tumor stromal cells, which are supposed to originate from tissue-resident fibroblasts, vMCs, blood-derived cells, and/or tumor cells after EMT [
35,
36]. Moreover, phenotypic plasticity of stromal cells could further complicate the situation [
8,
10]. In the current study, we have focused on a subset of tumor stromal cells, SM22-MCs. Because SM22α is reportedly expressed by vSMCs with relative specificity [
18‐
20], SM22-MCs mostly represent vessel-derived tumor stromal cells, although we could not provide definite evidence for this opinion. Functionally, SM22-MCs may regulate vascular structure and response to extracellular signals [
37]. In our study, genetic depletion of SM22-MCs promotes tumor malignancy, accompanied by exacerbated vessel anomalies, demonstrating that SM22-MCs are essential for vessel stability.
However, the vessel-stabilizing role of SM22-MCs appears to depend on TME. Indeed, vSMCs are highly plastic under various pathological conditions [
15,
38]. Our data have also shown that in response to diffusible factors from tumor cells, like other CAF types [
35], vSMCs lose their contractile phenotype and acquire a secretory phenotype, likely dampening their vessel-stabilizing capacity and induces tumor cell proliferation, migration, and invasion. Because this phenotypic transition of vSMCs is accompanied by alterations in Notch signaling, which regulates vSMCs phenotypic transition in other context [
14,
15], we tried to modulate the phenotypic transition of SM22-MCs in tumors by genetic modification of Notch signaling, specifically in SM22α
+ cells. Our results have shown that genetic activation of Notch signaling in SM22-MCs results in stabilized tumor vasculature, accompanied by enhanced contractile phenotype and dampened secretory phenotype, while blockade of Notch signaling by conditional knockout of RBPj or pharmaceutical inhibition of γ-secretase exhibits opposite effects. Jag1-KO embryonic vSMCs also show downregulated contractile phenotype-related genes and upregulated secretory phenotype-related genes, likely reflecting the conservation of Notch signaling. These results strongly suggest that in TME, SM22-MCs are likely to be deprived of their vessel-stabilizing ability due to phenotypic transition, which is controlled by Notch signaling and tends to promote tumor growth and metastasis. Of note, SM22-MCs could migrate into tumor stroma, which is likely a result of phenotypic transition, to promote proliferation and invasion of tumor cells and recruit immune cells via secreted cytokines and chemokines, but the contribution of these activities to tumor malignancy requires further investigation. Further studies are also required to decipher the involvement of canonical or non-canonical Notch signaling in this process. Moreover, TME in different types of tumors may respond to SM22-MCs in heterogenous manners, because vessel density showed a tendency of decrease in LLC (Fig.
4a) but increased in B16 tumors (Additional file
1: Figure S4E), although in both cases mural cell coverage of vessels decreased upon RBP-J ablation. More experiments are required to address this question.
The Notch signaling pathway participates in regulating TME development and remodeling, but the consequence of Notch activation is cell type- and/or context-dependent [
39]. In tumor stromal cells, loss of RBPj has been shown to promote fibroblast activation and conversion into CAFs and ensures keratinocyte-derived tumors [
17,
40]. Notch is also a well-recognized regulator of EMT, thus, influences the generation of cancer cell-derived tumor stromal cells [
41]. In vSMCs, Notch signaling plays a critical role in regulating development and plasticity [
15]. It is noteworthy that Notch3 is one of the most important regulator of vSMC in different districts [
42,
43], and it will be interesting to examine the differential role of each Notch receptor in future. Taking SM22α as a tentative vSMC marker, our results indicate that Notch inhibition promotes the transition of vSMCs from the contractile to secretory phenotype, leading to destabilized microvessels and vicious paracrine factors that facilitate tumor progression.
The mechanism underlying Notch-mediated regulation could be multiple. In TME, Notch signaling crosstalks with many other signals, including TGF-β, WNT, YAP, among others [
39]. These interactions form a robust network to elicit a pro-tumor stromal environment via promoting tumor growth and inhibiting anti-tumor immunity [
39]. Notch activation could inhibit vSMC proliferation and transdifferentiation via p27 and SOX9, respectively [
44]. Goruppi et al. have shown that autophagy, which is frequently activated in TME, regulates CAFs via Notch/RBPj signaling [
45]. Moreover, senescence, which is induced by over-proliferation and DNA damage, has been implicated in CAF activation and function [
46]. Recent reports have highlighted the role of Notch/RBPj signaling in the regulation of senescence. Procopio et al. have provided evidence that combined RBPj and p53 downregulation promotes CAF activation [
16]. Consistently, Bottoni et al. showed recently that RBPj controls telomere maintenance and genome stability in human dermal fibroblasts [
40]. These findings imply that blockade of Notch signaling would promote CAF activation and phenotype via promoting senescence. However, other reports have shown that activation of Notch signaling in cancer cells could also promote senescent phenotypes [
47]. More intensive studies, especially at the epigenetic level, could be important to reveal the mechanisms underlying Notch-mediated regulation of different subsets of tumor stromal cells.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.