Abstract

Introduction

   H3K27-altered diffuse midline glioma (DMG) is a highly aggressive
   glioma subtype, accounting for approximately 60% of pediatric
   high-grade gliomas, with a median survival of less than 12 months. Due
   to its predominant localization in the brainstem, conventional surgical
   resection is often unfeasible, underscoring the urgent need for
   alternative therapeutic strategies. While previous studies on DMG have
   primarily focused on regulatory mechanisms at the protein level, the
   role of alternative splicing in DMG remains largely unexplored. Given
   its potential impact on gene regulation and tumor progression, a
   comprehensive analysis of alternative splicing could provide novel
   insights into targeted or immune therapeutic strategies, complementing
   existing transcriptomic studies of DMG.

Methods

   To investigate the alternative splicing landscape of DMG, we performed
   transcriptome sequencing (RNA-seq) on patient-derived H3WT and
   H3K27-altered DMG cell lines, integrating these data with RNA-seq and
   single-cell transcriptomic (scRNA-seq) datasets from published sources.
   This comprehensive approach enabled us to delineate the alternative
   splicing landscape of H3K27-altered DMG and validate its distinct
   features at the cellular level.

Results

   Our multi-omics analysis revealed significant transcriptional
   alterations in H3K27-altered DMG compared to H3WT DMG, particularly in
   pathways related to neuro-regulation, metabolism, and immunity. Further
   in-depth analysis identified extensive alternative splicing changes in
   H3K27-altered DMG, predominantly associated with RNA modifications and
   key alterations in extracellular matrix and nucleotide metabolism.
   Integrating these findings, we characterized five RNA-associated
   proteins that enabled a binary classification of DMG into neural and
   immune subtypes, with each subtype exhibiting distinct prognostic and
   transcriptomic features. Notably, we identified RALYL as a potential
   key regulator in DMG progression.

Discussion

   Our findings indicate that H3K27-altered DMG exhibits significant
   alternative splicing alterations, which play crucial roles in
   tumorigenesis and progression. Additionally, our study identified an
   RNA-binding protein-based classification of DMG and characterized RALYL
   as a potential regulatory factor, highlighting its potential as a novel
   therapeutic target.

   Keywords: DMG, H3K27-altered, multi-omics, alternative splicing,
   metabolism, immune, posttranscriptional modifications, tumorigenesis

1. Introduction

   Diffuse midline gliomas (DMG), particularly those with H3K27M
   alterations, are the most common high-grade gliomas in children
   ([41]1). With a median survival of less than one year, DMG
   predominantly arises in midline structures and exhibits extensive
   invasiveness, rendering conventional surgical approaches largely
   ineffective ([42]2–[43]4). Current treatment strategies primarily rely
   on radiotherapy and chemotherapy, while emerging therapies, including
   CAR-T ([44]5), tumor vaccines ([45]6), and small-molecule drugs
   ([46]7), are being actively explored. However, treatment responses vary
   significantly among patients. Recent multi-omics studies—including
   transcriptomics, epigenomics, proteomics, and metabolomics—have
   provided valuable insights into the internal heterogeneity and
   molecular characteristics of DMG, laying the foundation for potential
   targeted interventions ([47]8–[48]10).

   Post-transcriptional regulation plays a crucial role in modulating gene
   expression at the mRNA level, encompassing processes such as
   conventional modifications, translation, and alternative splicing (AS)
   ([49]11). In eukaryotes, mRNA undergoes intricate processing before
   translation, including the splicing, modification, or selective
   inclusion/exclusion of introns and exons—a process known as alternative
   splicing ([50]12). Alternative splicing events generally include
   retained intron (RI), alternative 5’ splice site (A5SS), alternative 3’
   splice site (A3SS), mutually exclusive exon (MXE), and skipped exon
   (SE) ([51]13). This mechanism generates transcript diversity,
   contributing to protein structure and function variations, which may
   underlie intratumoral heterogeneity in tumors. Numerous studies have
   demonstrated that alternative splicing within tumors can significantly
   influence tumorigenesis, tumor progression, and immune evasion
   ([52]14–[53]16), including gliomas ([54]17). However, as a distinct
   glioma subtype, DMG remains largely unexplored in this context,
   highlighting the need for further investigation.

   In this study, we conducted an integrated analysis of H3WT and
   H3K27-altered DMG cell line samples, incorporating transcriptome
   sequencing (RNA-seq) data from patient tissue samples and single-cell
   transcriptome sequencing (scRNA-seq) data. This comprehensive approach
   enabled us to characterize the alternative splicing landscape of
   H3K27-altered DMG and identify distinct alternative splicing
   signatures. These events were predominantly associated with metabolic
   regulation, immune regulation, and RNA modification. By integrating
   RNA-binding protein (RBP) gene sets, we further demonstrated that
   H3K27-altered DMG could be dichotomized based on differentially
   expressed RBPs, with the two subtypes primarily linked to
   neuro-regulation and immune regulation, each representing distinct
   prognostic features. Notably, we identified RALYL as a key gene
   significantly correlated with poor prognosis in DMG, potentially
   contributing to tumor stemness and proliferation. In conclusion, our
   study provides the first comprehensive characterization of the
   alternative splicing landscape in DMG, unveils the intratumoral
   heterogeneity driven by alternative splicing, and highlights RALYL as a
   promising therapeutic target for DMG through alternative splicing-based
   mechanisms.

2. Results

2.1. H3WT and H3K27-altered DMG showed significant transcriptional landscape
differences

   To investigate transcriptomic alterations in H3K27-altered DMG, we
   performed RNA-seq by comparing H3WT (SF188) and H3K27M DIPG cell lines
   (SU-DIPGXIII and SU-DIPGXVII) ([55]18) in three repeat samples ([56]
   Figure 1A ). Differential gene expression analysis was conducted on the
   raw data, and a volcano plot was generated to highlight key
   differentially expressed genes between SF188 and
   SU-DIPGXIII/SU-DIPGXVII ([57] Figure 1B and [58]Supplementary Data
   Sheet 1 ), with COL9A1 and S100B showing significant upregulation ([59]
   Supplementary Figure 1 ). A heatmap of these differentially expressed
   genes ([60] Figure 1C ) further demonstrated distinct gene expression
   patterns between H3WT and H3K27M DIPG cell lines. Subsequently, we
   selected genes that were consistently upregulated in H3K27M cell lines
   for Gene Ontology (GO) ([61] Figure 1D ) and KEGG pathway enrichment
   analysis ([62] Figure 1E ). The enrichment results revealed that
   H3K27-altered DMG tumor cells exhibited significant upregulation of
   neural-related pathways, including synapse organization, glutamatergic
   synapses, axon guidance, etc. Additionally, as a more stem-like and
   aggressive tumor type, H3K27M DIPG demonstrated enhanced mesenchyme
   development, extracellular matrix remodeling, and altered glucose and
   lipid metabolism, further supporting its highly invasive nature.

Figure 1.

   [63]Figure 1
   [64]Open in a new tab

   Differences in the transcriptional landscape between H3WT and
   H3K27-altered DMG. (A) The workflow of analysis of RNA-seq data. (B) A
   volcano plot of differentially expressed genes between SF188 and
   SU-DIPGXIII/SU-DIPGXVII and the five most significantly
   up-regulated/down-regulated genes are shown. (C) The heatmap of
   differentially expressed genes between SF188 and
   SU-DIPGXIII/SU-DIPGXVII. (D) GO enrichment results of upregulated
   differential genes shared between SU-DIPGXIII and SU-DIPGXVII. (E) KEGG
   enrichment results of upregulated differential genes shared between
   SU-DIPGXIII and SU-DIPGXVII.

   In conclusion, our preliminary RNA-seq analysis confirmed previously
   reported characteristics of H3K27-altered DMG, including pronounced
   neuro-synaptic features, strong tumor stemness, invasive properties,
   and enhanced glucose and lipid metabolism.

2.2. Differential alternative splicing exhibits a unique transcriptional
landscape in DMG

   We then utilized rMATS ([65]19) to perform alternative splicing
   analysis on RNA-seq data from H3WT and H3K27M DIPG cell lines ([66]
   Supplementary Data Sheets 2 and [67]3 ). Our results revealed many
   alternative splicing events between H3WT and H3K27M cells, with SE
   being the most prevalent event type ([68] Figure 2A ). Notably, the
   differential alternative splicing events between SU-DIPGXIII and
   SU-DIPGXVII remained broadly consistent ([69] Figure 2B ). To assess
   potential event preferences among these alternative splicing genes, we