PPP4C: a potential molecular marker and therapeutic target in thyroid cancer and triple-negative breast cancer

Abstract

The protein phosphatase PPP4C has been implicated in oncogenic processes, but its role in thyroid carcinoma (THCA) and triple-negative breast cancer (TNBC) remains unclear. Here, we combined bioinformatic analyses with in vitro experiments to investigate the clinical and functional significance of PPP4C in these malignancies. PPP4C was significantly upregulated in both THCA and TNBC compared with normal tissues, and higher expression was associated with adverse clinicopathological features and poorer overall survival. Functional assays showed that PPP4C knockdown suppressed proliferation, migration, and invasion in TNBC and THCA cell lines, supporting a pro-tumorigenic role in vitro. Integrative analyses further suggested that elevated PPP4C expression is associated with oncogenic signaling pathways, RNA modification–related genes, and immune-related characteristics. Drug sensitivity prediction analysis indicated differential responsiveness to selected targeted agents, and experimental validation demonstrated that PPP4C depletion enhances sensitivity to specific inhibitors in vitro. Overall, PPP4C is consistently overexpressed and functionally relevant in THCA and TNBC. PPP4C may represent a candidate biomarker associated with tumor progression and therapeutic response, warranting further mechanistic and in vivo investigation.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-026-16083-2.

Introduction

Breast cancer remains one of the most prevalent cancers worldwide, with an estimated 319,750 new cases of invasive breast cancer diagnosed in the United States in 2025 [1]. Among breast cancer subtypes, triple-negative breast cancer (TNBC) accounts for approximately 15% of cases [2]. TNBC is characterized by the lack of estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) expression [3], making it one of the most aggressive and treatment-resistant forms of breast cancer. Despite advances in treatment, TNBC remains associated with poor prognosis, highlighting the need for novel therapeutic targets and biomarkers.

Interestingly, studies have suggested a bidirectional association between breast cancer and thyroid cancer (THCA). Breast cancer survivors have an increased risk of developing thyroid cancer, and conversely, patients with THCA are at a higher risk of developing breast cancer [4, 5]. This clinical link is thought to be driven by hormonal crosstalk, as thyroid hormones influence the growth of breast cells, while estrogen can promote tumorigenesis in thyroid cells [6]. Given the reported overlap in risk factors and hormonal influences between these cancers, there is growing evidence for a potential common etiology that warrants further investigation [7].

The molecular mechanisms driving both TNBC and THCA remain incompletely understood, with several key oncogenic factors still to be explored. Protein phosphatase 4 catalytic subunit (PPP4C) has been implicated in cancer progression, with previous studies indicating its involvement in the regulation of cancer cell proliferation, survival, and DNA repair [8]. In breast cancer, our previous research indicated that PPP4C may serve as a novel biomarker and therapeutic target. In contrast, THCA-specific evidence regarding PPP4C remains limited, and its expression pattern, clinical significance, and functional relevance in thyroid cancer have not been systematically investigated. Therefore, whether PPP4C may represent a convergent molecular feature associated with aggressive phenotypes in both TNBC and THCA remains unclear.

Biologically, PPP4C participates in fundamental processes relevant to malignant behavior, including cell-cycle control and genome maintenance programs. We therefore hypothesized that PPP4C upregulation may represent a convergent molecular feature associated with aggressive phenotypes in both TNBC and THCA by engaging shared oncogenic programs; importantly, our integrative analyses are designed to generate testable hypotheses rather than establish causality.

Therefore, this study aimed to investigate the expression pattern, clinical significance, and functional relevance of PPP4C in THCA and TNBC, and to determine whether PPP4C may represent a shared molecular feature associated with aggressive phenotypes in these two tumor types. By investigating the role of PPP4C in both cancers, we seek to identify shared molecular features that may inform future mechanistic investigations and therapeutic hypothesis generation.

Methods

PPI, KEGG, and GO analyses

The protein–protein interaction (PPI) network of PPP4C was constructed using the STRING database (https://string-db.org/, accessed on 8 August 2023), with the minimum required interaction score set to 0.4. The PPI network was visualized using Cytoscape. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses were performed using the “clusterProfiler” package in R. The pathways and biological processes significantly enriched with a false discovery rate (FDR) of less than 0.05 were considered relevant.

PPP4C messenger RNA (mRNA) and protein expression analysis

A Kruskal–Wallis analysis of PPP4C expression was performed to compare PPP4C mRNA expression in 31 normal tissues and 21 cancer tissues. Data from the TCGA database were used to compare the differences in PPP4C expression. We also analyzed PPP4C expression in GTEx. The Human Protein Atlas (HPA) online website (https://www.proteinatlas.org/) was used to analyze PPP4C in the "Tissue Atlas". P values were adjusted for multiple comparisons using the Benjamini–Hochberg method.

Differential and prognostic analysis

The median expression value of PPP4C was used to generate Kaplan–Meier survival curves, utilizing the "SurvMiner" and "Survival" R packages. The relationship between PPP4C expression and multiple human tumor survival rate was then examined using the "forestplot" R package. To explore the correlation between PPP4C expression and tumor stage, we employed UALCAN (https://ualcan.path.uab.edu/index.html).

Univariate and multivariate Cox regression

Univariate and multivariate Cox regression were performed to identify independent prognostic factors. For the BRCA cohort, covariates included PPP4C expression, age, pathological stage, pT, pN, and pM (when available). For the THCA cohort, covariates included PPP4C expression, age, gender, pathological stage, pT, pN, and pM (when available). Variables with P < 0.05 in univariate analysis were included in multivariate analysis. Nomograms were built based on the multivariate Cox model using the “rms” R package. C-index was used to evaluate discrimination; time-dependent ROC curves were generated.

RNA modification

The SangerBox database was utilized to perform an in-depth analysis of RNA methylation modifications, with a specific focus on N1-methyladenosine (m1A), 5-methylcytidine (m5C), and N6-methyladenosine (m6A). These modifications are associated with three distinct types of genes: writers, readers, and erasers. Additionally, we examined the expression of PPP4C in connection with genes involved in RNA methylation alterations across various types of tumors, providing crucial insights into the potential role of PPP4C in RNA regulatory processes within the context of cancer.

Genetic alteration

In our analysis, we employed the TCGA Pan-Atlas Cancer Genomics Dataset, which is available through the cBioPortal platform, to explore genetic alterations in PPP4C. This dataset offers a comprehensive and integrated resource for analyzing genomic data across a wide range of cancer types, providing a valuable tool for studying the molecular mechanisms underlying various cancers.

Immune analysis

To further our understanding of the tumor microenvironment and its relationship with PPP4C expression, we utilized the SangerBox online database. Specifically, we explored the correlation between seven types of immune cells, five categories of immune regulatory factors, 47 immune checkpoint genes, and PPP4C expression. These investigations provided critical insights into the potential link between PPP4C expression and various immune factors, further providing insights into how PPP4C expression is associated with immune-related characteristics, which may generate hypotheses for future immunotherapy-oriented studies.

Immunohistochemistry (IHC) experiment

To detect the expression of specific proteins, immunohistochemistry was performed on tissue sections. Tissue samples were first fixed in 4% paraformaldehyde at room temperature for 12 h. After fixation, tissues were dehydrated through a graded ethanol series and embedded in paraffin. Sections were then cut into 5 µm thick slices using a microtome and mounted on glass slides. The sections were dewaxed and rehydrated, followed by blocking endogenous peroxidase activity using 0.3% hydrogen peroxide solution. Antigen retrieval was performed via microwave heating in an appropriate retrieval buffer. Non-specific binding was blocked by incubating with 1% BSA. The sections were incubated overnight at 4 °C with the PPP4C antibody (1:200, ProteinTech, USA), followed by incubation with biotin-conjugated secondary antibody. Signals were developed using an immunohistochemical staining kit and observed under a light microscope. Positive signals were identified by brown-yellow staining, and the nuclei were counterstained with hematoxylin. IHC staining was independently assessed by two pathologists blinded to the clinical information. Immunoreactivity was semiquantitatively evaluated according to staining intensity (0–3) and the proportion of positive cells (0–4). The final IHC score was obtained by multiplying these two parameters, yielding a total score ranging from 0 to 12. In cases of discrepant scoring, a consensus score was reached after joint review.

Cell culture

The human thyroid cancer cell lines TPC-1 and KTC-1, and breast cancer cell lines MCF-10A, MCF-7, SK-BR-3, MDA-MB-468, and MDA-MB-231 were obtained from the American Type Culture Collection (ATCC, Manassas, VA, USA). These cell lines were maintained in RPMI-1640 (Gibco, USA) or DMEM (Gibco, USA) supplemented with 10% fetal bovine serum (FBS, Gibco, USA) and 1% penicillin/streptomycin (Gibco, USA) in a humidified incubator at 37 °C with 5% CO2. For transfection, cells were seeded at a density of 2 × 10^5 cells/well in 6-well plates and cultured in a CO2 incubator at 37 °C. Cells were allowed to adhere to the surface for 4–6 h before transfection.

siRNAs for knockdown

To investigate the functional role of PPP4C, two independent small interfering RNAs (siRNAs) targeting PPP4C were designed and synthesized by GenePharma (Shanghai, China; sequences provided in Supplementary Table 1). For transfection, cells were incubated with siRNA complexes formed using RNA TransMate reagent (Sangon Biotech, China), following the manufacturer's specifications. Transfection was performed in serum-free medium for 6 h, after which the medium was replaced with complete medium supplemented with 10% FBS. Cells were subsequently cultured for 24–48 h under standard incubation conditions prior to functional analysis.

Quantitative real-time PCR (qRT-PCR)

Total RNA was extracted from cells using TRIzol reagent (Thermo Fisher, USA) according to the manufacturer's protocol. RNA was reverse-transcribed into complementary DNA (cDNA) using 2 μg of total RNA per reaction. Quantitative real-time PCR was performed using SYBR Premix Ex Taq kit (Takara, Japan). ACTB was used as the internal reference gene. Relative gene expression levels were calculated using the 2^ − ΔΔCt method. Primer sequences for ACTB, and PPP4C are provided in Supplementary Table 2.

Western blot analysis

Cells were harvested and lysed in radioimmunoprecipitation assay (RIPA) buffer supplemented with phenylmethylsulfonyl fluoride (PMSF). Protein concentrations were quantified prior to analysis, and equal amounts of total protein were separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). Proteins were subsequently transferred onto polyvinylidene difluoride (PVDF) membranes (Millipore). After blocking with 5% bovine serum albumin (BSA) at room temperature, membranes were incubated overnight at 4 °C with the indicated primary antibodies. Following incubation with horseradish peroxidase (HRP)–conjugated secondary antibodies, protein bands were detected using an enhanced chemiluminescence (ECL) system (Millipore) and visualized with a Tanon-5220 imaging system.

The following primary antibodies were used: PPP4C (Abcam, ab195371, 1:1000), and Tubulin (CST, #3873, 1:5000), which served as the loading control. Band intensities were quantified using ImageJ software and normalized to Tubulin.

Hormone treatment assays

For hormone treatment experiments, estradiol (E2) and liothyronine (T3) were purchased from MCE (China). Stock solutions were prepared in dimethyl sulfoxide (DMSO) and stored at − 20°C. Cells were treated with increasing concentrations of estradiol (0.01–1 μM) or liothyronine (0.01–1 μM) for 48 h, with an equal volume of DMSO used as the vehicle control. The final concentration of DMSO was kept constant (≤ 0.1%) across all treatment groups. After treatment, cells were harvested for qRT-PCR and Western blot analyses to evaluate PPP4C expression at both mRNA and protein levels.

CCK-8 assay

Cell viability was assessed using the Cell Counting Kit-8 (CCK-8) assay (Abclonal, China). Thyroid cancer and breast cancer cells were seeded at a density of 3000 cells/well in a 96-well plate. After 48, 72, and 96 h of incubation, 10% CCK-8 solution was added to each well, and the plate was incubated for 2 h at 37 °C. Absorbance was measured at 450 nm using a microplate reader. All experiments were performed with at least three independent biological replicates and three technical replicates per condition.

For dose–response assays (IC50 determination), cells were treated with a series of concentrations of Foretinib (HY-10338, MCE), Afatinib (HY-10261, MCE), VX-11e (HY-14178, MCE), GDC0810 (HY-12864, MCE) and GSK591 (HY-100235, MCE) for 48 h, followed by cell viability measurement and IC50 estimation using nonlinear regression. Drugs were dissolved in DMSO to prepare stock solutions and diluted in complete medium immediately before use. The final DMSO concentration was kept identical across all groups and did not exceed 0.1% (v/v).

Colony formation assay

For the colony formation assay, cells were seeded at a density of 1000 cells/well in 6-well plates and cultured for two weeks at 37 °C, 5% CO2. The medium was replaced every 3 days. After two weeks, colonies were fixed with 4% paraformaldehyde for 20 min, stained with crystal violet for 30 min, and then photographed. Colonies were counted and quantified. Three biological replicates were performed, and for each replicate, three technical replicates were included (i.e., three wells per condition).

Transwell migration and invasion assay

For the migration assay, 5 × 10^4 cells were seeded in 200 μL serum-free medium and added to the upper chamber of a 24-well Transwell insert (Corning, USA). The lower chamber was filled with 600 μL complete medium containing 10% FBS. The cells were incubated at 37 °C with 5% CO2 for 24 h. After incubation, cells were fixed with 4% paraformaldehyde for 20 min and stained with crystal violet for 30 min. The upper chamber was gently wiped with sterile cotton to remove non-migrated cells. Migrated cells were counted under a microscope, and at least three random fields per well were quantified. For the invasion assay, 100 μL of extracellular matrix gel (Matrigel, Corning) was added to the upper chamber and incubated for 30 min at 37 °C to allow gel solidification. The invasion assay was performed as described above, with the only difference being the presence of Matrigel in the upper chamber. All assays were conducted with at least three biological replicates and three technical replicates.

Wound healing assay

For the wound healing assay, cells were seeded in 6-well plates and cultured until they reached 90–100% confluence. A scratch was made using a 200 μL pipette tip. The cells were washed twice with PBS to remove debris, and fresh low serum medium was added to avoid interference with cell growth. Wound healing was monitored by imaging the cells at 0 h and 24 h after scratching. The rate of wound closure was calculated by measuring the wound area at each time point using ImageJ software. Images were captured from at least three independent biological replicates and three random fields per well were quantified.

Transcription factors

PPP4C-related transcription factors were predicted using the hTFtarget database (https://guolab.wchscu.cn/hTFtarget), which provides curated and predicted TF–target regulatory relationships. PPP4C was used as the target gene, and all candidate TFs regulating PPP4C were retrieved as the initial TF candidate set. To identify TFs potentially relevant to PPP4C dysregulation in TNBC and THCA, differential expression analyses were performed separately in the TNBC and THCA cohorts. For TNBC, samples were selected from the TCGA-BRCA dataset according to the TNBC subtype definition (ER −/PR −/HER2 −), and THCA samples were obtained from the TCGA-THCA dataset. Gene expression matrices were normalized as described above. DEGs were identified using the "limma" package in R, with thresholds of |log2fold change|> 1 and adjusted P value (FDR) < 0.05.

The PPP4C-related TF list obtained from hTFtarget was intersected with DEGs from TNBC and THCA, respectively, to generate PPP4C-related differentially expressed TFs in each cancer type. Subsequently, the overlap between the TNBC and THCA TF sets was further intersected to identify common PPP4C-related TFs shared by both cancers. These shared TFs (n = 17) were visualized using a Venn diagram and summarized in the Results section. For visualization and relationship analysis of TFs, TF expression patterns (upregulated vs. downregulated) in TNBC and THCA were displayed using bar plots/heatmaps generated with the "ggplot2" and "pheatmap" R packages. In addition, TF–TF relationship networks were constructed by calculating pairwise expression correlations among the identified TFs within each cohort using Spearman's correlation. Edges were retained when |Spearman's ρ|> 0.3 and P < 0.05, and networks were visualized using Cytoscape or the R package "igraph". These TF results are predictive and hypothesis-generating; validation will require additional experiments.

PPP4C expression and drug sensitivity

The relationships between the half-maximal inhibitory concentration (IC50) of targeted therapeutic agents and the PPP4C expression levels were investigated using the "pRRophetic" R package. Drug response prediction was based on the Genomics of Drug Sensitivity in Cancer (GDSC) database. Predicted IC50 values were calculated separately for the TCGA-THCA cohort and the TCGA-BRCA cohort, with the TNBC subset defined according to ER −/PR −/HER2 − status. Patients were stratified into high- and low-PPP4C expression groups using the median expression value as the cutoff.

Statistical analysis

All statistical analyses were performed using R software (version 3.6.3) and GraphPad Prism (version 9.0). Data are presented as mean ± standard deviation (SD). Comparisons between two groups were conducted using Student's t-test. For comparisons involving more than two groups or multiple time points, one-way or two-way analysis of variance (ANOVA) followed by appropriate post hoc tests was applied. A two-sided P value < 0.05 was considered statistically significant unless otherwise specified. For high-dimensional analyses, P values were adjusted using the Benjamini–Hochberg method when applicable.

Results

PPI network and KEGG/GO enrichment of PPP4C-binding proteins

The STRING database and Cytoscape were used to identify 26 PPP4C-interacting proteins (Fig. 1A). KEGG enrichment analysis of these proteins revealed enrichment of several cancer-relevant pathways (Fig. 1B). Notably, pathway categories related to PI3K–Akt signaling, AMPK signaling, and Hippo signaling were among the most prominent enriched pathways. These pathways are well known to regulate tumor cell proliferation, metabolic adaptation, and survival. GO enrichment analysis further indicated that the major biological processes included DNA biosynthetic processes and mitotic G1/S transition checkpoint signaling. These findings suggest that PPP4C-associated networks may be involved in cell-cycle control and genome maintenance programs. The cellular component and molecular function analyses showed enrichment in phosphatase complexes and phosphatase regulator activity (Fig. 1C). Collectively, these results highlight two major functional themes of PPP4C-associated networks in cancer: regulation of cell cycle and genome stability, and modulation of oncogenic signaling pathways such as PI3K–Akt, AMPK, and Hippo signaling.

Fig. 1.

Protein–protein interaction network, KEGG and GO analysis of 26 targeted binding proteins of PPP4C. A Protein–protein interaction network; B The circos plot of KEGG analysis; C The dot plot of GO analysis. Adjusted P values were calculated using the Benjamini–Hochberg method; FDR < 0.05 was considered significant

PPP4C expression in normal tissues and tumors

We analyzed PPP4C expression in normal tissues from the GTEx database and found that PPP4C was expressed in all tissues, with the three most highly expressed tissues being the small intestine, breast, and stomach (Figure S1A). For TCGA tumors and adjacent normal tissues, PPP4C expression was only significantly downregulated in kidney chromophobe (KICH) while it was upregulated in almost all tumors, including glioblastoma multiforme (GBM), cervical squamous cell carcinoma and endocervical adenocarcinoma (CESC), lung adenocarcinoma (LUAD), colon adenocarcinoma (COAD), breast invasive carcinoma (BRCA), and thyroid carcinoma (THCA), etc. (Figure S1B).

Given that BRCA and THCA are among the most common tumors affecting women, we further focused on exploring the effect of PPP4C on these two tumors. PPP4C was significantly upregulated in tumor tissues compared with adjacent normal tissues in both breast cancer and thyroid cancer datasets (Fig. 2A–B). Immunohistochemical (IHC) results of the HPA database also showed that PPP4C was highly expressed in tumor tissues (Fig. 2C–D). IHC staining analysis of the cohort in our center showed that PPP4C IHC scores of TNBC tissues were higher than those of normal breast tissues (Fig. 2E–F). Similarly, Papillary thyroid carcinoma (PTC) tissue showed a comparable pattern (Fig. 2G–H). These findings suggest that PPP4C overexpression is associated with malignant tissues in both TNBC and THCA and may contribute to tumor-related phenotypes.

Fig. 2.

PPP4C expression in breast cancer and thyroid cancer tissues. A, B Volcanic map showed the expression difference between normal and tumor tissue in breast cancer (A) and thyroid cancer (B). C, D The protein levels of PPP4C in breast cancer (C) and thyroid cancer (D) by HPA database. E, G Representative IHC staining of PPP4C in adjacent normal and tumor tissues from breast (E) and thyroid (G). F, H Quantification of PPP4C protein expression in TNBC and adjacent normal breast tissues (F), and in PTC and adjacent normal thyroid tissues (H). Data are presented as mean ± SD. Two-tailed Student's t-test was used. *P < 0.05, **P < 0.01, ***P < 0.001

Correlations between PPP4C expression, clinicopathological features, and prognosis

The correlations between the PPP4C and the clinicopathological index of patients with BRCA and THCA from the TCGA dataset were investigated. We found no significant differences in PPP4C expression at different pathological tumor (pT) stages (Fig. 3A). However, the results demonstrated that patients with higher pathologic regional lymph node (pN) stages and pathological tumor, node, metastasis (pTNM) stages showed higher transcriptional levels of PPP4C than those with lower stages in BRCA (Fig. 3B–C). Meanwhile, in patients with THCA, the transcript levels of PPP4C were higher at higher pT, pN, and pTNM stages (Fig. 3D–F).

Fig. 3.

Correlation between PPP4C expression and clinicopathological and prognostic index across TCGA BRCA and THCA. A-C BRCA. A pT stages; B pN stages; C pTNM stages. D-F THCA. D pT stages; E pN stages; F pTNM stages. G, H Kaplan–Meier overall survival analysis of PPP4C-high and PPP4C-low groups in BRCA (G) and THCA (H). One-way ANOVA was used for stage comparisons; survival differences were assessed using the log-rank test

In terms of the prognostic relationship between PPP4C and BRCA and THCA, Kaplan–Meier analysis showed that higher PPP4C expression was associated with poorer overall survival in both the BRCA and THCA cohorts (Fig. 3G–H). These observations are association-based and do not establish causality.

Independent prognostic value of PPP4C in BRCA and THCA

To facilitate individualized prognosis prediction, we constructed nomograms integrating PPP4C expression with clinicopathological variables for patients with BRCA and THCA (Fig. 4A–B). Briefly, candidate variables were first screened by univariate Cox regression, and those of clinical relevance were subsequently included in multivariate Cox regression to identify independent prognostic factors. For the BRCA cohort, PPP4C expression and clinicopathological variables included in the multivariate Cox model remained significant predictors of survival (Fig. 4A). For the THCA cohort, PPP4C expression, age, pM, pN, pT, gender, and pathological stage were included in the multivariate analysis, and PPP4C also emerged as an independent prognostic indicator (Fig. 4B).

Fig. 4.

PPP4C-based nomograms and prognostic performance in BRCA and THCA. A Nomogram for predicting overall survival (OS) in the BRCA cohort constructed from the multivariate Cox regression model incorporating PPP4C expression and clinicopathological variables (age, pT, pN, pM, vital status, and pathological stage). B Nomogram for predicting OS in the THCA cohort constructed from the multivariate Cox regression model incorporating PPP4C expression and clinicopathological variables (age, gender, pT, pN, pM, and pathological stage). C Kaplan–Meier OS curves comparing the high- and low-risk groups stratified by the median nomogram-derived total risk score in the BRCA cohort. D Kaplan–Meier OS curves comparing the high- and low-risk groups stratified by the median nomogram-derived total risk score in the THCA cohort. E Discriminatory performance of the BRCA nomogram evaluated by Harrell's concordance index (C-index). F Discriminatory performance of the THCA nomogram evaluated by Harrell's concordance index (C-index). A higher C-index indicates better concordance between predicted and observed survival outcomes

In each nomogram, a point value was assigned to each predictor (e.g., PPP4C expression level and pathological stage), and the total points were calculated by summing the scores of all variables for a given patient. The total points were then converted into an estimated probability of overall survival at specific time points. Based on the total nomogram-derived risk score, patients were stratified into high- and low-risk groups using the median value as the cutoff.

Kaplan–Meier analysis showed that patients in the high-risk group had significantly worse overall survival (OS) than those in the low-risk group in both the BRCA and THCA cohorts (Fig. 4C–D). The discriminatory ability of the nomograms was assessed using Harrell's concordance index (C-index), with values of 0.760 for BRCA and 0.884 for THCA, indicating good agreement between predicted and observed outcomes (Fig. 4E–F). These results suggested that overexpressed PPP4C could be a prognostic biomarker candidate for BRCA and THCA.

Association of PPP4C expression with RNA modification-related genes in TNBC and THCA

The connection between PPP4C expression as well as RNA modification-related genes can be seen in Fig. 5A. PPP4C expression often showed a positive correlation with TNBC and THCA gene expression associated with m1A, m5C, and m6A (Fig. 5A). These correlation patterns suggest a potential link between PPP4C expression and RNA modification-related programs; mechanistic validation is required. The data in Fig. 5A are presented using a radar chart (Fig. 5B–G). PPP4C was associated with tumor progression at the transcriptional level and may also have a profound impact on tumor biological processes through the epigenetic regulatory pathway of RNA modification.

Fig. 5.

RNA epigenetic modification associated with PPP4C. A Spearman correlation between PPP4C expression and the expression of m1A-, m5C-, and m6A-related regulatory genes. B–D Associations between PPP4C expression and RNA modification scores in the TCGA-BRCA TNBC subset (BRCA-TNBC): B m1A, C m5C, and (D) m6A. E–G Associations between PPP4C expression and RNA modification scores in TCGA-THCA: E m1A, F m5C, and (G) m6A. Correlations were assessed using Spearman's correlation; group comparisons were performed using two-tailed Student's t-test. *P < 0.05, **P < 0.01, ***P < 0.001

Genetic alteration of PPP4C

To characterize the somatic alteration landscape of PPP4C across cancers, we analyzed single-nucleotide variants (SNVs) and copy number variations (CNVs) using TCGA pan-cancer data. As shown in Fig. 6A, PPP4C alterations were detected in multiple tumor types, with the highest mutation frequency observed in lung adenocarcinoma (LUAD), followed by colon adenocarcinoma (COADREAD) and stomach adenocarcinoma (STAD). Overall, PPP4C mutation frequency was low across most cancer types (0.2%–2.7%), and missense mutations were the predominant alteration type, followed by nonsense mutations and in-frame deletions.

Fig. 6.

Genetic alteration profiles of PPP4C in TCGA cohorts. A Single-nucleotide variant (SNV) frequencies of PPP4C across different cancer types in TCGA. B, C Waterfall plots comparing somatic mutation landscapes between PPP4C-low and PPP4C-high groups in the TCGA-BRCA TNBC subset (B) and TCGA-THCA (C). D The SNV classes of PPP4C in the BRCA-TNBC. E Copy number variation (CNV) analysis illustrating the frequency of CNV events across PPP4C genes. These panels are descriptive genomic analyses

PPP4C mutations were distributed across the protein sequence without clear recurrent hotspots, suggesting that PPP4C is not commonly affected by recurrent driver-type coding mutations. Given their low frequency, these variants are more likely to represent passenger alterations. In the TCGA-BRCA TNBC subset and TCGA-THCA cohort, waterfall plots showed differences in mutational landscapes between PPP4C-low and PPP4C-high groups (Fig. 6B–C), suggesting that PPP4C expression may be associated with distinct mutational contexts. In the BRCA-TNBC subset, missense mutation was the most common variant type, and C > G was the dominant base substitution (Fig. 6D). CNV analysis further showed that PPP4C copy number gains and losses were present in both TNBC and THCA, with gains occurring more frequently in THCA (Fig. 6E). These CNV changes may contribute to inter-sample variability in PPP4C expression; however, their biological consequences were not directly tested in this study.

Hormonal regulation of PPP4C expression

To investigate whether PPP4C expression is modulated by hormones commonly implicated in TNBC and THCA, estradiol (E2) and liothyronine (T3) treatment experiments were conducted. In the TNBC cell lines MDA-MB-468 and MDA-MB-231, treatment with increasing concentrations of estradiol (0–1 μM) for 48 h induced modest and cell line-dependent fluctuations in PPP4C mRNA expression at selected concentrations; however, PPP4C protein levels remained largely unchanged by Western blot analysis (Fig. 7A–B). In the PTC cell lines TPC-1 and KTC-1, exposure to liothyronine for 48 h produced limited variation in PPP4C mRNA expression, whereas no corresponding consistent alterations in PPP4C protein abundance were detected (Fig. 7C–D). Under the experimental conditions tested, short-term hormonal stimulation did not induce stable changes in PPP4C protein expression in these TNBC and PTC cell models. These findings do not exclude indirect, longer-term, or context-dependent hormonal regulation, nor potential functional effects not captured by PPP4C protein abundance.

Fig. 7.

Hormonal regulation of PPP4C expression in TNBC and THCA cells. A, B qRT-PCR (A) and Western blotting (B) of PPP4C in MDA-MB-468 and MDA-MB-231 cells treated with estradiol (0–1 μM, 48 h). C, D qRT-PCR (C) and Western blotting (D) of PPP4C in TPC-1 and KTC-1 cells treated with liothyronine (0–1 μM, 48 h). Data are presented as mean ± SD from three independent biological replicates. One-way ANOVA was used. ns, p ≥ 0.05; *, p < 0.05; **, p < 0.01;***, p < 0.001, ****, p < 0.0001

Immune characteristics of PPP4C in the tumor microenvironment

The tumor microenvironment consists of immune cells, stromal components, and various molecular factors that influence tumor initiation, progression, metastasis, and response to therapy. We analyzed immune-related characteristics associated with PPP4C expression in BRCA and THCA using ESTIMATE and CIBERSORT. PPP4C expression was negatively correlated with immune scores in BRCA (Fig. 8A), whereas no significant association was observed in THCA (Fig. 8B). Treg-related signatures were correlated with PPP4C expression in both tumor types (Figure S2A–B).

Fig. 8.

Correlation of PPP4C expression with immune score, and stromal score in BRCA and THCA. A Correlation of PPP4C expression with Immune Score, Stromal Score, ESTIMATE Score in BRCA. B Correlation of PPP4C expression with Immune Score, Stromal Score, ESTIMATE Score in THCA

These observations are based on computational inference and should therefore be interpreted as associations rather than direct evidence of PPP4C-mediated immune modulation. Notably, the direction and magnitude of PPP4C–immune associations differed between BRCA and THCA, likely reflecting tumor-type–specific microenvironmental contexts, including baseline immune infiltration, stromal composition, and distinct immune-escape strategies, rather than a uniform effect of PPP4C across cancers. Further mechanistic studies will be required to determine whether PPP4C directly influences immune cell recruitment or function.

PPP4C and immune checkpoint genes in BRCA and THCA

Correlation analyses were conducted between PPP4C expression and immune checkpoint genes. Results indicated that in BRCA, high PPP4C expression was negatively correlated with certain immunosuppressive and immunostimulatory genes, while in THCA, PPP4C expression showed positive correlation with these genes (Fig. 9A). Data in Fig. 9A are displayed in radar charts (Fig. 9B–E). These distinct association patterns suggest that PPP4C may reflect tumor-type–specific immune states rather than exert a uniform immunoregulatory effect across cancers.

Fig. 9.

Associations between PPP4C expression and immune checkpoint genes in BRCA and THCA. A Expression correlation analysis between PPP4C and immune checkpoint-related genes in BRCA and THCA. B The association between PPP4C expression and immunosuppressive genes levels in BRCA. C The association between PPP4C expression and immunostimulatory genes levels in BRCA. D The association between PPP4C expression and immunosuppressive genes levels in THCA. E The association between PPP4C expression and immunostimulatory genes levels in THCA

PPP4C expression was also associated with broader sets of immunoregulatory genes, including chemokines, receptors, MHCs, and immune-activating or immunosuppressive genes (Figure S3). These genes are critical for balancing immune activation and suppression, maintaining immune homeostasis, and enabling coordinated immune responses. Dysregulation of these genes is associated with immune disorders such as autoimmune diseases and immunodeficiencies. Collectively, these findings highlight PPP4C as a candidate marker associated with immune-related features in different tumor types, while emphasizing that the observations are correlative and warrant further mechanistic investigation in immunotherapy-relevant contexts.

Knockdown of PPP4C inhibits the growth of TNBC and THCA cells

To validate PPP4C expression across breast cancer molecular subtypes, we measured PPP4C mRNA and protein levels in a panel of mammary cell lines, including the non-tumorigenic MCF-10A line as well as ER + (MCF-7), HER2 + (SK-BR-3), and TNBC (MDA-MB-468 and MDA-MB-231) cell lines. qRT-PCR and Western blotting showed that PPP4C was broadly expressed in breast cancer cell lines and was elevated in multiple breast cancer cell lines compared with MCF-10A (Fig. 10A–B). To investigate the functional role of PPP4C in tumor cell proliferation, we transiently knocked down PPP4C in TNBC and THCA cell lines using two independent siRNAs. Knockdown efficiency was quantified by densitometry (normalized to Tubulin) and is shown alongside the representative blots (Fig. 10C). CCK-8 assays demonstrated that PPP4C silencing significantly reduced proliferative capacity in both TNBC and THCA cells (Fig. 10D–E). Consistently, colony formation assays showed a marked decrease in clonogenic growth upon PPP4C knockdown (Fig. 10F–G). Collectively, these results indicate that PPP4C promotes the proliferative capacity of TNBC and THCA cells in vitro.

Fig. 10.

PPP4C promotes the proliferative capacity of breast cancer and THCA cells. A PPP4C mRNA expression in MCF-10A, MCF-7, SK-BR-3, MDA-MB-468, and MDA-MB-231 cells. B Western blotting of PPP4C in the indicated breast cell lines. C Western blotting of PPP4C knockdown efficiency in TNBC and THCA cells. D, E CCK-8 assays after PPP4C knockdown. F, G Colony formation assays after PPP4C knockdown. Data are presented as mean ± SD from three independent biological replicates. Two-way ANOVA was used for CCK-8 assays; one-way ANOVA followed by Dunnett's multiple-comparisons test was used for colony formation assays. *P < 0.05, **P < 0.01, ***P < 0.001

Effect of PPP4C knockdown on migration and invasion of TNBC and THCA cells

Transwell assays showed that transient knockdown of PPP4C with two independent siRNAs markedly impaired the migration and invasion of TNBC and THCA cells (Fig. 11A–D). To further evaluate migration, we performed wound-healing assays under the same siRNA-mediated PPP4C silencing conditions; the results revealed a significant reduction in wound closure compared with control siRNA (Fig. 11E–H). Collectively, these data indicate that PPP4C positively regulates both the proliferative and invasive capacities of TNBC and THCA cells in vitro.

Fig. 11.

PPP4C downregulation inhibits breast cancer and thyroid cancer cells proliferation and invasion in vitro. A-D Transwell migration assay showing the ability of PPP4C knockdown cells to migrate. Migration was quantified from at least three independent biological replicates, and at least three random fields per biological replicate were quantified. E–H Wound healing assay showing the migration ability of PPP4C knockdown cells. The wound area was measured from three independent biological replicates, with at least three random fields per biological replicate. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001

Identification and analysis of PPP4C-related transcription factors in TNBC and THCA

Using hTFtarget (https://guolab.wchscu.cn/hTFtarget), potential transcription factors for PPP4C were predicted. By intersecting the TFs with differentially expressed genes from TNBC and THCA datasets, 17 common transcription factors were identified: E2F1, FOS, ZFP42, PAX5, E2F7, ATF3, RUNX2, ETV1, KLF9, ESR1, RUNX3, AR, EGR1, JUN, MECOM, FOXA2, MYH11 (Fig. 12A). In TNBC, the expression patterns of candidate TFs are shown in Fig. 12B, and their correlation network is shown in Fig. 12C. Similarly, in THCA, candidate TF expression patterns are shown in Fig. 12D, and their correlation network is shown in Fig. 12E. These analyses suggest that PPP4C expression may be associated with a set of shared transcriptional regulators in TNBC and THCA. However, these TFs represent computationally predicted candidates, and further experimental studies will be required to determine whether they directly regulate PPP4C transcription.

Fig. 12.

Prediction of transcription factors for PPP4C. A Common transcription factors of PPP4C in TNBC and THCA. B Upregulated and downregulated transcription factors in TNBC. C The relationships between transcription factors in TNBC. D Upregulated and downregulated transcription factors in THCA. E The relationships between transcription factors in THCA. TFs were predicted using hTFtarget and intersected with differentially expressed genes from TNBC and THCA dataset

PPP4C expression and drug sensitivity

The sensitivity to three commonly used targeted therapeutic agents was predicted across different PPP4C expression subgroups. Afatinib, an irreversible ErbB-family inhibitor [9], and Foretinib, a multi-kinase inhibitor targeting c-MET/VEGFR-related signaling [10], were among the candidate agents showing PPP4C-associated differences in predicted sensitivity. In the TCGA–BRCA TNBC subset, Afatinib exhibited significantly different predicted sensitivity between the two PPP4C strata (Fig. 13A). In the TCGA–THCA cohort, Foretinib also showed a significant PPP4C-dependent difference in predicted drug sensitivity (Fig. 13B).

Fig. 13.

Experimental validation of PPP4C-associated drug sensitivity to Foretinib and Afatinib. A, B Predicted sensitivity to Afatinib in the TCGA-BRCA TNBC subset (A) and Foretinib in TCGA-THCA (B) stratified by PPP4C expression. C–F Dose–response assays in MDA-MB-468 (C), MDA-MB-231 (D), TPC-1 (E), and KTC-1 (F) cells after PPP4C knockdown. Dose–response curves were fitted using nonlinear regression, and IC50 values were calculated accordingly. G–J Colony formation assays with PPP4C knockdown and drug treatment (1 μM); representative images in (G, I) and quantification in (H, J). K, L Transwell assays with PPP4C knockdown and drug treatment (1 μM) in TNBC (K) and THCA (L) cells. M, N Wound-healing assays with PPP4C knockdown and drug treatment (1 μM) in TNBC (M) and THCA (N) cells. Data are presented as mean ± SD from three independent biological replicates (n = 3). Two-tailed Student's t-test was used for two-group comparisons; one-way ANOVA was used for multiple-group comparisons. *P < 0.05, **P < 0.01, ***P < 0.001

To experimentally validate these predictions, dose–response assays were performed following PPP4C knockdown. PPP4C depletion enhanced responsiveness to Afatinib in TNBC cells and showed a more modest trend with Foretinib in THCA cells (Fig. 13C–F). The calculated IC50 values for these assays are summarized in Supplementary Table 3. Afatinib and Foretinib were prioritized for deeper phenotypic validation because they showed robust predicted differences and produced measurable dose-dependent growth inhibition in our models. Several additional candidate compounds displayed PPP4C-associated differences in predicted response (Figure S4).

Based on these results, we further assessed phenotypic effects at a fixed drug concentration (1 μM), used as an in vitro working dose for phenotypic comparison. This fixed concentration was used as an in vitro working dose to enable cross-condition comparisons in phenotypic assays rather than to model clinically achievable exposure. PPP4C knockdown together with Afatinib or Foretinib suppressed clonogenic growth, migration/invasion, and wound closure more effectively than drug treatment alone (Fig. 13G–N and Figure S4). Collectively, these findings support a functional role of PPP4C in modulating drug response and suggest that PPP4C expression may represent a candidate biomarker associated with targeted therapy response, warranting further clinical validation.

Discussion

PPP4C has been implicated in several cancer-relevant biological processes, including DNA damage repair [11], mitotic regulation [12], microtubule organization [13], and cell migration [14]. Previous studies have shown that dysregulated PPP4C is associated with tumor progression and poor prognosis in multiple malignancies, including ovarian cancer [15], diffuse large B-cell lymphoma [16], pancreatic cancer [17], and lung adenocarcinoma [18]. Collectively, these findings support the biological plausibility of PPP4C as a cancer-associated phosphatase involved in proliferation, survival, and stress adaptation.

In the present study, we combined TCGA-based analyses with in vitro functional validation in THCA and TNBC models. Our data showed that PPP4C was consistently upregulated in both THCA and TNBC, and that higher PPP4C expression was associated with adverse clinicopathological features and poorer prognosis. Functional assays further demonstrated that PPP4C knockdown suppressed proliferation, migration, and invasion in TNBC and THCA cells, supporting a potential pro-tumorigenic role of PPP4C in these malignancies. Importantly, our computational analyses should be interpreted as association-based and hypothesis-generating rather than causal evidence.

Compared with previous studies that mainly examined PPP4C in single tumor types or in the context of general phosphatase biology [8, 15–18], our study extends the current literature in three aspects. First, we identified PPP4C as a shared, consistently upregulated molecular feature in both THCA and TNBC. Second, we integrated bioinformatic analyses with functional validation in representative cell models of both cancer types. Third, we provide preliminary evidence that PPP4C expression may be associated with differential responsiveness to selected targeted agents, thereby suggesting a potential role for PPP4C in therapy-response stratification in preclinical settings.

Our enrichment analyses suggested that PPP4C-associated networks are linked to cancer-relevant pathways, particularly cell-cycle/genome maintenance programs and signaling pathways such as PI3K–Akt, AMPK, and Hippo signaling. These findings are consistent with previous studies showing that PPP4C participates in mitotic regulation and DNA repair [11–14, 19, 20], and support the possibility that PPP4C contributes to tumor progression by coordinating proliferative and survival-related signaling programs. However, the precise downstream mechanisms of PPP4C in THCA and TNBC remain to be defined.

The immune analyses revealed tumor-type–specific associations between PPP4C expression and immune-related features. In the BRCA/TNBC context, PPP4C expression was negatively correlated with immune scores, whereas no significant association was observed in THCA. PPP4C also showed distinct correlation patterns with immune checkpoint genes in the two cancer types. These differences may reflect fundamental differences in baseline immune infiltration, stromal composition, and immune-escape programs between breast and thyroid tumors, which may cause PPP4C to associate with different immune-related states in distinct tumor contexts [21, 22]. In addition, overall immune scores and individual immune checkpoint gene correlations capture different dimensions of the tumor microenvironment [23], which may partly explain the non-identical association patterns observed in our analyses. Therefore, these findings should be interpreted cautiously as computational associations rather than direct evidence that PPP4C regulates immune activity. This interpretation is also consistent with previous reports linking PPP4C to immune-related features in ovarian cancer and lung adenocarcinoma [15, 18].

Our transcription factor analysis identified 17 shared candidate TFs potentially associated with PPP4C dysregulation in both TNBC and THCA. Among them, E2F1 is a well-established regulator of cell-cycle progression [24], FOS is a classic stress-responsive immediate-early transcription factor [25], and ESR1 and AR are central mediators of hormone-associated transcriptional regulation [26, 27]. RUNX2 has also been implicated in transcriptional programs linked to hormone-related and tumor-associated regulation [28]. These observations provide a possible transcriptional context for PPP4C dysregulation; however, the TF–PPP4C relationships identified here remain computational predictions and require direct experimental validation.

With regard to therapeutic response, PPP4C expression was associated with predicted sensitivity differences for several targeted agents, and experimental validation showed that PPP4C depletion enhanced responsiveness to Afatinib in TNBC cells and Foretinib in THCA cells. Afatinib is an irreversible ErbB-family inhibitor that has been explored in breast cancer and TNBC-related contexts [29], whereas Foretinib is a c-MET/multi-target kinase inhibitor, and multi-kinase targeting strategies are biologically relevant in thyroid cancer [30]. These features, together with the predicted sensitivity differences observed in our datasets and the measurable dose-dependent growth inhibition in our models, supported their prioritization for further phenotypic validation. Additional candidate compounds were also tested in Supplementary Figure S4, although their effects were more modest or inconsistent. These data suggest that PPP4C expression may be associated with relative drug resistance in vitro and may have value as a candidate biomarker for therapy-response stratification in preclinical settings. This is in line with previous evidence linking PPP4C to chemotherapy resistance-related phenotypes [31].

Finally, short-term estradiol or liothyronine stimulation induced only modest and inconsistent changes in PPP4C mRNA expression in some cell models, whereas PPP4C protein levels remained largely stable. This mRNA–protein discordance is biologically plausible, as transient transcriptional responses do not necessarily translate into measurable changes in steady-state protein abundance within the same time window because of post-transcriptional regulation, translational control, or protein stability. Therefore, we interpreted these findings cautiously and did not infer direct hormonal regulation of PPP4C at the protein level.

Several limitations should be acknowledged. First, most integrative analyses were based on retrospective TCGA transcriptomic data and therefore remain association-based. Second, functional validation was limited to in vitro experiments, without in vivo confirmation. Third, only two THCA cell lines (TPC-1 and KTC-1) and two TNBC cell lines (MDA-MB-468 and MDA-MB-231) were used in the present study, which may limit the generalizability of the observed PPP4C-related phenotypes across the heterogeneity of these cancers. Fourth, rescue experiments were not performed, and formal drug-synergy modeling was not conducted. Fifth, the transcription factor analysis was computational and was not validated experimentally. Taken together, these limitations indicate that further mechanistic studies, validation in additional models, and in vivo investigations are needed to clarify the regulatory role and translational relevance of PPP4C in THCA and TNBC.

Conclusion

In summary, our findings indicate that PPP4C is consistently upregulated and functionally relevant in both thyroid carcinoma (THCA) and triple-negative breast cancer (TNBC). Elevated PPP4C expression is associated with adverse clinicopathological features and poorer prognosis in these malignancies. Functional experiments support a pro-tumorigenic role of PPP4C in promoting cell proliferation, migration, and invasion in vitro, and suggest that PPP4C expression may be linked to differential responsiveness to selected targeted agents.

Taken together, PPP4C may represent a candidate molecular biomarker associated with tumor progression and therapeutic response in THCA and TNBC. However, further mechanistic studies and in vivo validation are required to clarify its regulatory networks and to determine its potential clinical utility.

Abbreviations

PPP4C

Protein Phosphatase 4 Catalytic Subunit

THCA

Thyroid Cancer

TNBC

Triple-Negative Breast Cancer

TCGA

The Cancer Genome Atlas

GTEx

Genotype-Tissue Expression

IHC

Immunohistochemistry

CCK-8

Cell Counting Kit-8

PPI

Protein-protein interaction

KEGG

Kyoto Encyclopedia of Genes and Genomes

GO

Gene Ontology

TME

Tumor microenvironment

KM

Kaplan-Meier

IC50

Half-maximal inhibitory concentration

Authors’ contributions

Conceptualization, Maolin Yan, and Wenjun Xie; Data curation, Huashui Li, Yifan Chen, Guodong Guo, and Qiang Lin; Formal analysis, Huashui Li, Yifan Chen, Qinghua Luo, and Guodong Guo; Funding acquisition, Wenjun Xie, and Yifan Chen; Investigation, Maolin Yan, Qiang Lin, and Guodong Guo; Methodology, Wenjun Xie, Qinghua Luo, and Huashui Li; Project administration, Maolin Yan; Resources, Wenjun Xie, Huashui Li, Qiang Lin, and Guodong Guo; Software, Wenjun Xie, Qinghua Luo, and Huashui Li; Supervision, Maolin Yan; Validation, Wenjun Xie, Huashui Li, and Yifang Chen; Visualization, Wenjun Xie, Huashui Li, and Yifang Chen; Writing – original draft, Wenjun Xie, Huashui Li, and Yifang Chen; Writing – review & editing, all authors.

Funding

This work was supported by grants from Joint Fund Project for Science Research and Technology Innovation of Fujian Provincial Department of Science and Technology (2024Y9038), Fujian Provincial Natural Science Foundation (2024J08254), and Fujian Provincial Natural Science Foundation General Project (2023J011193).

Data availability

The datasets generated during this study are available from the corresponding author upon reasonable request. Public data sources: TCGA (https://portal.gdc.cancer.gov/), GTEx (https://gtexportal.org/) and HPA (https://www.proteinatlas.org/).

Declarations

Ethics approval and consent to participate

This study was approved by the Ethics Committee of Fujian Provincial Hospital (K2023-03–016) and conducted in accordance with the Declaration of Helsinki. All participants provided written informed consent. Tissue samples were anonymized before analysis.

Consent for publication

Not applicable. This study used anonymized tissue samples without identifiable personal information.

Competing interests

The authors declare no competing interests.