Utilizing genomics to identify novel immunotherapeutic targets in multiple myeloma high-risk subgroups

Abstract

Background: Immunotherapy is now standard of care for multiple myeloma (MM), where the most common targets are B cell maturation antigen, CD38, and G protein-coupled receptor class C group 5 member D (GPRC5D). However, additional novel targets are needed to counter tumor heterogeneity, therefore new strategies to identify additional targets are also required.

Methods: We utilized multi-omics data from two large datasets A framework that utilized prior knowledge of cell surface potential, expression in healthy organs, and expression level in MM cells was established to define novel immunotherapeutic targets. High confidence targets were prioritized for myeloma populations and subgroups, validated with flow cytometry and immunoblotting.

Results: Novel population-level candidate targets such as ITGA4 and LAX1, as well as subtype-specific targets including ROBO3 in t(4;14), CD109 in t(14;16), CD20 in t(11;14), CD180 in hyperdiploidy, GPRC5D in 1q gain, and ADAM28 in biallelic TP53 samples were identified. Candidate target surface expression was validated by flow cytometry and CRISPR-Cas9 knock-out models. Sub-clonal differences in expression were noted, using single-cell RNA-seq data. Additionally, alternative splicing of existing immunotherapy targets, such as FCRL5, was noted as a potential mechanism of antigen loss.

Conclusions: Our study presents a methodology to identify novel candidate immunotherapy targets. We also use known genomic data to identify subtype-specific targets that could be used either as complementary or alternative targets to existing treatments. We show that immunotherapy targets can have heterogenous expression within a patient, which can affect treatment efficacy. Taken together, our study establishes a robust methodology to identify novel therapeutic targets in MM, revealing critical insights that will inform the development of current and next-generation immunotherapies.

Fig. 1

Characteristics of candidate targets identified in ND and RR populations from two independent datasets. A General workflow of the target identification process. B A heatmap demonstrating all identified candidate genes in ND and RR population from MMRF and IU datasets with various annotations. C Expression level of selected genes. D Ranked expression of 5,092 proteins documented in Anderson et al. [36]. Numbers after gene names: rank. E A radar plot summarizing key characteristics among LAX1, ITGA4, and TNFRSF17/BCMA. Range (from center to edge): toxicity (healthy organs): 2~0; toxicity (blood cells): 0~1845; protein exp: 24.8~37221.6; essentiality: 0~ −1.75; hazard ratio (PFS): 1~1.28; mRNA exp: 3~7.6. Range in toxicity, protein expression, essentiality, hazard ratio, and mRNA exp indicated lowest to highest among 98 candidate genes. F Log₂-scaled median fluorescence intensity (MFI) of TNFRSF17/BCMA and ITGA4/CD49d detected by flow cytometry in 15 MM cell lines. G Density plots indicating MFI (blue peaks) of ITGA4/CD49d compared to the isotype control (grey peaks) across 6 MM cell lines. Log₂MFI: Log₂-scaled MFI. Highlighted genes in B: well-established targets or novel targets found in this study and validated by flow cytometry

Fig. 2

Characteristics of genes identified from primary subtypes. A UMAP plot indicating the expression heterogeneity of 98 population-based candidate targets among five primary subtypes. B A heatmap demonstrating all identified candidate genes from primary subtypes. C A Venn diagram indicating the common/unique candidate targets among primary subtypes. D Expression level of selected candidate targets uniquely/highly expressed in certain subtypes. E Protein expression of CD109 was detected in RPMI-8226 (t(14;16)) but not in KMS27 (t(11;14)) cell line. F Protein expression of ROBO3 and TNFRSF17/BCMA in t(4;14) (orange), t(14;16) (green), and t(11;14) (blue) cell lines. Statistical tests in D: Mann-Whitney U test; Significance level: *p < 0.05; **p < 0.01, ***p < 0.001

Fig. 3

Characteristics of candidate targets identified from high-risk subtypes. A Identified candidate targets for high-risk subtypes. B Four candidate genes demonstrating significantly elevated expression levels along with 1q copy number gain in the MMRF dataset. C A network plot indicating potential regulators of SELPLG, SPN, and GPRC5D found on 1q. D Three candidate genes demonstrating heterogenous expression among biallelic (-/-), monoallelic (+/-) and wild type (WT) TP53 subgroups in the MMRF dataset. E Three candidate genes exhibiting heterogenous expression in TP53 knock-out and WT triplicates of the AMO1 cell line. Protein expression of ADAM28 detected in TP53 wild-type (F) and knockout (G) AMO1 cell line measured by flow cytometry. Statistical test in boxplots: Mann-Whitney U test in B and D, T-test in E. Significance level: *p < 0.05, **p < 0.01, ***p < 0.001. PR/MF/MS: three of the seven expression subgroups defined as high-risk [26]

Fig. 4

Candidate targets were heterogeneously expressed among subclones identified from scRNA-seq data. A Transcriptomic landscapes of 49 IU patient samples with single-cell RNA-seq data. B Expression of ITGA4 among 49 samples. C Expression of CD79A is increased in t(11;14) and hyperdiploid 11q+ samples. D Fourteen subclones identified by scRNA-seq in a relapsed t(4;14) sample. E Uneven expression of CD38 among 14 subclones. F Expression of CD38 among 14 subclones. Statistical test: Mann Whitney U test in (F): Log2(CPM+1). Significance level: *p < 0.05, **p < 0.01, ***p < 0.001

Fig. 5

Impact of aberrant splicing towards candidate target gene expression. Summary of most significant alternative splicing events identified from MMRF (A) and IU (B) cohort, respectively. C Sashimi plots indicate inclusion levels of a cryptic exon in FCRL5 in different samples. Highlighted star: in-frame stop codon. −201, −202, −203, −206: transcript variants of FCRL5 defined in Ensembl. D A schematic plot indicating the truncated domains of a FCRL5-206 encoded protein. E The cryptic exon validated by RT-PCR and Sanger sequencing

Fig. 6

Overview of findings