1 Introduction
Esophageal adenocarcinoma (EAC) is a highly metastatic disease associated with poor clinical outcomes. Overall, its 5-year survival rate is only 20%. In Western countries, a significant and sustained rise in the incidence of EAC has been observed. The introduction of neoadjuvant treatment in the form of either chemotherapy or the combination of chemo- and radiotherapy improved the survival of potentially curable cases to around 40% [
1,
2]. Combining conventional therapy with inhibition of growth factors such as EGFR and ERBB2 has slightly improved outcomes [
3], but a major breakthrough in treatment has not been reached. A major problem is that none of the existing molecular therapies has been specifically developed for EAC. Therefore, novel and effective molecular targeting treatments specifically for EAC are an unmet clinical need.
The SMAD4 tumor suppressor gene is pivotal for downstream signaling of Bone Morphogenetic Proteins (BMPs). This pathway is activated by upstream ligands such as Sonic Hedgehog (SHH) [
4,
5]. Importantly, SMAD4 is frequently lost in gastrointestinal cancer [
6,
7] and in 8 to 10 % of EAC [
8‐
10]. SMAD4 loss is associated with non-canonical BMP signaling leading to a more metastatic phenotype, a poor prognosis and a poor response to treatment [
11]. Similar observations have been reported for aberrant activation of SHH, the upstream BMP ligand, in cancers of the pancreas, stomach and colon [
12]. SHH produced by cancer-associated fibroblasts (CAFs) in the microenvironment has been demonstrated to regulate cancer progression to trigger metastasis and chemoresistance [
12].
BMP2 and BMP4 (BMP2/4) are critically involved in malignant signaling leading to cancer progression [
13] and are well-known for their involvement in metastatic and invasive behavior of cancer [
14‐
18]. In hepatocellular carcinoma [
19,
20] gastric [
21,
22] and colon cancer [
23,
24]. BMP2 contributes to tumor cell migration, invasiveness and metastasis. Upregulated BMP4 has been reported in gastric cancer, hepatocellular carcinoma and colorectal cancer [
25‐
27]. Compared to normal esophageal squamous epithelium, BMP4 expression is also significantly upregulated in EAC and its precursor lesion, Barrett’s Esophagus (BE) [
28,
29]. Recently, our group demonstrated in a murine model that inhibition of BMPs effectively inhibits the regeneration of columnar epithelium and promotes the development of a neo-squamous epithelium from stem cells residing at the squamo-columnar junction (SCJ) following the ablation of normal columnar epithelium in this region [
30].
Focusing on BMP2/4, we recently developed two specific antibodies, C4C4 and C8C8, targeting BMP4 and BMP2/4, respectively. VHHs are Llama-derived single domain antibodies, which are low molecular weight molecules of around 15 kDa. As opposed to conventional antibodies, VHHs’ antigen-binding fragment is formed by only the variable domain of the heavy chain. This enables VHHs to bind specifically and with high affinity to their associated antigen as well as to hidden epitopes within grooves or cavities. Notably, the world’s first VHH applied in the clinic, caplacizumab, was approved in Europe and the US in 2018 for patients with acquired thrombotic thrombocytopenic purpura [
31]. The US Food and Drug Administration (FDA) considers it to be a first-in-class medication [
32]. Moreover, studies have shown that VHH could serve as a potential anti-COVID-19 agent because of its potent neutralizing ability and peculiar characteristics such as small size, low immunogenicity and high affinity and stability [
33,
34]. Above all, our previous studies have demonstrated that our VHHs targeting BMP2/4 have high specificity and affinity and low off-target effects compared to conventional antibodies or antagonists [
35,
36].
Although our previous studies have identified two superior antibodies targeted for BMP2/4, important questions remain to be studied. We hypothesized that SMAD4 loss is the driver for aggressive behavior in a subset of EAC and that it allows activation of malignant pathways when stimulated by BMP2/4. In this study, we set out to assess whether selective targeting of BMP2/4 could affect the process of malignant BMP signaling, cell migration, chemoresistance and growth in SMAD4 negative EAC. Importantly, we sought to test the feasibility of the VHHs for treatment of SMAD4 negative EAC. Toward these goals, BMP2/4 and SMAD4 pathways were interrogated using data from the TCGA as well as local patient samples. Confirmatory mechanistic studies were performed using a range of preclinical models.
2 Materials and methods
2.1 Ethics statement
This study was approved by the human ethics committee, the Peter MacCallum Cancer Centre Human Research Ethics Committee (08/30, 10/108 and/or 18/211). This study was approved by the animal ethics committee, the Peter MacCallum Cancer Centre Animal Experimentation Ethics Committee (E534 and E598).
2.2 Data sources
Clinical and RNA expression data from patients with EAC from the TCGA cohort were retrieved using the R package TCGAbiolinks [
37,
38]. Presence or absence of somatic mutations in the SMAD4 gene of tissue samples from the TCGA cohort were defined by results from the “muse” pipeline, and retrieved by the function GDCquery Maf [
39]. Expression data of patients with and without SMAD4 mutations were visualized by the pheatmap [
40]. This heatmap was restricted to show the expression of the 5000 most varying genes after normalization by variance stabilizing transformation by DESeq2 and calculation of z-scores [
41].
First, a broad analysis was performed and the GSEA tool from the Broad Institute [
42] was used to compare gene set enrichment for a large number of gene sets, derived from the REACTOME, KEGG and BIOCARTA databases, in patients with EAC from the TCGA cohort with and without a SMAD4 mutation [
42‐
45]. Using 1000 permutations, significance of gene set enrichment was set to a nominal
p-value < 0.05. Enrichment plots were shown for gene sets of interest, with a graphical view of the enrichment score per gene set. Thereafter, a subset of patients was selected for further analyses to investigate pathways associated with differences in SMAD4 mutation status and RNA expression. EAC patients with a SMAD4 mutation and a control cohort consisting of EAC patients which were sub-selected based on having the highest SMAD4 expression levels and without a SMAD4 mutation, were compared by differential expression analysis by DESeq2. Significance was set to an adjusted
p-value of ≤ 0.05. A heatmap of these subsets of patients and the differentially expressed genes was visualized, with expression of these genes defined by applying z-scores on the raw count data. Results served as input for Qiagen’s Ingenuity Pathway Analysis. Significance was defined by a -log(
p-value) of ≥ 1.3 and a zscore ≥ than 2.0 or ≤ than -2.0. Results from these analyses were visualized in the pheatmap. IPA pathways of interest were selected for visualization in a heatmap of their differentially expressed genes.
2.3 Study population and human tissues
Archival formalin fixed paraffin embedded (FFPE) resection specimens of 40 EAC cases from the Amsterdam UMC treated by surgery between 2006 and 2011 were used for immunohistochemistry (IHC) to assess SHH, BMP2, BMP4 and SMAD4 expression and for targeted sequencing to detect SMAD4 mutations.
2.4 Ethical considerations
For the use of the archival tissues from the Netherlands, the collection, storage and use of patient derived paraffin embedded tissue and data were performed in compliance with the “Code for Proper Secondary Use of Human Tissue in The Netherlands”, Dutch Federation of Biomedical Scientific Societies, the Netherlands and therefore no informed consent was required. For the fresh frozen biopsies from the Amsterdam UMC biobank used for RNA sequencing, patients provided written informed consent. The protocol for retrieval of archival EAC material was in accordance with the Medical Ethical Committee and/or Amsterdam UMC biobank committee of the Amsterdam UMC (AMC 2013_241).
2.5 Histopathological review
Pathology reports and FFPE tissues were obtained for histopathological review. Hematoxylin & Eosin (H&E) stained slides were evaluated to confirm the diagnosis of EAC and to determine the grade of differentiation and to select for high tumor density areas by a dedicated gastrointestinal pathologist, who was unaware of the research and clinical outcomes.
2.6 Targeted sequencing of a gene panel including the SMAD4 gene locus
Targeted sequencing was performed for the SMAD4 gene using an amplicon-based protocol with deep coverage (mean depth 4000x). Genomic DNA was extracted from matched FFPE tissue sections using a QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Targeted sequencing was performed from 50 ng of genomic DNA for the SMAD4 gene with deep coverage (mean depth 4000x) using a library preparation Lotus DNA Library Prep Kit (IDT, Integrated DNA Technologies, Inc., Coralville, Iowa, USA) and a custom-designed panel IDT2972740 (IDT) for enrichment, according to the manufacturer’s protocol. Libraries were sequenced with a mid-output v2.5 flow cell (300 Cycles) on a NexSeq500 sequencing machine (Illumina). Fastq raw data were mapped, filtered and analyzed according to our internal pipeline (doi 10.1093/brain/awab056). For SMAD4 variant validation, KAPA HiFi HotStart (Roche, Mannheim, Germany) was used for genomic DNA amplification after which the purified PCR products were analyzed by Sanger sequencing according to a previously described method [
46].
2.7 Detection of SMAD4 deletion in ISO76A cell line by Sanger sequencing
DNA isolation of the ISO76A cell line was performed using a NucleoSpin®Tissue Kit (BIOKE). A forward primer (5′- GATTTGCGTCAGTGTCATCG-3′) and a reverse primer (5′-GCTGGAGCTATTCCACCTACTG-3′) were used to amplify a certain SMAD4 gene fragment followed by Sanger sequencing.
2.8 Immunohistochemistry
Four μm tissue sections of FFPE tissue blocks were used for IHC. After rehydration, antigen retrieval was performed by incubating the slides for 20 minutes at 98°C in 10 mM sodium citrate buffer at pH 6.0 (BMP4) or 1 mM EDTA pH 9.0 (SMAD4 and SHH). Slides were allowed to cool down and endogenous peroxidases were blocked with Peroxidase Blocking Solution (Sakura Finetek) for 10 minutes at room temperature (RT). Nonspecific sites were blocked with UltraVision Protein Block (Thermo scientific) before primary antibody incubation. All primary antibodies used were diluted in Bright Diluent (VWR) (anti-BMP2 1:200 abcam ab6285 clone 65529.111; anti-BMP4 1:200 abcam ab124715 clone EPR6211; anti-SHH 1:200 abcam ab135240 clone 5H4; anti-SMAD4 1:200 Santa Cruz Biotechnology sc-7966 clone kappa light chain) and incubation was performed over-night at 4°C. Slides were washed in phosphate buffered saline (PBS) and incubated with a BrightVision 1 step detection system and anti-rabbit/mouse secondary antibodies (VWR). The peroxidase activity was visualized using 3,3'-Diaminobenzidine (DAB) chromogen (VWR). Finally, sections were counterstained with Mayer’s hematoxylin (VWR), dehydrated and mounted. The primary antibodies were excluded in the negative controls (PBS). H&E staining was performed in order to correlate the staining pattern against the tissue architecture.
2.9 Interpretation of IHC results
IHC results were scored independently by two researchers who were blinded to the experimental protocol. Scoring was performed semi quantitively for staining intensity (0 if no stain, + if weakly positive, ++ if moderately positive, +++ if strongly positive) and percentage of positive cells (‘1’ if no positive cells, ‘2’ if 1-33% positive cells, ‘3’ if 34-66% positive cells, ‘4’ if 67-100% positive cells). Both stromal and tumor cells were scored. For the final analysis, tumors were divided in low or high expression of BMP2 or BMP4, in case the intensity of the BMP expression was at least ++ in > 3. To define SMAD4 loss, cancer cells showed an intensity of 0 or + in more than 30% of cells, which is in line with previous studies [
47].
2.10 Selection of SMAD4 negative cell line from a Tissue Microarray (TMA)
A TMA was established using tissues from six separate EAC PDX lines. These PDXs were established using an intramuscular transplantation technique as previously described [
48]. After screening for SMAD4, a SMAD4 negative PDX named ISO76A [
48] was identified. A cell line from this PDX (IS076A) was recently established at the Peter MacCallum Cancer Centre, Melbourne, Australia. After splitting, an early passage of this cell line was transfected with Luciferase-eGFP
+.
2.11 Cell culturing, SMAD4 transfection and cell sorting
SMAD4(-) ISO76A cells were transfected with a SMAD4 expressing construct (ORF expression clone for SMAD4, NM_005359.5, Labomics) and HEK293 cells using Lipofectamine solution (Life Technologies) according to manufacturer’s instructions. To purify ISO76A cells and SMAD4 transfected ISO76A cells from fibroblasts and non-transfected cells, fluorescence-activated cell sorting (FACS) (BD FACSAria™) was performed to select E-cadherin positive and mCherry for sorting SMAD4(+) ISO76A cells, and E-cadherin positive only for sorting SMAD4(-) ISO76A cells. Both SMAD4(-) ISO76A and SMAD4(+) ISO76A cells were cultured in RPMI-1640 culture medium with 10% fetal calf serum (FCS), 2mM glutamine and 100 U penicillin/0.1 mg/ml streptomycin. The cells were placed in an incubator with a 5% CO2 concentration at 37°C. All cell lines were tested for Mycoplasma contamination once a month. All in vitro experiments were performed on cells growing exponentially.
2.12 Cell chemoresistance assay
5 × 104 cells/well were seeded in 96-well plates. After 12 hours, the medium was replaced with fresh medium without serum. Next, cells were treated with cisplatin and/or C4C4 and C8C8 for 24 hours at 37°C in a 5% CO2 atmosphere. Subsequently, the medium was removed and 100 μl CellTiter-Blue (Promega) solution was added into each well. The 96-well plates were then incubated for an additional 3 hours at 37°C in a 5% CO2 atmosphere. Fluorescence was analysed with a Synergy HT Multi-Mode Microplate Reader (Biotek) using 530(25)Ex/590(35)Em settings.
2.13 Cell migration assay
106 cells/well were seeded in 6-well plates and grown until 90% confluence. A sterile 200 μl pipette tip was held vertically to scratch (‘wound’) across each well. The detached cells were removed by washing with 2 ml PBS. Cells were treated in triplicate with BMP4, BMP2/4, BMP4 + C4C4, BMP2/4 + C8C8 and control at the specific concentrations for 24 hours. Sample images were obtained at 0 hour and 24 hours using a microscope at 5x magnification. Quantification of wound closure in each well was performed using Image J software.
2.14 Luciferase assay for testing BMP activity
SMAD4(-) ISO76A, OE19, OE33, SK-GT-4, OACM5, OACP4 and Flo-1 cells were cultured in 96-well plates at 5 x 103 cells/well under the same conditions and cells were placed in 37°C incubator overnight to attach. 100 μl RPMI-1640 medium with 0.1% FCS was added to each well. Cells were plated in triplicate. Wells with no cells were used as control. 100 μl luciferase substrate solution from a Bright-Glo Luciferase Assay System (Promega Benelux) was added to each well. After incubation for three minutes at RT, luciferase activity was measured using a Synergy HT Multi-Mode Microplate Reader (Biotek). Normalization of luciferase values was performed by subtracting background activity as measured in the control wells.
2.15 Western blotting
Cells were grown in 6-well plates after which they were washed with PBS and a cell lysis RIPA buffer (Thermo Scientific) containing protease and phosphatase inhibitors (Thermo Scientific) was added. The cells were lysed on ice for 20 minutes, scraped into 1.5 ml tubes, and centrifuged for 15 minutes at 14,000 rcf at 4°C. Protein levels were measured with a BCA kit (Thermo Scientific) using clear lysates. The protein levels in each sample were equalized, the lysates were mixed with 4x loading buffer (Bio-Rad) and boiled at 97°C for five minutes. The proteins were separated by sodium dodecyl sulfate gel electrophoresis (SDS-PAGE) and blotted onto a polyvinylidene difluoride (PVDF) membrane (Millipore, Billerica, MA, USA). The resulting blots were blocked in blocking buffer containing 5% Bovine Serum Albumin (BSA) dissolved in 1x Tris-buffered saline containing 0.1% Tween-20 (TBST) and washed three times in 1x TBST for five min each. Both primary and secondary antibodies were diluted in blocking buffer containing 5% BSA. The primary antibody was incubated overnight at 4°C. A secondary horseradish peroxidase (HRP) linked antibody was incubated for one hour at room temperature after which the blots were exposed to Lumilight+ (Roche, Basel, Switzerland) chemiluminescent substrate and visualized using a chemiluminescence imager (Bio-Rad).
2.16 Mice
Animal experiments were performed in accordance with the National Health and Medical Research Council Australian Code of Practice for the Care and Use of Animals for Scientific Purposes and approved by the Peter MacCallum Cancer Centre (PMCC) Animal Experimentation Ethics Committee. NOD-SCID IL-2RγKO (NSG) mice were bred in-house.
2.17 Tumor xenografts
To establish xenografts of the EAC cell line ISO76A (PDX model) transfected with e-GFP, 2.5 million cells suspended in 100 μl 1:1 PBS and Matrigel (BD Bioscience) were subcutaneously injected into the flank of 6-8 weeks female immunodeficient NSG mice. Subcutaneous tumor volumes were determined biweekly with calipers and calculated using the formula (length × width2)/2. All mice were euthanized when subcutaneous tumors reached ≥ 1500 mm3 or if humane endpoints were reached (labored breathing, bloated abdomen or weight loss in excess of 15% of baseline body weight).
2.18 In vivo bioluminescence imaging of PDX in NSG mice
To assess the establishment and growth of the PDX in the mice, in vivo imaging was performed. Luciferase-eGFP+ ISO76A cells (2.5 million cells per mouse) suspended in PBS and Matrigel were subcutaneously injected into NSG mice. Animals were imaged on a Xenogen IVIS 100 Imaging System (Caliper Life Science) for life imaging of the tumor xenografts. 100 μl of 20 mg/ml luciferin (Promega) in PBS was subcutaneously injected into each mouse 5 min before imaging. Imaging was performed under general anesthesia. Animals were shaved before imaging. Live imaging was performed at different endpoints and at the end of the study. At the study endpoint, the whole mouse and its organs were imaged to determine the tumor burden. The imaging exposure times were 60 s for whole animals and 5 min for organs. The bioluminescence signal was quantified using Living Image software.
2.19 In vivo imaging of BMP2/4 expression and delivery of VHHs in the PDX model
To assess the biodistribution of our recently developed highly specific VHHs (C4C4 and C8C8) within the PDX model, in vivo imaging was performed. To this end, C4C4 and C8C8 labeled with IRDye800cw (QvQ, Utrecht, the Netherlands) was used. At approximately two weeks post inoculation, three mice bearing tumors with a volume in excess of 1000 mm3 were used to investigate the delivery and retention of C4C4 and C8C8 within the tumor xenografts. Mice were administered 25-50 μg of labeled VHHs through tail vein injection and imaged after 30 minutes, 2 hours, 4 hours and 24 hours. Animals were imaged on a Xenogen IVIS 100 Imaging System (Caliper Life Science). At the end of the experiment, mice were sacrificed and their organs imaged to determine the retention of the drug in various organs in addition to the tumor xenograft. The imaging exposure times were 60 seconds for the whole animal and up to 5 minutes for ex vivo imaging of the organs. The infrared signal was quantified using the ‘region of interest’ function in Living Image software.
2.20 Treatment of the PDX model with C4C4 and C8C8
To test the anti-tumor activity of C4C4 and C8C8, both individually and in combination with cisplatin, xenografts were established subcutaneously in 30 NSG mice using firefly-luciferase-expressing SMAD4(-) ISO76A cells as previously described. The tumor growth was evaluated twice per week using electronic calipers. For treatment, animals were randomized into six groups: group 1 received vehicle (0.9% NaCl); group 2 received cisplatin 2 mg/kg once per week; group 3 received daily 100 μg C4C4 via intraperitoneal injection (IP); group 4 received daily 100 μg C8C8 IP; group 5 received cisplatin 2 mg/kg once per week in combination with C4C4, 100 μg daily IP, and group 6 received cisplatin 2 mg/kg once per week and C8C8, 100 μg daily IP. Treatment was started once the PDX reached a volume of at least 80 mm3 and continued for 4 weeks or until tumors reached a volume of 1500 mm3 or a humane endpoint was reached. After reaching the endpoints the mice were sacrificed and organs were harvested.
2.21 Survival study of SMAD4(-) ISO76A based PDTX model with continued treatment of C4C4 and C8C8
To evaluate long term outcomes, a survival study was performed using the same SMAD4(-) ISO76A PDX model. After an induction treatment for 4 weeks with C4C4, C8C8 and cisplatin (2 mg/kg), C4C4 and C8C8 treatment regimens were continued in doses of 100 μg five times per week, 100 μg three times per week and 100 μg once per week, in combination with cisplatin (2 mg/kg), until humane endpoints were reached. Control groups were animals treated with cisplatin (2 mg/kg) once per week for 4 weeks and animals treated with the vehicle (0.9% NaCl). Mice were euthanized when the PDX reached a volume of ≥ 1500 mm3 or when the mice showed signs of ill health or a weight loss of > 15% of their baseline body weight.
2.22 Statistical analysis
All data consolidation, organization and processing were carried out using Microsoft Excel. Image J was used for quantifying gray levels of bands in Western blotting and cell wound closure levels in cell migration assays. GraphPad Prism 9 was used for testing significant differences for cell chemoresistance and tumor volumes by paired t-test, and survival analysis by Log-rank (Mantel-Cox) test. For the remaining experiments, unpaired t-test and two-tailed distribution at the 5% level were used. IBM SPSS was used for performing survival analyses through the Kaplan Meier method.
4 Discussion
EACs are genetically highly instable, and characterized by a high mutational load [
50]. Due to the heterogeneity of these cancers, response to the current standard care of treatment is unpredictable, whereas the overall 5 year survival outcomes are poor. Identifying subsets of patients who can benefit from targeted therapies is, therefore, of paramount importance for improving EAC patient outcomes. The SMAD4 gene has been found to be frequently mutated in a subset of EAC [
8,
51]. Within the SHH-BMP-SMAD signaling cascade, SMAD4 is a downstream intracellular target, which together with several nuclear factors, regulates the transcription of factors involved in cell growth and development. Derangement of this signaling cascade through SMAD4 gene mutations can lead to alternative SHH-BMP signaling via non-canonical/oncological routes. Previous studies have shown that BMP signaling in the absence of SMAD4 leads to the activation of a broad range of non-canonical signaling pathways and to enhancement of invasion and metastasis in colorectal cancer [
7]. Our aim was to gain more insight into the biological mechanisms and to develop a targeted therapy for EACs that carry SMAD4 mutations and/or deletions. The mutational rate of this gene is generally found between 8-10% in EAC and to be associated with more frequent disease recurrences and a poor survival [
8‐
10]. We first analyzed a cohort of in house EAC cases and cases from the TCGA database for SMAD4 mutations, which confirmed the mutational rate to be between 5 and 10%. However, by IHC, we found decreased expression of SMAD4 in 20% of our cases and that this loss of SMAD4 expression was significantly correlated with poor survival. Discordance between SMAD4 mutational status and protein expression has been described earlier and could be due to epigenetic silencing mechanisms regulating gene expression [
47]. The finding of a higher number of patients with loss of SMAD4 expression, however, is of importance for the future selection of patients. It could mean that potentially more patients may benefit from treatments targeting tumors with a deficient SMAD4 status.
We found that all of the SMAD4 negative EAC cases exhibited preserved expression of BMP2 and/or 4. Targeting BMPs within the SHH-BMP signaling pathway is highly attractive, because the BMP molecules act extracellularly and, as such, can be easily reached by antibodies such as our recently developed highly selective low molecular weight llama VHHs. These VHHs specifically target BMP4 (C4C4) and BMP2/4 (C8C8) and are superior to conventional antibodies [
35,
36]. In this study, we focused on evaluating the effects of inhibiting BMP4 and BMP2/4 in SMAD4 mutated cells in comparison to SMAD4 expressing cells. Hereto, we selected a unique recently established SMAD4(-) EAC xenograft-derived cell line, ISO76A, which we transfected with a wild type SMAD4 expression construct to obtain a ‘SMAD4(+) ISO76A’ counterpart.
To investigate whether cases with a low SMAD4 expression exhibit increased non-canonical signaling, we performed GSEA analysis and compared a subset of EAC cases with the highest SMAD4 expression to those with the lowest expression and found that one of the known BMP non-canonical signaling pathways, the p38 MAPK pathway, was significantly enriched. We confirmed involvement of this pathway by stimulating SMAD4(-) ISO76A cells with BMP2 and BMP4, which enhanced the expression of p38 and pERK. In contrast, stimulation of SMAD4(+) ISO76A cells by these BMPs only increased the normal canonical BMP signaling. These results are in agreement with a previous report using the colorectal cell lines SMAD4(+) HCT116 and SMAD4(-) HCT116 [
7]. In contrast to other reports, BMP stimulation had no significant effect on the expression of p-NFkB. Also, BMP stimulation did not induce any changes in EMT markers. Thus, SMAD4 mutation and/or loss in EAC seems to be the switch that changes BMP signaling from tumor-suppressive canonical BMP signaling, to tumor-promoting non-canonical BMP signaling.
In search of the upstream signals which induce BMP2/4 expression, we found a significant correlation between SHH, BMP2 and BMP4 through the analysis of the TCGA database. Using IHC, we confirmed the presence of SHH in both stromal and cancer cells of EAC, while BMP2 and 4 were mostly expressed in the cytoplasm of the tumor cells. Upregulated SHH signaling is often observed in EAC [
52‐
54], but also in its precursor lesion known as BE [
4]. SHH is known to upregulate several types of BMPs including BMP4 [
55] and can be produced by cancer-associated fibroblasts in the cancer microenvironment [
12]. Under in vitro conditions, we found that cultured tumor cells, in the absence of a stromal compartment, failed to express BMPs and that this can be overcome under the influence of SHH. In addition, we not only demonstrated that SHH primarily induces BMP2 and BMP4 expression in ISO76A cells, but also that high SHH, BMP2 and 4 expression increases the chemoresistance of SMAD4(-) ISO76A cells. This is in line with a previous report [
56], which demonstrated that BMP4 is highly expressed in cisplatin-resistant gastric cancer cells and that targeting BMP4 sensitizes these cells to cisplatin. Of importance is that in our cell viability assay we showed that the chemoresistance induced by SHH, or directly by BMP2 and 4 stimulation, could be rescued by inhibiting BMP4 and BMP2/4 using C4C4 and C8C8, respectively. These effects were not seen in SMAD4(+) ISO76A cells. Patients with EAC notoriously exhibit a varied response to chemotherapy and they are relatively chemo resistant. Therefore, targeted therapies which can overcome chemoresistance are of importance.
EAC is also a highly metastatic cancer, which accounts for the high recurrence and mortality rates of this disease. Cancer cell migration assays can be used to investigate a cancer cell’s capacity to infiltrate in neighbouring tissues, to enter lymphatic and blood vessels and to disseminate into the circulation [
57]. We investigated whether the effects of the BMP inhibition would reduce the migration of ISO76A cells. Indeed, our migration assay indicated that BMP2/4 accelerated the migration of SMAD4(-) ISO76A cells, which could be effectively inhibited by our VHHs. These effects were absent in SMAD4(+) ISO76A cells.
The effects on migration and chemoresistance of the VHHs on the SMAD4 negative EAC cells, prompted us to further study the effects of the VHHs in vivo using SMAD4(-) ISO76A cells in a PDX model in NSG mice. This allowed us to investigate the anti-cancer effect of VHHs in an intact tumor micro-environment. IHC revealed that SHH, BMP2 and BMP4 were expressed in this PDX model. Delivery and effective targeting of the PDXs in this model was confirmed in real time through in vivo imaging using IRDye 800CW labeled VHHs and by ex vivo imaging of organs. We found that the labeled VHHs reached tumor sites within 30 minutes post tail vein injection and retention of VHHs in tumor sites was still observed 24 hours post injection. These results are in agreement with the physiological properties of the VHHs, which have a high target affinity and binding [
35,
36]. Also, rapid secretion of VHHs via kidneys and liver was observed two hours post injection, which is thought to be related to the low molecular weight of the VHHs.
Finally, we studied the tumor response in terms of tumor size by systemic administration of the VHHs. We found that the VHHs used alone, or in combination with cisplatin, significantly inhibited tumor growth. The combination of the VHHs with cisplatin showed a further decrease in tumor growth, which suggests that there is a synergetic effect between the VHHs and cisplatin in inhibiting tumor growth. These results are in line with our cell viability assay indicating that the VHHs decreased chemo-resistance to cisplatin. It is well known that cisplatin as an anticancer drug has unavoidable side effects in patients and animals [
58]. One of these side effects is mucosal damage, which can appear anywhere within the gastro-intestinal tract [
58,
59]. In the present study, we investigated the organs of the animals only at the end of the study period. At this time point, we found that the number of goblet cells in the colon of the mice were severely reduced in those treated with cisplatin. This effect was not seen in the group treated with the VHHs. Even more interesting is that the combination of the VHHs with cisplatin did not result in this side effect. This could mean that the anti-BMP VHHs protect against the side effects of cisplatin. The mechanism behind this observation needs to be further clarified. Another future priority includes further assessment of the potential side effects of VHHs, both when administered alone and in combination with chemotherapeutic agents.
In summary, through data analysis, imaging and functional assays in in vitro and in vivo preclinical models, we demonstrated the anti-cancer effects and feasibility of selectively targeting BMP2/4 in SMAD4 negative EAC by novel anti BMP4 and BMP2/4 VHHs. To translate these findings to the clinic, future patient trials are required. We anticipate that specific targeting of BMP2/4 using the VHHs could form a basis for personalised treatment in SMAD4 deficient EAC, and perhaps in other SMAD4 mutated cancers. Based on the present data, we believe that targeting BMP4 and BMP2/4 can potentially improve clinical outcomes of the highly aggressive SMAD4 deficient subgroup of EAC.
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.