Previous research results have shown that the nuclear localization sequence of ZNF32 may be located between Aa170 and Aa228. Among the previously constructed ZNF32 mutants, we found that the double-site mutations H^{179, 183}A will form NSs in 293T cells (Wei et al., 2016). To study the exact roles of ZNF32 NSs caused by mutations of histidine 179 and 183 in breast cancer cells, we first constructed green fluorescence protein (GFP) fusion expression plasmids with H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A of ZNF32 for transfection to breast cancer cells using ZR-75-30 breast cancer cells overexpressing ZNF32 (wild-type or WT) as the control. As expected, the three mutant cells (H¹⁷⁹A, H¹⁸³A, H^{179, 183}A) all showed ZNF32 NSs while the WT cells did not (Figure 1). In addition, we performed the same experimental verifications on two other breast cancer cell lines, namely MCF-7 and MDA-MB-231, and the results were consistent with those obtained with the ZR75-30 line (Supplementary Figure S1). Therefore, we speculate that ZNF32 H¹⁷⁹A, H¹⁸³A, H^{179, 183}A can lead to formation of NSs in breast cancer cells.

To better understand the causes of ZNF32 NS formation, lentiviral vectors were used to construct breast cancer cell lines with stable mutations at these sites, and the increases in ZNF32 expressions compared with endogenous levels were detected by reverse transcription quantitative real-time polymerase chain reaction (RT-qPCR) (Supplementary Figure S2). RNA-seq was conducted to analyze the DEGs associated with NS formation. ZNF32 H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A were compared with the control group, and three groups of DEGs were obtained after screening. Accordingly, a total of 1,414 upregulated and 1,379 downregulated DEGs were screened in the WT vs. H¹⁷⁹A, a total of 1,431 upregulated and 1,290 downregulated DEGs were screened in the WT vs. H¹⁸³A, and a total of 1,480 upregulated and 1,329 downregulated DEGs were screened in the WT vs. H^{179, 183}A (Figure 2A). To obtain DEGs with the same expression trends, we found the intersection of the three groups of upregulated DEGs and obtained 1060 genes. Similarly, the intersection of the three groups of downregulated DEGs showed 961 genes (Figure 2B). Thus, a total of 2021 DEGs with the same up-down-regulation trends were obtained. The details of these DEGs and their enrichment in each database are presented in Supplementary Table S1. The Kyoto Encyclopedia of Genes and Genomes orthology (KOG) enrichment analysis was performed on these DEGs, and it was found that the genes were mainly enriched in terms of functional classifications, such as signal transduction mechanisms, posttranslational modifications, transcription, as well as amino acid transport and metabolism (Figure 2C). Gene ontology (GO) enrichment analysis showed that the molecular functions of the DEGs mainly included growth factor activity, calcium binding, ATPase activity, transcription activator activity, ion-channel binding, and RNA polymerase II transcription factor activity (Figure 2D). Therefore, NS formation may affect cell proliferation, and the DEGs enriched for growth factor activity (GO:0008083) are shown in (Figure 2E). It has been reported that the appearance of NSs is related to the transcriptional activity of RNA polymerase II (Wei et al., 1999, Bregman et al., 1995). Therefore, we speculate that the DEGs related to RNA polymerase II (GO:0001228; GO:0004879) may play important roles in the formation of NSs, as shown in (Figure 2F). In addition, the DEGs in the KEGG pathway showed that the most significantly enriched pathways were those of glycine, serine, and threonine metabolisms, including the pathways related to cancer, MAPK, PI3K-Akt, Rapl, and RAS signaling, which are closely related to the proliferation of cancer cells. Together, these indicate that ZNF32 NS formation maybe related to the transcriptional activity of RNA polymerase II and growth factor activity and that these may affect the growth of breast cancer cells.

Metabolomics was used to further study the functions of ZNF32 H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A that cause NSs. We compared the ZNF32 H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A groups with the control WT group and screened out the differentially expressed metabolites (DEMs). To obtain DEMs with the same expression trends, we considered the intersections of the three groups to obtain three common upregulated and 21 common downregulated DEMs (Figure 3A, B). We plotted the set of screened differential metabolites as a cluster heatmap for display, and we believe that these 24 DEMs are likely to be related to the functions of the NSs (Figure 3C). The details of these DEMs and their enrichment in each database are presented in Supplementary Table S2. We also conducted a joint transcriptomics and metabolomics analysis to compare the KEGG pathways enriched by the 2021 DEGs obtained from the transcriptome analysis with those enriched by the 24 DEMs from the metabolome analysis. The DEMs were enriched in eight pathways and overlapped with 321 pathways enriched in terms of DEGs (Figure 3D). These eight signal pathways include choline metabolism in cancer, glycerophospholipid metabolism, beta-alanine metabolism, pantothenate and CoA biosyntheses, and arginine and proline metabolisms. A total of 15 DEGs and 3 DEMs were enriched in the choline metabolism pathway. Among these, the upregulated DEGs are PIK3R2, PDGFA, PLA2G4C, PDGFB, EGFR, and SLC22A4, while the downregulated DEGs are AC007192, FOS, PLA2G4A, PLPP3, PLD1, SLC44A2, RAC3, PDGFRB, and PIK3R3. The expression levels of the three DEMs, namely LysoPC (15:0), LysoPC (16:0), and LysoPC (17:0), were all downregulated. These results indicate that the formation of ZNF32 NSs is accompanied by changes in choline metabolism in cancer.

The results of the omics analyses indicate that NS formation is related to the transcriptional activity of RNA polymerase II and also causes changes in several cancer-related metabolic pathways. Studies have shown that abnormal choline metabolism is related to the growth, differentiation, invasion, and metastasis of cancer cells (Glunde et al., 2011; Cao et al., 2016). Previous works have reported that RNA polymerase II activity is associated with cell proliferation (Vervoort et al., 2022; Huang and Ji, 2023; Giakountis et al., 2017). Hence, we hypothesized that ZNF32 H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A causing NSs may lead to concomitant changes in cell proliferation. Consistent with this notion, ZNF32 H¹⁷⁹A and H¹⁸³A significantly increased the numbers of EdU positive cells (Figure 4A, B), but there was no statistical difference between the ZNF32 H^{179, 183}A and WT groups (Figure 4A, B). In addition, the results of the MTT assay (Figure 4C) and crystal violet staining (Figure 4D, E) were consistent with the EdU staining results. The above findings suggest that ZNF32 single-site mutations (H¹⁷⁹A, H¹⁸³A) can promote the proliferation of breast cancer cells.

To verify the consistency of the differential regulation of tumor cell proliferation by ZNF32 histidine 179 and 183 single-site and double-site mutations in vitro and in vivo, we constructed a subcutaneous xenograft tumor model in nude mice. Compared with the WT group, the tumor volumes in the ZNF32 histidine 179 and 183 single-site mutation groups were significantly higher while no significant change was noted in the ZNF32 H^{179, 183}A group (Figure 5A, B). Similarly, the tumor formation rates were higher in the ZNF32 H¹⁷⁹A and H¹⁸³A groups but not significant in the ZNF32 H^{179, 183}A group compared to the WT group (Figure 5C). Consistent with these results, the mRNA extracted from the above tumor tissues were used as the template for PCR amplification, and the product sequencing results showed that the corresponding site mutations of ZNF32 histidine were indeed present in the tumor cells (Figure 5D). Overall, these data indicate that ZNF32 histidine 179 and 183 single-site mutations can promote tumor growth in vivo.

As shown above, ZNF32 H¹⁷⁹A, H¹⁸³A, and H^{179, 183}A differentially regulate breast cancer cell proliferation in vivo and in vitro, and we explored whether the single-site mutations could regulate proliferation through specific regulation of the downstream gene expressions. We then analyzed the transcriptome sequencing data and found the DEGs in the single-site mutation groups to explain the stronger proliferation abilities of cancer cells versus the WT and H^{179, 183}A groups. According to our analysis results, there were nine upregulated (Figure 6A) and four downregulated (Figure 6B) differential genes. After removing the new and low-expression genes, three upregulated differential genes were found to be related to cell proliferation, namely CCDC39, ISY1-RAB43, and UPK3BL1, while SNX22 was the only downregulated DEG. We present the differential expressions of these genes for the different groups using Log2FC values, as shown in Table 1. Then, we used RT-qPCR to detect the relative expressions of the four DEGs and found that CCDC39, ISY1-RAB43, and UPK3BL1 expressions in the ZNF32 H¹⁷⁹A and H¹⁸³A groups were significantly higher than those in the WT and ZNF32 H^{179, 183}A groups, but there were no obvious differences between the WT and ZNF32 H^{179, 183}A groups (Figure 6C–E). The expressions of SNX22 in the ZNF32 H¹⁷⁹A and H¹⁸³A cells were significantly lower than those in the WT and ZNF32 H^{179, 183}A groups, and there were no obvious differences between the WT and ZNF32 H^{179, 183}A cells (Figure 6F). The potential transcriptional binding sequences of ZNF32 were found in the promoter areas of UPK3BL1, ISY1-RAB43, and SNX22 (Figure 6G–I). The dual luciferase reporter gene assay confirmed that ZNF32 H¹⁷⁹A and H¹⁸³A can transcriptionally activate ISY1-RAB43 and UPK3BL1 expressions while inhibiting the transcription of SNX22 (Figure 6J–L). Together, ZNF32 H¹⁷⁹A and H¹⁸³A promote the proliferation of breast cancer cells by differentially upregulating ISY1-RAB43 and UPK3BL1 as well as downregulating SNX22 expressions.

As mentioned above, both single-site and double-site mutations of ZNF32 could form NSs, but the single-site mutations promote proliferation of breast cancer cells by up-down-regulating the expressions of specific genes while the double-site mutation does not affect cell proliferation. Therefore, we consider that the protein structures of ZNF32 H^{179, 183}A as well as H¹⁷⁹A and H¹⁸³A could be inconsistent, thereby showing opposite regulation effects on the genes. As shown in Figure 7A, the mutation of histidine at Aa179 or Aa183 results in the loss of an imidazole ring in the zinc finger, while simultaneous mutations at these two sites can cause the loss of both imidazole rings (Figure 7A). ZNF32 has six typical C₂H₂-ZF domains, and the histone of the fourth zinc finger is situated at the 179 and 183 positions (Figure 7B). Therefore, we speculate that ZNF32 NS formation is largely related to the loss of the histidine imidazole ring in the zinc finger structure. Following this discovery, we mutated the histidine positions of the remaining five zinc finger structures of ZNF32 to alanine to confirm our hypothesis (Figure 7B). Interestingly, only ZNF32 H^{95, 99}A, H^{123, 127}A, and H^{151, 155}A formed NSs in the breast cancer cells (Figure 7C), whereas H^{207, 211}A showed no effect on ZNF32 localization and H^{235, 239}A promoted ZNF32 to shift from nuclear to diffuse localization of the cytoplasm (Figure 7C). We conducted the same experiments on the MCF-7 and MDA-MB-231 breast cancer cell lines, whose results showed that the double-site mutation of H^{207, 211}A had no effect on the localization of ZNF32 and that histidine double-site mutations at all the other zinc fingers formed NSs (Supplementary Figure S3). Thus, the different zinc finger structure mutations of ZNF32 show inconsistent formation of NS-like structures and may also have different effects on the proliferation of breast cancer cells.

The human breast cancer cell lines ZR-75-30, MCF-7, and MDA-MB-231 were obtained from the American Type Culture Collection (Manassas, VA, United States) and maintained in RPMI 1640 medium containing 10% fetal bovine serum (FBS; Gibco, United States) in a humidified atmosphere at 37°C and 5% CO₂ as well as tested regularly for mycoplasma to verify its negative status.

ZNF32 overexpressed (WT) and mutated (H^{179, 183}A, H¹⁷⁹A, and H¹⁸³A) lentivirus samples were purchased from Genepharm (Shanghai, China). All procedures were performed as per the manufacturer instructions. The stable mutated (ZNF32 H^{179, 183}A, H¹⁷⁹A, and H¹⁸³A) and WT cell lines were selected with puromycin.

GFP-ZNF32 plasmids were constructed and stored in our lab. The primers used are listed in Table 2, and all mutations were generated using Mut Express II Fast Mutagenesis Kit V2. (#C214-01 from Vazyme, Nanjing, China). Sequences containing promoter binding regions of ISY1-RAB43, UPK3BL1, and SNX22 were synthesized and constructed into PGL3-Basic vectors by NheI/HindIII (#CD01975871, #CD01975872, #CD01975873 from Tsingke Biotech, Beijing, China). All transfection experiments were performed with TurboFect Transfection Reagent (Thermo, Waltham, MA, United States) according to the manufacturer’s instructions.

For the 5-ethynyl-20-deoxyuridine (EdU) assay, the 5-8F-kiss1R and 5-8F-vehicle cells were seeded in 24-well plates at a density of 1 × 10⁵ cells/well, and the assay was carried out according to the instructions on the kit (Beyotime). The cells were transfected with kiss1 and control plasmids for 48 h as well as incubated for 2 h in a preheated EdU working solution (10 M) at 37°C. The cells were then washed thrice, and a permeable solution was added to the 24-well plates for incubation for 10–15 min. After washing three more times, approximately 200 μL of the click reaction mixture was added and the cells were incubated in the dark at room temperature for 30 min. The samples were then washed thrice, the cell nuclei were stained with DAPI for 5 min, and the cells were finally washed thrice with phosphate-buffered saline (PBS); the prepared cells were then observed and imaged with a microscope.

The MTT assay was performed as per manufacturer protocols using the MTT Cell Viability Assay Kit (Biotechwell WH1197). The cells were seeded in 96-well plates at 10⁴ cells/well and construct transfected 20 h post seeding, as indicated by the manufacturer. MTT was subsequently added to the culture medium and incubated for 2 h at 37°C. Then, the medium was discarded, and 150 μL of dimethylsulfoxide (DMSO) was added to each well. The absorbances of the samples were then measured at 490 nm.

The cells were seeded in 6-well plates and cultured for 2 weeks; during the incubation process, the medium was changed every 24 h. The cell colonies were fixed and stained using a buffer containing 0.05% w/v crystal violet, 1% formaldehyde, and 1% methanol in 1 × PBS at 20°C for 30 min. The samples were then thoroughly washed using ddH₂O and air-dried as per manufacturer protocols.

Six-week-old BALB/c female nude mice (Dashuo, Chengdu, China) were used in this study. About 5 × 10⁶ viable ZR-75-30 cells with the ZNF32 mutation, including WT, H^{179, 183}A, H¹⁷⁹A, and H¹⁸³A, were subcutaneously injected into the mice. One week following subcutaneous transplantation, we observed and recorded the tumor growth and formation rates. The Institutional Animal Care and Use Committee of Southwest Minzu University (Chengdu, China) approved this research project, and all animal experiments were conducted in line with the animal ethical treatment protocols.

When the cells were cultured to 90% in a 10-cm culture dish, the consumed culture medium was discarded, cells were washed twice with PBS before adding TRlzol for lysis, and lysed cells were transferred to an Eppendorf tube. The total RNA was extracted in accordance with the instruction manual of the TRlzol Reagent (Life Technologies, CA, United States). The concentration and purity of the RNA were measured using the NanoDrop 2000 (Thermo Fisher Scientific, Wilmington, United States), and the Agilent Bioanalyzer 2100 system (Agilent Technologies, CA, United States) was used to evaluate RNA integrity. The sequencing library was successfully constructed using the Hieff NGS Ultima Dual-Mode mRNA Library Prep Kit for Illumina (Yeasen Biotechnology, Shanghai, China). The quality of the library was evaluated using the AMPure XP system, and the library was sequenced on the Illumina NovaSeq platform to produce a double-ended read length of 150 bp. The sequencing data generated in this study have been deposited in the NCBI SRA database under the bioproject number PRJNA1135469 (https://www.ncbi.nlm.nih.gov/bioproject/?term=%20PRJNA1135469).

The expression level of each gene was normalized with the reads per kilobase per million (RPKM). To identify the DEGs, the edgeR package was used to filter the genes (Robinson et al., 2010). Following statistical analyses, we screened the DEGs with fold changes ≥3 by setting the false discovery rate (FDR) to <0.01.

GO enrichment analysis of the DEGs was performed using the GOseq R package (Young et al., 2010). All identified DEGs were then annotated using the KEGG database (Kanehisa et al., 2008). Additionally, a hypergeometric test was conducted to find the pathways that were significantly enriched in terms of the DEGs compared to the whole-genome background.

In this study, using the liquid chromatography quadrupole-time-of-flight (LC-QTOF) platform, a total of 24 samples from the mutation groups (ZNF32 H^{179, 183}A, H¹⁷⁹A, and H¹⁸³A) and WT group corresponding to six samples from each group were subjected to qualitative and quantitative metabolome analyses. The liquid chromatography mass spectrometry (LC/MS) system used for metabolomics analysis was composed of the Waters Acquity I-Class PLUS ultrahigh-performance liquid tandem Waters Xevo G2-XS QTOF high-resolution mass spectrometer. The column used was the Waters Acquity UPLC HSS T3 (1.8 μm, 2.1 × 100 mm). The positive and negative ion modes were both composed of 0.1% formic acid aqueous solution as mobile phase A and 0.1% formic acid acetonitrile as mobile phase B, with an injection volume of 1 μL.

The Waters Xevo G2-XS QTOF high-resolution mass spectrometer was used to collect primary and secondary MS data in the MSe mode using the MassLynx V4.2 (Waters) acquisition software. During each data acquisition cycle, dual-channel data acquisition was performed using both low and high collision energies at the same time. The low collision energy used was 2 V, while the high collision energy range was 10–40 V, with a scanning frequency of 0.2 s for a mass spectrum. The parameters of the electrospray ionization (ESI) source are as follows: capillary voltage of 2,000 V (positive ion mode) or −1,500 V (negative ion mode); cone voltage of 30V; ion source temperature of 150°C; desolvent gas temperature of 500°C; backflush gas flow rate of 50 L/h; desolvent gas flow rate of 800 L/h.

The raw data collected using MassLynx V4.2 was processed for peak extraction, peak alignment, and other data processing operations based on the Progenesis QI software online METLIN database and Biomark’s self-built library for identification; at this time, the theoretical fragment identification and mass deviation were both within 100 ppm.

A follow-up analysis was performed after normalizing the original peak area information with the total peak area. Principal component analysis and Spearman correlation analysis were used to assess the repeatability of the samples within the group and quality control samples. The identified compounds were searched for classification and pathway information in the KEGG, HMDB, and LIPID MAPS databases. Based on the grouping information, we calculated and compared the difference multiples. The R language package ropls was used to perform orthogonal partial-least-squares discriminant analysis (OPLS-DA) modeling, and 200-factor permutation tests was performed to verify the model reliability. The variable importance in projection (VIP) value of the model was calculated using multiple cross-validations. The difference multiple, p value, and VIP value of the OPLS-DA model were combined to screen the differential metabolites with thresholds of VIP ≥1 and fold change ≥1. The difference metabolites of the KEGG pathway enrichment significance were calculated using the hypergeometric distribution test.

To validate the DEGs discovered by transcriptome sequencing, RT-qPCR was performed. Primer Premier 5 software was used to design the sequence-specific primers for the selected genes (Table 3). Thereafter, RT-qPCR was performed with the qPCR SYBR Green SuperMix according to manufacturer instructions (Bimake, United States). The 18S rRNA gene was used as the endogenous reference to normalize the relative mRNA expression.

ZR-75-30 cells were seeded at 20,000 cells per well in 500 μL of medium in 24-well plates for 24 h. Using 1 μg firefly luciferase report plasmid (PGL3-ISY1-RAB43, PGL3-UPK3BL1, PGL3-SNX22) and 1 μg ZNF32 wild type or mutants (ZNF32 H¹⁷⁹A, ZNF32 H¹⁸³A, ZNF32 H^{179, 183}A) and 0.1 μg renilla plasmid pRL-TK were co-transfected into breast cancer cells. Forty-eight hours after transfection, cells were lysed and luciferase activity measured according to the manufacturer’s instructions (Dual Luciferase Reporter Assay Kit, #DL101, Vazyme, Nanjing, China). Finally, the luminescence signals of firefly luciferase and renilla luciferase were measured by a Varioskan™ LUX multimode microplate reader (#VLBLATGD2, Thermo Fisher Scientific, United States).

The quantitative PCR data were analyzed using the 2^-ΔΔCt method, and the data were expressed as mean ± standard deviation (Mean ± SD). The differences in the data were analyzed using one-way ANOVA, multiple comparison t-test, and student’s two-tailed t-test in GraphPad Prism 8.0 software. All experiments were repeated at least three times and were statistically significant when the p values were <0.05, *p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001.