SUMOylation of zebrafish transcription factor Zbtb21 affects its transcription activity

Plasmid construction

The zebrafish zbtb21 gene and its serial SUMOylation mutants were cloned into PCS2⁺ vector (Addgene, #34931, Watertown, MA). For the luciferase reporters, the −1.5 kb promoter of CDC6 gene and the −1.1 kb promoter of zbtb14 gene were cloned into the PGL3 basic vector (Promega, Promega, #E1751, Madison, Wisconsin, US). The primers used were listed in Table 1. All primers were designed using the Primer Premier software 5 (PREMIER Biosoft, Palo Alto, CA, US) and synthesized by Sangon (Shanghai, China).

Cell culture, transient transfection, luciferase reporter assay, and immunofluorescence

HEK293T was obtained from ATCC, and the cell line was tested negative for mycoplasma. HEK293T cells were maintained in DMEM (#11960044, Gibco, Waltham, MA, USA) with 10% fetal bovine serum (#16140071, Gibco, Waltham, MA, USA). Plasmid transfection was conducted utilizing Effectene Transfection Reagent (#301425, Qiagen, Hilden, Germany) following the manufacturer’s guidelines.

For the luciferase reporter assay, HEK293T cells were collected 48 h post-transfection and assessed using the Dual Luciferase Reporter Assay Kit (#E1910, Promega, Madison, Wisconsin, USA), in accordance with the manufacturer’s instructions.

For microscopic analyses, HEK293T cells were seeded on micro cover glasses in wells of six-well plates 24 h before transfection. 48 h after transfection, cells were immunostained with rabbit anti-HA monoclonal (#3724S, Cell Signaling Technology, Danvers, Massachusetts, USA) at a 1:1,000 dilution, and visualized with secondary antibody, Alexa Fluor 594 goat anti-rabbit IgG (#14708S, Cell Signaling Technology, Danvers, Massachusetts, USA), at a 1:1,000 dilution. Cell nuclei were stained by DAPI (1.0 mg/ml, #564907, BD Biosciences, Dubai, UAE) for 1 min. Fluorescent images were obtained by ECLIPSE E800 microscope (Nikon, Tokyo, Japan) with a SPOT RT Slider digital camera (SPOT Imaging Solutions, Sterling Heights, MI, USA).

Western blot and co-immunoprecipitation assays

HEK293T cells underwent transfection with specified plasmids (#301427, Qiagen, Hilden, Germany). Following 48 h post-transfection, the cells underwent three consecutive 1-min washes with phosphate-buffered saline (PBS) solution. Lysates were generated by treating the cells with RIPA lysis buffer (#P0013J, Beyotime, Jiangsu, China) containing proteinase inhibitor (#5892970001, Roche, Basel, Switzerland), followed by gentle agitation on ice for 30 min. Subsequently, the cells were harvested and centrifuged at 15,000 × g for 30 min at 4 °C. Rabbit anti-HA antibody (#3724S, Cell Signaling Technology, Danvers, Massachusetts, USA) was combined with the supernatant containing protein-G-agarose beads (30 µl) and incubated overnight at 4 °C. Subsequently, the beads were subjected to three washes with RIPA lysis buffer. Finally, proteins attached to the beads were released by the addition of 30 µl of 2 × SDS sample buffer and subsequently analyzed via immunoblotting using either anti-SUMO1 (#4930T, Cell Signaling Technology, Danvers, Massachusetts, USA) or anti-FLAG antibody (#F3165, Sigma, Burlington, MA, USA).

RNA-seq

Forty-eight h after transfection, mRNA was isolated from HEK293T cells by using Trizol reagent (#15596026, Ambion, Austin, Texas, US). Next, mRNA was isolated with Oligo(dT) magnetic beads and fragmented by fragmentation buffer, preparing it for first and second strand cDNA synthesis. After purification of the double-stranded cDNA, the ends were repaired, sequencing adapters were attached, and PCR amplification was performed. The prepared library underwent quality evaluation. Upon confirmation of library quality, sequencing was conducted using an MGI T7 sequencer (China, MGI, Inc). These data are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE255803.

ChIP-seq

The EpiTM Chromatin Immunoprecipitation Kit (#R1813, Epibiotek, Seoul, Korea) was utilized for ChIP experiments. Initially, 7 × 10⁶ fresh cells were fixed with 1% formaldehyde for 10 min, then quenched with 0.125 M glycine for 5 min. The cells were lysed using 1 mL of lysis buffer, and the nuclei were separated by centrifuging the lysate at 2,400 × g for 10 min at 4 °C. The nuclei were then resuspended in digestion buffer and incubated at 37 °C for 10–15 min for enzymatic digestion, fragmenting the chromatin to an average size of 200–500 bp. The chromatin fragments were collected by centrifuging at 18,000 × g for 10 min at 4 °C. The supernatant was then mixed with a ChIP reaction mixture including protein A/G magnetic beads, ChIP IP buffer, antibodies, and protease inhibitors, and incubated with rotation overnight at 4 °C. The following day, the protein A/G magnetic beads underwent washing and removal, while the chromatin was eluted through incubation in reverse cross-linking buffer at 65 °C for 3 h. Subsequently, the ChIP DNA underwent treatment with RNase A and Proteinase K at 37 °C for 30 min, followed by purification using phenol-chloroform extraction. Finally, the ChIP DNA was processed using the QIAseq Ultralow Input Library Kit (#180495, Qiagen, Hilden, Germany) according to the manufacturer’s instructions to generate the library. These data are available at https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE254892.

Statistical analysis

Data were analyzed by GraphPad Prism 9 using one-way ANOVA for comparisons among multiple groups. Differences were considered significant at P < 0.05. Data are expressed as mean ± standard error of the mean (SEM).

Zbtb21 is identified as a SUMOylated substrate

Bioinformatic prediction indicated that K419 and K845 are two potential SUMOylated lysine residues of the Zbtb21 protein, located within the linker and zinc finer regions, respectively (SUMOsp2.0 prediction software). To ascertain whether Zbtb21 is indeed a bona fide SUMOylated substrate, a series of mutants including Zbtb21^K419R, Zbtb21^K845R, and Zbtb21^K419/845R (lysine was mutated to arginine to disrupt the attachment with SUMO conjugates) was generated and transfected into HEK293T cells. Western blot analyses revealed four distinct bands corresponding to different forms of Zbtb21 protein. The lowest band represented unmodified Zbtb21 protein while the other three bands corresponded to SUMOylated forms of Zbtb21 which exhibited increased intensity in the presence of both UBC9 (SUMO conjugating enzyme) and SUMO1 (Fig. 1A, lane 2 and 3). Notably, specific SUMO adducts were absent for Zbtb21^K419R and Zbtb21^K845R mutant proteins, respectively (Fig. 1A, lane 4–7), whereas no adducts could be observed for Zbtb21^K419/845R double mutant (Fig. 1A, lane 8 and 9). These results from western blot analyses were further validated through immuno-coprecipitation (Co-IP) experiments using an anti-SUMO1 antibody (Fig. 1B). While wild type Zbtb21, Zbtb21^K419R and Zbtb21^K845R single mutants showed successful pull-down of their respective SUMOylated forms upon Co-IP analysis, no such interaction was observed for the Zbtb21^K419/845R double mutant.

Taken together, these observations suggest that Zbtb21 is a bona fide SUMOylated substrate, with K419 and K845 identified as two lysine residues conjugated by SUMO. Furthermore, the comparable protein levels of wild type Zbtb21 and Zbtb21^K419/845R double mutant imply that loss of SUMOylation does not significantly impact the stability of Zbtb21 protein.

SUMOylation of Zbtb21 affects its transcriptional activity

In most cases, SUMOylation of transcriptional factors is associated with transcription repressing through facilitating interaction with corepressors. Wang et al. (2005) demonstrated that ZBTB21 exerted transcriptional repression on the promoter of CDC6. Therefore, we conducted luciferase assays using wild-type Zbtb21 and SUMO-defective mutants to investigate their effects on the CDC6 reporter. Additionally, zbtb14, whose upstream regulatory region could be bound by Zbtb21 in our previous study (Deng et al., 2022), was also used for luciferase assay. Furthermore, considering that Zbtb14 can inhibit pu.1 expression (Deng et al., 2022), we aimed to explore whether Zbtb21 also possesses repressive function on the pu.1 promoter.

The results showed that while Zbtb21 exhibited a significant repression effect on the promoters of CDC6 and pu.1, it could activate the zbtb14 reporter (Figs. 2A–2C), suggesting that Zbtb21 is a context-dependent dual-function transcription factor. Importantly, both the repression and activation activities of Zbtb21 were impaired upon loss of SUMOylation (Figs. 2A–2C).

To gain insights into the global changes in the transcriptome, RNA-seq analyses were carried out in HEK293T cells transfected with equal amounts of wild type Zbtb21 and Zbtb21^K419/845R double mutant, respectively (Fig. 2D). Overall, we observed that HEK293T cells expressing wild type Zbtb21 displayed a total of 234 upregulated genes and 369 downregulated ones (Fig. 2E). However, cells expressing Zbtb21^K419/845R mutant showed only 146 upregulated genes and 105 downregulated ones (Fig. 2E). These findings suggest that compared to the SUMOylation-defective mutant, wild type Zbtb21 led to more downregulation of genes, implying that SUMOylation of Zbtb21 favors its transcriptional repression capacity.

Meanwhile, ChIP-seq analyses were conducted to define the genome-wide binding profiles of wild type Zbtb21 and Zbtb21^K419/845R mutant. The results showed that while regulatory regions of 2,041 genes were bound by wild type Zbtb21, only 845 genes were found for Zbtb21^K419/845R mutant (Fig. 2F). These observations indicate that SUMOylation of Zbtb21 can enhance its DNA binding capacity. Additionally, the regulatory regions of certain sets of genes were bound either by wild type Zbtb21 or Zbtb21^K419/845R mutant (1,829 genes vs. 633 genes) (Fig. 2F), suggesting that the SUMOylation status can alter the DNA binding property of Zbtb21.

Based on the results from ChIP-seq and RNA-seq analyses, we proceeded to analyze the expression levels of genes whose regulatory regions were bound by wild type Zbtb21 and Zbtb21^K419/845R, respectively. The results demonstrated that wild type Zbtb21 led to downregulation of 36 genes and upregulation of six genes (Fig. 2G). In contrast, Zbtb21^K419/845R mutant showed only seven downregulated genes and four upregulated genes. These findings further support the notion that SUMOylation of Zbtb21 tends to turn it to be a transcriptional repressor (Fig. 2G).

In most cases described to date, SUMOylation of transcriptional factors often leads to transcription inhibition through promoting interaction with corepressors. Notably, histone deacetylase corepressors have been identified as effectors of SUMOylation, playing important roles in transcription regulation (Yang & Sharrocks, 2004). Protein interaction database (BioGRID) showed that ZBTB21 can interplay with several corepressors including HDAC1/4/5/7. Therefore, we performed Co-IP analyses in HEK293T cells. Compared to SUMO-defective mutant, more endogenous HDAC4 molecules could be co-immunoprecipitated with wild type Zbtb21 (Fig. 2H), which would be the mechanism by which Zbtb21 inhibits transcription. Additionally, Co-IP analyses also demonstrated selective binding between wild type Zbtb21 and BRD4 coactivator rather than the SUMO-defective mutant (Fig. 2H).

Collectively, these results indicate that SUMOylation is essential for regulating the function of Zbtb21.

SUMOylation of Zbtb21 does not affect its subcellular localization

The consequence of protein SUMOylation includes alterations in protein activity, protein stability, and cellular localization (Chang & Yeh, 2020). The nuclear localization signal (NLS) of Zbtb21 is close to K419 SUMOylation site. To demonstrate whether SUMOylation plays a role in the subcellular localization of Zbtb21, immunofluorescence staining was performed to compare the localization of wild type Zbtb21 and SUMO-defective mutants. The results showed that HA-tagged wild type Zbtb21 was localized in the nucleus of HEK293T cells, and similar localization was found for Zbtb21^K419R, Zbtb21^K845R, and Zbtb21^K419/845R mutants (Fig. 3). These observations suggest SUMOylation of Zbtb21 is irrelevant to its nuclear distribution.

SUMOylation of Zbtb21 does not affect its interaction with its partners

In addition to DNA binding, C2H2-type zinc fingers can mediate protein-protein interactions. The interaction between ZBTB14 and ZBTB21 is mediated by both BTB and zinc finger domains (Wang et al., 2005). Since K845 of ZBTB21 is located within the zinc finger region, we set out to investigate whether K845 SUMOylation could affect its binding with ZBTB14. The results from Co-IP assays indicated that the ZBTB21 SUMO mutants could still interplay with ZBTB14 (Fig. 4A). Besides, SUMOylation did not affect the homodimer formation of ZBTB21, either (Fig. 4B).