Development and experimental validation of an M2 macrophage and platelet‐associated gene signature to predict prognosis and immunotherapy sensitivity in bladder cancer

Abstract

Platelets and M2 macrophages both play crucial roles in tumorigenesis, but their relationship and the prognosis value of the relative genes in bladder cancer (BLCA) remain obscure. In the present study, we found that platelets stimulated by BLCA cell lines could promote M2 macrophage polarization, and platelets were significantly associated with the infiltration of M2 macrophages in BLCA samples. Through the bioinformatic analyses, A2M, TGFB3, and MYLK, which were associated with platelets and M2 macrophages, were identified and verified in vitro and then included in the predictive model. A platelet and M2 macrophage‐related gene signature was constructed to evaluate the prognosis and immunotherapeutic sensitivity, helping to guide personalized treatment and to disclose the underlying mechanisms.

Platelets stimulated by UMUC3 could promote M2 macrophage polarization, and platelets were significantly associated with the infiltration of M2 macrophages in bladder cancer samples. Through the bioinformatic analyses, A2M, TGFB3, and MYLK, which were associated with platelets and M2 macrophages, were identified and verified in vitro and then included in the predictive model. A platelet and M2 macrophage‐related gene signature was constructed to evaluate the prognosis and immunotherapeutic sensitivity, helping to guide personalized treatment and to disclose the underlying mechanisms.

Abbreviations

A2M

alpha‐2‐macroglobulin

ARG‐1

arginase‐1

AUC

area under the receiver operating characteristic curve

BLCA

bladder cancer

CI

confidence interval

IF

immunofluorescence

IL

interleukin

MPRS

M2 macrophage–platelet risk score

MYLK

myosin light chain kinase

OR

odds ratio

OS

overall survival

qPCR

quantitative RT‐PCR

TCGA

The Cancer Genome Atlas

TEP

tumor‐educated platelet

TGFB

transforming growth factor β isoform

WB

western blot analysis

WGCNA

Weighted Gene Co‐expression Network Analyses

1. INTRODUCTION

Bladder cancer has become one of the most prevalent tumors worldwide, ranking ninth in 2022 in the morbidity of cancer. ¹ The most predominant pathological category of BLCA is uroepithelial carcinoma, which is divided into nonmuscle‐invasive and muscle‐invasive subtypes. ² Even though the rise of some novel treatment strategies, such as target therapy and immune checkpoint inhibitors, has improved BLCA patients' clinical outcomes, many patients still cannot benefit from these treatments. ³ Hence, seeking more reliable prognosis predictors is urgently demanded, which also provides novel therapeutic targets, helping to guide individualized treatment and drug development.

Macrophages are highly heterogeneous and show different phenotypes in different microenvironments. ⁴ Depending on the activation status and functions performed, macrophages are broadly classified into M1 macrophages (i.e. classically activated macrophages) and M2 macrophages (i.e. alternatively activated macrophages). ⁵ Unlike M1 macrophages, which promote the antitumor immune response, M2 macrophages have a weaker antigen‐presenting role, mainly exerting immunosuppressive functions. ⁶ Recent studies have revealed that tumor stem cells could recruit polarized M2 macrophages, thus forming an immune microenvironment that is conducive to tumor cells evading immune killing and completing the proliferation and metastasis process. ⁷ Supporting this hypothesis, several studies reported that the high abundance of M2 macrophages heralded unfavorable prognosis in BLCA. ⁸ M2 macrophage has a nonnegligible influence on the initiation and progression of malignant cancers, but our knowledge of the M2 macrophage in BLCA remains limited.

Platelets, one of the most abundant blood cells in the body, were previously thought to cause mainly tumor‐related cardiovascular complications such as thrombosis in tumor development. Nevertheless, emerging evidence suggested that platelets acted as local and systemic responders during tumorigenesis and cancer metastasis as well. ⁹ The platelets exposed to tumor‐induced education, resulting in altered platelets, were known as TEPs. ¹⁰ However, the potential functions of platelets in BLCA and their corresponding mechanisms have not been elucidated.

Platelets and M2 macrophages act as cancer‐promoting factors in many cancers. There is an argument about the direction of platelet‐promoting macrophage polarization in different diseases. The activated platelets could induce polarization of the M2 macrophages in chronic inflammatory diseases like rheumatoid arthritis, ¹¹ atherosclerotic disease, ¹² and lung cancer through miRNA ¹³ ; on the contrary, emerging evidence has shown that platelets facilitate polarization of monocytes toward M1 macrophages. ¹⁴ Hence, we need to urgently clarify the regulatory relationship between platelets and M2 macrophages in BLCA.

The present study implemented in vitro cell experiments, verification in clinical samples, and bioinformatical analyses to investigate the association between platelets and M2 macrophages and to construct a platelet and M2 macrophage‐related gene signature to predict prognosis and immunotherapeutic sensitivity.

2. MATERIALS AND METHODS

2.1. Data collection and processing

The platelet‐related genes were retrieved from the Molecular Signatures Database (https://www.gsea‐msigdb.org/gsea/msigdb/) with the keyword “platelet”. A sum of 478 platelet‐related genes was extracted, as displayed in Table S1. The transcriptome sequencing data of 414 BLCA samples were downloaded from TCGA (https://portal.gdc.cancer.gov/), as the training cohort. One hundred and sixty‐five BLCA samples from the GSE13507 dataset were used for external validation, obtained from the Gene Expression Omnibus (https://www.ncbi.nlm.nih.gov/geo/). The clinical parameters of these cohorts are shown in Table S2.

2.2. Cell culture

THP‐1 cells and cancer cell lines were provided by the Cell Bank of the Chinese Academy of Sciences. THP‐1, 786O, and C4‐2 cells were cultured in RPMI‐1640 medium and stimulated with 200 nmol concentration of PMA for 24 h before coculture with platelets or TEPs. Other cells were cultured in DMEM.

2.3. Isolation of platelets

The platelet‐rich serum was collected by centrifugation at 120 g for 20 min at room temperature, then centrifuged at 360 g for 20 min to separate the platelets from the supernatant.

2.4. Bladder cancer tissue IF analysis

Twenty‐one cancer tissues of BLCA patients were obtained from Nanfang Hospital between 2018 and 2022. The protocol of this study was reviewed and approved by the Ethics Committee of Nanfang Hospital, Southern Medical University, and the informed consent of each subject was obtained after admission. The Abs used in this experiment are shown in Table S3.

2.5. Western blot analysis

THP‐1 and its differentiated cells were washed three times with prechilled PBS buffer (300 g × 5 min), then lysed by adding 10% PMSF RIPA lysate. Details are explained in Appendix S1.

2.6. RNA isolation and qPCR assays

Total RNA was extracted from the cells by the TRizol (Thermo fisher, 15596026) method and its concentration and purity were measured by spectrophotometer. Real‐time qPCR was carried out in triplicate on a 7500 PCR system. Sequences of PCR primers are shown in Table S3.

2.7. Enzyme‐linked immunosorbent assay

The supernatants of cocultured (or single cell) cells were collected and centrifuged at 400 g for 20 min to adequately remove cell debris and platelets, and the supernatants were assayed for IL‐10 (EK110HS‐48; Multisciences) and ARG‐1 (ELK1790; ELK Biotechnology) according to the instructions provided by the ELISA kit vendor.

2.8. Weight coexpression network analysis

All samples in TCGA were included in the WGCNA analysis to screen the hub gene modules associated with M2 macrophages. The WGCNA is an algorithm to screen for coexpressed gene modules with high biological significance and explore the correlation between disease and gene modules. After rejecting outliers by hierarchical clustering analysis with the Hclust function in R language, we chose the optimal soft threshold to ensure a scale‐free network. Subsequently, the clustered adjacency matrix was used to identify hub modules. The four modules with the strongest positive correlation were selected for further analysis by calculating the Pearson correlation coefficients between each module and the CAF (cancer‐associated fibroblast) scores.

2.9. Construction and external validation of an M2 macrophage‐platelet‐associated risk model

Sequence‐based risk score models were constructed using a cross‐section of M2 macrophage‐related genes and platelet‐related genes. Univariate and multifactorial Cox regression analyses were undertaken to screen genes associated with BLCA prognosis and calculate their risk coefficients. Risk scores were calculated based on the expression of target genes and risk coefficients of each gene and named the MPRS: $\text{MPRS}={\sum }_i^n\left({\text{gene}}_i*{\mathrm{EXP}}_i\right)$, where EXP represents the mRNA expression value of the particular gene. The TCGA database and the GSE13507 cohort were used for training and external validation of the MPRS risk model, respectively.

2.10. Statistical analysis

Statistical analyses of the whole study were undertaken with R software (R Development Core Team, version 4.2.0) and GraphPad Prism (GraphPad Software Inc., version 8.0.1).

3. RESULTS

3.1. Tumor‐educated platelets stimulate polarization of M2 macrophages

The workflow of this study is shown in Figure 1. To investigate the role of BLCA cells in the process of platelet regulation of M2 macrophage polarization, platelets stimulated by the supernatants of BLCA cell lines SW780, UMUC3, and T24 were cocultured with PMA‐differentiated THP‐1 monocytes for 48 h (Figure 2A). Western blot analysis was used to detect CD80 (marker of M1 macrophages) and CD163 (marker of M2 macrophages) to determine the polarization trend of macrophages (Figure 2B). It was found that the expression of CD163 was upregulated after THP‐1 cells were treated with TEPs (stimulated by BLCA cells of three different degrees of malignancy) compared with normal platelets. In contrast, the expression of CD80 was downregulated in different degrees, that is, THP‐1 cells were more inclined to polarization to M2 macrophages, rather than M1 macrophages. To explore whether this platelet acclimation is BLCA‐specific, we added renal and prostate cancer cells and found that the renal cancer cell line 786O, but not so much in the prostate cancer cell line C4‐2, had the same acclimatization effect as BLCA, stimulating platelets to gain a function that promoted M2 macrophage polarization (Figure 2B). In addition, the supernatant of the coculture system was analyzed by ELISA for the secretion markers IL‐10 and ARG‐1 of M2 cells (Figure 2C), and immunofluorescence staining for macrophages (CD163 labeled M2 macrophages) (Figure 2D). It was further confirmed that TEP produced by BLCA cells and renal cancer cell line 786O significantly increased the proportion of M2 macrophages, while untreated platelets and TEP produced by prostate cancer cell line C4‐2 did not show this effect. Subsequently, we carried out further verification in vivo: IF staining of 21 BLCA samples showed that the platelet infiltration in BLCA tissues was positively correlated with M2 macrophages (Figure 2E). The quantification of these results is shown in Figure 2F (Spearman r = 0.754, p < 0.0001). All these results suggested that the interaction between platelets and BLCA cells enhanced the ability of platelets to promote the polarization of M2 macrophages. This phenomenon was not unique to BLCA cells, but also reflected in renal cancer cell line 786O.

FIGURE 1.

Workflow of this study. BCLA, bladder cancer; MPRS, M2 macrophage–platelet risk score; TCGA, The Cancer Genome Atlas; TEP, tumor‐educated platelet.

FIGURE 2.

Experimental validation of the relationship between platelets and M2 macrophages in the bladder cancer (BLCA) environment. (A) Experimental protocol and flow of experiments. (B) Protein extracts were obtained from macrophages treated or not with platelets or tumor‐educated platelets (TEPs). Western blot analysis was carried out to detect the markers of M1 (CD80⁺) and M2 (CD163⁺) macrophages. (C) ELISA analysis of arginase‐1 (ARG‐1) and interleukin‐10 (IL‐10) in the supernatants of cocultured cells (macrophages treated or not with platelets or TEPs). (D) Representative pictures of macrophages (red, CD163) cocultured with platelets (untreated) or TEPs, and the quantification of the proportion of M2 macrophage (CD163⁺). (E, F) Correlation between M2 macrophage (red, CD163) and platelet (green, CD42b) infiltration in (E) tumor tissues of BLCA patients and (F) the quantification. *p < 0.05; **p < 0.01; ***p < 0.001.

3.2. Construction of weighted gene coexpression network and functional analysis of critical module genes

A total of 36 gene modules were identified by WGCNA, with each color representing a different gene module. Among them, “blue” (CIBERSORT: r = 0.21, p < 0.05; CIBERSORT_ABS: r = 0.55, p < 0.0001; QUANTISEQ: r = 0.53, p < 0.0001; XCELL: r = 0.32, p < 0.01), “yellow” (CIBERSORT: r = 0.33, p < 0.001; CIBERSORT_ABS: r = 0.7, p < 0.0001; QUANTISEQ: r = 0.56, p < 0.0001; XCELL: r = 0.37, p < 0.001), “midnightblue” (CIBERSORT: r = 0.49, p < 0.0001; CIBERSORT_ABS: r = 0.52, p < 0.0001; QUANTISEQ: r = 0.47, p < 0.0001; XCELL: r = 0.4, p < 0.0001), and “lightcyan” (CIBERSORT: r = 0.21, p < 0.05; CIBERSORT_ABS: r = 0.75, p < 0.0001; QUANTISEQ: r = 0.61, p < 0.0001; XCELL: r = 0.64, p < 0.0001) gene module and the degree of M2 macrophage infiltration in BLCA suggested significant positive correlation in all four analysis methods, CIBERSORT, CIBERSORT_ABS, QUANTISEQ, and XCELL, and was selected as the BLCA M2 macrophage infiltration related module (Figure 3A). With MM >0.8 and GS >0.2 filterings, 40 genes in the “blue” module (Figure S1), 21 genes in the “yellow” module (Figure S2), 38 genes in the “midnightblue” module (Figure S3), and 10 genes in the “lightcyan” module (Figure S4) were considered as M2 macrophage‐related genes. Overall, a total of 109 M2 macrophage‐related genes were identified. The functional annotation indicated that the genes in the “blue” module were mainly enriched in “vascular smooth muscle contraction” and “myofibril assembly” (Figure 3B). The genes in the “yellow” module were mainly associated with “regulation of transmembrane receptor protein serine” and “cell–extracellular matrix interaction” (Figure 3C). The genes in the “midnightblue” and “lightcyan” modules are enriched in the “cellular response to growth factor stimulus” and “lymphocyte activation” (Figure 3D,E), respectively.

FIGURE 3.

Identification of M2 macrophage‐related genes in bladder cancer using Weighted Gene Coexpression Network Analysis. (A) Module–trait relationship diagram. Thirty‐six modules were generated; the MEblue, MEyellow, MEmidnightblue, and MElightcyan modules were most significantly related to M2 macrophage infiltration. (B–E) Metascape enrichment networks of (B) MEblue module, (C) MEyellow module, (D) MEmidnightblue module, and (E) MElightcyan module. Clustering annotations were color‐coded.

3.3. Tumor‐educated platelets upregulate A2M/MYLK/TGFB3 expression levels in M2 macrophages

We used the keyword “platelet” to manually obtain the gene signatures associated with platelets in the Molecular Signatures Database (https://www.gsea‐msigdb.org/gsea/msigdb/). By taking the intersection of the platelet‐related gene sets and M2 macrophage‐related genes, four overlapping genes were obtained: A2M, MYLK, TGFB3, and ACTIN1. With a multivariate Cox regression analysis, MPRS was calculated as follows: $\text{MPRS}=0.001*\mathrm{EXP}\left(\mathrm{A}2\mathrm{M}\right)+0.015*\mathrm{EXP}\left(\text{TGFB}3\right)-0.011*\mathrm{EXP}\left(\text{MYLK}\right)$ (Figure 4A). The pan‐cancer analyses at a single‐cell level of A2M, MYLK, and TGFB3 were shown in Figure 4B–D, respectively, which were obtained from the CancerSEA database (http://biocc.hrbmu.edu.cn/CancerSEA/). These results implied that these genes were associated with multiple malignant phenotypes across different cancers. Platelets are anucleated cells, which makes their protein expression less susceptible to change than M2 macrophages, so we suspected that the differences in expression of these three genes were mainly reflected in macrophages. To test this hypothesis, we undertook qPCR (Figure 4E) and WB experiments (Figure 4F) on A2M, MYLK and TGFB3 of macrophages in the above coculture system to detect the changes in the expression of these three genes at the transcriptome and protein level. The statistical analyses of WB are displayed in Figure 4G. The results confirmed that TEPs, which were acclimated from different malignant BLCA cell lines, could upregulate the expression of A2M, MYLK, and TGFB3 in macrophages in this coculture system to varying degrees. However, the protein expression level was not significantly increased in SW780 group, which might be because SW780 is a less malignant BLCA cell line. To test our hypothesis in vivo, IF staining of tumor tissues from BLCA patients was carried out. Figure 5 shows the colocalization of A2M, MYLK, TGFB3 with M2 macrophages in high and low grade BLCA tissues. The above results verified that the expression of A2M, MYLK, and TGFB3 in M2 macrophages of BLCA tissues was associated with the malignant phenotype of the tumor.

FIGURE 4.

Screening and experimental validation of M2 macrophage–platelet‐related genes. (A) The intersection of M2 macrophage‐related and platelet‐related genes, and the stepwise multivariable Cox regression analysis results. (B–D) Functional enrichments of key genes A2M, MYLK, and TGFB3 in malignancies in the CancerSEA database. (E) Real‐time quantitative PCR and (F) western blot (WB) analyses of A2M, MYLK, and TGFB3 mRNA and protein expression levels in macrophages (THP‐1 stimulated by PMA for 24 h) cultured alone, with platelets, or tumor‐educated platelets. (G) Statistical analysis of the results of the WB experiment. *p < 0.05; **p < 0.01; ***p < 0.001.

FIGURE 5.

Expression of (A) A2M, (B) MYLK, and (C) TGFB3 markers in M2 macrophages (CD163⁺) in bladder cancer tumor tissues detected by immunofluorescence staining.

3.4. Development and validation of an M2 macrophage and platelet‐associated risk model

The forest plot in Figure 6A displays the coefficients and the corresponding p values of A2M, MYLK, and TGFB3 in the MPRS risk model. To help clinicians and researchers better understand the established model, we constructed a prognostic nomogram incorporating these three genes (Figure 6B). Then we generated corresponding risk scores (MPRS) for all samples in the TCGA database and divided the samples into low‐ and high‐MPRS groups based on the optimal cut‐off value detected by the X‐tile, which equaled 0.8514. The scatter plot showed that more deaths and lower survival time were observed with the increase of MPRS, both in the training (Figure 6C) and external validation cohorts (Figure 6D). Kaplan–Meier survival analyses showed that the high MPRS heralded unfavorable prognosis in the training cohort (p < 0.001; Figure 6E) and the validation cohort (p < 0.05; Figure 6F). In the training cohort, approximately 43% of high‐MPRS patients eventually died (p < 0.001; Figure 6G), while the proportion was 59% in the external validation cohort (p < 0.05; Figure 6H).

FIGURE 6.

Construction and validation of M2 macrophage–platelet‐related gene signature associated with prognosis of patients with bladder cancer (BLCA). (A) Multivariate Cox regression models identified three M2 macrophage–platelet‐related genes to construct the prognostic signature. (B) Nomogram of the model. (C) Overall survival (OS) rate and status of The Cancer Genome Atlas (TCGA) BLCA patients and the distribution of risk scores for each patient. (D) OS rate and status of BLCA patients in the GSE13507 cohort of Gene Expression Omnibus (GEO) and the distribution of risk scores for each patient. (E) OS of low‐risk and high‐risk groups in the training set TCGA‐BLCA cohort. (F) OS of low‐risk and high‐risk groups in the training set TCGA‐BLCA cohort. Distribution of high and low M2 macrophage–platelet risk score (MPRS) in alive and dead groups in the training (G) and validation (H) sets, respectively. *p < 0.05; **p < 0.01.

To assess the accuracy and predictive value of this risk assessment model, we plotted the receiver operating characteristic curves to evaluate the overall survival rates in 1–10 years, as displayed in Figure 7A–J, respectively. Additionally, MPRS was an independent prognosis predictor, both in the univariate and multivariate Cox regression analyses (both p < 0.05; Table S4). In total, MPRS showed superiority to the routine clinicopathologic parameters, especially in the long‐term survival rates.

FIGURE 7.

Performance comparison among M2 macrophage–platelet risk score (MPRS) and other clinical prognostic variables. (A) 1‐year, (B) 2‐year, (C) 3‐year, (D) 4‐year, (E) 5‐year, (F) 6‐year, (G) 7‐year, (H) 8‐year, (I) 9‐year, and (J) 10‐year receiver operating characteristic (ROC) curves of MPRS. AUC, area under the ROC curve; OS, overall survival.

3.5. Correlation of MPRS with biological function and immune infiltration

Gene Set Enrichment Analysis showed that MPRS was closely connected with the various biological functions of platelets and M2 macrophages, proving the signature is specific to platelets and M2 macrophages (Figure 8A,B). Then we undertook a differential analysis of immune checkpoint expression in the high MPRS and low MPRS groups of BLCA tumor samples. We found that the expression of PDL1, PD1, LAG3, CTLA4, TIM‐3, and TIGIT was significantly higher in the high MPRS group than in the low MPRS group (all p < 0.05, Figure 8C,D). We then investigated the difference in MPRS between the three immune phenotypes in the IMvigor210 cohort and found that the MPRS in the immune‐excluded and immune‐inflamed groups were significantly higher than that in the desert group (all p < 0.01; Figure 8E). In addition, the complete response/partial response group had a lower MPRS than the stable disease/progressive disease) group (p < 0.05; Figure 8F), and the subjects with high MPRS showed worse prognoses (p < 0.05; Figure 8G) in the IMvigor210 cohort. These results suggested that MPRS was a promising tool to evaluate the immunotherapeutic effectiveness in BLCA.

FIGURE 8.

Comparison of the risk groups in our study with immune checkpoint genes and existing immune subtype and prediction of response to immunotherapeutic agents for different risk groups. (A) M2 macrophage and (B) platelet related functional enrichment analysis based on M2 macrophage–platelet risk score (MPRS). (C) Boxplots showing the difference in immune checkpoint genes between different risk groups. (D) Chord diagram showing the correlations between MPRS and immune checkpoint genes. (E) MPRS of patients in different immune phenotype groups of the IMvigor210 cohort. (F) Comparison of risk scores between the complete response (CR)/partial response (PR) group and the stable disease (SD)/progressive disease (PD) group in the IMvigor210 cohort. (G) Overall survival curves between high‐ and low‐MPRS subgroups in the IMvigor210 cohort. *p < 0.05; **p < 0.01; ***p < 0.001.

3.6. Mutational landscape

We explored the differences in mutations between two risk groups based on the TCGA dataset. The top 30 genes in terms of mutation rate in the high‐ and low‐MPRS subgroups are shown in Figure 9A,B, respectively. Univariate logistic regression analysis identified that MPRS was significantly associated with the mutational statuses of FGFR3 (AUC = 0.627; OR = 12.891; 95% CI = 2.297–111.158; p = 0.010), KDM6A (AUC = 0.566; OR = 4.411; 95% CI = 1.537–15.976; p = 0.013), and MUC16 (AUC = 0.520; OR = 3.055; 95% CI = 1.232–9.105; p = 0.028) (Figure 9C–E and Table S5).

FIGURE 9.

Relationship between M2 macrophage–platelet risk score (MPRS) and mutation. (A, B) Waterfall plots showing mutation information of the top 30 genes with high mutation frequency in the (A) high‐risk group and (B) low‐risk group. (C–E) Receiver operating characteristic (ROC) curves of mutated genes: FGFR3, KDM6A, and MUC16. These three genes were screened using univariate linear regression analyses (p < 0.05). AUC, area under the ROC curve; OR, odds ratio.

4. DISCUSSION

It has been claimed that platelets can be domesticated by tumor cells and thus, their genomic profiles and biological processes are altered. ¹⁵ In turn, platelets could also affect tumor development and immune escape. ¹⁶ , ¹⁷ Platelets are reportedly capable of regulating the repolarization of macrophages, but there is some debate about this (see above). ¹¹ , ¹³ Specifically, platelets modulate the secretion of cytokines like IL‐10 and TGFB in regulatory T cells to induce polarization of M2 macrophages in pulmonary infections, ¹⁸ the products from platelets enhance IL‐4‐mediated M2 polarization through PGE2 in atherosclerosis, ¹⁹ and platelets mediate macrophage polarization to an inflammatory M1‐like state by inducing Socs3 expression in atherosclerosis. ²⁰ Hence, given the fact that the number of studies focusing on the interaction between M2 macrophages and platelets remains limited in BLCA, exploring the regulatory relationship between these two molecules is urgently required.

The present study found that platelets treated with tumor cell domestication and stimulation led to a significant increase in the percentage of M2 macrophages. Importantly, we collected 21 BLCA samples from the local hospital and a significant positive association between platelet number and M2 macrophage infiltration levels was observed, implying the possible interaction between platelets and M2 macrophages in BLCA tissues. Subsequently, we use multiple bioinformatic analyses to investigate the hub genes associated with the interaction between platelets and M2 macrophages: A2M, MYLK, and TGFB3, based on which we developed a prognostic model MPRS. We undertook in vitro cell experiments to verify our findings. Interestingly, we found that the regulation of M2 macrophage polarization by platelets in the tumor microenvironment was not unique to BLCA; it was also observed in renal cancer cells but not in prostate cancer cells. Previous studies have reported that the renal cancer microenvironment promotes the polarization of M2 macrophages, ²¹ whereas prostate cancer polarizes macrophages in favor of M1 macrophages, ²² which is consistent with our experimental results; however, an opposing view has been published. ²³ Our experimental results provide evidence to clarify this controversial issue from the perspective of platelet involvement.

Alpha‐2‐macroglobulin is a protease inhibitor, considered to be relevant to nephrotic syndrome, diabetes mellitus, Alzheimer's disease, and hemostasis. ²⁴ Several previous studies suggested the diagnostic value of A2M for cancers. For instance, as a GALNT6 substrate, A2M promotes metastasis of breast cancer cells through the AKT signaling pathway. ²⁵ Macrophages produce A2M and express both A2M receptors. In turn, A2M promotes phagocytosis, cell migration, and the release of mediators of inflammation including the platelet‐activating factor of macrophages. ²⁶ Additionally, M2 macrophages have been reported to express A2M in severe COVID‐19, and A2M is associated with the activation of M2 macrophages. ²⁷ Alpha‐2‐macroglobulin is also involved in platelet degranulation mainly released from α granules and platelets ingest or bind A2M from serum during prolonged storage. ²⁸

Transforming growth factor‐β isoforms, including TGFB1, B2, and B3, have been shown to play key roles in tumor development. ²⁹ Transforming growth factor B3 has been reported to promote cell migration and invasion as a tumor promotor in many cancers like breast cancer, ³⁰ lung cancer, ³⁰ and colorectal cancer, ³¹ mainly by activating TGFB pathways and epithelial–mesenchymal transition. ³² Transforming growth factor B3 can be secreted by M2 microglia/peripheral macrophages. ³³ In addition, platelets are the primary carrier of TGF‐β in the body, and TGFB3 genes showed a positive correlation with the number of platelets in the blood. ³⁴

Myosin light chain kinase is a calcium ion (Ca2+)‐dependent enzyme, contributing to various physiological processes related to myosin activation, such as cell adhesion, migration, and division. ³⁵ Previous studies have reported that MYLK also regulates cell migration and tumor metastasis in many cancers, such as gastric, prostate, breast, and ovarian cancer, through phosphorylation promoting cell contraction and changes in the actin cytoskeleton. ³⁶ Expression of MYLK is limited in platelets but is involved in platelet activation. ³⁷ , ³⁸ However, the role of A2M, TGFB3, and MYLK in BLCA development, especially the interaction with M2 macrophages or platelets in the BLCA microenvironment, has not been elucidated. We have found for the first time that A2M, TGFB3, and MYLK in M2 macrophages might be regulated by TEPs in BLCA and correlate with the tumor immune microenvironment and poor prognosis in BLCA. Puntoni et al. reported that overexpression of vascular endothelial growth factor can be detected in the majority of cancers including BLCA, ³⁹ , ⁴⁰ and can upregulate the expression of A2M. ⁴¹ Studies have confirmed that MYLK expression is upregulated in BLCA. ⁴² MicroRNA‐93‐5p was upregulated in BLCA tissues and cell lines, ⁴³ while microRNA‐93‐5p promoted TGFB3 expression. ³¹ We considered that platelets act as a bridge between these regulators in BLCA and macrophages because platelets have an integrated endocytic mechanism for the uptake and storage of a variety of proteins and RNA from cancer cells. ⁴⁴ The clear elucidation of the mechanism still needs to be further studied in future work.

With the rise and development of big‐data analysis and high‐throughput sequencing, an increasing number of predictive models based on molecular expression levels have been proposed, helping to guide personalized treatment and reduce the costs of treatment. Recently, the establishment of diagnosis or prognosis models from the particular aspect of biological processes, such as ferroptosis, ⁴⁵ , ⁴⁶ necrosis, ⁴⁷ lipid metabolism, ⁴⁸ , ⁴⁹ and acetylation, ⁵⁰ and the tumor immune microenvironment, ⁵¹ , ⁵² has attracted increasing interest. Similar to these studies, the established model in our study was based on platelets and M2 macrophages. To the best of our knowledge, this is the first time that the platelet and M2 macrophage‐related gene signature was constructed in malignant cancers.

The flaws of the present study should be stated. First, the retrospective nature of this study limited the application of MPRS, and a prospective, double‐blind, large‐scale, and multicenter clinical trial would be beneficial to clarify the clinical usefulness of MPRS. Second, even though we verified our findings in the cell experiments and clinical samples, the biological functions of A2M, TGFB3, and MYLK, especially in the interaction between platelets and M2 macrophages, are still unclear. More in vitro, in vivo, and ex vivo experiments should be undertaken shortly.

In conclusion, platelets could induce the polarization of M2 macrophages in BLCA. Based on this finding, we constructed a platelet and M2 macrophage‐related gene signature to evaluate the prognosis and immunotherapeutic sensitivity in BCLA. Our findings helped to shed novel insights into the interaction of platelets and M2 macrophages and provided a potential tool in clinical practice.

AUTHOR CONTRIBUTIONS

Bahaerguli Muhuitijiang: Data curation; formal analysis; validation; writing – original draft; writing – review and editing. Jiawei Zhou: Methodology; validation; writing – review and editing. Ranran Zhou: Formal analysis; writing – original draft. Zhiyong Zhang: Data curation; investigation; writing – review and editing. Guang Yan: Methodology; writing – review and editing. Zaosong Zheng: Methodology; writing – review and editing. Xiangbo Zeng: Resources; visualization. Yuanchao Zhu: Resources; visualization. Haowei Wu: Data curation; investigation. Ruxi Gao: Data curation. Tianhang Zhu: Resources. Xiaojun Shi: Project administration; supervision. Wanlong Tan: Funding acquisition; project administration; resources.

FUNDING INFORMATION

This research was funded by the National Natural Science Foundation of China (Nos. 82073162 and 82372867) and the Science and Technology Program of Guangzhou (No. 202002030482).

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

ETHICS STATEMENT

Approval of the research protocol by an institutional review board: The Ethics Committee of Nanfang Hospital of Southern Medical University according to the Declaration of Helsinki.

Informed consent: All patients provided informed consent for the retention and anonymous analysis of their tissues for research purposes.

Registry and registration no. of the study/trial: N/A.

Animal studies: N/A.

Supporting information

Appendix S1

Muhuitijiang B, Zhou J, Zhou R, et al. Development and experimental validation of an M2 macrophage and platelet‐associated gene signature to predict prognosis and immunotherapy sensitivity in bladder cancer. Cancer Sci. 2024;115:1417‐1432. doi: 10.1111/cas.16113

DATA AVAILABILITY STATEMENT

The original contributions presented in the study are included in the article/supporting information. Further inquiries can be directed to the corresponding authors upon reasonable request.