A Novel MALT1-related Immune Prognostic Signature and Targeted Drug Screening for Glioblastoma

doi:10.21203/rs.3.rs-2866606/v1

Download PDF

Research Article

A Novel MALT1-related Immune Prognostic Signature and Targeted Drug Screening for Glioblastoma

https://doi.org/10.21203/rs.3.rs-2866606/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

As the most common intracranial malignancy in adults, glioblastoma (GBM) has limited improvement in prognosis with traditional treatment methods, such as surgery, chemotherapy, and radiotherapy. Currently, immunotherapy has revolutionized treatment outcomes for many cancers. However, special immunosuppressive microenvironment of GBM results in patients’ resistance to immunotherapy and poor prognosis. The mucosa-associated lymphoid tissue translocation gene 1 (MALT1) was reported to be involved in NF-κB activation and promote cancer cell’s proliferation and migration. And inhibition of MALT1 could attenuate the mesenchymal phenotype of GBM. Therefore, uncovering the role of MALT1 in the immunosuppressive microenvironment of GBM is of great importance. In this study, RNA-seq data of 169 GBM patients were downloaded from TCGA (The Cancer Genome Atlas) database and divided into MALT1_H (MALT1_High) and MALT1_L (MALT1_Low) groups based on MALT1’s expression level. First, the enrichment levels of the 29-immune signature were quantified in every GBM patient of MALT1_H and MALT1_L. Next, DEIGs (differentially expressed immune genes) were identified and used to establish an IPS (immune prognostic signature) by the LASSO (Least Absolute Shrinkage and Selection Operator) Cox regression analysis. PDYN was first found to be associated with GBM prognosis and was identified as a potential target of GBM. Based on three-gene IPS, we developed a predictive nomogram model to assess the prognosis of GBM patients. Additionally, MALT1 were proved to be a potential ideal therapeutic target for GBM. So, series of computer-aided technology were applied to screen favorable inhibitors of MALT1. In summary, we explored role of MALT1 in the suppressive immune microenvironment of GBM, established a novel MALT1-related nomogram model for prognostic prediction, and developed the targeted therapy for GBM in this study.

Glioblastoma (GBM)

Mucosa-associated lymphoid tissue translocation gene 1 (MALT1)

Immune prognostic signature (IPS)

Nomogram

Inhibitor

Glioblastoma (GBM) is the most common and aggressive primary brain tumor in adults. Traditional treatment strategies, such as surgery, chemotherapy, and radiotherapy, have limited clinical success and adverse effects on healthy tissue. Currently, immunotherapy has revolutionized treatment outcomes for many cancer patients and may provide new therapeutic approaches for the treatment of GBM. Therefore, detailed studies of the tumor immune microenvironment of GBM are required to discover novel therapeutic molecular targets and improve prognosis from an immunological perspective.

The human central nervous system (CNS) was once thought to be immune-privileged. Recently, studies have identified a functional network of lymphatic vessels lining the dural sinuses that flow into the deep cervical lymph nodes and serve as conduits for T cell transport between the periphery and the CNS cerebrospinal fluid (CSF) [1]. The brain undergoes constant immune surveillance and communication with the peripheral immune system, making immunotherapy possible as a means of treating diseases of the central nervous system.

Tumor immunotherapy aims to activate the human immune system, thereby relying on autoimmune function to overcome tumor immune resistance and kill cancer cells. A series of new and promising immunotherapeutic strategies have been applied to GBM, including vaccines, oncolytic viruses, immune checkpoint inhibitors, and adoptive cell transfer [2]. However, special immunosuppressive microenvironment of glioblastoma results in GBM patients’ resistance to immunotherapy and poor prognosis [3, 4].

The suppressive immune microenvironment of GBM is a major challenge for immunotherapy due to its low immunogenicity, suppressive immune cell infiltration, and lack of antigen-presenting potential and costimulatory antigens. In the GBM microenvironment, tumor cells secrete a large number of inhibitory factors that can inhibit effector T cells and promote tumor growth, such as ganglioside, kynurenine, transforming growth factor-β (TGF-β), vascular endothelial growth factor (VEGF), etc. In addition, recruited immunosuppressive cells such as Tregs and myeloid-derived suppressor cells (MDSCs) also secrete a large number of immunosuppressive factors, including TGF-β and IL-10, aggravating the immunosuppressive environment [5, 6].

Mucosal-associated lymphoma antigen 1 (MALT1) is a paracaspase that belongs to the caspase family of proteases with arginine-specific cysteine protease activity [7]. It mediates inhibitors of NF-κB signaling through proteolytic cleavages, such as TNFAIP3 (Tumor Necrosis Factor Alpha Induced Protein 3)/A20, and proteins that favor NF-κB activity, such as CYLD (Cylindromatosis) and RelB(Fig. 1) [8]. Furthermore, MALT1 acts as a scaffold to recruit additional signal transducers to activate NF-κB signaling [9]. NF-κB is a transcription factor belonging to the Rel family. After forming a dimer, it specifically binds to the promoters and enhancers of various genes and regulates the transcription of genes related to immunity, inflammation, cell proliferation and differentiation, and apoptosis. Activating NF-κB can promote tumor proliferation and invasion [22]. Moreover, activation levels of NF-κB in GBM signaling were found to be much higher than in non-GBM tissues, which is consistent with an increase in astrocytic tumor grade [23].

Recently, it has been reported that MALT1 is involved in EGFR-induced NF-κB activation and promotes EGFR-related tumor proliferation and migration [10]. The tumor-promoting role of MALT1 has been found in diffuse large B-cell lymphoma (DLBCL) and a subset of MALT lymphomas, suggesting that MALT1 is an attractive anticancer drug target [11] [12, 13]. In addition, studies have found that the mesenchymal phenotype of glioblastoma can be attenuated by direct inhibition of MALT1 [14]. At present, the main idea of GBM therapy targeting NF-κB is to prevent its ubiquitination degradation by inhibiting the phosphorylation of IκBα protein, thereby inhibiting the activity of NF-κB and binding to DNA. However, the effect of MALT1 on the GBM tumor microenvironment via NF-κB has not been investigated from an immunological perspective. Therefore, revealing the mechanism of action of MALT1 in the immunosuppressive microenvironment of GBM will be beneficial to provide prognostic prediction of tumor patients and discover new therapeutic targets.

In this study, RNA-seq data of 169 GBM patients were downloaded from TCGA (The Cancer Genome Atlas) database and divided into MALT1_H (MALT1_High) and MALT1_L (MALT1_Low) based on MALT1’s expression level. Firstly, the enrichment levels of the 29-immune signature were quantified in every GBM patient of MALT1_H and MALT1_L. Next, DEIGs (differentially expressed immune genes) were identified and used to establish an IPS (immune prognostic signature) by the LASSO (Least Absolute Shrinkage and Selection Operator) Cox regression analysis. Based on three-gene IPS, we developed a predictive nomogram model to assess the prognosis of GBM patients. Additionally, MALT1 was found to be a potential ideal therapeutic target for glioblastoma. So, a series of computer-aided technology, such as Discovery Studio 4.5, Schrodinger, and PyMol, were applied to screen favorable inhibitors of MALT1. In summary, we aimed to explore the role of MALT1 in the suppressive immune microenvironment of GBM, establish a prognostic prediction model, and develop targeted therapy drugs for GBM in this study.

Datasets of gene expression and processing of data

RNA-seq data of 169 GBM patients were downloaded from TCGA (The Cancer Genome Atlas, https://portal.gdc.cancer.gov/repository) database and divided into MALT1_H (n = 85) and MALT1_L (n = 84) based on MALT1’s expression level. The data included survival data and survival time over 30 days for GBM patients. RNA transcriptome profiling was calculated by FPKM values with log2-based transformation.

Immunogenomic analysis

First, we summarized the clinical features of the prognosis cohort associated with MALT1 expression in Supplementary Table 1. 29 sets of immune-associated genes were studied that represent different functions, pathways, and immune cell types (Supplementary Table 2). Then, the ssGSEA (single-sample gene-set enrichment analysis) score was calculated for assessing the enrichment levels of 29 immune signatures in each GBM patient dataset [15]. And ESTIMATE was applied to assess the stromal score (stromal content), tumor purity, and immune score (immune cell level) for every GBM patient [16]. Then, we compared the expression level of immune checkpoint genes and HLA genes between MALT1_H and MALT1_L by ANOVA test.

Survival analysis

OS (over survival) of GBM samples was compared between MALT1_H and MALT1_L. The log-rank test was used to analyze the significance of survival time differences with a threshold of P-value < 0.05. And the Kaplan–Meier curve was employed to directly present the differences in survival time.

Differential expression analysis

Differential expression analyses were carried out between MALT1_H and MALT1_L on RStudio by Wilcoxon Rank Sum and Signed Rank Tests [17]. Genes within the cut-off criterion of adjusted P values < 0.05 and |log2 fold change (FC)|>1 were chosen as DEIGs (Differentially expressed immune genes). And the ImmPort database (https://www.immport.org/) identified the DEIGs.

Analysis of functional enrichment

To calculate gene ontology and signal pathway enrichment of DEIGs, we uploaded DEIGs to Metascape (https://metascape.org/) to identify GO (Gene Ontology) terms and KEGG (Kyoto Encylopedia of Genes and Genomes) pathways. Metascape is an online website with gene function annotation analysis, visualization, and presentation of gene properties [18]. The GO analysis included classical pathway (Biological process), extracellular region (Cellular component), and antigen-binding (Molecular function). And KEGG analysis was carried out on human signaling pathways. The terms with P < 0.01 and the number of enriched genes ≥ 3 were considered significant and grouped separately according to their membership similarity. The term with the best p-value was selected from 20 clusters of no more than 15 terms each, for a total of 250 terms.

Construction and validation of the IPS (immune prognostic signature)

To construct the IPS (immune prognostic signature), we divided the TCGA dataset of GBM into training and validation sets in a random manner. The data from the training set was then subjected to LASSO Cox regression analysis using the "glmnet" R package to construct the IPS[19]. And IPS was established by weighting the Cox regression coefficients of every patient for calculating their risk scores [20]. Patients were divided into high-risk and low-risk groups according to the best cut-off values of risk score obtained by the “survminer” R package. Then we used the “survival ROC” R package to draw (ROC Receiver operating characteristic) curves to analyze the sensitivity and specificity. The AUC (Area under the curve) values were calculated to show the reliability. Furthermore, the same way was used to validate the IPS' prognostic prediction ability in the validation set.

Development of the nomogram

Multivariate and univariate Cox analyses were applied to assess the IPS' independent prognostic ability. Based on Cox analyses' results, an innovative nomogram was developed by the “rms” package. Furthermore, Calibration plots of observed vs. predicted probabilities of 1-, 3- years OS were performed to analyze the accuracy. The C-index (concordance index) was also calculated to evaluate the model's discrimination. And bootstraps got analyzed to correct the C-index.

Virtual screening using Libdock

We downloaded the structure of MALT1 with MLT-747 (ID:6F7I, 2.43 Å) from the Protein Database (PDB, https://www.rcsb.org/) [21]. Modifications of MALT1 were performed, including removal of water and other heteroatoms, the addition of hydrogen, protonation, ionization, and energy minimization. The chemical structure of MALT1 is shown in Fig. 5. The region where the MALT1 combined with MLT-747 was located and selected as the sphere binding region for virtual screening. And a total of 17799 natural, named, and purchasable molecules were downloaded from the ZINC15 database (https://zinc.docking.org/) to screen MALT1 inhibitors. MLT-747 is a well-studied known inhibitor and was chosen as the reference drug. The Libdock module of Discovery studio 4.5 (DS 4.5) is suitable for large-scale databases' rapid virtual screening based on rigid molecular docking. Therefore, it was applied for preliminary screening. Different postures of MLT-747 and ligands that docked with MALT1 were ranked according to the Libdock scores. Moreover, the compounds with the top 20 Libdock scores were selected for further pharmacological and toxicological properties analysis.

Pharmacological and toxicological properties analysis ADME and TOPKAT

ADME and TOPKAT modules of DS 4.5 were carried out for further pharmacological and toxicological analysis of the 20 selected molecules and MLT-747. The ADME (absorption, distribution, metabolism, and excretion) module was utilized to evaluate the water solubility, cytochrome P450 2D6 (CYP2D6) inhibition, plasma protein binding (PPB) level, human intestinal absorption, hepatotoxicity, and blood-brain barrier (BBB) permeability. TOPKAT module performed toxicity prediction analysis by constructing high-quality QSTR (Quantitative Structure Toxicity Relationship) models from the two-dimensional structural information of compounds. Toxicological properties analyses included Ames mutagenicity, developmental toxicity potential, and rodent carcinogenicity. Considering the above results, two compounds were selected as ideal inhibitors.

Binding interactions between selected compounds and MALT1

Precise docking between selected compounds and MALT1 was performed by the CDOCKER module of DS 4.5 based on the CHARMm36 force field. The ligand was flexible while the receptor remained rigid during the docking process. Each ligand generated ten docking poses to select an optimal pose based on a suitable docking orientation and high docking score [22, 23]. And the CDOCKER interaction energies, which reflect ligand binding affinity, were assessed separately for different poses of each tested molecule. Furthermore, the docking results of CDOCKER were visualized by Schrodinger and PyMol.

Pharmacophore analysis and Molecular dynamics simulation

The pharmacophores of selected molecules and MLT-747 were analyzed and displayed by the 3D-QSAR module of DS 4.5. Only those with energies below 10 kcal/mol were retained, with a maximum of 255 produced per molecule.

The optimal binding conformation of the ligand-MALT1 complex was selected for molecular dynamics simulations to evaluate the stability of the complex in the natural environment. First, the ligand-MALT1 complex was solvated through an explicit periodic boundary solvation water model. Then, solid chloride with an ionic strength of 0.145 was added to simulate the physical environment. The production module for molecular dynamics simulations was run with a time step of 1 fs and a simulation time of 100 ps with the temperature maintained at 300 K. Based on the initial reference settings, the trajectories of potential energies and RMSD were assessed through the analytical trajectory protocol of DS 4.5, and the corresponding images were plotted.

Immunogenomic analyses between MALT1_H and MALT1_L

As shown in Fig. 2A, functions, pathways, and immune cell types were more significantly enriched in MALT1_H than in MALT1_L. Additionally, the result of ESTIMATE indicated that the ESTIMATE score (Kruskal-Wallis test, P < 0.001), stromal scores (Kruskal-Wallis test, P < 0.001), and immune scores (Kruskal-Wallis test, P < 0.001) were higher in MALT1_H than in MALT1_L. Furthermore, the MALT1_L’s tumor purity was higher than MALT1_H (Fig. 2D). So, MALT1_H may consist of more stromal cells and immune cells but fewer tumor cells. Next, the expression level of HLA and immune checkpoint genes were analyzed. As the result showed, the expression level of HLA genes was higher in MALT1_H than in MALT1_L, such as HLA-DRA, HLA-DQP2, HLA-DRB6, HLA-DOA, HLA-DMB, HLA-DQA1, HLA-DQA2, HLA-DPB1, HLA-DRB1, HLA-DMA (ANOVA test, P < 0.01) and HLA-DPA1(ANOVA test, P < 0.05) (Fig. 2B). Moreover, most immune checkpoint genes had significantly higher expression levels in MALT1_H than in MALT1_L, such as CD274 (ANOVA test, P < 0.05), BTLA (ANOVA test, P < 0.05), and ICOS (ANOVA test, P < 0.01) (Fig. 2E). This result suggests that GBM patients in MALT1_H are immunocompromised and may be more likely to benefit from immune checkpoint inhibitors. Moreover, survival analyses showed that GBM patients in MALT1_L had a better survival prognosis than in MALT1_H (Fig. 2C).

Identification of DEIGs (differentially expressed immune genes) and functional enrichment

A total of 201 genes were identified as DEIGs (differentially expressed immune genes) (FDR < 0.05, criterion of |log2-fold change (FC)| ≥ 1), with 81 up-regulated genes and 120 down-regulated genes (Fig. 3A). Then, the DEIGs were uploaded to the Metascape website to identify KEGG pathways and GO terms. The DEIGs had higher enrichment in leukocyte migration, extracellular matrix organization, Cytokine-cytokine receptor interaction, collagen metabolic process, regulation of cell adhesion, regulation of vascular endothelial growth factor production, regulation of interleukin-1-mediated signaling pathway, negative regulation of cell motility, negative regulation of antigen processing and presentation (Fig. 3B). Each node was regarded as an enriched term. It was colored by its cluster-ID in Fig. 3C and colored by its P-value in Fig. 3D.

Construction and validation of the MALT1-related IPS (immune prognostic signature)

As shown in Fig. 4A-C, the LASSO Cox regression analysis was carried out to establish an IPS (immune prognostic signature) for the training set. And the risk scores of every patient were calculated (risk score = AREG*0.077 + PODNL1*0.317 + PDYN *0.296). We divided the training set’s patients into low-risk and high-risk groups according to the best cut-off value (0.7179717) of risk score evaluated by the “survminer” R package. The Kaplan-Meier analysis showed that the prognoses of patients in the high-risk group were worse than those in the low-risk group (Fig. 4D). And the AUC was 0.727 in the ROC (Receiver Operating Characteristic) curve analysis of the IPS, which indicated that the IPS had a promising ability to predict the OS of GBM patients (Fig. 4E). Risk score distribution and gene expression patterns in the training set are shown in Fig. 4F.

To validate the predictive ability of IPS on patient prognosis, we used the same formula to apply to the validation set. As shown in Fig. 4G, the result of the validation set was similar to the train set; patients in the high-risk group were more likely to have worse OS. Figure 4I showed gene expression patterns and risk score distribution in the validation set. The IPS was proved to be highly sensitive and specific in the ROC analysis (Fig. 4H). And AUC value was 0.652, demonstrating that the IPS had a high predictive value.

Establishment of a nomogram model based on IPS

The univariate Cox analysis showed that IPS had a significant association with OS (Hazard ratio: 2.064, 95% confidence interval: 1.393–3.060, P < 0.001) (Fig. 5A). In the multivariate Cox analysis, IPS was proved to be an independent prognostic factor (Hazard ratio: 2.044, 95% confidence interval: 1.377–3.035, P < 0.001). Ultimately, we constructed a nomogram model based on the IPS (Fig. 5B). The C-index (0.66) showed that the nomogram model could provide specific discrimination. Moreover, the observed and predicted 1-year and 3-year OS probabilities calibration plots showed good agreement (Fig. 5C).

Virtual screening of MALT1’s inhibitors using Libdock, ADME, TOPKAT

A total of 17799 natural, named, and purchasable molecules were downloaded from the ZINC 15 database for screening. According to the docking results of Libdock, a total of 6499 small molecules could dock with the MALT1(Fig. 6). Among them, 245 molecules had higher Libdock scores than MLT-747 (Libdock score: 119.848). The top 20 compounds by Libdock score were listed in Table 1 and chosen for further pharmacological and toxicological properties analysis by ADME and TOPKAT modules of DS 4.5. Finally, ZINC000019340795 (compound 1) and ZINC000004649679 (compound 1) were selected as ideal inhibitors of MALT1 due to non-CYP2D6 inhibition, no hepatotoxicity, low oncogenicity, and low mutagenicity in rodents (Table 2 and Table 3).

Table 1

Top 20 ranked compounds with higher libdock scores than MLT-747.
Number	Compounds	Libdock score	Number	Compounds	Libdock score
1	ZINC000100822245	174.177	11	ZINC000002528509	142.993
2	ZINC000011666642	159.417	12	ZINC000002126785	142.908
3	ZINC000002528510	151.66	13	ZINC000004654839	142.895
4	ZINC000031169598	147.637	14	ZINC000005157592	141.349
5	ZINC000000900047	147.068	15	ZINC000004098928	141.216
6	ZINC000002526389	145.779	16	ZINC000044417879	141.099
7	ZINC000014592909	145.731	17	ZINC000004649679	140.89
8	ZINC000027646631	145.329	18	ZINC000019340795	140.568
9	ZINC000032840903	145.175	19	ZINC000014767597	140.176
10	ZINC000014767590	143.067	20	ZINC000027646625	140.059

Table 2

ADME (Adsorption, Distribution, Metabolism, Excretion) properties of compounds.
Compounds	Solubility level^a	BBB level^b	CYP2D6^c	Hepatotoxic^d	Absorption level^e	PBB level^f
ZINC000100822245	2	4	1	0	3	0
ZINC000011666642	3	4	0	1	3	0
ZINC000002528510	2	4	1	1	0	1
ZINC000031169598	2	4	0	0	3	1
ZINC000000900047	2	1	0	1	0	1
ZINC000002526389	2	4	1	1	0	1
ZINC000014592909	1	4	0	0	3	1
ZINC000027646631	3	1	0	0	0	0
ZINC000032840903	3	4	1	0	1	0
ZINC000014767590	0	4	0	0	3	1
ZINC000002528509	2	4	1	1	0	1
ZINC000002126785	1	4	0	0	3	1
ZINC000004654839	0	4	0	0	3	1
ZINC000005157592	2	4	1	0	2	1
ZINC000004098928	2	3	0	1	0	1
ZINC000044417879	1	4	0	0	3	1
ZINC000004649679	3	4	0	0	0	1
ZINC000019340795	2	2	0	0	0	0
ZINC000014767597	0	4	0	0	3	1
ZINC000027646625	3	1	0	0	0	0
MLT-747	2	4	0	1	0	0

Table 3

Toxicities of compounds.
Number	Compounds	Mouse NTP^a		Rat NTP		AMES^b	DTP^c
Number	Compounds	Female	Male	Female	Male	AMES^b	DTP^c
1	ZINC000100822245	0.470	0.348	0.325	0.486	0.080	0.856
2	ZINC000011666642	0.428	0.548	0.315	0.586	0.095	0.863
3	ZINC000002528510	0.327	0.509	0.433	0.477	0.000	0.607
4	ZINC000031169598	0.585	0.547	0.279	0.304	0.027	0.414
5	ZINC000000900047	0.179	0.403	0.563	0.375	0.525	0.571
6	ZINC000002526389	0.327	0.509	0.433	0.477	0.000	0.607
7	ZINC000014592909	0.598	0.613	0.270	0.401	0.000	0.527
8	ZINC000027646631	0.646	0.614	0.280	0.363	0.007	0.361
9	ZINC000032840903	0.652	0.605	0.110	0.201	0.000	0.442
10	ZINC000014767590	0.668	0.627	0.278	0.581	0.000	0.518
11	ZINC000002528509	0.299	0.439	0.422	0.483	0.000	0.618
12	ZINC000002126785	0.775	0.781	0.267	0.255	0.001	0.772
13	ZINC000004654839	0.753	0.614	0.356	0.592	0.000	0.516
14	ZINC000005157592	0.588	0.650	0.165	0.314	0.001	0.487
15	ZINC000004098928	0.309	0.467	0.257	0.200	0.518	0.656
16	ZINC000044417879	0.519	0.721	0.136	0.114	0.000	0.445
17	ZINC000004649679	0.557	0.613	0.397	0.329	0.366	0.667
18	ZINC000019340795	0.138	0.232	0.421	0.364	0.000	0.609
19	ZINC000014767597	0.668	0.627	0.278	0.581	0.000	0.518
20	ZINC000027646625	0.646	0.614	0.280	0.363	0.007	0.361
21	MLT-747	0.411	0.431	0.438	0.462	0.505	0.454

Binding interactions between selected compounds and MALT1

Combinations between selected molecules, MLT-747 and MALT1, were analyzed by the CDOCKER module of DS 4.5. The CDOCKER interaction energies of compounds 1 and 2 were − 52.6735 and − 54.517 Kcal/mol, respectively, indicating their high affinity to MALT1. Additionally, Hydrogen bond, Carbon hydrogen bond, Halogen interaction, Pi-Alkyl interaction, Pi-Sultur interaction, and Alkyl interaction were assessed between selected molecules, MLT-747 and MALT1 in 2D and 3D structures (Figs. 7, 8, and Table 4, 5). ZINC000019340795 formed two Hydrogen bonds, two Carbon hydrogen bonds, five Pi-Alkyl interactions, one Pi-Sultur interaction, and four Alkyl interactions with MALT1(Table 6). ZINC000004649679 formed four Carbon hydrogen bonds and two Alkyl interactions with MALT1. MLT-747 formed one Hydrogen bond, six Carbon hydrogen bonds, one Halogen interaction, eight Pi-Alkyl interactions, and four Alkyl interactions with MALT1.

Table 4

CDOCKER interaction energy of compounds with MALT1.
Compounds	CDOCKER Interaction energy (Kcal/mol)
ZINC000019340795	-52.6735
ZINC000004649679	-54.517

Table 5

Hydrogen bond, Carbon hydrogen bond, and Halogen interaction parameters for each compound and MALT1 residues.
Receptor	Compound	Donor atom	Receptor atom	Distances (Å)
Hydrogen bond	MLT-747	Molecule:UNK0:H49	B:GLU397:OE1	2.39
	ZINC000019340795	B:GLN579:HE21	B:GLN579:HE21	2.07
	ZINC000019340795	ZINC000019340795:N23	B:GLU397:OE1	2.21
Carbon hydrogen bond	MLT-747	Molecule:UNK0:O21	B:GLN676:HA	2.95
		Molecule:UNK0:H34	B:ALA394:O	2.5
		Molecule:UNK0:H43	B:MET717:OCT1	2.48
		Molecule:UNK0:H47	B:GLU397:OE2	2.55
		Molecule:UNK0:H52	B:LEU715:O	2.38
		Molecule:UNK0:H53	B:GLN676:O	2.8
	ZINC000019340795	ZINC000019340795:H46	B:GLU397:O	2.71
	ZINC000019340795	ZINC000019340795:H54	B:GLU397:OE1	2.35
	ZINC000004649679	ZINC000004649679:H28	B:ALA345:O	2.91
		ZINC000004649679:H30	B:ALA345:O	2.78
		ZINC000004649679:H45	B:GLU390:OE2	2.58
		ZINC000004649679:H45	B:GLN676:O	2.59
Halogen interaction	MLT-747	Molecule:UNK0:CL27	B:LYS379:O	3.26

Table 6

Pi-Alkyl interaction, Pi-Sultur interaction and Alkyl interaction parameters for each compound and MALT1 residues
Receptor	Compound	Donor atom	Receptor atom	Distances (Å)
Pi-Alkyl	MLT-747	Molecule:UNK0	B:ALA394	4.88
		Molecule:UNK0	B:ILE712	5.07
		Molecule:UNK0	B:LEU715	5.2
		Molecule:UNK0	B:LEU346	5.01
		Molecule:UNK0	B:VAL381	3.96
		Molecule:UNK0	B:LEU401	5.3
		Molecule:UNK0	B:VAL381	4.06
		Molecule:UNK0	B:LEU401	4.65
	ZINC000019340795	ZINC000019340795	B:VAL381	4.28
		ZINC000019340795	B:ALA394	4.85
		ZINC000019340795	B:VAL344	5.12
		ZINC000019340795	B:VAL381	4.21
		ZINC000019340795	B:LEU401	4.75
	ZINC000004649679	ZINC000004649679	B:VAL381	4.56
	ZINC000004649679	ZINC000004649679	B:LEU401	5.28
Pi-Sultur	ZINC000019340795	ZINC000019340795:S11	B:PHE398	4.91
Alkyl	MLT-747	Molecule:UNK0:C11	B:ALA583	3.64
		Molecule:UNK0:C11	B:VAL381	4.6
		Molecule:UNK0:C11	B:LEU715	4.67
		Molecule:UNK0:CL19	B:LEU383	4.44
		Molecule:UNK0:CL19	B:LEU386	4.96
		Molecule:UNK0:CL19	B:ILE712	4.19
	ZINC000019340795	ZINC000019340795	B:LEU400	4.86
		ZINC000019340795:C1	B:ILE712	4.73
		ZINC000019340795:C1	B:LEU715	4.45
		ZINC000019340795	B:LEU346	5.25

Pharmacophore analysis and molecular dynamics simulation

The analysis of feature pharmacophores showed that compound 1 had 18 pharmacophores, including eight hydrogen bond receptors, four hydrophobic centres, one hydrogen donor, four ring aromatic centres, and one inozable positive. Compound 2 had 21 pharmacophores, including eight hydrogen bond receptors, four hydrophobic centres, five hydrogen donors, and four ring aromatic centres. The MLT-747 had 17 pharmacophores, including seven hydrogen bond receptors, five hydrophobic centres, four hydrogen donors, and one ring aromatic centres (Fig. 9).

In addition, molecular dynamics simulation was carried out to evaluate the stability of the compound 1- MALT1 complex and compound 2- MALT1 complex in a simulated natural environment. The RMSD and potential energy curves of compound 1, 2-MALT1 complexes were plotted and gradually stabilized at 60 ps. The above results show that ZINC000031169598 ZINC000014592909 can bind to MALT1 with favorable affinity and stability in the natural environment.

Currently, immunotherapy has revolutionized treatment outcomes for many cancers and may provide new therapeutic approaches to treat GBM. A series of new and promising immunotherapeutic strategies have been applied to GBM, including vaccines, oncolytic viruses, immune checkpoint inhibitors, and adoptive cell transfer [2]. However, the special immunosuppressive microenvironment of GBM results in GBM patients’ resistance to immunotherapy and poor prognosis [3, 4]. MALT1 was reportedly involved in EGFR-induced NF-κB activation and promoted EGFR-related proliferation, survival, and migration. However, the effect of MALT1 on the GBM tumor microenvironment via NF-κB has not been investigated from an immunological perspective. Therefore, detailed studies of the tumor immune microenvironment of GBM are required to discover novel therapeutic molecular targets and improve prognosis from an immunological perspective.

In this study, although the MALT1_H group was enriched in functional, pathway, and immune cells, the prognosis of GBM patients remained poor. In contrast, the prognosis of breast cancer patients was found to be positively correlated with immune status [24]. The poor prognosis of GBM has been reported possibly due to the fact that GBM tumor-associated lymphocytes are often poorly functioning and severely depleted, which is consistent with our results [25]. In addition, the effect of MALT1 expression levels on the immunosuppressive microenvironment of GBM may have contributed to the difference in prognosis between the two groups. It has been reported that MALT1 activates the expression of downstream genes in solid tumors by regulating the activation of the NF-kB signaling pathway in lymphocytes, thereby regulating the development of immune cells and the immune response of the body and promoting tumor cell proliferation[26] [27] [28, 29]. And studies have shown that when the MALT1 gene is knocked out, regulatory T cells are not generated in mice. In cancer, persistent inflammatory signaling diverts immature myeloid cells (IMCs) from normal differentiation and is pathologically activated to become myeloid-derived suppressor cells (MDSCs). Immature myeloid cells (IMCs) are a group of bone marrow-derived suppressor cells that are the precursors of dendritic cells (DCs), macrophages, and granulocytes. Compared with physiologically differentiated myeloid cells, MDSCs have distinct characteristics, such as immature phenotype and morphology, relatively weak phagocytic function, and anti-inflammatory and immunosuppressive functions. Therefore, MALT1 may promote the development of MDSCs and Treg cells by activating the NF-kB signaling pathway, thereby aggravating the inhibitory immune microenvironment of GBM and leading to the poor prognosis of GBM patients.

Moreover, 201 genes were identified as the DEIGs between MALT1_H and MALT1_L. According to the analysis of GO and KEGG, DEIGs were enriched in the extracellular matrix organization, collagen metabolism process, and cell adhesion regulation. The extracellular matrix is an essential component of the tumor microenvironment, and collagen is the main component of the extracellular matrix. Collagen cross-linking and deposition have been reported to increase the rigidity of the extracellular matrix, which can promote malignant transformation, invasion, and metastasis of tumors[30]. In addition, DEIGs are also enriched in negatively regulating cell motility and negatively regulating antigen processing and presentation. So, MALT1_H may contribute to the inhibitory immune microenvironment of GBM due to a lack of antigen-presenting cells or defective antigen-presenting cell function. Moreover, DEIGs are enriched in regulating vascular endothelial growth factor production, and angiogenesis is critical in promoting tumor metastasis. Therefore, MALT1 may promote tumor proliferation, invasion, and metastasis by regulating extracellular matrix, inhibiting antigen presentation, and promoting neovascularization, resulting in poor prognosis of GBM patients. And MALT1 is expected to be a prognostic marker and therapeutic target for GBM.

Additionally, the assessment of HLA expression indicated that most HLA genes had higher expression levels in MALT1_H than MALT1_L. HLA expression level has been reported to correlate with histological type and tumor grade[24]. Immunotherapy has a better therapeutic effect on GBM patients with low HLA-DR expression [31]. Therefore, GBM patients of MALT1_L are more likely to benefit from immunotherapy. In addition, most immune checkpoint genes had significantly higher expression levels in MALT1_H than in MALT1_L, such as CD274, BTLA, and ICOS. This result suggests that GBM patients in MALT1_H are immunocompromised and may be more likely to benefit from immune checkpoint inhibitors. So, this study provides a new reference metric for assessing the suitability of immunotherapy for GBM patients.

Furthermore, we established and validated a MALT1-related IPS that could help predict the prognosis of GBM patients with AUC values of 0.727 and 0.652 in the training and validation sets, respectively. The MALT1-related IPS was established and demonstrated to be an independent prognostic factor based on multivariate and univariate Cox analyses. And based on the three-gene IPS, we constructed a nomogram to predict 1- and 3-year survival in GBM patients from an immunological perspective.

In addition, AREG, PODNL1, and PDYN were identified as hub genes by LASSO Cox regression analysis. AREG is the potential target of multiple tumors, such as colorectal cancer and head and neck squamous cell carcinoma [32, 33]. In addition, AREG expression levels and methylation levels have been shown to correlate with patient prognosis and the degree of malignancy of GBM [34]. Targeting AREG has been reported to minimize the chemoresistance of cancer cells[35]. So, it is expected to be a prognostic marker and a therapeutic target for GBM. Moreover, studies have found that PODNL1 was overexpressed in GBM patients and associated with poorer survival [36]. Similarly, PODNL1 was more highly expressed in high-risk than in low-risk in our study, which is consistent with previous findings and demonstrates the model's reliability. Additionally, PDYN encodes a preproprotein that can be proteolytically processed to form secreted opioid peptides, such as dynorphin, leu-enkephalin, beta-neoendorphin, and so on [37]. Opioid peptides are essential regulators of the immune system, so it is speculated that it may affect prognosis by modulating the suppressive immune microenvironment of GBM. More importantly, PDYN was first found to be associated with GBM prognosis and was identified as a potential target of GBM.

Recently, studies on MALT1’s inhibitors have attracted a lot of attention. Several inhibitors of MALT1 have been reported, such as miR-181d, MI-2, MLT-747, etc [4, 38] [39]. But the research on MALT1 inhibitors is still in the research stage, so it is of great significance to develop favorable inhibitors and explore the mechanism of interaction between inhibitors and MALT1 for targeted therapy of GBM. Among these drugs known, MLT-747 is a well- studied inhibitor and was chosen as the reference drug.

Furthermore, a series of computer-aided technology, such as Discovery Studio 4.5, Schrodinger, and PyMol, were applied to screen favorable inhibitors of MALT1. Higher Libdock scores suggest better energy optimization and more stable conformation. First, the top 20 molecules by Libdock score were selected and subjected to further pharmacological and toxicological analysis. Then, ZINC000019340795 and ZINC000004649679 were selected as favorable inhibitors of MALT1 due to non-CYP2D6 inhibition, no hepatotoxicity, low oncogenicity, and low mutagenicity in rodents. In addition, both selected molecules and MLT-747 docked to MALT1 in the active pocket and formed various chemical bonds. The CDOCKER interaction energies of compounds 1 and 2 with MALT1 were − 52.6735 and − 54.517 Kcal/mol, respectively, indicating their high affinity to MALT1. Furthermore, both selected molecules formed multiple pharmacophores, which showed their potential and application prospect for GBM-targeted drug development. Additionally, according to the results of molecular dynamics simulations, the RMSD and potential energies gradually stabilized over time, suggesting that these two complexes can exist stably in the natural environment.

In summary, we explored the mechanism by which MALT1 affects the immune microenvironment of GBM, established a novel MALT1-related nomogram model for prognostic prediction and developed the targeted therapy for GBM in this study. However, we have to admit that this study has some limitations. Additional studies such as animal-controlled trials are required to verify the drug's efficacy and pharmacological and toxicological properties.

From an immunological perspective, our analysis analyzed the impact of MALT1 on the GBM immune microenvironment and the patient’s prognosis. Moreover, AREG, PODNL1, and PDYN were identified as the hub genes, based on which MALT1-related IPS was established and demonstrated to be an independent prognostic factor based on multivariate and univariate Cox analyses. More importantly, PDYN was first found to be associated with GBM prognosis and was identified as a potential target of GBM. Furthermore, we constructed a nomogram to predict 1- and 3-year survival in GBM patients. Additionally, MALT1 is considered an essential target by regulating the immune microenvironment of GBM. So, a series of computer-aided techniques were applied to screen for favorable MALT1 inhibitors. Finally, ZINC000019340795 and ZINC000004649679 were selected as ideal inhibitors of MALT1 with high affinity, stability and reasonable pharmacological and toxicological properties. In summary, we explored the mechanism by which MALT1 affects the immune microenvironment of GBM, established a novel MALT1-related nomogram model for prognostic prediction and developed the targeted therapy for GBM in this study.

Ethical Approval and consent to participate

Not applicable

Consent to Publication

All authors have approved the publication of this work.

Data Availability statement

Not applicable

Acknowledgment

We thanked the foundation of the open innovation experiment of the College of Basic Medical Sciences, Jilin University (2018)

Funding Information

Not applicable

Conflict of Interest

All authors declare no conflicts of interest related to this manuscript, and all authors have approved the publication of this work.

Author Contribution statement

The research was done through teamwork. Every author has made a lot of contributions to the study. Hui Li had come up with the conception. Additionally, Zhishan Du did the design work. Han Lu completed the experiment and data collection part. Jianxin Xi and Zhenhua Wang were involved in the analysis of data. Jun Chen contributed a lot to interpreting the data. Yutang Li was responsible for the creation of new software used in work. Zhishan Du had drafted the work. And Sheng Zhong had substantively revised it.

Availability of data and materials

TCGA (The Cancer Genome Atlas, https://portal.gdc.cancer.gov/repository)

the Protein Database (PDB, https://www.rcsb.org/)

the ZINC15 database (https://zinc.docking.org/)

Louveau, A., et al., Structural and functional features of central nervous system lymphatic vessels. Nature, 2015. 523(7560): p. 337-41.
Gudipati, V., et al., Inefficient CAR-proximal signaling blunts antigen sensitivity. Nat Immunol, 2020. 21(8): p. 848-856.
Lim, M., et al., Current state of immunotherapy for glioblastoma. Nat Rev Clin Oncol, 2018. 15(7): p. 422-442.
Do, A.S.S., et al., CD133 mRNA-Loaded Dendritic Cell Vaccination Abrogates Glioma Stem Cell Propagation in Humanized Glioblastoma Mouse Model. Mol Ther Oncolytics, 2020. 18: p. 295-303.
Kesarwani, P., et al., Metabolic remodeling contributes towards an immune-suppressive phenotype in glioblastoma. Cancer Immunol Immunother, 2019. 68(7): p. 1107-1120.
Kloepper, J., et al., Ang-2/VEGF bispecific antibody reprograms macrophages and resident microglia to anti-tumor phenotype and prolongs glioblastoma survival. Proc Natl Acad Sci U S A, 2016. 113(16): p. 4476-81.
Jaworski, M. and M. Thome, The paracaspase MALT1: biological function and potential for therapeutic inhibition. Cell Mol Life Sci, 2016. 73(3): p. 459-73.
Coornaert, B., et al., T cell antigen receptor stimulation induces MALT1 paracaspase-mediated cleavage of the NF-kappaB inhibitor A20. Nat Immunol, 2008. 9(3): p. 263-71.
Fontan, L., et al., Specific covalent inhibition of MALT1 paracaspase suppresses B cell lymphoma growth. J Clin Invest, 2018. 128(10): p. 4397-4412.
Zhu, X., et al., The antidepressant- and anxiolytic-like effects of resveratrol: Involvement of phosphodiesterase-4D inhibition. Neuropharmacology, 2019. 153: p. 20-31.
Konczalla, L., et al., Biperiden and mepazine effectively inhibit MALT1 activity and tumor growth in pancreatic cancer. Int J Cancer, 2020. 146(6): p. 1618-1630.
McAllister-Lucas, L.M., M. Baens, and P.C. Lucas, MALT1 protease: a new therapeutic target in B lymphoma and beyond? Clin Cancer Res, 2011. 17(21): p. 6623-31.
Hailfinger, S., et al., Essential role of MALT1 protease activity in activated B cell-like diffuse large B-cell lymphoma. Proc Natl Acad Sci U S A, 2009. 106(47): p. 19946-51.
Chen, Y.Y., et al., Upregulation of miR-125b, miR-181d, and miR-221 Predicts Poor Prognosis in MGMT Promoter-Unmethylated Glioblastoma Patients. Am J Clin Pathol, 2018. 149(5): p. 412-417.
Barbie, D.A., et al., Systematic RNA interference reveals that oncogenic KRAS-driven cancers require TBK1. Nature, 2009. 462(7269): p. 108-12.
Yoshihara, K., et al., Inferring tumour purity and stromal and immune cell admixture from expression data. Nat Commun, 2013. 4: p. 2612.
Servant, N., et al., EMA - A R package for Easy Microarray data analysis. BMC Res Notes, 2010. 3: p. 277.
Zhou, Y., et al., Metascape provides a biologist-oriented resource for the analysis of systems-level datasets. Nat Commun, 2019. 10(1): p. 1523.
Friedman, J., T. Hastie, and R. Tibshirani, Regularization Paths for Generalized Linear Models via Coordinate Descent. J Stat Softw, 2010. 33(1): p. 1-22.
Heagerty, P.J., T. Lumley, and M.S. Pepe, Time-dependent ROC curves for censored survival data and a diagnostic marker. Biometrics, 2000. 56(2): p. 337-44.
Preusser, M., et al., Prospects of immune checkpoint modulators in the treatment of glioblastoma. Nat Rev Neurol, 2015. 11(9): p. 504-14.
Ercheng, W., et al., Assessing the performance of the MM/PBSA and MM/GBSA methods. 10. Impacts of enhanced sampling and variable dielectric model on protein-protein Interactions. Physical chemistry chemical physics : PCCP, 2019. 21(35).
K, H.M., B. Hugues-Olivier, and H.R. E, Predicting fragment binding poses using a combined MCSS MM-GBSA approach. Journal of chemical information and modeling, 2011. 51(5).
Kaffes, I., et al., Human Mesenchymal glioblastomas are characterized by an increased immune cell presence compared to Proneural and Classical tumors. Oncoimmunology, 2019. 8(11): p. e1655360.
Woroniecka, K., et al., T-Cell Exhaustion Signatures Vary with Tumor Type and Are Severe in Glioblastoma. Clin Cancer Res, 2018. 24(17): p. 4175-4186.
Lucas, P.C., et al., Bcl10 and MALT1, independent targets of chromosomal translocation in malt lymphoma, cooperate in a novel NF-kappa B signaling pathway. J Biol Chem, 2001. 276(22): p. 19012-9.
Liu, X., et al., MALT1 is a potential therapeutic target in glioblastoma and plays a crucial role in EGFR-induced NF-κB activation. J Cell Mol Med, 2020. 24(13): p. 7550-7562.
Gewies, A., et al., Uncoupling Malt1 threshold function from paracaspase activity results in destructive autoimmune inflammation. Cell Rep, 2014. 9(4): p. 1292-305.
Yamasoba, D., et al., N4BP1 restricts HIV-1 and its inactivation by MALT1 promotes viral reactivation. Nat Microbiol, 2019. 4(9): p. 1532-1544.
Erler, J.T. and V.M. Weaver, Three-dimensional context regulation of metastasis. Clin Exp Metastasis, 2009. 26(1): p. 35-49.
Wu, Z., et al., HLA-E expression in diffuse glioma: relationship with clinicopathological features and patient survival. BMC Neurol, 2020. 20(1): p. 59.
Lee, M.S., et al., Association of CpG island methylator phenotype and EREG/AREG methylation and expression in colorectal cancer. Br J Cancer, 2016. 114(12): p. 1352-61.
Aderhold, C., et al., Targeting mTOR and AREG with everolimus, sunitinib and sorafenib in HPV-positive and -negative SCC. Anticancer Res, 2015. 35(4): p. 1951-9.
Steponaitis, G., et al., Significance of Amphiregulin (AREG) for the Outcome of Low and High Grade Astrocytoma Patients. J Cancer, 2019. 10(6): p. 1479-1488.
Xu, Q., et al., Targeting amphiregulin (AREG) derived from senescent stromal cells diminishes cancer resistance and averts programmed cell death 1 ligand (PD-L1)-mediated immunosuppression. Aging Cell, 2019. 18(6): p. e13027.
Shergalis, A., et al., Current Challenges and Opportunities in Treating Glioblastoma. Pharmacol Rev, 2018. 70(3): p. 412-445.
Gieryk, A., et al., Forebrain PENK and PDYN gene expression levels in three inbred strains of mice and their relationship to genotype-dependent morphine reward sensitivity. Psychopharmacology (Berl), 2010. 208(2): p. 291-300.
Zhao, D., J. Yang, and L. Yang, Insights for Oxidative Stress and mTOR Signaling in Myocardial Ischemia/Reperfusion Injury under Diabetes. Oxid Med Cell Longev, 2017. 2017: p. 6437467.
Fontan, L., et al., MALT1 small molecule inhibitors specifically suppress ABC-DLBCL in vitro and in vivo. Cancer Cell, 2012. 22(6): p. 812-24.

No competing interests reported.

Supportinginformation1.docx
Supporting information 1: The clinical features of the prognosis cohort associated with MALT1 expression.
Supportinginformation2.docx
Supporting information 2: The 29 immune signatures represented by 29 different gene sets.
Supportinginformation3.xlsx
Supporting information 3: Clinical Characteristics
Supportinginformation4.xlsx
Supporting information 4: RNA-seq data

Download PDF

Version 1

posted

You are reading this latest preprint version

A Novel MALT1-related Immune Prognostic Signature and Targeted Drug Screening for Glioblastoma

Status:

Version 1

Abstract

Figures

Introduction

Materials and methods

Datasets of gene expression and processing of data

Immunogenomic analysis

Survival analysis

Differential expression analysis

Analysis of functional enrichment

Construction and validation of the IPS (immune prognostic signature)

Development of the nomogram

Virtual screening using Libdock

Pharmacological and toxicological properties analysis ADME and TOPKAT

Binding interactions between selected compounds and MALT1

Pharmacophore analysis and Molecular dynamics simulation

Results

Immunogenomic analyses between MALT1_H and MALT1_L

Identification of DEIGs (differentially expressed immune genes) and functional enrichment

Construction and validation of the MALT1-related IPS (immune prognostic signature)

Establishment of a nomogram model based on IPS

Virtual screening of MALT1’s inhibitors using Libdock, ADME, TOPKAT

Binding interactions between selected compounds and MALT1

Pharmacophore analysis and molecular dynamics simulation

Discussion

Conclusion

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1