PPA1 is overexpressed in epithelial cells of CRC and correlates with patient prognosis
To explore tumor-specific transcriptional alterations between CRC cells and normal intestinal epithelial cells, we obtained the GSE200997 dataset from the GEO database, encompassing scRNA-seq data from 16 CRC tissues and 7 paired adjacent normal tissues. Following rigorous quality control (cells with>100 and <4000 measured genes, cells with <25% mitochondrial contamination) (Fig. S1A), we performed dimensionality reduction via PCA and UMAP, stratifying cells into nine distinct clusters (Fig. 1A). Cluster annotation was validated using canonical markers: B cells (CD79A, CD19), Plasma cells (CD79A, MZB1, IGHA1), T cells (CD3D, CD3E, IL7R), NK cells (GNLY, KLRD1, NKG7), Endothelial cells (PECAM1, VWF, CD34), Epithelial cells (EPCAM, KRT8), Fibroblasts (ACTA2, COL1A1, COL1A2), Myeloid cells (CD14, LYZ, CD68, CD163), Mast cells (KIT, MS4A2)(Fig. S1B) [31]. The "FindAllMarkers" function was employed to explore the DEGs between epithelial cells of CRC tissues and epithelial cells of adjacent normal tissues and a total of 437 DEGs met the thresholds of |log2foldchange | > 0.585 and adjusted p-value < 0.05 (Fig. S1C). To delineate metabolic reprogramming in CRC cells, we performed KEGG enrichment analyses using the "clusterProfiler" R package on 63 metabolism-associated DEGs. Strikingly, 20 DEGs were enriched in OXPHOS pathway, with majority demonstrating elevated expression in tumor-derived epithelial cells (Fig. 1B, Fig. S1D), which contrary to canonical Warburg effect [32]. Moreover, among the 20 OXPHOS-associated DEGs, PPA1 emerged as the most significantly upregulated in epithelial cells of CRC tissues compared to epithelial cells of adjacent normal tissues (Fig. 1C, Fig. S1D, E). Simultaneously, analysis of bulk-RNAseq data from the TCGA and GTEx databases revealed significantly higher PPA1 expression in colorectal cancer (CRC) tissues compared to normal colon tissues (Fig. 1D).
To further validate expression differences of PPA1 between CRC and adjacent normal tissues, we collected 48 paired CRC and adjacent normal tissues. The results of Western blotting and qPCR demonstrated that expression of PPA1 significant upregulated in CRC tissues versus matched adjacent normal tissues (Fig. 1E, F). In addition, immunohistochemical staining showed that more epithelial cells in CRC were positive for PPA1 compared with matched adjacent normal tissues (Fig. 1G). Next, Western blotting was used to detect the protein expression levels of PPA1 in normal intestinal epithelial cells (NCM460) and CRC cell lines (HCT8, HT116, SW480, SW620, and HCT29). As shown in Fig. 1H, all the expressions of PPA1 in the five CRC cells were remarkably higher than that in NCM460 cells.
Finally, to precisely validate the expression of PPA1 in epithelial cells of CRC and its correlation with clinical characteristics of CRC patients, we obtained a tissue microarray (TMA) containing 80 pairs of CRC and adjacent normal tissues. Dual immunofluorescence co-staining using EPCAM as an epithelial lineage marker was performed to evaluate PPA1 expression in epithelial cells of CRC. As shown in Fig. 1I and Fig. S1F, the mean fluorescence intensity (MFI) of PPA1 in epithelial cells of CRC tissues was significantly higher than that in the adjacent normal tissues. Patients with CRC were stratified into PPA1-high and PPA1-low groups based on the median expression level of PPA1. Combined analysis of PPA1 expression and the clinical data revealed that the CRC patients in PPA1-high group exhibited higher T stage (p = 0.0033; Fig. S1G), lymph node metastasis rate (p < 0.0001; Fig. S1H), and TNM stage (p < 0.0001; Fig. S1I), whereas PPA1 expression showed no correlation with age (p = 0.2318), gender (p = 0.2337), or M stage (p = 0.2862) of patients (Table S2). Kaplan-Meier survival analysis indicated that the overall survival rate of CRC patients in the PPA1-high group was significantly lower than that in the PPA1-low group (Fig. S1J).
In summary, we found that PPA1 was highly expressed in epithelial cells of CRC patients and it was associated with the survival rate of CRC patients. Therefore, we believe that PPA1 affects the malignant behavior of CRC and may be a potential target for the treatment of CRC patients.
To investigate the effects of PPA1 on the biological functions of CRC cells, we transfected HCT8 and HCT116 cells, which had the highest expression of PPA1, with ShRNA knockdown lentivirus for PPA1 and screened three stable cell lines including the control group (Sh-NC) and the knockdown group (Sh1-PPA1, Sh2-PPA1) cell lines.
Through CCK-8 and colony formation assays, we found that PPA1 knockdown significantly inhibited the proliferation and colony-forming ability of CRC cells compared to control group (Fig. 2A-D). Furthermore, under glucose restriction condition (0.25 mM glucose), PPA1 knockdown more strongly suppressed these effects (Fig. S2A, B). To investigate the metastatic potential of CRC cells following PPA1 knockdown, we performed Transwell migration and invasion assays. As shown in Fig. 2E, F, the number of migrating and invading cells in the Sh1-PPA1 and Sh2-PPA1 groups was significantly less than that in the Sh-NC group. At the same time, the wound healing assay further revealed that compared with the Sh-NC group, the migration rates of cells in the Sh1-PPA1 and Sh2-PPA1 groups were significantly decreased (Fig. 2G, H). Consistently, under glucose-restricted conditions, PPA1 knockdown further suppressed the migratory and invasive abilities of CRC cells (Fig. S2C-E). Finally, we verified the expressions of E-cadherin and N-cadherin, two EMT markers, in HCT8 and HCT116 cells by Western blotting. As shown in Fig. 2I, J, following the knockdown of PPA1, E-cadherin expression was upregulated markedly, whereas N-cadherin expression was downregulated significantly. These findings collectively demonstrate that PPA1 silencing attenuates proliferation, suppresses metastatic behaviors, and reverses EMT progression of CRC cell.
To functionally validate the tumor-promoting role of PPA1, lentiviral overexpression constructs encoding PPA1 were transduced into HCT8 and HCT116 cells. Similarly, overexpression of PPA1 promoted the proliferation and colony-forming ability of CRC cells, and this promoting effect was more significant under glucose restriction conditions (0.25 mM glucose) (Fig. 3A-D, Fig. S2F, G). Moreover, PPA1 overexpression significantly enhanced the migration and invasion capacities of CRC cells in the OE-PPA1 group compared to the OE-NC group, as demonstrated by Transwell assay and wound healing assay (Fig. 3E-H). Consistently, under glucose-restricted conditions (0.25 mM glucose), overexpression of PPA1 further enhanced the pro-invasive and migratory effects of CRC cells (Fig. S2H-J). These findings suggest that PPA1 may regulate proliferation, migration and invasion of CRC cells through glucose metabolism-related pathways. We further examined the alterations in PPA1 protein levels under glucose-restricted condition and found that the expression of PPA1 was significantly upregulated in this context (Fig. S2K, L). In addition, PPA1 overexpression significantly suppressed E-cadherin expression and enhanced N-cadherin expression, indicative of EMT progression (Fig. 3I, J).
Collectively, the above findings demonstrate that PPA1 functionally drives multiple oncogenic processes in CRC cells, and this oncogenic effect is amplified under glucose-restricted conditions.
While the aforementioned cellular functional assays provide preliminary validation of PPA1's critical role in CRC, the precise molecular mechanisms remain elusive. Notably, under glucose restriction conditions (0.25 mM glucose), both PPA1 knockdown and overexpression demonstrated more pronounced inhibitory or promotive effects on proliferation, migration, and invasion of CRC, respectively. These glucose-dependent phenotypic modulations strongly suggest that PPA1 regulates biology of CRC through glucose-related metabolism pathways, potentially involving metabolism reprogramming and energy homeostasis control.
Hence, we performed metabolomic profiling to assess metabolic alterations in HCT116 cells between Sh-PPA1 and Sh-NC groups cultured under glucose restriction conditions for 24 h. The KEGG analysis revealed that differentially altered metabolites were predominantly enriched in pathways including purine metabolism, amino acid metabolism, OXPHOS, and the tricarboxylic acid (TCA) cycle (Fig. 4A). Notably, OXPHOS-associated metabolites -- adenosine diphosphate (ADP), nicotinamide adenine dinucleotide (NAD + ), and succinate -- exhibited significantly lower levels in the Sh-PPA1 group compared to the Sh-NC group (Fig. 4B-D). Concurrently, we employed the "scMetabolism" R package to calculate metabolic pathway activity in epithelial cells from CRC tissues and adjacent normal tissues within the GSE20997 dataset. The results demonstrated that epithelial cells of CRC tissues exhibited significantly elevated activity in both OXPHOS and glycolytic pathways compared with their normal counterparts (Fig. 4E). These findings align with recent studies identifying a marked OXPHOS propensity in CRC cells to meet their substantial energetic demands, establishing OXPHOS as a crucial energy source in colorectal carcinogenesis [19, 20, 33]. Collectively, we believe that PPA1 may modulate the biological behaviors of CRC cells through the OXPHOS pathway.
Next, we evaluated OXPHOS activity in HCT116 and HCT8 cells after 24 h low-glucose culture using the Seahorse Mito Stress Test. In both cell lines, PPA1 knockdown significantly attenuated OXPHOS activity, with the Sh-PPA1 group showing reduced basal OCR and ATP production compared to the Sh-NC group. Conversely, PPA1 overexpression markedly enhanced OXPHOS activity, as evidenced by elevated OCR and ATP production in the OE-PPA1 group relative to the OE-NC group (Fig. 4F-I). Parallel measurements of intracellular ATP levels, NAD + , NAD + /NADH ratios, and mitochondrial respiratory chain complex I activity under low-glucose conditions revealed consistent patterns. The Sh-PPA1 group exhibited significantly decreased ATP production, diminished NAD + /NADH ratios, and suppressed complex I activity compared to Sh-NC group (Fig. 4J-N), corroborating our metabolomic findings. In contrast, PPA1 overexpression substantially increased these metabolic parameters in both CRC cell lines (Fig. 4J-N). We also found that the alteration in the NAD⁺/NADH ratio was due specifically to changes in NAD⁺ levels, indicating that PPA1 affects only NAD⁺ production without significantly influencing NADH in both CRC cell lines (Fig. 4L-N). Finally, we investigated the impact of PPA1 on mitochondrial ROS using MitoSOX™ Red fluorescence staining. Quantitative analysis revealed significantly higher mitochondrial ROS levels (measured by mean fluorescence intensity, MFI) in Sh-PPA1 group compared to Sh-NC group, whereas PPA1-overexpressing cells showed markedly reduced MFI relative to OE-NC groups (Fig. 4O). FCM analysis confirmed this bidirectional regulation: PPA1 knockdown substantially increased mitochondrial ROS accumulation, while PPA1 overexpression effectively suppressed ROS generation (Fig. 4P, Q).
In conclusion, our findings demonstrate that PPA1 modulates the malignant behaviors of CRC cells through regulation of the OXPHOS pathway.
To elucidate the specific mechanisms by which PPA1 regulates OXPHOS, we conducted integrated proteomic and phosphoproteomic profiling of HCT116 cells (Sh-NC vs. Sh-PPA1 groups) under low-glucose culture conditions. Phosphoproteomic analysis identified 665 differentially phosphorylated sites (Fig. S3A), mapping to 154 distinct proteins. These phosphoproteins revealed predominant enrichment in mitophagy and canonical autophagy pathways (Fig. S3B). Within the mitophagy pathway, ULK1 exhibited significantly reduced phosphorylation at three critical sites -- Ser467, Ser556, and Ser638 -- upon PPA1 knockdown, accompanied by diminished phosphorylation at Ser17 of FUNDC1 (Fig. S3C). ULK1 serves as the master initiator of autophagosome formation and directly coordinates mitophagy through phosphorylation cascades [34, 35]. FUNDC1, a pivotal receptor for ubiquitin-independent mitophagy, resides on the mitochondrial outer membrane where it recruits LC3 via direct binding to initiate mitophagy [36]. Crucially, ULK1-mediated phosphorylation enhances FUNDC1 phosphorylation at Ser17, establishing a hierarchical signaling axis that triggers mitophagy [37]. Our findings position PPA1 as an upstream regulator of this phospho-regulatory network, promoting ULK1/FUNDC1 phosphorylation to activate mitophagy.
To validate these findings, we conducted a series of functional assays of CRC ells under glucose restriction. As demonstrated in Fig. 5A and Fig. S3D, G, PPA1 knockdown did not significantly alter total protein expression levels of ULK1 or FUNDC1. However, it caused marked reductions in phosphorylation at three critical sites (Ser467, Ser556, and Ser638) of ULK1and the Ser17 site of FUNDC1 in the PPA1-Sh1 and PPA1-Sh2 groups compared to Sh-NC group. Concomitantly, the LC3II/LC3I ratio -- a key indicator of autophagic flux -- was substantially diminished in both PPA1-Sh1 and PPA1-Sh2 groups relative to the Sh-NC group (Fig. 5A, B, Fig. S3E, H).
To whether mitophagy levels were altered, we analyzed examined changes in LC3II and LC3I expression in mitochondrial and cytosolic fractions of HCT8 and HCT116 cells under glucose restriction. We found that LC3II protein aggregated on mitochondria was significantly reduced in PPA1-Sh1 and PPA1-Sh2 groups compared to the Sh-NC group across both cell lines (Fig. 5C, D, Fig. S3F, I), indicating impaired recruitment of LC3II -- a critical step in mitophagy initiation. Confocal microscopy revealed diminished LC3 puncta formation and attenuated colocalization with MitoTracker-labeled mitochondria in PPA1-Sh1 and PPA1-Sh2 groups (Fig. 5E, Fig. S3J-M). Then, we directly observed mitochondrial changes in CRC cells of each group by transmission electron microscopy (TEM). Under low-glucose culture conditions, cells in the Sh-NC group exhibited numerous autophagosomes containing mitochondria and autolysosomes with digested mitochondrial remnants. In contrast, cells in the PPA1-Sh1 and PPA1-Sh2 groups displayed markedly fewer autolysosomes, with the majority of mitochondria showing swelling and structural condensation -- morphological hallmarks of mitochondrial dysfunction that may predispose cells to apoptosis (Fig. 5F). Finally, we assessed mitochondrial membrane potential (ΔΨm) across each group using JC-1 fluorescent probe. As shown in Fig. S3N, the Sh-NC group exhibited a significantly higher MFI ratio of JC-1 aggregates (red fluorescence) to JC-1 monomers (green fluorescence) compared with PPA1-Sh1 and PPA1-Sh2 groups. FCM analysis confirmed this pattern, demonstrating that PPA1 knockdown substantially reduced ΔΨm (Fig. 5G). Severe ΔΨm dissipation disrupts mitochondrial functionality, ultimately triggering cell death [38]. We subsequently quantified apoptosis rates via FCM after 24 h glucose restriction. Both CRC cell lines showed significantly elevated apoptosis in PPA1-Sh1 and PPA1-Sh2 groups versus Sh-NC group (Fig. 5H).
In summary, we demonstrate that PPA1 knockdown suppresses ULK1-FUNDC1-mediated mitophagy in CRC cells under low-glucose culture, thereby inducing mitochondrial dysfunction and ultimately promoting apoptosis.
To identify which ULK1 phosphorylation site(s) regulate phosphorylation of FUNDC1 at Ser17 site under glucose restriction, the cells in the Sh-PPA1 group were transfected with either wild-type ULK1 overexpression wild-type plasmid (ULK1 WT group) or phosphorylation-deficient mutants targeting specific residues (S467A or S556 A, or S638A group). As shown in Fig. 6A, D, and G, phosphorylation levels at Ser467 or Ser556 or Ser638 site of ULK1 were significantly reduced in each mutant group compared to the ULK1 WT group. Furthermore, we observed that the phosphorylation level of FUNDC1 at the Ser17 site was markedly decreased in S467A and S556 A groups relative to the ULK1 WT group, whereas no significant change was detected in the S638A group (Fig. 6A, D, G). Similarly, the LC3II/LC3I ratio was substantially reduced in S467A and S556A groups compared to ULK1 WT, while the S638A group showed no alteration (Fig. 6A, E, H). In addition, we observed significantly diminished LC3II enrichment on mitochondria in both S467A and S556A groups compared to the ULK1 WT group, whereas LC3II accumulation in the S638A group remained unaltered. (Fig. 6B, C, F, I). Immunofluorescence analysis revealed significantly reduced LC3 puncta accumulation and diminished colocalization with mitochondria in S467A and S556A mutants compared to ULK1 WT and S638A groups (Fig. 6J). Consistently, JC-1 probe analysis demonstrated a substantially decreased MFI ratio of JC-1 aggregates (red fluorescence) to JC-1 monomers (green fluorescence) in S467A and S556A mutants versus ULK1 WT and S638A groups (Fig. 6K), indicating pronounced ΔΨm dissipation.
In conclusion, we demonstrate that under glucose restriction, phosphorylation of ULK1 at Ser467 and Ser556 sites specifically induces phosphorylation of FUNDC1 at Ser17 site, thereby promoting mitophagy in CRC cells.
Next, we further investigated whether pharmacological activation of ULK1 using its agonist LYN-1604 could directly induce mitophagy of CRC cells under glucose restriction.
As shown in Fig. 7A, D, and G, phosphorylation levels of ULK1 at Ser467 and Ser556 sites were significantly elevated in both HCT8 and HCT116 cells treated with Sh-PPA1 and LYN-1604(1 μM) compared to the Sh-PPA1 + DMSO group under glucose restriction. Similarly, the phosphorylation level of FUNDC1 at Ser17 site and the LC3II/LC3I ratio were markedly increased in Sh-PPA1 + LYN-1604 group (Fig. 7A, E, H). Mitochondrial fraction analysis revealed significantly enhanced mitochondrial recruitment of LC3II in the Sh-PPA1 + LYN-1604 group compared to the Sh-PPA1 + DMSO group (Fig. 7B, C, F, I). Confocal microscopy further demonstrated enhanced LC3 puncta formation and increased colocalization with MitoTracker-labeled mitochondria in Sh-PPA1 + LYN-1604 group (Fig. 7J, Fig. S4A-D). These results collectively validate that LYN-1604 effectively rescues PPA1 knockdown-induced deficits in ULK1 phosphorylation and mitophagy.
We further investigated whether ULK1/FUNDC1-mediated mitophagy exerts protective effects on OXPHOS functionality. Seahorse Mito Stress assays demonstrated that Sh-PPA1 + LYN-1604 group in both CRC cell lines exhibited significantly higher OXPHOS activity compared to Sh-PPA1 + DMSO group, with marked increases in basal OCR and ATP production (Fig. 7K-M). FCM analysis further revealed substantially reduced ROS accumulation in Sh-PPA1 + LYN-1604 group versus the Sh-PPA1 + DMSO group (Fig. 7N, O).
Finally, the effects of ULK1 agonist LYN-1604 on the malignant behaviors of CRC cells under glucose restriction were assessed by a series of functional assays. CCK-8 assay demonstrated significantly increased proliferative capacity in Sh-PPA1 + LYN-1604 group compared to Sh-PPA1 + DMSO and Sh-PPA1 groups in both HCT8 and HCT116 cells (Fig. S4E). Colony formation assay revealed a higher number of colonies in the Sh-PPA1 + LYN-1604 group than in Sh-PPA1 + DMSO and Sh-PPA1 groups (Fig. S4F). Wound healing assays showed greater migration distances and rates in Sh-PPA1 + LYN-1604-treated CRC cells compared to controls (Fig. S4G, H). Transwell assays further confirmed increased numbers of migrated and invaded cells in the Sh-PPA1 + LYN-1604 group relative to Sh-PPA1 + DMSO and Sh-PPA1 groups (Fig. S4I, J).
Collectively, these findings demonstrate that under glucose restriction, LYN-1604-activated ULK1 effectively promotes mitophagy, preserves OXPHOS functionality, and consequently drives malignant progression of CRC cells.
AMPK and mTOR represent classical upstream pathways regulating ULK1 phosphorylation [39]. To further investigate whether PPA1-mediated regulation of ULK1 depends on AMPK or mTOR signaling, we examined both protein expression and phosphorylation levels of mTOR and AMPK. As shown in Fig. 8A, B, PPA1 knockdown did not alter AMPK protein expression, but significantly reduced its phosphorylation at Thr172 site -- a critical site governing ULK1 phosphorylation. In contrast, neither protein level nor phosphorylation status of MTOR was affected (Fig. 8A). These results suggest that PPA1 modulates ULK1 phosphorylation potentially through regulating AMPK phosphorylation rather than via the MTOR pathway.
Next, we employed the AMPK agonist GSK621 to investigate whether AMPK activation could rescue the downstream effects induced by PPA1 knockdown. As demonstrated, GSK621 (30 μM) treatment markedly enhanced phosphorylation of AMPK at Thr172 site in both cell lines (Figs. 8C, S5A, B). Subsequently, phosphorylation levels at Ser467 and Ser555 sites of ULK1 and Ser17 site of FUNDC1 were correspondingly elevated (Fig. 8C, Fig. S5A, B). As expected, total autophagy (LC3II/LC3 I ratio) and mitophagy levels (LC3II level in mitochondrion) were significantly increased accordingly (Fig. 8C, D, Fig. S5C-F). Finally, we similarly observed that the Sh-PPA1 + GSK621 group exhibited increased LC3 puncta formation and enhanced co-localization with MitoTracker-labeled mitochondria compared to the Sh-PPA1 + DMSO group (Fig. 8E, Fig. S5G-J).
Finally, we also investigated the impact of the AMPK agonist GSK621 on the malignant behavior of CRC cells. As shown in Fig. S6A, B and C, the cells in Sh-PPA1 + GSK621 group exhibited significantly enhanced proliferation and colony formation capabilities compared to the Sh-PPA1 + DMSO group. Transwell assays further indicated a marked increase in the number of invading and migrating cells in the Sh-PPA1 + GSK621 group relative to Sh-PPA1 + DMSO group (Fig. S6D, E). Similarly, wound healing assays demonstrated that GSK621 treatment significantly promoted the migratory capacity of both CRC cell lines (Fig. S6F, G).
Collectively, these results indicate that PPA1 modulates ULK1-dependent mitophagy and enhances the malignant behaviors -- including proliferation, migration, and invasion -- in CRC cells via AMPK.
To further validate the impact of PPA1 on CRC biology in vivo, we established subcutaneous xenograft and hepatic metastasis models in nude mice.
Stably transfected HCT116 cells (Sh-PPA1, Sh-NC, OE-NC, and OE-PPA1 groups) were subcutaneously injected into nude mice. Tumor volumes were measured on days 5, 10, 15, 20, and 25. Growth curve analysis revealed that PPA1 knockdown significantly inhibited subcutaneous tumor growth, with the Sh-PPA1 group exhibiting smaller tumor volumes and lower weights compared to the Sh-NC group. Conversely, PPA1 overexpression markedly promoted tumor growth, as evidenced by larger tumor volumes and higher weights in the OE-PPA1 group versus the OE-NC group (Fig. 9A-C). Western Blotting analysis of tumor tissues revealed elevated E-cadherin and decreased N-cadherin expression in the Sh-PPA1 group compared to the Sh-NC group, whereas the OE-PPA1 group exhibited the opposite trend compared to the OE-NC group (Fig. 9D, E). A CRC liver metastasis model was established via splenic injection of stably transfected HCT116 cells (Sh-PPA1 and Sh-NC groups). Macroscopic examination showed fewer metastatic CRC lesions in the livers of the Sh-PPA1 group compared to the Sh-NC group (Fig. 9F). H&E staining further confirmed a reduced number of metastatic foci in the Sh-PPA1 group (Fig. 9G). As demonstrated in Fig. 9H, I, following the knockdown of PPA1, E-cadherin expression was upregulated markedly, whereas N-cadherin expression was downregulated significantly in CRC metastatic foci. The above results confirm that PPA1 can also promote the growth and metastasis of CRC cells in vivo.
In summary, PPA1 serves as a critical cytoprotective protein in CRC cells under glucose restriction. Under such metabolic stress, PPA1 promotes phosphorylation of AMPK at Thr172 site and thereby facilitates phosphorylation of ULK1 at Ser467 and Ser555 sites. These phosphorylation events specifically induce phosphorylation of Ser17 on FUNDC1, a mitophagy receptor localized at the mitochondrial outer membrane. Phosphorylated FUNDC1-Ser17 recruits LC3II proteins to initiate mitophagy. This selective mitochondrial clearance mechanism maintains intracellular homeostasis and mitochondrial pool stability, thereby preserving OXPHOS functionality to generate sufficient energy for driving CRC malignant behaviors -- including proliferation, migration, and invasion (Fig. 10).