Purinergic GPCR-integrin interactions drive pancreatic cancer cell invasion
Pancreatic ductal adenocarcinoma (PDAC) continues to show no improvement in survival rates. One aspect of PDAC is elevated ATP levels, pointing to the purinergic axis as a potential attractive therapeutic target. Mediated in part by highly druggable extracellular proteins, this axis plays essential roles in fibrosis, inflammation response, and immune function. Analyzing the main members of the PDAC extracellular purinome using publicly available databases discerned which members may impact patient survival. P2RY2 presents as the purinergic gene with the strongest association with hypoxia, the highest cancer cell-specific expression, and the strongest impact on overall survival. Invasion assays using a 3D spheroid model revealed P2Y2 to be critical in facilitating invasion driven by extracellular ATP. Using genetic modification and pharmacological strategies, we demonstrate mechanistically that this ATP-driven invasion requires direct protein-protein interactions between P2Y2 and αV integrins. DNA-PAINT super-resolution fluorescence microscopy reveals that P2Y2 regulates the amount and distribution of integrin αV in the plasma membrane. Moreover, receptor-integrin interactions were required for effective downstream signaling, leading to cancer cell invasion. This work elucidates a novel GPCR-integrin interaction in cancer invasion, highlighting its potential for therapeutic targeting.
In this manuscript, the authors address an important and urgent question: what molecular mechanisms drive the invasive behavior of pancreatic adenocarcinoma? Because these tumors have such a strong propensity for invasion and metastasis, identifying actionable targets is of high importance. Using a combination of in silico and in vitro modeling, they identify a role for purinergic G-protein coupled receptor P2Y2 as a critical node in mediating PDAC invasion, and they find that blocking the crosstalk between P2Y2 and αV integrins via a peptide inhibitor blocks PDAC invasion, which may have clinical utility. Thus their study provides insights into both the basic biology of PDAC but also identifies a new target.https://doi.org/10.7554/eLife.86971.sa0
PDAC, which accounts for 90% of diagnosed pancreatic cancer cases, has the lowest survival rate of all common solid malignancies. Surgery is the only potentially curative treatment, yet more than 80% of patients present with unresectable tumors (Kocher, 2023). Consequently, most patients survive less than six months after diagnosis, resulting in a five year survival rate of less than 5% when accounting for all disease stages (Bengtsson et al., 2020; Kocher, 2023). Despite continued efforts, this statistic has improved minimally in the past 50 years. Due to increasing incidence, late detection, and lack of effective therapies, pancreatic cancer is predicted to be the second most common cause of cancer-related deaths by 2040 (Rahib et al., 2021).
Failure to significantly improve clinical management is mainly a result of chemoresistance (Neuzillet et al., 2017), thus it is of vital importance to find new therapeutics that can improve patient survival. PDAC is characterized by its desmoplastic stroma, with dense fibrosis leading to impaired vascularisation and high levels of hypoxia (Koong et al., 2000; Di Maggio, 2016). Lack of oxygen leads to cellular stress and death, resulting in the release of purines such as ATP and adenosine into the tumor microenvironment (Forrester and Williams, 1977; Pellegatti et al., 2008). Extracellular ATP concentration in PDAC is 200-fold more than in normal tissue (Hu et al., 2019), suggesting that purinergic signaling could represent an effective therapeutic target in pancreatic cancer.
The proteins underpinning purinergic signaling comprise several highly druggable membrane proteins involved in the regulation of extracellular purines, mainly ATP and adenosine (Burnstock and Novak, 2012; Boison and Yegutkin, 2019; Yu et al., 2021). Extracellular ATP is known to promote inflammation (Kurashima et al., 2012), growth (Ko et al., 2012), and cell movement (Martínez-Ramírez et al., 2016). Contrastingly, adenosine is anti-inflammatory and promotes immunosuppression (Schneider et al., 2021). There are ongoing clinical trials in several cancers, including PDAC, for drugs targeting the ectonucleotidase CD73 (NCT03454451, NCT03454451) and adenosine receptor 2 A (NCT03454451) in combination with PD-1 checkpoint inhibitors and/or chemotherapy. However, a Phase II multi-cancer study evaluating an anti-CD73 and anti-PD-L1 combination was withdrawn due to minimal overall clinical activity (NCT04262388). This suggests that the oncogenic impact of purinergic signaling may act via pathways other than immunosuppression and highlights the need for the further mechanistic understanding of purinergic signaling in PDAC to exploit its full therapeutic potential.
Here, we combine bioinformatic, genetic, and drug-based approaches to identify a novel mechanism mediating ATP-driven invasion, uncovering a new therapeutic target in PDAC, a cancer of unmet clinical need. Beginning with an in-depth in silico analysis of the purinergic signaling transcriptome in PDAC, using publicly available patient and cell line databases, we build on bioinformatic data associating the purinergic receptor P2Y2 with PDAC. After validating the expression of P2Y2 in human PDAC, we focus on identifying the function of the receptor in cancer cells. In vitro data underline the importance of P2Y2 as a strong invasive driver, using a 3D physio-mimetic model of invasion. Finally, using a super-resolution imaging technique, DNA-PAINT, we characterize the behavior of P2Y2 in the membrane at the single molecule level, demonstrating the nanoscale distribution and interaction of this receptor with RGD-binding integrins in promoting pancreatic cancer invasion.
The PDAC extracellular purinome associates with patient survival, hypoxia score and cell phenotype
The extracellular purinome encompasses 23 main surface proteins, including pannexin 1, P2X ion channels, ectonucleotidases, and the P2Y, and adenosine GPCRs (Di Virgilio et al., 2018; Figure 1A). Interrogating public databases, we determined which purinergic signaling genes significantly impact pancreatic cancer survival. First, we examined the pancreatic adenocarcinoma (PAAD) database from The Cancer Genome Atlas (TCGA; n=177 patients), analyzing overall survival hazard ratios based on purinergic signaling gene expression (Figure 1B). Expression of five purinergic genes correlated with decreased patient survival, with high P2RY2 expression being associated with the highest hazard ratio (2.99, 95%, CI: 1.69–5.31, log-rank p=8.5 × 10–5). We then examined the mutational profile and mRNA expression level of purinergic genes in patients. Using cBioPortal (Gao, 2013), we generated OncoPrints of purinergic signaling genes from PAAD TCGA samples (Figure 1—figure supplement 1A), observing few genetic alterations in 0–3% of tumors and a heterogeneous percentage of tumors with high mRNA expression (z-score >1) for each purinergic gene. PDAC molecular subtypes associated with purinergic signaling genes were varied (Supplementary file 1). In the Bailey model, most genes were related to the immunogenic subtype except for NT5E, ADORA2B, PANX1, and P2RY2, which are related to squamous (Bailey et al., 2016). Collisson molecular subtyping showed several purinergic genes associated mostly with quasimesenchymal and exocrine subtypes (Collisson et al., 2011). The Moffitt subtypes were not strongly associated with purinergic genes except for ADA, NT5E, P2RY6, P2RY2, and PANX1 associated with the Basal subtype (Moffitt et al., 2015).
PDAC is known for its hypoxic environment (Koong et al., 2000; Yuen and Díaz, 2014), which is associated with worse overall survival (p=0.002, Figure 1—figure supplement 1B); hypoxia can lead to cellular stress and death, resulting in the increase of extracellular purines (Forrester and Williams, 1977). The Winter (Winter et al., 2007), Ragnum (Ragnum et al., 2015), and Buffa (Buffa et al., 2010) hypoxia scores were used to examine the correlation between the expression of purinergic genes and hypoxia in the PAAD TCGA database (Figure 1—figure supplement 1C). Samples were divided into low (n=88) or high (n=89) hypoxia score, using the median hypoxia score to perform a differential expression analysis. CD73 (NT5E), adenosine A2B receptor (ADORA2B), and P2Y2 (P2RY2) mRNA expression associated strongly with the high hypoxia score group for all three hypoxia scores (log2 ratio >0.5, FDR < 0.0001). P2Y2 had the highest log2 ratio in all hypoxia signatures compared to other purinergic genes. With a more extensive gene signature, the Winter hypoxia score (99 genes) allowed for a more comprehensive relative hypoxia ranking of tumor samples, compared to Ragnum (32 genes) and Buffa (52 genes) signatures. Hence, we used cBioPortal (Gao, 2013) to generate a transcriptomic heatmap of purinergic genes, ranked using the Winter hypoxia score and overlaid with overall survival data (Figure 1C). Taken together, these results show a direct correlation between Winter hypoxia score and decreased overall survival for high hypoxia score-related purinergic genes.
We hypothesized that genes related to high hypoxia scores would be expressed preferentially in the tumor cell compartment, as PDAC cells inhibit angiogenesis, causing hypo-vascularisation in the juxta-tumoral stroma (Di Maggio, 2016). Mining published RNA-seq data from 60 paired PDAC samples of stroma and tumor microdissections (GSE93326) (Maurer et al., 2019) and performing differential expression analysis, we observed that most genes related to high Winter hypoxia scores (P2RY2, ADORA2B, and NT5E) were expressed in the tumor epithelial tissue (Figure 1D), except for PANX1, encoding for pannexin 1, which is involved in cellular ATP release (Bao et al., 2004).
To elucidate the cell type-specific purinergic expression landscape, we used published data from TCGA PAAD compartment deconvolution, using DECODER (Peng et al., 2019) to plot purinergic gene weights for each cell type compartment (Figure 1E). The findings recapitulated the cell specificity data obtained from tumor microdissection analysis (Maurer et al., 2019; Figure 1D). Expression of purinergic genes in cancer cells was confirmed by plotting Z-scores of mRNA expression of PDAC cell lines from the cancer cell line encyclopedia (Ghandi et al., 2019) (CCLE; Figure 1—figure supplement 1D). Moreover, the expression of purinergic genes in normal tissue from the Genotype-Tissue Expression (GTEx) database compared to cancer tissue (PAAD TCGA) also mimicked the results found with DECODER (Figure 1—figure supplement 1E). P2RY2, encoding P2Y2 - a GPCR activated by ATP and UTP, was shown to be the purinergic gene most highly associated with cancer cell-specific expression in all our independent analyses (Figure 1D and E; Figure 1—figure supplement 1D, E). P2RY2 additionally showed the strongest correlation with all hypoxia scores (Figure 1C; Figure 1—figure supplement 1C). Most importantly, of all purinergic genes, P2RY2 expression had the biggest adverse impact on patient survival (Figure 1B). These independent in silico analyses encouraged us to explore the influence of P2Y2 on pancreatic cancer cell behavior.
P2RY2 is expressed in cancer cells and causes cytoskeletal changes
To validate our bioinformatic findings, based on microdissections from a 60 patient cohort (GSE93326) and from the deconvolution of 177 PAAD tissues from the TCGA, we performed RNAscope on human PDAC samples. This corroborated P2Y2 mRNA expression as being localized to the epithelial tumor cell compartment and not stroma, normal epithelium, or endocrine tissues (n=3, representative images of 2 different patients shown in Figure 2A and Figure 2—figure supplement 1A), matching our findings from larger publicly available cohorts, including P2Y2 IHC data from 264 patients in the Renji cohort (Hu et al., 2019). P2Y2 is known to be expressed at low levels in normal tissues but interestingly RNAscope did not detect this. This data suggests (1) the lower limits of the technique compounded by the challenge of RNA degradation in pancreatic tissue and (2) supports that in tumor tissue where it was detected there was indeed overexpression of P2Y2, in line with the bioinformatic data. Interrogating single-cell P2Y2 RNA expression in normal PDAC from https://www.proteinatlas.org/ (Karlsson et al., 2021), expression was found at low levels in several cells types, for example in endocrine cells and macrophages (Figure 2—figure supplement 1B). Using GEPIA (Tang et al., 2017), we analyzed PAAD TCGA and GTEx mRNA expression of tumor (n=179) and normal samples (n=171). Tumor samples expressed significantly higher (p<0.0001) P2Y2 mRNA levels compared to the normal pancreas (Figure 2B). Kaplan-Meier analysis from PAAD TCGA KMplot (Lánczky and Győrffy, 2021) showed a significant decrease in median overall survival in patients with high P2Y2 mRNA expression (median survival: 67.87 vs 17.27 months) (Figure 2C).
To predict P2Y2 function in PDAC, we performed gene set enrichment analysis (GSEA) of high vs low mRNA expressing P2Y2 tumor samples, divided by the median expression, for PAAD TCGA (n=177) and the PDAC Clinical Proteomic Tumor Analysis Consortium (CPTAC) (n=140) databases. The top gene set enriched in the PANTHER pathway database in both cohorts was the ‘integrin signaling pathway’ (Figure 2D). The top four enriched gene sets from the Gene Ontology ‘Molecular function’ functional database were associated with cell adhesion molecule binding, the cytoskeleton, protease binding, and extracellular matrix binding (Figure 2—figure supplement 1C). As preliminary validation of the GSEA results in vitro, we used the PDAC cell line AsPC-1, transduced with Lifeact, a peptide that fluorescently labels filamentous actin structures (Riedl et al., 2008), and monitored cell morphology using the Incucyte live-cell analysis system. Cells treated with ATP (100 µM) showed cytoskeletal rearrangements which were blocked by the selective P2Y2 antagonist AR-C118925XX (AR-C; 5 µM; Figure 2E; Muoboghare et al., 2019). Exposing cells to ATP at 100 µM resulted in the biggest change in cell area when testing six concentrations from 0.01 to 1000 µM (Figure 2—figure supplement 1D). ATP-driven morphological changes were fully reversed at 5 X (5 µM) the IC50 of AR-C (1 µM), while AR-C on its own had no effect on cell morphology (Figure 2—figure supplement 1E).
P2Y2 is the only P2Y GPCR possessing an RGD motif, located in the first extracellular loop (Figure 2F). P2Y2 has been shown to interact with αV integrins through this RGD motif (Erb et al., 2001), but the significance of this interaction has not been explored in cancer. Immunofluorescence (IF) showed colocalization of integrin αV and P2Y2 in the PDAC cell lines AsPC-1 as well as PDAC cell lines with strong epithelial morphology, BxPC-3 and CAPAN-2, while MIA PaCa-2 cells showed low expression of both proteins, and PANC-1 showed high integrin αV and low P2Y2, matching CCLE data (Figure 2G; Figure 2—figure supplement 1F, G). We hypothesized that P2Y2, through its RGD motif, could engage αV integrins in cancer cells in the presence of ATP, leading to increased migration and invasion.
Targeting P2Y2 and its RGD motif decreases ATP-driven invasion in PDAC cell lines
To evaluate the impact of P2Y2 in pancreatic cancer cell invasion, we used a 3D hanging drop spheroid model (Murray et al., 2022). PDAC cell lines were combined with stellate cells in a ratio of 1:2 (Kadaba et al., 2013), using an immortalized stellate cell line, PS-1 (Froeling et al., 2009) to form spheres (Figure 3A), recapitulating the ratios of the two biggest cellular components in PDAC. Stellate cells are crucial for successful hanging drop sphere formation (Figure 3—figure supplement 1A) and cancer cell invasion (Murray et al., 2022). Spheres were embedded in a Collagen type I and Matrigel mix and cultured for 48 hr until imaging and fixing (Figure 3A). Given that extracellular ATP concentration in tumors is in the hundred micromolar range (Pellegatti et al., 2008), spheres were treated with P2Y2 agonists ATP and UTP (100 µM). Both nucleotides increased invasion of the PDAC cell line AsPC-1 significantly compared to vehicle control (p<0.0001 and p=0.0013, respectively), and this was blocked by the P2Y2 selective antagonist AR-C (5 µM, p=0.0237, and p=0.0133; Figure 3B and C; Figure 3—figure supplement 1B). Treating spheres with AR-C on its own did not show significant effects on invasion (Figure 3—figure supplement 1B). Importantly, a non-hydrolyzable ATP (ATPγS;100 µM) showed similar effects to ATP, implicating ATP and not its metabolites as the cause of the invasion (Figure 3—figure supplement 1C). Of note, IF staining of PS-1 cells showed negligible expression of P2Y2 (Figure 3—figure supplement 1D). To determine whether integrin association was necessary for ATP-driven invasion, we treated spheres with 10 µM cyclic RGDfV peptide (cRGDfV), which binds predominantly to αVβ3 to block integrin binding to RGD motifs (Kapp et al., 2017), such as that in P2Y2 (Ibuka et al., 2015). cRGDfV treatment reduced ATP-driven motility significantly, both in 3D spheroid invasion assays (p<0.0001) (Figure 3B and C) and in 2D Incucyte migration assays (Figure 3—figure supplement 1E, F) as did treatment with AR-C. To ensure that this behavior was not restricted to AsPC-1 cells, experiments were corroborated in the epithelial-like BxPC-3 cell line (Figure 3—figure supplement 1G, H; Tan et al., 1986).
To further verify that ATP-driven invasion was dependent on P2Y2, we silenced P2Y2 expression in AsPC-1 cells using siRNA (Figure 3D; Figure 3—figure supplement 1I), abrogating the invasive response to ATP (p<0.0001). P2Y2 involvement in this phenomenon was confirmed by generating a P2Y2 CRISPR-Cas9 AsPC-1 cell line (P2Y2CRISPR), which displayed a significant decrease in invasion compared to a control guide RNA CRISPR cell line (CTRCRISPR) in both ATP-treated (p<0.0001) and non-treated (p=0.0005) conditions (Figure 3F and E). Additionally, we tested the off-target effects of AR-C in AsPC-1 P2Y2CRISPR spheres and confirmed no significant difference in invasion compared to the control (Figure 3—figure supplement 1J). Together, these findings demonstrate that P2Y2 is essential for ATP-driven cancer cell invasion.
To determine the importance of the RGD motif of P2Y2 in ATP-driven invasion, we obtained a mutant P2Y2 construct, where the RGD motif was replaced by RGE (P2Y2RGE), which has less affinity for αV integrins (Erb et al., 2001). This mutant was transfected into AsPC-1 P2Y2CRISPR cells and compared to cells transfected with wild-type P2Y2 (P2Y2RGD; Figure 3—figure supplement 1K). Only spheres containing P2Y2RGD transfected cells demonstrated a rescue of the ATP-driven invasive phenotype (p<0.0001; Figure 3G and H), with P2Y2RGE spheres not responding to ATP treatment. To ensure this behavior was not influenced by off-target CRISPR effects, we repeated the experiment in PANC-1 cell line, which expresses very low levels of P2Y2, but high levels of integrin αV (Figure 2—figure supplement 1F, G). No ATP-driven invasion was observed in PANC-1 cells transfected with an empty vector (EV) or with P2Y2RGE (Figure 3I and J). Only when transfecting PANC-1 cells with P2Y2RGD was ATP-driven invasion observed (p<0.0001). These results demonstrate that the RGD motif of P2Y2 is required for ATP-driven cancer cell invasion.
DNA-PAINT reveals RGD-dependent changes in P2Y2 and integrin αV surface expression
To interrogate how P2Y2 interacts with αV integrins, we examined the nanoscale organization of P2Y2 and αV proteins under different treatment conditions using a multi-color quantitative super-resolution fluorescence imaging method, DNA-PAINT. DNA-PAINT is a single-molecule localization microscopy (SMLM) method based on the transient binding between two short single-stranded DNAs - the ‘imager’ and ‘docking’ strands. The imager strand is fluorescently labeled and freely diffusing in solution, whilst the docking strand is chemically coupled to antibodies targeting the protein of interest. For DNA-PAINT imaging of P2Y2 and integrin αV, proteins were labeled with primary antibodies chemically coupled to orthogonal docking sequences featuring a repetitive (ACC)n or (TCC)n motif, respectively (Figure 4A). The benefit of such sequences is to increase the frequency of binding events, which in turn allows the use of relatively low imager strand concentrations without compromising overall imaging times, whilst achieving a high signal-to-noise ratio and single-molecule localization precision (Strauss and Jungmann, 2020).
The repetitive binding of imager and docking DNA strands in DNA-PAINT causes the same protein to be detected multiple times with nearly identical coordinates, originating a cluster of single-molecule localization around the true position of the protein. In contrast to other SMLM approaches, it is possible to take advantage of the DNA-binding kinetics to stoichiometrically calculate the number of proteins detected in each cluster of single molecule localizations, via an approach known as qPAINT (Schnitzbauer et al., 2017). As exemplified in Figure 4B (and detailed in the methods section), qPAINT relies on the first-order binding kinetics between the individual imager and docking strands to determine the number of copies of a protein that reside within a cluster of single-molecule localizations. The qPAINT index histograms obtained from P2Y2 and αV DNA-PAINT data sets were fitted with a multi-peak Gaussian function, identifying peaks located at multiples of a qPAINT index value of 0.011 Hz and 0.009 for the P2Y2 and αV docking-imager pairs, respectively (Figure 4C). These values were thus used to quantify the exact number of P2Y2 and αV proteins in all the clusters of single-molecule localization in the DNA-PAINT data sets. By combining qPAINT with spatial statistics, we recovered a good estimation of the ground truth position of all the proteins in the DNA-PAINT data and quantified protein clustering.
We have previously analyzed GPCR oligomerization quantitatively using DNA-PAINT super-resolution microscopy of P2Y2 in AsPC-1 cells (Joseph et al., 2021), where we observed a decrease in P2Y2 oligomerization upon AR-C treatment. Hence, we questioned whether the RGD motif in P2Y2 affected receptor distribution and clustering. We imaged AsPC-1 P2Y2CRISPR cells transfected with P2Y2RGD or P2Y2RGE in the absence or presence of 100 µM ATP for 1 hr (Figure 4D), observing a 42% decrease in the median density of P2Y2 proteins at the membrane when P2Y2RGD cells were treated with ATP, compared to control (p<0.0001; Figure 4E). In contrast, although a slight decrease in the density of P2Y2 proteins on P2Y2RGE cells was observed following ATP treatment, this was not statistically significant (p=0.1570). The density of P2Y2 proteins and protein clusters in both P2Y2RGD and P2Y2RGE controls were equivalent (Figure 4E and F; p>0.9999), indicating similar expression of the receptor at the surface in both control conditions. Interestingly, the density of P2Y2 clusters decreased significantly in both conditions when treated with ATP (Figure 4F; 43% decrease, p<0.0001 for P2Y2RGD, and 48% decrease, p=0.0002 for P2Y2RGE). We repeated these studies with normal AsPC-1 cells (untransfected and with unaltered P2Y2 expression) treated with ATP +/-cRGDfV, only observing a reduction of P2Y2 at the membrane with ATP alone (68% decrease, p<0.0001), while co-treatment with cRGDfV prevented this change (p>0.9999; Figure 4—figure supplement 1A, B). These findings highlight that the RGD motif is required for αV integrin to control P2Y2 levels at the membrane.
Turning to αV integrins, we observed an increase in the density of αV molecules and αV clusters at the membrane when stimulating P2Y2RGD with ATP (165 αV molecules/ROI, IQR = 162.75; 6.5 αV clusters/ROI, IQR = 8.75) compared to P2Y2RGD without stimulation (58 αV molecules/ROI, IQR = 41; 2.5 αV clusters/ROI, IQR = 2; p=0.0003; Figure 4G and H). This phenomenon was also observed with normal AsPC-1 cells, with significantly more αV molecules and clusters (p=0.0382 and p=0.0349) detected following ATP stimulation (Figure 4—figure supplement 1C, D). In absence of stimulation, P2Y2RGE transfected cells exhibited more αV molecules and clusters at the membrane (182 αV molecules/ROI, IQR = 262.75; 9 αV clusters/ROI IQR = 14) compared to P2Y2RGD cells (p=0.0003, p=0.0024, respectively). However, treating P2Y2RGE cells with ATP did not result in significant changes in αV molecules and clusters (p=0.7086; p=0.1846). When the number of clusters was normalized with the number of αV molecules, to obtain the percentage of αV in clusters (Figure 4—figure supplement 1E), there was no significant difference between conditions (p>0.9999), indicating that the increase in the number of αV clusters was due to an increase in the number of αV proteins at the membrane. Of note, the percentage of P2Y2 clusters significantly decreased in P2Y2RGE cells when treated with ATP compared to all other conditions (Figure 4—figure supplement 1F). Taken together, these data indicate an RGD motif-dependent function of activated P2Y2 in localizing integrin αV to the membrane.
Nearest neighbor distance (NND) was used to analyze homo and heterotypic protein-protein interactions between P2Y2 and αV. NND ranges were selected by using the approximate dimension of the antibodies (~14 nm) (Tan, 2008), integrins (5–10 nm) (Lepzelter et al., 2012), and GPCRs (~3 nm) (Figure 4—figure supplement 2A) and corroborating them with the NND histograms (Figure 4—figure supplement 2B) to predict the NND range in nm indicating a protein-protein interaction. We detected a higher percentage of integrin αV proteins in <50 nm proximity to P2Y2 in P2Y2RGD cells following ATP stimulation (Figure 4I; 103% increase, p=0.0143). In contrast, P2Y2RGE cells stimulated with ATP showed a 43% decrease (p=0.0101) in αV molecules in close proximity to P2Y2 in comparison to unstimulated cells. Analyzing the percentage of αV proteins with NND in the 20–100 nm range, we saw a similar pattern (Figure 4J). ATP-stimulated P2Y2RGD and unstimulated P2Y2RGE cells showed an increased percentage of αV proteins spaced at this range compared to untreated P2Y2RGD cells (98% increase with p=0.0132 and 89% increase with p=0.0181). No significant changes were observed in NND of <20 nm between αV proteins in any of the conditions (Figure 4K). In contrast, P2Y2RGD molecules were in significantly closer proximity to each other compared to P2Y2RGE in control and stimulated conditions (p<0.0001 and p=0.007) (Figure 4L). In summary, our SMLM studies demonstrate a reciprocal interaction between αV integrin and P2Y2 receptors, where P2Y2 can alter integrin localization to the plasma membrane while αV integrins influence activated P2Y2 membrane localization.
The RGD motif in P2Y2 is involved in integrin signaling
There is growing evidence of the importance of endosomal GPCR signaling and its potential relevance in disease and therapeutic opportunities (Calebiro and Godbole, 2018). As we identified the RGD motif in P2Y2 having a possible role in receptor internalization, integrin dynamics, and invasion, we proceeded to look at integrin signaling through phosphorylation of FAK (p-FAK) and ERK (p-ERK) from 0 to 1 hr after treating with 100 µM ATP. AsPC-1 cells displayed a significant increase of FAK and ERK phosphorylation after 15 min of ATP stimulation, which was abrogated by concomitant targeting of P2Y2 with AR-C (Figure 5A). When impairing the RGD motif function in P2Y2 with cRGDfV or by transfecting AsPC-1 P2Y2CRISPR cells with the P2Y2RGE mutant, p-FAK, and p-ERK levels decreased (Figure 5B and C). Collectively, targeting the RGD motif in P2Y2 impairs receptor signaling and inhibits pancreatic cancer cell invasion.
Improved molecular understanding of PDAC is vital to identify effective therapeutic approaches to improve patient survival. Purinergic signaling includes many druggable targets that have been related to hypoxia (Synnestvedt et al., 2002), immunosuppression (Fong et al., 2020), and invasion (Li et al., 2015), but have been relatively underexplored in PDAC. In this study, we used publicly available databases to identify purinergic signaling genes that could be promising targets for PDAC, determining P2Y2 as a driver of pancreatic cancer cell invasion. Extracellular ATP stimulated invasion in a 3D spheroid model of PDAC; an effect blocked by targeting P2Y2 genetically and pharmacologically. Mechanistically, we identified that the RGD motif in the first extracellular loop of P2Y2 is required for ATP-driven cancer invasion. Importantly, quantitative DNA-PAINT super-resolution fluorescence microscopy revealed the role of this RGD motif in orchestrating the number of P2Y2 and αV integrin proteins at the plasma membrane, upon ATP stimulation.
Purinergic signaling has been associated classically with hypoxia and immune function in cancer (Di Virgilio et al., 2018). One of the first reports of hypoxia-inducing ATP release in cells identified an increase of extracellular ATP in rat heart cells when kept in hypoxic conditions (Forrester and Williams, 1977). PDAC is a highly hypoxic cancer, with high levels of ATP reported in the tumor interstitial fluid of human and mouse PDAC tissues compared to healthy tissues (Hu et al., 2019). This vast release of ATP results in immune-mediated inflammatory responses via immune cells expressing purinergic signaling receptors (Chiarella et al., 2021). Expression of most purinergic genes was associated predominantly with immune cells, immunogenic PDAC subtype, and low hypoxia scores (Figure 1C and E). In contrast, expression of genes correlated with worse survival and hypoxia (PANX1, NT5E, ADORA2B, and P2RY2) was associated with tumor cells and the squamous PDAC subtype, correlating with hypoxia, inflammation, and worse prognosis (Bailey et al., 2016). The role of CD73 in PDAC has been examined in several studies (Yu et al., 2021) (NCT03454451, NCT03454451). In contrast, the adenosine A2B receptor has not been well studied. Adenosine A2B receptor requires larger agonist concentrations for activation compared to other receptors in the same family, such as adenosine A2A (Bruns et al., 1986; Xing et al., 2016), and receptor expression has been reported to increase when cells are subjected to hypoxia (Feoktistov et al., 2004). Moreover, HIF-1α has been shown to upregulate A2B and P2Y2 expression in liver cancer (Tak et al., 2016; Kwon et al., 2019). From our analyses, P2Y2 was associated with the worst patient overall survival, highest patient hypoxia scores, and strongest correlation with cancer cell expression compared to other purinergic genes. These observations were supported by published immunohistochemical staining of 264 human PDAC samples, showing that P2Y2 localized predominantly in cancer cells in human PDAC and that P2Y2 activation with ATP led to elevated HIF-1α expression (Hu et al., 2019). Hence, we decided here to explore P2Y2 in greater depth.
P2Y2 has been associated with cancer cell growth and glycolysis in PDAC (Ko et al., 2012; Hu et al., 2019; Wang et al., 2020). Combination treatment of subcutaneous xenografts of AsPC-1 or BxPC-3 cells with the P2Y2 antagonist AR-C together with gemcitabine significantly decreased tumor weight and resulted in increased survival compared to placebo or gemcitabine monotherapy control (Hu et al., 2019). Surprisingly, GSEA results of two different cohorts suggested a possible additional function of P2Y2 in invasion. Increased glycolysis and cytoskeletal rearrangements have been linked (Park et al., 2020), and both events could occur downstream of P2Y2 activation. P2Y2 has been implicated in invasive phenotypes in prostate, breast, and ovarian cancer (Jin et al., 2014; Li et al., 2015; Martínez-Ramírez et al., 2016). Moreover, high P2Y2 expression in patients was related to integrin signaling. The RGD motif in the first extracellular loop of P2Y2 results in a direct interaction of P2Y2 with RGD-binding integrins, particularly integrins αVβ3 and αVβ5 (Erb et al., 2001; Ibuka et al., 2015). This interaction can exert phenotypic effects – for example, the binding of P2Y2 to integrins via its RGD motif is necessary for tubule formation in epithelial intestinal cell line 3D models (Ibuka et al., 2015). We focus here on the importance of the RGD motif of P2Y2 and its key for integrin interaction in a cancer context. We were able to abrogate ATP-driven invasion using either the P2Y2 selective antagonist AR-C or by blocking P2Y2-integrin complexes using the selective αVβ3 cyclic RGD-mimetic peptide inhibitor cRGDfV. Likewise, spheres made using ASPC-1 P2Y2CRISPR or PANC-1 cells transfected with mutant P2Y2RGE, which decreases the affinity of P2Y2 for integrins, did not invade in response to ATP stimulation. Altogether, these results (1) support P2Y2 involvement in PDAC cell invasion, (2) show the RGD motif is essential for this function, and (3) identify the mechanism for this to be caused by P2Y2-integrin complexes. Despite efforts, there are currently no clinically efficacious P2Y2 antagonists, with poor oral bioavailability and low selectivity being major issues (Neumann et al., 2022). Our findings demonstrate that P2Y2 can also be targeted by blocking its interaction with RGD-binding integrins, due to its dependence on integrins for its pro-invasive function.
GPCR-integrin crosstalk is involved in many biological processes (Wang et al., 2005; Teoh et al., 2012). Only one study has directly examined the spatial distribution of integrins and GPCRs, however, this relied on IF analysis (Erb et al., 2001), where only changes in the micron scale will be perceived, hence losing information on the nanoscale distances and individual protein interactions. Here, we present a method to image integrin and GPCR dynamics using quantitative DNA-PAINT super-resolution fluorescence microscopy (Schnitzbauer et al., 2017), allowing spatial and quantitative assessment of P2Y2 and integrin αV interactions at the single protein level. Following ATP stimulation, the number of P2Y2 proteins at the plasma membrane decreased significantly after 1 hr, implying receptor internalization, in line with previous work showing P2Y2 at the cell surface was reduced significantly after 1 hr of UTP stimulation (Tulapurkar et al., 2005). Of note, cytoskeletal rearrangements, which we have also observed upon ATP stimulation (Figure 2E), were required for P2Y2 clathrin-mediated internalization, and authors noted that P2Y2 was most likely in a complex with integrins and extracellular matrix-binding proteins. Cells expressing RGE mutant P2Y2 or treated with cRGDfV, did not show significant changes in P2Y2 levels at the membrane upon ATP treatment, thus implicating the RGD motif in P2Y2 in agonist-dependent receptor internalization, though we have focused on motility phenotype in this work.
P2Y2 affecting cell surface redistribution of αV integrin has been reported, with αV integrin clusters observed after 5 min stimulation with UTP (Chorna et al., 2007). We observed an increased number of αV integrin molecules and clusters 1 hr after ATP stimulation, although this increase in clusters was mainly due to the increase in the total number of αV integrins at the membrane. The distance between αV integrin and P2Y2 molecules decreased (NND <50 nm) with ATP stimulation, indicating possible interaction. In contrast, with mutant P2Y2RGE, no significant ATP-dependent changes in the number of P2Y2 or αV integrin proteins at the membrane were observed. The same phenomenon was observed when treating normal AsPC-1 cells (untransfected and with no alteration to P2Y2) with cRGDfV and ATP. We speculate that by reducing the ability of integrins to bind to the RGD of P2Y2, through receptor internalization, RGE mutation or through cRGDfV treatment, there is less RGD-triggered integrin endocytosis, hence less integrin recycling and an increase of integrins at the cell surface. Western blot results supported our postulated role of the RGD motif in P2Y2 regulating downstream integrin signaling through FAK and ERK, leading to cancer cell migration and invasion (Figures 5 and 6). This is the first single-molecule super-resolution study to explore integrin and GPCR dynamics and to demonstrate a requirement for integrin-P2Y2 interactions in cancer cell invasion.
In summary, our study demonstrates that P2Y2, via its RGD motif, has a pivotal role in ATP-induced PDAC invasion by interacting with and regulating the number of αV integrins at the plasma membrane, revealing this critical axis as a promising therapeutic target.
Data mining and bioinformatic analysis
Hazard ratios and the P2Y2 Kaplan-Meier plot for overall survival were obtained using Kaplan-Meier Plotter (RRID:SCR_018753) (Lánczky and Győrffy, 2021) and the pancreatic adenocarcinoma dataset from the cancer genome atlas (PAAD TCGA, RRID:SCR_003193).
Using cBioPortal (RRID:SCR_014555) (Gao, 2013) and the database PAAD TCGA, mRNA differential expression analysis was performed for each Hypoxia Score (Winter et al., 2007; Buffa et al., 2010; Ragnum et al., 2015) by separating patients using the median hypoxia score. Results from purinergic genes were plotted in a volcano plot using VolcaNoseR (Goedhart and Luijsterburg, 2020). Significant hits were plotted in a heat map using cBioPortal (Gao, 2013). RNAseq raw counts from stromal and epithelial PDAC tissue from microdissections were downloaded from the GEO database (GSE93326) (Maurer et al., 2019) and a differential expression analysis was performed using DESeq2 (RRID:SCR_015687) (Love et al., 2014; Varet et al., 2016) in R.
Gene weight results from DECODER from PDAC tissues in the TCGA database were obtained from published results (Peng et al., 2019). Using GEPIA (RRID:SCR_018294) (Tang et al., 2017), mRNA expression of purinergic genes in normal tissue from the Genotype-Tissue Expression (GTEx, RRID:SCR_013042) compared to cancer tissue (PAAD TCGA) was obtained. PDAC cell line mRNA z-scores or mRNA reads per kilobase million (RPKM) were obtained using cBioPortal and the Cancer Cell Line Encyclopaedia (CCLE, RRID:SCR_013836) data (Gao, 2013).
For gene set enrichment analysis (GSEA), cBioPortal was used to separate PAAD TCGA or PDAC CPTAC patients into high and low P2RY2 by P2RY2 median expression and perform the differential expression analysis. Log ratio values were inserted in the WEB-based Gene SeT AnaLysis Toolkit (WebGestalt, RRID:SCR_006786) (Liao et al., 2019), where ‘GO: Molecular Function’ or ‘Panther’ with default analysis parameters were selected.
RNAscope in-situ hybridization
Formalin-fixed paraffin-mbedded (FFPE) sections (n=3) of PDAC with stroma and normal adjacent tissue were obtained from the Barts Pancreas Tissue Bank (Project 2021/02/QM/RG/E/FFPE). Sections were stained using the human P2RY2 probe (853761, ACD) and the RNAscope 2.5 HD Assay-RED (ACD) following the manufacturer’s instructions. Slides were imaged by NanoZoomer S210 slide scanner (Hamamatsu).
Cell lines and cell culture
The pancreatic cancer cell lines AsPC-1 (RRID:CVCL_0152), BxPC-3 (RRID:CVCL_0186), MIA PaCa-2 (RRID:CVCL_0428) and PANC-1 (RRID:CVCL_0480), in addition to the immortalized stellate cell line PS-1 (Froeling et al., 2009) were kindly donated by Prof. Hemant Kocher (Queen Mary University of London). Cell lines stably expressing fluorescently labeled histone subunits (H2B) or Lifeact (Riedl et al., 2008) were transduced with viral supernatant obtained from HEK293T cells co-transfected with pCMVR8.2 (Addgene #12263) and pMD2.G (Addgene #12259) packaging plasmids, and either H2B-GFP (Addgene #11680), H2B-RFP (Addgene #26001), or Lifeact-EGFP (Addgene # 84383) plasmids using FuGENE transfection reagent (Promega), following manufacturer’s guidelines. Successfully transduced cells were isolated using a BD FACS Aria Fusion cell sorter. AsPC-1 P2Y2CRISPR cells were generated by transfecting cells with a dual gRNA (TGAAGGGCCAGTGGTCGCCGCGG and CATCAGCGTGCACCGGTGTCTGG)CRISPR-CAS9 plasmid (VectorBuilder) with an mCherry marker which was used to select successfully transfected cells as above. Clonal expansion of single sorted cells was achieved with serial dilution cloning. Clones were evaluated by IF for P2Y2 compared to parental AsPC-1 cells. Cell lines were grown at 37 °C with 5% CO2 in DMEM (Gibco), RPMI-1640 (Gibco), or DMEM/F-12 (Sigma) supplemented with 10% fetal bovine serum (Sigma). Cells were monitored for mycoplasma contamination every six months.
Cell fixation and immunofluorescent staining
Cells were seeded on coverslips placed in a six well-plate (Corning) and fixed the next day in 4% paraformaldehyde (LifeTech) for 30 min and washed 3 x with phosphate-buffered saline (PBS). Coverslips were placed in 0.1% Triton X-100 (Avantor) for 10 min for permeabilization, followed by three PBS washes and blocking with 5% bovine serum albumin (BSA; Merck) for 1 hr. Coverslips were incubated at 4 °C overnight with anti-P2Y2 (APR-010, Alomone labs) and anti-integrin αV antibodies (P2W7, Santa Cruz) diluted in blocking solution (1:100 and 1:200, respectively). After three PBS washes, coverslips were incubated for 1 hr with Alexa Fluor 647 goat anti-mouse and Alexa Fluor 488 goat anti-rabbit (Invitrogen) or Alexa Fluor 546 goat anti-rabbit at 1:1000, diluted in blocking buffer. Following three PBS washes, 4’,6-diamidino-2-phenylindole (DAPI, Sigma-Aldrich) was used as a nuclear stain and was incubated for 10 min. Slides were mounted using Mowiol (Calbiochem) and imaged 24 hr later using a LSM 710 confocal microscope (Zeiss).
siRNA and plasmid transfection
Cells were seeded in six well plates at a density of 200,000 cells/well 24 hr before transfection. For siRNA experiments, cells were transfected with 20 nM pooled control or P2Y2-targeting siRNAs from a siGENOME SMARTpool (Dharmacon, GE Healthcare) with Lipofectamine 3000 (Invitrogen) following the manufacturer’s instructions. For P2Y2 plasmid expression experiments, cells were transfected with 500 nM P2RY2 (P2Y2RGD) or P2RY2D97E (P2Y2RGE) in pcDNA3.1 vector (Obtained from GenScript) or pcDNA3.1 alone (Empty vector, EV) together with lipofectamine 3000 and p3000 reagent (Invitrogen) as per manufacturer’s instructions. Plasmid concentration was selected by comparing AsPC-1 IF staining of P2Y2 with IF staining in AsPC-1 P2Y2CRISPR and PANC-1 cells with different concentrations of the plasmid to achieve a similar IF signal. Cells were split 48 hr post-transfection for experiments or imaged 72 hr post-transfection.
3D sphere model invasion assay
Spheres of PDAC cell lines with PS-1 cells were generated as described (Murray et al., 2022). Cancer cells at 22,000 cells/mL and PS-1 cells at 44,000 cells/mL were combined with DMEM/F-12 and 1.2% methylcellulose in a 4:1 ratio of methylcellulose (Sigma-Aldrich) and 20 µl drops, each containing 1000 cells, pipetted on the underside of a 15 cm dish lid (Corning) and hanging drops were incubated overnight at 37 °C. The next day, spheres were collected and centrifuged at 300 g for 4 min and washed with the medium. A mix of 2 mg/mL collagen (Corning), 175 µL/mL Matrigel, 25 µL/mL HEPES (1 M, pH 7.5), and 1 N NaOH (for neutral pH correction) was prepared with DMEM/F12 medium. Spheroids were re-suspended and seeded in low attachment 96-well plates (50 µl per well) with 40 µL previously gelled mix in the bottom of the wells. Once set, 150 µL of DMEM/F12 was added with treatments. Spheres were treated with 100 µM adenosine 5’-triphosphate trisodium salt hydrate (ATP, Sigma), uridine 5’-triphosphate trisodium salt hydrate (UTP, Sigma) or adenosine 5’-[γ-thio]triphosphate tetralithium salt (ATPγS, Tocris) alone or with 5 µM AR-C118925XX (AR-C, Tocris), or 10 µM cyclo(RGDfV) (cRGDfV, Sigma-Aldrich). Treatments were repeated 24 hr later. Spheres were imaged with a Zeiss Axiovert 135 light microscope at 10 x on day two after seeding. Cells were stained with 4’,6-diamidino-2-fenilindol (DAPI) (1:1000) for 10 min and imaged with a Zeiss LSM 710 confocal microscope. % Invasion was calculated by drawing an outline around the total area and central area of the spheres with ImageJ (Fiji) and using the equation:
Results were plotted in SuperPlots by assigning different colors to repeats and superimposing a graph of the average % Invasion with a darker shade of the assigned color as described previously (Lord et al., 2020).
IncuCyte migration assay
In IncuCyte ClearView 96-well cell migration plates (Essen BioScience), 40 μL medium with 5000 cells were seeded in each well. A solution of 20 μL medium with 15 µM AR-C or 30 µM cRGDfV was added on top of the wells to achieve a final concentration of 5 µM and 10 µM, respectively. Cells were allowed to settle for 15 min at room temperature and then placed at 37 °C for pre-incubation with the treatments for another 15 min. A volume of 200 μL of medium with or without 100 µM ATP was added in the appropriate reservoir wells and the plate was placed in the IncuCyte S3 (Essen BioScience) and was monitored every 4 hr for 39 hr (average doubling time of AsPC-1 cells Chen et al., 1982). Using the IncuCyte S3 2019 A software, the migration index was calculated by analyzing the average area occupied by the cells in the bottom well and was averaged with the initial average area occupied by cells in the top well.
RNA extraction and qPCR analysis
RNA was extracted using the Monarch RNA extraction kit (New England BioLabs) as instructed by the manufacturer. The extracted RNA was quantified using a Nanodrop One Spectrophotometer (ThermoFisher Scientific). Using LunaScript RT Supermix kit (BioLabs), cDNA was prepared in a 20 μL reaction according to the manufacturer’s instructions. The resulting cDNA was used in conjunction with MegaMix-Blue and P2RY2 primers (Eurogentec; Forward sequence: GCTACAGGTGCCGCTTCAAC, reverse sequence: AGACACAGCCAGGTGGAACAT) (Hu et al., 2019) for quantitative polymerase chain reaction (qPCR) at the manufacturer’s recommended settings in a StepOnePlus Real-Time PCR System (Applied Biosystems). The relative mRNA expression was calculated using the method (Livak and Schmittgen, 2001) and normalized to GAPDH.
DNA-antibody coupling reaction
DNA labeling of anti-αV antibody (P2W7, Santa Cruz, RRID:AB_627116) and anti-P2Y2 receptor antibody (APR-010, Alomone labs, RRID:AB_2040078) was performed via maleimidePEG2-succinimidyl ester coupling reaction as previously described (Simoncelli et al., 2020; Joseph et al., 2021). First, 30 µL of 250 mM DDT (Thermo Fisher Scientific) was added to 13 µL of 1 mM thiolated DNA sequences 5′-Thiol-AAACCACCACCACCA-3′ (Docking 1), and 5-Thiol-TTTCCTCCTCCTCCT-3’ (Docking 2) (Eurofins). The reduction reaction occurred under shaking conditions for 2 hrs. 30 min after the reduction of the thiol-DNA started, 175 µL of 0.8 mg/mL antibody solutions were incubated with 0.9 µL of 23.5 mM maleimide-PEG2-succinimidyl ester cross-linker solution (Sigma-Aldrich) on a shaker for 90 min at 4 °C in the dark. Prior to DNA-antibody conjugation, both sets of reactions were purified using Microspin Illustra G-25 columns (GE Healthcare) and Zeba spin desalting columns (7 K MWCO, Thermo Fisher Scientific), respectively, to remove excess reactants. Next, coupling of anti-P2Y2 with DNA docking 1 and anti-αV with DNA Docking 2 was performed by mixing the respective flow-through of the columns and incubate them overnight, in the dark, at 4 °C under shaking. Excess DNA was removed via Amicon spin filtration (100 K, Merck) and antibody-DNA concentration was measured using a NanoDrop One spectrophotometer (Thermo Fisher Scientific) and adjusted to 10 µM with PBS. Likewise, spectrophotometric analysis was performed to quantify the DNA-antibody coupling ratio and found to be ∼1.2 on average for both the oligo-coupled primary antibodies.
Cell fixation and immunofluorescence staining for DNA-PAINT imaging
Cells were seeded at 30,000 cells per channel on a six-channel glass-bottomed microscopy chamber (μ-SlideVI0.5, Ibidi) pre-coated with rat tail collagen type I (Corning). The chamber was incubated at 37 °C for 8 hr before treatments. Cells were treated with 100 μM of ATP (or the equivalent volume of PBS as control) in the medium for 1 hr and were fixed and permeabilized as described in the ‘Cell fixation and immunofluorescent staining’ section. Following permeabilization, samples were treated with 50 mM ammonium chloride solution (Avantor) for 5–10 min to quench auto-fluorescence and cells were washed 3x in PBS. Blocking was completed via incubation with 5% BSA (Merck) solution for 1 hr followed by overnight incubation at 4 °C with 1:100 dilutions of DNA labeled anti-P2Y2, and DNA labeled anti-αV antibody in blocking solution. The next day, samples were washed 3× in PBS and 150 nm gold nanoparticles (Sigma-Aldrich) were added for 15 min to act as fiducial markers for drift correction, excess of nanoparticles were removed by 3x washes with PBS. Samples were then left in DNA-PAINT imager buffer solution, prepared as described below, and immediately used for DNA-PAINT imaging experiments.
DNA-PAINT imager solutions
A 0.1 nM P2Y2 imager strand buffer solution (5-TTGTGGT-3’-Atto643, Eurofins) and a 0.2 nM αV imager strand buffer solution (5-GGAGGA-3’-Atto643, Eurofins) were made using 1x PCA (Sigma-Aldrich), 1x PCD (Sigma-Aldrich), 1x Trolox (Sigma-Aldrich), 1x PBS, and 500 mM NaCl (Merck) which facilitates the establishment of an oxygen scavenging and triplet state quencher system. Solutions were incubated for 1 hr in the dark before use. Stock solutions of PCA, PCD, and Trolox were prepared as follows: 40x PCA (protocatechuic acid) stock was made from 154 mg of PCA (Sigma-Aldrich) in 10 mL of Ultrapure Distilled water (Invitrogen) adjusted to pH 9.0 with NaOH (Avantor, Radnor Township, PA, USA). 100 x PCD (protocatechuate 3,4-dioxygenase) solution was made by adding 2.2 mg of PCD (Sigma-Aldrich) to 3.4 mL of 50% glycerol (Sigma-Aldrich) with 50 mM KCl (Sigma-Aldrich), 1 mM EDTA (Invitrogen), and 100 mM Tris buffer (Avantor). 100 x Trolox solution was made by dissolving 100 mg of Trolox in 0.43 mL methanol (Sigma-Aldrich), 0.345 mL 1 M NaOH, and 3.2 mL of Ultrapure Distilled water.
Exchange-PAINT Imaging Experiments
Exchange DNA-PAINT imaging was performed on a custom-built total internal reflection fluorescence (TIRF) microscope based on a Nikon Eclipse Ti-2 microscope (Nikon Instruments) equipped with a 100×oil immersion TIRF objective (Apo TIRF, NA 1.49) and a Perfect Focus System. Samples were imaged under flat-top TIRF illumination with a 647 nm laser (Coherent OBIS LX, 120 mW), that was magnified with custom-built telescopes, before passing through a beam shaper device (piShaper 6_6_VIS, AdlOptica) to transform the Gaussian profile of the beam into a collimated flat-top profile. The beam was focused into the back focal plane of the microscope objective using a suitable lens (AC508-300-A-ML, Thorlabs), passed through a clean-up filter (FF01-390/482/563/640-25, Semrock), and coupled into the objective using a beam splitter (Di03-R405/488/561/635-t1-25 × 36, Semrock). Laser polarization was adjusted to circular after the objective. Fluorescence light was spectrally filtered with an emission filter (FF01-446/523/600/677-25, Semrock) and imaged on an sCMOS camera (ORCA-Flash4.0 V3 Digital, Hamamatsu) without further magnification, resulting in a final pixel size of 130 nm in the focal plane, after 2 × 2 binning. For fluid exchange, each individual chamber of the ibidi µ-SlideVI0.5 was fitted with elbow Luer connector male adaptors (Ibidi) and 0.5 mm silicon tubing (Ibidi). Each imaging acquisition step was performed by adding the corresponding imager strand buffer solution to the sample. Prior to the imager exchange, the chamber was washed for 10 min with 1 x PBS buffer with 500 mM NaCl. Before the next imager strand buffer solution was added, we monitored with the camera to ensure the complete removal of the first imager strand. Sequential imaging and washing steps were repeated for every cell imaged. For each imaging step, 15,000 frames were acquired with 100ms integration time and a laser power density at the sample of 0.5 kW/cm2.
Super-resolution DNA-PAINT image reconstruction
Both P2Y2 and αV Images were processed and reconstructed using the Picasso (Schnitzbauer et al., 2017) software (Version 0.3.3). The Picasso ‘Localize’ module was used to identify and localize the x, y molecular coordinates of single molecule events from the raw fluorescent DNA-PAINT images. Drift correction and multi-color data alignment were performed via the Picasso ‘Render’ module, using a combination of fiducial markers and multiple rounds of image sub-stack cross-correlation analysis. Localizations with uncertainties greater than 13 nm were removed and no merging was performed for molecules re-appearing in subsequent frames. Super-resolution image rendering was performed by plotting each localization as a Gaussian function with a standard deviation equal to its localization precision.
Protein quantification via qPAINT analysis
To convert the list of x, y localizations into a list of x, y protein coordinates the data was further processed using a combination of DBSCAN cluster analysis, qPAINT analysis, and k-means clustering.
First, 21 randomly selected, non-overlapping, 4 × 4 µm2 regions of interest (ROIs) for each type of cell and cell treatment were analyzed with a density-based clustering algorithm, known as DBSCAN. To avoid suboptimal clustering results; ROIs were selected such that they do not intersect with cell boundaries and the regions were the same for P2Y2 and αV images. Single-molecule localizations within each ROIs were grouped into clusters using the DBSCAN modality from PALMsiever (Pengo et al., 2015) in MATLAB (Version 2021a)(Pengo et al., 2015). This clustering algorithm determines clusters based on two parameters. The first parameter is the minimum number of points (‘minPts’) within a given circle. For minPts, we chose a parameter in accordance with the binding frequency of the imager strand and acquisition frame number; in our case this was set to 10 localizations for all the experiments. The second parameter is the radius (epsilon or ‘eps’) of the circle of the cluster of single molecule localizations. This is determined by the localization precision of the super-resolved images and, according to the nearest neighbor based analysis was ca. to 10 nm for all the images.
For qPAINT analysis, we used a custom-written MATLAB (Version 2021a) code: https://github.com/Simoncelli-lab/qPAINT_pipeline (Joseph and Simoncelli, 2023). Briefly, localizations corresponding to the same cluster were grouped and their time stamps were used to compile the sequence of dark times per cluster. All the dark times per cluster were pooled and used to obtain a normalized cumulative histogram of the dark times which was then fitted with the exponential function 1 – exp(t/τd) to estimate the mean dark time, τd, per cluster. The qPAINT index (qi) of each cluster was then calculated as the inverse of the mean dark time, 1/τd.
Calibration was then performed via a compilation of all qPAINT indexes obtained from the DNA-PAINT data acquired for each protein type into a single histogram. Only qPAINT indices corresponding to small clusters (i.e. clusters with a maximum point distance of 150 nm) were considered. This histogram was fitted with a multi-peak Gaussian function to determine the qPAINT index for a cluster of single molecule localizations corresponding to one protein (qi1).
The calibration value obtained with this method was used to estimate the number of P2Y2 and αV proteins in all the single-molecule localizations clusters identified by DBSCAN, as this corresponds to the ratio between qi1 and the qPAINT index of each cluster. Finally, k-means clustering was used to recover a likely distribution of the proteins’ positions in each cluster of single molecule localizations, where k is equal to the number of proteins in that cluster. This information allowed us to quantify the protein density and level of protein clustering.
Nearest neighbor analysis
Nearest neighbor distances (NND) for P2Y2 – P2Y2 and αV-αV were calculated using the recovered P2Y2 and αV-protein maps as described above via a custom-written MATLAB (Version 2021a) script: https://github.com/Simoncelli-lab/qPAINT_pipeline (Joseph and Simoncelli, 2023). For colocalization analysis, the NND for each protein of one dataset with respect to the reference dataset was calculated (i.e. P2Y2 - αV) using a similar MATLAB script. To evaluate the significance of the NND distributions, we randomized the positions of P2Y2 and αV for the comparison of P2Y2 – P2Y2 and αV-αV NND distributions, respectively, and the positions of one of the two proteins for the comparison of the NND between P2Y2 - αV protein distributions. The resulting histogram of the nearest neighbor distances for both the experimental data sets and the randomly distributed data was normalized using the total number of NND calculated per ROI to calculate the percentage of the population with distances smaller than a set threshold value.
Cell lysates were extracted using RIPA buffer and 20 µg denatured protein per sample was loaded and separated using an 8% SDS-PAGE gel. Gels were run at 150 V for 2 hr and transferred into a nitrocellulose membrane (GE Healthcare) at 100 V for 1 hr. Following blocking with 5% milk (Sigma) in 0.1% TBS-T for 1 hr, membranes were incubated with 1:1000 dilution of antibodies against phosphorylated FAK (Tyr397, 3283, Cell Signaling, RRID:AB_2173659), phosphorylated ERK 1/2 (S217/221, 9154, Cell Signaling, RRID:AB_2138017), P2Y2 (APR-010, Alomone Labs, RRID:AB_2040078), HSC 70 (SC7298, Santa Cruz, RRID:AB_627761), or α-tubulin (T5168, Sigma-Aldrich, RRID:AB_477579) with 5% BSA in 0.1% TBS-T overnight at 4 °C. Membranes were probed with anti-Mouse-HRP (P0447, DAKO, RRID:AB_2617137), or Anti-Rabbit-HRP (P0448, DAKO, RRID:AB_2617138) at 1:5000 in 5% milk in TBS-T for 1 hr at room temperature. Images were captured by using Luminata Forte Western HRP substrate (Millipore) and imaged with an Amersham Imager 600 (GE Healthcare).
For the statistical analysis of the number and colocalization of DNA-PAINT images, a minimum of five 4 × 4 µm2 regions obtained from AsPC-1 cells were analyzed per condition. For all experiments, normality tests were performed and the non-parametric Kruskal-Wallis test for significance was calculated. All graphs and statistical calculations of experimental data were made using Prism 9.4.1 (GraphPad).
All data generated or analysed during this study are included in the manuscript and supporting file, or online resources are fully referenced. Human PDAC tumour data were generated by TCGA Research Network (https://www.cancer.gov/tcga) and by the Clinical Proteomic Tumour Analysis Consortium (https://www.proteomics.cancer.gov). The Genotype-Tissue Expression (GTEx) Project was used for the analysis of normal pancreatic tissue samples (https://gtexportal.org).
Characterization of the A2 adenosine receptor labeled by [ 3H ] NECA in rat striatal membranesMolecular Pharmacology 29:331–346.
Large meta-analysis of multiple cancers reveals a common, compact and highly prognostic hypoxia metageneBritish Journal of Cancer 102:428–435.https://doi.org/10.1038/sj.bjc.6605450
Purinergic signalling in the pancreas in health and diseaseThe Journal of Endocrinology 213:123–141.https://doi.org/10.1530/JOE-11-0434
Internalization of G-protein-coupled receptors: implication in receptor function, physiology and diseasesBest Practice & Research. Clinical Endocrinology & Metabolism 32:83–91.https://doi.org/10.1016/j.beem.2018.01.004
P2Y2 receptors induced cell surface redistribution of alpha (V) integrin is required for activation of ERK 1/2 in U937 cellsJournal of Cellular Physiology 211:410–422.https://doi.org/10.1002/jcp.20946
Extracellular ATP and P2 purinergic signalling in the tumour microenvironmentNature Reviews. Cancer 18:601–618.https://doi.org/10.1038/s41568-018-0037-0
Release of adenosine triphosphate from isolated adult heart cells in response to hypoxiaThe Journal of Physiology 268:371–390.https://doi.org/10.1113/jphysiol.1977.sp011862
Imbalance of desmoplastic stromal cell numbers drives aggressive cancer processesThe Journal of Pathology 230:107–117.https://doi.org/10.1002/path.4172
A single-cell type transcriptomics map of human tissuesScience Advances 7:eabh2169.https://doi.org/10.1126/sciadv.abh2169
Pancreatic tumors show high levels of hypoxiaInternational Journal of Radiation Oncology, Biology, Physics 48:919–922.https://doi.org/10.1016/s0360-3016(00)00803-8
HIF-1α regulates A2B adenosine receptor expression in liver cancer cellsExperimental and Therapeutic Medicine 18:4231–4240.https://doi.org/10.3892/etm.2019.8081
Web-based survival analysis tool tailored for medical research (kmplot): development and implementationJournal of Medical Internet Research 23:e27633.https://doi.org/10.2196/27633
Integrin clustering in two and three dimensionsLangmuir 28:5379–5386.https://doi.org/10.1021/la203725a
WebGestalt 2019: gene set analysis toolkit with revamped uis and apisNucleic Acids Research 47:W199–W205.https://doi.org/10.1093/nar/gkz401
SuperPlots: communicating reproducibility and variability in cell biologyThe Journal of Cell Biology 219:e202001064.https://doi.org/10.1083/jcb.202001064
The P2RY2 receptor induces carcinoma cell migration and EMT through cross-talk with epidermal growth factor receptorJournal of Cellular Biochemistry 117:1016–1026.https://doi.org/10.1002/jcb.25390
Characterisation of p2y2 receptors in human vascular endothelial cells using AR-C118925XX, a competitive and selective p2y2 antagonistBritish Journal of Pharmacology 176:2894–2904.https://doi.org/10.1111/bph.14715
Discovery of p2y2 receptor antagonist scaffolds through virtual high-throughput screeningJournal of Chemical Information and Modeling 62:1538–1549.https://doi.org/10.1021/acs.jcim.1c01235
Lifeact: a versatile marker to visualize F-actinNature Methods 5:605–607.https://doi.org/10.1038/nmeth.1220
Super-resolution microscopy with DNA-PAINTNature Protocols 12:1198–1228.https://doi.org/10.1038/nprot.2017.024
Ecto-5’-nucleotidase (CD73) regulation by hypoxia-inducible factor-1 mediates permeability changes in intestinal epitheliaThe Journal of Clinical Investigation 110:993–1002.https://doi.org/10.1172/JCI15337
Upregulation of P2Y2 nucleotide receptor in human hepatocellular carcinoma cellsThe Journal of International Medical Research 44:1234–1247.https://doi.org/10.1177/0300060516662135
Characterization of a new primary human pancreatic tumor lineCancer Investigation 4:15–23.https://doi.org/10.3109/07357908609039823
A nanoengineering approach for immobilizationACS Nano 2:2374–2384.https://doi.org/10.1021/nn800508f
GEPIA: a web server for cancer and normal gene expression profiling and interactive analysesNucleic Acids Research 45:W98–W102.https://doi.org/10.1093/nar/gkx247
Integrin and GPCR crosstalk in the regulation of ASM contraction signaling in asthmaJournal of Allergy 2012:341282.https://doi.org/10.1155/2012/341282
Endocytosis mechanism of P2Y2 nucleotide receptor tagged with green fluorescent protein: clathrin and actin cytoskeleton dependenceCellular and Molecular Life Sciences 62:1388–1399.https://doi.org/10.1007/s00018-005-5052-0
Richard M WhiteSenior and Reviewing Editor; Memorial Sloan Kettering Cancer Center, United States
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
[Editors' note: this paper was reviewed by Review Commons.]https://doi.org/10.7554/eLife.86971.sa1
1. General Statements [optional]
We are grateful to the reviewers for highlighting the novelty of the mechanism we describe for P2Y2 in driving RGD-binding integrin-dependent invasion, and acknowledging its potential in cancer therapy. We thank the reviewers for their valuable and detailed comments, which have allowed us to prepare a significantly stronger and clearer manuscript.
2. Point-by-point description of the revisions
Reviewer #1 (Evidence, reproducibility and clarity (Required)):
The study identifies P2Y2 as a purinergic receptor strongly associated with hypoxia, cancer expression and survival. A link is found between P2Y2-integrin interaction and cancer invasion, highlighting this as a novel therapeutic target. The mechanism is interesting and general well explored.
We thank the reviewer for acknowledging the novelty of the therapeutic target presented in this work.
As P2Y2 is highly expressed by other cell types found with tumours, including vascular endothelium and leukocytes, the authors should reflect on this as a confounding factor in the analysis of adrenocarcinoma gene expression analysis. I appreciate the RNAscope work may resolve this issue to some extent.
We agree that P2Y2 is known to be expressed in other cell types. RNAscope did not show convincing staining in PDAC normal adjacent tissue (was similar to negative staining), perhaps due to the challenging nature of pancreatic tissue with respect to RNA degradation. We have resolved this issue by including single cell RNA-seq of normal human pancreas for P2Y2 from Protein Atlas (Sup. Figure 2B), which shows expression in several cell types, mainly endocrine cells, and macrophages. We now mention this in line 142 : “P2Y2 is known to be expressed at low levels in normal tissues but interestingly RNAscope did not detect this. This data suggest (1) the lower limits of the technique compounded by the challenge of RNA degradation in pancreatic tissue and (2) supports that in tumour tissue where it was detected there was indeed overexpression of P2Y2, in line with the bioinformatic data. Interrogating single cell P2Y2 RNA expression in normal PDAC from proteinatlas.org (Karlsson et al., 2021), expression was found at low levels in several cells types, for example in endocrine cells and macrophages (Sup. Figure 2B).”
The authors correctly identify that the level of ATP in the tumour microenvironment can be very high, typically 100uM or so. However, these concentrations are supramaximal for P2Y2 activation, at which ATP has an approximate EC50 of 100nM. Coupled with the fact that many cell types, including cancer cells, constitutively secrete ATP, there is an opportunity to explore the effects of lower ATP concentrations in some assays, or provide some concentration-response relationship to give more confidence of P2Y2-dependent effects.
We thank the reviewer for raising this point and we agree that 100 μm can be a high concentration, albeit one that is frequently used throughout the literature. We have now included a concentration-response relationship (Sup. Figure 2D) showing that ATP causes cytoskeletal changes that are P2Y2 dependent most prominently at 100 uM, the concentration that, as the reviewer has also corroborated, is similar to the concentration of ATP found in tumours.
Also, the authors describe the use of cancer cells where P2Y2 has been knocked out using CRISPR. Does this KO have an effect on cancer invasion? The effect of ARC should be absent in these cells and give confidence the effects of ARC are P2Y2-dependent, as some off-target effects of this antagonist have been reported. To explore the influence of constitutive P2Y2 activity, the authors should explore the effects of ARC alone in some assays.
We agree that including more AR-C only experiments would be informative, so we have included a 3D sphere invasion assay with our CRISPR cell line treated with and without AR-C that shows no effect in invasion (p = 0.4413) (Sup. Figure 3J). We have now also included images of AsPC-1 cells transfected with Lifeact, showing no changes in morphology with AR-C only (Sup. Figure 2E). We apologise for missing a ‘+’ in one of the supplementary figures which shows AR-C only in AsPC-1 cells has no effect on its own.
The effects of the CRISPR cell line in invasion are shown in Figure 3F, showing a significant reduction (p = 0.0005) in invasion.
The title of the manuscript implies extracellular ATP drives cancer invasion, though in my opinion this statement is not fully explored. Though ATP/UTP are applied at supramaximal concentrations for P2Y2 activation, the influence of ATP in the cell culture microenvironment without exogenous application is not explored. One would predict that scavenging extracellular ATP with apyrase would negatively impact invasiveness and the proximity of integrin and P2Y2 without ATP/UTP application if constitutively secreted ATP is involved. Pharmacological manipulation of ectonucleotidase activity is an alternative. Experimental route to explore this.
We agree and have changed the title of our article to “Purinergic GPCR-integrin interactions drive pancreatic cancer cell invasion”. Our 3D sphere experiments with the CRISPR cell line show a reduction in invasion without exogenous application of ATP, which we also see to a lesser extent in our siRNA P2Y2 cell line. We have tested our sphere model with apyrase but unfortunately, the buffer used for apyrase to work is not compatible with our gel composition. Pharmacological manipulation would be a very good alternative if the cells used expressed high levels of CD39 or PANX1, which unfortunately they don’t. We hypothesise that most basal extracellular ATP in our 3D spheres comes from hypoxic areas that cause cell death, just as is postulated for tumours.
Immunoprecipitation experiments of native proteins would be more convincing data that P2Y2 and integrin physically interaction, as opposed to being in close proximity. This would also overcome artifacts of interaction that can be attributed to receptor overexpression.
We attempted immunoprecipitation experiments but unfortunately ran into several technical difficulties, including the anti-αV antibody working poorly for Western blot. Immunoprecipitation of these proteins has been reported by others (PMID: 25908848), supporting the proposed interaction.
DNA-PAINT super resolution microscopy allows for quantification of nanoscale distances, and we used this to calculate the distances where physical interaction occurs. The possibility of this close proximity being by chance is accounted for in the computational nearest neighbour distance calculation by calculating points randomly distributed. This random distribution calculation also helps in overcoming artifacts of interaction due to overexpression, as the random distributed points are the same number of points as the proteins detected in each condition for each region of interest. Importantly, we also performed DNA-PAINT in using untransfected AsPC-1 thus endogenous levels (no receptor overexpression or alteration) and saw similar results (Sup. Figure 4A-D), thus we are confident of the interactions reported.
Finally, we alter the RGD motif, which underpins the physical interaction, and see significant changes that match observations in previous publications using the P2Y2 agonist UTP, mentioned in the discussion: Line 398 “Following ATP stimulation, the number of P2Y2 proteins at the plasma membrane decreased significantly after one hour, implying receptor internalisation, in line with previous work showing P2Y2 at the cell surface was reduced significantly after one hour of UTP stimulation (Tulapurkar et al., 2005).” and Line 408: “P2Y2 affecting cell surface redistribution of αV integrin has been reported, with αV integrin clusters observed after 5 min stimulation with UTP (Chorna et al., 2007)”
It is currently not clear what the mechanistic relationship between P2Y2 activity, P2Y2-integrin proximity and RGD motif is. Do the authors suggest the RGD domain becomes exposed upon receptor activation? The mechanism is not fully articulated in the discussion.
We apologise for any lack of clarity in our postulated mechanism, we have now included a more detailed explanation of the mechanism in the discussion : Line 417 “We speculate that by reducing the ability of integrins to bind to the RGD of P2Y2, through receptor internalisation, RGE mutation or through cRGDfV treatment, there is less RGD-triggered integrin endocytosis, hence less integrin recycling and an increase of integrins at the cell surface.”
Reviewer #1 (Significance (Required)):
General assessment: A novel mechanism is presented for therapeutic intervention of cancer. The study relies on supramaximal concentrations of agonist and overexpressed receptors. Role of endogenous P2Y2 not fully explored. The study lacks in vivo evidence of the importance of this mechanisms. Cell developed in the study could be used in mouse models to explore effect on tumour growth.
Advance: Integrin and P2Y2 interactions are already documented but not in context of cancer.
Audience: basic research
We thank the reviewer for crediting this work as a novel mechanism for therapeutic intervention of cancer. We trust that the new data provided (as discussed above) have resolved the concerns of the reviewer as we now have provided an explanation for the concentrations used. We do rely on overexpressed receptors for a small portion of our experiments, however, all experiments with overexpressed receptors were then tested in cells with endogenous expression of P2Y2 and used pharmacological means to show the same behaviour. We have now clarified this. We have also included in the discussion a sentence about the mouse experiment performed by Hui et al. with regards to reduced tumour growth when targeting P2Y2: Line 365: “Combination treatment of subcutaneous xenografts of AsPC-1 or BxPC-3 cells with the P2Y2 antagonist AR-C together with gemcitabine significantly decreased tumour weight and resulted in increased survival compared to placebo or gemcitabine monotherapy control (Hu et al., 2019).”
Reviewer #2 (Evidence, reproducibility and clarity (Required)):
Considering the fact that most PDAC are characterized by a high level of extracellular purines content, authors decided to study the expression of the 23 genes coding for membrane proteins involved in the binding or transport of purines in available PDAC transcriptomic cohorts. This approach led to the identification of P2Y2, a GPCR, as the best predictor for the worst survival of patients. Using in vitro models, they show that P2Y2 expression is associated with increased invasion capacity of pancreatic cancer cells and that this pro-invasive effect is dependent on the interaction of P2Y2 with αV integrin via the RGD motif.
It is not clear to me why authors decided at one point to perform a GSEA comparing low and high mRNA expression of P2Y2 and why they decided to focus on the potential interaction of P2Y2 with integrin αV. As a GPCR, activation of P2Y2 leads to the activation of several downstream signaling pathways that may directly impact the adhesion, migration, and invasion properties of cells. Moreover, despite the presence of the RGD motif in P2Y2, it is not excluded that it may bind (maybe more efficiently) to other "cell adhesion" molecules.
We apologise if the link between the GSEA figure and focusing on the potential integrin interaction was not clear. We have now performed GSEA using the panther gene set library, which includes a “Integrin signalling pathway” gene set. This was the top ranked gene set in both cohorts and we have substituted the GSEA figure for this instead (Figure 2D). We trust that the narrative of the manuscript and our rationale to pursue the importance of integrin interaction is now clear.
We agree with the reviewer and believe that P2Y2 may bind to other molecules important in cell adhesion. We studied integrin interactions due to the clear relationship of P2Y2 and integrins in patient data, which was not as evident with other binding partners. Furthermore, this relationship is unexplored in cancer and offers novel therapeutic strategies.
Similarly, if αV can regulate P2Y2 signaling, what about the regulation of αV signaling pathways by P2Y2? αV integrin has to bind to a β subunit and, depending on the identity of the β subunit, may have distinct regulations and so different impact on cell invasion. How P2Y2 can interfere with these α/β ratios?
We thank the reviewer for this comment, and have now included western blots showing the impact of P2Y2 treatment on integrin signalling through FAK and ERK (Figure 5). We agree that the β subunit may have distinct regulation and outputs, but this is outwith the scope of our current study.
While it has been shown in other studies, in this work, there is no real proof of the interaction between P2Y2 and αV. Only in Figure 4I, where the authors look at the NND <20nm between both proteins, we can see that only 1 to 2 % of αV is in close proximity with P2Y2, which seems very low.
We thank the reviewer for raising this point as it has made us realise that our chosen NND of <20 nm, in an attempt of being cautious, consistant, and only select true physical interaction was too strict for some conditions. We have adjusted this to the maximum NND required for the proteins to physically touch based on individual protein and antibody dimensions (Sup. Figure 5A). The resulting changes are <50 nm for P2Y2-αV, <20 nm for αV-αV and <40 nm for P2Y2-P2Y2, giving more realistic percentages for each condition. We then verified this range using the NND histograms now included in the manuscript, which additionally provide information about the distribution of the NNDs in each protein-protein dynamic (Sup. Figure 5B). One important point is that integrins are part of a large number of protein complexes and interact with many proteins, hence only a small percentage will be interacting with P2Y2.
Surprisingly, in the absence of ATP, P2Y2 RGE mutant, which should no more interact with αV, show a 2 to 3 fold more vicinity to αV compared to WT P2Y2. How can the authors explain this?
We agree that this is a surprising, but robust discovery. By altering the RGD motif, there may be less RGD-triggered integrin endocytosis, leading to increased integrins at the surface. We have included this hypothesis in the discussion in Line 417. The RGE mutation has less affinity to integrins, meaning it still retains some ability to bind to integrins. Hence by chance, a higher number of integrins will result in a higher number of interactions with the RGE. We are planning to interrogate the internalisation dynamics in a future study.
For DNA-PAINT experiments, the authors only focus on membrane proteins whose amounts are balanced by internalization, recycling and export from internal compartment. As claimed, but not demonstrated by the authors, interaction of P2Y2 and αV may interfere with all these steps, thereby increasing or decreasing the cell surface expression of both proteins. Hence, it would be useful to (1) control proteins levels by western blot, especially for the overexpressed P2Y2, to be sure that they are the same, (2) block internalization and/or export to decipher the important steps.
In fact, all these main questions are raised by the authors in the end of the discussion but so far, they only show that the RGD motif has an impact on the biological role of P2Y2 (cell invasion) and on the membrane dynamic of αV and itself.
We thank the reviewer for the suggestions:
1) In the course of our attempts to perform co-IP for P2Y2 and αV we could confirm that P2Y2 expression levels were equivalent (see Author response image 1), but the problems with anti-αV antibodies prevented completion of the experiment. We also show IF staining showing similar levels of P2Y2 for both overexpressed conditions (Sup. Figure 3K).
2) As the reviewer highlights, in this work we have focused on the role of P2Y2 in PDAC invasion and have looked at single-molecule resolution membrane dynamics of αV and P2Y2. The different steps of P2Y2 and integrin αV interactions in internalisation, recycling and export are certainly interesting to study but beyond the scope of the current manuscript and in our future aims. We include these ideas in the discussion as suggestions for future research and as a possible explanation for the dynamics observed.
Figure 2A, authors use RNAscope in order to reveal P2Y2 mRNA expression and distribution in tumor versus normal tissue from 2 patients. They rather show the protein expression, using the antibody they used in other experiments, by standard IHC and in a higher number of patients, including short and long survival, to confirm that the results they obtain by bioinformatics study of transcriptomic data are real.
We now explicitly mention a paper (PMID: 30420446) that performed IHC of P2Y2 in 264 patients showing that P2Y2 was predominantly found in the tumour area, matching our bioinformatics study: Line 141 “matching our findings from larger publicly available cohorts, including P2Y2 IHC data from 264 patients in the Renji cohort (Hu et al., 2019).” and Line 359 “These observations were supported by published immunohistochemical staining of 264 human PDAC samples, showing that P2Y2 localised predominantly in cancer cells in human PDAC…”
Some figure legends are incorrectly numbered or described, such as the figure 4.
We apologise for the incorrectly described figures in figure 4, this has now been corrected.
Can we reasonably talk about OMIC while studying 23 genes? In fact, as described by Timothy A. J. Haystead in 2006 (PMID: 16842150) the purinome is constituted of about 2000 genes coding for proteins binding to purines (including all kinases for example). Author should redefine they pool of genes as perhaps purines receptors/transporter?
We agree with the reviewer and have redefined the pool of genes to ‘purinergic signalling genes’ or ‘(part of the) extracellular purinome’.
P2Y2 and ADORA2B associated with worse survival while P2Y11 and ADORA2A are associated with better survival (Figure 1B). Would it be more interesting to understand why proteins of the same family act in opposite ways?
We have now included text exploring this idea in the discussion. Both P2Y2 and ADORA2B show increased expression with HIF-1α and/or hypoxia and the inverse happens with ADORA2A, for example. Line 352: “Adenosine A2B receptor requires larger agonist concentrations for activation compared to other receptors in the same family, such as adenosine A2A (Bruns, Lu and Pugsley, 1986; Xing et al., 2016), and receptor expression has been reported to increase when cells are subjected to hypoxia (Feoktistov et al., 2004). Moreover, HIF-1α has been shown to upregulate A2B and P2Y2 expression in liver cancer (Tak et al., 2016; Kwon et al., 2019).”
Figure 1C, any value for the correlation with Survival? Cause this is not so obvious in the figure.
We agree this correlation needs strengthening with a numeric value, we have now included a Kaplan-Meier curve of high vs low Winter hypoxia score PDAC patients showing significantly lower survival with higher Winter hypoxia score (Sup. Figure 1B).
Regarding the correlation of P2Y2 and ADORA2B with hypoxia scores, any HIF1 responsive element in promoter? What happens regarding the expression level of these genes when cells are transferred to low oxygen conditions?
We thank the reviewer for these questions. The relationship of P2Y2 and ADORA2B with hypoxia and/or HIF-1α has been explored in other publications which are now cited in the discussion. Line 356: “Moreover, HIF-1α has been shown to upregulate A2B and P2Y2 expression in liver cancer (Tak et al., 2016; Kwon et al., 2019).” Of note, a HIF1-α responsive element has been reported for A2B, but as yet not for P2Y2.
Figure 4 E to M are too small.
We apologise and have now increased the size of the graphs and the figure.
In Supp Figure 4, what are the "Non-altered AsPC-1 cells"?
We apologise for the confusion that may have arisen from calling normal AsPC-1 cells “Non-altered AsPC-1 cells”. We have changed this to ‘Normal AsPC-1 cells (untransfected and unchanged P2Y2 expression).
Reviewer #2 (Significance (Required)):
Strengths: All the data shown are experimentally and statistically strong.
Limitations: This study remains largely descriptive with no real molecular mechanism that could at least partially explain the biological role of P2Y2 regarding cell invasion.
We thank the reviewer for noting the experimental strength of the paper.
After the suggested changes, including integrin signalling experiments, and strengthening our DNA-PAINT results, the molecular mechanism presented in this work has been strengthened and clarified significantly. These changes have also helped greatly in the mechanistic explanation of the role of P2Y2 in cell invasion.
Reviewer #3 (Evidence, reproducibility and clarity (Required)):
The authors concentrate on the members of the purinome and attempt to identify members of the pathway that are especially relevant for PDAC biology, especially invasion and metastatic spread. Using the in silico analysis of transcriptome data from publicly available PDAC patient cohorts, the authors identify P2Y2 as being the most prominent in terms of cancer cell expression and with highest impact on patient survival. The authors than take an effort in functional characterization of P2Y2 and demonstrate that downregulation/deletion of P2Y2 leads to abrogation of ATP activated invasion in hanging drop spheroid model system in a very reasonable and scientifically good way. Finally, the authors postulate that the P2Y2 actions go over interaction with integrin AlphaV and modulations of the cellular cytoskeleton and show via DNA PAINT that a direct interaction of the 2 molecules. The hypothesis is experimentally elaborated in a sound way mostly using cell culture as a system.
The study is solid communicated, the number of experiments seems to be fine. For my understanding, the study relies much on mRNA data (gene expression in cell lines and patient samples), I would suggest providing evidence on protein level what might have been challenging due to potential lack of specific antibody.
We thank the reviewer for acknowledging our experimentally elaborated hypothesis and our solid communication of the study. As mentioned before, we now explicitly mention a paper (PMID: 30420446) that performed IHC of P2Y2 in 264 patients showing that P2Y2 was predominantly found in the tumour area, matching our bioinformatics study.
Reviewer #3 (Significance (Required)):
To strengthen the hypothesis experimentally, I would suggest the experiments listed below:
Figure 1: The authors took a solid bioinformatic effort and analyzed expression of different genes of the purinome pathway in different PDAC patient and cell gene expression databases. In this part, the authors rely a lot on correlation of hypoxia and define high hypoxia scores and low hypoxia scores from previously published datasets. Although hypoxia surely plays an important biological condition in the PDAC, I am not sure I get the connection between purinome pathway and hypoxia. Few sentences give a broad introduction about hypoxia-purinome connection in the discussion part of the manuscript, but I think the readership would benefit from more specific statements (which drug, which hypoxic target, which system-mouse/human/cells, what was the exact discovery) and connect those specific statements to the work that has been done here.
We agree with the reviewer that the study can benefit from more information about the hypoxia-purinergic signalling link. Hence, we have now included more detailed explanations of how hypoxia and purinergic signalling are related in the discussion, giving more information about the cell types and the exact discovery. Line 338: “Purinergic signalling has been associated classically with hypoxia and immune function in cancer (Di Virgilio et al., 2018). One of the first reports of hypoxia inducing ATP release in cells identified an increase of extracellular ATP in rat heart cells when kept in hypoxic conditions (Forrester and Williams, 1977). PDAC is a highly hypoxic cancer, with high levels of ATP reported in the tumour interstitial fluid of human and mouse PDAC tissues compared to healthy tissues (Hu et al., 2019).”
Do the authors attempt to state here that hypoxic PDACs are those with worse prognosis and more aggressive and thus try to associate members of the purine pathway with those "worse" PDACs? Surprisingly, there is relatively little knowledge about hypoxia in PDAC and I would not suggest using it in this context as a predictor. Reports do suggest that hypoxia forces the emerging of resistant phenotypes but if the authors want to use hypoxic signatures, they have to fortify better (with literature) why do they choose hypoxia and what is the hypothesis that connects hypoxia to purinome, what makes this connection worth investigating.
We thank the reviewer for raising the question of PDAC and worse prognosis with hypoxia. We have now included a Kaplan-Meier curve of high vs low Winter hypoxia score PDAC patients showing significantly lower survival with higher Winter hypoxia score (Sup. Figure 1B). The significant link with poor survival shown with hypoxia and the inclusion of more detailed explanation of the links with hypoxia and purinergic signalling proteins (mentioned above), now clarify the reasoning for investigating this connection.
I find the statement "hypoxia in tumor core" a bit tricky, acute and chronic hypoxia can occur anywhere in the tumor, to my knowledge there are no reports saying only the tumor core suffers from hypoxia in PDAC. PDAC being especially rich in stroma in all of its parts is probably more prone to overall hypoxia and not only in tumor core.
We agree that “hypoxia in tumour core” can be a tricky statement. We have changed “tumour core” to tumour cell compartment and have cited data that demonstrate hypo-vascularisation found in the juxta-tumoural stroma, due to PDAC cells inhibiting angiogenesis (PMID: 27288147). This paper supports our hypothesis of distribution of oxygen being reduced in the tumour area. Hence why we hypothesise that purinergic genes would be preferentially expressed in the tumour area: Line 112 “We hypothesised that genes related to high hypoxia scores would be expressed preferentially in the tumour cell compartment, as PDAC cells inhibit angiogenesis, causing hypo-vascularisation in the juxta-tumoural stroma (Di Maggio et al., 2016).”
We would like to clarify that we do not believe that only the tumour core suffers from hypoxia, we hypothesise that there is more hypoxia in the tumour cell areas. Although there are no reports of only the tumour core suffering from hypoxia, there is evidence of the tumour epithelial region of the cancer having a greater range of hypoxia (1-39%) compared to the stromal (1-13%) (PMID: 26325106). Moreover, all our analyses point to most purinergic genes differentially expressed in patients with high hypoxic scores being also related to cancer cells and the tumour region. These bioinformatic results linking certain genes like P2RY2 and ADORA2B with hypoxia are also supported in published work cited in the discussion (Line 354 and 356).
I would suggest that the authors rely on published subtyping of PDAC
patient cohorts (Collisson et al., 2010; Bailey et al; Moffit et al., 2015; Chan-Sen-Yue, 2020)
and correlate the expression of purinome genes with the QM/basal-like PDAC subtype that has been confirmed multiple times as the "bad predictor" and use those subtypes for correlation with purinome pathway members. In figure 1E is also shown that P2RY2 is high in expression in basal-like subtype.
We thank the reviewer for this suggestion and have included the subtyping of patients in the PAAD-TCGA cohort in Sup. Table 1 and added comments about the genes related to the different subtypes in the text: Line 88 “In the Bailey model, most genes were related to the Immunogenic subtype except for NT5E, ADORA2B, PANX1 and P2RY2, which related to Squamous (Bailey et al., 2016). Collisson molecular subtyping showed several purinergic genes associated mostly to quasimesenchymal and exocrine subtypes (Collisson et al., 2011). The Moffit subtypes were not strongly associated with purinergic genes except for ADA, NT5E, P2RY6, P2RY2 and PANX1 associated with the Basal subtype (Moffitt et al., 2015).” and Line 345 “Expression of most purinergic genes was associated predominantly with immune cells, immunogenic PDAC subtype and low hypoxia scores (Figure 1C, E). In contrast, expression of genes correlated with worse survival and hypoxia (PANX1, NT5E, ADORA2B and P2RY2) was associated with tumour cells and the squamous PDAC subtype, correlating with hypoxia, inflammation and worse prognosis (Bailey et al., 2016).”
We did not include the subtyping of Chan-Sen-Yue, 2020, due to the similarities with Moffit and the lack of correlation of basal/classical types with purinergic signalling genes as many of them are not expressed in cancer cells.
Figure 2: In further course of the paper the authors elaborate on possible functions of P2RY2 in PDAC. Although the mRNA data is pretty elaborate, the RNA SCOPE ISH has been performed on only 3 (!) patient PDAC samples. To demonstrated the mRNA is really found in tumor and not in normal adjacent tissue or stroma, I would strongly suggest to increase the number of samples here. The authors should perhaps try to co-localize ISH signals with IF/IHC for some other cancer cell marker, e.g. PanCK or GATA6/KRT81 in human samples to differentiate basal-like from classical samples; if possible, I would even suggest to perform immunohistochemistry instead of RNA scope and confirm the presence of the receptor. If there is an issue with the antibody availability, please state so in the manuscript so that it is clear to the readers why mRNA expression is favored over protein.
We thank the reviewer for these suggestions.
RNAscope was used to verify our transcriptomic bioinformatic results of location of expression P2Y2 in the tumour from publicly available data of 60 pairs of laser microdissection of PDAC epithelial and stromal tissue and the PAAD TCGA deconvolution of 177 patients. We have experienced issues with RNAscope due to the RNA degradation in pancreatic tissue and other technical difficulties which unfortunately led to only having 3 samples showing staining with the positive control. All three successful samples showed P2Y2 expression located in cancer cells. The images presented show the location of P2Y2 RNA expression in the tumour region, which was the aim of the RNAscope experiment.
RNAscope only captures mRNA expression above a specific threshold, and we are aware that P2Y2 will be expressed in other cell types in the normal adjacent as seen in the deconvolution. We have now included in supplementary single cell RNAseq data of normal PDAC tissue to counteract this issue (Sup. Figure 2B).
We also cite a publication that has performed P2Y2 IHC in 264 patients and showed that P2Y2 protein expression was predominantly shown in the epithelial tumour region (PMID: 30420446), hence staining of P2Y2 in a high number of patients has already been performed: Line 359 “These observations were supported by published immunohistochemical staining of 264 human PDAC samples, showing that P2Y2 localised predominantly in cancer cells in human PDAC”
As shown in Figure 1 E, P2Y2 is associated with basal and classical tumour cells, not just exclusively to basal, hence the staining to differentiate subtypes is not pertinent to the focus of this paper.
The GSEA data indicated that high P2Y2 expression relates to processes of adhesion/ECM/cytoskeleton organization where the authors draw the conclusion (based also on published data mostly on neuronal/astrocyte work) that P2Y2 may interact with integrins over the RGD domain and thus contribute to invasion an migration. Since this is a very important assumption, I would strongly suggest to expand the experiments of figure 2E and 2G on at least 2 more PDAC cell line, if possible include some with originally epithelial morphology (eg. HPAFII, HPAC…).The visualization of filaments can be done with common IF staining, eg. phalloidin, no need for stable expression.
Perhaps the reviewer missed Sup. Figure 2F, where data from Figure 2G are recapitulated in 3 different cell lines. We support the idea of the reviewer in including epithelial morphology cells, hence we added an extra cell line to have 2 cells with epithelial morphology, BxPC-3 and CAPAN-2.
We have tried repeating the experiment in Figure 2E in epithelial cells, but the way the epithelial cells grow in clusters (Sup. Figure 2F) make it very difficult to evaluate the morphology of individual cells and get quantifiable results. Nonetheless, we show phenotypic similarities of BxPC-3 to AsPC-1 cells in the invasion assays.
I would also be in favor of investigating the expression of EMT markers upon ATP stimulation.
We thank the reviewer for the suggestion, although feel this is out of scope for our study. There have been recent controversies with reference to EMT and cancer metastasis (PMID:31666716) but more importantly we see changes in cell morphology 1 hour after ATP treatment, indicating it is not/not just EMT.
How was 100µM/5µM chosen as a working concentration?
We have now included figures showing different concentrations of ATP (Sup. Figure 2D) and AR-C (Sup. Figure 2E) to illustrate how the concentrations were selected based on the greatest change in morphology for ATP and the full recovery of original cell morphology for AR-C.
AsPC-1 is also known as the cell line that gladly migrates and invades, usually used in metastatic modeling of PDAC. Would be interesting to see if another cell line that is not that migrative (HPAF II) presents the same effect…
This is an interesting point, although we haven’t performed experiments with low migrative cells, later on the work, invasion assays with the epithelial cell line BxPC-3, which has a very different migrative nature, presented the same effect (Sup. Figure 3G, F). We also perform invasion assays with PANC-1 cells, which also recapitulate an invasive phenotype when transfected with P2Y2.
Is treatment with ATP inducing expression of P2RY2 maybe? What is happening with Intergrin expression upon ATP treatment? Since the hypothesis is that extracellular ATP is driving the invasion, I would certainly suggest to investigate if ATP treatment induces expression of P2RY2 in a time and dose dependent manner.
We thank the reviewer for this suggestion. We have now changed the title to “Purinergic GPCR-integrin interactions drive pancreatic cancer cell invasion”, hence shifting from a focus on extracellular ATP and focusing on the effects of the RGD motif in invasion.
The authors made very good efforts here to provide functional evidence that P2Y2 is really involved and essential for ATP induced invasion in PDAC cells. They performed an 3D hanging drop spheroid model for invasion in co-culture with stellate cells and show that ATP treatment leads to invasive behavior that is than blocked by addition of P2Y2 antagonist or RGD blocking peptides. Although stellate cells are a nice add-on, keeping in mind the very complex tissue micro-environment of the PDAC, I don't rate the presence of stellate cells here as essential. Are the results the same when experiments are performed without stellate cells?
We thank the reviewer for raising this point, as it has allowed us to clarify that the stellate cells are crucial for this assay to work as they are essential for the formation of the cancer spheres due to their matrix deposition. We have included the hanging drop with and without stellate cells to illustrate this point (Sup. Figure 3A)
EMT markers increase upon ATP stimulation, do not increase under siRNA downregulation of P2Y2?
As mentioned above, we thank the reviewer for the comment, but we are not focusing on EMT, given the rapidity of the phenotype we observe.
Furthermore, the authors downregulate the P2Y2 using the siRNA/CRISPR-Cas9 approach and confirm that the P2Y2 is really involved in the invasive spread also using the specific RGD block. Experiments in the figure 3 are fairly done and provide functional evidence for the hypothesis. I would suggest that for clarity reasons on every panel (A, B,C…) is written which cell line is used (mostly Aspc1) and for the siRNA experiment I would suggest writing directly on the figure the time points (48h-72h post transfection) and shortly explain in the text why was mRNA evaluated as the measure of siRNA efficacy and not the protein? Probably the antibody problem, though western-blot applicable antibodies are available.
We thank the reviewer for acknowledging that the experiments in figure 3 provide functional evidence for our hypothesis. We agree with the reviewer and for clarity have included the cell line in each panel and the time point post transfection. We now include a Western blot showing protein levels in the siRNA P2Y2 treatment (Sup. Figure 3I).
Furthermore, for providing higher impact, I would encourage the experiments to be performed (at least in part) in a PDAC cell line with epithelial morphology (eg. HPAF II or any other that expresses the P2Y2 to a reasonable level).
We agree that performing this experiment with an epithelial morphology cell line provides higher impact, hence why we performed the experiment in BxPC-3 cell lines, perhaps missed in Sup. Figure 3G and H. We now highlight that they are epithelial-like in the text.
Figure 5: By using the DNA-PAINT method, the authors demonstrated that integrin av and P2Y2 physically interact in the cell membrane over the RGD domain and these interactions are essential for ATP induced P2Y2 mediated invasion in Aspc1 cells. The performed work seems plausible, however, I leave the technical evaluation of this experiment to experts in the field.
I believe the work would benefit from a clinical/patient perspective if the authors show by immunohistochemistry in PDAC tissue samples that P2Y2 is localized at the invasive front/or metastasis. Is there a surrogate marker that can be used to label ATP rich regions in the tumor, are those regions at the invasive front? Are the P2Y2 positive cells those cells at the invasive front?
This is an interesting suggestion but immunostaining has already been performed on a large cohort of 264 PDAC patients (PMID: 30420446) and expression was consistent throughout the tumour cells.https://doi.org/10.7554/eLife.86971.sa2
Article and author information
Cancer Research UK (A27781)
- Edward P Carter
Cancer Research UK (A25137)
- Edward P Carter
- Richard P Grose
Medical Research Council (MRC0227)
- Elena Tomas Bort
Biotechnology and Biological Sciences Research Council (BB/T008709/1)
- Megan D Joseph
Royal Society (DHF\R1\191019)
- Sabrina Simoncelli
Royal Society (RGS\R2\202038)
- Sabrina Simoncelli
Medical Research Council (MR/N014308/1)
- Nicolas J Roth
Pancreatic Cancer Research Fund (Tissue Bank Grant)
- Hemant M Kocher
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We thank the Barts Pancreatic Tissue Bank (BPTB) for providing pancreatic tissue slides presented in this work. BPTB is supported by Pancreatic Cancer Research Fund and we thank all its members, in particular, Prof. Claude Chelala, Christine Hughes, Ahmet Imrali, and Amina Hughes for help, as well as Consultant Pathologist Dr. Joanne Chin-Aleong and members of Tissue Access Committee and Operations Group. We thank Dr. Ann-Marie Baker for her expertise in RNAscope experiments. This work was supported by a Medical Research Council (MRC) iCase award to PJM and RPG from Barts Charity and the MRC Doctoral Training Programme for ETB at Queen Mary University of London (Project MRC0227). NJR acknowledges the QMUL MRC Doctoral Training Program (MR/N014308/1). MDJ acknowledges support from the BBSRC (BB/T008709/1) via the London Interdisciplinary Doctoral Programme and SS acknowledges financial support from the Royal Society through a Dorothy Hodgkin fellowship (DHF\R1\191019) and a Research Grant (RGS\R2\202038). This work was supported by Cancer Research UK (CRUK) awarded to EPC and RPG (A27781) and a CRUK Centre grant to Barts Cancer Institute (A25137). Diagrams were generated using BioRender.
Senior and Reviewing Editor
- Richard M White, Memorial Sloan Kettering Cancer Center, United States
- Preprint posted: October 24, 2022 (view preprint)
- Received: February 14, 2023
- Accepted: March 17, 2023
- Accepted Manuscript published: March 21, 2023 (version 1)
- Version of Record published: April 3, 2023 (version 2)
© 2023, Tomas Bort et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.
- Page views
Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.
Downloads (link to download the article as PDF)
Open citations (links to open the citations from this article in various online reference manager services)
Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)
- Cancer Biology
In embryonal rhabdomyosarcoma (ERMS) and generally in sarcomas, the role of wild-type and loss or gain-of-function TP53 mutations remains largely undefined. Eliminating mutant or restoring wild-type p53 is challenging; nevertheless, understanding p53 variant effects on tumorigenesis remains central to realizing better treatment outcomes. In ERMS, >70% of patients retain wild-type TP53, yet mutations when present are associated with worse prognosis. Employing a kRASG12D-driven ERMS tumor model and tp53 null (tp53-/-) zebrafish, we define wild-type and patient-specific TP53 mutant effects on tumorigenesis. We demonstrate that tp53 is a major suppressor of tumorigenesis, where tp53 loss expands tumor initiation from <35% to >97% of animals. Characterizing three patient-specific alleles reveals that TP53C176F partially retains wild-type p53 apoptotic activity that can be exploited, whereas TP53P153D and TP53Y220C encode two structurally related proteins with gain-of-function effects that predispose to head musculature ERMS. TP53P153D unexpectedly also predisposes to hedgehog expressing medulloblastomas in the kRASG12D-driven ERMS-model.
- Cancer Biology
The presence of lymph node metastasis (LNM) affects treatment strategy decisions in T1NxM0 colorectal cancer (CRC), but the currently used clinicopathological-based risk stratification cannot predict LNM accurately. In this study, we detected proteins in formalin-fixed paraffin-embedded (FFPE) tumor samples from 143 LNM-negative and 78 LNM-positive patients with T1 CRC and revealed changes in molecular and biological pathways by label-free liquid chromatography tandem mass spectrometry (LC-MS/MS) and established classifiers for predicting LNM in T1 CRC. An effective 55-proteins prediction model was built by machine learning and validated in a training cohort (N=132) and two validation cohorts (VC1, N=42; VC2, N=47), achieved an impressive AUC of 1.00 in the training cohort, 0.96 in VC1 and 0.93 in VC2, respectively. We further built a simplified classifier with nine proteins, and achieved an AUC of 0.824. The simplified classifier was performed excellently in two external validation cohorts. The expression patterns of 13 proteins were confirmed by immunohistochemistry, and the IHC score of five proteins was used to build an IHC predict model with an AUC of 0.825. RHOT2 silence significantly enhanced migration and invasion of colon cancer cells. Our study explored the mechanism of metastasis in T1 CRC and can be used to facilitate the individualized prediction of LNM in patients with T1 CRC, which may provide a guidance for clinical practice in T1 CRC.