E3 ligase MKRN3 is a tumor suppressor regulating PABPC1 ubiquitination in non–small cell lung cancer

Lung cancer is the leading cause of cancer-related mortality worldwide, with 2,093,876 new cases and 1,761,007 deaths globally in 2018 (Hellmann et al., 2019; Siegel et al., 2020; Teixeira et al., 2019). Approximately 85% of all lung cancer cases are non–small cell lung cancers (NSCLCs; Campbell et al., 2016; Peifer et al., 2012). NSCLCs mostly include lung adenocarcinoma and lung squamous cell carcinoma (Campbell et al., 2016). Although tyrosine kinase inhibitors and immunotherapy have contributed to significant survival benefits in some patients, the overall survival rates for NSCLCs remain low. In particular, patients with NSCLC that is driven by KRAS mutations are often unresponsive to tyrosine kinase inhibitors and have a poor prognosis (Mainardi et al., 2018). For patients with NSCLC who harbor mutations in the epidermal growth factor receptor (EGFR) or in anaplastic lymphoma kinase (ALK) fusions, targeted therapies are now the first-line standard of care (Govindan et al., 2012; Imielinski et al., 2012; Pao and Chmielecki, 2010; Shaw and Engelman, 2013); in contrast, targeted therapy against mutant KRAS-driven tumors (in up to 30% of NSCLCs) has proved challenging (Hellmann et al., 2019; Mainardi et al., 2018). Indeed, although allele-specific inhibitors for the KRAS^G12C mutant have entered phase I clinical trials, a general strategy that targets all KRAS mutants remains elusive (Hong et al., 2020). There is therefore an urgent need to identify new driver genes and develop therapeutic strategies to benefit a broader patient population, especially those with KRAS mutations.

Central precocious puberty (CPP) is largely caused by germline mutations in the Makorin ring finger protein 3 (MKRN3: reference transcript, NM_005664; reference protein, NP_0056550) gene (Abreu et al., 2013). The molecular function of MKRN3 in CPP is a subject of considerable investigation (Abreu et al., 2013; Abreu et al., 2020; Shin, 2016). Interestingly, CPP has been epidemiologically linked to various diseases in adulthood, including cancers (Day et al., 2017; Holmes, 2017). Cohorts of individuals with CPP show an increased risk (∼10–25%) of malignancies such as lung cancers (Ben Khedher et al., 2017; Calcaterra et al., 2013; Kreuzer et al., 2003). There is also compelling evidence that sex steroid hormones and reproductive factors play a role in the genesis of lung cancer (Ben Khedher et al., 2017; Kreuzer et al., 2003), yet the mechanisms are unclear.

Cancer cells require high global protein translation rates to maintain their fast proliferation behavior. Protein synthesis is a major metabolic event that controls cancer cellular growth and proliferation, but its precise regulatory mechanisms in cancer are not well understood (Bretones et al., 2018; Ebright et al., 2020; Lindqvist et al., 2018; Nguyen et al., 2018; Pelletier et al., 2018).

To investigate whether MKRN3 functions in human cancers, we analyzed the public data from cancer genomics studies and found recurrent inactivating genomic MKRN3 aberrations in NSCLCs. We further present genetic, functional, genetically engineered mouse model and mechanistic data that identify the MKRN3 gene at 15q11 as a bona fide tumor suppressor in NSCLCs and uncover a potential therapeutic target in KRAS-mutant lung cancer.

To investigate whether CPP-associated MKRN3 gene is mutated in human cancers, we queried The Cancer Genome Atlas (TCGA) Pan-Cancer genomic datasets. Frequent MKRN3 alterations were detected in the TCGA Pan-Cancer genomic dataset containing 566 samples of lung adenocarcinoma, the most common type of NSCLC. MKRN3 alterations were identified in 5.0% (28 of 566) of lung adenocarcinoma samples (Fig. 1 A). We further detected a 2–6% MKRN3 alteration rate in four additional NSCLC cohorts consisting of lung adenocarcinomas and lung squamous cell carcinomas similarly examined by whole-genome/whole-exome sequencing (Fig. 1 A). Genomic MKRN3 aberrations are significantly enriched in human NSCLC samples harboring oncogenic KRAS mutations (Fig. S1 A; P < 0.05, discrete independence statistic controlling for observations with varying event rate [DISCOVER] test; Table 1). Collectively, these data established an MKRN3 mutation rate of 3.8% (67/1,758) in NSCLC samples (n = 1,758; Table S1), which included nonsense mutations, frameshift deletions, frameshift insertions, splice mutations, missense mutations, and genomic deletions (Fig. 1 B and Table S1). Across all 1,758 NSCLC samples, nonsense mutations, frameshift deletions, frameshift insertions, splice mutations, and deletions accounted for 45% of genomic alterations in MKRN3 (Fig. 1 B and Table S1), with 55% of mutations being missense, around 58% of which were classed as disruptive by both SIFT and PolyPhen-2 software tools that predict the functional effects of nonsynonymous protein coding single nucleotide polymorphisms (Fig. 1 B and Table S1; Adzhubei et al., 2010). We further provided evidence to show that the missense mutations are loss-of-functional (see below, Fig. 5 D). MKRN3 is maternally imprinted, and the paternal allele is expressed (Abreu et al., 2013). We extended the mutations analysis to the intragenic mutation of MKRN3 in the NSCLC cell line in which it was possible to compare the nucleotide sequence of PCR products generated from genomic DNA or cDNA. Although the MKRN3 mutation was heterozygous in genomic DNA, only the mutant sequence was detected in the lung cancer cell–derived transcripts, consistent with monoallelic expression of the mutant copy from the nonimprinted allele (Fig. S1 B).

It has been reported that MKRN3 is expressed in the hypothalamic arcuate nucleus of mice during postnatal development (Abreu et al., 2013). We demonstrated that MKRN3 is also expressed in both mouse and human lung tissues (Fig. 1 C). MKRN3 inactivation was found across all lung cancer stage subtypes, including early-stage (stage I and II) lung cancer, suggesting that genomic MKRN3 alterations can arise around disease initiation (Table S1). To explore a possible role of MKRN3 in lung cancer, we assessed MKRN3 expression in humans with lung cancer. As shown in Fig. 1, D–F; Fig. S1 D; and Table S2, immunohistochemistry staining and Western blotting analysis revealed that the levels of MKRN3 protein were decreased in the tumor samples as compared with the adjacent normal counterparts. Immunohistochemistry staining and Western blots on a panel of human lung cancer cell lines demonstrate that MKRN3 is expressed in lung cancer cells (Fig. 1 D and Fig. S2 A). The identification of MKRN3 inactivation in TCGA patients and Asian patients shows that the MKRN3 inactivation in lung cancer is generalizable and is not restricted to a specific ethnic group. Although MKRN3 is a maternally imprinted gene, there are no gender differences in MKRN3 mutations (P = 0.922, Fisher's test) and expression (P = 0.673, Fisher's test). Strikingly, lower MKRN3 expression is associated with a significantly shorter overall survival time in two separate lung cancer patient cohorts (Fig. 1 G). As indicated by the hazard ratios, MKRN3 expression levels serve as an independent predictor for risk stratification of overall survival and relapse-free survival in patients (Fig. 1 G). Together, these findings highlight MKRN3 as a potential biomarker for lung cancer and suggest a role of genomic MKRN3 inactivation in lung tumorigenesis.

The biological function of MKRN3 was investigated using various human NSCLC models. MKRN3 biological function was evaluated by reexpressing MKRN3 in MKRN3-inactivated NSCLC cells. A human NSCLC cell line (H1703, lung squamous cell carcinoma) was identified to contain a MKRN3 frameshift deletion (c.636_636delC, p.Y212Yfs*73; Ghandi et al., 2019; Fig. S1 B). Lentivirus-mediated MKRN3 transduction into MKRN3-inactivated H1703 cells induced MKRN3 expression (Fig. 2 B and Fig. S2 A). MKRN3 reexpression reduced the number of viable cells (Fig. 2 A) and proliferative properties (Fig. 2, B and C), but increased cell apoptosis (Fig. S2, C and D). MKRN3 restoration also decreased the cell cycle, increasing the proportion of cells in G2/M phase (Fig. 2 D). MKRN3 reexpression inhibited anchorage-independent growth in MKRN3-inactivated lung cancer cells (Fig. 2 E). To determine whether the inhibition of cell proliferation is manifested in vivo, we generated both control and MKRN3-restored H1703 xenografts in nude mice. MKRN3 restoration markedly attenuated tumor growth (Fig. 2, F and G). To further test the role of MKRN3 in lung cancer, we extended these findings to a second NSCLC cell line (H1437, lung adenocarcinoma) with MKRN3 inactivation (Fig. S2 A). Consistent with the above data, MKRN3 reexpression reduced the cell growth and proliferation in vitro and in vivo (Figs. 2 and S3). Collectively, these results demonstrate that MKRN3 inactivation promotes lung cancer tumorigenesis.

The incidence of cancer has been observed to increase markedly in tumor suppressor–deficient mice upon exposure to environmental insult. We speculated that inactivation of MKRN3 might also render mice susceptible to a second cancer trigger. To test this hypothesis, we challenged Mkrn3 KO (Li et al., 2020) and control littermates with urethane, a DNA alkylating agent that is classified as a chemical carcinogen and is widely used to induce lung tumor formation in mice (McLoed et al., 2016; Rex et al., 2016; To et al., 2008; Fig. 3 A and Fig. S4 B). Consistent with previous reports (McLoed et al., 2016), injections of urethane resulted in a very low incidence of lung tumors from mice with WT genotypes (Fig. 3, B and C). Only a low percentage of WT mice developed lung tumors (Fig. 3, B–D). Notably, almost all Mkrn3 KO (Mkrn3^p-/m+) mice developed large and rapidly growing lung tumors in response to urethane (Fig. 3, B, D, and E). Within the same mice, multifocal lung tumor development was observed, suggesting multiple transformation events (Fig. 3, B and C). Tumors were histologically classified as adenocarcinomas (Fig. 3 C). During the observation period, we did not observe metastatic spread of these primary lung tumors. Our results show that loss of MKRN3 renders mice susceptible to urethane-induced lung cancer.

MKRN3 genomic alterations are enriched in human lung cancer samples harboring oncogenic KRAS mutations (Fig. S1 A and Table 1). To genetically validate the synergistic effect of the two alterations in the lung tumors studied here, we performed mouse studies. The commonly used K-ras^LSL-G12D/+ (K) mouse model harbors Cre-inducible alleles of K-ras gene and develops lung tumors. To enable inducible deletion of Mkrn3 specifically in lung cells, mice with a floxed Mkrn3 exon 1 allele were generated. We combined the Mkrn3^pfl/m+ alleles with K mouse and targeted Cre expression to the lungs (Fig. S4, C and D) using intranasal adenovirus delivery (DuPage et al., 2009; Xie et al., 2018; Fig. 4 A).

Next, cohorts of Ad-Cre–infected K-ras^LSL-G12D/+ (K) and K-ras^LSL-G12D/+; Mkrn3^pfl/m+ (KM) mice were followed over time. The KM mice showed earlier onset and a more rapid progression of the alveolar lesions. Histopathological analysis of their lungs at 12 mo after Ad-Cre infection revealed a significant increase in tumor burden compared with that in K controls (Fig. 4, B–E). Moreover, KM mice displayed advanced adenocarcinomas, a tumor stage that is extremely infrequent at this time in tumors driven by K-rasG12D alone (Fig. 4, B–E). To quantify the extent of progression of KM tumors, we devised a grading system by which to evaluate the stage of every tumor in each mouse as previously described (Jackson et al., 2005). Using this grading scheme, we confirmed that MKRN3 inactivation resulted in a markedly more severe tumor phenotype in KM mice (Fig. 4 E and Fig. S4 E). KM mice developed lung adenocarcinomas, identified by positive staining for TTF1⁺ (immunostaining biomarker for adenocarcinoma) and negative staining for p40 (immunostaining biomarker for squamous cell carcinomas; Yatabe et al., 2019; Fig. 4 F). Altogether, these observations demonstrate that MKRN3 inactivation in the presence of K-ras (G12D) mutations accelerated lung tumor progression.

What are the molecular mechanisms underlying MKRN3 tumor suppression? MKRN3 is a ubiquitin (Ub) E3 ligase that facilitates the ubiquitination of target proteins (Abreu et al., 2013; Li et al., 2021; Li et al., 2020). To identify the potential substrates for MKRN3 in the cancer context, Flag-tagged MKRN3 was transduced into MKRN3-inactivated H1703 lung cancer cells as bait. MKRN3 was recovered along with proteins that potentially formed complexes with MKRN3 through coimmunoprecipitation with anti-Flag beads, and the samples were subjected to mass spectrometry–based proteomics screening. As shown in Fig. 5 A, PABPC1 protein emerged as the hit with the highest confidence. The interaction between endogenous PABPC1 and restored Flag-MKRN3 was validated in H1703 cells (Fig. 5 B) and was further confirmed in an additional human lung adenocarcinoma cell line H1437 (Fig. 5 B).

We next sought to determine whether MKRN3 ubiquitinates PABPC1 and whether cancer-derived mutations in MKRN3 alter its Ub ligase activity. We started with an assay in HEK293T cells to measure the ubiquitination function of MKRN3. MKRN3 was found to interact with and ubiquitinate ectopically expressed PABPC1 in the HEK293T model (Fig. 5 C and Fig. S5 A) and significantly increase the ubiquitination of the endogenous PABPC1 in lung cancer cells (Fig. 5 E). MKRN3 missense mutations identified in NSCLC patients were found to substantially compromise MKRN3-mediated PABPC1 ubiquitination (Fig. 5 D).

Previous studies showed that several proteins interact with PABPC1 via PABP-interacting motif 2 (PAM2; Xie et al., 2014), which contained a conserved peptide sequence xxLNxxAxEFxPxxx (Fig. 5 F). Thus, we speculated that MKRN3 interacts with PABPC1 via PAM2 at amino acid positions 197–211. To define the sites in MKRN3 that are responsible for recognizing PABPC1, we cotransfected HEK293T cells with Flag-PABPC1 and either WT or mutant MKRN3 (A203S, F206A, P208A, A203S/F206A, or A203S/F206A/P208A). Flag-PABPC1 coimmunoprecipitation showed that WT, MKRN3^A203S, and MKRN3^P208A had equivalent interactions with PABPC1, while the interaction between MKRN3^F206A and PABPC1 was significantly impaired (Fig. 5 G). Furthermore, MKRN3^A203S/F206A and MKRN3^{A203S/F206A/P208A} exerted weaker interactions with PABPC1 than WT MKRN3 (Fig. 5 G). Strikingly, MKRN3^{A203S/F206A/P208A} decreased the ubiquitination of the PABPC1 protein (Fig. 5 G) and attenuated MKRN3 tumor suppression properties in lung cancer cells (Fig. 5, H and I). Overall, these results illustrate that MKRN3 directly interacts with PABPC1, and the A203/F206/P208 of MKRN3 are crucial interaction residues mediating PABPC1 ubiquitination.

Interestingly, MKRN3-mediated ubiquitination of PABPC1 seemed to have no effect on the stability of PABPC1 (Fig. 5 B). Transfection of MKRN3 in HEK293T cells resulted in increased PABPC1 ubiquitination in the presence of WT Ub or Ub that only forms K63 linkages (UbK63), one type of poly-Ub chains that typically leads to signal transduction (Komander and Rape, 2012; Kwon and Ciechanover, 2017), compared with HEK293T cells transfected with empty vector (Fig. 6 A). In contrast, the expression of MKRN3 together with Ub that only forms K48 linkages (UbK48), one type of poly-Ub chains that typically leads to protein degradation (Komander and Rape, 2012; Kwon and Ciechanover, 2017), did not result in PABPC1 ubiquitination in HEK293T cells (Fig. 6 A). In addition, a cycloheximide chase experiment also showed that compared with MKRN3-C258*, MKRN3 did not induce PABPC1 degradation in H1703 cells (Fig. 6 B).

To further elucidate the molecular mechanisms underlying MKRN3-mediated ubiquitination of PABPC1, we found that the second RNA recognition motif (RRM) in PABPC1 was required for PABPC1 ubiquitination (Fig. 6 C and Fig. S5 B). We then mutated the eight lysine residues individually in RRM2. Our strategy was to mutate these putatively ubiquitinated lysines to arginines, since arginine cannot be ubiquitinated but retains the positive charge. Single mutations of K104R, K108R, K113R, K129R, K138R, K157R, and K174R showed reductions in the overall level of PABPC1 ubiquitination (Fig. 6 D). The double mutant (K104R/K108R; 2KR), triple mutant (K104R/K108R/K113R; 3KR), quadruple mutant (K104R/K108R/K113R/K129R; 4KR), 5KR mutant (K104R/K108R/K113R/K129R/K138R), 6KR mutant (K104R/K108R/K113R/K129R/K138R/K157R), and 7KR mutant (K104R/K108R/K113R/K129R/K138R/K157R/K174R) were then constructed. PABPC1 ubiquitination was at the lowest level detected in this 7KR mutant (Fig. 6 E), indicating that all seven lysines are subject to PABPC1 ubiquitination. Interestingly, these seven lysine residues are conserved across human, rabbit, rat, mouse, zebrafish, and frog (Fig. S5 C).

To determine whether MKRN3 acts as a tumor suppressor by mediating ubiquitination of PABPC1, a series of rescue experiments was performed in a lung cancer context. Endogenous PABPC1 was knocked out in H1703 cells with a CRISPR/Cas9 approach, followed by reconstitution of WT or 7KR mutant PABPC1 (Fig. 6 F). In this MKRN3-inactivated lung cancer context, MKRN3 restoration promoted ubiquitination of WT PABPC1 and exhibited tumor suppression properties, while enforced MKRN3 expression in PABPC1 mutant lung cancer cells attenuated various MKRN3 tumor suppression properties (Fig. 6, G and H). These data show that MKRN3 functions as a tumor suppressor in lung cancer largely through PABPC1 ubiquitination.

PABPC1 is one of the 1,000+ RNA-binding proteins found in humans (Xie et al., 2014). Previous studies established that PABPC1 and eIF4G are direct interacting components in the translation initiation complex (TIC), and the PABPC1-eIF4G interaction is important for both the TIC formation and overall mRNA translation efficiency (Tritschler et al., 2010; Xie et al., 2014). Then we explored whether MKRN3 affects the interaction between PABPC1 and eIF4G. As shown in Fig. 7 A, the same amounts of antibody enriched almost equal PABPC1 proteins in both groups, while less eIF4G was present in the MKRN3-restored group. Therefore, MKRN3 restoration hinders the formation of PABPC1-eIF4G complex.

We next evaluated whether MKRN3 inactivation can affect global protein translation. Protein synthesis is globally repressed in MKRN3 restored lung cancer cells as measured by L-azidohomoalanine (L-AHA) incorporation (Fig. 7 B). Since MKRN3 restoration induced cell cycle arrest in G2/M phase (Fig. 2 D), we investigated transcription and expression level of CCNB1, which is a cell cycle regulator of G2/M phase. We found that PABPC1 does bind to the 3′ poly(A) tail of CCNB1 mRNA in the lung cancer context (Fig. 7 C). CCNB1 encodes cyclin B1, which functions as a cell-cycle regulator, and its overexpression has been shown to correlate with tumor progression (Fang et al., 2014). With the particular interest in the homeostasis of CCNB1 mRNAs, then we asked whether MRKN3-mediated PABPC1 ubiquitination would have any effect on their binding to the poly(A) tail of CCNB1 mRNA. As shown in Fig. 7 D, RNA immunoprecipitation (IP) assay indicated that the same amounts of anti-Flag (pulling down Flag-PABPC1 protein) did pull down significantly less CCNB1 mRNAs in MKRN3-restored lung cancer cells (Fig. 7 D). Down-regulation of cyclin B1 by MKRN3-PABPC1 was further validated at the protein level (Fig. 7 E) but not the mRNA level (Table S3).

All these data demonstrated that the ubiquitination of PABPC1 by MKRN3 attenuates its binding to 3′ poly(A) tails of mRNAs, and decreases the TIC formation, thereby repressing global protein synthesis and maintaining lung cancer cells with limited proliferative capacity (Fig. 7 F). In contrast, inactivated MKRN3 reduces PABPC1 ubiquitination, promotes its binding to 3′ poly(A) tails of mRNAs, and thereby accelerates global protein synthesis and promotes lung cancer proliferation and progression (Fig. 7 F).

The molecular mechanism underlying the progression of NSCLC is not fully understood. Our findings of recurrent genomic alterations, together with various functional, mouse model, and mechanistic data herein, highlight the MKRN3 gene as a bona fide tumor suppressor gene contributing to NSCLC progression. Inactivated MKRN3 is unable to ubiquitinate PABPC1 and leads to an increase in the global translation rate and altered pattern of mRNA translation, thereby contributing to the tumor proliferation and progression. We believe our findings have important implications for understanding the molecular events shaping oncogenesis in NSCLC and for elucidating the genetic drivers of PABPC1 ubiquitination. A deep understanding of the mechanisms underlying E3 ligase regulation and function in tumorigenesis is expected to identify novel prognostic markers and to enable the development of the next generation of anticancer therapies (Hyer et al., 2018; Kim et al., 2017; Popovic et al., 2014; Senft et al., 2018).

It is intriguing that MKRN3 inactivating mutations are so frequent in NSCLC given that MKRN3 is ubiquitously expressed (Fig. S4 A). Certainly this is one limitation of our results. Nonetheless, our results indicate that the MKRN3-PABPC1 pathway plays a prominent role in lung cancer pathogenesis. It is also striking that humans with germline MKRN3 mutations only develop an overt pathology within the neuroendocrine brain that ultimately regulates the reproductive axis, a disease called CPP (Abreu et al., 2013; Latronico et al., 2016; Shin, 2016). From a pathophysiologic standpoint, it seems that inactivation of certain genes, such as MKRN3, leads to distinct physiological outcomes depending on the cellular context. Indeed, we previously demonstrated that the DMD and DEPDC5 genes function as tumor suppressors (Pang et al., 2019; Wang et al., 2014), but their germline mutations result in muscular dystrophy and focal epilepsy, respectively (Hoffman et al., 1987; Ishida et al., 2013). Notably, cohorts of individuals with CPP show an increased risk of malignancies such as lung cancers (Ben Khedher et al., 2017; Calcaterra et al., 2013; Kreuzer et al., 2003). It is hard to imagine that two entirely different classes of disease, such as lung cancer and CPP, can converge at a critical point attributable to a single gene, MKRN3. There is increasing evidence that sex steroid hormones exert effects in nonreproductive organs, such as the lungs (Ben Khedher et al., 2017; Kreuzer et al., 2003). Sex steroid hormones and reproductive factors also play a role in the genesis of lung cancer (Ben Khedher et al., 2017; Kreuzer et al., 2003). Nonetheless, our findings provide a potential molecular mechanism. This mysterious and hidden connection may prove a boon in disguise and has raised hopes that studying the biology of one disease may help to identify novel therapeutic targets for the other.

Germline MKRN3 mutations also cause Prader–Willi syndrome in humans. Reports of spontaneous lung cancer in patients with Prader–Willi syndrome suggest that germline MKRN3 inactivation may predispose to lung cancer (Nenekidis et al., 2011) and suggest a possible link between genetic lesions present in Prader–Willi syndrome and enhanced/early-onset carcinogenesis in lung tumors. MKRN3 interacts with p53-regulated metabolic genes in a precocious puberty context (Yellapragada et al., 2019). Interestingly, genomic MKRN3 aberrations occur frequently in human NSCLC samples harboring TP53 mutations, suggesting that TP53 and MKRN3 mutations are not necessary in a mutually exclusive manner in the lung cancer context.

In summary, our genomic, functional, genetically engineered mouse model and mechanistic findings herein demonstrate that the tumor suppressor roles of MKRN3 contribute to NSCLC progression. MKRN3 dysregulation warrants evaluation as a potential point of therapeutic attack in KRAS-driven NSCLCs.

We curated freely available somatic mutations from the data portal of TCGA (Cerami et al., 2012; Gao et al., 2013) and other published studies (Campbell et al., 2016; George et al., 2015; Govindan et al., 2012; Hellmann et al., 2018; Hoadley et al., 2018; Imielinski et al., 2012; Jamal-Hanjani et al., 2017; Peifer et al., 2012; Rudin et al., 2012; Shi et al., 2016).

De-identified snap-frozen tumor biopsies and matched normal samples were from lung cancer patients at Shanghai Changzheng Hospital (approval no. 2019SL009). All samples were collected with institutional review board approval. Informed written consent was obtained from all human participants.

To study the possible relationship between MKRN3 gene expression and survival, we first sorted patients from the TCGA and GEO datasets into MKRN3-high and MKRN3-low groups according to the MKRN3 mRNA relative expression levels (top 50% versus bottom 50%). Next, we analyzed the difference in survival between these two groups using GraphPad Prism 6.

HEK293T cells (American Type Culture Collection [ATCC]; catalog #ACS-4500) and the NSCLC cell lines NCI-H1703 (ATCC; catalog #CRL-5889) and NCI-H1437 (ATCC; catalog CRL-5872) were purchased from ATCC. HEK293T, NCI-H1703, and NCI-H1437 cells were maintained in RPMI 1640 (HyClone; #SH30027.01) medium containing 10% FBS (Thermo Fisher Scientific #10099141) and penicillin/streptomycin (Thermo Fisher Scientific; #15140122). All these cells were cultured at 37°C in a 5% CO₂ humidified atmosphere. None of the cell lines in this study appeared in the misidentified cell line list maintained by the International Cell Line Authentication Committee. All cell lines were routinely tested for microbial contamination (including mycoplasma).

The MKRN3 and PABPC1 expression plasmids were constructed by cloning the corresponding cDNAs into the p3×FLAG-CMV^TM-7.1 vector (Sigma-Aldrich; #E7533). PABPC1-ΔRRM1 (deleting aa 11–89), PABPC1-ΔRRM2 (deleting aa 99–175), PABPC1-ΔRRM3 (deleting aa 191–268), and PABPC1-ΔRRM4 (deleting aa 294–370) were generated by subcloning the corresponding cDNAs into the p3×FLAG-CMV^TM-7.1 vector. Hemagglutinin (HA)-Ub, HA-Ub-K48, and HA-Ub-K63 vectors were generously provided by Dr. Shao-Cong Sun (Department of Immunology, MD Anderson Cancer Center, The University of Texas, Houston, TX). PABPC1 K to R point mutants and MKRN3 point mutants were generated by introducing these mutations into the WT expression vector using the QuikChange Lightning Site-Directed Mutagenesis Kit (#210518). The PABPC1 single guide RNA (sgRNA; 5′-GCC​CGG​CGC​TCA​CCG​TCC​GC-3′) vector was generated by cloning the PABPC1 targeting sgRNA into the lentiCRISPRv2 vector (Addgene; plasmid #52961). MKRN3 and PABPC1 lentiviral constructs were generated by cloning the corresponding cDNAs into the pCDH-CMV-MCS-EF1-Puro lentiviral expression vector (System Biosciences; catalog #CD510B-1). Lentivirus particles were generated by cotransfecting these lentiviral constructs and helper virus packaging plasmids pCMVΔR8.9 and pHCMV-VSV-G into HEK293T cells using Lipofectamine 3000 (Invitrogen; #L3000015). Lentiviruses were harvested after 24, 36, 48, and 60 h and frozen at −80°C in aliquots at appropriate amounts for infection.

Tissues were homogenized in TRIzol Reagent (Invitrogen; #15596026), followed by total RNA isolation using the standard protocol. RNA was further reverse-transcribed into cDNA using the HiScript III first Strand cDNA Synthesis Kit (+ genomic DNA wiper; Vazyme; #R312-01). PCR was performed for gene expression analysis of MKRN3 using 2×Taq Master Mix (Dye Plus; Vazyme; #P112-01); samples were run with non-RT or nontemplate control. Quantitative PCR was performed for target gene expression analysis using the ChamQ Universal SYBR qPCR Master Mix (Vazyme; #Q711-02). Samples were run in triplicate with non-RT or non–template control. Amplification accuracy was verified by melting curve analysis.

6-well plates were first layered with 0.6% bottom agar (Noble agar; BD Difco; #214220) containing RPMI 1640 medium with 10% FBS and penicillin/streptomycin. H1703 (10,000 cells per well) and H1437 (10,000 cells per well) cells were transduced with control or MKRN3 lentivirus and seeded in 0.35% top agar containing 10% FBS and penicillin/streptomycin. Cells were allowed to grow for 4 wk and then stained with 1 ml of 1 mg/ml methyl thiazol tetrazolium (Sigma-Aldrich; #M5655) for 3 h. Colonies were counted by ImageJ software (National Institutes of Health). All assays were performed in triplicate wells, with the entire study replicated three times. A colony formation assay was conducted by seeding H1703 (500 cells per well) cells or H1437 (500 cells per well) cells transduced with the WT and truncating MKRN3 lentivirus into 6-well plates and allowed to grow for 3 wk. Then, the cells were fixed with 4% paraformaldehyde for 30 min and stained with crystal violet solution (Wuhan Servicebio Technology Co.; #G1014) for 3 h. All assays were performed in in triplicate wells, with the entire study replicated three times. Images were obtained using a scanner (Microtek; TMA 1600III).

For cell cycle analysis, cells were grown until 70% confluence, serum-starved for 8 h (H1703) or 40 h (H1437) and washed with PBS. Cells were harvested and fixed in 70% ethanol for 24 h, resuspended in 50 µg/ml of propidium iodide (Sigma-Aldrich; #P4170) and 100 µg/ml of RNaseA (TIANGEN; #RT405) containing PBS solution after centrifugation, then were analyzed using the Gallios Flow Cytometer (Beckman Coulter) and ModFit LT software. For apoptosis analysis, cells were grown to 80% confluence, serum-starved for 3 d, then washed with PBS and harvested. Cells were stained with the APC Annexin V Apoptosis Detection Kit with 7-AAD (BioLegend; #640930) and evaluated by flow cytometry (Beckman Coulter; Gallios Flow Cytometer) according to the manufacturer's protocol.

Whole-cell lysates from cell lines or frozen tissues were prepared using IP buffer (50 mM Tris-HCl, pH 8.0, 1% NP-40, 100 mM sodium fluoride, 2 mM sodium molybdate, 30 mM sodium pyrophosphate, 5 mM EDTA, and 2 mM sodium orthovanadate) containing protease inhibitors (10 µg/ml leupeptin, 10 µg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride). The lysates were then rocked for 8 h (cell lines) or overnight (frozen tissues) at 4°C and cleared by centrifugation at 14,000 rpm for 30 min at 4°C. The protein concentrations in the lysate were determined using a Quick Start Bradford 1× Dye Reagent (Bio-Rad; #5000205). Electrophoresis and Western blotting were performed using standard techniques. The hybridization signals were detected by chemiluminescence (Immobilon Western, Millipore Corporation) and captured using an Amersham Imager 600 imagers (GE Healthcare; #29083461). The primary antibodies were list as follows: MKRN3 (Sigma-Aldrich, #HPA029494; Abcam, #ab177203), PABPC1 (ProteinTech; #10970-1-AP), Flag (Sigma-Aldrich; #F-1804), HA (ProteinTech; #51064-2-AP), Ub (Santa Cruz; #sc-8017), PCNA (Santa Cruz; #sc-56), β-actin (Sigma-Aldrich; #A4700), and GAPDH (Sigma-Aldrich; #G8795). Relative protein quantification was performed with Image Quant TL 8.1 (GE Healthcare) software. For cycloheximide chase assay, cells were treated with cycloheximide at 10 µM and harvested at the indicated time.

Immunohistochemistry was performed on tissue and tumor sections using MKRN3 (Sigma-Aldrich; #HPA029494), p40 (Maxim Biotech; #RMA-0815), or TTF1 (Maxim Biotech; #MAB-0599). Four-micron slides were deparaffinized in xylene and hydrated in a graded series of alcohol. Slides were then boiled by microwave for 12 min in citrate buffer (pH 6). Immunohistochemistry reactions were visualized by diaminobenzidine staining, using an EnVision+ system (Dako).

The animal experiments were approved by the Institutional Animal Care and Use Committee of the Shanghai Institutes for Biological Sciences, Chinese Academy of Science (approval no. SIBS-2017-WYX-1). All genetically engineered mice were from the C57BL/6 background. MKRN3 straight KO (Mkrn3^p-/m+) mice were generated using the transcription activator-like effector nuclease–based approach to create a 2-bp deletion as previously described (Li et al., 2020; Fig. S4 B). K-ras^LSL-G12D/+ mice were purchased from The Jackson Laboratory (stock #008179). Mkrn3^pfl/m+ mice were generated as described in the conditional KO mouse model.

NCI-H1703 (2 × 10⁶ cells) and NCI-H1437 (2 × 10⁶ cells) cells transduced with MKRN3 WT and control lentivirus were injected subcutaneously into 6-wk-old male BALB/c nude mice, and tumor xenografts were allowed to grow for 4–5 wk. The resulting tumors were measured every 3 d. Tumor volume was calculated using the following formula: tumor volume = length × width × width/2. Once the largest tumor diameter reached the maximal tumor diameter allowed under our institutional protocol, all mice were killed, and tumors were collected. The maximal tumor diameter allowed by the Institutional Animal Care and Use Committee was 2.0 cm.

13 MKRN3^+/+ and 11 MKRN3^p-/m+ mice were given four intraperitoneal injections of urethane (Sigma-Aldrich; #U2500; 1 g/kg of body weight) weekly from 4 wk of age. Urethane was dissolved in 0.9% NaCl. Mice were euthanized by dislocation 12 mo after the last injection. Tumors were counted manually after dissection.

Mice with a floxed Mkrn3 exon 1 allele on the C57BL/6 background were generated as follows. We prepared a targeting construct including a 5.2-kb 5′ arm, a floxed fragment containing Mkrn3 exon 1 and a neomycin positive selection cassette flanked by flippase recognition target sites, a 6.1-kb 3′ arm, and a thymidine kinase negative selection cassette. JM8A3 C57BL/6 embryonic stem cells were electroporated with the targeting vector and selected using G418 and ganciclovir. Three clones were identified with long-range PCR from 144 clones. The clonally expanded embryonic stem cells were injected into C57BL/6 blastocysts to generate chimeric mice. The Mkrn3^pfl/m+ mice were generated by crossing the chimeric mice with expressing Flp recombinase mice. KM mice were generated by crossing MKRN3^pfl/m+ male mice with K-ras^LSL-G12D/+ female mice. K-ras^LSL-G12D/+ mice, Mkrn3^pfl/m+ mice, and K-ras^LSL-G12D/+; Mkrn3^pfl/m+ mice were infected by intranasal instillation with adenovirus as previously described (DuPage et al., 2009). Mice were euthanized by dislocation 12 mo after intranasal instillation.

Lungs were perfused through the trachea with 4% paraformaldehyde, fixed overnight, transferred to 70% ethanol, and subsequently embedded in paraffin. Sections were cut at a thickness of 4 µm and stained with H&E for pathological examination. H&E-stained slides were scanned by an Aperio Imagescope (Leica).

Cells were lysed by IP buffer containing protease inhibitors (10 µg/ml leupeptin, 10 µg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride), mixed with 1 µg anti-Flag antibody and 20 µl protein G-sepharose (Thermo Fisher Scientific; #101242), incubated overnight, and eluted by boiling with SDS loading buffer. The eluted samples were detected by SDS-PAGE followed by Coomassie staining (Invitrogen; #LC6025; Colloidal Blue Staining Kit). For mass spectrometry, IP samples were eluted by shaking with 8 M urea and 100 mM Tris-Cl, pH 8.0, and analyzed by mass spectrometry.

Cells were lysed by IP buffer containing protease inhibitors (10 µg/ml leupeptin, 10 µg/ml aprotinin, and 1 mM phenylmethylsulfonyl fluoride), mixed with 2 µg anti-Flag antibody and 20 µl protein G-sepharose, and incubated overnight. Immunoprecipitates were eluted by boiling with SDS loading buffer. IP samples and whole-cell lysates were analyzed by Western blotting.

293T cells were transfected with MKRN3-WT or mutants, Flag-PABPC1-WT or mutants, and HA-Ub-WT or mutants, and then whole-cell lysates were immunoprecipitated with 2 µg anti-Flag antibody and analyzed by Western blotting with an anti-HA antibody. To analyze the ubiquitination of PABPC1 in H1703 cells, H1703 cells were infected with MKRN3 or control lentivirus, and then whole-cell lysates were immunoprecipitated with 4 µg anti-PABPC1 antibody and analyzed by Western blotting with an anti-Ub antibody.

Total RNA was isolated from cells using standard Trizol protocol. Paired-end sequencing (2 × 100 bp) was performed with a BGI-500 instrument to obtain at least 20 million reads for each sample. The sequence data were processed and mapped to the human reference genome (hg19) using Bowtie2. Gene expression was quantified to fragments per kilobase per million mapped fragments using RNA sequencing by expectation maximization.

RIP assay was performed using the EZ-Magna RIP Kit (Merck; #17-701). Cells were cultured in a 10-cm dish to 80–90% confluence. One RIP reaction required 100 µl of cell lysate from ∼2.0 × 10⁷ cells. Cells were washed by PBS, collected by scraping, and suspended by RIP Lysis Buffer (Merck; #CS203176). Next, 5 µg of purified antibodies or corresponding IgG was added to the 100-µl cell lysate, and the mixture was incubated with rotation overnight at 4°C. Anti-Flag (Sigma-Aldrich; #F-1804) and normal mouse IgG (Merck; #CS200621) were used for RIP assay. The immunoprecipitated RNA was purified and analyzed with RT quantitative PCR.

300,000 cells were seeded and transduced with MKRN3 and control lentivirus. 60 h later, the medium was replaced with methionine-free medium (Thermo Fisher Scientific; #A1451701) for 2 h. Cells were incubated with 50 µM L-AHA (Clickchemistrytools; #1066-25) for 8 h and harvested in lysis buffer. L-AHA was detected using click chemistry with 50 µM Biotin Alkyne (Clickchemistrytools; #1266–5) with the Click-&-Go Protein Reaction Buffer Kit (Clickchemistrytools; #1262) per the manufacturer’s protocol. The extracted protein was detected by Western blotting as described above. The blots were probed with a horseradish peroxidase–conjugated streptavidin (Proteintech; #SA0001) and developed with Enhanced Chemiluminescence (GE; #RPN2232).

The DISCOVER method was used to determine significant mutual exclusivity and cooccurrence between MKRN3 and KRAS mutations in the five NSCLC cohorts as previously described (Canisius et al., 2016).

Statistical analysis was performed using GraphPad Prism 6 (GraphPad Software). The difference between two groups was analyzed by paired sample t test. The differences among multiple groups were analyzed by one-way ANOVA.

Fig. S1 presents the genomic MKRN3 aberrations in human NSCLCs. Fig. S2 shows effects of 5-aza-2′-deoxycytidine on MKRN3 expression in NSCLC cell lines and cell apoptosis caused by MKRN3 restoration in MKRN3-inactivated lung cancer cells. Fig. S3 demonstrates that MKRN3 restoration in MKRN3-inactivated H1703 lung cancer cells reduces cell proliferation. Fig. S4 displays MKRN3 inactivation in Mkrn3 straight KO (Mkrn3^p-/m+) mice and Mkrn3 conditional KO (Mkrn3^pfl/m+) mice. Fig. S5 shows that MKRN3 mediates PABPC1 ubiquitination. Table S1 shows clinicopathologic classification for 67 NSCLCs with genomic MKRN3 aberrations from TCGA lung cancer datasets. Table S2 shows clinicopathologic classification and MKRN3 protein expression in 30 NSCLCs from Asian patients. Table S3 shows gene expression profiling by RNA sequencing 96 hours after MKRN3 restoration in MKRN3-deficient/MKRN3-mutant H1703 cells. Table S4 shows primers used in the study.

RNA sequencing datasets for MKRN3-deficient/MKRN3 mutant H1703 cells are available in the National Omics Data Encyclopedia (https://www.biosino.org/node/) under accession no. OEP002179.

We thank Drs. Ronggui Hu and Chuanyin Li (State Key Laboratory of Molecular Biology, Shanghai Institute of Biochemistry and Cell Biology, Center for Excellence in Molecular Cell Science, Chinese Academy of Sciences, Shanghai, China) for providing MKRN3 straight KO mice and sharing PABPC1 as a substrate of MKRN3 in the precocious puberty context. We also thank Dr. Jun Qin (Shanghai Institute of Nutrition and Health, Chinese Academy of Sciences, Shanghai, China) for Kaplan–Meier survival analysis and all members of the Wang laboratory for thought-provoking discussion, technical assistance, and help with manuscript preparation.

This work was supported by grants from the National Natural Science Foundation of China (82072974, 81572642), the Basic Research Project of Shanghai Science and Technology Commission (20JC1419200), the National Key Research and Development Program of China (2016YFC1302100), the China Medicine Education Association (2021-002), the Key Laboratory of Tissue Microenvironment and Tumor of the Chinese Academy of Sciences (202002), the Chinese Academy of Sciences, and Shanghai Changzheng Hospital.

Author contributions: Y. Wang and K. Li conceived and designed the research. K. Li, X. Zheng, H. Tang, Y.-S. Zang, C. Zeng, X. Liu, Y. Shen, Y. Pang, S. Wang, F. Xie, X. Lu, Y. Luo, Z. Li, W. Bi, X. Jia, T. Huang, Q. Zhu, X. Zhang, and Y. Wang performed experiments. H. Tang, R. Wei, K. Huang, and Z. Chen provided samples and clinical data. K. Li, X. Zheng, H. Tang, Y.-S. Zang, C. Zeng, X. Liu, Y. Shen, Y. Pang, S. Wang, F. Xie, X. Lu, Y. Luo, Z. Li, W. Bi, X. Jia, T. Huang, Q. Zhu, Y. He, M. Zhang, Z. Gu, Y. Xiao, X. Zhang, and Y. Wang analyzed data. Y. Zhang, M. Zhang, Z. Gu, Y. Xiao, X. Zhang, and J.A. Fletcher provided scientific advice and helpful comments on the project. K. Li, X. Zheng, and Y. Wang wrote the manuscript. All authors read and approved the final manuscript.