Abstract

Background

   Given the limitations of conventional therapies in prostate cancer
   (PCa) management, identifying novel biomarkers capable of predicting
   tumor prognosis and immunotherapy response is critically important.
   This article revealed the prognosis, immunological characteristics, and
   potential mechanisms of HKR1 in PCa via bulk and single-cell RNA
   sequencing (scRNA-Seq).

Methods

   Bulk and scRNA-Seq analyses of HKR1 in PCa were collected from online
   databases. Differential expression and Cox regression analyses were
   carried out to evaluate its expression and prognosis values in PCa,
   respectively. Correlation analyses were performed to evaluate
   associations between HKR1 expression and enriched pathways, immune cell
   infiltration, and other relevant biological processes.

Results

   HKR1 showed higher expression in PCa than in normal tissues, as
   verified by qPCR in both PCa cell lines and tissue samples (p < 0.05).
   ScRNA-seq analysis demonstrated HKR1 expression in malignant cells,
   epithelial cells, and immune cell populations. Moreover, PCa sufferers
   with higher HKR1 expressions were linked with poorer prognoses, and Cox
   regression analysis suggested it was an independent indicator in PCa
   (p < 0.05). Further, we shed light on the fact that the toll-like
   receptor, the TGF-beta, and the p53 pathways were significantly related
   to HKR1 expression in PCa. HKR1 was also found to be markedly linked to
   immunity in PCa (p < 0.05). Notably, we characterized two novel
   lncRNA-RBP-HKR1 regulatory axes that potentially modulate HKR1
   transcriptional dynamics in prostate carcinogenesis.

Conclusions

   HKR1 played an undeniable role in the prognosis and immunological
   potential of PCa, providing evidence for the molecular mechanisms of
   HKR1 in PCa.

Supplementary Information

   The online version contains supplementary material available at
   10.1186/s12885-025-14230-9.

   Keywords: HKR1, Prostate cancer, Immunotherapy, Prognosis,
   ScRNA-sequencing

Introduction

   Prostate cancer (PCa) was the most common type of urinary malignant
   tumor, with approximately 268,490 new cases diagnosed in 2022 [[32]1].
   Numerous epidemiologic studies had demonstrated several risk factors of
   PCa, involving age, family history, lifestyle, environmental or
   occupational exposures, and so on [[33]2]. The clinical adoption of
   prostate-specific antigen (PSA) testing in the early 1990s marked a
   transformative advancement in prostate cancer care, significantly
   enhancing early detection capabilities and therapeutic outcomes
   [[34]3]. Currently, various clinical interventions were developed for
   PCa therapy, including radiation therapy, androgen deprivation therapy,
   chemotherapy, and novel immunotherapies [[35]4]. Although early-stage
   PCa sufferers’ five-year overall survival (OS) rate was satisfactory, a
   majority of patients were still prone to relapse and develop
   castration-resistant prostate cancer after surgery or endocrine therapy
   [[36]5]. Meanwhile, false-negative results in PCa patients with normal
   serum PSA levels could affect timely treatment and lead to a poor
   prognosis [[37]6].

   PCa exhibits distinct biological features—including pronounced tumor
   heterogeneity, a heavily immunosuppressive microenvironment, and
   minimal lymphocytic infiltration—that collectively contribute to the
   limited efficacy of conventional immunotherapies [[38]7]. Immune
   checkpoint inhibitor monotherapy demonstrated limited clinical
   efficacy, while combination therapy with the PD-1 inhibitor nivolumab
   and the CTLA-4-targeted agent ipilimumab exhibited a partial response
   in clinical trials [[39]8]. In addition, the predominant
   tumor-associated antigens in PCa include prostate-specific antigen
   (PSA), prostate-specific membrane antigen (PSMA), and six-transmembrane
   epithelial antigen of the prostate-1 (STEAP1) et al., which garnered
   significant interest in therapeutic development, with major approaches
   encompassing vaccine-based strategies and CAR-T cell therapies [[40]9].
   However, most of these therapeutic approaches remained in Phase I/II
   clinical trials, indicating a substantial pathway to clinical
   implementation [[41]10]. Therefore, it was essential to investigate new
   prognostic biomarkers to predict the occurrence of PCa and explore
   their clinical applications in PCa.

   Members of the Krüppel-associated box domain zinc finger (KRAB-ZNF)
   family served as epigenetic suppressors, increasing DNA methylation,
   which was discovered to be involved in proliferation, development, and
   even the carcinogenesis of cancer [[42]11, [43]12]. The human
   kruppel-related gene 1 (HKR1), as one member of the KRAB-ZNF family,
   was responsible for recruiting histone deacetylase complexes to domains
   of DNA-binding sites [[44]13, [45]14]. As reported, HKR1 played crucial
   roles in regulating DNA methylation, and the longitudinal association
   between HKR1 methylation level and physical development was also
   revealed [[46]15, [47]16]. Studies demonstrated a significant
   association between HKR1 methylation levels and phosphorylated tau
   deposition load [[48]17]. Moreover, Oguri et al. proposed that HKR1 was
   up-regulated in lung carcinoma in comparison with normal tissues
   [[49]18]. One whole-exome sequencing on PCa patients with a family
   history of the disease was performed, and a truncation mutation of HKR1
   was identified, which might contribute to the occurrence of PCa
   [[50]19]. In conclusion, these findings revealed the intrinsic
   connection between HKR1 and various diseases, including PCa.

   Nevertheless, the associations between HKR1 and PCa remained unknown.
   More investigations were necessary to be carried out on the potential
   functions of HKR1 in PCa based on public databases. Meanwhile, the
   “starBase” database was utilized to identify potential upstream
   mechanisms of HKR1 in PCa. This study identified a novel prognostic
   biomarker for Pca and characterized its potential clinical utility in
   therapeutic decision-making.

Materials and methods

Acquisition and analysis of PCa bulk RNA-seq data

   The TCGA-PRAD bulk RNA-seq and clinical data
   ([51]http://cancergenome.nih.gov/; accessed on January 2023) were
   employed for the current study, with 499 PCa samples and 52 normal
   samples. Those samples without critical clinical data were eliminated.
   These data were further normalized and processed in the R statistical
   environment, as previously described [[52]20, [53]21]. We analyzed HKR1
   mRNA expression across multiple cancer types using the TIMER database
   ([54]https://cistrome.shinyapps.io/timer/), while protein expression in
   prostate cancer was assessed through the HPA
   ([55]https://www.proteinatlas.org/) [[56]22]. Differential expression
   analysis of HKR1 among PCa and normal sample tissues was conducted via
   the “limma” R program with a threshold of p-values below 0.05.

Acquisition and analysis of PCa single-cell RNA sequencing (scRNA-seq) data

   The PCa scRNA-seq data for HKR1 was acquired from the [57]GSE141445
   dataset containing 13 PCa tumor samples from the GEO online website
   ([58]https://www.ncbi.nlm.nih.gov/geo/; accessed on January 2023). The
   prior article was cited for information on data processing and cell
   cluster annotations [[59]23]. Briefly, the measurement data were merged
   with the “Seurat” R package and subjected to principal component
   analysis. Using Seurat’s FindNeighbors function (v4.0, default
   parameters), we first computed the k-nearest neighbor (kNN)
   relationships between cells, then transformed these into a shared
   nearest neighbor (SNN) graph for subsequent graph-based clustering.
   With local neighborhoods of cells, the “Seurat” R package’s
   “FindClusters” function was used to cluster cells identified with
   accepted gene markers.

Prognostic abilities assessment

   Patients from the TCGA-PRAD dataset were grouped into a low- or
   high-HKR1 group with the median expression level of HKR1, and K-M
   survival analysis was carried out to discover whether the OS rate was
   different between these two subgroups. Moreover, to investigate the
   clinical relevance of HKR1 in PCa, we used clinical information from
   the TCGA-PRAD dataset, including age, lymph nodes, tumor status, PSA
   value, Gleason’s grade, stage, T, and N. Due to the small sample size
   and non-normal distribution of the data, pairwise comparisons between
   groups were performed using the Wilcoxon rank-sum test, while
   comparisons involving more than two groups were conducted using the
   Kruskal-Wallis test. Further, COX regression analyses were conducted to
   investigate relationships among OS and above clinical characteristics,
   as well as HKR1 expression, to identify independent prognostic markers
   for PCa. A nomogram was depicted with the clinical characteristics
   mentioned above using the “rms” R software package, with ROC curves,
   C-index, and calibration curves evaluating its predictive performance.

Gene Set Enrichment Analysis (GSEA) and protein-protein interaction (PPI)
network

   To investigate the biological role of HKR1 in PCa, GSEA version 4.0.0
   ([60]http://www.gsea-msigdb.org/gsea/login.jsp) was employed to perform
   KEGG pathway enrichment analysis. This approach identifies key
   signaling pathways associated with specific biological processes
   [[61]24]. The KEGG pathway analysis using the
   c2.cp.kegg.v7.1.symbols.gmt dataset revealed enriched pathways
   associated with HKR1 expression profiles in the TCGA-PRAD dataset. The
   cut-off criteria consisted of NES (normalized enrichment score) > 1.5
   as well as p values < 0.05. Additionally, the STRING database was
   utilized to draw a network of PPI networks for HKR1
   ([62]https://cn.string-db.org/). To obtain information on protein
   interactions, we set the organism filter to “Homo sapiens” and
   specified the minimum required interaction score as the median
   confidence level (0.4).

Immune assessment and drug sensitivity

   Following the methodology described in previous study [[63]25], we
   assessed correlations between HKR1 expression and immunogenomic
   features—including immune cell infiltration, tumor neoantigen burden
   (TNB), microsatellite instability (MSI), tumor mutational burden (TMB),
   and immune checkpoint molecule (ICM) expression—using the Sangerbox
   platform ([64]http://www.sangerbox.com/tool). Statistical significance
   was set at p < 0.001. The tumor microenvironment, containing stromal
   and immune scores, was estimated using TCGA-PRAD transcriptome data by
   means of the “ESTIMATE” algorithm, and the immune cell infiltration
   levels of HKR1 in PRAD were conducted by the “TIMER” website. We also
   utilized the “CellMiner” online website
   ([65]https://discover.nci.nih.gov/cellminer/) to find correlations
   among HKR1 and the IC50 values of common molecular drugs.

StarBase v2.0 website identified HKR1-related mechanisms in PCa

   To investigate upstream mechanisms regulating HKR1, the LncRNA/RNA
   binding protein (RBP)/HKR1 network was identified through the
   “StarBase” website ([66]https://starbase.sysu.edu.cn/), as described in
   previously published articles [[67]26, [68]27]. First, we conducted a
   medium stringency (≥ 5) and pan-cancer cut-off criterion (≥ 22) to
   search for HKR1-targeted RBPs. Further, the RBP-relevant LncRNAs were
   identified with a threshold of pan-cancer ≥ 10, medium stringency
   (≥ 5), and hub LncRNAs in TCGA-PRAD (|log[2](fold change)|≥1,
   FDR < 0.05, OS p < 0.05). In addition, LncRNAs were negatively
   correlated with the expression of HKR1 (p < 0.001; correlation
   coefficient > 0.3). Finally, we visualized these interaction networks
   using Cytoscape software (version 3.8.2).

Cell culture and clinical samples

   PRAD cell lines (LNCaP and DU145) and a human normal prostate cell line
   (WPMY-1) were purchased from the Cell Bank of the Shanghai Institute of
   Biological Science (SIBCB; China) and cultured according to the
   instructions described on the SIBCB website ([69]https://www.atcc.org).
   The PRAD and paraneoplastic tissues were obtained from PRAD patients
   who underwent surgery. Our current study has been approved by the
   institutional review board, and all patients involved signed an
   informed consent form for the research only.

RNA extraction and real-time qPCR

   Total RNAs from PCa samples and cell lines were acquired by TRIzol
   (Takara, Kusatsu, Japan). The PrimeScript RT reagent kit (Novabio,
   China) was utilized to synthesize the complementary DNA. Gene
   expression profiling was conducted via SYBR Green-based quantitative
   RT-qPCR, with amplification data acquired and analyzed using StepOne
   Software v2.3 for relative quantification. The relative expression
   level was normalized through the ACTB expression. The primer sequences
   were listed as follows: ACTB primer (forward):
   5’-CATGTACGTTGCTATCCAGGC-3’; ACTB primer (reverse):
   5’-CTCCTTAATGTCACGCACGAT-3’ (reverse); HKR1 primer (forward):
   5’-CTCTGTCCAGAATCGAAGCCA-3’; HKR1 primer (reverse):
   5’-GAGGATTTCCTGCCCATAAACTT-3’.

Statistical analysis

   The R programming language version 4.2.1 was used in this article. To
   examine the differences between two or more subgroups, the “Wilcoxon”
   test was used. Statistical analyses considered results with
   p-values < 0.05 to be significant.

Results

Bulk RNA-seq data of HKR1 expression and its prognosis in PCa

   The HKR1 expressions in pan-cancer were described in Fig. [70]1A. It
   revealed that HKR1 was down-regulated in malignancies including CHOL,
   COAD, READ, LIHC, and STAD, while up-regulated in tumors like PRAD,
   LUSC, UCEC, THCA, and LUAD (p < 0.05). The qRT-PCR analysis revealed
   significantly higher HKR1 mRNA expression in LNCaP cells compared to
   WPMY-1 cells (Fig. [71]1B). Also, HKR1 was up-regulated in PRAD tissues
   in contrast to normal tissues (Fig. [72]1C). Figure [73]1D
   unequivocally demonstrated that HKR1 was up-regulated in PRAD samples
   in comparison to normal prostate tissues in the TCGA-PRAD dataset
   (p = 9.982e-06). Kaplan-Meier analysis revealed HKR1 expression level
   as a significant prognostic factor, with low-expressing patients
   exhibiting prolonged overall survival compared to high-expressing cases
   (p = 0.014, Fig. [74]1E). The HPA database suggested that HKR1 protein
   expression was low in normal prostate tissues compared to PRAD samples
   (Fig. [75]1F-G).

Fig. 1.

   [76]Fig. 1
   [77]Open in a new tab

   Bulk RNA-seq data and survival prognosis of HKR1 expression in PRAD.
   (A) Pan-cancer analysis revealed HKR1 was up-regulated in PRADS with
   TCGA database; (B) qRT-PCR results indicated that HKR1 was up-regulated
   in PCa cell line LNCaP compared with normal WPMY-1 cells; (C) qRT-PCR
   results indicated that HKR1 mRNA expression was higher in PRAD tissues
   compared with adjacent tissues; (D) HKR1 was up-regulated in tumor
   tissues compared with normal tissues in TCGA-PRAD; (E) Survival
   analysis revealed high HKR1 expression is correlated with a poor
   prognosis in TCGA-PRAD; (F-G) The protein expression of HKR1 was
   highly-expressed in PCa tissues compared with tissues from HPA
   database. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001

scRNA-seq data of of HKR1 expression in PCa

   To evaluate the HKR1 expression in PRAD at the single-cell level, the
   [78]GSE141445 dataset containing 13 PRAD samples was further enrolled.
   After data processing, the cell clusters were annotated by UMAP plot,
   as presented in Fig. [79]2A. The accepted markers utilized for each
   cell cluster were detailed in Figure [80]S1. Single-cell resolution
   analysis revealed the expression pattern of HKR1 in PRAD, visualized
   through UMAP dimensionality reduction (Fig. [81]2B) and violin plots
   (Fig. [82]2C). The results indicated that HKR1 could be expressed in
   various cell clusters, including malignant cells, epithelial cells, and
   immune cells.

Fig. 2.

   [83]Fig. 2
   [84]Open in a new tab

   [85]GSE141445 scRNA-seq data of HKR1 expression in PRAD. (A) PRAD
   samples were annotated with eight cell clusters by UMAP plot; (B) The
   expression of HKR1 in PRAD at single-cell levels by UMAP plot; (C) HKR1
   had relatively higher expressions in epithelial cells, and malignant
   cells at single-cell levels

Correlation analysis between HKR1 expression and the clinicopathological
characteristics

   By using clinical information from the TCGA-PRAD dataset, patients were
   grouped with different clinicopathological characteristics, containing
   age, race, PSA value, T, M, N, and so on. Correlation analyses between
   HKR1 expression and clinicopathological information in PRAD were
   further conducted, and the Wilcoxon rank sum and Kruskal-Wallis tests
   were utilized for comparison. Notably, we found that the expression of
   HKR1 was significantly linked with age, PSA value, and T stage in
   Fig. [86]3 (all p < 0.05).

Fig. 3.

   [87]Fig. 3
   [88]Open in a new tab

   Relationship of HKR1 expression with clinicopathological
   characteristics in PRAD. Correlation analysis between HKR1 expression
   with (A) age; (B) race; (C) PSA-value; (D) T; (E) M; (F) N; (G)
   gleason-score; (H) biochemical-recurrence; (I) cancer-status

Prognostic values of HKR1 in PRAD

   Through univariate Cox regression analysis, we identified that HKR1
   expression and PSA value could serve as prognostic indicators in PRAD.
   Meanwhile, Cox regression analysis demonstrated that HKR1 might be an
   independently prognostic biomarker in PRAD (Table [89]1, p < 0.05).
   These findings indicated the prognostic value of HKR1 in PRAD.
   Prognostic modeling was performed through nomogram construction (R
   ‘rms’ package, version 6.2-0) incorporating: (i) established clinical
   predictors (age, Gleason score, T stage, N stage, PSA), (iii)
   recurrence history, and (iv) HKR1 expression levels et al.,
   demonstrating integrated predictive performance for PRAD survival
   outcomes (Fig. [90]4A). The AUC of this nomogram’s 5-year ROC curve
   measured 0.974 (Fig. [91]4B). The performance of this nomogram was
   proven by 5-year calibration curves (Fig. [92]4C).

Table 1.

   Univariate and multivariate Cox analysis based on HKR1 and
   clinicopathologic characteristics of OS for PRAD
   ID Univariate analysis Multivariate analysis
   HR HR.95 L HR.95 H P value HR HR.95 L HR.95 H P value
   Age 0.972997 0.855266 1.106936 0.677408 1.041754 0.872796 1.24342
   0.650502
   Staged T 2.940790 0.470691 18.373505 0.248550 3.515307 0.11587
   106.649139 0.470266
   Staged N 5.648441 0.791027 40.333501 0.084305 1.284547 0.004915
   335.687756 0.929734
   Cancer status 5.238172 0.868125 31.606547 0.070957 4.737859 0.22449
   99.992521 0.317409
   Lymphnodes 1.981361 0.925045 4.243891 0.078496 1.158299 0.093983
   14.275514 0.908701
   Gleason score 1.31901 0.514518 3.381395 0.564308 0.467868 0.068884
   3.177833 0.437105
   Psa value 1.213885 1.031111 1.429057 0.019916 1.374506 0.949736
   1.989256 0.091693
   HKR1 1.636443 1.223566 2.18864 0.000900 1.729961 1.227366 2.438365
   0.001749
   [93]Open in a new tab

Fig. 4.

   [94]Fig. 4
   [95]Open in a new tab

   Prognostic values of HKR1 in PCa in PRAD. (A) A nomogram was construct
   to evaluate its performance in prediction the prognosis for PRAD; (B)
   The 5-year ROC curve of nomogram; (C) The 5-year calibration curve was
   constructed to evaluate the performance of the model

HKR1 significantly associated with pathways in PRAD

   GSEA was utilized by us to obtain HKR1 significantly associated with
   pathways in PRAD. It revealed HKR1 was significantly associated with
   toll-like receptor (NES=-1.818, p < 0.05), TGF-beta (NES=-1.786,
   p < 0.05), nod-like receptor (NES=-1.829, p < 0.05), p53 (NES=-1.741,
   p < 0.05), cytosolic DNA sensing (NES=-1.772, p < 0.05), chemokine
   (NES=-1.962, p < 0.05), and B cell receptor (NES=-2.015, p < 0.05)
   pathways, as presented in Fig. [96]5; Table [97]2.

Fig. 5.

   [98]Fig. 5
   [99]Open in a new tab

   HKR1-related signaling pathways through GSEA. GSEA was performed to
   identify HKR1-relevant signaling pathways in PRAD including (A) B cell
   receptor signaling pathway; (B) chemokine signaling pathway; (C)
   cytosolic DNA sensing pathway; (D) nod-like receptor signaling pathway;
   (E) p53 signaling pathway; (F) TGF-beta signaling pathway; (G)
   toll-like receptor signaling pathway; (H) all of these seven pathways

Table 2.

   Gene set enrichment analysis results
   MSigDB collection Gene set name NES NOM P-val FDR q-val
   c2.cp.kegg.v7.1 symbols.gmt KEGG_B_CELL_RECEPTOR_SIGNALING_PATHWAY
   -2.015 <0.05 0.011
   KEGG_CHEMOKINE_SIGNALING_PATHWAY -1.962 <0.05 0.015
   KEGG_CYTOSOLIC_DNA_SENSING_PATHWAY -1.772 <0.05 0.046
   KEGG_NOD_LIKE_RECEPTOR_SIGNALING_PATHWAY -1.829 <0.05 0.034
   KEGG_P53_SIGNALING_PATHWAY -1.741 <0.05 0.054
   KEGG_TGF_BETA_SIGNALING_PATHWAY -1.786 <0.05 0.042
   KEGG_TOLL_LIKE_RECEPTOR_SIGNALING_PATHWAY -1.818 <0.05 0.036
   [100]Open in a new tab

Correlations between HKR1 expression and molecular characteristics

   A PPI network was depicted via the “STRING” website tool, and the ten
   most relevant proteins of HKR1 were identified (Fig. [101]6A). Using
   Sangerbox, we examined potential associations between HKR1 expression
   and MSI, TMB, and TNB. However, no significant correlations were
   observed in PRAD (Fig. [102]6B-D).

Fig. 6.

   [103]Fig. 6
   [104]Open in a new tab

   Correlations between HKR1 expression with molecular characteristics.
   (A) A PPI networks was established to identify ten most HKR1-relevant
   genes; Correlations between HKR1 expression with (B) MSI; (C) TNB; (D)
   TMB

Correlations between HKR1 expression and tumor immunity

   The immune subtype composition between two subgroups with HKR1
   expression was compared, and it showed that there was no difference
   between the two subgroups (Fig. [105]7A, p = 0.065). The correlation
   analyses among immune cell infiltration levels and HKR1 expression were
   illustrated in Fig. [106]7B. Our analysis revealed significant
   correlations between HKR1 expression levels and tumor-infiltrating
   immune cells, particularly CD8 + T cells and CD4 + T cells (p < 0.05).
   Moreover, HKR1 expression was found to have remarkable associations
   with ESTIMATE, Immune, and Stroma Scores (Fig. [107]7C, p < 0.05). The
   associations between HKR1 expression and immunological checkpoint
   molecules were depicted in Fig. [108]7D. We discovered HKR1 expression
   was correlated with CD86, CD70, CD44, CD40, CD274, CD200, and CD27
   (p < 0.05). Furthermore, HKR1 expression showed significant
   associations with multiple activated immune cell populations, including
   B cells, CD8 + T cells, CD4 + T cells, dendritic cells, and central
   memory CD4 + T cells (Fig. [109]7E, p < 0.05).

Fig. 7.

   [110]Fig. 7
   [111]Open in a new tab

   Correlations between HKR1 expression with tumor immunity. (A) The
   composition of pan-cancer immune subtypes between two subgroups based
   on HKR1 expression. Correlations analysis between HKR1 expression and
   (B) immune cells infiltration; (C) tumor microenvironment; (D) immune
   checkpoint molecules; (E) immune cells. *P < 0.05; **P < 0.01;
   ***P < 0.001

HKR1-related drugs based on cellminer database

   We evaluated the correlations between HKR1 expression and drug
   sensitivities (IC50 values) through “CellMiner”, with cut-off values of
   correlation coefficient ≥ 0.3 and p < 0.01. As shown in Fig. [112]8,
   the HKR1 expression in the samples was positively correlated with the
   IC50 values of chelerythrine, cladribine, and fludarabine (p < 0.01).
   Patients with low HKR1 expression showed better response to molecular
   drugs’ chemotherapies.

Fig. 8.

   [113]Fig. 8
   [114]Open in a new tab

   Correlation analysis of HKR1 expression and drug sensitivity of
   chemotherapy. Associations between HKR1 expression with IC50 values of
   (A) chelerythrine; (B) cladribine; (C) fludarabine in the CellMiner
   database

Identification of LncRNA/RBP/HKR1 mRNA networks

   In this research, the network of LncRNA/RBP/HKR1 was evaluated through
   the website StarBase v2.0. The specific flowchart is summarized in
   Fig. [115]9A. The LncRNA/HNRNPA2B1/HKR1 was discovered following a
   second search of a chosen RBP-relevant LncRNA with Venn diagrams
   (Fig. [116]9B). Finally, we identified two regulatory
   axes—LINC00342/HNRNPA2B1/HKR1 and AP00468.1/HNRNPA2B1/HKR1—as potential
   upstream mechanisms controlling HKR1 expression in PRAD (Fig. [117]9C).

Fig. 9.

   [118]Fig. 9
   [119]Open in a new tab

   Identification of LncRNA/RBP/HKR1 mRNA networks. (A) Flowchart and (B)
   Venn diagrams of identification for HKR1-related mechanism of
   LncRNA/RBP/HKR1 mRNA networks. (C) LncRNA/RBP/HKR1 mRNA networks were
   pictured by Cytoscape 3.6.1

Discussion

   Millions of men were diagnosed with PCa disease around the world
   [[120]6]. On the basis of the genetic characteristics of tumors, it was
   feasible to evaluate the occurrence and progression of PCa. As noted,
   HKR1 was a member of the KRAB-ZNF gene family, playing important roles
   in various cancers [[121]28]. Our preliminary data indicated
   upregulated HKR1 expression in PCa, yet the functional and clinical
   significance of this observation requires systematic investigation.

   The scRNA-seq technology could detect the transcriptional landscape of
   target tissues at the single-cell level, which enabled the fine
   division of cell subtypes and revealed cellular heterogeneity
   [[122]29]. Nowadays, its application was very extensive, enabling
   researchers to gradually deepen their understanding of the
   tumorigenesis, tumor microenvironment, and immune response in PCa
   [[123]23, [124]30]. Our analysis of existing single-cell RNA sequencing
   (scRNA-seq) data revealed HKR1 was mainly expressed in epithelial and
   malignant cell subtypes in PRAD based on the existing scRNA-seq data.
   Song et al. had reported the heterogeneities of tumor-related
   epithelial cell states in PRAD [[125]31]. Nevertheless, whether
   abnormal expression of HKR1 in epithelial cells promotes prostate
   cancer needed further exploration.

   To shed light on various signaling pathways related to the targeted
   genes, GSEA was further employed [[126]32]. Moreover, we identified
   several signaling pathways associated with HKR1, including toll-like
   receptor, TGF-β, p53, cytosolic DNA sensing, NOD-like receptor,
   chemokine, and B cell receptor signaling pathways. Toll-like receptor 4
   was reported to play a vital role in tumor progression through
   regulating the RAS/MAPK signaling pathway [[127]33]. Deng et al.
   reported that the TGF-beta-related regulatory network played a vital
   role in PCa progression [[128]34]. Consistent with findings by Wan and
   the colleagues, elevated p53 protein expression in prostate cancer
   cells demonstrated significant associations with enhanced
   proliferative, migratory, and adhesive capacities [[129]35]. Chemokines
   and their receptors were proven to contribute survival, proliferation,
   and invasion capabilities to PCa cells in both the primary and
   metastatic sites of cancer [[130]36]. The regulation of chemokines and
   their receptors may potentially contribute to combating tumor
   development. These findings suggested that HKR1 might take part in the
   occurrence or progress of PCa through the above signaling pathways, and
   further experiments were needed to confirm it in vitro.

   PCa, often characterized as an immunologically “cold” tumor, generally
   shows limited response to immune-related therapies, although a subset
   of patients may still benefit from immunotherapy. Thus, the evaluation
   of the correlation between tumor cells and the tumor microenvironment
   was crucial for understanding the occurrence, progression, and
   therapeutic response of PCa [[131]37]. In addition, immune cell
   infiltration, which was a vital component of the tumor
   microenvironment, was closely correlated with the progression and
   immunological responses of the tumor [[132]38]. Our results indicated
   that HKR1 was positively linked with the number of CD8^+ T cells. Also,
   HKR1 expression showed a strong correlation with ESTIMATE, stromal, and
   immune Scores, indicating the relevance of the microenvironment for
   prognosis and precision immunotherapy of PCa. A high level of stromal,
   immune and estimate scores had been proven to be linked to a poor
   prognosis in PCa [[133]39]. Moreover, we found HKR1 expression was
   significantly associated with immune cells and ICM, including CD86,
   CD70, CD44, CD40, CD274, CD200, activated CD4 T cells, and etc. These
   findings demonstrate a strong correlation between HKR1 and tumor
   immunity in PRAD, highlighting its potential as a target for
   immunotherapy.

   To discover HKR1-related drugs in PCa, the “CellMiner” dataset was
   applied to facilitate the process of researching and selecting
   anticancer medications [[134]40, [135]41]. Our results demonstrated a
   significant positive correlation of HKR1 expression with the IC50 of
   chelerythrine, cladribine, and fludarabine, which had the potential to
   restrain the progression of PRAD. As reported, chelerythrine could
   suppress the proliferation of PCa cells via regulating NF-κB pathways
   [[136]42]. Fludarabine was found to be a potential target of N-MYC
   overexpressing neuroendocrine PCa [[137]43]. Overall, it meant PRAD
   patients with abnormal expression of HKR1 were more likely to benefit
   from these molecular drug therapies.

   Furthermore, the “StarBase” website was used to explore the upstream
   mechanisms of HKR1 in PCa, revealing its correlation with several RBPs
   and associated LncRNAs. Eventually, we identified the
   LINC00342/HNRNPA2B1/HKR1 and [138]AP004608.1/HNRNPA2B1/HKR1 axes as
   potential mechanisms of HKR1 in PCa. Heterogeneous nuclear
   ribonucleoproteins A2B1 (HNRNPA2B1) was a pre-mRNA-binding protein that
   acted as a crucial RNA regulator of RNA N6-methyladenosine (m^6A),
   which was involved in the progress and occurrence of various tumors
   [[139]44, [140]45]. Analysis of datasets from the TCGA database
   suggests that HNRNPA2B1 may function as a crucial m6A regulator in
   prostate cancer, and knockdown of HNRNPA2B1 can suppress the migration
   of PCa cells in vitro [[141]46]. As for the related lncRNA, colorectal
   tumor cells were induced to metastasize and proliferate by LINC00342
   via sponging miR-19a-3p to regulate NPEPL1 expression [[142]47], and
   silence of LINC00342 could inhibit the invasion, proliferation, and
   migration of AGS cells in vitro [[143]48]. Besides, [144]AP004608.1 was
   found to be up-regulated in PCa tissues compared with benign prostate
   tissues [[145]49]. Overall, we had identified two LncRNA/RBP/HKRI
   network axes that might act on PRAD. Nevertheless, the two upstream
   mechanisms of HKR1 still needed further experiments to verify in PRAD.

   These limitations should also be pointed out in this research. Firstly,
   most parts of the data were obtained using bioinformatics analysis, and
   the potential function of HKR1 expression and related mechanisms in PCa
   needed to be investigated in vivo. Secondly, the value of HKR1 in
   immunotherapy intervention for PCa still needed further clinical trials
   to verify.

Conclusions

   In conclusion, HKR1 was found to be up-regulated in PCa, and it could
   serve as a potential prognostic biomarker for PCa. Moreover, HKR1 was
   closely associated with several signaling pathways, such as the
   toll-like receptor, TGF-beta, and p53, in PCa. Moreover, HKR1 was found
   to be significantly correlated with immunity, which could act as a
   latent target for the clinical intervention of PCa. Two LncRNA/RBP/HKR1
   axes (LINC00342/HNRNPA2B1/HKR1 and AP00468.1/HNRNPA2B1/HKR1) were
   identified for their potential upstream regulatory mechanisms of HKR1
   in PCa. These findings provide a novel prognostic indicator for PCa and
   reveal its potential impact on immunotherapy in this disease.

Electronic supplementary material

   Below is the link to the electronic supplementary material.
   [146]12885_2025_14230_MOESM1_ESM.tif^ (870.9KB, tif)

   Figure S1. Accepted cell markers utilized for each cell cluster by
   dotplot in the GSE141445 scRNA-seq data of PRAD

Acknowledgements