Oncogenic KRAS promotes pancreatic ductal adenocarcinoma (PDAC) through post-transcriptionally regulated KRAS-induced granules (KGs)

doi:10.21203/rs.3.rs-3064215/v1

Download PDF

Article

Oncogenic KRAS promotes pancreatic ductal adenocarcinoma (PDAC) through post-transcriptionally regulated KRAS-induced granules (KGs)

https://doi.org/10.21203/rs.3.rs-3064215/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Over 90% of pancreatic ductal adenocarcinoma (PDAC) tumors harbor mutations in KRAS, which promote many hallmark characteristics of cancer. How a single driver mutation causes the malignant properties observed in PDAC is poorly understood. We discovered that oncogenic KRAS causes an accumulation of mRNA, which drives the localization of components of the mRNA degradation machinery into novel post-transcriptional condensates termed KRAS-induced granules (KGs). Mechanistically, we find that KG formation depends on the phosphorylation of Argonaute2 at tyrosine393 and requires EGFR, but not MEK and PI3K signaling. Targeted transcriptomic analysis suggests that KGs are enriched for transcripts targeted by misregulated miRNAs, leading to decreased mRNA turnover and increased translation of pro-tumor transcripts. Using in vivo and in vitro models, we find that genetic ablation of KGs leads to delayed tumorigenesis, reduced angiogenesis, and decreased innervation. Our findings reveal an essential and therapeutically targetable role for altered post-transcriptional regulation in PDAC tumorigenesis.

Biological sciences/Cancer/Oncogenes

Biological sciences/Cell biology/Mechanisms of disease

Biological sciences/Molecular biology/Non-coding RNAs/miRNAs

Despite major advances in cancer therapy over the last 20 years, pancreatic ductal adenocarcinoma (PDAC) remains refractory to most treatments and has a dismal 5-year overall survival rate of 12%¹. Over 90% of PDAC tumors harbor driver mutations in the Kristen rat sarcoma viral oncogene homolog (KRAS)^2,3 that lead to increased activation of many pro-tumor pathways such as proliferation, metastasis, and metabolic plasticity. However, how oncogenic KRAS exerts such a wide array of alterations that lead to malignancy is still not fully understood. Mutations in KRAS drive extensive cellular dysfunction resembling other oncogenes and tumor suppressors (MYC, TP53, RB, etc.)^4,5. It is not unexpected that many mutant oncogenes and tumor suppressors can drive major metaplastic changes, as they mostly function in the nucleus to control transcription and cell cycle progression^4,6. However, KRAS is a plasma membrane-localized GTPase, so its ability to induce the hallmark characteristics of cancer represents a conundrum, as it is not a transcription factor, nor does it function in the nucleus.

One regulatory mechanism that explains the ability of KRAS to globally activate oncogenic pathways is post-transcriptional regulation by miRNAs. The intersection of post-transcriptional regulation and KRAS begins during embryonic development. Pluripotent cells exhibit remarkable cellular plasticity as they proliferate during early development. Many of the molecular changes that occur during pluripotent differentiation, such as metabolic rewiring, epigenetic remodeling, and the epithelial-to-mesenchymal transition, have been shown to depend on both KRAS and post-transcriptional regulation^7–9. Interestingly, the molecular mechanisms driving cancer mirror many of the cellular changes that occur during development, suggesting that post-transcriptional misregulation plays an integral role in driving tumorigenesis.

One critical aspect of how transcripts are regulated by miRNAs is the spatial localization of miRNA-bound mRNAs within the cell. Transcript degradation occurs most efficiently when the miRNA and degradation machinery are either freely diffusible in the cytoplasm or localized to small, dynamically motile processing bodies (PBs)¹⁰. As such, factors involved in mRNA turnover are less enzymatically active when localized to larger, static PBs, which are linked to decreased mRNA degradation efficiency ^11–13. Interestingly, several studies have shown that oncogenic KRAS misregulates aspects of the miRNA pathway, such as miRNA maturation and localization¹⁴. However, none of these studies have explained how mutant KRAS leads to the coordinated misregulation of so many pro-tumor processes. In addition, little is known about if or how oncogenic KRAS alters mRNA localization or the activity of the machinery involved in mRNA turnover.

In this report, we describe how oncogenic KRAS reduces miRNA abundance, leading to mRNA accumulation within the cell. This post-transcriptional misregulation leads to the formation of large, PB-like granules, which we term KRAS-induced granules (KGs). Mechanistically, we find that the EGFR-dependent phosphorylation of Argonaute 2 (AGO2)^Y393 drives KG formation. Furthermore, we utilize targeted transcriptional analysis to identify that KGs are enriched for upregulated tumorigenic transcripts whose repression by miRNAs is disrupted. We also find that protein levels of KG-localized transcripts are increased with mutant KRAS. This suggests that although the remaining miRNAs can still interact with their target mRNAs, disruptions in their stoichiometric ratios promote their localization to large, immobile granules, which are associated with decreased mRNA degradation efficiency. Finally, we demonstrate that inhibiting KGs leads to delayed tumor growth, reduced tumor angiogenesis, and reduced tumor innervation. Our data suggest that mutant KRAS disrupts the physiological ratios of miRNAs and their targets, leading to the sequestration of mRNA turnover machinery into KGs, where degradation is less efficient. Overall, this post-transcriptional misregulation leads to increased translation of pro-tumor transcripts.

Oncogenic KRAS induces downregulation of miRNA abundance

It has previously been shown that mutated KRAS leads to the misregulation of several miRNAs¹⁵. First, we sought to establish how mutant KRAS globally affects miRNA expression in an unbiased manner. We performed small RNA sequencing on isogenic mouse pancreatic cancer cells harboring a doxycycline (DOX)-inducible Kras^G12D mutation (mPDAC iKras^G12D)¹⁶ (Supplemental Fig. 1A) to determine differentially expressed (DE) miRNAs in Kras^G12D cancer cells compared with control. The presence of Kras^G12D led to a considerable misregulation of miRNAs, with over twice as many miRNAs significantly downregulated as upregulated (63 vs 28) (Fig. 1A, Supplemental Fig. 1B, Supplemental Table 1). Small RNA sequencing in NIH/3T3 cells ectopically expressing Kras^G12V (Supplemental Fig. 1C) showed a similar pattern, with over three times as many miRNAs significantly downregulated as upregulated (Supplemental Fig. 1D), suggesting that multiple Kras^G12 mutations are sufficient to induce miRNA misregulation. Additionally, we compared miRNA downregulation in our datasets with two published miRNA expression datasets comparing human PDAC to normal pancreatic tissue to determine whether the pattern of reduced miRNA abundance in our data was generally consistent with that seen in human PDAC tumors^17,18. Indeed, our data largely overlapped with data comparing miRNA expression in PDAC tumors vs. normal pancreatic tissue (Supplemental Fig. 1E).

Oncogenic KRAS causes an accumulation of pro-tumor mRNA transcripts

We next sought to determine whether mutant KRAS-dependent miRNA misregulation was responsible for mRNA transcript expression changes. To do this, we performed bulk mRNA sequencing on the mPDAC iKras^G12D and NIH/3T3 Kras^G12V cells along with their corresponding controls (Supplemental Fig. 1F, G). We discovered that transcripts increased 1.4-fold in relative abundance vs. decreased in the mPDAC iKras^G12D cells (960 vs. 686) (Fig. 1B) and increased 1.3-fold in relative abundance vs. decreased in NIH/3T3 Kras^G12V cells (737 vs. 562) (Supplemental Fig. 1H), consistent with our data showing that oncogenic KRAS led to an overall reduction in miRNA abundance (Supplemental Table 1). In addition, overrepresentation analysis revealed that many of the significantly overrepresented Gene Ontology biological process (GOBP) and MsigDB Hallmark pathways included those involved in tumorigenic processes including angiogenesis, metabolism, differentiation, immune response, and extracellular matrix reorganization (Fig. 1C).

Next, to further probe the functional consequences of the KRAS-induced miRNA misregulation, we cross-referenced the DE mRNA transcripts with the list of DE miRNAs. We determined that the expression patterns of the top DE transcripts and miRNAs were consistent with the notion that miRNAs negatively regulate their targets. As such, predicted targets of downregulated miRNAs increased and vice versa (Fig. 1B, Supplemental Fig. 1H). Furthermore, we identified miRNA binding sites in the DE mRNAs using the MIENTURNET (microRNA Enrichment TURned NETwork) analysis tool and performed an overrepresentation network analysis based on miRTarBase experimentally validated miRNA-mRNA interactions (Supplemental Fig. 2A)¹⁹. Overall, we found that DE mRNAs were enriched for predicted miRNA binding sites in a pattern consistent with differential miRNA expression. We further utilized LIF expression as a functional readout for miRNA misregulation, since LIF expression levels are upregulated in KRAS-mutated pancreatic cancer cell lines and negatively regulated by miRNAs¹⁶ (Supplemental Fig. 2B, C). We first confirmed that the knockdown of KRAS led to a decrease in LIF mRNA and protein expression (Supplemental Fig. 2D-F). Next, we identified that miR-26a and miR-181a (predicted to target LIF) expression levels were decreased in KRAS-mutated PDAC cell lines compared to controls (Supplementary Fig. 2G, H). The ectopic expression of miR-26a and miR-181a in PANC-1 and MIA PaCa-2 cells resulted in decreased LIF protein expression 72 hours after transfection (Supplemental Fig. 2I). Together, these data suggest that changes in miRNA expression are highly predictive of the observed changes in mRNA transcript abundance and protein levels in cancer cells.

Oncogenic KRAS leads to the subcellular reorganization of AGO2

The localization of factors involved in mRNA turnover dictates their enzymatic function. For example, decapping enzymes have been shown to be catalytically less active in PBs, further supporting that the degradation efficiency of mRNAs localized to PBs is decreased¹². In addition, abnormalities in the localization of mRNA and RNA-binding proteins have been implicated in tumorigenesis²⁰. Thus, we sought to determine if mutant KRAS led to alterations in the subcellular distribution of Argonaut2 (AGO2). Not only is AGO2 a critical component of the miRNA regulatory pathway, but the function of AGO2 has also recently been shown to be modulated by oncogenic KRAS^15,21,22. Indeed, immunofluorescence (IF) of KRAS-mutant human PDAC cells revealed a reorganization of AGO2 localization which included increased plasma membrane localization (Supplemental Fig. 3A). In KRAS^WT cells, AGO2 localizes to the endoplasmic reticulum (ER)¹⁵, as AGO2 regulates mRNA turnover at the ER²³. However, AGO2 relocalizes from the ER to the plasma membrane in mutant KRAS-dependent PDAC, which is substantiated by our data (Supplemental Fig. 4A)¹⁵. The ER also modulates the dynamics of PBs, which can alter the enzymatic function of the proteins localized to PBs^24–26. Thus, we co-stained our cells with the PB protein, DCP1A to determine if mutant KRAS alters AGO2 localization to DCP1A + foci. Interestingly, we noticed a distinct increase in AGO2 localization to a subset of foci that co-stained with DCP1A (Fig. 1D, E).

Oncogenic KRAS leads to the formation of distinct post-transcriptional granules

Recent evidence suggests that granule size is also associated with decreases in degradation efficiency. One proposed mechanism explaining this is that mRNAs localized to larger, immobile PBs are more stable than when localized to small PBs, where they are more accessible to degradation^13,27. Thus, we next sought to establish if mutant KRAS altered post-transcriptional granule dynamics, such as their size and spatial mobility. In addition to increased AGO2 localization to DCP1A + foci (Fig. 1D, E), we also discovered that the AGO2 + DCP1A + foci in KRAS-mutant cells were significantly larger than any DCP1A + foci observed in KRAS^WT cells (Fig. 1D, F). In KRAS^WT human PDAC (BxPc-3), mPDAC, and NIH/3T3 cells, the area of individual DCP1A + foci was generally less than 400 nm² (Fig. 1G, Supplemental Fig. 3B, C). However, in KRAS^G12V, KRAS^G12D, and KRAS^G12C cells, there was also a significant population of DCP1A + foci larger than 400 nm², which we defined as KRAS-induced granules (KGs) (Supplemental Fig. 4B, C). In addition, the DCP1A + foci less than 400 nm² in KRAS-mutant cells were still on average, larger than those in KRAS^WT cells, suggesting that degradation efficiency may also be decreased in other foci aside from KGs. We found that the size and number of KGs in KRAS-mutant cells were significant enough to be reflected by quantifying the average size of foci across cells (Fig. 1H). The presence of KGs was also observed in NIH/3T3 cells ectopically expressing Kras^G12V (Supplemental Fig. 3D) and mPDAC iKras^G12D cells (Supplemental Fig. 3E) suggesting that the formation of KGs depends on KRAS mutations and not other confounding mutations found in human PDAC cell lines. We also co-stained with other canonical PB markers and identified that KGs also contain EDC4, DDX6, LSM14A, and 4E-T (Fig. 1I, Supplemental Fig. 3F-I). Next, we performed live imaging on clonal mPDAC iKras^G12D cells stably expressing low levels of EGFP-DCP1B (another canonical PB protein) to investigate the spatial mobility of KGs. Of note, clones with low expression were chosen to prevent any interruption to granule formation and DCP1b-GFP + foci also stained positive for other PB/KG markers (Supplemental Fig. 3M) 72 hours after induction of Kras^G12D, we observed the formation of KGs, which were nearly completely static and exhibited almost no movement over the course of 5 minutes. However, smaller, nearly sub-microscopic foci were more dynamically active (Supplemental Video 1). Together, these data support our hypothesis that mutant KRAS leads to altered post-transcriptional granule dynamics, indicative of a decrease in mRNA repression by miRNAs.

GW182 has been shown to promote the aggregation of miRNA machinery (such as AGO2) into PBs. This is likely due to the ability of GW182 to bind multiple AGO2 proteins, which can promote the stability of mRNA targets^28,29. Thus, we next examined the subcellular distribution of GW182. Indeed, GW182 showed increased localization into KGs. Since transcript degradation has been shown to occur more efficiently when degradation machinery is not localized to large foci¹⁰, these data suggest that KGs are more likely sites of mRNA storage as opposed to sites of active mRNA degradation. (Supplemental Fig. 3J). Additionally, DICER, another member of the miRNA processing machinery, has been shown to passively bind mRNAs, promoting their localization to PBs³⁰. We hypothesized that increased mRNA abundance would increase the passive binding of DICER to mRNA, and thus localization to KGs. Co-staining of KRAS-mutant cells with DICER and EDC4 indeed showed increased localization of DCER to KGs (Supplemental Fig. 3K) Finally, we aimed to determine if KGs were distinct from stress granules (SGs), another type of post-transcriptional condensate that is involved in therapy resistance in PDAC^31,32. To do this, we treated mPDAC iKras^G12D cells with sodium arsenite (NaAsO2), which is a potent SG inducer, and then costained for DDX6 and G3BP1, to mark PBs and SGs, respectively. Indeed, SGs and KGs occupied separate subcellular areas, indicating that KGs are a distinct class of granules from SGs (Supplemental Fig. 3L).

Next, we wanted to investigate if KG formation in 2D culture can recapitulate what occurs in vivo in 3D tumor samples. Thus, we assayed mouse and human PDAC tumors for the formation of KGs. Indeed, we identified KGs in tumors harvested from LSL-Kras^G12D/+;LSL-Trp53^R172H/+;Pdx-1-Cre (KPC) mice³³ (Supplemental Fig. 4A-C). Finally, we identified the presence of KGs in sequencing-confirmed (Tempus) KRAS-mutated human PDAC tumor samples (Supplemental Table 3) to substantiate both the in vitro and mouse data in a human PDAC tumor context (Supplemental Fig. 4D-F).

KG assembly requires oncogenic KRAS and miRNA reduction

To formally test the dependence of KG formation on mutant KRAS, we utilized a KRAS^G12C-specific small-molecule inhibitor, ARS-1620. After 12 hours of treatment, inhibition of KRAS^G12C in MIA PaCa-2 cells (measured by p-ERK inhibition) (Fig. 2A) prevented KG formation, demonstrated by the disappearance of large DCP1A + foci and decreased DCP1A + AGO2 + co-localization (Fig. 2B-D). However, treatment of KRAS^G12D mutated PANC-1 and KRAS^WT BxPc-3 cells showed no difference in DCP1A + foci compared to DMSO controls, supporting that specific inhibition of mutant KRAS is responsible for KG ablation (Supplemental Fig. 5A, B). Additionally, we performed siRNA-mediated knockdown of Kras in NIH/3T3 Kras^G12V and NIH/3T3 Kras^WT cells. Knockdown of Kras in Kras^G12V cells (Supplemental Fig. 5C) prevented KG formation (Supplemental Fig. 5D, E), but Kras knockdown in Kras^WT (Supplemental Fig. 5F) led to no difference in DCP1A + foci, suggesting that during homeostasis, PBs are not altered by the loss of KRAS (Supplemental Fig. 5G). We also sought to determine if KG formation depended on miRNA downregulation seen with KRAS mutations. To test this, we performed siRNA-mediated knockdown of DGCR8, which functionally recapitulates global miRNA downregulation³⁴. With DGCR8 knockdown in KRAS^WT BxPc-3 (Fig. 2E) and NIH/3T3 cells (Supplemental Fig. 5H), we observed a similar induction of KG-like foci (Fig. 2F and Supplemental Fig. 5I) whereby DGCR8 knockdown led to a 2-fold increase in the average DCP1A + foci size (Fig. 2G). Interestingly, DGCR8 knockdown in cells harboring a KRAS mutation (MIA PaCa-2 and NIH/3T3 Kras^G12V) lead to the appearance of even larger KGs (Supplemental Fig. 5J-M). Together, these data demonstrate that mutant KRAS is required for KG formation and that the induction of KGs occurs due to the miRNA downregulation caused by oncogenic KRAS.

Phosphorylation of tyrosine 393 in AGO2 is required and sufficient for KGs

AGO2 function is known to be modulated by the phosphorylation of multiple residues, and serine 387 and tyrosine 393 have been specifically implicated in the miRNA function of AGO2^35,36. Indeed, p-Ago2^S387 and p-Ago2^Y393 are both increased upon the induction of Kras^G12D (Fig. 2H). Because these phosphorylation sites are critical for miRNA function and because we see altered AGO2 localization in KGs, we hypothesized that AGO2 phosphorylation is critical for KG formation. To determine if phosphorylation of Ago2 at either residue is required for KG formation, we ectopically expressed AGO2^WT as well as both phospho-mimic and phospho-deficient mutants of S387 (S387E and S387A) and Y393 (Y393E and Y393A) in the mouse PDAC iKras^G12D cell line (Supplemental Fig. 6A). After the generation of clonal lines, we determined that neither AGO2^WT nor AGO2^S387E mutants were sufficient to induce KGs (Fig. 2I-J, Supplemental Fig. 6B, C). However, AGO2^Y393E mutants phenocopied KG formation despite the absence of mutant KRAS (Fig. 2I-J, Supplemental Fig. 7B, C). Additionally, AGO2^S387A phospho-deficient mutants did not affect foci size (Supplemental Fig. 6D, E). However, ectopic expression of phospho-deficient AGO2^Y393A prevented the formation of KGs upon activation of oncogenic KRAS (Supplemental Fig. 7D, E). Moreover, we found that only AGO2^Y393E mutants, and not WT or AGO2^S387E mutants were sufficient to induce cell proliferation (Supplemental Fig. 6F, G), cell migration (Supplemental Fig. 6H, I), and tumorsphere formation (Supplemental Fig. 6J, K). These data suggest that increased phosphorylation of AGO2^Y393 and downstream effects of p-AGO2^Y393 (such as KG formation) are, at least in part, responsible for specific pro-survival and metastatic characteristics of oncogenic KRAS-driven pancreatic tumors.

KGs require EGFR but not MEK and PI3K signaling

AGO2^Y393 is phosphorylated by EGFR after KRAS-dependent localization of AGO2 to the plasma membrane³⁷. To test if KG formation is dependent on EGFR signaling (and thus EGFR-dependent AGO2^Y393 phosphorylation) we treated MIA PaCa-2 cells with Gefitinib, an EGFR inhibitor (EGFRi) known to prevent p-AGO2^{Y393 35}. Indeed, EGFRi (Fig. 3A) prevented KG formation in MIA PaCa-2 cells, suggesting that KG formation occurs downstream of EGFR-dependent p-AGO2^Y393 (Fig. 3B, C). Although EGFRi did not prevent AGO2 localization to the membrane where it presumably interacts with EGFR, EGFRi did prevent AGO2 localization to KGs (Fig. 3B). Additionally, the treatment of AGO2^Y393E phospho-mimic lines with EGFRi and KRpep-2d, a KRAS^G12D inhibitor (Fig. 3D) failed to inhibit KG formation, further substantiating that mutant KRAS- and EGFR-dependent phosphorylation of AGO2^Y393 are both necessary and sufficient for KG formation (Fig. 3E, F).

Downstream KRAS effectors ERK and AKT have also been shown to play a role in regulating AGO2 function^{36, 38–41}. However, inhibiting ERK and AKT signaling (Fig. 3G, H) did not alter KG formation (Fig. 3I, J), further supporting our hypothesis that KGs are distinct, PB-like condensates that form independent of downstream MAPK and PI3K signaling pathways.

KGs are enriched for oncogenic transcripts

Although we identified that oncogenic KRAS reduces miRNAs abundance, miRNAs are not completely lost and thus, may still interact with their targets. However, the stoichiometric ratios of proteins and RNAs that condense into RNP granules dictate the enzymatic activity of proteins in condensates²⁴. We hypothesize that a fraction of transcripts with increased abundance due to the loss of miRNAs are targeted by the miRNAs that are still present. Targeting by miRNAs leads these transcripts to be localized to KGs, where they are stored instead of actively degraded. Thus, the net result is a feed-forward loop of increased translation due to elevated levels of mRNAs and decreased efficiency of mRNA degradation due to the localization of transcripts into KGs. To identify transcripts enriched in KGs, we aimed to utilize a proximity-based labeling technique. Although multiple proximity-based labeling protocols have been recently published, certain limitations, including the necessity for exogenous protein expression, still exist^42,43. Thus, we devised tyramide

signal amplification RNA-sequencing (TSA-RNA-seq) as a straightforward technique that allows for the isolation of sub-cellular RNA populations without any genetic manipulation or in vivo chemistry that might alter biology (Fig. 4A)⁴⁴. After optimizing fixation and labeling conditions for foci vs cytoplasmic specificity and RNA quality in MIA PaCa-2 and BxPc-3 human PDAC cells (Supplemental Fig. 7A-H), we were able to specifically label condensate-resident transcripts with biotin for streptavidin enrichment (Fig. 4B). After confirmation of RNA biotinylation and isolation of intact RNA (Supplemental Fig. 7I, J), biotinylated RNA was enriched and sequenced. Differential enrichment analysis revealed that KGs are transcriptionally distinct from foci in KRAS^WT cells with over 2000 transcripts significantly enriched in KGs (Fig. 4C, Supplemental Table 1). GOBP and MsigDB Hallmark analysis revealed that transcripts involved in mutant-KRAS-dependent processes in PDAC were significantly overrepresented in the KG-enriched transcript set. For example, many of the significantly overrepresented GOBP pathways included those involved in cell proliferation, WNT signaling, and metabolism (Fig. 4D, Supplemental Table 4). MsigDB Hallmark analysis similarly showed an overrepresentation of pathways involved in cancer, such as the epithelial-to-mesenchymal transition (EMT), translation, mTORC1 signaling, and cell cycle progression (Fig. 4E, Supplemental Table 4). Interestingly, most of the top 10 most significantly enriched transcripts have been shown to play fundamental roles in promoting mutant KRAS-specific characteristics of PDAC, further supporting the overrepresentation analysis in Fig. 4D and E (Fig. 4F). In parallel, TSA-RNA-seq was conducted in NIH/3T3 Kras^G12V KGs v Kras^WT cells (Supplemental Fig. 8A). Similarly, we observed a distinct KG-enriched transcriptome, with miRNA-targeted pro-tumor transcripts localized to KGs (Supplemental Fig. 8B-E).

To validate our results, we utilized single-molecule inexpensive RNA fluorescence in situ hybridization (smiRNA-FISH) to visualize transcripts localizing to KGs. After using previously published probes known to localize to DCP1A + foci in cells as well as negative controls to ensure a specific signal (Supplemental Fig. 8F), we probed for SPP1, one of the most differentially enriched KG transcripts identified in Fig. 4H. Indeed, we detected significant enrichment of SPP1 transcripts in KGs compared with DCP1A + foci in KRAS^WT cells (Supplemental Fig. 8G-I).

KG mRNAs are targets of downregulated miRNAs and encode proteins upregulated by mutant KRAS

Next, we asked whether KG-enriched transcripts were targets of miRNAs with reduced abundance. To answer this, the top 500 most abundant KG-enriched transcripts (p [adj] > 0.05 & LOG2FC > 1) were included for MIENTURNET miRNA-target network analysis. Indeed, transcripts predicted to be targeted by miRNAs with lower abundance were over-represented in the KG transcriptome (Fig. 4G). These data further support our hypothesis that a fraction of accumulated transcripts in KRAS-mutant cells are being targeted to KGs, where they are protected from degradation. To further delineate this, we performed bulk RNA-sequencing of ΔKG NIH/3T3 Kras^G12V (Supplemental Fig. 8J), which revealed an increase in the number of downregulated vs upregulated transcripts (1247 vs. 1104) (Supplemental Fig. 8K) compared with controls. These data suggest that interruption of KGs restores some level of mRNA degradation.

Next, to determine whether protein levels of KG-enriched transcripts are increased, we analyzed protein expression levels of SPP1 after the induction of mutant KRAS. SPP1 encodes the protein osteopontin (OPN), a secreted phosphoprotein whose expression is a negative prognostic factor for survival in The Cancer Genome Atlas Pancreatic Adenocarcinoma patient cohort (TCGA PAAD) (overall survival HR = 1.94, log-rank P = 0.002 and regression-free survival HR = 3.76, log-rank P = 0.00078) (Fig. 4H)^45,46. After 72 hours of DOX induction, western blot analysis for SPP1/OPN in mPDAC iKras^G12D cells revealed nearly a 2-fold increase in OPN protein expression with mutant KRAS (Fig. 4I). Lastly, we correlated our western blot data with IF imaging of KRAS-mutant human PDAC tumors, which also revealed a significant increase in SPP1/OPN expression in human PDAC tumors compared with healthy control tissue (Fig. 4J, K).

KG inhibition delays tumor growth

We have established that mutant KRAS increases pro-tumor protein levels presumably by decreasing miRNA repression and promoting KG formation. We also identified that p-AGO2Y³⁹³ (which induces KGs) promotes cancer cell proliferation and migration (Supplemental Fig. 6), thus we hypothesize that disrupting KGs would decrease tumor growth and severity. To inhibit KG formation without simultaneously disrupting PBs, we ectopically expressed the PB-resident protein DCP1A. DCP1A is suggested to be a scaffolding protein that acts as a binding platform for multiple PB proteins and the ectopic expression of DCP1A does not affect PB function or the ability of cells to proliferate or migrate^47–50. Thus, we ectopically expressed mNeonGreen-tagged DCP1A (DCP1A-mNG) in MIA PaCa-2 (Fig. 5A), PANC-1, BxPc-3c, NIH/3T3 Kras^WT and NIH/3T3 Kras^G12V cells (Supplemental Fig. 9A). Indeed, ectopic expression of DCP1A in KRAS^MUT cells inhibited KGs (ΔKG) (Fig. 5A and Supplemental Fig. 9C, E) without changing the structure of other condensates in KRAS^WT cells. (Supplemental Fig. 9D, F). We next wanted to determine whether the genetic ablation of KGs would prevent tumor growth in vivo. To do this, we utilized a heterotopic tumor model where ΔKG and control MIA PaCa-2 cells were engrafted into SCID/Bg mice. Indeed, ΔKG MIA PaCa-2 tumors grew significantly slower than control tumors (Fig. 5C), leading to a significant increase in survival (Fig. 5D).

Inhibition of KGs prevents tumor vascularization and innervation

Upon harvest, ΔKG tumors also exhibited more regular tumor borders (Fig. 5B), less necrosis, less vascularity, and less innervation compared with control tumors (Fig. 5E, F). To further delineate how KGs promote PDAC tumors, we quantified cell proliferation in ΔKG and control MIA PaCa-2, PANC-1, and BxPc-3 cell lines as well as ΔKG and control NIH/3T3 Kras^G12V cell lines. The disruption of KGs led to a 2-fold decrease in proliferation in MIA PaCa-2 cells (Supplemental Fig. 9B), a 1.5-fold decrease in proliferation in PANC-1 cells (Supplemental Fig. 9C), but no significant difference in BxPc-3 cells (Supplemental Fig. 9D). Similarly, KG disruption in NIH/3T3 Kras^G12V cells led to a 3-fold decrease in proliferation (Supplemental Fig. 9E), but no significant difference in NIH/3T3 Kras^WT cells (Supplemental Fig. 9F). The disruption of KGs also led to a marked decrease in cell motility of nearly 2-fold in MIA PaCa-2 cells (Supplemental Fig. 9G) and NIH/3T3 Kras^G12V cells (Supplemental Fig. 9H). These data suggest that KRAS-mutant cells depend on KGs for their survival advantage and metastatic capability. Thus, inhibiting KGs or upstream factors driving KG formation may represent a novel avenue for treatment development in the future.

Although increasing evidence suggests that post-transcriptional misregulation contributes significantly to malignancy, it is still an underappreciated aspect of PDAC. Here, we identify novel, PB-like granules in pancreatic cancer cells termed KRAS-induced granules (KGs). PDAC cells depend on KGs to drive various aspects of tumor propagation, thus they represent a previously unidentified vulnerability in PDAC. We identify p-AGO2^Y393 as a critical mediator of KG formation and propose that pro-tumor transcripts that lose miRNA regulation downstream of mutant KRAS accumulate in KGs. The formation of KGs represents a feedforward mechanism of reducing miRNA regulation, thus leading to increased translation in PDAC (Fig. 5G). Interestingly, SGs, another type of post-transcriptional granule, have previously been shown to be upregulated by oncogenic KRAS and play a role in driving cancer cell survival and therapy resistance in PDAC^31,32. However, SGs are not present in KRAS^MUT cells without additional stressors, such as cytotoxic therapeutics. Thus, KGs, which are present in KRAS^MUT cells during otherwise homeostatic conditions, may analogously provide a survival advantage for pancreatic cancer cells akin to SGs during the stress response.

Mutant KRAS exerts a profound influence on the tumor microenvironment (TME) in PDAC, contributing to disease progression and therapy resistance. One way mutant KRAS has been shown to do this is by promoting translation to drive cellular plasticity⁵¹. Thus, we speculate that transcripts stored in KGs, which are enriched in tumor-promoting pathways known to be activated in KRAS-mutant PDAC, may act to support the increases in translation and protein synthesis that are known to occur in PDAC⁵², especially since some of the changes in the tumor microenvironment, metabolism, and immune signaling have been shown to occur very rapidly⁵³. One area where this is of particular interest includes the exceptionally immunosuppressive TME in PDAC, which is in part, due to a high degree of infiltration of suppressive immune cells, such as regulatory T cells (Tregs), myeloid-derived suppressor cells (MDSCs), and tumor-associated macrophages (TAMs)^54–56. These immune cell populations exhibit immunosuppressive properties and actively suppress the activity of cytotoxic T lymphocytes (CTLs) and other immune effector cells, preventing efficient tumor clearance⁵⁷. Furthermore, oncogenic KRAS influences the stromal components of the TME, leading to the deposition of a dense extracellular matrix and the activation of cancer-associated fibroblasts (CAFs), which provide physical and biochemical support for tumor growth and metastasis^58,59. This suppressive TME also contributes to the failure of PDAC to respond to immunotherapeutics such as checkpoint inhibitors, such as PD-1 (Programmed Cell Death Protein-1, PD-L1 (Programmed Cell Death Ligand-1), and CTLA-4 (Cytotoxic T Cell Antigen-4) inhibitors. Checkpoint inhibitors have revolutionized cancer treatment in various liquid and solid malignancies yet remain ineffective against PDAC^60,61.

Our data have revealed a link between KGs and the immunosuppressive milieu within the PDAC TME. Many of the most abundant KG-enriched transcripts, including PCOLCE, SPP1, POTEF, S100A4, LGALS1, SPARC, POTEE, VIM, TPM1, and FN1 (Fig. 4), encode factors that have been implicated in promoting a suppressive TME. Most interestingly is the enrichment for secreted factors in this gene list, including PCOLCE (procollacen-C endopeptidase enhancer), SPP1 (secreted phosphoprotein 1/ osteopontin), S100A4 (s100 calcium binding protein A4), LGALS1 (galectin-1), SPARC (secreted protein acidic and cysteine rich), and FN1 (fibronectin 1). PCOLCE, LGALS1, SPARC, and FN1 all encode glycoproteins that have known roles in the PDAC TME. For example, galectin-1 has been implicated in promoting immunosuppression by inhibiting T cell activation, inducing Treg expansion, and suppressing antitumor immune responses^62–64. SPARC has also been shown to induce immunosuppression, as well as PDAC cell proliferation and metastasis^65–67. Other secreted factors, such as SPP1, promotes the recruitment and immunosuppressive functions of MDSCs^68–71, with other roles in promoting proliferation⁴⁶, and metastasis⁷². The enrichment of these transcripts within KGs suggests a decrease in their miRNA regulation, thus increasing their translation and secretion to promote immunosuppression in the PDAC TME.

Another interesting observation from our heterotopic tumor experiments was that inhibiting KGs led to a significant decrease in intratumoral innervation (Fig. 5). In PDAC, intratumoral innervation comprises another critical aspect of the TME and includes a complex network of sensory, sympathetic, and parasympathetic nerve fibers⁷³. Studies have demonstrated that PDAC tumors exhibit a high degree of innervation compared to normal pancreatic tissue, which is associated with increased tumor aggressiveness, metastatic potential, and resistance to therapy^73,74. Intratumor innervation is also tightly linked with the metabolic plasticity exhibited by PDAC tumors. Oncogenic KRAS can upregulate autophagy and other scavenging pathways to utilize metabolites, such as serine, that are secreted from surrounding neurons to support PDAC cells in a metabolically limited environment⁷⁵. Interestingly, genes involved in both neuron recruitment (axonogenesis, neuron migration, etc.) and serine biosynthesis are also significantly enriched within KGs (Supplemental Table 4). Therefore, PDAC cells may use KGs to inhibit the turnover of neurotrophic factors, leading to increased secretion and thus, increased intratumoral innervation. In addition, KGs may also represent an adaptive mechanism for PDAC cells to promote their swift adjustment to metabolically hostile conditions.

Despite these promising results, our study is not without some limitations. Our TSA-RNA-seq data set does utilize a common marker between PBs and KGs to perform our comparison. Because PBs are still present in KRAS^MUT cells, we are unable to fully eliminate PB-resident transcripts from our KG-enriched dataset. However, we have optimized our labeling protocol for larger foci (Supplemental Fig. 14) and our differential analysis does normalize between samples, which should minimize any PB resident transcripts that are present in KRAS^WT and KRAS^MUT cells. In addition, questions remain regarding how KG-localized transcripts are regulated. Although the miRNAs targeting these transcripts are downregulated, they are not completely absent. The relative decrease in miRNA abundance may lead to stoichiometric changes in miRNA-transcript interactions, which could alter the recruitment or binding of other RNA-binding proteins. miRNAs can also target many transcripts; thus, another possibility is that decreases in the relative expression of certain miRNAs may act as a limiting reagent, and only allow for the regulation of transcripts with higher affinity for miRNA binding.

From a therapeutic perspective, our results identify a three-pronged approach for future interventions. First, KGs represent a previously unidentified vulnerability in KRAS-mutant PDAC. Because of the distinct phenotype induced upon acquisition of an activating KRAS mutation, high-throughput screening for molecules that may inhibit KG formation could be a simple strategy to identify previously unknown compounds that may be able to be used alone or in combination to treat PDAC. Additionally, our data reveal that targeting AGO2^Y393 phosphorylation may also be a potential therapeutic avenue, especially for use in combination therapy, which is proving to be necessary for treating PDAC. The use of in silico cheminformatics platforms has increased over recent years to help identify novel small molecule inhibitors that may be able to target specific residues in AGO2. Finally, our data also reveal a subset of spatially regulated transcripts that may unveil novel, targetable pathways for future therapeutic consideration.

Cell culture

All cells were maintained at 37°C in a humidified incubator containing 5% (v/v) CO2. Human pancreatic cancer cell lines BxPc-3, MIA PaCa-2, and PANC-1 were obtained from and cultured as indicated by American Type Culture Collection (ATCC). PaTu-8902 was obtained from DSMZ Germany. NIH/3T3 cell lines (controls and Kras^G12V) and the isogenic mouse iKras^G12D pancreatic cancer cell lines were a kind gift from Dr. Man-Tzu Wang (EpiBiologics). All cell lines were cultured in DMEM/F12 medium with 10% fetal bovine serum and 1X GlutaMAX™ (ThermoFisher). To generate mPDAC iKRAS AGO2 mutants, site-directed mutagenesis (SDM) of pLJM1-AGO2 was used to generate AGO2^S387A, AGO2^S387E, AGO2^Y393A, and AGO2^Y393E (Supplemental Table 4). mPDAC iKRAS cells were then transduced with AGO2 mutant constructs and then selected with puromycin (10ug/mL). mPDAC iKras^G12D cells that had integrated the mutant constructs were obtained by single-cell plating and clonal selection based on protein overexpression measured by western blot. All iKras^G12D expression experiments were performed on day 3 of doxycycline (2 µg/ml) induction, except for sequencing data, which was performed on cells maintained over time in doxycycline. To generate cells stably overexpressing DCP1A-mNG, NIH/3T3 KRAS^WT, NIH/3T3 Kras^G12V, BxPc-3, PANC-1, and MIA PaCa-2 cell lines were transduced with mNEONGREEN-DCP1A-pHAGE-pgk-cFLAG-HA lentiviral expression vectors and then selected with puromycin (10ug/mL). Cells with integrated constructs were obtained by single-cell plating and clonal selection based on protein overexpression measured by western blot.

RNA isolation

Total RNA was isolated from cell lines in 1 ml of Trizol (Invitrogen, Carlsbad, CA) and purified using the miRNeasy micro kit according to the manufacturer’s instructions (Qiagen, Germantown, MD). RNA concentration was determined using the DeNovix RNA Assay kit (DeNovix Inc., Wilmington, DE), and RNA integrity was evaluated using the high-sensitivity RNA screen tape with the Agilent 2200 TapeStation System according to the manufacturer’s instructions (Agilent Technologies). RNA samples with a RIN ≥ 8 were included in the study for bulk mRNA and miRNA sequencing, and > 6 for TSA-RNA-sequencing.

cDNA library preparation and sequencing

Small RNA libraries were generated using NEXTFlex® Small RNA Library Prep Kit v3 (Cat #5132–06) according to the manufacturer’s instructions. Briefly, for each sample, 1 ug of total RNA was reverse transcribed into cDNA, which was amplified with 15 cycles of PCR. Then the libraries were purified by running them on a 5% Criterion™ TBE polyacrylamide gel for 30 min at 200V. Bands between 150 to 170 bp were excised from the gels and purified. Library concentration was determined using the DeNovix dsDNA High Sensitivity Assay kit (DeNovix Inc., Wilmington, DE). Library size was determined using the Agilent TapeStation System according to the manufacturer’s instructions (Agilent Technologies) and subsequently sequenced on the Illumina NextSeq 550 (Illumina, Inc.).

mRNA libraries were generated using the TruSeq® stranded mRNA Library Prep Kit (Illumina, Inc., Cat 20020594) according to the manufacturer’s instructions. Briefly, for each sample, 1 ug of total RNA was reverse transcribed into cDNA, which was amplified with 15 cycles of PCR. Then the libraries were then purified using Clean NGS beads. Library concentration was determined using the DeNovix dsDNA High Sensitivity Assay kit (DeNovix). Library size was determined using the Agilent TapeStation System according to the manufacturer’s instructions (Agilent Technologies, Palo Alto, CA) and subsequently sequenced on the Illumina NextSeq 550 (Illumina, Inc. San Diego, CA).

miRNA and mRNA sequencing data analysis

mRNA:

Fastq files were downloaded using the BaseSpace™ Sequence Hub downloader (Illumina, Inc.). SmallRNA adapters were trimmed using cutadapt⁷⁶ (version 3.5), and low-quality sequences were removed before data analysis. Sample quality and read numbers were assessed using FastQC⁷⁷. Sequences > = 16 bp and < = 35 bp length were considered for further analysis. The STAR package⁷⁸ (version 2.7.9a) was used to align reads to a reference genome (GRCm39) using default parameters. The GenomicAlignments package⁷⁹ (version 1.28) was used to determine read counts, and the DESeq2 package⁸⁰ (version 3.14) was used to determine DE miRNAs with Benjamin-Hochberg correction. miRNAs were considered DE if the p.adj value was < 0.05.

mRNA:

Fastq files were downloaded using the BaseSpace Sequence Hub downloader. Low-quality sequences were removed before data analysis. Sample quality and read numbers were assessed using FastQC⁷⁷. The STAR package⁷⁸ (version 2.7.9a) was used to align reads to a reference genome (GRCm39) using default parameters. The GenomicAlignments package⁷⁹ (version 1.28) was used to determine read counts and the DESeq2 package⁸⁰ (version 3.14) was used to determine DE mRNAs. The ComBat⁸¹ package (version 3.42) was used to correct for any batch effect. Gene Ontology (GO) analysis and mSigDB Hallmark were performed using Enrichr^82–84. miRNA targets were identified using TargetScanMouse 7.2⁸⁵. Visualization of bioinformatic analysis was performed using ggplot2⁸⁶ and GraphPad Prism version 9.2.0 for Mac OS X (GraphPad Software, San Diego, CA).

Immunohistochemistry

Cells were grown on poly-d-lysine coated coverslips, fixed, and immunostained with antibodies for the established PB markers DDX6, EDC4, and DCP1A, 4E-T, and LASM-14A as previously described²⁵.

In mouse tissue: KPC or control mice were euthanized and pancreata were excised, fixed in 4% formaldehyde for 1 hr at RT, and embedded in O.C.T. (Sakura Finetek, Torrance, CA). Pancreata were then sectioned at 10 µM thickness and slides were stored at − 80°C until staining. For immunofluorescence, slides were warmed to RT, then washed three times with 0.1% Triton-X in PBS (PT) for 5 minutes each. Blocking (5% goat serum, 1% BSA in PBS) was done for 1 hr at RT. The primary abs were centrifuged at 15,000 x g for 5 min at 4°C before being diluted to a concentration of 10 µg/mL and incubated overnight at 4°C. After incubation, slides were washed 5 times with PT for 5 min, then incubated for 2 hr at RT with fluorescently conjugated secondary antibodies. Slides were washed 5 times with PT for 5 min and incubated in DAPI (1 µg/mL, Sigma) for 10 min. After three 5‐min washes with PT, slides were mounted and images were visualized on an inverted fluorescent microscope (LSM 990 confocal microscope; Zeiss, Thornwood, NY, USA).

In human tissue: Sections (10 µm) were cut from formalin-fixed paraffin-embedded samples for immunostaining. A total of 6 PDACs (KRAS^MUT), and 4 normal tissues adjacent to PDACs were analyzed. Hematoxylin and eosin (H&E) staining was conducted according to standard protocols. For immunofluorescence, slides were deparaffinized and sodium citrate antigen retrieval was conducted according to standard protocols. Slides were then blocked (5% goat serum, 1% BSA in PBS) for 1 hr at RT. The primary abs were centrifuged at 15,000 x g for 5 min at 4°C before being diluted to a concentration of 10 µg/mL and incubated overnight at 4°C. After incubation, slides were washed 5 times with PT for 5 min, then incubated for 2 hr at RT with fluorescently conjugated secondary antibodies. Slides were washed 5 times with PT for 5 min and incubated in DAPI (1 µg/mL, Sigma) for 10 min. After three 5-min washes with PT, slides were mounted and imaging was performed using an LSM-900 confocal microscope system (Zeiss), using a 63X Objective from Zeiss. Serial optical sections were captured every 0.2-µm and processed using the ZenBlue software and projected into one maximum intensity projection picture (Zeiss). Images from each data set were acquired on the same day using the same exposure times.

Condensate quantification

Condensates were quantified using Image J (http://rsb.info.nih.gov/ij). Briefly, images were randomly acquired in 4 different fields across at least three separate experiments with at least 5 cells/field. Background intensity was subtracted from each image. The plug-in “Analyze Particles” was used to measure the average size of condensates and the number of condensates/cells. The cell area was delineated and measured through Image J selection and measure area tools.

Small molecule inhibitor treatments

For KRAS^G12C inhibition, cells were treated with either vehicle alone or ARS-1620 (Selleck Chemicals, Houston, TX) for 12 hours at 10 uM (10 mM stock dissolved in DMSO). For tyrosine kinase inhibition, cells were treated with either vehicle alone or Gefitinib (Sigma-Aldrich, St. Louis, MO) for 24 hours at 1uM (1 mM stock dissolved in DMSO). For MEK inhibition, cells were treated with either vehicle alone or U0126 (Selleck Chemicals) for 24 hours at 20 uM (20 mM stock dissolved in DMSO). For PI3K inhibition, cells were treated with either vehicle alone or LY294002 (Sigma-Aldrich) for 24 hours at 20 uM (20 mM stock dissolved in DMSO).

Cell proliferation and cell tracking

For cell proliferation and tracking assays, cells were seeded at a density of 2000 cells/well in 24-well plates. Time-lapse imaging was performed every 30 minutes for 40 hours with a Lumascope LS720™ inverted microscope (Etaluma, Inc., Carlsbad, CA) using a 10X Phase UPlan Fluorite objective (Olympus). Analysis of cell growth and cell tracking was performed using the cell proliferation and phase cell tracking recipes, respectively, with Lumaquant™ imaging software (Etaluma).

Tumorsphere assay

Cells were harvested, counted, and seeded onto Ultra-Low Attachment Culture 96-well plates (Corning Life Science, Catalog number 3261) at 100 per well in DMEM/F12 with 1% FBS and 10% matrigel. The number of formed spheres was counted 3 weeks after seeding.

siRNA gene knockdown

The siRNA oligos targeting KRAS and DGCR8 were purchased from ThermoFisher. siRNA IDs: KRAS s68936. DGCR8. Scramble cat# AM4611. According to the manufacturer's protocol, cells were reverse transfected with 50 pM of control siRNAs, KRAS or DGCR8 siRNA with Lipofectamine RNAiMAX (Invitrogen). Cells were harvested for protein extraction two days after siRNA transfection to confirm successful siRNA knockdown. Cells were simultaneously transfected and plated on poly-d-lysine coated glass coverslips in 24-well plates for imaging.

Single-molecule inexpensive RNA-FISH

KRAS^WT and KRAS^MUT cells were plated on glass coverslips, fixed, and prepared for hybridization as previously published⁸⁷. Briefly, probe/fluorescent probe hybridized duplexes were incubated on cells at 37°C overnight, then washed and co-stained with PB-specific primary abs/ species-specific secondary abs conjugated to AlexaFluor-546. Cells were co-stained with DAPI and mounted and imaged on the LSM 900 confocal microscope.

Constructs

pLJM1-AGO2 phospho-deficient and phospho-mimic vectors were generated using site-directed mutagenesis of pLJM1-FH-AGO2-WT, which was a gift from Dr. Joshua Mendell (the University of Texas Southwestern Medical Center)³⁶ (Addgene plasmid # 91978; http://n2t.net/addgene:91978 ; RRID:Addgene_91978). The constructs encoding EGFP-DCP1B-pHAGE-pgk-cFLAG-HA and mNEONGREEN-DCP1A-pHAGE-pgk-cFLAG-HA were gifts from Dr. Jason Lee. The Flag-tagged full-length or mutated AGO2 and EGFP-DCP1B-pHAGE-pgk-cFLAG-HA and mNEONGREEN-DCP1A-pHAGE-pgk-cFLAG-HA lentiviral expression vectors were used to establish stable cell lines with the packaging plasmids pMD2G and psPAX2.

Human PDAC samples

The use of human tissue was reviewed and approved by the Institutional Review Board of Baylor College of Medicine, and samples (provided by the Tissue Acquisition Service under IRB H-45971) were obtained after informed consent.

Animal work

Animal tissue sample collection and injection were approved by the Baylor College of Medicine Institutional Animal Care and Use Committee (AN-7682) and performed in accordance with regulations and established guidelines.

All schematics were produced using Biorender illustrations.

Acknowledgments and Funding

We thank the Elkins Pancreas Center at the Dan L Duncan Comprehensive Cancer Center and the Baylor College of Medicine Tissue Acquisition and Pathology Shared Resource Core for providing the human tissue for this work. Specifically, we thank Dr. William Fisher, Dr. George Van Buren, Dr. Michael Ittmann, Dr. Patricia Castro, and the entire support staff for their support. This work was supported by the Cancer Prevention and Research Institute of Texas Scholar in Cancer Research (CPRIT RR150106 to RJP), the Andrew McDonough B+ Foundation (RJP), and the V Foundation V Scholar in Cancer Research (#V2017-017 to RJP). The Parchem lab is also supported by the NICHD (R01HD099252 and R01HD098131) and the NHLBI (T32 HL092332 to ASB and MGC). This work was also supported by the Baylor College of Medicine Medical Scientist Training Program.

Conflicts of Interest

All authors except for Frank McCormick (FM) declare no competing interests. FM is a consultant for the following companies: Amgen, Daiichi, Exuma Biotech, Frontiers Med, Ideaya Biosciences, Kura Oncology, Leidos Biomedical Research, PellePharm, Pfizer, P.M.V. Pharma, Quanta Therapeutics. FM is a consultant and co-founder for the following companies: BridgeBio, Olema Pharmaceuticals, and Quartz. FM has received research grants from Daiichi Sankyo and Gilead Sciences and has a current grant from Boehringer-Ingelheim.

Author contributions

ASB, MTW, FM, and RJP designed the study. ASB contributed to all figures in this study. ZY contributed to Figure 1 and Supplemental Figure 5. MGC contributed to Figure 4. MTW contributed to Supplemental Figure 3. RJP and FM supervised various aspects of the work. ASB wrote the paper with critical input from all other authors who approved the final version.

Siegel, R. L., Miller, K. D., Wagle, N. S. & Jemal, A. Cancer statistics, 2023. CA Cancer J. Clin. 73, 17–48 (2023).
Waters, A. M. & Der, C. J. KRAS: the critical driver and therapeutic target for pancreatic cancer. Cold Spring Harb. Perspect. Med. 8, (2018).
Collins, M. A. et al. Oncogenic Kras is required for both the initiation and maintenance of pancreatic cancer in mice. J. Clin. Invest. 122, 639–653 (2012).
Chen, H., Liu, H. & Qing, G. Targeting oncogenic Myc as a strategy for cancer treatment. Signal Transduct. Target. Ther. 3, 5 (2018).
Knight, J. R. P. & Sansom, O. J. Tuning protein synthesis for cancer therapy. Mol. Cell. Oncol. 8, 1884034 (2021).
Hafner, A., Bulyk, M. L., Jambhekar, A. & Lahav, G. The multiple mechanisms that regulate p53 activity and cell fate. Nat. Rev. Mol. Cell Biol. 20, 199–210 (2019).
Lu, R. et al. Systems-level dynamic analyses of fate change in murine embryonic stem cells. Nature 462, 358–362 (2009).
Altshuler, A. et al. RAS Regulates the Transition from Naive to Primed Pluripotent Stem Cells. Stem Cell Reports 10, 1088–1101 (2018).
Kubara, K. et al. Status of KRAS in iPSCs Impacts upon Self-Renewal and Differentiation Propensity. Stem Cell Reports 11, 380–394 (2018).
Horvathova, I. et al. The Dynamics of mRNA Turnover Revealed by Single-Molecule Imaging in Single Cells. Mol. Cell 68, 615–625.e9 (2017).
Eulalio, A., Behm-Ansmant, I., Schweizer, D. & Izaurralde, E. P-body formation is a consequence, not the cause, of RNA-mediated gene silencing. Mol. Cell. Biol. 27, 3970–3981 (2007).
Hubstenberger, A. et al. P-Body Purification Reveals the Condensation of Repressed mRNA Regulons. Mol. Cell 68, 144–157.e5 (2017).
Pitchiaya, S. et al. Dynamic recruitment of single rnas to processing bodies depends on RNA functionality. Mol. Cell 74, 521–533.e6 (2019).
Bortoletto, A. S. & Parchem, R. J. KRAS hijacks the microrna regulatory pathway in cancer. Cancer Res. (2023) doi:10.1158/0008-5472.CAN-23-0296.
Shankar, S. et al. KRAS engages AGO2 to enhance cellular transformation. Cell Rep. 14, 1448–1461 (2016).
Wang, M.-T. et al. Blockade of leukemia inhibitory factor as a therapeutic approach to KRAS driven pancreatic cancer. Nat. Commun. 10, 3055 (2019).
Lemberger, M. et al. MicroRNA profiling of pancreatic ductal adenocarcinoma (PDAC) reveals signature expression related to lymph node metastasis. Oncotarget 10, 2644–2656 (2019).
Tesfaye, A. A., Azmi, A. S. & Philip, P. A. miRNA and Gene Expression in Pancreatic Ductal Adenocarcinoma. Am. J. Pathol. 189, 58–70 (2019).
Licursi, V., Conte, F., Fiscon, G. & Paci, P. MIENTURNET: an interactive web tool for microRNA-target enrichment and network-based analysis. BMC Bioinformatics 20, 545 (2019).
Jiang, S., Fagman, J. B., Chen, C., Alberti, S. & Liu, B. Protein phase separation and its role in tumorigenesis. eLife 9, (2020).
McKenzie, A. J. et al. KRAS-MEK Signaling Controls Ago2 Sorting into Exosomes. Cell Rep. 15, 978–987 (2016).
Waninger, J. J. et al. Biochemical characterization of the interaction between KRAS and Argonaute 2. Biochem. Biophys. Rep. 29, 101191 (2022).
Efstathiou, S. et al. ER-associated RNA silencing promotes ER quality control. Nat. Cell Biol. 24, 1714–1725 (2022).
Case, L. B., Zhang, X., Ditlev, J. A. & Rosen, M. K. Stoichiometry controls activity of phase-separated clusters of actin signaling proteins. Science 363, 1093–1097 (2019).
Lee, J. E., Cathey, P. I., Wu, H., Parker, R. & Voeltz, G. K. Endoplasmic reticulum contact sites regulate the dynamics of membraneless organelles. Science 367, (2020).
Currie, S. L. et al. Quantitative reconstitution of yeast RNA processing bodies. Proc Natl Acad Sci USA 120, e2214064120 (2023).
Moon, S. L. et al. Multicolour single-molecule tracking of mRNA interactions with RNP granules. Nat. Cell Biol. 21, 162–168 (2019).
Elkayam, E. et al. Multivalent recruitment of human argonaute by GW182. Mol. Cell 67, 646–658.e3 (2017).
Briskin, D., Wang, P. Y. & Bartel, D. P. The biochemical basis for the cooperative action of microRNAs. Proc Natl Acad Sci USA 117, 17764–17774 (2020).
Rybak-Wolf, A. et al. A variety of dicer substrates in human and C. elegans. Cell 159, 1153–1167 (2014).
Grabocka, E. & Bar-Sagi, D. Mutant KRAS enhances tumor cell fitness by upregulating stress granules. Cell 167, 1803–1813.e12 (2016).
Fonteneau, G. et al. Stress Granules Determine the Development of Obesity-Associated Pancreatic Cancer. Cancer Discov. 12, 1984–2005 (2022).
Hingorani, S. R. et al. Trp53R172H and KrasG12D cooperate to promote chromosomal instability and widely metastatic pancreatic ductal adenocarcinoma in mice. Cancer Cell 7, 469–483 (2005).
Wang, Y., Medvid, R., Melton, C., Jaenisch, R. & Blelloch, R. DGCR8 is essential for microRNA biogenesis and silencing of embryonic stem cell self-renewal. Nat. Genet. 39, 380–385 (2007).
Shen, J. et al. EGFR modulates microRNA maturation in response to hypoxia through phosphorylation of AGO2. Nature 497, 383–387 (2013).
Golden, R. J. et al. An Argonaute phosphorylation cycle promotes microRNA-mediated silencing. Nature 542, 197–202 (2017).
Shankar, S. et al. An Essential Role for Argonaute 2 in EGFR-KRAS Signaling in Pancreatic Cancer Development. BioRxiv (2017) doi:10.1101/227264.
Bridge, K. S. et al. Argonaute Utilization for miRNA Silencing Is Determined by Phosphorylation-Dependent Recruitment of LIM-Domain-Containing Proteins. Cell Rep. 20, 173–187 (2017).
Zeng, Y., Sankala, H., Zhang, X. & Graves, P. R. Phosphorylation of Argonaute 2 at serine-387 facilitates its localization to processing bodies. Biochem. J. 413, 429–436 (2008).
Horman, S. R. et al. Akt-mediated phosphorylation of argonaute 2 downregulates cleavage and upregulates translational repression of MicroRNA targets. Mol. Cell 50, 356–367 (2013).
Rüdel, S. et al. Phosphorylation of human Argonaute proteins affects small RNA binding. Nucleic Acids Res. 39, 2330–2343 (2011).
Padrón, A., Iwasaki, S. & Ingolia, N. T. Proximity RNA Labeling by APEX-Seq Reveals the Organization of Translation Initiation Complexes and Repressive RNA Granules. Mol. Cell 75, 875–887.e5 (2019).
Sears, R. M., May, D. G. & Roux, K. J. BioID as a Tool for Protein-Proximity Labeling in Living Cells. Methods Mol. Biol. 2012, 299–313 (2019).
Zhang, L. et al. TSA-seq reveals a largely conserved genome organization relative to nuclear speckles with small position changes tightly correlated with gene expression changes. Genome Res. (2020) doi:10.1101/gr.266239.120.
Tu, Y., Chen, C. & Fan, G. Association between the expression of secreted phosphoprotein - related genes and prognosis of human cancer. BMC Cancer 19, 1230 (2019).
Nallasamy, P. et al. Pancreatic Tumor Microenvironment Factor Promotes Cancer Stemness via SPP1-CD44 Axis. Gastroenterology 161, 1998–2013.e7 (2021).
Rzeczkowski, K. et al. c-Jun N-terminal kinase phosphorylates DCP1a to control formation of P bodies. J. Cell Biol. 194, 581–596 (2011).
Dougherty, J. D., Reineke, L. C. & Lloyd, R. E. mRNA decapping enzyme 1a (Dcp1a)-induced translational arrest through protein kinase R (PKR) activation requires the N-terminal enabled vasodilator-stimulated protein homology 1 (EVH1) domain. J. Biol. Chem. 289, 3936–3949 (2014).
Mugridge, J. S., Ziemniak, M., Jemielity, J. & Gross, J. D. Structural basis of mRNA-cap recognition by Dcp1-Dcp2. Nat. Struct. Mol. Biol. 23, 987–994 (2016).
Mugridge, J. S., Tibble, R. W., Ziemniak, M., Jemielity, J. & Gross, J. D. Structure of the activated Edc1-Dcp1-Dcp2-Edc3 mRNA decapping complex with substrate analog poised for catalysis. Nat. Commun. 9, 1152 (2018).
Suzuki, T. et al. Mutant KRAS drives metabolic reprogramming and autophagic flux in premalignant pancreatic cells. Cancer Gene Ther. 29, 505–518 (2022).
Chan, K. et al. eIF4A supports an oncogenic translation program in pancreatic ductal adenocarcinoma. Nat. Commun. 10, 5151 (2019).
Kemp, S. B. et al. Efficacy of a Small-Molecule Inhibitor of KrasG12D in Immunocompetent Models of Pancreatic Cancer. Cancer Discov. 13, 298–311 (2023).
Siret, C. et al. Deciphering the Crosstalk Between Myeloid-Derived Suppressor Cells and Regulatory T Cells in Pancreatic Ductal Adenocarcinoma. Front. Immunol. 10, 3070 (2019).
Nishikawa, H. & Koyama, S. Mechanisms of regulatory T cell infiltration in tumors: implications for innovative immune precision therapies. J. Immunother. Cancer 9, (2021).
Väyrynen, S. A. et al. Composition, spatial characteristics, and prognostic significance of myeloid cell infiltration in pancreatic cancer. Clin. Cancer Res. 27, 1069–1081 (2021).
Goulart, M. R., Stasinos, K., Fincham, R. E. A., Delvecchio, F. R. & Kocher, H. M. T cells in pancreatic cancer stroma. World J. Gastroenterol. 27, 7956–7968 (2021).
Tape, C. J. et al. Oncogenic KRAS regulates tumor cell signaling via stromal reciprocation. Cell 165, 910–920 (2016).
Velez-Delgado, A. et al. Extrinsic KRAS signaling shapes the pancreatic microenvironment through fibroblast reprogramming. Cell. Mol. Gastroenterol. Hepatol. 13, 1673–1699 (2022).
Hsu, S.-K. et al. Culprits of PDAC resistance to gemcitabine and immune checkpoint inhibitor: Tumour microenvironment components. Front. Mol. Biosci. 9, 1020888 (2022).
Marei, H. E., Hasan, A., Pozzoli, G. & Cenciarelli, C. Cancer immunotherapy with immune checkpoint inhibitors (ICIs): potential, mechanisms of resistance, and strategies for reinvigorating T cell responsiveness when resistance is acquired. Cancer Cell Int. 23, 64 (2023).
Orozco, C. A. et al. Targeting galectin-1 inhibits pancreatic cancer progression by modulating tumor-stroma crosstalk. Proc Natl Acad Sci USA 115, E3769–E3778 (2018).
Nambiar, D. K. et al. Galectin-1–driven T cell exclusion in the tumor endothelium promotes immunotherapy resistance. The Journal of Clinical Investigation (2019).
Jiang, Z., Zhang, W., Sha, G., Wang, D. & Tang, D. Galectins are central mediators of immune escape in pancreatic ductal adenocarcinoma. Cancers (Basel) 14, (2022).
Vaz, J., Ansari, D., Sasor, A. & Andersson, R. SPARC: A potential prognostic and therapeutic target in pancreatic cancer. Pancreas 44, 1024–1035 (2015).
Pan, K., Huang, X. & Jia, X. SPARC promotes pancreatic cancer cell proliferation and migration through autocrine secretion into the extracellular milieu. Oncol. Lett. 21, 485 (2021).
Luan, H., He, Y., Zhang, T., Su, Y. & Zhou, L. The identification of liver metastasis- and prognosis-associated genes in pancreatic ductal adenocarcinoma. BMC Cancer 22, 463 (2022).
Sangaletti, S. et al. Osteopontin shapes immunosuppression in the metastatic niche. Cancer Res. 74, 4706–4719 (2014).
Shurin, M. R. Osteopontin controls immunosuppression in the tumor microenvironment. J. Clin. Invest. 128, 5209–5212 (2018).
Wei, T. et al. The significance of secreted phosphoprotein 1 in multiple human cancers. Front. Mol. Biosci. 7, 565383 (2020).
Cheng, M. et al. Immunosuppressive role of SPP1-CD44 in the tumor microenvironment of intrahepatic cholangiocarcinoma assessed by single-cell RNA sequencing. J. Cancer Res. Clin. Oncol. (2022) doi:10.1007/s00432-022-04498-w.
Erikson, D. W., Burghardt, R. C., Bayless, K. J. & Johnson, G. A. Secreted phosphoprotein 1 (SPP1, osteopontin) binds to integrin alpha v beta 6 on porcine trophectoderm cells and integrin alpha v beta 3 on uterine luminal epithelial cells, and promotes trophectoderm cell adhesion and migration. Biol. Reprod. 81, 814–825 (2009).
Gola, M., Sejda, A., Godlewski, J., Cieślak, M. & Starzyńska, A. Neural component of the tumor microenvironment in pancreatic ductal adenocarcinoma. Cancers (Basel) 14, (2022).
Renz, B. W. et al. β2 Adrenergic-Neurotrophin Feedforward Loop Promotes Pancreatic Cancer. Cancer Cell 33, 75–90.e7 (2018).
Banh, R. S. et al. Neurons Release Serine to Support mRNA Translation in Pancreatic Cancer. Cell 183, 1202–1218.e25 (2020).
Martin, M. Cutadapt removes adapter sequences from high-throughput sequencing reads. EMBnet j. 17, 10 (2011).
Andrews, S. Babraham Bioinformatics - FastQC A Quality Control tool for High Throughput Sequence Data. https://www.bioinformatics.babraham.ac.uk/projects/fastqc/.
Dobin, A. et al. STAR: ultrafast universal RNA-seq aligner. Bioinformatics 29, 15–21 (2013).
Hervé Pagès, V. O. GenomicAlignments. Bioconductor (2017) doi:10.18129/b9.bioc.genomicalignments.
Love, M. I., Huber, W. & Anders, S. Moderated estimation of fold change and dispersion for RNA-seq data with DESeq2. Genome Biol. 15, 550 (2014).
Zhang, Y., Parmigiani, G. & Johnson, W. E. ComBat-seq: batch effect adjustment for RNA-seq count data. NAR Genom. Bioinform. 2, lqaa078 (2020).
Chen, E. Y. et al. Enrichr: interactive and collaborative HTML5 gene list enrichment analysis tool. BMC Bioinformatics 14, 128 (2013).
Kuleshov, M. V. et al. Enrichr: a comprehensive gene set enrichment analysis web server 2016 update. Nucleic Acids Res. 44, W90-7 (2016).
Xie, Z. et al. Gene Set Knowledge Discovery with Enrichr. Curr. Protoc. 1, e90 (2021).
Lewis, B. P., Burge, C. B. & Bartel, D. P. Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell 120, 15–20 (2005).
Wickham, H. ggplot2 - Elegant Graphics for Data Analysis. (Springer-Verlag New York, 2016). doi:10.1007/978-0-387-98141-3.
Tsanov, N. et al. smiFISH and FISH-quant - a flexible single RNA detection approach with super-resolution capability. Nucleic Acids Res. 44, e165 (2016).

Yes there is potential Competing Interest. FM is a consultant for the following companies: Amgen, Daiichi, Exuma Biotech, Frontiers Med, Ideaya Biosciences, Kura Oncology, Leidos Biomedical Research, PellePharm, Pfizer, P.M.V. Pharma, Quanta Therapeutics. FM is a consultant and co-founder for the following companies: BridgeBio, Olema Pharmaceuticals, and Quartz. FM has received research grants from Daiichi Sankyo and Gilead Sciences and has a current grant from Boehringer-Ingelheim.

Download PDF

Version 1

posted

You are reading this latest preprint version

Oncogenic KRAS promotes pancreatic ductal adenocarcinoma (PDAC) through post-transcriptionally regulated KRAS-induced granules (KGs)

Status:

Version 1

Abstract

Figures

Introduction

Results

Oncogenic KRAS induces downregulation of miRNA abundance

Oncogenic KRAS causes an accumulation of pro-tumor mRNA transcripts

Oncogenic KRAS leads to the subcellular reorganization of AGO2

Oncogenic KRAS leads to the formation of distinct post-transcriptional granules

KG assembly requires oncogenic KRAS and miRNA reduction

Phosphorylation of tyrosine 393 in AGO2 is required and sufficient for KGs

KGs require EGFR but not MEK and PI3K signaling

KGs are enriched for oncogenic transcripts

KG mRNAs are targets of downregulated miRNAs and encode proteins upregulated by mutant KRAS

KG inhibition delays tumor growth

Inhibition of KGs prevents tumor vascularization and innervation

Discussion

Methods

Cell culture

RNA isolation

cDNA library preparation and sequencing

miRNA and mRNA sequencing data analysis

Immunohistochemistry

Condensate quantification

Small molecule inhibitor treatments

Cell proliferation and cell tracking

Tumorsphere assay

siRNA gene knockdown

Single-molecule inexpensive RNA-FISH

Constructs

Human PDAC samples

Animal work

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1