Abstract

Background

   Colon adenocarcinoma (COAD) is increasingly prevalent among patients
   under 50 years old, and the 5-year survival rate for patients with
   metastasis is less than 20%. Identifying significant biomarkers and
   therapeutic targets is crucial. We investigated the expression of LRP2
   in COAD and its prognostic value utilizing single-cell sequencing and
   transcriptomics datasets, which was conducted preliminary validation at
   the patient samples and cellular levels as well.

Methods

   Based on differential gene expression of tumor samples and normal
   tissues in The Cancer Genome Atlas (TCGA), we performed consensus
   clustering, univariate and multivariate Cox regression analysis
   applying 1,234 mitochondrial metabolism-related genes (MMRGs) to
   identify some essential genes associated with poor prognosis in COAD
   patients. We validated survival outcome and biological function of the
   target gene leveraging single-cell sequencing and transcriptomics
   datasets from Gene Expression Omnibus (GEO), and evaluated the value of
   the target gene in the clinical pathology stage of COAD patients.
   Simultaneously, the expression levels of critical gene were detected in
   the diverse tissues of COAD by immunohistochemistry (IHC) staining.
   Transcriptomics data was continuously implemented to compare the
   discrepancy between the expression levels of the target gene and
   somatic mutation burden, inspecting the key pathways of the target gene
   by gene set enrichment analysis (GSEA) and examining its drug
   sensitivity synthetically in the CellMiner databases. The proliferative
   capacity augmented in LRP2-overexpressed colon cancer cells was
   determined employing cell counting kit-8 (CCK-8) and flow cytometry
   assays.

Results

   LRP2 served as a key mitochondrial metabolism-related gene was assessed
   clinical prognosis in COAD patients according to the TCGA database.
   High expression of LRP2 was prominently associated with poor prognosis
   in COAD patients (P < 0.05), which was validated by GEO databases, and
   the expression levels of LRP2 were positively related to clinical
   pathological stage simultaneously (P < 0.05). Some specific cell types
   were clustered and proliferation pathways were immensely enriched,
   which were correlated with LRP2 in two single-cell sequencing datasets.
   The mutation profiles displayed remarkable differences in two levels of
   LRP2, we also observed high expressions of LRP2 were immensely
   correlated with high tumor mutation burden (TMB) and unfavorable
   prognosis in these patients (P < 0.05). LRP2 was significantly enriched
   in multiple cancer proliferation-related pathways, and the noteworthy
   correlation between LRP2 and the sensitivity to various drugs was
   identified (P < 0.05). The expression levels of LRP2 were multifarious
   in different COAD patients based on IHC staining. LRP2-overpression
   could stimulate the proliferation capability of HCT116 and SW480 cell
   lines markedly (P < 0.05).

Conclusion

   The expression levels of LRP2 were intimately correlated with gene
   mutations, prognosis, pathological stage and the sensitivity to
   anticancer drugs in COAD. Augmented levels of LRP2 would manifest poor
   prognosis, which furnished novel insights for clinical diagnosis and
   treatment in COAD. LRP2 could extensively facilitate the proliferation
   ability of colon cell lines.

Supplementary Information

   The online version contains supplementary material available at
   10.1007/s12672-025-02568-2.

   Keywords: Colon cancer, Prognosis, LRP2, Multi-omics, Cell
   proliferation

Introduction

   The incidence and mortality rates of colorectal cancer ranked third
   among cancers and had become the second leading cause of death among
   cancer patients [[38]1]. Although the incidence and mortality rates of
   colorectal cancer were slowly declining, the proportion of patients
   under 50 years old was increasing, with a higher occurrence in the left
   colon [[39]2]. Approximately 20% of colorectal cancer patients would
   develop metastasis, and the five-year survival rate was less than 20%
   [[40]3]. Patients diagnosed with early-stage colon cancer could undergo
   surgical treatment, but there was a 40% recurrence rate in those with
   clinical stage III [[41]4]. Therefore, identifying meaningful
   biomarkers would provide a new perspective for the prognosis assessment
   and treatment of colon cancer patients.

   Tumor cells can alter their metabolic pathways to prompt energy
   production including metabolic reprogramming to maintain their stable
   survival and rapid proliferation capability [[42]5]. Mitochondria are
   crucial bioenergetic hubs for cellular metabolism and responsible for
   oxidative phosphorylation (OXPHOS), fatty acid oxidation (FAO),
   tricarboxylic acid (TCA) cycle, and electron transport chain (ETC)
   processes, all of which may promote the development and maintenance of
   malignant phenotypes in tumor cells [[43]6]. Dysregulation of
   mitochondrial metabolic pathways caused by mitochondrial
   metabolism-related genes may facilitate tumor onset, progression, and
   immune evasion [[44]7]. However, there was little systematic research
   that the relevant genes could stimulate mitochondrial metabolic
   disturbances and concentrate on the risk assessment and targeted
   therapy for colon carcinoma patients.

   Single-cell sequencing and transcriptomics analysis complemented novel
   insights in colorectal cancer research [[45]8, [46]9]. With the
   developments in single-cell technology, we could obtain a deeper
   understanding of the metabolic abnormalities, tumor microenvironment,
   and heterogeneity in tumor development and identify personalized
   therapeutic strategies corresponding with COAD.

   In the present study, it had revealed that mitochondrial metabolic
   pathways participated in accommodating colon cancer progression
   [[47]10, [48]11]. By comparing the differential expression of 1,234
   mitochondrial metabolism-related genes in colon cancer patients and
   their prognostic impact, LRP2 was identified as a key gene for
   assessing the prognosis of colon cancer patients. Simultaneously, we
   employed single-cell RNA sequencing and microarray datasets from the
   Gene Expression Omnibus (GEO) and mRNA-seq data from The Cancer Genome
   Atlas (TCGA) database to explore LRP2 affect COAD metabolic activities.

   LRP2 was an endocytic receptor that combined and internalized various
   ligands, such as vitamin-binding protein complexes [[49]12], lipid
   transport proteins, protease-protease inhibitor complexes, hormones,
   lipoproteins, and lipocalins [[50]13–[51]15]. Previous studies had
   illustrated that silencing LRP2 could inhibit the proliferation of
   thyroid cancer cells and reduce the survival rate of melanoma cells
   [[52]16, [53]17]. Therefore, it was speculated that LRP2 may be
   involved in regulating colon cancer occurrence and progression.
   Conducting multi-omics bioinformatics analysis of the pathway
   enrichment and drug sensitivity of LRP2 in COAD samples, these results
   unraveled regulatory mechanisms of LRP2 in the occurrence and
   development of colon cancer, and which was preliminarily validated
   performing immunohistochemical staining and COAD cell experiments. LRP2
   certainly enhanced the developmental activity of two colon carcinoma
   cell lines.

   This study provided a theoretical support for exploring the application
   of mitochondrial metabolism-related genes in the prevention, early
   diagnosis, and effective treatment of COAD. LRP2 may serve as a novel
   target for the progression assessment and personalized treatment of
   colon carcinoma patients.

Materials and methods

Data sources and preprocessing

   COAD mRNA-seq data and relevant clinical information (including 476
   colon cancer samples and 41 adjacent normal tissues) were obtained from
   TCGA database. The extensive datasets (single-cell sequencing data:
   [54]GSE132465 and [55]GSE179784; transcriptome data: [56]GSE39582 and
   [57]GSE38832) utilized in this study were downloaded from GEO
   databases, which was employed for verification analysis of the selected
   gene. In the scRNA-seq datasets, low-quality cells were excluded with
   mitochondrial genes ≤20%, UMI ≤500, and gene count ≥50,000. The more
   significant variance genes were proceeded for downstream analysis.
   Transcriptome databases were standardized using the log2 (X + 1) method
   in R 4.2.3, and differential analysis was performed using the “limma”
   package to compare colon cancer and normal samples with the cutoff
   criteria: false discovery rate (FDR) <0.05 and |log2 fold change (FC)|
   >1.

Screening of DEGs associated with mitochondrial metabolism-related and
cellular differentiation-related

   According to literature report, 1,234 mitochondrial metabolism-related
   genes (MMRGs) were obtained [[58]18]. Additionally, we distinguished
   2,264 cellular differentiation-related genes (CDRGs) from single-cell
   sequencing datasets. The intersection was taken with expression
   differences of CDRGs and MMRGs from COAD and adjacent normal tissues. A
   univariate Cox regression analysis was conducted on the related genes
   possessed significant prognosis assessment prospects.

Authentication of molecular subtype characteristics and construction of
prognostic model based on MMRGs and CDRGs

   Based on these prognostic genes, we inspected the molecular subtype
   utilizing an unsupervised consensus clustering approach employing the
   “ConsensusClusterPlus” package, which introduced a K-means clustering
   algorithm to identify robust clusters, setting the maximum number to
   10. The optimal number of clusters was assessed employing the Consensus
   Matrix and Cumulative Distribution Function (CDF) curves. To further
   detect critical prognostic signatures, the related genes were analyzed
   employing least absolute shrinkage and selector operation (LASSO)
   regression model by “glmnet” package. Subsequently, these identified
   prognostic genes were used to construct the related gene scores for
   each COAD sample. The related gene scores were calculated as the
   following formula: SMARCD3 × 0.106 + ACBD4 × 0.096 + MBOAT1 × 0.052
   + LRP2 × 0.154 + OSBPL1A × 0.060 + PNPLA7 × 0.167 + RAB6B × 0.045
   + CYP11A1 × 0.009 + TNFAIP8L3 × 0.047 + ALOX12B × 0.100 + CHAT × 0.056
   + SLCO1A2 × 0.069 + ACSL6 × − 0.025 + PPARGC1A × − 0.244. Additionally,
   the 1-, 3-, and 5-year prediction accuracy of the model was
   investigated by plotting the receiver operating characteristic (ROC)
   curves [[59]19]. The differences in Overall Survival (OS) both diverse
   clusters and models were assessed performing Kaplan Meier survival
   analysis in TCGA, [60]GSE39582, and [61]GSE38832 datasets.

Identification of key target signature and gene ontology (GO) enrichment
analysis

   To further investigate the decisive genes in the predictive model, we
   conducted a multivariate Cox regression analysis. The target biomarker
   upregulating in COAD and preforming a high hazard ratio (HR) would be
   percolated. We validated its significance of clinical prognosis
   evaluation by incorporating survival information and clinical
   pathological features from multiple datasets. Furthermore, the
   single-cell RNA data associated with critical signature was
   log-normalized and dimensionality was lessened conducting Principal
   component analysis (PCA), followed by t‐distributed stochastic neighbor
   embedding (t‐SNE) and uniform manifold approximation and projection
   (UMAP) for visualization [[62]20]. We manually annotated Cell types to
   define distinct cellular populations, focusing on variable epithelial
   and immune cells. With | log2 FC |> 0.585 and FDR <0.05, differential
   analysis was performed on different differentiation states. The
   filtered cell differentiation‐related genes relevant with the selective
   biomarker were implemented GO enrichment analysis to further comprehend
   the biological significance.

GSEA analysis

   Gene Set Enrichment Analysis (GSEA) was utilized to identify key
   pathways involving the target gene in the occurrence and development of
   colon cancer [[63]21]. GSEA analysis was carried out on pathways
   annotated in Gene Ontology (GO), Kyoto Encyclopedia of Genes and
   Genomes (KEGG), and HALLMARK. Differential analysis of COAD samples
   with multifarious risk scores and expression levels of critical genes
   in TCGA were performed using the “limma” package (FDR <0.05 and |log2
   FC| >1). The essential pathways related to prognostic signature were
   sorted by their normalized enrichment scores, selecting pathways with
   p < 0.05 for further analysis.

Mutation analysis

   Somatic mutation data for COAD was acquired from TCGA database. The
   “maftools” package was used to create waterfall plots to illustrate the
   mutation landscape for COAD patients of multitudinous expression levels
   of target gene, calculating the mutation burden score for each sample.

Drug sensitivity analysis

   Integration and analysis of molecular and pharmacological data from the
   NCI-60 tumor cell lines were conducted in the CellMiner database
   ([64]www.discover.nci.nih.gov) [[65]22]. Clinical trial and
   FDA-approved drugs were sieved, and we calculated the Pearson
   correlation coefficient between the expression of the selected
   prognostic genes and these drugs, with P < 0.05 as the threshold for
   selection of results utilizing “impute” and “limma” packages.

Immunohistochemistry (IHC) staining

   Target gene was experimentally verified by IHC staining. We collected
   diverse colon cancer tissue samples to perform IHC staining from
   Tianjin Medical University Cancer Institute and Hospital. Ethical
   approval was received from the hospital ethics committee and all
   patients signed informed consent forms. Two independent pathologists
   evaluated and scored the immunohistochemical staining. Staining
   intensity was graded from 0 to 3, with 0 implicating no staining, 1
   representing weak staining, 2 implicating moderate staining, and 3
   indicating strong staining, separately.

Cell culturing and lentiviral transfection

   Human colon cancer HCT116 and SW480 cell lines were purchased from ATCC
   and cultivated in DMEM medium supplemented with 10% FBS and 1%
   penicillin–streptomycin under the condition of 37 °C in a humidified
   atmosphere (≥95%) with 5% CO[2]. The negative control (NC) group and
   overexpression (LRP2-OE) group were established for constructing empty
   and LRP2 overexpression lentiviral vectors (Beyotime, China),
   respectively. The complete culture medium was replaced after 12 h, and
   the cells were cultured about 72 h continuously until they could be
   applied for subsequent experiments.

Real-time quantitative PCR (RT-qPCR)

   RT-qPCR was performed to detect the expression of two selected genes:
   LRP2 and the internal control gene glyceraldehyde-3-phosphate
   dehydrogenase (GAPDH). RNA was extracted from HCT116 and SW480 cell
   lines and then cDNA was synthesized using a reverse transcription kit
   (Takara, Japan). The relative gene expression of LRP2 was assessed
   based on the 2 − ΔΔCT method. Primer sequences were as follows: GAPDH
   Forward: 5′-TAGAGCGGCCGCCATGTT-3′, Reverse:
   5′-CGCCCAATACGACCAAATCAGA-3′; LRP2 Forward: 5′-TTTTGGCCACTAGGAGCTGG-3′,
   Reverse: 5′-CGCACTGTCACATTCTTGGC-3′.

Western blot

   The lysate was separated using SDS-PAGE and then transferred to PVDF
   membranes. The membranes were washed five times with 1 × TBST, which
   were then blocked by 5% skimmed milk. They were incubated with primary
   antibodies for two hours, with antibodies including LRP2 (1:1000,
   ab313741, Abcam) and GAPDH (1:1000, ab181602, Abcam). The membranes
   were washed five times with 1 × TBST again and transferred to secondary
   antibodies for overnight incubation. Protein detection was performed
   using an ECL reagent (Beyotime, China) on the Tanon-5000 system and
   gray values were examined by ImageJ software.

CCK-8 assay for cell viability

   Cells in the logarithmic growth phase were selected and seeded into a
   96-well plate at a density of 2,000 cells/well. After being cultured
   for 0, 24, 48, 72, and 96 h, cell lines were added by the CCK-8 reagent
   and incubated at 37 °C with 5% CO[2] for 2 h. The optical density (OD)
   was measured at a wavelength of 450 nm.

Flow cytometry

   Logarithmic phase cells were seeded into 6-well plates at a density of
   5 × 10^5 cells/well and continued to culture about 48 h. The cell
   suspension was collected and subsequently treated with APC-EdU and
   FITC-Ki67 for 30 min in the dark at room temperature. Flow cytometry
   was used to detect cell proliferation activity.

Results

Identification of mitochondrial metabolism-related and cellular
differentiation-related clusters in COAD

   This study was performed to identify mitochondrial metabolism-related
   genes associated with colon cancer prognosis and to further investigate
   their roles in the progress of COAD. A comparison of differential
   expression in colon cancer samples and adjacent tissues yielded 8,987
   differential genes (including 1,467 upregulated genes and 7,520
   downregulated genes) (Fig. S1 A). The intersection of 1,234
   mitochondrial metabolism-related genes with the 8,987 differential
   genes and 2266 cellular differentiation-related genes produced 423
   common genes (Fig. [66]1A). Univariate Cox analysis identified 31 genes
   significantly associated with prognosis (P < 0.05) (Fig. [67]1B). We
   conducted consensus clustering analysis in 476 COAD patients from the
   TCGA database, through clusters (K) from 2 to 10. When k = 2, the
   optimal grouping was strongly determined on the expression matrix of 31
   related genes in the training cohorts (TCGA-COAD) (Fig. [68]1C, D and
   S1B, C). Compared with cluster 2, the shorter survival time of cluster
   1 indicated a worse outcome (P < 0.05) (Fig. [69]1E). The same result
   was further validated in the [70]GSE39582 dataset (P < 0.05)
   (Fig. [71]1F and S1D). The intragroup correlation was fierce at k = 2
   and cluster 1 owned a poorer prognosis in the [72]GSE38832 database
   (Fig. [73]1G, H).

Fig. 1.

   [74]Fig. 1
   [75]Open in a new tab

   Screening mitochondrial metabolism-related genes as assured prognostic
   biomarkers in COAD. A Comparison of MMRGs; B Examination of 31 critical
   genes using univariate Cox analysis; C The cumulative distribution
   function curve of MMRGs clustering; D Consensus matrices of the TCGA
   dataset for k = 2; E Kaplan–Meier (KM) curves of COAD patients in two
   different clustering groups; F Outcome of COAD patients between cluster
   1 and cluster 2 in [76]GSE39582; G The consensus matrix of the k-means
   clustering (k = 2) in [77]GSE38832; H Validation for the overall
   survival (OS) of two different COAD cohorts in [78]GSE38832

Estimation of the potential elements

   To screen critical signatures, we operated LASSO logistic regression to
   identify 14 biomarkers, including SMARCD3, ACBD4, MBOAT1, LRP2,
   OSBPL1A, PNPLA7, RAB6B, CYP11A1, TNFAIP8L3, ALOX12B, CHAT, SLCO1A2,
   ACSL6, PPARGC1A (Fig. [79]2A, B). The expression heatmap of 14 genes
   between tumor samples and normal tissues represented a prominent
   distinction (Fig. S1E). It divided different colon cancer patients into
   two groups. Compared with the low-risk group, more deaths and shorter
   survival time of the COAD patients with the high risk were observed
   simultaneously in TCGA (P < 0.0001), [80]GSE39582 (P = 0.036), and
   [81]GSE38832 (P = 0.013) datasets (Fig. [82]2C–E). Next, the predictive
   value of the prognostic model was evaluated executing ROC curves and
   the AUC figures of the construction model at 1, 3, and 5 years were
   0.71, 0.722, and 0.757, respectively (Fig. [83]2F). We continued to
   apply GSEA analysis for investigating GO, KEGG and HALLMARK pathways to
   deeper understand the significance of these potential biomarkers in
   colon carcinoma occurrence and development. Some proliferation-related
   pathways were extensively enriched, including extracellular matrix
   structural constituent, GTPase activator activity, growth factor
   activity, growth cone, nuclear periphery, spindle, BMP signaling
   pathways, Wnt signaling pathways, calcium signaling pathway, Rap1
   signaling pathway, cGMP-PKG signaling pathway, PI3K-Akt signaling
   pathway, MAPK signaling pathway, FoxO signaling pathway, myogenesis,
   hedgehog signaling, TNFA via NFκB signaling, G2M checkpoint and DNA
   repair, etc. (Fig. [84]2G, H and S1 F).

Fig. 2.

   [85]Fig. 2
   [86]Open in a new tab

   Construction of MMRGs prognostic model for COAD patients. A Least
   absolute shrinkage and optimal coefficient profiles of the selective
   gene sets; B Determination of the number of factors by the LASSO
   regression analysis; C–E Kaplan–Meier curves between high-risk score
   and low-risk score cohorts in the TCGA (C), [87]GSE39582 (D), and
   [88]GSE38832 (E) sets; F ROC curves indicate the 1-, 3-, 5-years
   survival of COAD patients in TCGA; G GO results related to various risk
   score groups employing GSEA analysis; H KEGG results employing GSEA
   analysis connected with risk scores

Evaluation between LRP2 expression level and prognosis of COAD patients

   We preformed univariate Cox analysis aiming at 14 biomarkers obtained
   above again, and 9 signatures with crucially prognostic values were
   sieved from 379 colonic adenocarcinoma samples based on TCGA database
   (P < 0.05) (Fig. [89]3A). The multivariate Cox analysis displayed that
   both LRP2 (OR = 1.219, 95% CI: 1.069–1.389, P = 0.00307) and PPARGC1A
   (OR = 0.795, 95% CI: 0.707–0.893, P = 0.00012) might be utilized as
   independent factors of prognosis for COAD (Fig. [90]3B). Afterwards, 3
   genes impacting patient survival were identified employing a univariate
   Cox regression analysis in [91]GSE38832 (P < 0.05) (Fig. S1G).
   Interestingly, all of these genes including LRP2 (OR = 1.605, 95% CI:
   1.045–2.465, P = 0.03078), CYP11A1 (OR = 1.450, 95% CI: 1.015–2.071,
   P = 0.04097) and PPARGC1A (OR = 1.219, 95% CI: 1.069–1.389, P =
   0.01036) still behaved specific prognostic values in the multivariate
   Cox regression analysis (P < 0.05) (Fig. [92]3C). Compared to adjacent
   normal tissues, LRP2 was markedly upregulated, while the expression
   levels of CYP11A1 and PPARGC1A were fairly reduced in colon carcinoma
   samples (P < 0.0001) (Fig. [93]3D). CYP11A1 represented a positive
   association with an increased risk of colon cancer in [94]GSE38832, but
   the same result was seemingly not obtained by multifactorial analysis
   in TCGA dataset. Moreover, the expression level of CYP11A1 was lessened
   in tumor samples, which signified it may not be an ideal prognostic
   marker. We continued to detect the prognostic evaluation roles of LRP2
   and PPARGC1A in colon cancer patients.

Fig. 3.

   [95]Fig. 3
   [96]Open in a new tab

   LRP2 was sieved from prognostic biomarkers of COAD patients. A 9
   potential biomarkers were further selected from 14 biomarkers (P <
   0.05); B The results of multivariate cox analysis of 9 markers in TCGA;
   C The results of multivariate cox analysis of 3 markers in
   [97]GEO38832; D Expression levels of 3 candidate genes between tumor
   samples and adjacent normal tissues; E–G KM curves of high and low LRP2
   expression in TCGA (E), [98]GSE39582 (F), and [99]GSE38832 (G). H–K The
   correlation of expression levels of LRP2 in COAD patients with pTNM
   stage (H), T stage (I), N stage (J), and M stage (K) in TCGA cohort.
   ****P < 0.0001

   From TCGA database (after excluding missing samples, 448 colon
   carcinoma samples remained), survival curve revealed that colon cancer
   patients with higher LRP2 expression had poorer prognoses (P = 0.0017)
   (Fig. [100]3E). In the [101]GSE39582 dataset, colon cancer patients
   with LRP2 upregulation also exhibited significantly reduced survival
   time (P = 0.046) (Fig. [102]3F). In the [103]GSE38832 database, LRP2
   higher expression in colon cancer patients was also associated with
   momentously worse prognosis (P = 0.036) (Fig. [104]3G). The ROC plot
   confirmed LRP2 owned an independent appraisal function, and the AUC
   values of LRP2 at 1, 2, and 3 years were 0.617, 0.628, and 0.648,
   respectively (Fig. S1H). Interestingly, PPARGC1A reflected an immensely
   opposite tendency that the incremental set could lengthen the survival
   rate in the TCGA (P = 0.00042), [105]GSE39582 (P = 0.23), and
   [106]GSE38832 (P = 0.023) (Fig. S2 A–C). It was reasonably speculated
   that two genes may be playing crucial roles in the progression of COAD,
   while we decided to pay attention to further investigate the relevant
   significance of LRP2 in regulating tumor status and prognosis
   assessment in COAD, due to its significantly elevated expression and
   high HR value in colon cancer tissues. To validate the prognostic value
   of LRP2 in colon cancer patients, we conducted a comprehensive analysis
   according to clinical features across various pathologic stages, N
   staging, M staging, T staging, gender, age, and CEA level in the
   disparate expression of LRP2 cohorts (Table S1).

   LRP2 expression levels were strikingly associated with the pathologic
   stage (P < 0.05) (Fig. [107]3H), T stage (P < 0.05) (Fig. [108]3I), N
   stage (P < 0.05) (Fig. [109]3J), and M stage (P < 0.05) (Fig. [110]3K).
   Kaplan–Meier survival analyses revealed that the higher LRP2 expression
   group had worse overall survival compared to the lower group in
   different strata of clinical characteristics including stage II-IV,
   T2–T4, M0 or M1, and patients younger (<65 years) or older (≥65 years)
   (Fig. S2D–G). These results manifested that LRP2 possessed a prominent
   robustness for prognosis assessment of COAD.

scRNA-seq analysis unveiled the components of diverse cells and GO pathways
contacted with LRP2 in COAD

   After processing and refining the data, multitudinous expressions of
   marker genes were conducted to distinguish various cellular groupings
   in [111]GSE132465 (Fig. [112]4A). After the tSEN and UMAP
   non-dimensionality carrying out strict quality control and filtration,
   a total of 63,689 cells were obtained. We identified 6 distinct cell
   subsets utilizing UMAP dimensionality reduction in [113]GSE132465,
   including epithelial cell (18,798), myeloid cell (6788), B cell (8861),
   T cell (23,095), mast cell (206) and stromal cells (5941)
   (Fig. [114]4B). DEGs between high and low expression levels of LRP2
   were displayed in the form of a volcano (Fig. [115]4C). Based on the
   annotation of known classical markers, 12, 118 cells were acquired in
   [116]GSE179784 and 6 diverse cell types were categorized, which mainly
   concerned on epithelial cell (8953), fibroblasts (273), myeloid cell
   (204), endothelial cells (534), B cell (403) and T cell (1751)
   (Fig. [117]4D, E). We observed that 208 down-regulated and 75
   up-regulated genes were identified in [118]GSE179784 (Fig. [119]4F).

Fig. 4.

   [120]Fig. 4
   [121]Open in a new tab

   Identification of cell subtypes and DEGs associated with LRP2 in the
   scRNA-seq data. A Eleven cell clusters were obtained applying the UMAP
   algorithm in [122]GSE132465; B Six primary cell types were annotated by
   the given cell marker in [123]GSE132465; C Differential expression
   genes of both higher and lower LRP2 expression groups in
   [124]GSE132465; D Ten cell clusters were acquired using UMAP algorithm
   in [125]GSE179784; E The annotation of 6 prominent cell types in
   [126]GSE179784; F Volcano plot of diverse LRP2 levels in [127]GSE179784

   In the GO enrichment analysis through these DEGs, some
   proliferation-related pathways were prominently up-regulated in
   [128]GSE132465, including positive regulation of cell proliferation,
   cell cycle, mitochondrial protein-containing complex, organelle outer
   membrane, mitochondrial translation, and regulation of wound healing
   (Fig. [129]5A). At the same time, we employed differentially expressed
   genes to conduct GO analysis in [130]GSE179784. Concerning the BP,
   positive regulation of cell proliferation, cell cycle process, cell
   cycle, and sterol homeostasis contributed to tumor progression.
   Regarding CC, p granule, myofilament, GABA-ergic synapse, and cell
   division site were abundantly enriched. In terms of MF, carbonate
   dehydratase activity, myristoyl-CoA hydrolase activity, and GTPase
   activating protein binding were observed, separately (Fig. [131]5B).
   Therefore, we conjectured LRP2 might trigger colon carcinoma growth
   capacity according to the pathways associated with tumor germination.

Fig. 5.

   [132]Fig. 5
   [133]Open in a new tab

   GO analysis at single-cell sequencing level and TMB landscapes of LRP2.
   A–B GO annotation correlated with LRP2 in [134]GSE132465 (A) and
   [135]GSE179784 (B) datasets. C The waterfall plot of TMB in the high
   LRP2 expression cohort; D The landscapes portrayed the top 15 highly
   mutated genes in the low LRP2 expression group; E The KM curve of high
   and low TMB; F–G The KM plot related with LRP2 in high (F) and low (G)
   tumor mutation burden cohorts

Mutation profiles of COAD patients with different LRP2 expression levels

   Genomic sequencing had fundamentally changed the understanding of
   somatic mutations in cancer, which would provide a detailed map of gene
   mutations and the processes driving cancer mutations [[136]23]. Among
   the top 15 most frequently mutated genes, APC, TP53, TTN, KRAS, PIK3
   CA, SYNE1, MUC16, OBSCN, RYR2, FAT4, and DNAH5 were shared
   high-frequency mutated genes across two groups, but the mutation
   frequencies of APC, TP53, KRAS, LRP1B, PCLO, FBXW7, and RYR3 were
   significantly increased in the high expression group (Fig. [137]5C, D.
   Surprisingly, the higher TMB tended to behave a worse OS compared with
   the lower TMB (P = 0.039) (Fig. [138]5E). Moreover, low expression of
   LRP2 group would reveal a better outcome compared with high expression
   group in high TMB subgroups (P = 0.00072) (Fig. [139]5F). The
   consistent results were represented in the low TMB cohorts, while
   without a statistically sufficient difference (P = 0.14)
   (Fig. [140]5G). The combination of LRP2 and TMB would represent nicely
   predictive prognosis values in COAD patients as well.

GSEA enrichment analysis of LRP2-related pathways in COAD

   Differential analysis was implemented in colon cancer samples with
   different LRP2 expressions, which produced the identification of 430
   DEGs (including 421 upregulated genes and 9 downregulated genes) (Fig.
   S3 A). GSEA was used to conduct GO, KEGG, and HALLMARK analyses to
   inspect the regulatory pathways of LRP2 in COAD. The results of the GO
   analysis indicated that LRP2 primarily participated in biological
   process (BP) such as epithelial to mesenchymal transition,
   developmental cell growth, BMP, Wnt, TGF-β, and notch signaling
   pathways. In cellular component (CC), it was mainly enriched in DNA
   packaging complex, growth cone, cullin-RING ubiquitin ligase complex,
   adherens junction, and mitotic spindle. In molecular function (MF), it
   was predominantly associated with extracellular matrix structural
   constituent, ATP-dependent activity, acting on DNA, helicase activity,
   cytoskeletal motor activity, and ATP hydrolysis activity
   (Fig. [141]6A).

Fig. 6.

   [142]Fig. 6
   [143]Open in a new tab

   Analysis of GSEA, drug susceptibility, and immunohistochemical staining
   in different LRP2 expression groups. A GO pathways of GSEA for LRP2; B
   KEGG pathways of GSEA for LRP2; C Enrichment score of Wnt signaling
   pathway; D Drug susceptibility results; E Survival probability of
   clinical samples; F Immunohistochemical staining from collecting
   patient tissues

   According to KEGG pathway enrichment analysis, LRP2 may mainly be
   involved in the focal adhesion, transcriptional misregulation in
   cancer, Wnt, PI3K-Akt, mTOR, MAPK, cell cycle, and Rap1 signaling
   pathways (Fig. [144]6B). In the HALLMARK analysis, the signaling
   pathways were primarily enriched in Hedgehog signaling, angiogenesis,
   mitotic spindle, G2M checkpoint, TNFA/NFκB, E2 F targets, myogenesis
   and apical junction (Fig. S3B). Notably, both GO and HALLMARK
   enrichment analyses showed a high enrichment score for the Wnt
   signaling pathway and other proliferative pathways, and the Wnt
   enrichment plot was displayed in Fig. [145]6C.

   Collectively, LRP2 certainly generated the energy to trigger the
   proliferative capability of COAD. Afterwards, we speculated that LRP2
   may affect the progression of colon cancer modulating the Wnt pathway
   and further impact the prognosis of colon cancer patients.

Drug sensitivity analysis and immunohistochemical staining in COAD

   Chemotherapy is one of the most significant treatment methods for colon
   cancer, and drug resistance remains a key research focus. Using the
   CellMiner database, we analyzed the complex relationship between LRP2
   expression and cancer treatment drugs, revealing that high LRP2
   expression was positively correlated with anti-cancer drugs, namely
   Tipifarnib, Vemurafenib, SB-590885, PLX-4720, Dabrafenib, Hypothemycin,
   PD-98059, and Haloperidol, and which indicated that high LRP2
   expression generated colon carcinoma cells more sensitivities to these
   medications (Fig. [146]6D and S3 C–J). Conversely, the expression of
   LRP2 was negatively contacted with Celecoxib, suggesting a potential
   resistance with LRP2 expression arising (Fig. [147]6D and S3 K).

   Additionally, immunohistochemistry analysis manifested the expression
   levels of LRP2 in colon cancer tissues were classified as 4 stages
   among different COAD patients, including absent, weak, moderate and
   strong (Fig. [148]6F). After integrating the clinical information about
   disparate patients, we found diminished expression of LRP2 could
   lengthen survival time contrasted with those patients of higher LRP2
   expression (P = 0.014) (Fig. [149]6E). These results suggested that
   LRP2 would provide a striking insight for the clinical diagnosis and
   treatment of colon cancer.

LRP2 overexpression facilitated proliferative capacity of HCT116 and SW480
cell lines

   Bioinformatics analysis unveiled that LRP2 was principally associated
   with multiple pathways promoting tumor cell proliferation, thereby
   impacting colon cancer progression. To verify the effect of LRP2 on the
   biological behavior of colon cancer cells, LRP2-overexpressing cell
   lines containing HCT116OE and SW480OE were constructed using lentiviral
   vectors in the colon cancer cell lines between HCT116 and SW480,
   respectively. Both qPCR detection and Western Blot assay results showed
   that the expression levels of LRP2 were significantly increased in the
   overexpression colon carcinoma cell lines (P < 0.01) (Fig. [150]7A, B).

Fig. 7.

   [151]Fig. 7
   [152]Open in a new tab

   LRP2 stimulated proliferative capability of HCT116 and SW480. A qPCR
   detection of LRP2 mRNA levels in HCT116 and SW480 overexpressed cell
   lines; B Western Blot estimation in two overexpression colon cancer
   cells; C CCK8 assay examined the proliferation ability of LRP2-OE colon
   cell lines; D–E EDU assay in overexpressed SW480 (D) and HCT116 (E);
   F–G Ki67 values in LRP2-overexpression SW480 (F) and HCT116 (G). **:
   P < 0.01; ***: P < 0.001; ****: P < 0.0001

   Subsequently, CCK-8 and flow cytometry assays were utilized to examine
   the effect of LRP2 overexpression on the proliferative capacity of
   HCT116 and SW480 cells. In the CCK-8 cell proliferation experiments,
   the OD values of HCT116 and SW480 cells with LRP2 overexpression were
   crucially upregulated (P < 0.0001) (Fig. [153]7C). Flow cytometry
   results also demonstrated that APC-EdU values extensively arose with
   the LRP2 overexpression in SW480 and HCT116 cell lines (P < 0.0001)
   (Fig. [154]7D, E). Consistently, the overexpression of LRP2 upgraded
   the percentage of FITC-Ki67 relative to the control of both two colon
   cancer cells (Fig. [155]7F, G). These results indicated that LRP2
   overexpression might enforce the proliferation capacity of colon cancer
   cells.

Discussion

   Colon cancer had become the third most common cancer globally and the
   second leading cause of death among cancer patients [[156]24, [157]25].
   In recent decades, the incidence of colorectal cancer in individual
   under 50 had significantly increased, with a considerable proportion of
   colorectal cancer patients being diagnosed with colon cancer [[158]25,
   [159]26]. Approximately 15–30% of colon cancer patients experienced
   metastasis, and those with metastatic colon cancer had a poor
   prognosis, with a five-year survival rate being only 15% [[160]27,
   [161]28]. Colon cancer continuously poses a threat to human health,
   particularly for younger populations, with its rising incidence and
   mortality rates requiring urgent attention. It was crucial to explore a
   biomarker that could monitor the progression of colon cancer and serve
   as a potential therapeutic target.

   Metabolic reprogramming is a hallmark of tumor development. It not only
   represented the functional state of tumors but also played an assured
   role in the mechanisms underlying various cancers [[162]29]. The
   Warburg effect described the abnormal aerobic glycolysis of malignant
   cells, which was the first discovery linking tumor biology with cell
   metabolism [[163]30]. In recent years, the importance of
   mitochondrial-associated activities in the occurrence and development
   of tumors had gained increasing attentions [[164]31].

   We selected 31 mitochondria-related genes possessing clinical
   prognostic value by univariate Cox regression methods based on
   multi-omics data. These 31 prognostic-related genes were preformed
   consensus clustering analysis in TCGA and GEO datasets, and we observed
   that optimal clustering value occurred at K = 2. Survival analysis
   across the three datasets revealed that cluster 1 had a worse survival
   probability compared to cluster 2. Meanwhile, we constructed a risk
   assessment model through Lasso regression analysis and 14 important
   genes were identified, namely SMARCD3, ACBD4, MBOAT1, LRP2, OSBPL1A,
   PNPLA7, RAB6B, CYP11A1, TNFAIP8L3, ALOX12B, CHAT, SLCO1A2, ACSL6, and
   PPARGC1A. The clinical prognosis assessment potentiality of COAD was
   further validated in utilizing two external validation cohorts. GSEA
   analysis of this model exerted partial biological effects on colon
   tumor. A significant enrichment of pathways related with tumor
   proliferation were observed, indicating that these gene sets might be
   helpful in facilitating carcinogenesis and growth. Undoubtedly, the Wnt
   signaling pathway was likewise noteworthy enrichment between GO and
   KEGG analysis. To further detect the crucial gene that occupied a
   well-established position in this gene set, we conducted multivariate
   Cox regression analysis and inferred that LRP2 played a striking role
   in the progression of colon cancer.

   LRP2 was an important gene associated with mitochondrial metabolism,
   which was a multi-ligand endocytic receptor expressed in the apical
   membranes of various epithelial cells, including alveolar cells
   [[165]32, [166]33], breast [[167]34], thyroid [[168]35], colon,
   prostate [[169]36], and gallbladder [[170]37] epithelial cells. LRP2
   was a specific homodimer, which accelerated recycling to the cell
   surface from endosomes and enabled ligand to deliver and dissociate
   efficiently [[171]15, [172]38]. This study confirmed the prognostic
   assessment value of mitochondrial metabolism-related genes in COAD
   through TCGA and GEO databases, discerning LRP2 as a crucial gene that
   may make a significant impact on the prognosis of COAD at the
   transcriptomic level. The study found that the expression of LRP2 was
   associated with the pathological staging of colon cancer, and high
   levels of LRP2 indicated a poor prognosis, suggesting that LRP2 may
   play a critical role in the development of colon cancer. In-depth
   research on LRP2 could assist to elucidate the pathogenesis of COAD.

   With the advancements of single-cell technology, we obtained novel
   insights about tumor microenvironment and heterogeneity in CRC
   [[173]39, [174]40]. In this study, to further investigate the prominent
   value of LRP2 in COAD, a total of 63,689 cells were acquired in the
   [175]GSE132465 dataset, which was primarily classified into six
   categories based on marker gene sets. In the [176]GSE179784, 6 major
   cell types were clustered similarly. Interestingly, the heterogeneity
   related to LRP2 was obvious in various cell types, with T cells
   displaying vitally assembled in [177]GSE132465 and [178]GSE179784. It
   was reported that single-track mitochondrial transportation occurred
   from T cells to neighboring cancer cells, checking the nanotube-based
   mutual effect between cancer and immune [[179]41]. The mitochondria
   provided the essential energies for T cell activation and implicated it
   could further affect T cell dysfunction [[180]42]. Simultaneously, it
   was recently observed cancer cells were metabolically empowered in the
   process of near-unidirectional mitochondrial transfer from T cells to
   tumor cells while immune cells were exhausted, supplying novel
   perspectives about immune evasion and tumor-T cell interaction
   [[181]43]. In addition, an adequate number of patients were studied by
   mitochondrion transfer score, illustrating that mitochondrion transfer
   may produce a negative impact on patient survival by enhancing the
   proliferative ability of cancer cells [[182]43]. Synthesizing these
   researches above and our study, it was reasonably supposed that
   mitochondrial function was positive with the dysfunction of T cells,
   resulting in the crumble of the cancer immune supervisory system.
   Collectively, LRP2, T cells, and mitochondria were inevitably
   connected, which required deeper research work to investigate.

   We observed the biological roles of differential expression genes
   related with LRP2 by GO analysis from two single-cell sequencing
   datasets. These results exhibited that LRP2 might positively
   participate in mechanism regulation of cell proliferation, cell cycle,
   and GTPase activating protein binding, unraveling the LRP2 might play
   potential roles in accelerating colon tumor germination and
   development.

   GSEA pathway enrichment analysis indicated that LRP2 was significantly
   enriched in pathways related to BMP, Wnt, TGF-β, Notch, TOR, Rap1,
   cGMP-PKG, PI3K-Akt, MAPK, cAMP, mTOR, Hedgehog, angiogenesis, mitotic
   spindle, and NFκB/TNFA in colon cancer samples. The regulatory
   adaptation of cancer cells has been associated with various molecules
   metabolic activity, such mTOR, AKT, Wnt, MAPK, and Notch
   [[183]44–[184]47].

   These pathways were related to tumor progression regulation, where both
   GO and KEGG analyses showed the enrichment of the Wnt signaling
   pathway, indicating that LRP2 probably prompted tumor occurrence and
   germination, consistent with the findings that higher LRP2 level
   implicated a perishing prognosis in colon cancer. Interestingly, the
   Wnt signaling pathway had finished a noteworthy enrichment in the
   prognostic model, which underlined that LRP2 certainly occupies a
   commanding position synthesizing multivariate Cox regression analysis
   in the TCGA and GEO databases again. Drug sensitivity analysis
   represented high sensitivity to certain chemotherapy drugs with high
   expression of LRP2. These results suggested that LRP2 may be connected
   with the progression of colon cancer and could guide potential
   directions for cancer treatment.

   In this immunohistochemical staining, the role of LRP2 was clearly
   demonstrated in dissimilar COAD tissue samples. The results manifested
   that high expression of LRP2 was a risk factor associated with poor
   prognosis in COAD patients. CCK-8 and flow cytometry assays indicated
   that LRP2 overexpression indeed enhanced the proliferative capacity of
   colon cancer cells, accordant with multi-omics analysis results. In
   cell experiments, we focused on LRP2 overexpressed cell lines, and
   established their direct connection with colon carcinoma cells.
   However, the uncertainty that LRP2 silencing cell lines would repress
   carcinogenesis and development of HCT116 and SW480 was reserved in this
   study. Moreover, investigating how LRP2 modulates tumorigenesis in vivo
   and in vitro is mediated. Such as the potential involvement of cancer
   cell dissemination and progression in COAD and even other carcinomas,
   will be our future work.

   Based on previous bioinformatics analyses regarding the relationship
   between LRP2 and COAD, several enrichment results displayed that LRP2
   was likely to be involved in the regulation of the Wnt signaling
   pathway in colon carcinoma germination and progression. However, this
   inference should be validated through additional experimental studies.
   In future studies, we will continue to focus on exploring its potential
   mechanism in regulating metabolic activities and its clinical value as
   a therapeutic target in COAD. Certainly, we also expect more demanded
   findings interacted with this study would be replenished and further
   translated them into the solid clinical value to better serve COAD
   populations.

Conclusion

   LRP2 established an impregnable relationship with gene mutations,
   prognosis, pathological stage, and susceptibility to anticancer drugs
   in COAD. LRP2 exactly participated in the regulation of multitudinous
   proliferative pathways based on single-cell sequencing and
   transcriptomics datasets. Enhanced expression levels of LRP2 would tend
   towards poor prognosis, which complemented novel perspectives for
   clinical diagnosis and treatment in COAD. LRP2 might strikingly
   accelerate the growth activity of HCT116 and SW480 cell lines.

Supplementary Information

   [185]Additional file 1^ (11.7MB, pdf)
   [186]Additional file 2^ (422.1KB, docx)
   [187]Additional file 3^ (358.8KB, docx)
   [188]Additional file 4^ (329.4KB, docx)
   [189]Additional file 5^ (20.9KB, docx)

Author contributions

   Jie Zhang and Xia Zhang designed the study; Jie Zhang and Yongjie Xie
   performed primary bioinformatic analyses and wrote the manuscript;
   Xiaobin Shang and Zhenyu Shi designed bioinformatic analyses and
   commented on the manuscript; Ziyun Liu and Xiaoqing Ma performed the
   cellular and immunohistochemical experiments; Jing Zhao designed the
   experiments employed in this study; Jie Zhang, Ziyun Liu and Xiaoqing
   Ma contributed to the collection of clinical samples. All authors
   reviewed the manuscript.

Funding

   Beijing-Tianjin-Hebei Basic Research Project of Tianjin Science and
   Technology Bureau (Grant Numbers: 22 JCZXJC00040); Public Health
   Science and Technology Major Project of the Tianjin Science and
   Technology Bureau (Grant Numbers: 21ZXGWSY00020); Tianjin Key Medical
   Discipline (Specialty) Construction Project (Grant Numbers:
   TJYXZDXK-011 A).

Data availability

   The datasets in this study can be publicly obtained from TCGA-COAD and
   GEO. The original contributions presented in the study are included in
   the article/Supplementary Material, further inquiries can be directed
   to the corresponding authors.

Declarations

Ethics approval and consent to participate

   All experimental protocols in this study were approved by the Ethics
   Committee of Tianjin Medical University Cancer Institute and Hospital.
   The authors affirm that all procedures contributing to this work comply
   with the ethical standards of relevant national and institutional
   committees on human experimentation.

Statement of compliance

   We hereby confirm that all methods employed in our project have been
   conducted in accordance with the relevant guidelines and regulations.
   We ensure that every aspect of our work adheres to established
   standards.

Competing interests

   The authors declare no competing interests.

Footnotes

   Publisher's Note

   Springer Nature remains neutral with regard to jurisdictional claims in
   published maps and institutional affiliations.

   Jie Zhang, Ziyun Liu and Xiaoqing Ma have contributed equally.

Contributor Information

   Yongjie Xie, Email: 2335940013@qq.com.

   Xiaobin Shang, Email: shangxiaobin@tmu.edu.cn, Email:
   shangxiaobin626@live.cn.

   Xia Zhang, Email: zhangxia864@126.com.

References