Targeting ALDOA to modulate tumorigenesis and energy metabolism in retinoblastoma

Summary

This study aims to elucidate the pivotal role of aldolase A (ALDOA) in retinoblastoma (RB) and evaluate the potential of the ALDOA inhibitor itaconate in impeding RB progression. Utilizing single-cell RNA sequencing, ALDOA consistently exhibits overexpression across diverse cell types, particularly in cone precursor cells, retinoma-like cells, and retinoblastoma-like cells. This heightened expression is validated in RB tissues and cell lines. ALDOA knockdown significantly diminishes RB cell viability, impedes colony formation, and induces notable metabolic alterations. RNA-seq analysis identifies SUSD2, ARHGAP27, and CLK2 as downstream genes associated with ALDOA. The application of itaconate demonstrates efficacy in inhibiting RB cell proliferation, validated through in vitro and in vivo models. This study emphasizes ALDOA as a promising target for innovative RB therapies, with potential implications for altering tumor energy metabolism.

Graphical abstract

Highlights

• •ALDOA overexpression in various retinoblastoma cell types and tissues
• •ALDOA knockdown reduces retinoblastoma cell viability and alters metabolism
• •Itaconate inhibits retinoblastoma proliferation in both in vitro and in vivo models
• •Targeting ALDOA holds potential for advancing retinoblastoma therapies

Molecular biology; Cancer; Transcriptomics

Introduction

Retinoblastoma (RB) is the most prevalent ocular malignancy in infants and young children. It constitutes 2–4% of all pediatric tumors, wielding a profound impact on both the visual acuity and life expectancy of affected individuals.¹ The incidence of RB typically ranges from 1 in 20,000 to 1 in 15,000, yielding approximately 8,000–9,000 new cases among children annually.² Among the RB patients who experience metastasis, an alarming 80% eventually face a fatal outcome, with the invasive extraocular extension emerging as pivotal contributors to the metastatic process.³ Recent years have witnessed notable progress in the diagnosis and management of RB, leading to substantial improvements in both survival and rates of eye preservation. However, it is crucial to note that the advanced RB remain refractory to current therapeutic modalities, resulting in a less favorable prognosis.

The phenomenon, known as aerobic glycolysis or the Warburg effect, has evolved beyond a mere adaptation to the environment, now recognized as an integral component of malignant phenotypes and tumor metabolic reprogramming.⁴ Enhanced glycolysis in tumors not only fuels their proliferation but also provides intermediates for synthesizing critical biomolecules like nucleotides, lipids, and proteins.⁵ Consequently, targeting tumor glycolysis has arisen as a promising strategy in cancer therapy, as tumor cells adjust their energy metabolism to meet the demands of growth, biosynthesis, and redox reactions. Aldolase is a pivotal enzyme within the glycolysis pathway, facilitating the breakdown of 1,6-diphosphate-D-fructose into 3-phosphate-D-glyceraldehyde, and subsequently interacting with α-dihydroxyacetone phosphate.⁶ The aldolase family comprises three members: ALDOA, ALDOB, and ALDOC. Notably, ALDOA has been observed to be overexpressed in a variety of malignant tumors, including gastric cancer, hepatocellular carcinoma, colorectal cancer, and cervical cancer.^7,8,9,10

Our recent research into the metabolic microenvironment of RB has revealed an upregulation of the glycolytic pathway.¹¹ In this study, we aim to investigate the expression of key glycolytic enzymes in RB at the single-cell level. By integrating single-cell RNA sequencing (scRNA-seq) datasets from RB and normal retinal tissues, we compared gene expression profiles and confirmed the upregulation of ALDOA in RB. Subsequent in vitro and in vivo experiments demonstrated the effectiveness of targeted ALDOA inhibition in suppressing RB growth.

Results

Elevated ALDOA expression in retinoblastoma

GSEA of DEGs between retina and RB revealed significant enrichment in gene sets related to amide biosynthesis, glycolysis, catalytic activity/metabolism, and ATP synthesis (Figure 1A). This suggests a significant alteration in energy metabolism in RB, highlighting the potential crucial role of the glycolytic pathway in RB. Expression patterns of key genes within the glycolytic pathway was showed in heatmap (Figure 1B). Through dimensionality reduction and subsequent clustering analysis utilizing Uniform Manifold Approximation and Projection (UMAP), we visualized the expression profiles of Aldolase A (ALDOA) in RB and normal retina (Figure 2A). The analysis unveiled a high expression level of ALDOA in RB (Figure 2B), particularly elevated in extraocular RB compared to intraocular cases (Figure 2C). Branched expression analysis modeling unveiled a clear upregulation of ALDOA expression in retinoma-like cells as they progressed through pseudotime. However, this unique expression profile was not evident in other cell types (Figure S1). To validate ALDOA expression in RB, we examined its levels in human RB tissue samples and RB cell lines (WERI-RB1, Y79). The results demonstrated upregulated ALDOA mRNA and protein levels in RB cells compared to ARPE-19 (Figures 3A and 3B). Additionally, we analyzed ALDOA mRNA expression using the GSE125903 dataset¹² from the Gene Expression Omnibus (GEO) database, distinguishing between invasive and non-invasive RB cases. The analysis revealed higher ALDOA expression in invasive RB cases compared to non-invasive ones (Figure 3C). Proteins extracted from tissues of 5 intraocular and 5 extraocular RB patients revealed higher ALDOA protein expression in extraocular RB, as detected by western blotting (Figure 3D). IHC results further confirmed significantly higher ALDOA expression in extraocular RB compared to intraocular RB and normal retina (Figure 3E). Clinicopathological analysis revealed a trend toward higher ALDOA expression in cases with high-risk features like scleral and optic nerve invasion. Notably, ALDOA expression was significantly lower in younger patients (<2 years old) (Figure S2).

Figure 1.

Enrichment analysis and glycolytic pathway expression in retinoblastoma

(A) Gene set enrichment analysis (GSEA) of differentially expressed genes (DEGs) between normal retina and retinoblastoma (RB) samples.

(B) Heatmap representation of the expression patterns of key genes within the glycolytic pathway, highlighting their differential expression in RB compared to normal retina.

Figure 2.

Visualization of ALDOA expression profiles in retinoblastoma and normal retina

(A) Visualization of the expression profiles of Aldolase A (ALDOA) in retinoblastoma (RB) and normal retina using uniform manifold approximation and projection (UMAP).

(B) Comparison of ALDOA expression between RB and normal retina.

(C) Comparison of ALDOA expression between extraocular RB and intraocular case.

(D) ALDOA expression across various cell types.

Figure 3.

ALDOA overexpression in various retinoblastoma cell types and tissues

(A and B) Quantitative PCR (A) and western blot assays (B) revealed significantly higher messenger RNA (mRNA) and protein levels of ALDOA in retinoblastoma (RB) cells compared to normal cells.

(C) ALDOA mRNA expression levels in non-invasive and invasive retinoblastoma cases from the GSE125903 dataset.

(D) Western blot analysis of ALDOA protein expression in tissues obtained from 5 intraocular and 5 extraocular RB patients, demonstrating elevated ALDOA protein levels in extraocular RB.

(E) Immunohistochemistry (IHC) staining and a statistical analysis of ALDOA expression in normal retina, intraocular RB tissue, and extraocular RB tissue. R, retina; RB, retinoblastoma; S, sclera; VC, vitreous cavity; ON, optic nerve. Data are presented as means ± standard deviations. Statistical significance was determined using Student’s t test, ∗p < 0.05, ∗∗p < 0.001, ∗∗∗p < 0.001, ∗∗∗∗p < 0.0001.

Disruption of retinoblastoma cell growth and energy metabolism by ALDOA downregulation

Given the observed ALDOA overexpression in RB, we investigated the functional consequences of ALDOA downregulation in RB cell lines. The qRT-PCR and western blotting results showed a decrease in ALDOA mRNA and protein expression levels after transfection (Figures 4A and 4B). The CCK-8 assay indicated that upon ALDOA knockdown in Y79 and WERI-RB1 cells, noticeable alterations in cell viability were observed from the second day, with more pronounced differences becoming evident by the third day when compared to the si-NC groups (Figure 4C). Additionally, the EdU assay results showed a decrease in the proliferation ability of Y79 and WERI-RB1 cells following ALDOA knockdown (Figure S3). The colony formation assays revealed a significant decrease in RB cells colony formation abilities following the knockdown of ALDOA (Figure 4D). Lactate assay results demonstrated a marked reduction in extracellular lactate level in RB cells following ALDOA knockdown (Figure 4E), accompanied by elevated levels of reactive oxygen species and decreased ATP levels (Figures 4F and 4G).

Figure 4.

Functional implications of ALDOA downregulation in retinoblastoma cell lines

(A and B) Quantitative reverse transcription-polymerase chain reaction (qRT-PCR) (A) and western blotting analysis (B) revealed a substantial reduction in ALDOA mRNA and protein expression levels following transfection of RB cell lines.

(C) The CCK-8 assay demonstrated inhibited cell proliferation in response to ALDOA downregulation.

(D) Colony formation experiments indicated a reduced tumorigenic potential of RB cell lines after ALDOA knockdown.

(E) Extracellular lactate analysis exhibited a notable decrease in lactate levels in RB cells treated with siALDOA for 48 h compared to the negative control.

(F) Levels of reactive oxygen species (ROS) in RB cells treated with siALDOA for 48 h were increased compared to the negative control.

(G) Intracellular ATP levels in RB cells treated with siALDOA for 48 h were reduced in comparison to the negative control. Data are presented as means ± standard deviations. Statistical significance was determined using Student’s t test, with ∗p < 0.05, ∗∗p < 0.001, and ∗∗∗p < 0.001 denoting significance.

Signaling alteration in response to ALDOA downregulation in retinoblastoma

Next, we performed transcriptome sequencing on both the si-ALDOA and si-NC WERI-RB1 to explore the downstream alterations resulting from ALDOA downregulation. Our transcriptome analysis revealed significant changes in gene expression patterns, highlighting 162 upregulated genes including PHRF1, SUSD2, IGFN1, UNC79, and 136 downregulated genes including PTP4A3, ARHGAP27, MYT1, CLK2 (Figures 5A and 5B). Further pathway enrichment analysis of these differentially expressed genes (DEGs) indicated potential shifts in biological functions in RB following ALDOA downregulation. Significantly, the upregulated DEGs were notably enriched in pathways related to “Organic acid transport”, “Amino acid transport”, and “Carboxylic acid transport” (Figure 5C). Conversely, the downregulated DEGs were predominantly associated with pathways involving “DNA-binding transcription activator activity” and “histone methyltransferase activity” (Figure 5D). Gene set enrichment analysis (GSEA) further revealed that after ALDOA knockdown, upregulated DEGs were enriched in pathways such as glycosphingolipid biosynthesis, while downregulated DEGs were enriched in metabolic pathways like oxidative phosphorylation, glutathione metabolism, and fatty acid metabolism (Figure S4). To validate the expression levels of the identified DEGs, we conducted qRT-PCR on selected genes displaying significant expression differences. The results demonstrated a substantial upregulation of SUSD2, ARHGAP27, and CLK2 mRNA levels in RB cell lines compared to the ARPE-19 (Figure 5E). Furthermore, in RB cell lines where ALDOA was knocked down, these genes exhibited altered mRNA expression patterns, with increased expression of SUSD2 and decreased expression of ARHGAP27 and CLK2 (Figure 5F). Western blot results showed that after ALDOA knockdown, the protein level of SUSD2 increased while the levels of CLK2 and ARHGAP27 decreased in WERI-RB1 cells. However, in Y79 cell lines, an increase in SUSD2 protein level was not detected after ALDOA knockdown, and the trends for the other two proteins were also decreasing (Figure 5G). Furthermore, to individually knock down SUSD2, ARHGAP27, and CLK2, three pairs of siRNAs were designed for each gene (Figure 5H). CCK8 experiments revealed that knocking down ARHGAP27 led to a decrease in the proliferation capacity of RB cells (Figure 5I).

Figure 5.

Effects of ALDOA downregulation on WERI-RB1 cell line

(A) RNA-seq analysis was employed to examine alterations in the mRNA expression profile of WERI-RB1 cells following ALDOA downregulation.

(B) Heatmap analysis illustrating differentially expressed genes (DEGs) in WERI-RB1 cells subjected to ALDOA knockdown.

(C and D) Pathway enrichment analysis focused on the upregulated DEGs (C) and downregulated DEGs (D). Upregulated DEGs were significantly associated with pathways related to “Organic acid transport”, “Amino acid transport”, and “Carboxylic acid transport”, while downregulated DEGs were primarily linked to “DNA-binding transcription activator activity” and “Histone methyltransferase activity”.

(E) Quantitative PCR analysis of selected DEGs (SUSD2, ARHGAP27, and CLK2).

(F) Quantitative PCR analysis of SUSD2, ARHGAP27, and CLK2 in RB cell lines with ALDOA knockdown.

(G) Western blot analysis showing changes in SUSD2, ARHGAP27, and CLK2 protein levels following ALDOA knockdown.

(H) Western blot analysis of SUSD2, ARHGAP27, and CLK2 protein knockdown.

(I) Proliferation changes in WERI-RB1 cells following knockdown of SUSD2, ARHGAP27, and CLK2. Statistical significance was assessed using Student’s t test, with ∗p < 0.05 and ∗∗p < 0.001 indicating significance.

Itaconate suppresses retinoblastoma cell proliferation and inhibits WERI-RB1 xenograft growth

Itaconate is an enzymatic inhibitor of ALDOA, which suppresses the catalytic activity of ALDOA in the glycolytic pathway through covalent modification of Cys73 and Cys339, without altering the protein expression level of ALDOA.¹³ Hence, we utilized itaconate as a means to inhibit the enzymatic activity of ALDOA, while keeping its protein expression level unaltered (Figure 6A). Results from the CCK8 assay demonstrate a reduction in RB cell viability upon the addition of itaconate (Figure 6B). Moreover, the introduction of itaconate led to alterations in energy metabolism within RB cells. Mass spectrometry analysis revealed alterations in cellular metabolite levels upon treatment with itaconate in RB cell lines. Specifically, there was a downward trend in the levels of metabolites involved in glycolysis pathways such as 3-Phospho-D-glycerate, Beta-D-Fructose 6 phosphate, and D-Glucose-6-phosphate, although not statistically significant (Figure 6F). However, metabolites associated with oxidative phosphorylation pathways including L-Malate acid, Fumarate, and Oxaloacetate exhibited a significant decrease (Figure 6G). To further investigate, an RB murine model was established by injecting WERI-RB1 cells into the eyes of nude mice. 14 days post-grafting, the animals were randomized into groups and received intraocular administrations of either PBS, itaconate, or melphalan, with enucleation performed on day 40 (Figures 7A and 7B). Analysis of eye weight revealed that both the itaconate and melphalan treatment groups had lower eye weights compared to the control group, although these differences were not statistically significant (Figure 7C). External examination of the eyes and H&E staining showed significant corneal neovascularization and extraocular invasion in the PBS group, with extensive tumor cell infiltration in the vitreous body and anterior chamber. In contrast, the itaconate treatment group exhibited tumor cells primarily located anterior to the retina. The melphalan-treated mice had fewer intraocular tumor cells but showed signs of ophthalmatrophia and retinal structural abnormalities (Figure 7D). Histological analysis revealed ophthalmatrophia in 40% of the melphalan-treated eyes and 20% of the itaconate-treated eyes, while extraocular invasion was observed in 30% of the PBS group and 10% of the itaconate group (Figure 7E). TUNEL staining showed a higher incidence of TUNEL-positive cells in the tumors of the itaconate group compared to PBS controls (Figure 7F). Additionally, the itaconate group had a lower proportion of TUNEL-positive cells in the retina compared to the melphalan-treated group (Figure 7G). These observations indicate that itaconate is effective in targeting intraocular RB cells similarly to melphalan, while maintaining a more favorable safety profile for retinal tissue.

Figure 6.

Itaconate-mediated inhibition of ALDOA and its impact on retinoblastoma in vitro and in vivo

(A) Western blotting analysis Itaconate serves as an enzymatic inhibitor of ALDOA while preserving ALDOA’s protein expression level.

(B) CCK8 assay reveal a reduction in RB cell viability following the addition of itaconate.

(C, D, and E) Introduction of itaconate induces alterations in energy metabolism within RB cells, leading to reduced lactate production (C), increased generation of reactive oxygen species (ROS) (D), and decreased ATP generation (E).

(F) Metabolite changes in the glycolytic pathway in RB cells treated with itaconate.

(G) Metabolite changes in the oxidative phosphorylation pathway in RB cells treated with itaconate. Statistical significance was assessed using Student’s t test, with ∗p < 0.05 and ∗∗p < 0.001 indicating significance.

Figure 7.

Itaconate treatment in an RB murine model

(A) Study schematic illustrating the experimental design.

(B) Gross appearance of eyeballs 40 days post-xenograft.

(C) Eye weight analysis (n = 10).

(D) Representative views of enucleated eyes and H&E staining from each treatment group.

(E) Proportion of eyes with ophthalmatrophia and extraocular RB in the different treatment groups.

(F) Representative TUNEL images and quantitative analysis of intratumoral cells comparing the itaconate and melphalan groups.

(G) Representative TUNEL images and quantitative analysis of retinal cells comparing the itaconate and PBS groups. Data are presented as means ± SD. Statistical significance was determined by unpaired, two-tailed Student’s t test, with ∗p < 0.05 and ∗∗p < 0.01 indicating significance.

Discussion

In this study, we integrated the scRNA-seq datasets of normal retinal tissue and RB tissue, uncovering a significant upregulation of the glycolytic enzyme ALDOA in tumor tissue. Additionally, GSEA identified enrichment of gene sets containing ALDOA related to RB, particularly those associated with ATP generation, protein synthesis, and glycolysis. Next, we confirmed the elevated expression of ALDOA in both cell lines and patient tissues, and noted a significant increase in extraocular RB. Through the application of siRNA interference and the ALDOA inhibitor itaconate, we demonstrated that ALDOA inhibition alter tumor energy and metabolism and efficiently diminished the viability of RB cells and the tumor progression in the RB mouse model. Furthermore, by using transcriptome sequencing, we uncovered possible downstream targets under the control of ALDOA, such as SUSD2, ARHGAP27, and CLK2.

The mutation or loss of RB1 is a fundamental etiological factor in the development of RB. RB1 encodes the RB protein, which directs cellular metabolism reprogramming through E2F-induced transcriptional mechanisms, involving processes such as glucose oxidation phosphorylation, fatty acid oxidation, and amino acid synthesis.¹⁴ Nonetheless, the exact role of metabolic reprogramming in tumor progression and the specific metabolic enzymes implicated in the development of RB remain incompletely understood. ALDOA, a member of the aldolase enzyme family, has been the subject of extensive research, shedding light on its intricate interactions with hypoxia-inducible factor-1 (HIF-1) and its crucial role in modulating the glycolytic pathway and the process of epithelial-mesenchymal transition (EMT).¹⁵ It is worth noting that ALDOA expression levels have been thoroughly examined across various malignancies, revealing a direct link between ALDOA expression and the aggressiveness and invasiveness of tumors. Furthermore, comprehensive survival analyses consistently emphasize a strong positive correlation between increased ALDOA expression and an adverse prognosis.¹⁶ In line with previous findings, our research results validate a noticeable upregulation of ALDOA in RB tissue. Importantly, in cases of extraocular RB, ALDOA demonstrates higher expression levels compared to intraocular cases, highlighting the pivotal role of ALDOA in promoting the initiation and progression of RB. Recent research has increasingly focused on RB invasiveness, identifying targets such as UBE2C, SOX4, and the MCM family.^17,18,19 However, additional investigation is needed to clarify the specific role of ALDOA in the extraocular invasion of RB. Liquid biopsies, especially aqueous humor biopsies, show promise for predicting RB prognosis and assessing chemotherapy efficacy.^20,21,22,23 Investigations utilizing aqueous humor biopsy have shown potential in aiding RB prognosis and assessing chemotherapy efficacy.^24,25 Thus, further exploration of tumor-related glycolytic enzymes in aqueous humor or blood samples may provide additional support for the diagnostic significance of ALDOA in extraocular RB.

In the course of RB development, distinct cell types play varying roles. Our analysis of scRNA-seq dataset reveals a pronounced upregulation of ALDOA in specific cell types, particularly cone cell precursors, RB-like cells, and retinoma like cells. Cone precursor like cells represent an early-stage cell population with the inherent potential for differentiation into cones or malignant cells.²⁶ RB-like cells embody a highly proliferative malignant cell population characterized by a reduction in photoreceptor characteristics.²⁷ Retinoma like cells serve as transitional cell stages between premalignant cone precursors and fully developed tumor cells.²⁸ The specific high expression of ALDOA within these three cell types suggests its potential oncogenic role in RB. Apart from tumor cells, the tumor microenvironment comprises a complex network of cells, stroma, blood vessels, and immune cells surrounding the tumor.^29,30 Within the tumor microenvironment, the expression levels of ALDOA may be regulated, and ALDOA can influence the metabolism and survival status of tumor cells. Specifically, ALDOA may influence the function and status of other cells within the tumor microenvironment by affecting the energy metabolism and signaling pathways of tumor cells.^31,32 However, in this study, the impact of ALDOA on other cells within the RB microenvironment, such as immune cells, stromal cells, and endothelial cells, has not yet been investigated. Additionally, ALDOA may modulate factors such as oxygen concentration, pH balance, and nutrient supply within the tumor microenvironment, affecting the growth and chemoresistance of tumor cells.^33,34,35 Moreover, previous studies have implicated the POU2F1-ALDOA axis in promoting tumor cell proliferation and chemoresistance.³⁶ In this study, RNA sequencing revealed a significant alteration in POU2F1 expression following ALDOA knockdown, suggesting that the POU2F1-ALDOA axis may play a critical regulatory role in RB.

Itaconate has the capability to inhibit the catalytic activity of ALDOA in the glycolytic pathway without altering the expression level of the ALDOA protein. Initially discovered as a metabolite activated in macrophages, itaconate plays a critical role in the negative regulation of inflammatory responses by suppressing the production of inflammatory factors associated with macrophage activation.³⁷ This regulatory process primarily functions through the alkylation of KEAP1, leading to the activation of NRF2.³⁸ In the research of tumor, there has been a hypothesis that itaconate might indirectly influence tumor progression by its impact on macrophages.³⁹ However, it remains uncertain whether itaconate directly regulates tumor progression. In this study, we found that although the intravitreal injection of itaconate did not result in a reduction in the rate of intraocular tumorigenesis, there was a notable decrease in the size of intraocular tumors, as well as a reduction in structural damage to the eyeball following the treatment. These observations suggest the potential of itaconate as an adjunctive therapy for RB. One of the primary challenges in translating preclinical findings into clinical applications is navigating the complex regulatory landscape. For itaconate to be considered for clinical use in treating RB or other tumors, detailed pharmacokinetic studies must be conducted. Additionally, thorough investigation of potential side effects is necessary to ensure patient safety.

Genes such as PHRF1, SUSD2, IGFN1, and UNC79 were identified as the top upregulated genes in ALDOA knockdown samples. SUSD2, situated on chromosome 22, encodes a transmembrane protein that plays a crucial role in cell-cell interactions and cell-matrix adhesion. Studies have indicated that SUSD2 acts as a tumor suppressor in RB by regulating the occurrence, development, and angiogenesis of RB, making it a potential target for RB treatment.⁴⁰ Conversely, genes including PTP4A3, ARHGAP27, MYT1, and CLK2 were identified as the top downregulated genes in ALDOA knockdown samples. ARHGAP27 functions as a Rho GTPase-activating protein and is associated with pediatric chronic myeloid leukemia as well as ovarian cancer.^41,42 CLK2, a phosphorylation-involved splicing protein kinase, exerts a promotional role in various cancers.^43,44 These differentially expressed genes might play important roles in RB progression, part of which were validated in this study, while the intrinsic interaction network in RB still requires further exploration.

In conclusion, our study illuminates the important role of ALDOA in driving tumor growth in RB. These findings not only offer insights into the metabolic dynamics of RB but also present potential target candidates for the diagnosis and therapy of RB patients.

Limitations of the study

Our study has certain limitations that warrant discussion. First, the relatively small number of clinical samples used may constrain the analysis, potentially limiting the generalizability of our findings. Expanding the sample size and diversifying patient demographics would strengthen the robustness of our conclusions. Second, incorporating primary cells into future investigations could better capture the heterogeneity of RB, thus enhancing the validity of our results. Additionally, further exploration into patient outcomes and correlation with relevant clinical parameters would provide deeper insights into the clinical significance of our findings. Lastly, while our study successfully identifies downstream targets, a more comprehensive investigation is needed to elucidate the mechanistic intricacies of how ALDOA influences these targets and the broader signaling pathways involved.

Resource availability

Lead contact

Further information and requests for resources should be directed to and will be fulfilled by the lead contact, Rong Lu (lurong@gzzoc.com).

Materials availability

This study did not generate new unique reagents.

Data and code availability

• •The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive in National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA008175) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human. Original western blot images have been provided in the supplementary material.
• •This paper does not report original code.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

This work was supported by the National Natural Science Foundation of China (82203175), the Natural Science Foundation of Guangdong Province, China (2022A1515011609), and the Science and Technology Program of Guangzhou, China (2024A03J0266, 2024A03J0267).

Author contributions

Conceptualization, Y.W. and R.L.; Methodology, Y.W., J.T., Y.M., Z.Z., S.C., and Y.L.; Software, H.Z. and Y.L.; Investigation, J.L., Y.G., C.L., C.C., S.S., and R.L.; Sample collection: S.A., Y.M., P.Z., and R.L.; Visualization, J.T., H.Z., and Y.L.; Writing – original draft, J.T. and Y.L; Writing – review and editing, Y.W., X.W., J.L., and R.L.; Funding acquisition, Y.L. and R.L.; Supervision, R.L.

Declaration of interests

The authors declare no conflict of interest.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
Recombinant Anti-SUSD2 antibody	Abcam	Cat# ab182147, RRID: AB_3082988
Anti-Aldolase antibody	Abcam	Cat# ab252953, RRID: AB_3099487
CLK2 antibody	Proteintech	Cat# 11092-1-AP, RRID: AB_2082254
ARHGAP27 antibody	Proteintech	Cat# 22077-1-AP, RRID: AB_2878985
Beta Tubulin antibody	Proteintech	Cat# 66240-1-Ig, RRID: AB_2881629
Beta Actin Antibody	Proteintech	Cat# 66009-1-lg, RRID: AB_2782959
Chemicals, peptides, and recombinant proteins
RPMI 1640	Corning	#10-040-CV
DMEM/F-12	Gibco	#11320033
Fetal Bovine Serum	Gibco	#10099141C
Penicillin-streptomycin	Gibco	#15070063
HiPerFect Reagent	Qiagen	# 301704
Melphalan	Selleck	#S8266
Itaconic acid	Selleck	#S3095
Critical commercial assays
EZ-press RNA Purification Kit	EZBioscience	#B0004D
4× Reverse Transcription Master Mix	EZBioscience	#A001GQ
TUNEL Apoptosis Assay Kit	Beyotime	#C1088
Cell Counting Kit-8	Vazyme	#A311-01
BeyoClick™ EdU Cell Proliferation Kit with Alexa Fluor 488	Beyotime	# C0071S
Lactate Colorimetric Assay Kit	Abbkine	#KTB1100
ADP/ATP Ratio Assay Kit	Abnova	#KA1673
Deposited data
Raw data	Liu et al.18	GSE249995
Raw data	Wu et al.27	GSE168434
Raw data	Rajasekaran et al.12	GSE125903
Raw data	Lukowski et al.45	E-MTAB-7316
Raw data	This study	HRA008175
Experimental models: Cell lines
WERI-RB1	American Type Culture Collection	# HTB-169
Y79	American Type Culture Collection	# HTB-18
ARPE-19	American Type Culture Collection	# CRL-2302
Experimental models: Organisms/strains
Mouse: BALB/C nude	BesTest Bio-Tech	N/A
Oligonucleotides
Primers for Q-RT PCR, see Table S2	This paper	N/A
Software and algorithms
GraphPad Prism	GraphPad	V 8.0.2
R (v 4.0.3)	R	r-project.org
DESeq2	Love et al.46	https://bioconductor.org/packages/release/bioc/html/DESeq2.html
GSEA	UC San Diego and Broad Institute	http://gsea-msigdb.org

Experimental model and study participant details

Clinical samples

Experiments involving human retinoblastoma tissues were ethically approved by the Committee of Ethics at Zhongshan Ophthalmic Center in Guangzhou, China, ensuring compliance with the guidelines set forth in the Declaration of Helsinki (Ethical Approval Number: 2023KYPJ308). All patients included in the study provided informed consent. Samples from retinoblastoma and uveal melanoma patients were obtained from individuals undergoing surgery at the ocular tumor division of Zhongshan Ophthalmic Center. A section of uveal melanoma patient tissue was utilized as the control representing normal retina in the study. Extraocular RB refers to a tumor that extends beyond the eye, involving structures such as the orbit, optic nerve, meningeal spaces around the optic nerve, and full-thickness invasion of the sclera.⁴⁷ It may also invade adjacent adipose tissue, extraocular muscles, bones, conjunctiva, or eyelids. Comprehensive clinical data for the patients involved were documented in Table S1. All patients included in this study were Chinese, of Asian ethnicity, ensuring demographic homogeneity and relevance to the specific population under study.

scRNA-seq data processing and integration

We integrated scRNA-seq datasets from multiple sources for our analysis. Specifically, scRNA-seq data from 4 RB samples (2 intraocular and 2 extraocular RB patients) from our previous cohort were accessed via the GEO database under accession ID GSE249995.¹⁸ Additionally, scRNA-seq data from 7 RB samples (4 intraocular and 3 extraocular RB patients) were obtained from the GEO database with accession number GSE168434.²⁷ For comparative analysis, we included publicly available scRNA-seq datasets from 5 normal retina samples, which were sourced from ArrayExpress under the accession number E-MTAB-7316.⁴⁵ The procedures for unsupervised clustering analysis, dimensionality reduction, and cell type assignment were meticulously followed in accordance with established protocols.¹⁸ The SeuratR package (version v4.0.3) was employed for scRNA-seq data integration.⁴⁸ Differentially expressed genes were defined with a significance threshold of p < 0.05. Gene Set Enrichment Analysis (GSEA) was performed using the OmicStudio cloud platform (https://www.omicstudio.cn)⁴⁹ to understand the biological significance of the differentially expressed genes between RB and normal retina samples. The analysis utilized gene sets from the Molecular Signatures Database (MSigDB) and provided enrichment scores and p-values to determine the significance of each gene set. Graphical representations were generated to visualize the results, highlighting key pathways and biological processes altered in retinoblastoma. The Monocle3 software (version 1.2.9) was utilized to infer the pseudo-time trajectory.⁵⁰ DEGs between different states were determined using the fit_models function, employing a Q value threshold of <0.001. For cell type assignment, we gathered known gene markers from previous studies representing various cell types.^51,52,53 Using a semi-supervised machine learning method, we mapped the expression patterns of these markers across all cell clusters. Cells showing distinct expression levels of marker genes were assigned corresponding cell types, with manual adjustments made based on marker gene expression profiles. Visualization of the trajectory was conducted through UMAP graphs, while dynamic expression heatmaps were generated using the Heatmap function within the ComplexHeatmap package.

Cell lines and reagents

Two RB cell lines, WERI-RB1 and Y79, were employed, along with a non-tumor human retinal pigment epithelial cell line known as ARPE-19 (American Type Culture Collection, Manassas, VA, USA). Y79 and WERI-RB1 cells were cultured in RPMI 1640 medium (Corning, USA). ARPE-19 were cultured in Dulbecco’s Modified Eagle Medium/Nutrient Mixture F-12 (DMEM/F-12) (Gibco, Thermo Fisher Scientific, USA). These cultures were all supplemented with 10% fetal bovine serum (Gibco, Thermo Fisher Scientific, USA) and 1% penicillin-streptomycin (Invitrogen, USA). All cell cultures were maintained at 37°C humidified incubator with 5% CO₂. Itaconic acid was purchased from Selleck (S3095, Houston, TX, USA), and stock solutions were diluted in DMSO at an initial concentration of 1 mM. Melphalan was purchased from Selleck (S8266, Houston, TX, USA).

Tumor xenograft model

All animal experiments adhered to the ARVO Statement for the Use of Animals in Ophthalmic and Vision Research and were approved and supervised by the Institutional Animal Care and Use Committee at the Zhongshan Ophthalmic Center. Female BALB/C nude mice were procured from Zhuhai BesTest Bio-Tech Co (Zhuhai, China). At four weeks of age, the mice were randomly allocated into three groups: PBS (N = 10 eyes), Itaconate (N = 10 eyes), and melphalan (N = 10 eyes). Intravitreal injections of 1 × 10ˆ5 freshly prepared WERI-RB1 cells were administered using a 33g Hamilton needle, with neomycin ointment applied to the eyes afterward. Mice were monitored every three days for survival and ocular changes. On day 14, mice were examined under a microscope to confirm intraocular tumor presence prior to allocation. Each mouse was then randomly assigned an identification number to ensure unbiased group assignment before receiving intravitreal injections of PBS, 1 mM itaconate, or 1mM melphalan. On day 40, mice were euthanized, and their eyeballs were enucleated, fixed, and paraffin-embedded for subsequent H&E staining. TUNEL staining was performed using the One-Step TUNEL Apoptosis Assay Kit as per the manufacturer’s instructions. The percentage of apoptotic cells per tumor-bearing eye was calculated by averaging the number of TUNEL-positive nuclei divided by the total number of nuclei in the field.

Method details

Western blotting

Cells and RB tissues were collected and then lysed in RIPA buffer (K1020, APE×Bio, USA), supplemented with a protease inhibitor. The protein lysates obtained were assessed using a standard Western blotting protocol to detect the expression levels of ALDOA (1:1000, Abcam, ab252953), SUSD2 (1:1000, Abcam, ab182147), CLK2 (1:500, Proteintech, 11092-1-AP), ARHGAP27 (1:500, Proteintech, 22077-1-AP), β-actin (1:10000, Proteintech, 66009-1-lg) and β-tubulin (1:10000, Proteintech, 66240-1-Ig). Following this, the bound antibodies were detected by incubating the membranes with HRP-conjugated Affinipure Goat Anti-Rabbit IgG(H + L) (1:10000, ProteinTech, SA00001-2), and subsequently, the membranes underwent three rounds of 10-min washing in TBS-0.1% Tween 20. Finally, the bands were visualized using enhanced chemiluminescence (ECL; Tanon, China) and the Tanon 5200 MultiImage System (Tanon, China).

RNA extraction and quantitative real-time qPCR

RNA was extracted using the EZBioscience EZ-press RNA Purification Kit (B0004D, EZBioscience, USA). For cDNA synthesis, 1000 ng of total RNA was utilized along with the EZBioscience 4× Reverse Transcription Master Mix (A001GQ). The resulting reverse transcription product was then combined with other components in a 10 μL reaction for PCR amplification. GAPDH was chosen as the reference gene, and the relative expression levels of the target genes were calculated using the 2-ΔΔCT method. Each experiment was performed in triplicate. Primer sequences for each gene were provided in Table S2.

siRNA transfection of cell line

The small interfering RNA (siRNA) target sequences were synthesized by Shanghai GimaGen Corporation. For transfection, Y79 and WERI-RB1 cells were cultured in 1640 culture medium. On the day of transfection, 100 μL of the same medium was added to each well of a 24-well plate containing 2 × 10ˆ5 cells. 750 ng of siRNA, resulting in a final concentration of 100 nM, was diluted in 100 μL of serum-free 1640 culture medium, and 6 μL of HiPerFect Reagent (301704, Qiagen, Netherlands) was added. After incubation at room temperature for 5–10 min, the transfection complex was added to the cells, followed by co-culture for 6 h. Subsequently, 400 μL of complete 1640 culture medium was added, and cells were cultured for 24–48 h before collection for further experiments. Sequences of siRNA were provided in Table S3.

Cell viability assay

Cell viability of Y79 and WERI-RB1 cells was assessed using the Cell Counting Kit-8 (CCK-8) assay (A311-01, Vazyme, China). In brief, cells were seeded in 96-well plates at a density of 5000 cells per well and subsequently treated with siRNA and itaconic acid. Next, 10 μL of the CCK-8 reagent was introduced into each well, and the plates were incubated for an additional 4 h. The absorbance of the samples was then measured at a wavelength of 450 nm. Cell viability inhibition was calculated using GraphPad Prism software (version 8.0; GraphPad Software, La Jolla, CA, USA), and all experiments were performed in triplicate.

EdU assay

Cells were cultured in 6-well plates and allowed to recover overnight before treatment. A 2X EdU working solution was prepared by diluting EdU (10 mM) 1:500 in cell culture medium to achieve a final concentration of 20 μM. This was added in equal volume to the wells to obtain a final EdU concentration of 10 μM. Cells were incubated for 2 h, then fixed with 1 mL of 4% paraformaldehyde for 15 min at room temperature. Cells were washed three times with 1 mL of wash solution (3–5 min each), permeabilized with 1 mL of PBS containing 0.3% Triton X-100 for 10–15 min, and washed again. The Click Additive Solution was prepared as per the manufacturer’s instructions, and the Click reaction mixture was added (0.5 mL per well) and incubated at room temperature in the dark for 30 min. Cells were washed three more times, and nuclear staining was performed. Fluorescence microscopy was used to observe the cells, with Azide 488 having a maximum excitation wavelength of 495 nm and an emission wavelength of 519 nm. Three biological replicates were conducted.

Cell colony formation assay

Following transfection, Y79 and WERI-RB1 cells were plated onto poly-D-lysine-coated coverslips within 6-well plates, with a cell density of 1×10⁵ cells per well. This culture was maintained for a 24-h period, with regular medium replacements every 3 days. Subsequently, after a 7-day incubation, the cells underwent staining with 0.1% crystal violet dye for a duration of 1 h. Colonies were identified as cell clusters consisting of more than 50 cells and were quantified. Each experiment was repeated three times.

Measurement of extracellular lactate level

The extracellular lactate content was quantified using the Lactate Colorimetric Assay Kit (KTB1100, Abbkine, China). Following cell lysis with lactate extract buffer, the intracellular lactate levels were measured in accordance with the manufacturer’s instructions. Specifically, a total of 5 × 10⁶ cultured cells were washed with PBS and the resulting supernatant (0.8 mL) was collected in a sterile tube post-centrifugation. Standard curves were constructed per the manufacturer’s protocol, and the samples were incubated with the reaction mixture for 30 min. After centrifugation to remove insoluble materials, the supernatants were dissolved in ethanol and the absorbance was measured at a wavelength of 570 nm. Calculation of lactate levels was based on the standard curve.

ADP/ATP ratio assay

The ADP/ATP ratio was determined using the ADP/ATP Ratio Assay Kit (KA1673, Abnova) following the manufacturer’s instructions. Briefly, 1 × 10⁴ treated cells were plated in white opaque 96-well plates. 90 μL of ATP reagent was added to each well, and luminescence (RLU A) was measured after a 1-min incubation using a luminometer. Ten minutes later, luminescence (RLU B) was measured again. Subsequently, 5 μL of ADP reagent was added to each well, and luminescence (RLU C) was quantified after another 1-min incubation. The ADP/ATP ratio was calculated using the formula: ADP/ATP ratio = ((RLU C) - (RLU B))/(RLU A). All experiments were performed in triplicate to ensure accuracy.

LC‒MS analysis

Liquid chromatography‒mass spectrometry (LC‒MS) (Novogene, China) was utilized for mass spectrometry analysis. Cell samples were processed, and supernatants were collected after centrifugation. Following drying and reconstitution, samples were analyzed on a Waters UPLC BEH Amide column with ESI source conditions set at 600°C. Multiple reaction monitoring (MRM) was employed for scanning, and data processing was performed using SCIEX OS software. Metabolite structures were identified through mass spectra analysis and database retrieval, followed by standardization and statistical analysis for data interpretation.

RNA sequencing

Total RNA was recovered from cell cultures using Trizol Reagent (Thermo Fisher Scientific-Invitrogen, Carlsbad, CA), and frozen at −80°C prior to use for RNA sequencing. Sequencing libraries were generated using NEBNext Ultra RNA Library Prep Kit for Illumina (NEB, USA, Catalog #: E7530L) following manufacturer’s recommendations and index codes were added to attribute sequences to each sample. Library quality was assessed on the Agilent 5400 system (Agilent, USA) and quantified by QPCR (1.5 nM). The qualified libraries were pooled and sequenced on the Illumina NovaSeq 6000 using 2 × 150 paired-end reads. Fastp (v0.23.4) was used to trim adapter sequences and low-quality bases in raw FASTQ files.⁵⁴ The survived reads were mapped onto GRCh38. p13 through STAR (v2.7.11b) and then counted based on GENCODE v39 GTF.^55,56 Finally, RSEM (v1.3.3) was utilized to estimate gene expression levels from alignment files.⁵⁷ The RSEM outputs summarized expected count, transcripts per million (TPM), and fragments per kilobase per million mapped fragments (FPKM). For downstream analysis, genes with zero TPM value were excluded. Identification of differentially expressed genes was performed using R package DESeq2 (v1.39.8).⁴⁶ The identified DEGs were sorted based on their P-values for further analysis. Annotation, visualization, and integrated discovery were performed using the DAVID.⁵⁸ This analytical platform was utilized for Gene Ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis, enabling a comprehensive exploration of the molecular functions of the DEGs. GO analysis was conducted within the biological process (BP) category, with statistical significance defined at p < 0.05. Additionally, Gene Set Enrichment Analysis (GSEA) was performed to further elucidate the functional significance of the DEGs.⁵⁹ The raw sequence data reported in this paper have been deposited in the Genome Sequence Archive⁶⁰ in National Genomics Data Center,⁶¹ China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences (GSA-Human: HRA008175) that are publicly accessible at https://ngdc.cncb.ac.cn/gsa-human.

Histological and immunohistochemistry

Four-micrometer-thick paraffin sections obtained from both human and mouse tissues underwent standard Hematoxylin and Eosin (H&E) staining and immunohistochemistry (IHC) procedures following optimized protocols. Tissue specimens for IHC analysis were collected from a cohort comprising 15 extraocular retinoblastoma (RB) patients and 15 intraocular RB patients who had received treatment at the Zhongshan Ophthalmic Center. The assessment of IHC staining in RB samples was conducted by two authors (J.T. and Y.W.), under the supervision of pathologists (P.Z. and S.C.). ALDOA expression levels were evaluated using a 1:100 dilution of anti-ALDOA antibody (Abcam, ab252953), and quantification was performed by determining the proportion of positively stained tumor cells.

Quantification and statistical analysis

The data are presented as means ± standard deviation (SD). Data analysis and presentation were conducted using R software (R Foundation for Statistical Computing, Vienna, Austria, http://www.r-project.org) and GraphPad Prism Software v 8.0.2 (GraphPad, Inc., La Jolla, CA, USA). Statistical analyses were performed utilizing unpaired two-tailed Student’s t test, one-way ANOVA, or Fisher’s exact test as appropriate. Significance levels are indicated in the respective figures.

Published: August 15, 2024