Background
In normal tissues, fibroblasts are the major cellular component of connective tissue and are key participants in maintaining homeostasis of the extracellular matrix (ECM), regulating epithelial differentiation, inflammation and wound healing. Fibroblasts not only synthesize the major constituents of the ECM, but they release ECM-degrading proteinases to assure normal matrix turnover and function. Fibroblasts also secrete multiple growth factors and support mesenchymal-epithelial interactions via paracrine and juxtacrine signaling. Within the tumor stroma, subpopulations of fibroblasts emerge and exhibit an "activated" phenotype, whereupon they acquire characteristics that can be distin guished from normal fibroblasts and often portend a bad prognosis [
1]. These activated fibroblasts, also referred to as peritumoral fibroblasts, cancer-associated fibroblasts, reactive stromal fibroblasts and tumor-associated fibroblasts, are characterized by the expression of myofibroblast-like cell markers, including alpha smooth muscle actin (α-SMA) and desmin, and secrete factors that generally promote cell growth and proliferation (e.g. hepatocyte growth factor (HGF), epidermal growth factor (EGF), vascular endothelial growth factor (VEGF), insulin growth factor 2 (IGF2), fibroblast-like growth factor 2 (FGF2), transforming growth factor-beta (TGF-β), ECM-degrading proteinases such as MMPs, cytokines such as tumor necrosis factor (TNF)-αand interleukin (IL)-1β and chemokines [
2‐
4].
Tumor-associated fibroblasts are believed to originate from tissue-resident fibroblasts and mesenchymal stem cells, by recruitment of bone marrow-derived cells from the circulation [
5] and/or by epithelial-to-mesenchymal transition [
6]. The mechanisms by which fibroblasts become activated are not well-defined, although TGF-β, EGF, platelet-derived growth factor (PDGF)-BB, FGF2, reactive oxygen species, complement factors, and integrins have all been implicated [
7‐
9]. Although there are major gaps in our understanding of the mechanisms by which tumor-associated fibroblasts evolve, cell surface markers that are specific to these cells are attractive candidate targets for therapy.
CD248, also referred to as endosialin or tumor endothelial marker 1 (TEM1), is a highly sialylated cell surface glycoprotein [
10‐
12] that has been shown to be restricted to activated stromal and perivascular fibroblasts [
13‐
16]}. During normal embryonic development, CD248 is highly expressed [
17,
18], but by full-term, CD248 has almost entirely disappeared. Postnatally, expression is retained only in the endometrium, in bone marrow fibroblasts and in the corpus luteum [
11,
15,
19]. However, CD248 is frequently upregulated in tumors [
10,
15,
20], with particularly high expression in tumor associated stromal fibroblasts in sarcomas [
19] and primary and secondary brain tumors [
21]. CD248 is also expressed in human mesenchymal stem cells from bone marrow, which may differentiate into tumor stromal fibroblasts [
22]. In breast cancer and neuroblastomas, CD248 expression levels have been directly correlated with tumor grade, invasiveness and poor prognosis [
23,
24]. The physiologic importance of CD248 in cancer progression and its potential utility as a therapeutic target is further highlighted by the finding that lack of CD248 in mice results in resistance to the growth and metastasis of some tumors [
25]. Therefore, delineating the hitherto unknown mechanisms by which CD248 regulates tumor growth is important for the development of therapeutic strategies.
The human
CD248 gene is intronless and encodes a 95-kDa multi-domain type I transmembrane protein of 757 amino acids [
26]. The protein comprises an N-terminal C-type lectin-like domain, a Sushi domain, three epidermal growth factor (EGF)-like repeats, a mucin-like region, a single transmembrane segment and a 51 amino acid residue cytoplasmic tail with potential sites for phosphorylation [
27]. CD248 belongs to a family of proteins containing C-type lectin-like domains which have functions in cell adhesion and regulation of inflammation [
28,
29].
Few analyses have been performed to elucidate the mechanisms by which CD248 regulates tumor growth.
In vitro studies suggest that the extracellular region of CD248 may interact with ECM proteins, thereby facilitating activation of MMP-9, cell migration and metastasis formation [
30,
31]. We recently demonstrated that the highly conserved cytoplasmic domain of CD248 mediates signals that regulate stromal fibroblast function in an experimental model of rheumatoid arthritis [
32] and hypothesized that it would play a similarly important role in modulating tumor growth. We show that lack of the cytoplasmic domain of CD248 in transgenic mice results in reduced tumor growth, with alterations in fibroblast signaling via TGF-β, PDGF-BB, and Notch pathways, and establishment of a pattern of gene expression favoring tumor suppression. The findings extend previous reports of the importance of CD248 in tumor growth and point to the cytoplasmic domain of CD248 as a potential therapeutic target in neoplasia.
Methods
Mice
Transgenic mice lacking CD248 (CD248
KO/KO) or the cytoplasmic domain of CD248 (CD248
CyD/CyD) were previously generated and genotyped as reported [
32]. Mice were maintained on a C57Bl6 genetic background and corresponding wild-type mice (CD248
WT/WT), generated from siblings during breeding of the CD248 transgenic lines, were used as controls.
In vivo tumor models
Heterotopic implantation of Lewis Lung Carcinoma (LLC) tumor fragments was performed as described in [
25]. Briefly, 0.5 × 10
6 LLC cells were injected subcutaneously (s.c.) into the right flank of 5-week-old CD248
WT/WT mice. After 20 days, mice were sacrificed and tumors dissected and cut into 1 mm
3 pieces. 6-7 week old mice were anaesthetized with isoflurane and the cecum exteriorized via a small incision parallel to the midline. A single tumor fragment was implanted on the serosal surface of the cecum. Tumors were dissected 15 days after implantation, and tumor volume and weight were measured. Volume was calculated using the formula, length × width
2 × π/6.
For T241 fibrosarcoma studies, 1 × 106 cells in 200 μL PBS or 7.5 × 104 LLC cells in 50 μL PBS were injected s.c. into the right flank or footpad of 7-9 week-old mice. Tumor size was evaluated every 2 days using a caliper and weights were obtained after dissection. Studies were performed in a blinded manner to the investigator.
The model of orthotopic growth and metastasis of pancreatic adenocarcinoma in mice was performed exactly as reported [
33]. Briefly, mice were anaesthetized with isoflurane and the stomach exteriorized via an abdominal midline incision. 1 × 10
6 PancO2 pancreatic adenocarcinoma cells in 25 μL PBS were injected into the head of the pancreas. At day 11, primary tumors were dissected, and tumor volume and weight were determined.
Immunohistochemistry and quantification of tumor vessel density
Tissue samples were fixed with 2% paraformaldehyde overnight at 4°C, dehydrated and embedded in paraffin. Serial 7 μm sections were cut for histological analysis. Immunohistochemical detection was performed using the following antibodies: rat anti-CD31 (BD Pharmingen, Erembodegem, Belgium), mouse anti-SMA-Cy3 (Dako, Glostrup, Denmark), phospho-histone H3 (Cell Signaling Technology, Bioké, Leiden, the Netherlands) and rabbit anti-caspase 3 (Abcam, Cambridge, MA, USA). Morphometric analyses were performed using a Zeiss Imager Z1 or AxioPlan 2 microscope with KS300 image analysis software. For all studies, 6-8 optical fields per tumor section, at 40× or 80× magnification, were randomly chosen and analyzed. Cell proliferation was calculated as the number of phospho-histone H3 per mm2. Vessel density was calculated as the number of CD31-positive vessels per mm2 and pericyte coverage as the percentage of CD31-positive vessels that are covered by SMA-positive cells. Vessel distribution was determined by calculating the frequency distribution of vessels with different areas.
In vitro studies with murine embryonic fibroblasts
Murine embryonic fibroblasts were isolated from E13.5 embryos as previously reported [
32]. Fibroblasts were cultured in DMEM with 10% fetal calf serum (FCS) and used at passages 2-5. To assess the response to exogenous growth factors, 1.5 × 10
5 fibroblasts were seeded in 6-well plates. After 18 hours of serum-starvation, cells were stimulated with 20 ng/mL recombinant rat PDGF-BB (R&D Systems, Abingdon, UK) for 30 minutes or 3 ng/mL recombinant human TGF-β1 (R&D Systems) for 72 hours. To assess the role of direct contact of endothelial cells, 1.5 × 10
5 fibroblasts were mixed with an equal number of the human endothelial cell line, EaHy926 [
34] and cultured in DMEM containing 10% FCS for 24 hours. Cells were finally processed for reverse transcription PCR.
Murine embryonic fibroblast migration studies
2.5 × 10
4 fibroblasts were seeded in the upper chamber of 8 μm-pore size transwells (Costar, Elscolab, Kruibeke, Belgium). DMEM/1% FBS with or without 20 ng/mL PDGF-BB was added to the bottom well to stimulate migration. After 18 hours of incubation, cells were fixed in 1% paraformaldehyde and stained with 0.5% crystal violet solution. The number of migrated cells was quantified by counting five high-power magnification fields per transwell [
14]. All studies were repeated with 3 independent clones of fibroblasts from each genotype, yielding comparable results. Thus, representative results were reported.
Tumor cell survival assays
5 × 103 T241 fibrosarcoma cells or PancO2 cells were seeded in 96-well plates and serum-starved for 6 hours before adding conditioned medium of murine embryonic fibroblasts grown in DMEM containing 10% FCS or stimulated with serum-free DMEM containing 3 ng/mL recombinant human TGF-β1 for 24 hours. The number of viable cells was determined 24, 48 and 72 hours after using the CellTiter 96 Aqueous One Solution (Promega, Leiden, The Netherlands) according to manufacturer's instructions.
Quantitative reverse transcription (qRT)-PCR
RNA was extracted from cells using Qiagen RNeasy kit. 0.5-1 μg of total RNA was used for reverse transcription with QuantiTect Reverse Transcription kit (Qiagen, KJ Venlo, the Netherlands). qRT-PCR was performed using TaqMan Fast Universal PCR Master Mix (Applied Biosystems, Halle, Belgium) and commercially available or home-made primers and probes for the genes of interest (Table
1). Analyses were performed using ABI7500 Fast Real-Time PCR System (Applied Biosystems, Halle, Belgium). All experiments were performed a minimum of 3 times, each in at least triplicate.
Table 1
List of primers and probes used for qRT-PCR
Hes | TCAGCGAGTGCATGAACGA | CCTCGGTGTTAACGCCCTC | TGACCCGCTTCCTGTCCACGTG |
Gene
|
Sequence ID (Commercially available primers)
|
β-actin | Mm00607939_s1 |
CD248 | Mm0054785_s1 |
Hey1 | Mm00468865_m1 |
Jagged1 | Mm00496902_m1 |
hJagged1 | Hs01070028_g1 |
Notch3 | Mm00435270_m1 |
SM22α | Mm00441660_m1 |
Statistics
Data represent mean ± standard error of the mean (SEM) of experiments performed at least in duplicate. Statistical significance was calculated by t-test or two-way ANOVA (Prism 5.0), with p < 0.05 considered statistically significant.
Animal care
All experimental animal procedures were approved by the Institutional Animal Care and Research Advisory Committee of the K. U. Leuven.
Discussion
CD248 was originally believed to be a tumor endothelial cell marker, and thus referred to as endosialin or TEM1 [
10‐
12]. CD248 is now recognized to be primarily expressed on the surface of mesenchymal stem cells, activated stromal fibroblasts and pericytes [
16], cells that may contribute to fibrovascular network expansion and tumor progression [
5]. While several investigators have shown that CD248 plays an important role in tumor growth and stromal expansion [
13‐
15,
21,
25] with expression levels that have been correlated with tumor progression [
23,
24], the mechanisms by which CD248 functions and the key structural domains involved, have remained a mystery. In our studies, we established that the cytoplasmic domain of CD248 is a key regulator of tumor growth and that mice lacking this domain are resistant to growth of T241 fibrosarcoma tumors and heterotopic LLC tumors.
Activated fibroblasts dynamically interact with constituents of the stromal compartment and participate in tumor progression via several mechanisms, including for example, remodeling the ECM, promoting the recruitment of inflammatory cells, enhancing nutritional support of the stromal microenvironment, and by secreting an array of pro-lymph/angiogenic and autocrine and paracrine acting cellular growth factors [
6,
56‐
58]. The cytoplasmic domain of CD248 is a critical enabler for stromal fibroblast activation, endowing the fibroblast with several tumor-promoting properties. Tumor fibroblasts can achieve and maintain an activated state in the tumor microenvironment by acquiring epigenetic and/or genetic changes that mitigate the function of tumor suppressor genes, such as p53 and PTEN in breast cancer [
41,
42]. In this manner, activated fibroblasts can exhibit a phenotype that favors proliferation and an enhanced response to pro-survival and migratory cues released by neighboring fibroblasts and other tumor and non-tumor stromal cells. In this regard, we evaluated the relationship between the cytoplasmic domain of CD248 and SM22α, a tumor suppressor gene that when dysregulated, is implicated in the progression and metastasis of cancers of the colon, breast and prostate [
45,
59]. SM22α was considered a likely candidate for CD248-dependent expression because it is known to be upregulated by TGF-β, and because we had previously shown that CD248
CyD/CyD fibroblasts release less TGF-β and are resistant to TGF-β-mediated angiogenic and pro-inflammatory activities [
32]. CD248
CyD/CyD fibroblasts expressed significantly higher transcript levels of SM22α, further upregulated by TGF-β to levels exceeding those seen with CD248
WT/WT fibroblasts. Although we have not directly determined whether SM22α dampens activation of fibroblasts, it is reasonable to consider that elevated levels might contribute to the smaller tumors in the CD248
CyD/CyD mice by maintaining the fibroblasts in a more quiescent state and by indirectly reducing TGF-β-induced MMP-9 release, overall preventing microenvironmental changes that would facilitate cell migration and cancer progression.
Since SM22α is co-ordinately regulated by Notch and TGF-β [
60], we hypothesized that the cytoplasmic domain of CD248 would similarly regulate expression of Hes and Hey1, downstream effectors of Notch that also exhibit context-specific tumor suppressor properties [
53,
54]. Indeed, Hes and Hey1 transcript levels in CD248
CyD/CyD fibroblasts were significantly higher than in CD248
WT/WT fibroblasts, in line with the role of CD248 in promoting tumor growth. Notably from these studies, we uncovered a novel mechanism by which fibroblast CD248 is itself regulated, i.e., it is markedly suppressed by direct contact with endothelial cells. Although the molecular mechanisms remain to be clarified, identification of strategies to downregulate CD248 will be important for the design of therapies to reduce both tumor growth and inflammation.
In addition to the role of the cytoplasmic domain of CD248 in imparting fibroblast sensitivity to the effects of TGF-βα, our studies also show that this domain of CD248 is crucial for optimal migratory response of activated fibroblasts to PDGF-BB. Our observations are in line with recent reports showing that CD248-deficient fibroblasts or pericytes also have defects in migration and proliferation that may [
14,
61] or may not [
49] depend on PDGF-BB. The apparent discordant findings in the literature likely reflect differences in experimental approaches. Studies by Tomkowicz et al. suggest that CD248 may recruit Src/PI-3 Kinase and cFos pathways to enhance PDGF-BB-induced signals emanating from the PDGF-receptor [
61]. Further study will be necessary to elucidate the intracellular pathways responsible for the reduced PDGF-BB induced CD248
CyD/CyD fibroblast migration.
In exploring the mechanisms underlying increased resistance to arthritis induction in CD248
CyD/CyD mice [
32], we recently showed that CD248
CyD/CyD fibroblasts are less adherent to monocytes and express reduced levels of VEGF, PlGF and VEGFR-1. These CD248-dependent alterations serve to reduce leukocyte infiltration, synovial fibroblast migration, proliferation and inflammation in arthritis [
62,
63]. The analogy between cellular proliferation and metastasis formation in cancer and synovial hyperplasia and invasion in rheumatoid arthritis is well-recognized [
64]. For that reason, and because tumor associated macrophages also contribute to cancer progression [
4], we examined tumors from CD248
WT/WT and CD248
CyD/CyD mice for leukocyte infiltration (data not shown). There were, in fact, fewer leukocytes in the tumors from CD248
CyD/CyD mice (1000 + 28 cells/mm
2 in CD248
CyD/CyD versus 1200 + 86 cells/mm
2 in CD248
WT/WT, n = 7, p = 0.0578), although the differences were not statistically significant. Nonetheless, the findings suggest that the cytoplasmic domain of CD248 might also participate in the regulated release by activated fibroblasts of pro-inflammatory factors such as IL-1β, monocyte chemotactic protein (MCP)-1 and IL-8 [
7], thereby further tipping the balance of the stromal microenvironment toward one that favours tumor initiation and progression.
A striking observation made by investigators who previously evaluated CD248-deficient mice in tumor models was the smaller tumor size associated with a seemingly paradoxical increase in microvessel density [
21,
25]. In spite of their finding that pericyte coverage was not altered, Nanda et al. postulated that CD248-deficient blood vessels may fail to mature properly, hence favoring the sprouting of small-caliber vessels [
25]. Surprisingly, we did not observe any differences in vessel density or size distribution, nor in pericyte coverage in T241 fibrosarcoma tumors from the CD248
CyD/CyD mice or CD248
KO/KO mice. Several factors could explain why the angiogenic responses were different in our studies
versus others, particularly for the CD248-deficient mice. First and foremost, the xenograft tumor cell lines that we examined were different from those of Nanda et al. [
25] and Carson-Walter et al. [
21]. Second, one of the groups [
21] used immunodeficient mice for their xenograft studies. Finally, the mice were exposed to different environmental factors and were generated from distinct genetic backgrounds. In spite of these differences, the remarkable finding in the CD248
CyD/CyD mice and the CD248
KO/KO mice (irrespective of the source) was that tumor angiogenesis was not reduced, in spite of the tumors being smaller than in their wild-type counterparts. This paradoxical lack of reduced tumor angiogenesis in the setting of smaller tumors is not without precedent. For instance, Gas6-deficient mice grew smaller tumors as compared to their wild type counterparts, while microvessel density, vessel lumen area and pericyte coverage did not change [
37]. In that case, it was determined that tumors cells induce infiltrating leukocytes to produce the mitogen Gas6. Since CD248 is not expressed by tumor cells, but rather by activated fibroblasts, it is interesting to note that conditioned media from CD248
CyD/CyD fibroblasts dampened the proliferative potential of T241 fibrosarcoma cells. Thus, it is likely that the cytoplasmic domain of CD248 facilitates fibroblast release of soluble factors that promote tumor growth. Several candidates could be considered, including for example, IGF-1, HGF and FGF, all of which favor tumor cell survival and proliferation [
2‐
4,
65].
Interestingly, we and others have shown that not all tumors depend on CD248 for growth. Moreover, the same tumor does not necessarily progress in the same CD248-dependent manner in different anatomical sites [
25]. Several groups have demonstrated that tumor microenvironment and location functionally influence tumor growth and metastasis. And tumor stromal heterogeneity may be associated with multiple factors including differences in hypoxia-induced vascular response [
66], vessel maturation [
67] and activation of tumor associated fibroblasts [
68]. Such heterogeneity has considerable clinical relevance and may explain, at least to some extent, unresponsiveness of some tumors to anti-VEGF therapy [
69]. The variable contribution of CD248 to tumor growth highlights the importance of understanding and establishing multiple targets for the design of effective therapies for given tumors.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
MM participated in the design of the project, coordinated and performed most of the experiments, and helped in the preparation of the manuscript. AV, TJ and MM provided technical support in managing mice, preparing cells, performing assays and analysing data. RJL and JT helped in the design of the studies and preparation of the manuscript. EMC conceived of the study and its design, supervised all aspects of the work, and prepared the manuscript. All authors read and approved the final manuscript.