Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2025 Dec 1.
Published in final edited form as: J Pathol. 2024 Oct 22;264(4):411–422. doi: 10.1002/path.6356

Suppression of dystroglycan function accompanies pancreatic acinar-to-ductal metaplasia and favours dysplasia development

Ge Huang 1, Luke Ternes 1, Christian Lanciault 2, Kevin MacPherson-Hawthorne 3, Young Hwan Chang 1,4, Rosalie C Sears 3,4,5, John L Muschler 1,4,5,*
PMCID: PMC11560643  NIHMSID: NIHMS2021822  PMID: 39435649

Abstract

The basement membrane (BM) is among the predominant microenvironmental factors of normal epithelia and of pre-cancerous epithelial lesions. Evidence suggests that the BM functions not only as a barrier to tumour invasion but also as an active tumour-suppressing signalling substrate during premalignancy. However, the molecular foundations of such mechanisms have not been elucidated. Here we explore potential tumour-suppressing functions of the BM during pre-cancer evolution, focusing on the expression and function of the extracellular matrix receptor dystroglycan in the pancreas and pancreatic disease. We show that the dystroglycan protein is highly expressed in the acinar compartment of the normal pancreas, but lower in the ductal compartment. Moreover, there is a strong suppression of dystroglycan protein expression with acinar-to-ductal metaplasia in chronic pancreatitis and in all stages of pancreatic pre-cancer and cancer evolution, from acinar-to-ductal metaplasia to dysplasia to adenocarcinoma. The conditional knockout of dystroglycan in the murine pancreas produced little evidence of developmental or functional deficiency. However, conditional deletion of dystroglycan expression in the context of oncogenic Kras expression led to a clear acceleration of pancreatic disease evolution, including accelerated dysplasia development and decreased survival. These data establish dystroglycan as a suppressor of pancreatic dysplasia development, and one that is muted in chronic pancreatitis and at the earliest stages of oncogene-induced transformation. We conclude that dystroglycan is an important mediator of the tumour suppressing functions of the BM during pre-cancer evolution and that reduced dystroglycan function increases cancer risk, highlighting the dynamics of cell-BM interactions as important determinants of early cancer progression.

Keywords: pre-cancer, extracellular matrix, basement membrane, dysplasia

Introduction

The tumour microenvironment regulates cancer progression at all stages [1,2]. There are many potent microenvironmental factors, including diverse infiltrating cells, secreted soluble factors, and extracellular matrices (ECMs), and these factors are dynamically regulated with disease progression [1,2]. Dynamic changes in the tissue microenvironment can result from oncogenesis or they can precede oncogenesis, as observed in chronic inflammatory conditions, creating a higher risk environment for cancer development in the event of oncogenic insult [2–4].

The basement membrane (BM) is among the predominant microenvironmental factors of normal and premalignant epithelia. This thin specialized layer of ECM separates the epithelial and stroma compartments and persists from the normal cell stage through high grade dysplasia, with escape through the BM defining early invasive disease [5]. The cell-BM interface serves as a barrier to cell invasion and also as a complex signalling interface populated by an array of ECM receptors and ligands [5,6]. Like other microenvironmental factors, this interface is a source of signals that can either suppress or favour disease progression [5,7]. Several prominent BM receptors are strongly implicated in tumour promotion, including the β1 and β4 integrins [7,8]. Conversely, as a regulator of normal tissue architecture and homeostasis, cell-BM interactions are also believed to enforce tumour suppression, however the molecular underpinnings of such tumour suppressor functions have not clearly emerged.

Dystroglycan (DG) is a prominent BM receptor that has been implicated as a tumour suppressor in diverse cancers [9–16]. DG is a nearly ubiquitous cell-surface receptor for multiple ECM ligands. It is encoded by a single gene product and auto-proteolysed into an extracellular α-DG subunit and transmembrane β-DG subunit that remain associated at the cell surface [17]. DG binding to its multiple ECM ligands, including laminins, perlecan and agrin, depends on unique α-DG glycosylation [16,18]. Loss of DG function is a prominent cause of congenital muscular dystrophies and is studied extensively in the context of those diseases [19]. However, DG expression and function are also frequently suppressed in diverse cancers, ranging from breast, pancreatic, colon, and prostate, and in some cases, this suppression is associated with poor outcomes [9–11,16,20]. Suppression of DG function can occur by reduced core protein expression or by altered DG glycosylation that disrupts its ligand-binding functions [11,14,16,21,22]. In vitro studies show that loss of DG function correlates with more aggressive cancer subtypes, and restoration of function regulates growth, epithelial polarity, laminin assembly and internalization, and invasion [13–15,23]. In vivo interrogation of DG function in cancer development has been thus far limited to xenograft experiments where restoration of DG function inhibited growth of breast and prostate cancers [12–15,24].

Here we interrogate the expression and function of DG in pancreatic pre-cancer and cancer evolution in human and murine tissues, and by genetic perturbation in autochthonous murine models of pancreatic ductal adenocarcinoma (PDAC). Results presented here establish DG as a dynamically regulated BM receptor that is suppressed in pancreatic disease settings beginning with acinar-to-ductal metaplasia (ADM) in pre-cancers and in chronic pancreatitis. Deletion of dystroglycan function in murine disease models exhibits a clear acceleration of pancreatic pre-cancer evolution, establishing DG as a tumour suppressor acting at the cell-BM interface during the earliest steps of disease evolution.

Materials and methods

Animal models

All animal work was performed in accordance with Institutional Animal Use and Care Committee (IACUC) guidelines of the Oregon Health & Science University (OHSU) and approved by the IACUC at OHSU. Transgenic mouse lines used here were the Ptf1a-Cre (p48-Cre), Ptf1a-CreERTM (p48-CreERTM), LSL- KrasG12D, LSL-Trp53R172H, and Dag1fl (DGfl) lines [25–28]. All transgenic lines used were backcrossed at least 5 generations onto the C57Bl/6J background. For tamoxifen induction of Cre recombinase activity, mice bearing the p48-CreERTM transgene were treated with 5 consecutive daily intraperitoneal injections of tamoxifen at 7 weeks of age using 75 mg tamoxifen per kg body weight.

Human tissues

Formaldehyde-fixed and paraffin-embedded (FFPE) human tissues were obtained from tissue archives at the Oregon Health & Science University (OHSU). FFPE tissue blocks were sectioned and H&E stained at the OHSU Histopathology Shared Resource. A pancreatic tissue microarray was purchased from US Biomax (BIC4011b) containing additional specimens of pancreatic dysplasia, pancreatitis and PDAC. This study received IRB approval under OHSU protocol IRB00003609.

Immunofluorescence staining

FFPE tissue sections were subject to antigen retrieval methods following deparaffinization and rehydration. The Dako Target Retrieval Solution, pH9, (preparation: S236784–2, Agilent, Santa Clara, CA, USA) was used for antigen retrieval when immunostaining for β-DG, amylase, or cytokeratins. FFPE sections were immersed in the Dako solution and heated in a pressure cooker. For α-DG immunostaining, FFPE sections were immersed in 20 ug/ml proteinase K dissolved in TE buffer (50 mM Tris Base, 1 mM EDTA, 0.5% Triton X-100, pH 8.0) and incubated at 37 °C for 45 min. Following antigen retrieval, slides were rinsed in PBS buffer and then blocked with blocking buffer (1X PBS, 1% BSA, 2% normal horse serum, 0.3% Triton™ X-100). Primary antibodies used were the anti-Dystroglycan (EMD Millipore, Billarica, MA, USA: 05–593), anti-β-Dystroglycan (Santa Cruz, Houston, TX, USA: sc-33701), anti-“pan-cytokeratin” (Santa Cruz: sc-15367) that preferentially stains ductal cytokeratins, and anti-amylase (Santa Cruz: sc-12821) antibodies. Sections were incubated overnight at 4 °C with primary antibodies, rinsed with PBS, and incubated with secondary antibodies (Invitrogen: A-21043, A10037, A-21206, or A32849) at room temperature for 1.5 h. Lastly, the slides were rinsed with PBS and mounted under coverslips using DAPI-containing Prolong® gold anti-fading mounting medium (Invitrogen: P36931). Fluorescence and H&E staining images were acquired using a Carl Zeiss Axioscan Z1 slide scanner (Zeiss, Oberkochen, Germany).

Immunostaining quantification

Immunofluorescence signals were quantified using ImageJ software (https://imagej.net/ij/download.html). The threshold tool was applied manually to select the amylase-, ductal cytokeratin-, or DAPI-positive tissue regions. The area of amylase, ductal cytokeratin, and DAPI staining was measured in pixels. The percentage of amylase or ductal cytokeratin positive area was expressed relative to the total DAPI-positive area. Any non-pancreatic tissues (e.g. lymph nodes or intestine) were manually cropped and excluded from calculations. Tissue assessments by VISTA were performed as described previously [29].

Single cell RNA-seq (scRNA-seq) data analysis

DAG1 expression was interrogated across cell types at the single cell level utilizing a previously published PDAC scRNA-seq atlas curated and analysed by Oh and colleagues [30]. The epithelial compartment results, which include cell type annotation and clustering results, were analysed using the R package Seurat (v4.3.0). The data was subset to retain acinar (AMY2A, PRSS1), normal duct (CFTR, SCTR), metaplastic (MMP7, MUC6), and neoplastic (KRT17, TFF3) cells. After RNA count data were normalized and scaled, DAG1 differential expression was performed between acinar cells and other cell states using the Seurat function FindMarkers. FindMarkers parameters were set to 0.25 minimum log2 fold-change, to not threshold results by minimum expressing cells, and to test differential expression via a Wilcoxon test.

Statistical analysis

IBM SPSS® 25 software (IBM, Armonk, NY, USA) was used to perform statistical analyses. Difference of means was assessed by independent sample t-test or one-way ANOVA. The Kaplan–Meier survival curve function was used to assess differences in survival times between cohorts. Differences in the incidence of PDAC and metastases were assessed by χ2 test. A two-tailed p-value equal to or less than 0.05 was considered significant.

Results

Low or reduced dystroglycan protein expression is evident in pancreatic cancers and dysplasia.

DG protein expression was assessed first in normal human pancreatic tissues by immunofluorescence staining. This staining showed prominent DG protein expression at the cell-BM interface in acinar epithelial cells, and detectable but consistently lower expression at the ductal cell-BM interface (Figure 1A and supplementary material, Figure S1). DG immunostaining was also weak or undetectable in the endocrine cells of the islets of Langerhans (supplementary material, Figure S1A). This staining was achieved using an anti-β-DG antibody which detects the core protein of the DG transmembrane subunit, and with the anti-α-DG monoclonal antibody clone IIH6 which detects the functional carbohydrate modifications that determine the ligand-binding properties of the α-DG subunit [16,18]. Positive IIH6 antibody staining in the acinar cells indicates that the DG protein is expressed as a functional ECM receptor in pancreatic acini. Weaker immunostaining in the ductal compartment for both β-DG staining (core protein) and α-DG staining (IIH6 antibody) indicates that reduced core protein expression was the primary cause for the lower signal, and not altered glycosylation alone [16,18]. Among known DG ligands, laminin-111 is detected prominently in the exocrine pancreas (supplementary material, Figure S1A), and multiple laminin isoforms and perlecan have been detected in the pancreas by proteomics [31].

Figure 1.

Figure 1.

Open in a new tab

DG protein expression is high in normal acinar cells but low in ductal cells and in PDAC. (A) H&E staining and immunofluorescence staining for pan-CK, β-DG and α-DG are shown for normal human pancreatic tissue. Yellow arrowheads point to acinar cells and blue arrows point to ductal cells. (B) Representative H&E and immunostaining of human PDAC tissue for the same markers shown in (A). Red arrow heads point to pancreatic cancer cells. Scale bars, 100 μm.

Immunostaining for DG protein expression in human PDAC tissues revealed low or undetectable DG protein levels in the PDAC cells using antibodies for either β-DG or α-DG (IIH6) (Figure 1B). Among 15 PDAC samples stained, all showed low or undetectable DG expression by core protein staining with the anti-β-DG antibody. These results are consistent with prior work where reduced dystroglycan mRNA and protein expression was reported in pancreatic cancers and consistent with reports of reduced DG expression in a wide range of other solid tumours [9,10,16,20].

DG protein expression patterns were next assessed at earlier, pre-invasive stages of human cancer evolution, examining high- and low-grade dysplasia appearing at the periphery of PDAC biopsies. Staining for DG was consistently low, relative to normal acinar cells, in both high and low-grade dysplasia, whether assayed using anti-β-DG protein or anti-α-DG (IIH6) monoclonal antibodies (Figure 2). Heterogeneity was observed among dysplasias, with some expressing relatively higher levels of DG, although always at lower levels than detected in normal acinar cells (Figure 2, column b). Staining of intraductal papillary mucinous neoplasm (IPMN) samples also showed lower DG expression relative to adjacent normal acinar cells (supplementary material, Figure S2). These data indicate that DG protein expression and function are high in the pancreatic acinar compartment, lower in the ductal compartment, and are consistently low or suppressed throughout PDAC evolution beginning at the earliest, pre-cancerous stages.

Figure 2.

Figure 2.

Open in a new tab

Suppression of DG protein expression is evident in all stages of human pancreatic cancer evolution, from preneoplastic lesions to adenocarcinoma. H&E and immunofluorescence staining are shown for diverse human tissue regions including (a) normal acinar cells and low grade dysplasia, (b) low grade dysplasia, (c) high grade dysplasia, and (d) carcinoma cells, corresponding to the boxed regions on the left. Immunostaining is shown for β-DG, α-DG and cytokeratins (CK). The asterisk in column d markers a nerve that is positive for DG staining. Scale bars, 500 μm.

Suppression of dystroglycan protein expression accompanies ADM associated with pancreatic cancers and chronic pancreatitis.

The aetiology of lower DG expression levels observed in dysplasia and adenocarcinoma could be consistent with a ductal origin of these lesions or with suppression of dystroglycan expression in lesions originating from acinar cells having undergone ADM, an adaptive cellular program arising in response to tissue stress [3]. ADM occurs with pancreatitis, being transient with acute inflammation, and persistent in chronic disease. ADM also occurs with oncogene-induced transformation and is an intermediate on the path of acinar cell conversion to dysplasia [3].

DG protein expression was next assessed in pancreatic tissues exhibiting ADM. Regions of ADM were identified at the periphery of human PDAC samples by histology and by molecular markers of ADM in the pancreas, including the loss of amylase immunostaining and induction of ductal cytokeratin staining [32,33]. Immunostaining in these tissues showed clearly reduced DG protein expression in cells having undergone ADM (Figure 3A). DG staining was strong at the cell-BM interface of amylase-expressing acinar cells, but significantly reduced or undetectable around cells with positive ductal cytokeratin staining. Notably, the suppression of DG protein expression was so tightly correlated with ADM that, even within hybrid structures exhibiting both acinar and ductal components, DG staining was detected uniquely in the amylase-positive cells and suppressed in the ductal cytokeratin-positive cells (Figure 3A).

Figure 3.

Figure 3.

Open in a new tab

DG protein expression is suppressed with acinar to ductal metaplasia (ADM). H&E and immunofluorescence staining of human tissues are shown for regions exhibiting ADM adjacent to PDAC tissue (A) and ADM associated with chronic pancreatitis (B). Immunostaining is shown for amylase (AMY), cytokeratins (CK) and β-DG. Boxed regions of A and B are enlarged in the bottom row. Yellow and orange arrows point to hybrid structures of acinar cells and ADM. Scale bars, 100 μm.

DG protein expression is clearly suppressed with ADM in pancreatic specimens containing adenocarcinoma, however, both inflammation and oncogene-induced ADM can occur at the periphery of PDAC tissues, and these data do not indicate whether DG suppression is oncogene-dependent or independent. To address this question, chronic pancreatitis tissue samples were tested. As observed in PDAC-adjacent tissues, DG immunostaining was markedly suppressed or absent in areas of ADM with background chronic pancreatitis (Figure 3B). The suppression of DG protein expression was again observed in hybrid lesions displaying both acinar and ductal phenotypes with DG detected uniquely on the amylase-positive acinar cells (Figure 3B). These data demonstrate that suppression of DG protein expression is tightly coordinated with ADM induced either by oncogenesis or an inflammatory state.

Suppression of dystroglycan protein expression in human ADM and PDAC is replicated in murine models of oncogene-induced pre-cancers and PDAC.

Pancreatic dysplasia and evolution to adenocarcinoma is well modelled in mice by conditional expression of the KrasG12D mutation in the pancreas and accelerated by the inclusion of a Trp53 mutation [25,26]. We therefore tested in these murine PDAC models whether DG expression was similarly altered with disease progression driven by oncogenic Kras. As observed in normal human tissues, DG was strongly expressed and functional in the normal acinar compartment of the normal murine pancreas and expressed at lower levels in normal ducts and endocrine cells (supplementary material, Figure S3).

We next employed the p48-Cre; LSL-KrasG12D (KC) and the p48-Cre; LSL-KrasG12D; LSL-Trp53R172H (KPC) models (Figure 4A) [25–27]. Immunostaining of tissues from the KC and KPC mouse models showed DG expression patterns that closely mirrored those observed in human pancreatic disease. DG was markedly reduced in PDAC cells, relative to the normal acinar cells (Figure 4B). As in human tissues, this reduced expression was also observed in high and low-grade dysplasia (Figure 4B,C). The heterogeneity of DG staining observed in high-grade human dysplasia was also observed in the murine models; some dysplasia showed higher levels of DG staining and heterogeneity even within single lesions (Figure 4B). In addition, the loss of DG protein expression was clearly evident in cells exhibiting ADM (Figure 4B,C).

Figure 4.

Figure 4.

Open in a new tab

The suppression of DG protein expression observed in human pancreatic disease is replicated in murine models. (A) Schematic describing the transgenic mouse lines used to generate the KC, KPC, KC-ΔDG (KC-ΔDGfl/fl), KPCER, and KCER-ΔDG mouse models. (B) H&E and immunofluorescence staining is shown for β-DG and cytokeratins (CK) in tissue from induced KPCER mice where both normal acinar cells and cancer cells were evident. Blue arrowheads indicate acinar cells, purple arrowheads indicate ADM, yellow arrows indicate dysplasia and red arrows indicate PDAC cells. (C) H&E and immunofluorescence stain is shown for β-DG and cytokeratins (CK) in the pre-cancerous lesions. Representative images are shown for ADM (left column), low grade dysplasia (middle column) and high grade dysplasia (right column). Scale bars, 100 μm.

The mechanism for loss of DG protein expression was probed by interrogation of a previously published PDAC single-cell RNA-seq atlas curated and analysed by Oh and colleagues [30]. These data showed that dystroglycan mRNA expression was generally low relative to other genes, even in normal acinar cells. Most importantly, the data revealed no significant change in dystroglycan mRNA levels between normal acinar and ductal cells, between acinar and preneoplastic cells, and between acinar cells and cancer cells, indicating that suppression of DG in cancer evolution is post-transcriptional (supplementary material, Figure S4).

Deletion of dystroglycan expression accelerates pre-cancer evolution, including ADM and dysplasia, and leads to reduced survival.

The suppression of DG protein expression during the early stages of disease could have a significant impact on pre-cancer and cancer evolution, however, a functional role for DG during pre-cancer evolution in vivo has not been assessed. To address this question, we employed the conditional deletion of DG gene expression in the autochthonous KC mouse model. The KC model was chosen because it exhibits the key features of pancreatic pre-cancer and PDAC evolution, and exhibits these in a temporally prescribed and slowly evolving progression over more than a year-long timeframe [25,26]. The floxed DG mouse line (DGfl)[28] was crossed with the p48-Cre mice to generate the pancreas-specific DG knockout (p48-Cre-DGfl/fl or ΔDG), and crossed with the KC mice allowing conditional DG deletion simultaneous with induction of oncogenic KrasG12D expression (KC-ΔDG) (Figure 4A).

Deletion of DG alone from the murine pancreas had little apparent impact on the histology or function of the pancreas. Conditional DG knockout mice were in good health, were of comparable weight to control animals, histologically normal, and displayed normal differentiation markers (supplementary material, Figure S5). However, the relative pancreas weight was significantly smaller in the ΔDG mice, revealing a subtle perturbation of normal tissue growth or homeostasis that remains to be elucidated (supplementary material, Figure S5B).

Disease progression was then compared between KC mice and KC-ΔDG mice. Pancreatic tissues were isolated from KC and KC-ΔDG mice at 2 and 5 months of age and assessed for altered disease progression by whole pancreas H&E staining and immunostaining. Tissue changes were first quantified by immunostaining of whole pancreas tissue sections for amylase and ductal cytokeratin expression, marking normal acini and transformed cells (ADM and dysplasia), respectively. At 2 months of age, there was already a clear divergence in disease progression, with the KC-ΔDG mice exhibiting accelerated loss of normal acinar tissue and enhanced ADM and dysplasia development relative to the KC mice (Figure 5A). These differences were perpetuated and significant also in cohorts at 5 months of age (Figure 5B).

Figure 5.

Figure 5.

Open in a new tab

Deletion of DG expression accelerates pancreatic dysplasia development. (A) H&E stain and immunofluorescence stain are shown for amylase (AMY) and ductal cytokeratin (CK), and for nuclei (DAPI) in two-month-old KC and KC-ΔDG mice. Quantification of the relative amylase and pan-CK staining (% of total pancreatic tissue area) is shown at right for both cohorts (n=6 and 6, respectively). Scale bars, 2 mm. (B) H&E stain and immunofluorescence stain is shown for amylase (AMY) and ductal cytokeratin (CK), and for nuclei (DAPI) in five-month-old KC and KC-ΔDG mice. Quantification of the relative amylase and pan-CK staining (% of total pancreatic tissue area) is shown for both cohorts (n=7 and 6, respectively). Scale bars, 2 mm. (C) Representative images generated by the VISTA tool for quantification of pancreatic tissue pathology, segmenting normal acinar (light blue), ADM (red), dysplasia (yellow), and other pancreatic tissue regions (dark blue). Scale bar, 2 mm. (D) Quantification of tissue pathology in cohorts of two-month-old KC and KC-ΔDG mice using the VISTA tool, calculating acinar cell area, ADM, and dysplasia in KC and KC-ΔDG mice (n=6 and 6, respectively). (E) Quantification of tissue pathology in cohorts of five-month-old KC and KC-ΔDG mice using the VISTA tool, calculating acinar cell area, ADM, and dysplasia in KC and KC-ΔDG mice (n=7 and 6, respectively).

To resolve and specifically quantify dysplasia development, we employed a recently developed artificial intelligence tool, called VISTA, which can reliably and objectively quantify the major histologic features of pancreatic pre-cancers (normal acini, ADM, and dysplasia), based on computational semantic segmentation of H&E stained whole pancreatic tissue sections [29]. The prevalence of normal acini, ADM, and dysplasia were each quantified in the tissue sets of 2- and 5-month old mice using the VISTA tool. These quantifications too showed that the presence of normal acini was significantly reduced in the KC-ΔDG mice at both 2 and 5 months of age, and further showed that ADM and dysplasia were both significantly enhanced in the KC-ΔDG mice, relative to KC mice (Figure 5C–E).

The accelerated development of dysplasia observed in the KC-ΔDG mice at 2 and 5 months of age could result in accelerated PDAC development. Disease and survival were assessed in a set of 44 KC and KC-ΔDG mice (n=21 and 23, respectively) up to one year of age. A statistically significant decrease in survival was observed for the KC-ΔDG, with several animals succumbing to PDAC prior to one year of age (Figure 6A). An enhanced development of cystic lesions that was observed in the 2 and 5 month old mice KC-ΔDG (Figure 5A,B) was also evident in the 1 year old cohort, with larger cystic lesions (> 500 μm in diameter) being significantly more prevalent in the KC-ΔDG mice than KC mice at this age (Figure 6B–D, and supplementary material, Figure S6). Animals surviving to one year of age were further assessed by necropsy for the development of PDAC, which is uncommon in the KC mice at this age [25]. Tissue assessment by a pathologist unaware of the experimental group revealed a trend towards accelerated PDAC development in the KC-ΔDG mice, concluding that 30% (7 of 23) of the KC-ΔDG mice displayed either focal or advanced PDAC at or before one year of age, compared to 14% (3 of 21) among KC mice (Figure 6E,F). Liver metastases were also observed in 9% (2 of 23) of the KC-ΔDG mice, but in none of the KC mice (Figure 6E,G). Although the incidence of PDAC and metastases were both higher in KC-ΔDG mice than KC mice, these differences did not reach statistical significance within the number of animals evaluated (p=0.202 and p= 0.167 respectively).

Figure 6.

Figure 6.

Open in a new tab

Deletion of DG expression causes reduced survival and trends towards increased PDAC evolution (A) Survival plots for KC and KC-ΔDG mice up to one year of age (n=32 and 26 respectively). (B) A plot of large cystic lesions (> 500 μm diameter) counted per tissue section in one-year-old KC and KC-ΔDG mice (n= 21 and 17, respectively). (C) Representative H&E stain showing the pancreatic lesions in one-year-old KC mice. Scale bar, 2 mm. (D) Representative H&E stain showing the pancreatic lesions in one-year-old KC-ΔDG mice. Scale bar, 2 mm. (E) Graphs of PDAC incidence and liver metastasis incidence at or before one year of age in KC and KC-ΔDG mice (n= 21 and 23, respectively) where black designates the detection of either PDAC or liver metastases. (F and G) Representative H&E stain of PDAC (F) and liver metastasis (G) in one-year-old KC-ΔDG mice. Scale bars, 200 μm.

Acinar-specific deletion of DG in adult animals accelerates dysplasia formation.

The KC mouse model employed in the experiments above relied on DNA recombination induced by the p48-Cre transgene, which is first expressed in the pancreas at E9.5 of embryonic development and is expressed in both the acinar and ductal lineage at this stage, as well as in pancreatic islets [25]. Consequently, the use of this model prompts questions about possible developmental stage-specific and lineage-specific effects of the genetic perturbation. To address these questions we employed the tamoxifen-inducible p48-CreER transgene which allows the conditional Cre transgene activation in the adult via intraperitoneal tamoxifen injection, thereby avoiding developmental influences [27]. An additional merit of this inducible system is that expression of the p48-CreER transgene is confined to the acinar cells in adult mice, allowing assessment of disease progression originating uniquely from the acinar compartment, where DG is expressed most prominently [27].

KCER mice (p48-CreER; LSL-KrasG12D) and KCER-ΔDG mice were generated and Cre-mediated DNA recombination was induced by tamoxifen treatment in adult mice between 7 and 8 weeks of age. Disease assessment was performed in H&E-stained whole pancreas tissue sections at 3 and 6 months post tamoxifen treatment using the VISTA tool to quantify the presence of normal acini, ADM and dysplasia development. This quantification revealed a significant acceleration of dysplasia development at both 3 and 6 months post-treatment in the KCER-ΔDG mice (supplementary material, Figure S5). At 6 months post tamoxifen treatment a significant decrease in normal acinar tissue was measured along with a significant increase in both ADM and dysplasia development (supplementary material, Figure S7C).

In summary, the deletion of the DG gene in the KC mouse models shows a clear acceleration of pancreatic disease, including an acceleration of developing ADM and dysplasia, decreased overall survival, and a trend toward increased rates of PDAC and metastasis. The deletion of the DG gene in the adult- and acinar-specific KCER model produced a similar acceleration of ADM and dysplasia development.

Discussion

The work presented here demonstrates that the ECM receptor DG functions as a tumour suppressor in the pancreas during pre-cancer evolution. The conditional knockout of DG in the pancreas produced limited evidence of structural or functional changes, and this is consistent with knockouts of DG in other epithelial compartments, including the kidney, which produced no phenotype, and the mammary gland, where defects were primarily associated with rapid tissue outgrowth and function during lactation [34,35]. However, in the context of oncogenic Kras expression, knockout of DG in the pancreas led to a clear acceleration of pancreatic pre-cancer evolution. In addition, data presented here reveal that DG protein expression is dynamically regulated in early disease evolution, being strongly suppressed with ADM at the earliest stages of oncogenesis and with chronic pancreatitis. Combined, these findings have important implications for understanding pancreatic disease aetiology, microenvironmental controls of cancer evolution, and disease risk factors.

The functional interrogations within this study were achieved by genetic perturbation within well-established autochthonous models of pancreatic pre-cancer and cancer evolution. These models exhibit the pre-cancer phase in a slow, more than one-year-long, progression to invasive disease and replicate many features of human disease pathology [25,26]. Conditional deletion of DG from the epithelial compartment within these models clearly accelerated dysplasia development. Accelerated dysplasia development was evident in multiple animal cohorts at various ages and with two distinct Cre transgene drivers; acceleration was evident in models relying on the p48-Cre transgene (expressed in all pancreatic cells) and in models relying on the inducible p48-CreER which specifically targets the acinar cell compartment, where DG expression is the highest. Accelerated disease progression was also evident by a significant reduction in survival by one year of age, coincident with an increase in PDAC development. Prior in vivo studies of DG function have relied solely on xenografts of manipulated breast and prostate cancer cells [12–15,24]. These xenograft studies showed that elevated DG function can slow the growth of invasive tumour cells following implantation. Our data, combined with these prior studies, indicate that DG can suppress tumour development and growth in both early and later stage disease, and in carcinomas of diverse tissue origin.

The mechanisms by which DG suppresses dysplasia development are as yet unclear, and obscured by DG’s remarkably diverse functions. DG is a multifaceted molecule involved not only in direct signalling functions, but also mediation of complex molecular assemblies at its cytoplasmic, transmembrane, and extracellular domains, and the consequent linkages between the ECM and cytoskeleton [16,36]. DG has been implicated in functions ranging from the regulation of Notch and YAP signalling [37–39] to epithelial polarization and STAT signalling [13,35,40], to filopodia formation [41,42]. DG has been shown to initiate cell surface binding and assembly of laminins in diverse cell types, and to mediate laminin endocytosis, with implications for the overall assembly and organization of the BM [14,23,35,40,43–45]. Intriguingly, DG’s roles as a linkage between the BM and actin cytoskeleton have been implicated in a “molecular corset” model in Drosophila that regulates epithelial growth and morphogenesis by mechanical constraint [45,46].

DG’s extracellular and cytoplasmic domains can function in concert or independently. The β-DG subunit is observed to translocate to the nucleus where it has been implicated in nuclear architecture and function [47–50]. In mammary epithelial cells, DG’s laminin assembly function was associated with signalling for epithelial polarization and STAT5 pathway activation, yet, unexpectedly, all of these functions were shown to operate independently of the DG cytoplasmic domain, suggesting that DG can function as a co-receptor for microenvironment signalling through modification of ECM assembly [35,40]. In the light of these data on ECM assembly functions, cytoplasmic domain functions, and the molecular corset model, it will be important to determine whether DG’s function in the suppression of dysplasia development is dependent on its cytoplasmic domain functions, ECM binding functions, or both.

Additional clues to DG’s mechanism of action may be gleaned from the subtle phenotype in the ΔDG mice and the notably cystic phenotype of the KC-ΔDG mice. The smaller relative weight of the pancreas in the ΔDG mice demonstrates tissue perturbations that warrant detailed analyses, interrogating altered signalling networks and possible compensatory mechanisms that adapt to the loss of DG. The cystic phenotype in the KC-ΔDG mice is the most distinctive feature exhibited in the pancreatic DG knockouts. Similar cystic phenotypes can be observed in other gene knockout studies and offer potentially related pathways to explore [51,52]. Notable among these is the knockout of Nr5a2, encoding a transcription factor that regulates development and differentiation, and complexes with pancreas transcription factor 1A [51].

The pre-cancerous microenvironment is a strong determinant of cancer risk [2], and the tumour suppressive function of DG identified here points to DG as a potentially important factor. Having established DG as a tumour suppressor in pancreatic pre-cancers, it is significant to also discover that DG expression is suppressed at the earliest stages of pancreatic pre-cancer evolution, and is suppressed with chronic pancreatitis, a known risk factor for pancreatic cancer development. This suppression of DG is replicated across pancreatic lesions in distinct pre-cancer subtypes in humans (e.g. IPMN and PanINs) and replicated in oncogenic Kras-driven animal models of pancreatic pre-cancer. Our study is the first to detail DG protein expression in early pancreatic disease and just two other studies have approached this topic in other cancers; one study in oral cancers showed that α-DG detection was lost in the transition from oral carcinoma in situ to invasive squamous cell carcinoma while another study included evidence for loss of α-DG immunostaining in a prostatic intraepithelial neoplasia [24,53]. More work is warranted to define the patterns of DG regulation and function in pre-cancerous lesions of diverse origin, especially in conditions of chronic inflammation associated with increased cancer risk. This path may help to identify novel indicators of disease prognosis. Another important question to address is whether augmentation of DG function can suppress disease progression, which could have therapeutic implications. A host of cancer-promoting factors are implicated in the inflammatory microenvironment, mostly originating from the stromal compartment [2]. Evidence presented here demonstrates that dynamic signalling changes also arise at the cell-BM interface with chronic inflammation, and these changes too can be highly consequential.

Supplementary Material

Supinfo

Figure S1. DG protein expression in normal human pancreatic tissue

Figure S2. Suppression of DG protein expression is evident in a human intraductal papillary mucinous neoplasm

Figure S3. DG protein expression patterns in the normal murine pancreas tissue mirror that of normal human pancreatic tissues

Figure S4. Single cell RNA-seq analysis shows DAG1 mRNA expression is low and unchanged across human pancreatic epithelial and malignant cell types

Figure S5. Deletion of DG expression alone had no apparent impact on murine pancreas development, architecture, or function, but showed a significant reduction in the relative pancreas weight.

Figure S6. Details of the cystic histology of the KC-ΔDG pancreas at 1 year of age

Figure S7. Acinar-specific deletion of DG expression in adult mice accelerates pancreatic neoplasia development

Acknowledgments

The authors wish to thank the OHSU Histopathology Shared Resource for assistance with tissue preparations, the OHSU Knight BioLibrary and the Oregon Pancreas Tissue Registry for tissue access, the OHSU Advanced Light Microscopy Core for imaging assistance, Dr Kevin Campbell (University of Iowa) for making available the floxed dystroglycan mice, and Dr Heidi Feiler and Dr Ellen Langer (OHSU) for critical reading of the manuscript. This work was supported by funding from the Brenden Colson Center for Pancreatic Care at OHSU (to JLM, RCS and YHC), by NIH grants U54CA209988 and 1U01 CA224012 (to Y.H.C and R.C.S.), and the OHSU Knight Cancer Institute NCI Cancer Center Support Grant P30CA069533.

Footnotes

No conflicts of interest were declared

Data availability statement

Data supporting the findings of this study are available within the article and its supplementary material files. Other information or data related to this paper can be obtained through contacting the corresponding author.

References

  • 1.Bissell MJ, Hines WC. Why don’t we get more cancer? A proposed role of the microenvironment in restraining cancer progression. Nat Med 2011; 17: 320–329. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2.Shalapour S, Karin M. Immunity, inflammation, and cancer: an eternal fight between good and evil. J Clin Invest 2015; 125: 3347–3355. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Storz P Acinar cell plasticity and development of pancreatic ductal adenocarcinoma. Nat Rev Gastroenterol Hepatol 2017; 14: 296–304. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Coussens LM, Werb Z. Inflammation and cancer. Nature 2002; 420: 860–867. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Sekiguchi R, Yamada KM. Basement Membranes in Development and Disease. Curr Top Dev Biol 2018; 130: 143–191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Kelley LC, Lohmer LL, Hagedorn EJ, et al. Traversing the basement membrane in vivo: a diversity of strategies. J Cell Biol 2014; 204: 291–302. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Hamidi H, Ivaska J. Every step of the way: integrins in cancer progression and metastasis. Nat Rev Cancer 2018; 18: 533–548. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Ramovs V, Te Molder L, Sonnenberg A. The opposing roles of laminin-binding integrins in cancer. Matrix Biol 2017; 57–58: 213–243. [DOI] [PubMed] [Google Scholar]
  • 9.Jiang X, Rieder S, Giese NA, et al. Reduced alpha-dystroglycan expression correlates with shortened patient survival in pancreatic cancer. J Surg Res 2011; 171: 120–126. [DOI] [PubMed] [Google Scholar]
  • 10.Sgambato A, Brancaccio A. The dystroglycan complex: from biology to cancer. J Cell Physiol 2005; 205: 163–169. [DOI] [PubMed] [Google Scholar]
  • 11.Esser AK, Miller MR, Huang Q, et al. Loss of LARGE2 disrupts functional glycosylation of alpha-dystroglycan in prostate cancer. J Biol Chem 2013; 288: 2132–2142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12.Sgambato A, Camerini A, Faraglia B, et al. Increased expression of dystroglycan inhibits the growth and tumorigenicity of human mammary epithelial cells. Cancer Biol Ther 2004; 3: 967–975. [DOI] [PubMed] [Google Scholar]
  • 13.Muschler J, Levy D, Boudreau R, et al. A Role for Dystroglycan in Epithelial Polarization: Loss of Function in Breast Tumor Cells. Cancer Res 2002; 62: 7102–7109. [PubMed] [Google Scholar]
  • 14.Akhavan A, Griffith OL, Soroceanu L, et al. Loss of cell-surface laminin anchoring promotes tumor growth and is associated with poor clinical outcomes. Cancer Res 2012; 72: 2578–2588. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Bao X, Kobayashi M, Hatakeyama S, et al. Tumor suppressor function of laminin-binding alpha-dystroglycan requires a distinct beta3-N-acetylglucosaminyltransferase. Proc Natl Acad Sci U S A 2009; 106: 12109–12114. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Quereda C, Pastor A, Martin-Nieto J. Involvement of abnormal dystroglycan expression and matriglycan levels in cancer pathogenesis. Cancer Cell Int 2022; 22: 395. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Akhavan A, Crivelli SN, Singh M, et al. SEA domain proteolysis determines the functional composition of dystroglycan. FASEB J 2008; 22: 612–621. [DOI] [PubMed] [Google Scholar]
  • 18.Kanagawa M, Saito F, Kunz S, et al. Molecular recognition by LARGE is essential for expression of functional dystroglycan. Cell 2004; 117: 953–964. [DOI] [PubMed] [Google Scholar]
  • 19.Mercuri E, Muntoni F. The ever-expanding spectrum of congenital muscular dystrophies. Ann Neurol 2012; 72: 9–17. [DOI] [PubMed] [Google Scholar]
  • 20.Henry MD, Cohen MB, Campbell KP. Reduced expression of dystroglycan in breast and prostate cancer. Hum Pathol 2001; 32: 791–795. [DOI] [PubMed] [Google Scholar]
  • 21.Singh J, Itahana Y, Knight-Krajewski S, et al. Proteolytic enzymes and altered glycosylation modulate dystroglycan function in carcinoma cells. Cancer Res 2004; 64: 6152–6159. [DOI] [PubMed] [Google Scholar]
  • 22.Beltran-Valero de Bernabe D, Inamori KI, Yoshida-Moriguchi T, et al. Loss of alpha-dystroglycan laminin binding in epithelium-derived cancers is caused by silencing of large. J Biol Chem 2009; 284: 11279–11284. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Leonoudakis D, Huang G, Akhavan A, et al. Endocytic trafficking of laminin is controlled by dystroglycan and disrupted in cancers. J Cell Sci 2014; 127: 4894–4903. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Sgambato A, De Paola B, Migaldi M, et al. Dystroglycan expression is reduced during prostate tumorigenesis and is regulated by androgens in prostate cancer cells. J Cell Physiol 2007; 213: 528–539. [DOI] [PubMed] [Google Scholar]
  • 25.Hingorani SR, Petricoin EF, Maitra A, et al. Preinvasive and invasive ductal pancreatic cancer and its early detection in the mouse. Cancer Cell 2003; 4: 437–450. [DOI] [PubMed] [Google Scholar]
  • 26.Hingorani SR, Wang L, Multani AS, et al. Trp53R172H and KrasG12D cooperate to promote chromosomal instability and widely metastatic pancreatic ductal adenocarcinoma in mice. Cancer Cell 2005; 7: 469–483. [DOI] [PubMed] [Google Scholar]
  • 27.Kopinke D, Brailsford M, Pan FC, et al. Ongoing Notch signaling maintains phenotypic fidelity in the adult exocrine pancreas. Dev Biol 2012; 362: 57–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Cohn RD, Henry MD, Michele DE, et al. Disruption of DAG1 in differentiated skeletal muscle reveals a role for dystroglycan in muscle regeneration. Cell 2002; 110: 639–648. [DOI] [PubMed] [Google Scholar]
  • 29.Ternes L, Huang G, Lanciault C, et al. VISTA: VIsual Semantic Tissue Analysis for pancreatic disease quantification in murine cohorts. Sci Rep 2020; 10: 20904. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 30.Oh K, Yoo YJ, Torre-Healy LA, et al. Coordinated single-cell tumor microenvironment dynamics reinforce pancreatic cancer subtype. Nat Commun 2023; 14: 5226. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 31.Li Z, Tremmel DM, Ma F, et al. Proteome-wide and matrisome-specific alterations during human pancreas development and maturation. Nat Commun 2021; 12: 1020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Wong CH, Li YJ, Chen YC. Therapeutic potential of targeting acinar cell reprogramming in pancreatic cancer. World J Gastroenterol 2016; 22: 7046–7057. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 33.Zhu L, Shi G, Schmidt CM, et al. Acinar cells contribute to the molecular heterogeneity of pancreatic intraepithelial neoplasia. Am J Pathol 2007; 171: 263–273. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Jarad G, Pippin JW, Shankland SJ, et al. Dystroglycan does not contribute significantly to kidney development or function, in health or after injury. Am J Physiol Renal Physiol 2011; 300: F811–820. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Leonoudakis D, Singh M, Mohajer R, et al. Dystroglycan controls signaling of multiple hormones through modulation of STAT5 activity. J Cell Sci 2010; 123: 3683–3692. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Moore CJ, Winder SJ. Dystroglycan versatility in cell adhesion: a tale of multiple motifs. Cell Commun Signal 2010; 8: 3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.McClenahan FK, Sharma H, Shan X, et al. Dystroglycan Suppresses Notch to Regulate Stem Cell Niche Structure and Function in the Developing Postnatal Subventricular Zone. Dev Cell 2016; 38: 548–566. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Sirour C, Hidalgo M, Bello V, et al. Dystroglycan is involved in skin morphogenesis downstream of the Notch signaling pathway. Mol Biol Cell 2011; 22: 2957–2969. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 39.Morikawa Y, Heallen T, Leach J, et al. Dystrophin-glycoprotein complex sequesters Yap to inhibit cardiomyocyte proliferation. Nature 2017; 547: 227–231. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Weir ML, Oppizzi ML, Henry MD, et al. Dystroglycan loss disrupts polarity and {beta}-casein induction in mammary epithelial cells by perturbing laminin anchoring. J Cell Sci 2006; 119: 4047–4058. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Eyermann C, Czaplinski K, Colognato H. Dystroglycan promotes filopodial formation and process branching in differentiating oligodendroglia. J Neurochem 2012; 120: 928–947. [DOI] [PubMed] [Google Scholar]
  • 42.Batchelor CL, Higginson JR, Chen YJ, et al. Recruitment of Dbl by ezrin and dystroglycan drives membrane proximal Cdc42 activation and filopodia formation. Cell Cycle 2007; 6: 353–363. [DOI] [PubMed] [Google Scholar]
  • 43.Henry MD, Campbell KP. A role for dystroglycan in basement membrane assembly. Cell 1998; 95: 859–870. [DOI] [PubMed] [Google Scholar]
  • 44.Colognato H, Winkelmann DA, Yurchenco PD. Laminin polymerization induces a receptor-cytoskeleton network. J Cell Biol 1999; 145: 619–631. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Cerqueira Campos F, Dennis C, Alegot H, et al. Oriented basement membrane fibrils provide a memory for F-actin planar polarization via the Dystrophin-Dystroglycan complex during tissue elongation. Development 2020; 147. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Haigo SL, Bilder D. Global tissue revolutions in a morphogenetic movement controlling elongation. Science 2011; 331: 1071–1074. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 47.Oppizzi ML, Akhavan A, Singh M, et al. Nuclear Translocation of beta-Dystroglycan Reveals a Distinctive Trafficking Pattern of Autoproteolyzed Mucins. Traffic 2008; 9: 2063–2072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Azuara-Medina PM, Sandoval-Duarte AM, Morales-Lazaro SL, et al. The intracellular domain of beta-dystroglycan mediates the nucleolar stress response by suppressing UBF transcriptional activity. Cell Death Dis 2019; 10: 196. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Velez-Aguilera G, de Dios Gomez-Lopez J, Jimenez-Gutierrez GE, et al. Control of nuclear beta-dystroglycan content is crucial for the maintenance of nuclear envelope integrity and function. Biochim Biophys Acta Mol Cell Res 2018; 1865: 406–420. [DOI] [PubMed] [Google Scholar]
  • 50.Sciandra F, Desiderio C, Vincenzoni F, et al. Analysis of the GFP-labelled beta-dystroglycan interactome in HEK-293 transfected cells reveals novel intracellular networks. Biochem Biophys Res Commun 2024; 703: 149656. [DOI] [PubMed] [Google Scholar]
  • 51.Cobo I, Iglesias M, Flandez M, et al. Epithelial Nr5a2 heterozygosity cooperates with mutant Kras in the development of pancreatic cystic lesions. J Pathol 2021; 253: 174–185. [DOI] [PubMed] [Google Scholar]
  • 52.Izeradjene K, Combs C, Best M, et al. Kras(G12D) and Smad4/Dpc4 haploinsufficiency cooperate to induce mucinous cystic neoplasms and invasive adenocarcinoma of the pancreas. Cancer Cell 2007; 11: 229–243. [DOI] [PubMed] [Google Scholar]
  • 53.Ahsan MS, Yamazaki M, Maruyama S, et al. Differential expression of perlecan receptors, alpha-dystroglycan and integrin beta1, before and after invasion of oral squamous cell carcinoma. J Oral Pathol Med 2011; 40: 552–559. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supinfo

Figure S1. DG protein expression in normal human pancreatic tissue

Figure S2. Suppression of DG protein expression is evident in a human intraductal papillary mucinous neoplasm

Figure S3. DG protein expression patterns in the normal murine pancreas tissue mirror that of normal human pancreatic tissues

Figure S4. Single cell RNA-seq analysis shows DAG1 mRNA expression is low and unchanged across human pancreatic epithelial and malignant cell types

Figure S5. Deletion of DG expression alone had no apparent impact on murine pancreas development, architecture, or function, but showed a significant reduction in the relative pancreas weight.

Figure S6. Details of the cystic histology of the KC-ΔDG pancreas at 1 year of age

Figure S7. Acinar-specific deletion of DG expression in adult mice accelerates pancreatic neoplasia development

NIHMS2021822-supplement-Supinfo.docx (6MB, docx)

Data Availability Statement

Data supporting the findings of this study are available within the article and its supplementary material files. Other information or data related to this paper can be obtained through contacting the corresponding author.

RESOURCES