3.1 Characterization of cell–cell interactions and AA-related gene expression in the CRC TME

In this study, we utilized data from the GEO and TCGA databases, concentrating on three single-cell RNA sequencing datasets of CRC (GSM4994385, GSM5688708, and GSM5688711) to enhance our understanding of the tissue landscape associated with this disease, as previously illustrated in Fig. 1A. Our dataset comprised 33,694 cells from the TME, meticulously categorized into primary cell types, including T cells, B cells, myeloid cells, plasma cells, epithelial cells, and fibroblasts (Fig. 1B).

Fig. 1

Tumor microenvironment of CRC. A This study's overall design and data were obtained from the GEO dataset (GSM4994385, GSM5688708, GSM5688711); B cell type annotation of 33,694 cells was performed using Seurat's distribution-based random neighbor embedding (t-SNE) plots; C The expression of aggregated autophagy genes TUBA1B and HSP90AA1 in normal and tumor group; D cell chat analysis show variety correlation of TME cell types in T Cell, B Cell, Myeloids Cell, Plasma Cell, Epithelial Cell, and Fibroblasts Cell; E Heat map show that gene variety

Through a detailed analysis of cell–cell communication, we elucidated the intricate interactions among these TME cell types, revealing their complex interrelationships. Notably, our analysis uncovered significant associations between AA and various TME cell types, encompassing T cells, B cells, myeloid cells, plasma cells, epithelial cells, and fibroblasts (Fig. 1C). A heatmap further demonstrated the dynamic expression patterns of all AA-related genes throughout different stages of tumor progression, highlighting notable genes such as TUBA1A, TUBA1B, PARK7, TUBA4B, DYNC1I2, and HSP90AA1 (Fig. 1E). In addition, the AA-related genes HSP90AA1 and TUBA1B also exhibit widespread expression across different cell subpopulations (Fig. 1D).Our comprehensive findings suggest that elevated expression levels of TUBA1B, and HSP90AA1 may potentially accelerate the progression of CRC. These insights not only enhance our understanding of the molecular mechanisms underpinning CRC but also pave the way for targeted therapeutic interventions.

3.2 Role of AA-related genes TUBA1B and HSP90AA1 in CAFs activation and CRC progression

The AA-related gene has been implicated in the proliferation of CAFs, which play a crucial role in promoting vascular inflammation and tissue stiffness, thereby significantly influencing the progression of CRC. To further investigate this relationship, we isolated highly abundant fibroblasts from tumor tissues. Pseudotime analysis revealed that the AA genes TUBA1B and HSP90AA1 exhibited low expression levels at earlier stages, while DYNC1LI2 and HSP90AA1 were expressed at later stages (Fig. 2A). Utilizing non-negative matrix factorization (NMF) analysis, we identified four distinct clusters of CAFs within the tumor microenvironment (TME) based on AA gene expression: TUBA1B + CAF-C1 (n = 40), HSP90AA1 + CAF-C2 (n = 125), DYNC1I2 + CAF-C3 (n = 94), and non-aggregated CAF-C0 (Fig. 2B). Cell-Chat analysis revealed variable molecular signaling interactions among these CAF clusters, as well as interactions with epithelial cells and other fibroblasts (Fig. 2C). Notably, both TUBA1B and HSP90AA1 not only received signals but also initiated signaling pathways, suggesting a potent regulatory role in the TME (Fig. 2D). Furthermore, the expression of TUBA1B and HSP90AA1 was found to inhibit the function of macrophage migration inhibitory factor (MIF) and enhance the expression of the adhesion protein periostin within tumors (Fig. 2E).

Fig. 2

Aggregate autophagy gene set factors alter the characteristics of fibroblasts; A Pseudo-time analysis of the role of AA-related gene sets in fibroblasts (n = 746); B, C Four CAF clusters, DYNC1I2 + CAF-C6 (n = 94), HSP90AA1 + CAF-C2 (n = 125), TUBA1B + CAF-C1 (n = 40), and the percentage of HSP90AA1 + CAF-C2 was higher in tumors than in normal mucosa (p < 0.05); D heat map using GAS pathway analysis ( p < 0.05); F,G Transcription factor analysis between four clusters; H correlation of 6 TFS with inflammation demonstrated; The heatmap illustrates variations in the mean expression levels of essential genes involved in common signaling pathways across four distinct clusters. These pathways include essential components such as collagen, extracellular matrix (ECM), matrix metalloproteinases (MMPs), changing growth factor beta (TGF-β), neo-angiogenesis, contraction, RAS signaling, and pro-inflammatory mediators

Our investigation into gene regulatory networks highlighted significant differences in the expression of 28 transcription factors (TFs) across the four CAF clusters. In particular, the TUBA1B and HSP90AA1 cluster exhibited elevated levels of TFs such as KLF2, REL, STAT3, STAT1, and CREB3 (Fig. 2F). The identification of Pan-CAF signatures, as reported in previous studies, reinforced our findings, revealing a strong correlation between the TUBA1B and HSP90AA1 cluster and inflammatory cancer-associated fibroblasts (iCAFs-2). These iCAFs-2 subtypes are characterized by their secretion of various growth factors, including pro-inflammatory chemokines and cytokines such as CXCL2, CXCL1, CCL2, IL-6, and IL-7 (Fig. 2G, H). Given the critical role of CAFs in the tumor microenvironment, the observed association between TUBA1B and HSP90AA1 expression and CAF activation underscores the significance of TUBA1B in CRC progression. Specifically, TUBA1B and HSP90AA1 may drive CAFs to enhance the TME through the release of inflammatory factors, thereby significantly influencing tumor development and progression.

3.3 Impact of AA-related gene expression on T cell dynamics in CRC progression

Recent evidence highlights the critical presence of T cells as the most abundant and characteristic components of the TME in CRC. Intriguingly, the progression of CRC through AA is marked by alterations in T cell dynamics. We isolated high-abundance CD8 + T cells (n = 4753) and B cells (n = 2327) from CRC samples to investigate these changes. Pseudotime analysis indicated that genes TUBA1A, TUBA1B, and DYNC1H1 showed elevated expression in the late stages of CRC (Fig. 3A). Using NMF analysis, we delineated distinct clusters of CD8 + T cells based on AA gene expression, including TUBA1A + CD8 + T cells-C1 (n = 11), PARK7-CD8 + T cells-C2 (n = 7), TUBA1B + CD8 + T cells-C3 (n = 8), TUBB4B + CD8 + T cells-C4 (n = 208), and others (Fig. 3B). Cell-chat analysis revealed varying levels of ligand-receptor interactions among these clusters, particularly between epithelial cells and T cells (Fig. 3C). The TME may drive the plasticity of tumor subtypes, and we analyzed the communication network between cells through cellChat. The results showed that epithelial cells and T cells acted as communication hubs through VISFATIN signals (Fig. 3D).

Fig. 3

NMF clusters of AA regulatory factors in T cells, B cells and epithelial cells; A Pseudo-time analysis play the role of AA-related gene sets in T cells, B cells (CD8 + T Cell n = 4753,B cell n = 2327); B, C Role of ligand-receptor relationships in AA-related CD8 + T cells clusters; E, F Transcription factor analysis in AA-related CD8 + T cells clusters; G CD8 + T cells killing and exhaustion rate correlation is demonstrated; Heat map displaying the the expression of ICB in AA-related CD8 + T cells clusters

During CRC progression, interactions between AA TME cells and transcription factors such as JUND, JUNB, FOS, JUN, FOSB, RORA, BATF, IRF1, RBPJ, and RPDM1 were observed. These factors were up-regulated in TUBA1A + CD8 + T cells-C1 and PARK7-CD8 + T cells-C2 clusters, while in TUBA1B + B cells-C3, decreased transcription factor activity was linked to immune evasion (Fig. 3E). Additionally, TUBA1B + CD8 + T cells-C3 showed reduced association with T cell immune factors (Fig. 3G). Heatmap analysis further demonstrated significant variations in the expression of key immune checkpoint molecules across different clusters. Notably, TIGIT showed high expression in tumor-infiltrating lymphocytes (TILs), which interacts with receptors like CD155, D112, and CD113, enhancing T cell activation but dampening cytotoxicity. Conversely, TNFRSF15 (GITR) showed low expression in TUBA1A + CD8 + T cells-C1 and PARK7-CD8 + T cells-C2, promoting tumor development through co-stimulatory effects (Fig. 3F). Comparative analysis of AA-mediated interactions among CD8 + T cells revealed that the presence of TUBA1A + CD8 + T cells-C1 and PARK7-CD8 + T cells-C2 in the TME contributes to T cell exhaustion and immune escape, thereby accelerating tumor metastasis (Fig. 3G). In summary, the interplay of T cell exhaustion, and immune escape mechanisms within the TME significantly accelerates tumor development and metastasis in CRC.

3.4 Role of AA-related gene expression in macrophage profiling

During tumor progression, the expression of genes such as TUBA1C, HSP90AA1, TUBB4B, and DYNLL1 shows a marked increase in the later stages, while TUBA1B is expressed at earlier phases (Fig. 4A). A comprehensive analysis of AA-related gene sets across various macrophage types within the TME revealed six distinct macrophage clusters: TUBA1B + Mac-C1 (n = 72), DYNLL1 + Mac-C2 (n = 35), TUBA1C + Mac-C3 (n = 300), TUBB4B + Mac-C4 (n = 40), HSP90AA1-Mac-C5 (n = 63), and Non-Aggre-Mac-C6. These clusters displayed varying degrees of ligand-receptor interactions with epithelial cells (Fig. 4B, C).

Fig. 4

Identification of clustered AA-related gene in macrophages; A pseudo-time analysis of AA-related gene sets in macroautophagy (n = 1088); B,C ligand-receptor relationship roles among the AA-related gene in MAC clusters; D Metabolism anslysis in six MAC clusters

Further examination of the metabolic activities of these macrophage clusters demonstrated that the TUBA1B-Mac-C1 cluster exhibited elevated expression of critical metabolic pathways, including the pentose phosphate pathway, citrate cycle (TCA cycle), and pyruvate metabolism (Fig. 4D). This indicates that the aggregation of autophagy-associated macrophages significantly enhances TCA cycle energy metabolism, thereby facilitating tumor development. Such metabolic reprogramming not only meets the heightened energy demands of the tumor but also contributes to its aggressive growth and in vasiveness.

3.5 Expression profiling of TUBA1B and HSP90AA1 were involvement in metabolic pathways in CRC

Next, we analyzed the expression of TUBA1B and HSP90AA1 from GSE37182 and GSE41258. Our results indicated that the expression of AA-related genes, specifically TUBA1B and HSP90AA1, is significantly increased in COAD (Fig. 5A). We also observed that both genes exhibited favorable AUC values (Fig. 5B). Coincidentally, KEGG analysis revealed that these genes are associated with cellular development and metabolic processes, such as the TCA cycle and cell cycle regulation (Fig. 5C).

Fig. 5

High expression of TUBA1B and HSP90AA1 accelerates tumor progression. A Validation of TUBA1B and HSP90AA1 expression in GSE37182 and GSE41258. B Diagnostic efficacy analysis, with an AUC greater than 0.75 indicating high diagnostic efficacy. C KEGG enrichment analysis of differentially expressed genes in GSE37182

3.6 Influence of TUBA1B and HSP90AA1 on the exhaustion of CD8 T and CD4 cells in CRC

Subsequently, we conducted a further analysis of TUBA1B and HSP90AA1. Our findings revealed that the expression of TUBA1B is associated with the promotion of exhaustion in CD8 T and CD4 cells, with blue indicating a negative correlation and red indicating a positive correlation (Fig. 6A). Conversely, higher expression levels of HSP90AA1 were found to accelerate the functional exhaustion of CD8 T and CD4 cells, as well as impair the activity of neighboring cells, including macrophages and mast cells, showing a positive correlation in their expression (Fig. 6B).

Fig. 6

TUBA1B and HSP90AA1 induce immune dysfunction, accelerating immune evasion. A, B Immune correlation analysis of TUBA1B and HSP90AA1, with red indicating positive correlation and blue indicating negative correlation

3.7 Inhibition of TUBA1B and HSP90AA1 reduces tumor cell viability and alters inflammatory factors in CRC

Previous analyses have identified that the expression of TUBA1B and HSP90AA1 is a key factor accelerating the progression of CRC. To further study this, we employed siRNA to inhibit the expression of TUBA1B and TUBA1A. The CCK-8 assay confirmed that silencing TUBA1B and HSP90AA1 significantly reduced the viability of tumor cells (Fig. 7A). Additionally, we observed a specific decrease in the expression of TUBA1B and HSP90AA1 in CRC following transfection with siTUBA1B and siHSP90AA1, which was accompanied by a reduction in the expression of IL6, IL7, CXCL1, and CXCL2 (Fig. 7B, C). This suggests that chronic inflammatory factors are critical components involved in the complex TME. Inhibiting the expression of TUBA1B and HSP90AA1 may alleviate the complexities of the TME while concurrently suppressing tumor cell proliferation.

Fig. 7

Low expression of TUBA1B and HSP90AA1 inhibits tumor growth. A CCK8 assay validated the effect of transfecting si-TUBA1B and si-HSP90AA1 on tumor cell viability in Caco-2 Colo-205; B Expression levels of TUBA1B and HSP90AA1 in Caco-2 Colo-205; C Expression levels of inflammatory factors IL6, IL7, CXCL1, and CXCL2 in Caco-2 Colo-205. **P < 0.01, ***P < 0.001, ****P < 0.0001

3.8 AA-related gene clusters in CRC prognosis and immunotherapy response

To enhance the accuracy of prognostic predictions for CRC and to assess patient outcomes following immunotherapy. Our analysis revealed a downregulation in the expression of several gene clusters, including DYNC1I2 + CAF-C3, TUBB4B + CD8 + T cells-C4, TUBA1B + Mac-C1, and TUBA1C + Mac-C3, alongside an up-regulation of clusters such as TUBA1B + CD8 + T cells-C3, PARK7 + CD8 + T cells-C2, TUBB4B + Mac-C4, and HSP90AA1 + Mac-C5, suggesting significant variability within CRC (Fig. 8A).

Fig. 8

Illustrates the general prognosis and response to immunotherapy of AA cell types; A Predict the expression of different groups by GSVA score; B Single factor regression analysis that twelve aggregation autophagy cluster by TCGA and GEO; C Analysis of immunotherapy response (analysis of twelve cell clusters with response rate TCGA-COAD); D Global landscape of the tumor microenvironment.*P < 0.05,**P < 0.01,***P < 0.001,****P < 0.0001

Further analyses using Cox regression and bulk transcriptome data indicated that the expression of gene clusters DYNC1I2 + CAF-C3, HSP90AA1 + Mac-C5, TUBA1B + CD8 + T cells-C3, TUBA1B + Mac-C1, TUBA1C + Mac-C3, and TUBB4B + CD8 + T cells-C4 plays a crucial role in resisting immune escape and inhibiting tumor growth and metastasis (Fig. 8B). Additionally, our evaluation of immunotherapy responses, utilizing data from TCGA-COAD and GEO-GSE39582, demonstrated that AA-related genes, such as DYNC1I2 + CAF-C3, HSP90AA1 + CAF-C2, TUBA1A + CD8 + T cells-C1, TUBA1B + CAF-C1, TUBA1B + Mac-C1, and TUBA1C + Mac-C3, are implicated in promoting tumor cell proliferation and migration. Conversely, clusters such as TUBA1B + CD8 + T cells-C3, TUBB4B + CD8 + T cells-C4, and PARK7 + CD8 + T cells-C2 exhibit protective functions, contributing to increased cell killing and exhaustion, thereby facilitating effective anti-tumor responses (Fig. 8C, D). These findings elucidate the differential roles of autophagy-related gene clusters and enhance our understanding of their impact on CRC progression and response to immunotherapy.