Tumor-secreted anterior gradient-2 binds to VEGF and FGF2 and enhances their activities by promoting their homodimerization

Guo, H; Zhu, Q; Yu, X; Merugu, S B; Mangukiya, H B; Smith, N; Li, Z; Zhang, B; Negi, H; Rong, R; Cheng, K; Wu, Z; Li, D

doi:10.1038/onc.2017.132

Published: 08 May 2017

Tumor-secreted anterior gradient-2 binds to VEGF and FGF2 and enhances their activities by promoting their homodimerization

H Guo^1,2^na1,
Q Zhu^1,2^na1,
X Yu³,
S B Merugu¹,
H B Mangukiya¹,
N Smith⁴,
Z Li⁴,
B Zhang^1,2,
H Negi¹,
R Rong¹,
K Cheng⁵,
Z Wu¹ &
D Li^1,2

Oncogene volume 36, pages 5098–5109 (2017)Cite this article

1541 Accesses
31 Citations
3 Altmetric
Metrics details

Subjects

Abstract

The importance of the tumor microenvironment in targeted anticancer therapies has been well recognized. Various protein factors participate in the cross-talk between tumor cells and non-malignant cells. Anterior gradient-2 (AGR2) is overexpressed in diverse human adenocarcinomas and it exists in both intracellular and extracellular spaces. Although intracellular AGR2 has been intensively investigated, the function of secreted AGR2, especially its exact mechanism of action is still poorly understood. Here we report that the secreted AGR2 promotes the angiogenesis and the invasion of vascular endothelial cells and fibroblasts by enhancing the activities of vascular endothelial growth factor (VEGF) and fibroblast growth factor 2 (FGF2). Further study indicated that AGR2 directly binds to these extracellular signaling molecules, and enhances their homodimerization. The extracellular AGR2 activity can be blocked to reduce angiogenesis and inhibit tumor growth in vitro and in vivo by a monoclonal antibody targeting the AGR2 self-dimerization region, and combined treatment with bevacizumab produced maximum inhibition effect. In conclusion, our investigation reveals a mechanism that directly links the secreted AGR2 with extracellular signaling networks, and we propose that the secreted AGR2 is a blockable molecular target, which acts as a chaperon-like enhancer to VEGF and FGF2.

Download PDF

Introduction

The malignant characteristics of tumor cells cannot develop without the cross-talk between tumor cells and their local environment, which contains various kinds of supporting cells and soluble protein factors.¹ Nearly half of the primary tumor mass is composed of non-malignant cells. These cells can be categorized into three general classes: angiogenic vascular endothelial cells, cancer-associated fibroblasts and infiltrating immune cells.^{2, 3} These cells are organized around tumor cells to support tumor vascularization, to maintain tumor structure or to form an immunosuppressive environment. The formation of this complex microenvironment is dependent on intercellular communication through extracellular signaling molecules. Some of these signaling molecules have already been used as drug targets such as vascular endothelial growth factor (VEGF) and fibroblast growth factor 2 (FGF2).^{4, 5, 6, 7, 8} However, their clinical effects are often limited by the compensation effects of other extracellular signaling molecules, demonstrating the importance of cross-talk among signaling molecules in tumor local environment.^{9, 10, 11}

Human anterior gradient-2 (AGR2) is a homolog of Xenopus XAG2, a secreted protein that has a critical role in cement gland differentiation and limb regeneration.^{12, 13} It contains both the secretory signal-peptide and the endoplasmic reticulum retention sequence KTEL, and is overexpressed in diverse human cancers.^{14, 15} Tumor-related functions of intracellular AGR2 have been intensively investigated.^{15, 16, 17} We and others have also reported several extracellular AGR2 functions, including the promotion of angiogenesis and fibroblasts coordinated tumor cells invasion.^{18, 19, 20} However, the exact mechanism of how secreted AGR2 performs its function is not understood, especially how AGR2 interacts with other extracellular signaling molecules, such as VEGF and FGF2, which are the major players in tumor angiogenesis.^{21, 22, 23, 24}

Here, we report a mechanism for the tumor-promoting function of extracellular AGR2. We show that extracellular AGR2 directly binds to VEGF and FGF2 and enhances their activities. This enhancement is dependent on both the AGR2 self-dimerization region and the signaling pathways of VEGF and FGF2. We also show that extracellular AGR2 activity can be blocked, at least partially, by a monoclonal antibody targeting the AGR2 self-dimerization region in vitro and in vivo. Our results suggest that the secreted AGR2 is a potential anti-tumor target, which influences the activities of multiple extracellular signaling molecules.

Results

AGR2 is secreted from tumor cells and accumulated in the tumor interstitial fluid (TIF) in the ovarian cancer xenograft tumors

AGR2 overexpression in ovarian cancers has been reported in several clinical studies.^{25, 26, 27} To target AGR2 against ovarian cancer, it is critical to investigate the AGR2 expression levels in different ovarian cancer cell lines and whether AGR2 is secreted into the tumor microenvironment. We, therefore, measured the AGR2 expression levels in three ovarian cancer cell lines and their corresponding xenograft tumors. Our results showed that these cell lines had three different expression patterns (Figures 1a–c). There was no AGR2 expression in both normal cultured (DMEM supplied with 10% serum) ES-2 cells and ES-2 xenograft tumors (Figure 1a), whereas both cultured A2780 cells and A2780 xenograft tumors expressed significant amounts of AGR2 (Figure 1b). The SKOV3 cells had little AGR2 expression under normal culture condition but they produced large amounts of AGR2 during the generation of xenograft tumors (Figure 1c). However, the expression of AGR2 was reduced to a minimal level when tumor cells were isolated from the solid tumors and re-cultured in vitro (Figure 1c, re-cultured tumor cells). These tumor-isolated SKOV3 cells could be induced to express AGR2 again after re-inoculation on nude mice to form a passage 2 tumor (Figure 1c, SKOV3 passage 2 tumor). These results suggest that AGR2 expression in SKOV3 cells is induced during tumor formation at protein level and this induction is environmentally dependent.

It has been reported that AGR2 expression is regulated by hypoxia and serum limitation in breast cancer cells,^{28, 29} but this is still unclear in ovarian cancer cells. Thus, we studied the regulation of AGR2 expression by hypoxia and serum deprivation in three ovarian cancer cell lines (Figures 1d and e). Western blot analysis showed that AGR2 expression was not induced by hypoxia in all three ovarian cell lines (Figure 1d and Supplementary Figure S1). However, serum deprivation significantly increased the AGR2 expression in A2780 and SKOV3 cells (Figure 1e), both of which expressed AGR2 in xenograft tumors. These results indicate that AGR2 induction in ovarian xenograft tumors may be related to the serum limitation but independent of hypoxia.

To further explore the influence of AGR2 induction in SKOV3 tumors, two stable AGR2 knockdown SKOV3 clones were created, in which AGR2 could not be induced by serum deprivation (Figure 1f). Xenograft result showed that AGR2 knockdown reduced tumor size by an average of 37.2% compared with the control SKOV3 tumors (Figure 1g). Real-time PCR analysis showed that the induced AGR2 was produced by human tumor cells but not by the mouse tumor stromal cells, and the AGR2 mRNA expression level in AGR2 knockdown tumors (shAGR2) was 3.5 times lower than that of the control groups (shCtrl; Figure 1h). The concentration of secreted AGR2 in the TIF was also measured using ELISA (Figure 1i). Data showed that the secreted AGR2 accumulation in the TIF of control SKOV3 tumor is 4.6 fold higher than that of AGR2 knockdown tumors (2.672±0.488 μg/ml for shCtrl tumors and 0.581±0.216 μg/ml for shAGR2 tumors). These results suggest that the ability to induce AGR2 expression in tumor cells creates an extracellular microenvironment with high local AGR2 level.

AGR2 enhances VEGF and FGF2 activities through various degrees of cross-talk with VEGF and FGF2 signals

In consideration of the angiogenic characteristics of SKOV3 tumor,^{30, 31} investigating how AGR2 functions in the tumor angiogenesis signaling networks may facilitate better understanding of the role of secreted AGR2 in extracellular signal networks. So we used transwell assays to measure the pro-migration effect of externally added AGR2 on human umbilical vein endothelial cells (HUVECs) and to assess the cross-talk between AGR2 and five major angiogenic factors (Figure 2a and Supplementary Figure S4). Results showed that AGR2 alone had no significant effect on HUVECs migration. There were also no additional migrations when cells were treated with AGR2 in combination with angiopoietin 1, platelet-derived growth factor and placental growth factor, relative to the cells treated with these growth factors alone. However, AGR2 combined with VEGF and FGF2 significantly enhanced the cell migration relative to VEGF or FGF2 treated alone, and this effect was observed to be AGR2 dose-dependent.

An EdU incorporation assay was also performed to assess the AGR2 function in HUVECs proliferation (Figure 2b). AGR2 alone did not promote HUVECs proliferation even at the 500 ng/ml concentration. However, the proliferation rates of combined groups with 500 ng/ml AGR2 were 1.41 and 1.39 times higher than groups treated with FGF2 alone and VEGF alone, respectively. This enhancement effect was observed to be AGR2 dose-dependent.

As migration of stroma cells toward cancer mass is critical to the development of the tumor microenvironment and tumor growth, chemoattractant of HUVECs, 3T3 fibroblasts, THP-1 macrophages and primary mouse peritoneal cavity immune cells were investigated in an agarose spot invasion assay with central spots, which formed sustainable specific protein release sources (Figures 2c–e and Supplementary Figure S5). Results showed that AGR2 spot only induced slight invasion of HUVECs into the agarose compared with FGF2 and VEGF spots. However, AGR2 combined with VEGF or FGF2 induced a much greater invasion than that of VEGF or FGF2 alone (Figure 2c). In 3T3 cells, the invasion induced by AGR2 alone was much more obvious than that in HUVECs, and the addition of FGF2 significantly increased the rate of invasion, but VEGF did not (Figure 2d). It has been known that TGFβ1 is one of the important growth factors for fibroblasts, so we investigated whether AGR2 can prime the effect of TGFβ1 (Figures 2e). Agarose invasion result showed that there was no extra invasion when AGR2 was combined with TGFβ1 (Figure 2e). We also found that AGR2 treatment could not improve the αSMA expression (Figure 2f), indicating that the fibroblasts invasion induced by AGR2 is independent of fibroblast activation. In addition, both AGR2 alone and AGR2 in combination with VEGF or bFGF were not able to induce THP-1-differentiated macrophages and primary peritoneal immune cells to migrate into the agarose (Supplementary Figure S5). These data indicate that AGR2 requires existing VEGF or FGF2 to enhance vascular endothelial cell invasion, but AGR2 also acts as both an enhancer of FGF2 and an independent promoter for fibroblasts invasion.

The improvement of VEGF and FGF2 activities by AGR2 can be blocked by a specific antibody

To better understand the effect of AGR2 in angiogenesis, an in-house generated anti-AGR2 mouse monoclonal antibody, 18A4 was employed. Western blot and immunoprecipitation results confirmed that 18A4 could specifically bind to the secreted AGR2 (Figure 3a). Because reorganization is one of the key steps during new vessel formation, a tube formation assay using MCF7-conditioned medium was performed. Result showed that the tube formation which induced by MCF7-conditioned medium was inhibited by 18A4 in a dose-dependent manner but not by control IgG. This indicates that AGR2 in MCF7-conditioned medium is necessary for HUVECs to form capillary tubes properly and 18A4 can effectively block this process (Figure 3b).

To assess the cross-talk effect of AGR2 within paracrine signaling networks in vivo, a gel plug assay and hemoglobin quantification were used to test new blood vessel formation in nude mice (Figures 3c and d). Results showed that plugs with AGR2 alone had little hemoglobin content whereas both FGF2 and VEGF plugs contained much greater amounts of hemoglobin than controls. When AGR2 was combined with FGF2 or VEGF, the hemoglobin contents were further increased by 48.8% and 52.3%, respectively. These AGR2 enhancing effects could be largely abolished by the addition of 18A4 (71.2 and 76.3%). Furthermore, CD31 staining (vascular endothelial cell marker) showed that combinations of AGR2 with FGF2/VEGF implants had significantly more CD31-positive cells than implants containing FGF2/VEGF alone, whereas AGR2 alone had little positive cells (Figure 3e).

The improvement of VEGF and FGF2 activities by AGR2 requires VEGF-VEGFR and FGF2-FGFR signal transductions

In order to explore the influence of AGR2 on VEGF and FGF2 signal transductions, we detected several VEGF and FGF2 downstream signaling pathways on HUVECs (Figure 4a). Results showed that AGR2 treatment alone had little effects on the level of pPLCγ (Y783), pAKT (S473), pMEK½ (S217/221) and pERK½ (T202/Y204) but significantly enhanced VEGF and FGF induced phosphorylation. Because ERK½ is a common downstream target of both FGF2 and VEGF pathways, pERK½ was selected as a reporter to elucidate the further mechanism by which secreted AGR2 affects endothelial cells (Figures 4b and g). Our results showed that there was little induction of ERK½ phosphorylation when HUVECs were treated with increasing amounts of AGR2 alone, but the addition of AGR2 enhanced the FGF2-induced ERK phosphorylation in a dose-dependent manner (Figure 4b). In addition, we also investigated the changes in VEGFR2 phosphorylation (Y1175 and Y951) with the combinatorial treatment of VEGF and AGR2, result showed increased VEGFR2 phosphorylation in combinational treatment (Figure 4c).

Furthermore, the enhancement of FGF2 activity by AGR2 was found to be largely reversed by both the FGFR1 inhibitor PD173074 and the AGR2 blocking antibody 18A4 (Figures 4d and e). The enhancement of VEGF activity by AGR2 was also diminished by the addition of VEGFR inhibitor axitinib, as well as VEGF blocking antibody bevacizumab and 18A4 (Figures 4f and g).

These results indicate that AGR2-induced improvement of VEGF and FGF2 activities requires the interaction of AGR2 with the extracellular components of VEGF-VEGFR2 and FGF2-FGFR1 pathways.

AGR2 binds to VEGF and FGF2, but not to their receptors

To determine how AGR2 influences the VEGF and FGF2 signals, the interactions of exogenous AGR2 with the extracellular components of VEGF and FGF2 pathways were investigated (Figure 5). Results showed that purified FGF2 was specifically co-immunoprecipitated using anti-AGR2 antibody only when purified AGR2 was added, and the purified FGF2 could mediate the co-immunoprecipitation of purified AGR2 by anti-FGF2 antibody (Figure 5a). Purified AGR2 also mediated the co-immunoprecipitation of VEGF by anti-AGR2 antibody and anti-VEGF antibody was able to co-precipitate AGR2 when VEGF was added (Figure 5b). These co-immunoprecipitation results indicate that the interactions are direct without the requirement of any other mediators. To determine whether the AGR2-FGF2 and AGR2-VEGF interactions can be detected in physiological conditions, we used a mixture of cell conditioned medium containing naturally secreted AGR2, VEGF and FGF2 by T47D cells (secreting AGR2) and by SKOV3 cells (secreting VEGF and FGF2; Figures 5c and d). Co-immunoprecipitation of both AGR2-FGF2 (Figure 5c) and AGR2-VEGF (Figure 5d) can be detected. In addition, the interaction between AGR2 and FGF2 was also confirmed in the SKOV3 TIF (Supplementary Figure S3a), whereas the interaction between AGR2 and VEGF was not detected owing to the low concentration of VEGF in the SKOV3 TIF (Supplementary Figure S3b). However, the interactions of AGR2-VEGFR2 and AGR2-FGFR1 were not detected by co-immunoprecipitation. (Supplementary Figure S6).

The AGR2-FGF2 and AGR2-VEGF interaction properties were further measured using a biolayer interferometry technology (Figures 5e and f). The association and dissociation rates between AGR2 and VEGF, AGR2 and FGF2 were measured using purified proteins. The results showed that although AGR2 was indeed able to bind to both VEGF and FGF2, the binding affinity for FGF2 (KD=117±9 nM) was much stronger compared with the predominant isoform of VEGF–VEGF165³² (KD=315±55 nM), which is diffusible, although a significant fraction remains bound to the cell surface and extracellular matrix. The binding affinity of AGR2 for VEGF165 is just a little lower than the affinity of TGF-β 3 for TGF-β receptor II (KD=290 nM)³³ (Supplementary Table S1). We also explored the interactions between AGR2 and other two common VEGF isoforms, VEGF121, could not is a freely diffusible protein, and VEGF189, which is almost completely sequestered in the extracellular matrix (Figure 5f). Results showed that the binding between AGR2 and VEGF121 could not be detected, but the binding affinity between AGR2 and VEGF189 (KD=93.2±3 nM) is about threefold higher than the affinity between AGR2 and VEGF165 (KD=315±55 nM). We also investigated the ability of AGR2 binding to IGF-1, EGF and TGFβ1. However, the binding of AGR2 to all of these growth factors could not be detected (Figures 5e and f).

AGR2 self-dimerization region is required for increasing the homodimerization of VEGF and FGF2

FGF2 molecules are naturally monomeric and do not self-dimerize. However, FGFR can only be activated by the FGF2 dimers but not by the monomeric molecules.³⁴ So we investigated whether AGR2 exerted its enhancement function by influencing the homodimerization of FGF2 and VEGF (Figures 6c and d). Results showed that the homodimer levels of both VEGF and FGF2 were increased by the presence of AGR2 and this increase was negated by the presence of 18A4. As the AGR2 self-dimerization region (60-EALYK-64)³⁵ is one of the blocking sites of 18A4 (Figures 6a and b), a self-dimerization region mutant △AGR2 (Figure 6a) was employed to determine whether AGR2 self-dimerization is needed for increasing the homodimerization of VEGF and FGF2. Results showed that △AGR2 indeed had less homodimerization promotion of VEGF and FGF2 (Figures 6c and d). It was also less effective both in the ERK phosphorylation (Figure 6e) and in the agarose spot assays (Figures 6f and g).

Targeting AGR2 reduces the infiltration of vascular endothelial cells and fibroblasts in the AGR2-positive ovarian tumors

To investigate the effect of blocking secreted AGR2 signal on ovarian tumors, 18A4 antibody was employed in xenograft tumors from three tumor cell lines (Figure 7). Results showed that 18A4 treatment had no effect on ES-2 tumors (Figures 7a and b), which does not express AGR2 (refer to Figure 1a), but 18A4 reduced average 35.4% of the size of A2780 xenograft tumors (Figures 7e and f), which express AGR2 constitutively (refer to Figure 1b). In addition, both 18A4 and bevacizumab reduced the SKOV3 tumor size effectively (63.8% and 72.3%, respectively), and the tumor volumes were further repressed when 18A4 and bevacizumab treatment were combined (Figures 7i and j). Cell-type specific immuno-staining showed that 18A4 treatment reduced the tumor-associated vascular epithelial cells (CD31) and activated fibroblasts (αSMA) in the AGR2-positive tumors, especially when combined with bevacizumab (Figures 7g, h, k and i). Considering AGR2 had no effect on the αSMA expression level (Figure 2f), this result suggested that the blocking of AGR2 reduced the infiltration of activated fibroblasts. All these results indicate that AGR2 is a promising tumor target for the treatment of AGR2-positive ovarian cancers.

Discussion

In this research, we focused on the relationship between secreted AGR2 and the extracellular parts of VEGF and FGF2 signaling pathways. We found that the AGR2 paracrine function partially requires the presence of VEGF or FGF2 signals. Specifically, AGR2 can directly bind to VEGF and FGF2 and enhance their activities, which leads to the angiogenesis and tumor growth (Figure 7m).

Unlike the VEGF and FGF2, which work at the picomolar levels, the effective AGR2 concentration is at the nanomolar range (100–500 ng/ml) in our experiments. This may be related to its moderate binding affinity with VEGF (315 nM) and FGF2 (117 nM). It also indicates that the AGR2 function is effective only in the areas immediately around the AGR2-secreting cells, such as the TIF, where the tumor-secreted proteins have several orders of magnitude higher concentration compared with the plasma.^{36, 37}

In our investigation, we reported that AGR2 could bind to FGF2 and enhance its activity. This is similar to the pattern of interaction between fibroblast growth factor-binding protein and FGF.^{38, 39} The difference is that fibroblast growth factor-binding protein only binds to FGF family members, whereas AGR2 binds to not only FGF2 but also VEGF. This indicates that AGR2 may act like a chaperon that is capable of cross-talking with multiple existing growth factors and influencing kinds of cell types.

It is interesting to note that the signaling of Xenopus AGR2 is also dependent on an intact FGF signal transduction pathway.¹² Furthermore, the homology analysis of AGR2 shows that the self-dimerization region is conserved among vertebrates. Both findings support that the mechanism of AGR2 paracrine signaling we reported here may be evolutionarily conserved.

VEGF have several isoforms. Among them, VEGF121, VEGF165 and VEGF189 are the major isoforms existing in the physiological conditions.⁴⁰ Our results showed that the binding of these VEGF isoforms to AGR2 has a similar pattern with the binding of them to the extracellular matrix. The relationship between these two events may worth further investigation.

Several researches have reported that intracellular AGR2 affects tumor cells by promoting their survival, proliferation, migration, transformation and drug resistance.¹⁵ Our investigation reveals that secreted AGR2 also can influence the tumor-surrounding cells through cross-talk with existent growth factors. Secreted AGR2 possess several characteristics suitable for therapeutic antibody targeting, including extracellular localization for antibody access, induced expression during tumor formation and its multifunctional influence on both tumor cells and tumor supporting cells. Our antibody blocking data showed significant xenograft tumor inhibition activity especially when combined with bevacizumab, suggesting that AGR2 is a potential target for antibody-based anti-tumor therapy.

Materials and methods

Chemical inhibitors, growth factors and antibodies

IgG control antibody was purchased from Pierce (Thermo Scientific, Waltham, MA, USA). VEGF antibody (ab46154), HIF1α antibody (ab51608) and FGF2 antibody (ab208687) were from Abcam Company (Abcam, Cambridge, MA, USA). β-actin antibody (sc-47778) and mouse anti-AGR2 antibody (sc-101211) were from Santa Cruz Biotechnology (Santa Cruz, CA, USA). FGFR1 antibody (9740S), VEGFR2 antibody (2479S), pAKT (S473) antibody (4060S), p-p44/42 MAPK (T202/Y204) antibody, pMEK1/2 (S217/221) antibody (9154P), p-PLC gamma (Y783) antibody (2821T), p-VEGFR2 (Y1175) and p-VEGFR2 (Y951) antibodies (2478S and 4991T) were from Cell Signaling Technology (Danvers, MA, USA). Rabbit anti-AGR2 polyclonal antibody was purchased from Abgent Biotech (Wuxi AppTec, Suzhou, China). Bevacizumab (Avastin) was obtained from Shanghai Roche Pharmaceutical (Roche, Shanghai, China). Mouse anti-AGR2 monoclonal antibody 18A4 was generated in-house and has been described previously.⁴¹ VEGF189 was purchased from ReliaTech GmbH (ReliaTech, Wolfenbüttel, Germany), all other growth factors were obtained from Peprotech Company (Peprotech, Rocky Hill, NJ, USA). Recombinant VEGF protein used in this paper is VEGF165 unless specifically mentioned. His-AGR2 was generated in-house and the purity has been detected by sodium dodecyl sulfate polyacrylamide gel electrophoresis (Supplementary Figure S2). FGFR inhibitor PD173074 and VEGFR inhibitor Axitinib were purchased from Selleck Chemical (Selleck, Houston, TX, USA).

Cell culture

All cells were routinely authenticated by short tandem repeat profiling and tested for mycoplasma contamination. HUVECs were purchased from AllCells Company and maintained in HUVEC complete medium (AllCells, Shanghai, China). Cells were passaged for fewer than six passages. THP-1 macrophages were differentiated via 48 h of 50 ng/ml PMA induction. Mouse peritoneal cavity immune cells were separated according to Ray’s protocol.⁴² A2780, SKOV3, ES-2 and MCF7 were purchased from Procell (Procell, Wuhan, China). All cell lines were cultured in the appropriate culture medium supplemented with 10% fetal bovine serum, 1% antibiotics (Solarbio, Beijing, China), with or without insulins (Solarbio) according to the ATCC guidelines. To create a physical hypoxic environment, cells were seeded in the six-well plates and placed in a MIC-101 modular incubator chamber (Billups-Rothenberg, Del Mar, CA, USA) that was sealed and flushed for 10 min with a gas mixture of 0.5% O₂, 5% CO₂ and 94.5% N₂ at a rate of 20 l/min. The hypoxic chamber was re-flushed after 1 h of incubation then sealed tightly and placed in an incubator (at 37 °C in humidified 5% CO₂/95% air) for 24 h. Using this method, the O₂ concentration of the culture media was measured to be 0.8–1.0% with a Dissolved Oxygen Meter (Model 5509, Lutron, Taiwan). Cells were harvested after treatment for 24 h. To create a chemical induced hypoxia, cells were treated with 200 μM CoCl₂ and cultured for 24 h in a 5% (v/v) CO₂ humidified incubator at 37 °C. For serum deprivation, cells were seeded with complete medium overnight for attachment, then medium was changed with serum-free medium and incubated for another 72 h in the humidified incubator before collected. For fibroblasts activation measurement, 3T3 cells were seeded into the 24 well plate (3 × 10⁴/well) and incubated overnight for attachment, then cells were starved for 6 h in DMEM with 5% serum and treated with AGR2 alone or combined with TGFβ1 (5 ng/ml) for another 12 h before RNA isolation.

Generation of stable AGR2 knockdown SKOV3 cell line

Short hairpin RNAs targeting AGR2 were cloned into a UCPK lentiviral vector. Lentiviruses were produced by transfecting 293T. Supernatants containing lentiviruses were collected 72 h after transfection, mixed with polybrene (6 μg/ml) and used to infect SKOV3 cells. Fresh medium containing puromycin (2 μg/ml) was added after 24 h, and red fluorescent cells were selected. Stable knockdown cells were tested via serum deprivation for 3 days. The shAGR2-1 sequence was 5′-GTCCTCCTCAATCTGGTTTAT-3′, and the shAGR2-2 sequence is 5′-GATATTCAAATCGTCTCTATG-3′.

RNA isolation and real-time PCR

RNA isolation and real-time PCR were performed as previously described.⁴³ The 5′ primer for human AGR2 was 5′-ATCCAGAAATTGGCAGAGCAGTTTGTC-3′ and the 3′ primer was 5′-CTAACTGTCAGAGATGGGTCAACAAACATAATC-3′. The 5′ primer for mouse AGR2 was 5′-TCTCCAGAGGTTGGGGCGATCAG-3′ and the 3′ primer was 5′-AAGGCTTGACTGTGTGGGCATTCG-3′. The 5′ primer for mouse αSMA was 5′-CCTCTGGACGTACAACTGGTATTGTGC-3′ and the 3′ primer was 5′-CAGACGCATGATGGCATGAGGC-3′.

Transwell assay

HUVECs transwell assay was performed as previously described.⁴⁴ Starved HUVECs were seeded in the upper chamber (gelatin coated), whereas recombinant AGR2 (100 ng/ml or 300 ng/ml) or growth factors including angiopoietin 1 (30 ng/ml), platelet-derived growth factor-BB (20 ng/ml), placental growth factor (30 ng/ml), VEGF (2 ng/ml) and FGF2 (1 ng/ml) coupled with or without AGR2 were placed in the lower chamber. After 18 h of incubation, the migrated cells were fixed with cold methanol, stained with 2% (w/v) crystal violet and photographed using an Olympus microscope. Migrated cells were counted by microscopic inspection from two typical fields of three independent wells of each treatment group (n=6).

Proliferation assay

HUVECs proliferation was measured with EdU Assay kit (Invitrogen, Carlsbad, CA, USA) according to the kit instructions. Cells were treated with indicated growth factors for 40 h and EdU was added 24 h before proliferation was measured using Varioskan Flash Microplate Reader (Thermo Scientific).

Tube formation assay

Tube formation was performed as previously described.⁴⁵ Starved HUVECs were resuspended in serum-free MCF7-conditioned medium and seeded onto pre-solid Matrigel in the presence of AGR2 antibody 18A4 or control IgG. After 8 h of incubation, cell networks were photographed and the tube lengths were analyzed using WimTube software on the Wimasis image analysis platform.

Matrigel plug assay and hemoglobin analysis

Growth factor reduced Matrigel (BD Biosciences, San Jose, CA, USA) mixed with AGR2 (1 μg/ml), growth factors (500 ng/ml) and 18A4 (20 μg/ml) as indicated were added into the angioreactors and incubated at 37 °C for 1 h. The angioreactors were implanted on the posterior dorsal flanks of nude mice. After 14 days, they were taken out and photographed. Gel plugs were divided into two equal parts. Half were embedded for immunofluorescence analysis. Another half were used for hemoglobin analysis with Drabkin's reagent (Sigma-Aldrich, St Louis, MO, USA) as previously described.⁴⁶ The concentration of hemoglobin was calculated by standard hemoglobin curve assayed in parallel.

In vitro phosphorylation assay

After 6 h of starvation, the indicated concentrations of inhibitors were added and incubated for 30 min. AGR2, growth factors and antibodies were mixed and incubated for 30 min before they were added. Cells were stimulated with mixtures for the indicated periods (details are in figure legends). The levels of signaling protein phosphorylation were detected using Western blot and data were quantified using Quantity One software.

Co-immunoprecipitation

For purified protein inputs, 400 ng AGR2, 200ng VEGF or 200ng FGF2 were mixed in 200 μl PBS as indicated in Figure 5 and incubated for 4 h at 4 °C. For conditioned medium inputs, 10 ml T47D conditioned medium was mixed with 90 ml SKOV3-conditioned medium and incubated for 4 h at 4 °C. Concentrated input was produced through ultrafiltering 100 ml conditioned medium mixture to 400 μl using Amicon Ultra-15 Centifugal Filter Concentrators (NMWL: 3 KDa; Millipore, Bedford, MA, USA) and used as positive control. A total of 3 μg corresponding antibodies were added into corresponding inputs as indicated in Figure 5 and incubated for 4 h at 4 °C. Then, 20 μl protein G beads (Pierce, Thermo Scientific) were added into each tubes. After 1 h of incubation, beads were washed three times, mixed with loading buffer and boiled for western blot analysis.

Dimerization assay

A total of 200 ng VEGF or FGF2 were mixed with 70 ng AGR2, △AGR2 or AGR2 plus 18A4 (600 ng) and incubated for 30 min. Then, 20 μM BS³ cross-linker (Thermo Scientific) was added and incubated for 20 min to stabilize the dimers. After quenching for 15 min (50 mM Tris), loading buffer was added and samples were boiled for western blot analysis. Data were quantified using Quantity One software.

Optical biosensor biolayer interferometry

Protein–protein interactions were measured using Octet RED96 system (Pall ForteBio, Fremont, CA, USA). His-AGR2 (25 μg/ml) was loaded on to the Ni-NTA biosensors, after equilibration with kinetics buffer, association data were recorded during introduction of AGR2-loaded biosensors into solutions containing 1 μM VEGF121, VEGF165, VEGF189, FGF2, EGF, IGF-1 or TGFβ1, respectively. Then, dissociation data were recorded during washing the AGR2-loaded biosensors in kinetics buffer. Data were analyzed using the Octet Analysis software.

Agarose spot assay

An agarose spot assay was performed as previously described.^{47, 48} In brief, Agarose spots containing VEGF (3 μg/ml), FGF2 (3 μg/ml), TGFβ1 (3 μg/ml), AGR2 (150 μg/ml or mentioned in the figure) or combined were dropped at the center of slides. Cells were seeded around the spots in complete medium 2 h for attachment. Then cells were cultured in serum-free medium for 24 h. Images were taken and the invasion area data were derived from two random fields of three independent wells of each treatment group (n=6) and analyzed using Image J software.

Xenograft studies

Animal studies were approved by the Institutional Animal Care and Use Committee of Shanghai Jiao Tong University. Six-week-old female nude mice were used for all animal models. For tumorigenesis, 2 × 10⁶ shAGR2 or shCtrl SKOV3 cells were injected subcutaneously. For antibody treatment, 2 × 10⁶ ES-2 cells or A2780 cells were injected subcutaneously on the posterior dorsal flanks of 12 mice, which were randomly, equally separated to two groups for treatment of 8 mg/kg IgG and 18A4. For combination treatment, 2 × 10⁶ SKOV3 cells were injected subcutaneously on the posterior dorsal flanks of 24 mice, which were randomly, equally separated to four groups for treatment of 8 mg/kg IgG, 18A4, bevacizumab or 18A4 coupled with bevacizumab. Antibodies were injected (i.p.) twice per week, 2 days after cells injection. Tumors were monitored and tumor volumes were measured by caliper (1/2 × length × width²) every 3 days. Subcutaneous tumors were harvested for immunofluorescence studies and TIF studies. Immunofluorescence analysis was accomplished on frozen sections of xenograft tumors and data were quantified using image J software. For all animal work, at least six mice were randomized and separated into each experimental group, and all animals used were included in the analysis because we did not observe abnormalities on size, weight or apparent disease symptoms that pre-established before performing the experiment. Animal studies were not blinded during data analysis.

TIF separation and AGR2 concentration analysis

TIF was separated as previously described.⁴⁹ The SKOV3 tumors were centrifuged through a 40 μm cell-strainer (STEMCELL, Vancouver, BC, Canada) at 1500 rpm for 10 min. Then, the filtrate was collected and centrifuged at 6000 rpm for 1 min and supernatant TIF was stored at −80 °C. The AGR2 in the TIF was measured with the mouse AGR2 ELISA kit (Mybiosource, San Diego, CA, USA) according to the kit instructions. The concentration of AGR2 was calculated by standard AGR2 curve assayed in parallel.

Statistics

Data were analyzed statistically using GraphPad Prism 5, Image J and Quantity One software. Samples or animals would be excluded if observing any abnormalities in terms of size, weight or apparent disease symptoms, or deviations caused by obvious technical mistakes that were pre-established before performing the experiments. At least three repeats were included for every experiment and quantified data represented at least three independent biological replicates. Detailed sample sizes have been described in figure legends. And the unpaired two-tailed Student's t-test was used to confirming differences between groups (*, P<0.05 and **, P<0.01).

Abbreviations

AGR2:: anterior gradient-2
HUVECs:: human umbilical vein endothelial cells
BEV:: bevacizumab
TIF:: tumor interstitial fluid
αSMA:: alpha smooth muscle actin.

References

Sounni NE, Noel A . Targeting the tumor microenvironment for cancer therapy. Clin Chem 2013; 59: 85–93.
CAS Article Google Scholar
Balkwill FR, Capasso M, Hagemann T . The tumor microenvironment at a glance. J Cell Sci 2012; 125: 5591–5596.
CAS Article Google Scholar
Hanahan D, Coussens L . Accessories to the crime: functions of cells recruited to the tumor microenvironment. Cancer Cell 2012; 21: 309–322.
CAS Article Google Scholar
Cross MJ, Claesson-Welsh L . FGF and VEGF function in angiogenesis: signalling pathways, biological responses and therapeutic inhibition. Trends Pharmacol Sci 2001; 22: 201–207.
CAS Article Google Scholar
Fang H, DeClerck YA . Targeting the tumor microenvironment: from understanding pathways to effective clinical trials. Cancer Res 2013; 73: 4965–4977.
CAS Article Google Scholar
Hagedorn M, Bikfalvi A . Target molecules for anti-angiogenic therapy: from basic research to clinical trials. Crit Rev Oncol Hematol 2000; 34: 89–110.
CAS Article Google Scholar
Joyce JA . Therapeutic targeting of the tumor microenvironment. Cancer Cell 2005; 7: 513–520.
CAS Article Google Scholar
Carmeliet P, Jain RK . Molecular mechanisms and clinical applications of angiogenesis. Nature 2011; 473: 298–307.
CAS Article Google Scholar
Lieu C, Heymach J, Overman M, Tran H, Kopetz S . Beyond VEGF: Inhibition of the fibroblast growth factor pathway and antiangiogenesis. Clin Cancer Res 2011; 17: 6130–6139.
CAS Article Google Scholar
Petrillo M, Scambia G, Ferrandina G . Novel targets for VEGF-independent anti-angiogenic drugs. Expert Opin Investig Drugs 2012; 21: 451–472.
CAS Article Google Scholar
Moserle L, Jiménez-Valerio G, Casanovas O . Antiangiogenic therapies: going beyond their limits. Cancer Discov 2014; 4: 31–41.
CAS Article Google Scholar
Aberger F, Weidinger G, Grunz H, Richter K . Anterior specification of embryonic ectoderm: The role of the Xenopus cement gland-specific gene XAG-2. Mech Dev 1998; 72: 115–130.
CAS Article Google Scholar
Kumar A, Godwin JW, Gates PB, Garza-Garcia AA, Brockes JP . Molecular basis for the nerve dependence of limb regeneration in an adult vertebrate. Science 2007; 318: 772–777.
CAS Article Google Scholar
Brychtova V, Vojtesek B, Hrstka R . Anterior gradient 2: a novel player in tumor cell biology. Cancer Lett 2011; 304: 1–7.
CAS Article Google Scholar
Chevet E, Fessart D, Delom F, Mulot A, Vojtesek B, Hrstka R et al. Emerging roles for the pro-oncogenic anterior gradient-2 in cancer development. Oncogene 2013; 32: 2499–2509.
CAS Article Google Scholar
Brychtova V, Mohtar A, Vojtesek B, Hupp TR . Mechanisms of anterior gradient-2 regulation and function in cancer. Semin Cancer Biol 2015; 33: 16–24.
CAS Article Google Scholar
Higa A, Mulot A, Delom F, Bouchecareilh M, Nguyên DT, Boismenu D et al. Role of pro-oncogenic protein disulfide isomerase (PDI) family member anterior gradient 2 (AGR2) in the control of endoplasmic reticulum homeostasis. J Biol Chem 286: 44855–44868.
CAS Article Google Scholar
Hong XY, Wang J, Li Z . AGR2 expression is regulated by HIF-1 and contributes to growth and angiogenesis of glioblastoma. Cell Biochem Biophys 2013; 67: 1487–1495.
CAS Article Google Scholar
Li D, Wu Z, Zhu Q, Guo H, Gao G, Mashausi DS et al. Agtuzumab, a humanized monoclonal antibody, blocks AGR2 function through conformational epitopes around its catalytic center. Cancer Res 2013; 73: 4320–4320.
Google Scholar
Tsuji T, Satoyoshi R, Aiba N, Kubo T, Yanagihara K, Maeda D et al. Agr2 mediates paracrine effects on stromal fibroblasts that promote invasion by gastric signet-ring carcinoma cells. Cancer Res 2015; 75: 356–366.
CAS Article Google Scholar
Ferrara N . VEGF and the quest for tumour angiogenesis factors. Nat Rev Cancer 2002; 2: 795–803.
CAS Article Google Scholar
Teoh DGK, Secord AA . Antiangiogenic therapies in epithelial ovarian cancer. Cancer Control 2011; 18: 31–43.
Article Google Scholar
Gavalas NG, Liontos M, Trachana SP, Bagratuni T, Arapinis C, Liacos C et al. Angiogenesis-related pathways in the pathogenesis of ovarian cancer. Int J Mol Sci 2013; 14: 15885–15909.
Article Google Scholar
Schmitt J, Matei D . Targeting angiogenesis in ovarian cancer. Cancer Treat Rev 2012; 38: 272–283.
CAS Article Google Scholar
Armes JE, Davies CM, Wallace S, Taheri T, Perrin LC, Autelitano DJ . AGR2 expression in ovarian tumours: a potential biomarker for endometrioid and mucinous differentiation. Pathology 2013; 45: 49–54.
CAS Article Google Scholar
Edgell TA, Barraclough DL, Rajic A, Dhulia J, Lewis KJ, Armes JE et al. Increased plasma concentrations of anterior gradient 2 protein are positively associated with ovarian cancer. Clin Sci 2010; 118: 717–725.
CAS Article Google Scholar
Rice GE, Edgell TA, Autelitano DJ . Evaluation of midkine and anterior gradient 2 in a multimarker panel for the detection of ovarian cancer. J Exp Clin Cancer Res 2010; 29: 62.
Article Google Scholar
Li Z, Zhu Q, Hu L, Chen H, Wu Z, Li D . Anterior gradient 2 is a binding stabilizer of hypoxia inducible factor-1α that enhances CoCl2-induced doxorubicin resistance in breast cancer cells. Cancer Sci 2015; 106: 1041–1049.
CAS Article Google Scholar
Zweitzig DR, Smirnov DA, Connelly MC, Terstappen LWMM, O'Hara SM, Moran E . Physiological stress induces the metastasis marker AGR2 in breast cancer cells. Mol Cell Biochem 2007; 306: 255–260.
CAS Article Google Scholar
Bamias A, Pignata S, Pujade-Lauraine E . Angiogenesis: a promising therapeutic target for ovarian cancer. Crit Rev Oncol Hematol 2012; 84: 314–326.
CAS Article Google Scholar
Nadkarni NJ, Geest K, Neff T, Young B, Bender DP, Ahmed A et al. Microvessel density and p53 mutations in advanced-stage epithelial ovarian cancer. Cancer Lett 2013; 331: 99–104.
CAS Article Google Scholar
Ferrara N, Gerber HP, LeCouter J . The biology of VEGF and its receptors. Nat Med 2003; 9: 669–676.
CAS Article Google Scholar
De Crescenzo G, Hinck CS, Shu Z, Zuniga J, Yang J, Tang Y et al. Three key residues underlie the differential affinity of the TGFbeta isoforms for the TGFbeta type II receptor. J Mol Biol 2006; 355: 47–62.
CAS Article Google Scholar
Plotnikov AN, Schlessinger J, Hubbard SR, Mohammadi M . Structural basis for FGF receptor dimerization and activation. Cell 1999; 98: 641–650.
CAS Article Google Scholar
Gray TA, Murray E, Nowicki MW, Remnant L, Scherl A, Muller P et al. Development of a fluorescent monoclonal antibody-based assay to measure the allosteric effects of synthetic peptides on self-oligomerization of AGR2 protein. Protein Sci 2013; 22: 1266–1278.
CAS Article Google Scholar
Ahn SM, Simpson RJ . Body fluid proteomics: prospects for biomarker discovery. Proteomics Clin Appl 2007; 1: 1004–1015.
CAS Article Google Scholar
Haslene-Hox H, Madani A, Berg KCG, Woie K, Salvesen HB, Wiig H et al. Quantification of the concentration gradient of biomarkers between ovarian carcinoma interstitial fluid and blood. BBA Clin 2014; 2: 18–23.
Article Google Scholar
Abuharbeid S, Czubayko F, Aigner A . The fibroblast growth factor-binding protein FGF-BP. Int J Biochem Cell Biol 2006; 38: 1463–1468.
CAS Article Google Scholar
Tassi E, Al-Attar A, Aigner A, Swift MR, McDonnell K, Karavanov A et al. Enhancement of fibroblast growth factor (FGF) activity by an FGF-binding protein. J Biol Chem 2001; 276: 40247–40253.
CAS Article Google Scholar
Ferrara N . Role of vascular endothelial growth factor in the regulation of angiogenesis. Kidney Int 1999; 56: 794–814.
CAS Article Google Scholar
Wu ZH, Zhu Q, Gao GW, Zhou CC, Li DW . Preparation, characterization and potential application of monoclonal antibody 18A4 against AGR2. Xi Bao Yu Fen Zi Mian Yi Xue Za Zhi 2010; 26: 49–51.
CAS PubMed Google Scholar
Ray A, Dittel BN . Isolation of mouse peritoneal cavity cells. J Vis Exp 2010; 35: 1488.
Google Scholar
Li B, Meng Y, Zheng L, Zhang X, Tong Q, Tan W et al. Bispecific antibody to ErbB2 overcomes trastuzumab resistance through comprehensive blockade of ErbB2 heterodimerization. Cancer Res 2013; 73: 6471–6483.
CAS Article Google Scholar
Zhang X, Gao F, Yu LL, Peng Y, Liu HH, Liu JY et al. Dual functions of a monoclonal antibody against cell surface F1F0 ATP synthase on both HUVEC and tumor cells. Acta Pharmacol Sin 2008; 29: 942–950.
CAS Article Google Scholar
Xia Y, Song X, Li D, Ye T, Xu Y, Lin H et al. YLT192, a novel, orally active bioavailable inhibitor of VEGFR2 signaling with Potent antiangiogenic activity and antitumor efficacy in preclinical models. Sci Rep 2014; 4: 6031.
CAS Article Google Scholar
Kim JY, Al-Hilal TA, Chung SW, Kim SY, Ryu GH, Son WC et al. Antiangiogenic and anticancer effect of an orally active low molecular weight heparin conjugates and its application to lung cancer chemoprevention. J Control Release 2015; 199: 122–131.
CAS Article Google Scholar
Wiggins HL, Rappoport JZ . An agarose spot assay for chemotactic invasion. Biotechniques 2010; 48: 121–124.
CAS Article Google Scholar
Hwang H, Kim EK, Park J, Suh PG, Cho YK . RhoA and Rac1 play independent roles in lysophosphatidic acid-induced ovarian cancer chemotaxis. Integr Biol 2014; 6: 267–276.
CAS Article Google Scholar
Wiig H, Aukland K, Tenstad O . Isolation of interstitial fluid from rat mammary tumors by a centrifugation method. Am J Physiol Heart Circ Physiol 2003; 284: H416–H424.
CAS Article Google Scholar

Download references

Acknowledgements

This research was supported by National Natural Science Foundation of China No. 81373319; Shanghai Science and Technology Commission Foundation No. 14431903400; Guangdong Major Science and Technology Projects Foundation No. 2012A080202014.

Author information

H Guo and Q Zhu: These authors contributed equally to this work.

Affiliations

School of Pharmacy, Shanghai Jiao Tong University, Shanghai, China
H Guo, Q Zhu, S B Merugu, H B Mangukiya, B Zhang, H Negi, R Rong, Z Wu & D Li
Engineering Research Center of Cell and Therapeutic Antibody of Ministry of Education, School of Pharmacy, Shanghai Jiao Tong University, Shanghai, China
H Guo, Q Zhu, B Zhang & D Li
Department of Neurology, Shanghai Jiao Tong University Affiliated Sixth People's Hospital, Shanghai, China
X Yu
Department of Molecular Pharmacology and Chemical Biology, School of Medicine, University of Pittsburgh, Pittsburgh, PA, USA
N Smith & Z Li
Department of Respiratory Medicine, Shanghai Pulmonary Hospital, Tongji University School of Medicine, Shanghai, China
K Cheng

Authors

H Guo
View author publications
You can also search for this author in PubMed Google Scholar
Q Zhu
View author publications
You can also search for this author in PubMed Google Scholar
X Yu
View author publications
You can also search for this author in PubMed Google Scholar
S B Merugu
View author publications
You can also search for this author in PubMed Google Scholar
H B Mangukiya
View author publications
You can also search for this author in PubMed Google Scholar
N Smith
View author publications
You can also search for this author in PubMed Google Scholar
Z Li
View author publications
You can also search for this author in PubMed Google Scholar
B Zhang
View author publications
You can also search for this author in PubMed Google Scholar
H Negi
View author publications
You can also search for this author in PubMed Google Scholar
R Rong
View author publications
You can also search for this author in PubMed Google Scholar
K Cheng
View author publications
You can also search for this author in PubMed Google Scholar
Z Wu
View author publications
You can also search for this author in PubMed Google Scholar
D Li
View author publications
You can also search for this author in PubMed Google Scholar

Corresponding authors

Correspondence to Z Wu or D Li.

Ethics declarations

Competing interests

The authors declare no conflict of interest.

Additional information

Supplementary Information accompanies this paper on the Oncogene website

Supplementary information

Supplementary Information (DOCX 1373 kb)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Guo, H., Zhu, Q., Yu, X. et al. Tumor-secreted anterior gradient-2 binds to VEGF and FGF2 and enhances their activities by promoting their homodimerization. Oncogene 36, 5098–5109 (2017). https://doi.org/10.1038/onc.2017.132

Download citation

Received: 20 June 2016
Revised: 18 March 2017
Accepted: 27 March 2017
Published: 08 May 2017
Issue Date: 07 September 2017
DOI: https://doi.org/10.1038/onc.2017.132