https://orcid.org/0000-0002-0477-7365
Figures
Abstract
Exploiting the immune system to eradicate cancer cells is an area of intense clinical study. However, the mechanisms that shape the tumor-immune microenvironment are incompletely understood. Here, we identify Muscleblind-like (MBNL) proteins as novel modulators of the tumor-immune microenvironment across diverse cancers. We demonstrate that loss of tumor MBNL expression results in an attenuated response to interferon gamma and reduced tumor antigen presentation in melanoma, breast cancer, and colorectal cancer cells. Parallel experiments in a syngeneic mouse melanoma model revealed that MBNL loss reduces tumor cell killing by CD8 + T cells in vitro and facilitates tumor escape from cytotoxic CD8 + T cell infiltration in vivo. Finally, we extended these studies to 29 human cancer types to find that MBNL expression levels are strongly associated with gene expression signatures of T cell tumor infiltration. These insights suggest that MBNL proteins play important roles in shaping the immune landscape across diverse malignancies.
Citation: Gabel AM, Crosse EI, Belleville AE, Hogg SJ, McKellar SA, Abdel-Wahab O, et al. (2025) Muscleblind-like proteins are novel modulators of the tumor-immune microenvironment. PLoS One 20(4): e0321148. https://doi.org/10.1371/journal.pone.0321148
Editor: Atsushi Asakura, University of Minnesota Medical School, UNITED STATES OF AMERICA
Received: April 19, 2024; Accepted: March 2, 2025; Published: April 30, 2025
Copyright: © 2025 Gabel et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The RNA-sequencing data reported in this paper have been deposited in the Gene Expression Omnibus (GEO accession number: GSE261415).
Funding: A.M.G and S.A.M are ARCS Foundation scholars. E.I.C is a Damon Runyon fellow. J.D.T. was supported by the NIH/NCI (K99 CA263168). R.K.B. was supported in part by the NIH/NCI (R01 CA251138), NIH/NHLBI (R01 HL128239, R01 HL151651) and the Blood Cancer Discoveries Grant program through the Leukemia & Lymphoma Society, Mark Foundation for Cancer Research, and Paul G. Allen Frontiers Group (8023-20). R.K.B is a Scholar of The Leukemia & Lymphoma Society (1344-18) and holds the McIlwain Family Endowed Chair in Data Science. Computational studies were supported in part by FHCC’s Scientific Computing Infrastructure (ORIP S10 OD028685). Experimental studies were supported in part by the Experimental Histopathology, Flow Cytometry, and Genomics Shared Resources of the Fred Hutch/University of Washington Cancer Consortium (NIH/NCI P30 CA015704). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: RKB is a founder and scientific advisor of Codify Therapeutics and Synthesize Bio and holds equity in both companies. OA-W is a founder and scientific advisor of Codify Therapeutics and holds equity in this company. RKB and OA-W have received research funding from Codify Therapeutics unrelated to the current work. OA-W has served as a consultant for Foundation Medicine Inc., Merck, Prelude Therapeutics, Amphista Therapeutics, MagnetBio, and Janssen, and is on the Scientific Advisory Board of Envisagenics Inc., Harmonic Discovery Inc., and Pfizer Boulder; OA-W has received prior research funding from H3B Biomedicine, Nurix Therapeutics, Minovia Therapeutics, and LOXO Oncology unrelated to the current manuscript. This does not alter our adherence to PLoS One policies on sharing data and materials. The remaining authors declare no competing interests.
Introduction
Immunotherapy, which exploits the power of the immune system to fight malignant cells, has transformed cancer treatment within the past decade. Immunotherapies are exemplified by both chimeric antigen receptor (CAR) T cells, designed with high specificity for tumor antigens, and immune checkpoint blockade inhibitors, which block tumor immune evasion mechanisms from cytotoxic T cell killing of cancer cells. The number of different strategies is rapidly growing, including those that recruit natural killer (NK) cells and macrophages, but the currently most clinically advanced therapies are designed to stimulate a robust T cell response directed against the tumor cells [1]. This growing field has demonstrated remarkable successes, such as the complete remission of patients with B cell acute lymphoblastic leukemia following treatment with CAR-T cells targeting CD19 + B cells [2], and immune checkpoint inhibitors such as Ipilimumab (anti-CTLA-4), Pembrolizumab, and Nivolumab (anti-PD-1) have significantly improved survival in patients with metastatic melanoma [3].
Despite these clinical breakthroughs, the efficacy of immunotherapies varies significantly across cancer types and among individual patients. Particularly for immune checkpoint blockade, a crucial determinant for immunotherapeutic success is the extent of T cell immune infiltration within the tumor microenvironment. Indeed, ‘hot’ tumors, characterized by substantial adaptive immune responses, have been linked with more effective responses to checkpoint blockade [4]. Thus, deciphering the mechanisms that shape the tumor-immune microenvironment is essential for devising strategies to transform ‘cold’ tumors, those with minimal immune infiltration, into ‘hot,’ immunoreactive ones.
In this context, our report unveils a role for Muscleblind-like (MBNL) proteins as novel regulators of resistance to T cell-mediated tumor killing through alterations in antigen presentation. In the human genome, the three MBNL genes encode RNA-binding proteins with roles including regulation of alternative RNA splicing, polyadenylation, and localization [5–7]. The pathogenesis of MBNL dysregulation is best described for myotonic dystrophy [8], a neuromuscular disease caused by loss of MBNL function that results in myotonia, cardiomyopathy, and central nervous system abnormalities. Additionally, there is growing evidence for MBNL’s role in cancer. MBNL expression is downregulated in colorectal [9] and prostate [10] cancers. In the case of prostate cancer, however, there is additionally increased inclusion of MBNL exon 7, associated with pro-tumor activity and indicative of a splice isoform-specific role in cancer. Conversely, MBNL1 is consistently overexpressed in MLL-rearranged leukemia [11]. Furthermore, MBNL1 has been reported to play roles in the inhibition of glioblastoma tumor initiation [12] and suppression of metastasis in breast [13] and colon [14] cancers.
While most studies have focused on cancer cell-intrinsic roles of MBNLs, little is known about how they may shape the tumor microenvironment. In this study, we investigated and characterized a novel role for MBNL proteins in anti-tumor immune responses. Using mouse models of melanoma, we determine that MBNL loss attenuates MHC Class I antigen presentation, response to immune stimulation by interferon gamma, and subsequent cancer cell killing by cytotoxic T cells. Furthermore, we demonstrate that our results from mouse melanoma extend to multiple additional mouse cancer models and are consistent with transcriptional signatures of tumor-immune interactions from diverse clinical human cancers. The correlation between MBNL and T cell infiltration that we initially identified in mouse cancer models extends across many human cancer types based on genomic analyses of human clinical datasets. Overall, our study highlights an important role for MBNL proteins in shaping the tumor microenvironment, broadening our understanding of the intricacies of tumor-immune interactions.
Results
MBNL loss confers resistance to cytotoxic T cell killing of mouse melanoma cells
In a prior publication by Pan et al. [15], a genome-wide CRISPR-Cas9 screen was carried out on B16-F10 mouse melanoma cells to identify genes whose loss conferred resistance to antigen-specific killing by cytotoxic T cells. We mined this dataset to identify novel modulators of anti-tumor immune responses that were not highlighted or further validated in Pan et al.’s study. The screen first involved a pre-treatment with or without 1 ng/mL of IFNγ for 24 hours to stimulate MHC Class I surface expression, as B16-F10 cells have inherently low cell-surface MHC Class I levels. Then, B16-F10 cells were co-cultured with either of two types of cytotoxic T cells: (i) Pmel-1, which recognizes the endogenous melanoma antigen gp100, or (ii) OT-1, which, while having a low affinity for gp100, strongly recognizes Ova peptide (OVA), which was provided exogenously to tumor cells.
As expected and reported by Pan et al, investigation of the ranked list of genes enriched and depleted in the Pmel-1 screen by log fold change identified B2m as a top hit conferring resistance to T cell-mediated tumor killing (Fig 1A). B2M is a component of the MHC Class I complex and thus critical for T cell receptor recognition. We were intrigued to find that two members of the Muscleblind-like protein family, MBNL1 and MBNL2, also ranked very highly, indicative of a previously undescribed role for MBNL proteins in mediating cellular response to cytotoxic T cell killing.
(A) Rank plot of log fold-change of gRNA enrichment/depletion from a genome-wide CRISPR-Cas9 screen in B16-F10 melanoma cells under cytotoxic T cell selection pressure (data from Pan et al., 2018). gRNA enrichment indicates a gene whose loss confers resistance to cytotoxic T cell killing. (B) Log10 fold-change of normalized counts for B2m, Mbnl1 and Mbnl2 gRNA knockouts for T cell selection assays with or without irrelevant T cells, Pmel-1 T cells and with or without IFNγ stimulation.(C) Log10 fold-change of normalized counts for B2m, Mbnl1 and Mbnl2 gRNA knockouts for T cell selection assays stimulated with IFNγ with or without OT-1 T cells and with or without exogenous antigen OVA peptide.
A focused examination of B2m, Mbnl1, and Mbnl2 demonstrated a significant enrichment for gRNAs targeting all three of these genes when in the presence of Pmel-1 CD8 + cells (Fig 1B) following treatment with IFNγ. Of the two Mbnl genes enriched in the screen, Mbnl1 exhibited the larger positive log fold-change relative to both tumor cells without T cells and those co-cultured with irrelevant T cells that do not recognize the tumor antigen (Fig 1A, B). The same trend was seen for co-culture with OT-I T cells when stimulated with the exogenous antigen, OVA peptide (Fig 1C). Taken together, these data indicate that Mbnl1 and, to a lesser degree, Mbnl2, have direct roles in modulating MHC Class I-dependent killing of cancer cells by antigen-specific T cells.
MBNL modulates antigen presentation in response to IFNγ
Motivated by our re-analysis of the data from Pan et al., we next sought to further elucidate the role of MBNL proteins in shaping the tumor-immune microenvironment. We generated CRISPR-Cas9 knockout models of Mbnl1 (“1KO”) and Mbnl2 (“2KO”) in B16-F10 melanoma cells (Fig 2A). Although western blotting of polyclonal cells did not reveal complete absence of protein, we confirmed that the majority of alleles in the polyclonal population harbored indels via next-generation DNA sequencing (Fig S1A–S1D in S1 File). Additionally, we generated a double knockout (DKO) of both Mbnl1 and Mbnl2, as MBNL1 and MBNL2 are partially functionally redundant, and there is evidence that MBNL2 can be upregulated to compensate for MBNL1 loss [16]. Indeed, we observed an increase in MBNL2 protein levels with Mbnl1 KO (Fig 2A), emphasizing the importance of an Mbnl DKO model system to assess the role of MBNL proteins in IFNγ signaling.
(A) Western blot of B16-F10 Cas9 cells treated with the indicated gRNA or pgRNA (NTC = non-targeting control, 1KO = Mbnl1 KO, 2KO = Mbnl2 KO, DKO = Mbnl1 and Mbnl2 double KO). (B) Rank plot of log2 fold-change of genes in non-targeting control (NTC) treated vs untreated. Hallmark IFNγ response genes are highlighted in purple and selected genes also involved in MHC antigen presentation are labeled. (C) Heatmap of log2 fold-change for either NTC or double knockout (DKO) stimulated with IFNg compared to their unstimulated counterparts. Genes involved in antigen processing, transport to endoplasmic reticulum and components of the major histocompatibility complex are indicated with purple, green and red labels respectively. (D) Transcripts per million normalized B2m expression for both NTC and DKO with and without IFNγ stimulation. (E) Surface MHC Class I (MHC I) expression measured by flow cytometry after stimulation with an IFNγ dilution series. (F) Surface MHC I expression with and without 1 ng IFNγ stimulation for NTC, Mbnl2 knockout (2KO), Mbnl1 knockout (1KO) and DKO in a B16-F10 polyclonal cell line. (G) Percentage viable NTC and DKO B16-F10 tumor cells pulsed with OVA peptide after in vitro co-culture with OT-I CD8 + T cells.
To assess transcriptional alterations in response to IFNγ, we performed RNA-seq of Mbnl DKO cells compared to B16-F10 cells treated with a non-targeting control (NTC) gRNA treated with or without 1 ng/mL IFNγ for 24 hours. We first validated our IFNγ treatment by comparing treated and untreated cells and found that IFNγ response was the top enriched pathway in treated cells (Fig S2A in S1 File; S1 Table). As expected, top-ranking genes contributing to the pathway’s enrichment are involved in antigen presentation (Fig 2B). This includes proteasomes involved in antigen processing for presentation (Psmb2, Psmb8, Psmb9, Psmb10, Psme1, Psme2, and Psma3), Tap1, which transports antigens to the ER for attachment to MHC components, and components of the MHC Class I complex itself, including B2m and H2-D1.
We next used these data to determine how MBNL loss altered the IFNγ response by conducting a comparative analysis of the fold change in these enriched IFNγ response genes between treated and untreated groups for both NTC and Mbnl DKO cells. This revealed a substantial attenuation of the interferon response in Mbnl DKO cells (Fig 2C), including a diminished IFNγ-dependent induction of genes involved in antigen presentation. Notably, B2m expression exhibited significant attenuation in response to IFNγ for DKO cells (Fig 2D), suggesting that MBNL proteins contribute to the modulation of the tumor-immune interaction in part via alterations in expression levels of components of MHC Class I antigen presentation.
We then set out to validate that reduced levels of MHC Class I in MBNL-deficient cells contribute to the reduced T cell cytotoxic activity seen in the Pan et al. screen. We found that Mbnl DKO cells had significantly reduced levels of surface MHC Class I compared to NTC cells across all tested IFNγ concentrations, again demonstrating a suppression of induction of MHC Class I components in response to immune stimulation (Fig 2E). The single knockouts also reduced MHC Class I surface expression, with Mbnl1 KO (1KO) having a larger effect than Mbnl2 KO (2KO). However, Mbnl DKO had the most substantial reduction in MHC Class I, again suggestive of a functional redundancy between Mbnl1 and Mbnl2 (Fig 2F). As these experiments were performed in a polyclonal cell line with varying degrees of MBNL protein loss (Fig 2A), we generated genetically validated monoclonal cell lines and repeated the experiments for certainty of the results (Figure S2B-C in S1 File). Once again, surface expression of MHC Class I was significantly reduced (p = 0.0013) in the monoclonal DKO as compared with NTC cells (Fig S2D in S1 File).
We next tested whether a dependence on MBNL for MHC induction following IFNγ stimulation was a common feature of cancer cell lines beyond the B16-F10 model. We generated Mbnl single and double knockout lines using CRISPR-Cas9 in cell lines representing two different cancer types, E0771 breast cancer and MC38 colorectal carcinoma lines. For both of these cell lines, we observed a reduction in surface MHC expression for 1KO, 2KO and DKO (Figure S2E-F in S1 File). As expected from the B16-F10 cell data, Mbnl DKO reduced surface MHC Class I levels to a greater degree than Mbnl1 or Mbnl2 KO alone. Taken together, this provides additional validation and broader relevance for the role of MBNL in regulating tumor antigen presentation across multiple cancer types.
We next sought to test whether the impaired MHC antigen presentation in Mbnl DKO B16-F10 cells affected anti-tumor immune activity. Our re-analysis of data from Pan et al strongly suggested that MBNL loss should potentiate immune evasion, but as those results were conducted in the context of a highly multiplexed screen with single gene knockouts, we sought to test how dual knockout of Mbnl1/2 influenced T cell evasion in a focused assay. We pulsed B16-F10 Cas9 edited cells with OVA peptides and then co-cultured each cell line with OT-I specific CD8 + T cells. We observed a significant increase in the percentage of viable tumor cells after 24 hours in DKO cells compared to NTC cells (Fig 2G). These data indicate that MBNL deficiency and the resultant impaired MHC antigen presentation promotes cancer cell evasion of antigen-specific CD8 + T cell-mediated cytotoxicity.
MBNL loss facilitates resistance to T cell infiltration in vivo
Having established a role for MBNL in IFNγ response, MHC Class I antigen presentation and T cell-mediated tumor cell killing in vitro, we hypothesized that downregulation of MBNL would cause reduced tumor immune infiltration in vivo. We transplanted NTC, 1KO, and DKO cells into immunocompetent mice (Fig 3A). We did not observe any significant changes in overall tumor volume (Fig S3A in S1 File); however, we did observe a significant increase in host survival for DKO mice (Figure S3B in S1 File). Given that perturbation of Mbnl proteins in disease is known to be deleterious to cell health, we were not surprised to observe a slight reduction in tumor growth in DKO tumors. For example, Mbnl1 and Mbnl2 sequestration is known to contribute significantly to cellular dysfunction via RNA misprocessing in neuromuscular disease [17–19]. We screened publicly available data to find that CRISPR knockout of MBNL1 does not directly affect cancer cell viability (Figure S3C in S1 File). Thus, the combination of Mbnl1 and Mbnl2 DKO is likely more deleterious to cancer cell growth than the knockout of either gene alone, again highlighting the likely functional redundancy of these proteins.
(A) Experimental design schematic. (B-C) Surface MHC Class I (MHC I) expression measured by flow cytometry across genotypes for B16-F10 tumors; non-targeting control (NTC), B2m knockout, Mbnl1 knockout (1KO) and double knockout (DKO) ( * = p < 0.05 by a two-sided Student’s t-test). (D) Frequency of tumor immune infiltrate populations measured by flow cytometry; T cells, CD8 + frequency within T cells and NK cells ( * = p < 0.05 by a two-sided Student’s t-test). (E) Quantification of CD8 + cells infiltrates in tumors across genotypes by histology analysis ( * = p < 0.05, *** p < 0.0005 by a one-sided Wilcoxon signed-rank test). (F) Representative histology images of B16-F10 tumors across genotypes with CD8 + in brown DAB stain.
Analysis of the dissected tumors showed a reduction in MHC Class I surface expression to levels approaching that of a B2m KO control (Fig 3B and 3C). We assessed immune infiltration of the tumors by flow cytometry (Fig S3D in S1 File) and found a reduction in T cells (Fig 3D). Within this T cell population, there was a reduction in CD8 + cells and, to a lesser degree, CD4 + cells (Fig 3D; Figure S3E in S1 File). Similarly, there was a significant reduction in NK cells, potentially indicating a reduced anti-tumor immune activity that is not restricted to T cells (Fig 3D). Correspondingly, we assessed CD8 + cell frequency in B16-F10 tumors using immunohistochemistry and found a significant reduction in the number of CD8 + cells in DKO B16-F10 tumors relative to NTC B16-F10 tumors (Fig 3E-F). We did not observe any difference in 1KO B16-F10 tumors, which we speculate is due to the corresponding upregulation of Mbnl2 in Mbnl1 KO cells (Fig 2A). Taken together, these data strongly support the hypothesis that MBNL proteins help shape the tumor microenvironment by attenuating anti-tumor immune cell recruitment.
MBNL1 expression is strongly correlated with anti-tumor immune activity in clinical human cancers
We next sought to determine whether our findings in mouse tumor models are relevant for human cancers. In the human genome, as in mice, there are three MBNL family genes: MBNL1, MBNL2, and MBNL3. Using the TCGA melanoma (SKCM) RNA-seq dataset, we correlated MBNL expression with CD8A gene expression as a proxy for tumor T-cell infiltration (S2 Table). This showed a significant positive correlation (R = 0.27, p = 1.8e-08) for MBNL1, but not for MBNL2 or MBNL3 (Fig 4A).
(A) Correlation of MBNL1-3 expression and tumor infiltration indicated by CD8A expression (TPM = transcripts per million) for melanoma (SKCM), pancreatic adenocarcinoma (PAAD) and thymoma (THYM). (B) Gene Ontology (GO) terms enriched in the upper MBNL1 expression tercile bottom tercile within the SKCM patient cohort. (C) Plot of Pearson correlation coefficients for MBNL1-3 expression with tumor infiltrate (CD8A expression) across TCGA cancer datasets. (D) Rank plot of the median Pearson correlation coefficient for all gene expression with CD8A expression across all TCGA datasets. MBNL genes and selected immune infiltration related genes are labeled. (E) Waterfall plot of median Pearson correlation coefficient (r) across all analyzed TCGA cohorts comparing the indicated gene expression with gene expression of CD8A. Dotted lines indicated the top 5th, 10th and 25th percentiles of positively correlated genes.
We stratified the SKCM TCGA dataset by low (bottom tercile) and high (top tercile) MBNL1 expression and quantified changes in gene expression between the two bins. Higher MBNL1 expression was associated with significantly higher expression of genes related to immune pathways, particularly those relating to T cell activity (Fig 4B). Extending the analysis to other cancer types, we discovered that MBNL1 expression also had a high positive correlation with CD8A expression for several other cancer types, including Testicular Germ Cell Tumors (TCGT) and Pancreatic Ductal Adenocarcinoma (PAAD) (Fig 4A).
We correlated MBNL expression across all TCGA cancer datasets with tumor T cell infiltration by either CD8A expression (Fig 4C–D; S3 Table) or cytolytic T-cell (CYT) score (Fig S4A–B in S1 File). The CYT score is calculated as the geometric mean of the expression of two key cytolytic effectors that are significantly upregulated upon CD8 + T cell activation: granzyme B (GZMB) and perforin (PRF1) [20]. Strikingly, MBNL1 showed a significant positive correlation for T cell infiltration (based on CD8A expression) for 27 of 29 analyzed TCGA cancer types. To a lesser degree, MBNL3 had a positive correlation with predicted T cell infiltrations across some cancers, including BRCA and LGG, though it should be noted that MBNL3 expression is fairly low across all assayed tissues (Fig 4C–D; Figure S4A-B in S1 File Supplementary Figures). Conversely, MBNL2 trended to have a negative correlation with T cell infiltration (Fig 4C–D; Figure S4A-B in S1 File), which we hypothesize is largely due to the anti-correlated nature of MBNL1 and MBNL2 expression given their functional redundancy.
We correlated all genes included in the Pan et al. screen CRISPR library with CD8A expression across all cancers in the TCGA datasets. MBNL1 ranked within the top 5% of genes, indicating that MBNL1 is significantly more positively correlated with CD8A expression than are most genes in the genome (Fig 4E). Scoring above it were genes directly involved in interferon response (IFNγ, B2M) and T cell activation (PRF1, GZMA, GZMB, CD274, PDL1). This analysis predicts that MBNL1 has a profound effect on anti-tumor T cell infiltration, similar to that of genes with explicit roles in T cell-mediated killing of cancer cells.
Discussion
Our study highlights a critical role for MBNL proteins in shaping the tumor-immune microenvironment. In mouse melanoma, this was first revealed through a CRISPR knockout screen, where Mbnl1 and Mbnl2 emerged as top hits conferring tumor resistance to cytotoxic T cell-mediated killing [15]. Using our own CRISPR knockout models, we validated this finding by demonstrating reduced antigen-specific CD8 + T cell killing of Mbnl-deficient (DKO) cells. Beyond cytotoxic resistance, we observed a dampened transcriptional response to IFNγ stimulation, particularly in genes encoding the antigen presentation machinery and MHC class I complex. In vivo, melanoma tumors lacking Mbnl exhibited lower MHC class I expression and significantly reduced CD8 + T cell infiltration. Our findings suggest that MBNL loss could contribute to the downregulation of interferon signaling and MHC class I that constitutes a well-documented mechanism of immune evasion found in 40–90% of human tumors [21].
Extending our analysis to human cancers, we found that MBNL1 expression positively correlates with CD8A expression and cytolytic T cell activity, suggesting a conserved role for MBNL proteins in regulating T cell infiltration and function. Additionally, we present multiple lines of evidence supporting the previously described functional redundancy between MBNL1 and MBNL2 [16, 19]. Collectively, these findings position MBNL proteins as key regulators of the tumor-immune microenvironment.
While our study establishes a link between MBNL expression, antigen presentation, and T cell-mediated tumor killing, the precise molecular mechanisms remain unclear. MBNL proteins are well-characterized modulators of RNA metabolism, influencing mRNA splicing, localization, and stability [5–7, 19]. In several solid tumors, reduced MBNL1 expression leads to exon skipping in MAP2K7, promoting stem cell-like phenotypes and correlating with poor prognosis [22]. Conversely, MBNL1 overexpression in MLL-rearranged leukemias drives widespread intron removal, altering the expression of key epigenetic modifiers involved in leukemogenesis [11]. It is plausible that MBNL proteins regulate the RNA processing of MHC class I components or transcription factors that govern IFN responses. For instance, alternative splicing of STAT1 generates distinct isoforms that activate unique gene expression programs downstream of IFNγ signaling [23, 24]. Elucidating these mechanisms in future studies will be crucial for fully understanding MBNL’s role in anti-tumor immunity.
This emerging role of MBNL in tumor-immune interactions may present new therapeutic opportunities with broad clinical implications. MBNL proteins are known to drive the pathogenesis of myotonic dystrophy, where a trinucleotide repeat expansion sequesters MBNL, leading to widespread RNA splicing defects [17, 18,25–27]. Interestingly, two hallmark features of the disease—muscle wasting and cataract formation—have been linked to aberrant inflammatory activation [28, 29]. Combined with our findings, this suggests that MBNL1 could potentially regulate intrinsic immune responses in diverse cell types. Given that MBNL expression correlates with markers of T cell infiltration and activity in most human cancers, leveraging this relationship for therapeutic benefit could have far-reaching implications.
Materials and methods
Cell lines and culture
B16-F10 and E0771 cells were obtained from ATCC (CRL-6475 and CRL-3461) and MC38 cells were obtained from Sigma-Aldrich (SCC172). All cell lines were cultured per the manufacturer’s instructions. Respective cell lines were infected at 0.3 MOI with Cas9 lentivirus (Addgene 52962-LV) and selected with 2 ug/mL Blasticidin for 7 days. Polyclonal cells were then single cell sorted, and individual clones were assessed for on-target editing efficiency, all subsequent CRISPR assays were completed with the respective Cas9 expressing clone with the highest editing rate.
To generate Cas9 edited mutant lines, we followed methods as previously described [30]. In brief, Cas9 expressing cells were transduced with a lentivirus containing a single guide RNA targeting either the respective gene of interest (Mbnl1, Mbnl2, etc.) or a paired-guide RNA targeting the two genes of interest (Mbnl1 and Mbnl2) in the pLentiguide-Puro backbone (Addgene #52963). Following transduction, cells with stable integration were selected using 1 ug/mL Puromycin for 72 hours.
Animal use
All animal work and procedures were completed in accordance with the Guidelines for the Care and Use of Laboratory Animals and were approved by the Fred Hutchinson Cancer Center Institutional Animal Care and Use Committee (IACUC). Six week old male C57BL/6 mice were obtained from the Jackson Laboratory. In brief, animals for tumor studies were anesthetized using isoflurane gas until they were unresponsive for all initial tumor cell injections as approved by the Fred Hutchinson Cancer Center IACUC. Animals were then routinely monitored throughout experiments by both core staff supported by Comparative Medicine and co-authors. When tumors reached the end point (1.5 cm in any dimension), the animals were sacrificed in accordance with institutional IACUC approved methods using CO2 inhalation for 5 minutes followed by cervical dislocation to attempt to minimize any suffering.
In vivo validation studies
Cas9 expressing B16-F10 cells were grown per the manufacturer’s recommendations and then transduced with lentivirus containing indicated pgRNAs and then selected in 1ug/mL Puromycin for 72 hours. 5x106 cells were then injected subcutaneously into each flank of adult male C57BL/6 mice, and monitored using calipers. Animals were euthanized when a tumor reached 1.5 cm in any dimension. Tumor material was isolated either for an archived flash frozen sample, fixed in 10% formalin at room temperature for histology studies, or placed in TRIzol reagent (Thermo Fisher 15596018) for subsequent RNA isolation.
Immunohistochemistry
Tissues from tumors were processed, embedded, and stained through the Fred Hutch Experimental Histopathology core. Mouse Cd8a (CST Clone D4W27 1:200 dilution) staining was performed using rabbit monoclonal antibodies. Staining was performed with a BOND RX autostainer (Leica Biosystems) and images were acquired with an Aperio ImageScope at 40x magnification (Leica Biosystems). Image analysis was completed using HALO Image Analysis software.
Western blotting
Total protein lysates were isolated in 1x RIPA buffer and quantified with the Pierce 660 nm Protein Assay Reagent (Thermo Fisher Scientific 22660). Total protein lysates were separated electrophoretically then transferred to a nitrocellulose membrane using the NuPAGE system (Thermo Fisher Scientific). Each membrane was blocked for 1 hour at room temperature and then probed with primary antibody diluted in a blocking buffer overnight at 4 degrees Celsius. Mbnl1 (DSHB Clone MB1(4A8)), Mbnl2 (Santa Cruz Biotechnologies Clone 3B4), and Gapdh (Abcam ab9485) primary antibodies were used. Anti-mouse or anti-rabbit IRDye (LI-COR Biosciences) secondary antibodies and the Odyssey CLx Imager (LI-COR Biosciences) were utilized for detection and imaging.
In vitro flow cytometry and IFNγ stimulation
B16-F10 Cas9-expressing cells were then infected and grown for 48 hours. Cells were then plated and treated with standard DMEM with 10% FBS with indicated concentration of recombinant mouse IFNγ (Cell signaling technology 39127S). Following the indicated time course, all cells were washed 3x with PBS and then resuspended in 100 uL of PBS and stained with respective antibodies and LIVE/DEAD Fixable Violet (Thermo Fisher L34955) stain per the manufacturer’s instructions. Cells were then passed through a 40-micron filter to generate a single cell suspension and then run on a BD FACSCelesta. Single cells were gated for live cells and then the signal was measured in the appropriate channel. Data were analyzed in FlowJo v10.
In vivo immune cell and tumor cell flow cytometry
Tumors were dissected from mice and mechanically dissociated. Then 0.5 g of tissue were placed in 2-3 mL of tumor digestion buffer (20 mg/mL collagenase and 0.25 mg/mL DNAseI in 1x PBS) and shaken for 30 minutes at 37 degrees. The mixture was then subjected to centrifugation at 1200 rpm for 5 minutes, resuspended in 1x PBS, and then passed through a 70 micron filter (Corning 352250). Samples were then spun down and washed with 1x PBS and resuspended in 1x red blood cell lysis buffer (Thermofisher 00-4333-57). Cells were then washed with 1x PBS, these samples were then used for flow cytometry of whole tumor cells. Cells were then incubated in 1x FACS buffer (1 mM EDTA and 2% FBS in 1x PBS) and 1:100 anti-H 2Kb/H-2Db antibody (BioLegend 114606).
For immune cell populations, digested and RBC lysed samples were then incubated with mouse CD45 microbeads (Miltenyi 130-052-301) and CD45 + cells were purified using positive selection columns per manufacturer’s instructions (Miltenyi 130-042-401). Cells were then washed with PBS and then incubated with an antibody panel to identify T cell subsets and NK cells (CD90.2 BV786, CD8a eFluor 450, CD45 FITC, NK1.1 PE, CD4 BV510, anti-TCR-b chain Alexa Fluor 700) before proceeding with flow cytometry run on a BD FACSCelesta and analyzed in in FlowJo v10.
gDNA PCR and on-target editing verification
gDNA was extracted with the DNeasy Blood and Tissue kit (Qiagen) according to the manufacturer’s protocol. A window around the anticipated on-target cut site for the respective gene (Mbnl1, Mbnl2) was identified to allow for PCR amplification. PCR product was then purified and submitted for AmpliconEZ sequencing (Azenta/Genewiz). Reads were trimmed and mapped and the estimated fraction of reads with a specific indel were identified using the CRISPresso2 software using default settings [31].
RNA-seq library preparation
Cells were first pelleted and then RNA was isolated using the Qiagen ReasyA kit (Qiagen 74104) and RNA purity was confirmed via 4200 TapeStation System. Poly(A)-selected, unstranded Illumina libraries were prepared according to the TruSeq protocol per manufacturer’s instructions. The final library size and distribution was confirmed with a 4200 TapeStation System before sequencing on an Illumina HiSeq as 2x50 bp to generate ~ 40 million reads per sample.
RNA-seq data analysis
RNA-seq was analyzed as previously described [32]. RNA-seq reads were mapped to an annotated transcriptome that was created using Ensembl 71 [33], UCSC knownGene [34] and Misov2.0 [35] annotations using RSEM version 1.2.4 [36] (modified to call Bowtie [37] with option ‘-v 2’). Unaligned reads were mapped to the corresponding genome (hg19/GRCh37 assembly, mm10/GRCmc38 assembly) as well as a database containing all possible pairings of 5’ and 3′ splice sites per gene contained in out merged transcriptome annotation using TopHat version 20.8b [38]. Lastly, mapped reads were merged and input into MISO v2.0. For TCGA studies, we analyzed 9,045 available samples across 29 cancer types.
Cytolytic T cell correlations
RNA-seq data was analyzed as above and for each sample we quantified CD8A gene expression. For each individual cancer subtype we then correlated CD8A gene expression with expression of every coding gene in our lab annotation using Pearson correlation and generated a Pearson correlation coefficient. Then, for each gene we calculated the median Pearson correlation coefficient across all 29 analyzed cancer types.
CRISPR screen re-analysis
CRISPR screen data was downloaded from previously published complementary screens in B16-F10 murine melanoma cells under distinct conditions to identify resistance and sensitizing mutations to antigen specific cytotoxic T cell killing [15].
OT-1 cytotoxicity assay
Bulk splenocytes were isolated from OT-1 animals and cultured for three days in media containing 100 U/mL murine IL-2 and 100 μg/mL SIINFEKL peptide to activate CD8 + T cells. Cultures were then washed several times with media to remove ova peptide and rested for at least 24 hours prior to use. OT-1 cells were passaged in T cell media with 50 U/mL IL-2 for no more than seven days after animal sacrifice before use. For the cytotoxicity assay, tumor cells alone or tumor cells + OT-1 cells (1:1 ratio) were incubated in T cell media for 18 hours under standard conditions and then analyzed by flow cytometry to quantify killing. OT-1 cells and other hematopoietic cells were excluded using CD45, CD3, and CD8 staining. Tumor cell viability was measured with DAPI staining.
Supporting information
S1 Table. Differential induction of IFNγ induced gene expression in NTC versus DKO B16-F10 Cas9 cells.
Differentially expressed genes from B16-F10 Cas9 cells infected with the indicated pgRNAs exposed to no or 1 ng/mL IFNγ for 24 hrs. Median gene expression per condition expressed in transcripts per million (TPM) along with the computed log2(fold-change) between the indicated comparison as well as the adjusted P value.
https://doi.org/10.1371/journal.pone.0321148.s003
(XLSX)
S2 Table. Computed gene expression of MBNL proteins, CD8A, PRF1, and GZMB across TCGA RNA-seq samples.
Gene expression values (in transcripts per million, TPM) computed for all analyzed TCGA RNA-seq samples as well as the computed geometric mean of GZMB and PRF1 used as a proxy for cytolytic T cell activity.
https://doi.org/10.1371/journal.pone.0321148.s004
(XLSX)
S3 Table. Genome wide correlation of gene expression with CD8A expression across 29 cancer types.
Computed Pearson correlation coefficients and associated P values comparing gene expression of the indicated gene with CD8A expression in the indicated cancer subtype. Each subtype is analyzed independently and the number of samples analyzed is documented per cancer subtype.
https://doi.org/10.1371/journal.pone.0321148.s005
(XLSX)
Acknowledgments
The results in this publication are based in part on data from The Cancer Genome Atlas Research Network (http://cancergenome.nih.gov). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
References
- 1. Liu C, Yang M, Zhang D, Chen M, Zhu D. Clinical cancer immunotherapy: current progress and prospects. Front Immunol. 2022;13:961805. pmid:36304470
- 2. Brentjens R, Davila M, Riviere I, Park J, Wang X, Cowell L. CD19-targeted T cells rapidly induce molecular remissions in adults with chemotherapy-refractory acute lymphoblastic leukemia. Science Translational Medicine. 2013;5(177):177ra38.
- 3. Wolchok J, Chiarion-Sileni V, Gonzalez R, Rutkowski P, Grob J, Cowey C. Overall survival with combined nivolumab and ipilimumab in advanced melanoma. New England J Medicine. 2017;377(14):1345–56.
- 4. Bruni D, Angell HK, Galon J. The immune contexture and immunoscore in cancer prognosis and therapeutic efficacy. Nat Rev Cancer. 2020;20(11):662–80. pmid:32753728
- 5. Ho T, Charlet-B N, Poulos M, Singh G, Swanson M, Cooper T. Muscleblind proteins regulate alternative splicing. EMBO J. 2004;23(15):3103–12.
- 6. Batra R, Manchanda M, Swanson MS. Global insights into alternative polyadenylation regulation. RNA Biol. 2015;12(6):597–602. pmid:25892335
- 7. Wang ET, Cody NAL, Jog S, Biancolella M, Wang TT, Treacy DJ. Transcriptome-wide regulation of pre-mRNA splicing and mRNA localization by muscleblind proteins. Cell. 2012;150(4):710–24.
- 8. Lee JE, Cooper TA. Pathogenic mechanisms of myotonic dystrophy. Biochem Soc Trans. 2009;37(Pt 6):1281–6. pmid:19909263
- 9. Navvabi N, Kolikova P, Hosek P, Zitricky F, Navvabi A, Vycital O, et al. Altered Expression of MBNL family of alternative splicing factors in colorectal cancer. Cancer Genomics Proteomics. 2021;18(3):295–306.
- 10. Tabaglio T, Low DH, Teo WKL, Goy PA, Cywoniuk P, Wollmann H, et al. MBNL1 alternative splicing isoforms play opposing roles in cancer. Life Sci Alliance. 2018;1(5):e201800157. pmid:30456384
- 11. Itskovich SS, Gurunathan A, Clark J, Burwinkel M, Wunderlich M, Berger MR, et al. MBNL1 regulates essential alternative RNA splicing patterns in MLL-rearranged leukemia. Nat Commun. 2020;11(1):2369. pmid:32398749
- 12. Voss D, Sloan A, Spina R, Ames H, Bar E. The alternative splicing factor, MBNL1, inhibits glioblastoma tumor initiation and progression by reducing hypoxia-induced stemness. Cancer Research. 2020;80(21).
- 13. Fish L, Pencheva N, Goodarzi H, Tran H, Yoshida M, Tavazoie S. Muscleblind-like 1 suppresses breast cancer metastatic colonization and stabilizes metastasis suppressor transcripts. Genes & Development. 2016;30(4):386–98.
- 14. Tang L, Zhao P, Kong D. Muscleblind‑like 1 destabilizes Snail mRNA and suppresses the metastasis of colorectal cancer cells via the Snail/E‑cadherin axis. Int J Oncol. 2019;54(3):955–65. pmid:30664186
- 15. Pan D, Kobayashi A, Jiang P, Ferrari de Andrade L, Tay R, Luoma A. A major chromatin regulator determines resistance of tumor cells to T cell-mediated killing. Science. 2018;359(6377):770–5.
- 16. Nitschke L, Hu R, Miller A, Lucas L, Cooper T. Alternative splicing mediates the compensatory upregulation of MBNL2 upon MBNL1 loss-of-function. Nucleic Acids Research. 2023;51(3):1245–59.
- 17. Lin X, Miller JW, Mankodi A, Kanadia RN, Yuan Y, Moxley RT, et al. Failure of MBNL1-dependent post-natal splicing transitions in myotonic dystrophy. Hum Mol Genet. 2006;15(13):2087–97. pmid:16717059
- 18. Goodwin M, Mohan A, Batra R, Lee K-Y, Charizanis K, Fernández Gómez FJ, et al. MBNL Sequestration by Toxic RNAs and RNA Misprocessing in the Myotonic Dystrophy Brain. Cell Rep. 2015;12(7):1159–68. pmid:26257173
- 19. Konieczny P, Stepniak-Konieczna E, Sobczak K. MBNL proteins and their target RNAs, interaction and splicing regulation. Nucleic Acids Res. 2014;42(17):10873–87. pmid:25183524
- 20. Rooney MS, Shukla SA, Wu CJ, Getz G, Hacohen N. Molecular and genetic properties of tumors associated with local immune cytolytic activity. Cell. 2015;160(1–2):48–61. pmid:25594174
- 21. Cornel AM, Mimpen IL, Nierkens S. MHC Class I downregulation in cancer: underlying mechanisms and potential targets for cancer immunotherapy. Cancers (Basel). 2020;12(7):1760. pmid:32630675
- 22. Ray D, Yun YC, Idris M, Cheng S, Boot A, Iain TBH, et al. A tumor-associated splice-isoform of MAP2K7 drives dedifferentiation in MBNL1-low cancers via JNK activation. Proc Natl Acad Sci U S A. 2020;117(28):16391–400. pmid:32601196
- 23. Semper C, Leitner NR, Lassnig C, Parrini M, Mahlakõiv T, Rammerstorfer M, et al. STAT1β is not dominant negative and is capable of contributing to gamma interferon-dependent innate immunity. Mol Cell Biol. 2014;34(12):2235–48. pmid:24710278
- 24. Parrini M, Meissl K, Ola M, Lederer T, Puga A, Wienerroither S, et al. The C-terminal transactivation domain of STAT1 has a gene-specific role in transactivation and cofactor recruitment. Front Immunol. 2018;9:2879.
- 25. Miller JW, Urbinati CR, Teng-Umnuay P, Stenberg MG, Byrne BJ, Thornton CA, et al. Recruitment of human muscleblind proteins to (CUG)(n) expansions associated with myotonic dystrophy. EMBO J. 2000;19(17):4439–48. pmid:10970838
- 26. Mankodi A, Logigian E, Callahan L, McClain C, White R, Henderson D, et al. Myotonic dystrophy in transgenic mice expressing an expanded CUG repeat. Science. 2000;289(5485):1769–73. pmid:10976074
- 27. Mankodi A, Urbinati C, Yuan Q, Moxley R, Sansone V, Krym M, et al. Muscleblind localizes to nuclear foci of aberrant RNA in myotonic dystrophy types 1 and 2. Human Molecular Genetics. 2001;10(19):2165–70.
- 28. Rhodes JD, Lott MC, Russell SL, Moulton V, Sanderson J, Wormstone IM, et al. Activation of the innate immune response and interferon signalling in myotonic dystrophy type 1 and type 2 cataracts. Hum Mol Genet. 2012;21(4):852–62. pmid:22062891
- 29. Franck S, Couvreu De Deckersberg E, Bubenik J, Markouli C, Barbé L, Allemeersch J. Myotonic dystrophy type 1 embryonic stem cells show decreased myogenic potential, increased CpG methylation at the DMPK locus and RNA mis-splicing. Biol Open. 2022;11(1):bio058978.
- 30. Thomas JD, Polaski JT, Feng Q, De Neef EJ, Hoppe ER, McSharry MV, et al. RNA isoform screens uncover the essentiality and tumor-suppressor activity of ultraconserved poison exons. Nat Genet. 2020;52(1):84–94. pmid:31911676
- 31. Clement K, Rees H, Canver MC, Gehrke JM, Farouni R, Hsu JY, et al. CRISPResso2 provides accurate and rapid genome editing sequence analysis. Nat Biotechnol. 2019;37(3):224–6. pmid:30809026
- 32. Dvinge H, Ries RE, Ilagan JO, Stirewalt DL, Meshinchi S, Bradley RK. Sample processing obscures cancer-specific alterations in leukemic transcriptomes. Proc Natl Acad Sci U S A. 2014;111(47):16802–7. pmid:25385641
- 33. Flicek P, Ahmed I, Amode MR, Barrell D, Beal K, Brent S, et al. Ensembl 2013. Nucleic Acids Res. 2013;41(Database issue):D48-55. pmid:23203987
- 34. Meyer LR, Zweig AS, Hinrichs AS, Karolchik D, Kuhn RM, Wong M, et al. The UCSC Genome Browser database: extensions and updates 2013. Nucleic Acids Res. 2013;41(Database issue):D64-9. pmid:23155063
- 35. Katz Y, Wang ET, Airoldi EM, Burge CB. Analysis and design of RNA sequencing experiments for identifying isoform regulation. Nat Methods. 2010;7(12):1009–15. pmid:21057496
- 36. Li B, Dewey C. RSEM: accurate transcript quantification from RNA-Seq data with or without a reference genome. BMC Bioinformatics. 2011;12:323.
- 37. Langmead B, Trapnell C, Pop M, Salzberg SL. Ultrafast and memory-efficient alignment of short DNA sequences to the human genome. Genome Biol. 2009;10(3):R25. pmid:19261174
- 38. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctions with RNA-Seq. Bioinformatics. 2009;25(9):1105–11.