KRAS Hijacks the miRNA Regulatory Pathway in Cancer

Abstract

Extensive studies have focused on the misregulation of individual miRNAs in cancer. More recently, mutations in the miRNA biogenesis and processing machinery have been implicated in several malignancies. Such mutations can lead to global miRNA misregulation, which may promote many of the well-known hallmarks of cancer. Interestingly, recent evidence also suggests that oncogenic Kristen rat sarcoma viral oncogene homolog (KRAS) mutations act in part by modulating the activity of members of the miRNA regulatory pathway. Here, we highlight the vital role mutations in the miRNA core machinery play in promoting malignant transformation. Furthermore, we discuss how mutant KRAS can simultaneously impact multiple steps of miRNA processing and function to promote tumorigenesis. Although the ability of KRAS to hijack the miRNA regulatory pathway adds a layer of complexity to its oncogenic nature, it also provides a potential therapeutic avenue that has yet to be exploited in the clinic. Moreover, concurrent targeting of mutant KRAS and members of the miRNA core machinery represents a potential strategy for treating cancer.

Introduction

Posttranscriptional regulation through miRNAs is a highly conserved and ubiquitous process known to be misregulated across all cancers. More recently, miRNA core machinery mutations have been implicated in several malignancies (1–4). Interestingly, the oncogenic effects of global miRNA misregulation overlap with the functional consequences of mutations in the Kristen rat sarcoma viral oncogene homolog (KRAS). KRAS mutations drive tumor initiation in several types of cancer, including pancreatic ductal adenocarcinoma (PDAC), non–small cell lung cancer (NSCLC), and colorectal cancer (5–8). In fact, lung, colorectal, and pancreatic cancer now represent the top three causes of cancer-related deaths in the United States. Thus, the need for identifying novel ways to treat KRAS-driven cancers is greater than ever. KRAS is the most mutated oncogene among all cancers, and activating KRAS mutations lead to the constitutive activation of downstream signaling cascades, which induce various oncogenic characteristics such as sustained proliferation, sustained self-renewal, and increased vascularization (6, 9–12). Although much is known about how miRNAs influence KRAS expression and function, less is known about how KRAS modulates the global miRNA landscape (13). In recent years, oncogenic KRAS has been found to influence aspects of miRNA biogenesis and effector function. Similar to consequences seen with mutations in miRNA biosynthesis machinery, mutations in KRAS can analogously lead to global miRNA misregulation. Here, we review what is currently known about how mutations in the miRNA biosynthesis machinery contribute to cancer formation and relate this to how oncogenic KRAS modifies the miRNA regulatory pathway to promote tumorigenesis. Understanding this additional layer to the complex genetic misregulation induced by oncogenic KRAS may elucidate novel targetable pathways for therapeutic advancement. Moreover, with the development of enhanced small-molecule inhibitors of KRAS on the horizon, concurrent targeting of mutant KRAS and individual members of the miRNA biogenesis and processing machinery thus represents the potential for additive or synergistic therapies to treat several types of cancer (14).

Overview of miRNA Biogenesis

miRNAs are small, noncoding RNAs that range between 20 and 25 nucleotides in size and are involved in posttranscriptional gene regulation (15). miRNAs typically exert their function by binding to their target mRNA through the interaction of the seed region of the miRNA with the 3′ untranslated region (UTR) of the target transcript. The interaction of miRNAs with their target then regulates gene expression by promoting translational inhibition or mRNA decay. miRNA biosynthesis begins in the nucleus (Fig. 1A), where canonically, pri-miRNAs are transcribed by RNA polymerase II and processed into pre-miRNAs by the nuclear microprocessor complex, which comprises DROSHA and DGCR8. DCGR8 is an RNA-binding protein responsible for binding the pri-miRNA in complex with other members of the microprocessing machinery such as DEAD-box helicase 5 (DDX5), DexH-box helicase 9 (DHX9), and DEAD-Box helicase 17 (DDX17; refs. 15–17). DROSHA is another nuclear miRNA microprocessing complex member and a binding partner of DGCR8. After transcription by RNA pol II, DROSHA cleaves off the stem-loop of the pri-miRNA bound by DGCR8, generating a pre-miRNA.

Figure 1.

The miRNA regulatory pathway, associated cancers, and the effect of KRAS^MUT on miRNA processing. A, Nuclear miRNA processing and cancers associated with mutations in the microprocessor. B, Cytoplasmic miRNA processing and cancers associated with mutations in cytoplasmic factors. C, KRAS^MUT-induced effects on miRNA biogenesis, processing, and function. i, Increased oncomirs and decreased tumor suppressive miRNAs via KIMAT1 and DHX9. ii, Decreased miRNA export. iii, Decreased cytoplasmic DICER processing function. iv, Decreased exosomal miRNA secretion and increased PB localization of AGO2. v, Inhibition of AGO2 miRNA processing function. vi, Inhibition of AGO2 slicing activity. vii, Upregulation of SG formation and increased cell survival and therapy resistance. viii, Upregulation of UPR and shift from tubular to cisternal ER, causing decreased PB fission events and decreased PB numbers.

After pri-miRNAs are processed into pre-miRNAs, they exit the nucleus via Exportin-5 and GTP-bound Ras-related nuclear protein (RAN-GTP; ref. 18). They are then processed into mature duplex miRNAs by the endoribonuclease DICER (Fig. 1B). Next, argonaute (AGO) proteins bind to one strand of the mature miRNA duplex and GW182, a scaffolding protein that promotes the recruitment of other RNA-binding proteins. In humans, the AGO family of proteins consists of AGO1–4 and is involved in miRNA-mediated mRNA silencing (19, 20). Although all four members of AGO are equally loaded with miRNAs, AGO1, and AGO2 are generally expressed at higher levels than AGO3 or AGO4 (21–23). The AGO-bound miRNA and GW182 constitute the core miRNA–induced silencing complex (miRISC) that is guided to the target mRNA for silencing. Although miRNA-mediated regulation typically results in translational inhibition, certain miRNA families can induce AGO-slicing activity, resulting in miRNA degradation.

Oncogenic mutations in the nuclear microprocessing complex and export machinery

Mutations in the factors involved in miRNA biosynthesis and processing have been implicated in many cancer types (Table 1; Fig. 1A and B). Mutations in DGCR8 have been correlated with increased cancer incidence and have been shown to act as driver mutations in Wilms tumors, certain thyroid cancers, pineoblastoma, and familial tumor susceptibility syndromes (24–29). One study identified DGCR8 missense, frameshift, and nonsense mutations in a significant percentage of Wilms tumors, all leading to nonfunctional DGCR8 peptides (24). One of the mutational hotspots identified in this study (E518K, which resides within the RNA-binding domain) has similarly been identified in follicular thyroid tumors and cases of familial multinodal goiter with schwannomatosis (26, 28–30). These studies suggest that a decrease in DGCR8 function leads to reduced miRNA biosynthesis, thus promoting the misregulation of their target transcripts.

Table 1.

miRNA machinery mutations and the effects of oncogenic KRAS on miRNA regulation in cancer.

Gene	Function	Mutation	Cancer type		Ref.
DGCR8	Nuclear microprocessing complex member. Binds pri-miRNAs to facilitate processing by DROSHA	E518K	Wilms tumor thyroid carcinoma, FMGS		30
		rs417309 G/A	Laryngeal cancer, breast cancer		26, 29
		p.R32fs			25
		Copy-number loss, p.S92fsa	Pineoblastoma		27
RNASEN/DROSHA	Nuclear microprocessing complex member. Ribonuclease that cleaves pir-miRNA to pre-miRNAs	rs640831 C/A	Lung adenocarcinoma		31
		rs1110386 G/Aa	Ovarian cancer		32, 33
		rs486732 C/Aa	Bladder cancer		34
		p.E500*	Breast cancer		35
		p.R277C	Thyroid carcinoma		37
		p.Q136*a	Pineaoblastoma		27
DHX9	Nuclear microprocessing complex member. Helicase that modulates nuclear miRNA processing	p.S625C, p.P89A	Breast cancer		42
XPO5	Facilitates pre-miRNA export from the nucleus to the cytoplasm	p.T1182del	Colon cancer, gastric cancer, endometrial cancer		46
		rs2257082 A/G	Gastric cancer		48
DICER1	Member of the miRISC. Endoribonuclease that cleaves the stem-loop from pre-miRNAs, leading to the formation of mature miRNAs	Germline:
		p.Y1180*	Fetal lung adenocarcinoma, Sertoli-Leydig ovarian tumor, FMGS		54
		p.Y819H	Hepatocellular carcinoma		67
		p.D1713Aa	Wilms tumor		55
		p.R187*a	Pleuropulmonary blastoma		52
		p.G803Ra	Wilms tumor		56
		p.E503Xa	Embryonal ovarian cancer		51
		Somatic:
		p.A872Ta	Wilms tumor		55
		p.E428Ka	Hepatocellular carcinoma		67
		p.E813Qa	Sertoli-Leydig cell tumor		58
		p.D1709Na	Yolk-sac tumor		57
		p.D1709G	Juvenile granulosa-cell tumor
		p.D1810Y	Teratoma
		p.E1813Qa	Sertoli-Leydig cell tumor		57
		p.Y1701*a	Pineoblastoma		27
TARBP1	Member of the miRISC. Binding partner of DICER that facilitates RISC assembly	p.P144fs	Colon cancer		68
		p.R353fs	Wilms tumor		41
		p.M145fs	Gastric cancer		71
		Copy-number gain	Adrenal cancer		70
PRKRA	Member of the miRISC. Binding partner of DICER that facilitates RISC assembly	p.P127L	Ovarian cancer		71
AGO2	Member of the miRISC. Endonuclease that has mRNA cleavage function	Copy-number gain	Head and neck cancer		74
			Multiple myeloma		76
			Breast cancer		75
TNRC6A	Member of the miRISC. Binding partner of AGO2 required for mRNA silencing	p.P115-Q118del	Esophageal cancer		78
		p.R1183fs, p.Trp804fs	Gastric cancer, colorectal cancer		71, 73
miRNA process	Influence by KRAS			Cancer type	Ref.
Nuclear miRNA processing	Increased expression of lncRNA KIMAT1 leads to increased expression of oncomiRs and decreased expression of tumor-suppressive miRNAs			Lung adenocarcinoma	83
miRNA export	Phosphorylation of Exportin-5 by ERK inhibits miRNA export from the nucleus			HCC	84
miRNA processing by DICER	Phosphorylation of DICER by ERK inhibits processing function and increases nuclear localization			NSCLS, PDAC, cholangiocarcinoma, endometrioid tumors	87–92
miRNA processing by AGO2	Binding of KRAS to AGO2 inhibits miRNA unwinding and slicing function			PDAC, CRC, NSCLS	94, 99, 101
	Phosphorylation of AGO2 by AKT increases localization to PBs and decreases export via exosomes
Upregulation of SGs	Upregulates SG formation and transcript storage for stress protection and therapy resistance			PDAC, multiple myeloma	118, 121
Upregulation of the UPR	Upregulates UPR to compensate for ER stress, which promotes survival			PDAC	118, 136
	UPR upregulation also induces SG formation

^aOne mutation represented from citation.

Mutations in RNASEN, the gene encoding DROSHA, have also been identified in various tumor types. SNPs in RNASEN have been identified and correlated with decreased survival in subpopulations of patients with lung, ovarian, and bladder cancer (31–34). Other studies have identified possible oncogenic mutations in RNASEN in breast and thyroid cancer (35–37). In addition, studies of copy-number variation in genes encoding the miRNA machinery have identified RNASEN as one of the most frequently amplified genes in lung and cervical cancer, correlating with decreased survival (38, 39). Finally, mutations in RNASEN have been shown to act as driver mutations in Wilms tumors, similar to DGCR8 (40). Multiple studies have also demonstrated how several RNASEN mutations lead to decreased catalytic function and reduced miRNA biogenesis, suggesting that the overall outcome of miRNA processing inhibition is the promotion of tumorigenesis regardless of the mutation that causes it (24, 41).

DHX9 is a helicase with diverse functions, including modulating miRNA processing in the nucleus. Multiple point mutations in DHX9 have been identified in breast cancer. These mutations are thought to drive tumorigenesis in a subset of BRCA1/BRCA2 wild-type (WT) tumors, as DXH9 interacts with BRCA proteins as part of the miRNA processing function of BRCA1/BRCA2 (42, 43). DDX17, another helicase involved in nuclear miRNA processing, has also been implicated in cancer. Although the expression of DDX17 is misregulated in various cancers, including colon, lung, breast, prostate, and liver cancer, whether this misregulation is driven by genetic mutations, copy-number variations, or posttranslational modification is unknown (44, 45).

The gene XPO5 encodes exportin-5. Like members of the microprocessor, XPO5 mutations have been reported across multiple types of cancer. One study reported the presence of inactivating mutations in XPO5 in colon, endometrial, and stomach cancer cell lines (46). Various SNPs in the XPO5 gene have also been correlated with increased breast, esophageal, and gastric cancer incidence and poorer gastric cancer outcomes, while others have been associated with better survival outcomes in lung and colon cancer (47, 48). The expression of the binding partner of exportin-5, RAN-GTP, is generally upregulated across most types of cancer. However, no cancer-specific mutations have been identified within RAN-GTP as of yet (49).

Oncogenic mutation in the cytoplasmic components of miRNA core machinery

More tumor-promoting mutations have been identified in the cytoplasmic components of the miRNA biosynthesis machinery compared with their nuclear counterparts. Both germline and somatic mutations in DICER1, the gene encoding the DICER protein, have been shown to promote multiple cancers. Germline loss-of-function mutations in DICER1 have been shown to cause embryonal tumors in pediatric populations, such as Wilms tumors, pleuropulmonary blastoma, cystic nephroma, embryonal rhabdomyosarcoma, ovarian Sertoli-Leydig cell tumor, ciliary body medulloepithelioma, multinodular goiter, thyroid adenomas, pituitary blastoma, pineoblastoma, and renal sarcoma (50–58). Germline DICER1 mutations can also cause familial genetic tumor predisposition syndromes, such as DICER1 syndrome, where patients may develop multiple of the above mentioned tumor types (50, 59–65). The propensity for DICER1 mutations (and certain mutations in other miRNA core machinery factors) to promote congenital and childhood cancers may be due to the vital role that miRNAs play in the spatiotemporal regulation of critical developmental processes. However, the misregulation of these processes can also promote tumor formation in adults.

Somatic mutations in DICER1 can also promote the formation of similar tumors seen in the germline syndromes, such as adult-onset pulmonary blastoma, pineoblastoma, Sertoli-Leydig cell tumors, Wilms tumors, and hepatocellular carcinoma (HCC; refs. 27, 40, 63, 64, 66, 67). In addition, SNPs in DICER1 have been significantly correlated with poorer outcomes in certain gastric tumors (48). trans-activation response element (TAR)-binding protein (TARBP1 or TRBP) and the protein activator of protein kinase R (PACT or PRKRA) are DICER-binding partners that modulate its miRNA processing ability. Whole genome sequencing of various tumor types has identified TARBP1 mutations in gastrointestinal cancer, colorectal cancers, and Wilms tumors associated with decreased expression of TARBP1 (68, 69). However, copy-number variations that lead to increased expression have also been reported in adrenal cancers (70). While less is known about PRKRA, hotspot mutations in this gene have been identified in ovarian cancer (71).

Although four AGO proteins exist in humans, only AGO2 has been implicated in cancer regarding its role in the miRNA regulatory pathway. While AGO1 does display oncogenic function in multiple cancers, it does so outside of its miRNA processing and effector function (72). Although some loss-of-function somatic AGO2 mutations have been identified in gastric and colorectal cancers, most cancers with AGO2 misregulation show genetic amplification and increased protein expression (73). Breast cancer, multiple myeloma, HCC, bladder cancer, ovarian cancer, gastric cancer, colorectal cancer, and head and neck squamous cancers are all types of cancer where AGO2 copy-number amplification or protein overexpression has been shown to drive proliferation and cancer progression (74–77). Because AGO2 is involved in both miRNA biosynthesis and effector function, the overexpression of AGO2 may affect how transcripts are regulated. AGO2 overexpression also does not necessarily lead to increased miRNA processing, as AGO2 is heavily regulated by posttranslational modification. As the various AGO family members are differentially loaded with miRNAs, AGO2 may also promote the processing of oncogenic miRNAs to a greater degree than tumor-suppressive miRNAs. Trinucleotide repeat-containing adapter 6A (TNRC6A) encodes for the AGO2-binding partner, GW182. Interestingly, a recent analysis showed that TNRC6A is the most frequently mutated miRNA biogenesis gene across 33 types of cancer. However, further studies on TNRC6A mutations in cancer are needed, as TNRC6A mutations have only been significantly correlated with poor outcomes in esophageal, gastric, and colorectal cancers (71, 73, 78).

KRAS and miRNA Regulation

Clearly, mutations in members of the miRNA processing machinery can promote cancer. Save for a few exceptions, most of these mutations have been linked to a decrease in miRNA biosynthesis, thus leading to the global misregulation of miRNAs and the mRNA transcripts they target. Interestingly, recent evidence suggests that mutant KRAS is similarly able to modulate the activity of multiple members of the core miRNA machinery as an additional mechanism to promote malignant transformation. Several studies have shown that miRNA biogenesis and global miRNA expression levels are grossly misregulated in KRAS-mutated cancers (79–82). However, over the last decade, KRAS has been shown to play a more significant role in driving miRNA misregulation by directly regulating critical factors involved in the miRNA regulatory pathway (Table 1). This additional regulatory mechanism disrupted by mutant KRAS may represent a novel vulnerability in KRAS-driven cancers that has yet to be targeted.

KRAS and nuclear miRNA biosynthesis

Recently, KIMAT1, a KRAS-dependent long noncoding RNA, was shown to regulate DHX9 (83). Increased KIMAT1 expression can stabilize DHX9 and promote an increase in the relative expression of oncogenic miRNAs while simultaneously preventing the expression of tumor-suppressive miRNAs in lung tumors (Fig. 1C, i). This suggests that oncogenic KRAS may influence the differential processing of miRNAs by modulating the microprocessing complex. Increased levels of phosphorylated ERK have also been shown to increase the phosphorylation of exportin-5 in HCC, which inhibits its ability to bind and export pre-miRNAs (Fig. 1C, ii; ref. 84). Nuclear export of pre-miRNAs also depends on RAN, a small GTPase that binds to Exportin-5. Because RAN activation occurs downstream of KRAS, mutant KRAS may increase RAN activity or expression, which is known to occur across many types of cancer (49). Thus, mutant KRAS may affect not only the nuclear processing of miRNAs but also whether those miRNAs can exit the nucleus.

KRAS and cytoplasmic miRNA processing

Once pre-miRNAs exit the nucleus, they are further processed into mature miRNAs by DICER. During C. elegans oogenesis, ERK-dependent phosphorylation of DICER at serine 1705 and 1833 (1728 and 1852 in humans) coordinates its nuclear localization and also inhibits its miRNA processing function (85, 86). As mutant KRAS increases ERK signaling, increased nuclear localization and decreased nuclease activity of DICER would be two logical consequences of KRAS mutations in cancer. The sequestration of DICER away from cytoplasmic pre-miRNAs and inhibition of its nuclease activity would presumably lead to a relative decrease in mature miRNA production (Fig. 1C, iii). Interestingly, the nuclear localization of DICER and its phosphorylation at Serine 1712 and 1836 (1728 and 1852 in humans) has indeed been shown to promote tumor formation and treatment resistance in various KRAS-driven mouse cancer models, including NSCLC and PDAC models (87–92).

In addition to DICER, oncogenic KRAS can affect other RNA-binding proteins involved in cytoplasmic miRNA processing, such as AGO2. The N-terminal domain of AGO2 can bind the switch II domain in KRAS, and this interaction inhibits AGO2-dependent miRNA maturation and stabilization (93–95). Although WT KRAS can bind AGO2, it is the interaction between AGO2 and mutant KRAS that drives cellular transformation, implying that there is some type of feed-forward mechanism between AGO2 and mutant KRAS (94). AGO2 is also required for pancreatic intraepithelial neoplasia (PanIN) progression to PDAC via EGFR–KRAS signaling (96). Indeed, increased AGO2 expression and phosphorylation have been found to promote tumorigenesis and metastasis in other cancer models, such as HCC and NSCLC (97–99). Mutant KRAS also influences the function of AGO2 by modulating AGO2-mediated miRNA secretion via exosomes. In colorectal cancer cells, mutant KRAS was found to promote the differential sorting of miRNAs into exosomes through posttranslational modifications of AGO2 mediated by increases in MAPK pathway activation (100, 101). Thus, oncogenic KRAS may exert much of its miRNA-modulating activity through direct and indirect interactions with AGO2.

KRAS and miRNA-mediated silencing and decay

Beyond its effect on miRNA biogenesis, oncogenic KRAS has been shown to modulate how miRNAs regulate their target transcript. In addition to interfering with miRNA maturation, the mutant KRAS/AGO2 interaction can also misregulate the effector function of miRNAs, as AGO2 and the miRISC are essential for miRNA localization and miRNA-mediated target silencing. AGO2 function is regulated via posttranslational modification, including phosphorylation at various residues. Cells harboring an activating KRAS mutation have higher levels of AGO2 phosphorylation at serine 387 (AGO2^S387; ref. 101). As previously mentioned, AGO2-bound miRNAs can be secreted from the cell via exosomes. However, ERK and AKT-mediated phosphorylation of AGO2^S387 prevents this secretion and shifts AGO2 localization to processing bodies (PB; Fig. 1C, iv; refs. 101, 102). AGO2^S387 phosphorylation also increases the association of AGO2 with GW182 (102, 103). This interaction is necessary for miRNA-mediated silencing and the localization of AGO2 to PBs (103–105). Although increasing the AGO2/GW182 interaction would presumably lead to an increase in miRNA-mediated gene silencing, it is not entirely known how mutated KRAS affects this process.

AGO2 is also regulated by phosphorylation at tyrosine 393 (AGO2^Y393), which also increases in cells harboring KRAS mutations, such as in PDAC and NSCLC. AGO2^Y393 phosphorylation inhibits the miRNA unwinding and miRISC loading function of AGO2 (Fig. 1C, v; ref. 96). During physiologic conditions, AGO2^Y393 is phosphorylated by tyrosine kinases, such as EGFR and c-Src kinase (106, 107). Phosphorylation of AGO2^Y393 leads to the dissociation of KRAS^WT and AGO2 but is insufficient to interrupt the mutant KRAS–AGO2 interaction, thus preventing proper miRNA binding by AGO2 (96). In addition, mutant KRAS has been shown to prevent AGO2^Y393 dephosphorylation by protein tyrosine phosphatase 1B (PTP1B), further interfering with AGO2 function (108). Although there are multiple other sites in which AGO2 may be phosphorylated, it is not yet known how oncogenic KRAS affects these sites.

Despite the conservation of the N-terminal domain across the four AGO members, KRAS has only been shown to interact with and modulate the activity of AGO2. AGO2 can catalyze the degradation of target transcripts through its mRNA-slicing function. However, this only occurs in certain contexts with specific miRNAs that bind with perfect sequence complementarity to their target transcripts. Increased phosphorylation of AGO2^S387 was also shown to prevent the slicing activity of AGO2. (Fig. 1C, vi; ref. 102). Although the consequences of mutant KRAS inhibiting AGO2 slicing are unknown, one possibility is that transcripts usually degraded by AGO2 slicing may be stabilized, such as the HOX family of genes. HOX genes are regulated by miR-196-directed cleavage by AGO2 (109). Many HOX genes have been shown to act as oncogenes in KRAS-driven cancers such as lung, colorectal, and pancreatic cancers (110). Thus, inhibition of HOX transcript cleavage by AGO2 may contribute to HOX upregulation in cancer.

KRAS and intracellular condensates

Beyond directly regulating members of the miRNA regulatory pathway, oncogenic KRAS may also coordinate the storage of miRNA-targeted transcripts in intracellular condensates. Two types of intracellular condensates involved in transcript regulation are stress granules (SG) and PBs; refs. 111–114). Components of these RNP granules can modulate the expression of transcripts that control disrupted functions in cancer, such as signal transduction, metabolism, and cellular stress. Thus, understanding their misregulation in cancer has become of great interest (115).

SGs are stress-induced and are storage sites for translationally arrested mRNAs (116). miRNAs and members of the miRNA core machinery have been shown to localize to SGs, suggesting that SGs may regulate miRNA-bound transcripts during cell stress (117). Mutated KRAS has been shown to upregulate SG formation in PDAC and multiple myeloma via the upregulation of 15-Deoxy-delta (12, 14)-prostaglandin J (15d-PGJ2; ref. 2), which subsequently targets eIF4A for inactivation, promoting SG formation (118–120). This upregulation of SGs has been shown to promote survival and enhanced stress protection, both of which drive therapy resistance (Fig. 1C, vii; refs. 118, 121). Moreover, another SG protein, DDX3, has been shown to enhance the transcription of oncogenic KRAS (122). This suggests that although mutated KRAS likely affects components of SGs, factors that localize to SGs may also affect the expression of KRAS. This is not surprising, considering that most of these factors have some function in regulating mRNA transcript levels.

PBs are composed of decapping enzymes, decapping activators, exonucleases, deadenylases, translational repressors, and miRNA machinery. The specific roles of these proteins and how PBs function under physiologic conditions have been extensively described in other reviews (113, 123–125). As stated above, oncogenic KRAS affects the posttranslational modification of AGO2, leading to its increased localization to PBs. KRAS has also been shown to interact with DDX6, one of the few components required for PB formation. This interaction was found to play a role in modulating KRAS signaling downstream of HER2 by promoting the translation of HER2 in gastric cancer (126). In addition, it has been recently shown that receptor tyrosine kinase fusion proteins have the capacity to drive oncogenic KRAS signaling at membraneless protein granules (127). Similar to SGs, this suggests that KRAS may not only affect RNP granule dynamics, but signaling from RNP granules may also regulate KRAS.

KRAS has also been identified as an upstream regulator of mRNA transcripts targeted by the CCR4-NOT deadenylase machinery in PBs (128). In addition to regulating the expression of CCR4-NOT–bound transcripts, KRAS has been shown to regulate TOB, a cofactor in the CCR4–NOT complex, which regulates cyclin D1 and prevents proliferation (129). Interestingly, TOB has been shown to act as a tumor suppressor in certain cancers (130). MAPK and c-Jun JNK, both downstream effectors of KRAS, can phosphorylate TOB, decreasing its antiproliferative function and preventing its tumor suppressive function. JNK also phosphorylates multiple PB components, such as DCP1A and 4E-T. Upon phosphorylation by JNK, DCP1a is redistributed from small, punctate PB to larger cytoplasmic inclusions (131). JNK can also phosphorylate 4E-T, increasing PB size during cell stress (132). Together, these data indicate that mutant KRAS may affect multiple components of granules involved in posttranscriptional regulation, further supporting the notion that some interplay exists between oncogenic KRAS, miRNAs, and PBs.

KRAS and the endoplasmic reticulum

Regulation of mRNA translation and silencing is also tightly connected to the endoplasmic reticulum (ER). AGO2, which oncogenic KRAS heavily influences, was recently shown to regulate mRNA degradation at the ER (133). Interestingly, oncogenic KRAS also affects ER homeostasis. Mutations in KRAS, along with other RAS family members, have been shown to lead to ER expansion via the upregulation of the unfolded protein response (UPR; Fig. 1C, viii; ref. 134). The UPR is also critically important in KRAS-mutated PDAC, as KRAS is required to activate the UPR in response to ER stress (135). Under physiologic conditions, the upregulation of the UPR can lead to cell-cycle arrest or stress-induced apoptosis. However, in KRAS-transformed cancers, chronic upregulation of the UPR can lead to tumor cell resistance to acute stress–induced apoptosis (134, 136). Thus, some of the transformative characteristics of mutant KRAS likely depend on transcript regulation by AGO2 and the ER. Activation of the UPR also induces SG formation, which KRAS-mutated cells have coopted to promote survival (118). Interestingly, the ratio of cisternal to tubular ER also plays a role in the fission of PB granules, thus affecting the size and number of PBs (137). When induced, the UPR leads to a decrease in PB numbers, presumably due to a shift to a more cisternal-dominant ER. However, how this phenomenon alters posttranscriptional control through PBs has yet to be discovered.

Conclusion and Future Perspectives

Recently, more effective inhibitors against mutant KRAS subtypes have been developed, such as MRTX849 (adagrasib), which has shown potent inhibition of KRAS^G12C (138). Despite its success and current status in clinical trials, a significant percentage of preclinical models still showed no effective response. In addition, most of the successes in targeting mutant KRAS subtypes have centered on KRAS^G12C. While KRAS^G12C is the predominant driver mutation in lung cancers, KRAS^G12D driver mutations occur far more frequently in PDAC and colorectal cancer (139). Although MRTX1133 was recently announced (January 2023) as a novel KRAS^G12D inhibitor cleared to enter clinical trials, it will still be years before clinical implementation if it is proven effective in human patients (140).

In addition to KRAS-specific inhibitors, other inhibitors of both upstream (i.e., EGFR) and downstream (i.e., MEK and AKT) KRAS signaling pathways have shown promise in treating specific subsets of NSCLC and other cancer types where KRAS is less frequently mutated (139). Regardless of instances where clinical success is seen, almost all treatment modalities are plagued by nonresponders or the development of resistance mechanisms. Because of these issues, identifying novel targetable pathways for therapeutic development in KRAS-driven cancers is still vitally important. As summarized in this review, the miRNA regulatory pathway represents one such pathway that has yet to be therapeutically exploited. Most intriguing is the numerous ways oncogenic KRAS can influence several steps of miRNA biogenesis and function simultaneously, thus presenting several potential avenues for treating KRAS-driven cancers that have yet to be explored.

Although intriguing, targeting the miRNA regulatory pathway in KRAS-driven cancers would not come without challenges. Because individual miRNAs can target multiple transcripts, inhibiting, or activating members of the miRNA core machinery could induce off-target responses. Despite this concern, there may be ways to mitigate potential adverse effects. Many of the ways oncogenic KRAS modulates members of the miRNA regulatory pathway involve posttranslational modifications, such as the phosphorylation of DICER and AGO2 on various residues. Kinase inhibitors are one class of drugs that may modulate the activity of KRAS-induced phosphorylation of DICER and AGO2. However, kinase inhibitors have classically failed to treat KRAS-driven cancers, often due to a lack of specificity or resistance mechanisms. Another appealing option for targeting posttranslationally modified proteins involves using an in silico approach to identifying potential interactions between small-molecule inhibitors and site-specific phosphorylation events in proteins. With the advancement of various cheminformatics platforms, high-throughput screening of small molecules has been used to identify inhibitors of distinct phosphorylation events on specific proteins (141). A similar approach may be able to identify site-specific small-molecule inhibitors of DICER, AGO2, and other posttranslationally modified proteins involved in the miRNA-driven oncogenicity of KRAS. In addition, next-generation sequencing technology has advanced significantly over the last ten years, thus allowing for fast and accurate sequencing of the small RNA transcriptome. Because of this, it is possible to identify potential off-target responses that occur if members of the miRNA regulatory pathway are inhibited/ activated (i.e., the undesired upregulation or downregulation of specific miRNAs). Off-target changes in miRNA expression can be controlled using miRNA mimics or antagomirs (142).

Another possible strategy to inhibit miRNA-dependent oncogenesis of mutant KRAS could involve targeting how miRNAs or transcripts are spatially misregulated by oncogenic KRAS. Recently, the subcellular localization of specific transcripts was shown to be critical for driving tumor invasion (143). Similarly, specific miRNAs or miRNA-targeted transcripts can be differentially enriched in subcellular compartments such as the surface of the ER or PBs, which could also be targeted in KRAS-mutant cancer cells. Although studies are beginning to identify how mutant KRAS alters some of these regulatory pathways, many questions remain to be answered. What are the key downstream effectors driving miRNA misregulation in KRAS-dependent cancers? Can the miRNA regulatory pathway be targeted therapeutically without significant off-target effects? What other role do SGs, PBs, and the ER play in oncogenesis? Are there targetable interactions between oncogenic KRAS, its downstream effectors, and proteins involved in miRNA regulation? Does oncogenic KRAS lead to differential regulation of transcripts in SGs and PBs? Answering these questions and understanding precisely how mutant KRAS alters posttranscriptional regulation through miRNAs seem essential to identifying novel treatment modalities for KRAS- and other oncogene-driven cancers.

Authors' Disclosures

No disclosures were reported.