N-Terminal deleted isoforms of E3 ligase RNF220 are ubiquitously expressed and required for mouse muscle differentiation

Abstract

Four isoform peptides of the novel E3 ligase ring finger protein 220 (RNF220) have been identified in humans. However, all of the previous studies have predominantly focused on isoform 1 (the full-length form), which consists of 566 amino acids. Here, we show that a shorter isoform, which is 308 amino acids lacking most of the N-terminus (human isoform 4; mouse isoform 3; ΔN-RNF220), is the predominant and ubiquitously expressed variant that warrants functional investigation. Both isoform 1 and ΔN-RNF220 are expressed in the brain; however, ΔN-RNF220 is the major isoform expressed in all other tissues in mice. Consistently, H3K4me3 ChIP-seq data from ENCODE reveal that the transcription start site for ΔN-RNF220 demonstrates broader and stronger activity across human tissues than that of isoform 1. ΔN-RNF220 produces 2 peptides (4a and 4b) through alternative translation initiation, with isoform 4b displaying distinct subcellular localization, subnuclear structures and interaction with a nuclear protein WDR5. Notably, during embryonic stem cell differentiation into neural stem cells, isoform 1 expression increases, whereas ΔN-RNF220 expression decreases. In murine myoblasts, ΔN-RNF220 is the sole expressed isoform and is required for MyoD and myogenin expression, as well as for muscle differentiation. Our findings highlight ΔN-RNF220 as the ubiquitously and highly expressed variant, likely playing a fundamental role across tissues while exhibiting functional differences from isoform 1. These results emphasize the critical importance of ΔN-RNF220 in future studies investigating the biological functions of RNF220.

INTRODUCTION

Ring finger protein 220 (RNF220) is an E3 ubiquitin ligase involved in neural development and synaptic transmission (Ma et al., 2020a, Ma et al., 2019, Ma et al., 2022, Song et al., 2020, Wang et al., 2022). Furthermore, it has roles in carcinogenesis and immunity (Deng et al., 2023, Guo et al., 2021, Pan et al., 2021, Yan et al., 2021).

Human RNF220, located on chromosome 1p34.1 (NCBI Gene ID: 55182), produces 4 isoforms using 5 transcription start sites and alternative splicing (Fig. 1a). Shorter isoforms 2 and 4 lack N-terminal regions present in longer isoforms 1 and 3. The protein isoforms from the same gene may mediate unique (Kubickova et al., 2023) or even opposing biological functions. As seen with LEF1 (de Klerk and 't Hoen, 2015), MYC, STAT3 (Aigner et al., 2019), and RUNX1 (Davuluri et al., 2008), RNF220 isoforms may also exhibit distinct biological functions. However, most studies on RNF220 have focused on the “full-length” 63 kDa isoform 1, in both gain-of-function and loss-of-function experiments. The widely used RNF220 knockout mice, developed by deleting exon 2 (Ma et al., 2019), may retain other short isoforms. Should these short isoforms remain expressed in this model, they may be obscuring as-yet unappreciated physiological roles for RNF220.

Fig. 1.

Comparative analysis of RNF220 isoforms: structure, expression, and subcellular localization. (a) Exon structure of human RNF220 isoform 1 mRNA (NM_018150.4) and other RNF220 isoforms and schematic diagram of RNF220 isoforms in human, mouse, chicken, zebrafish, and fruit fly based on their amino acid sequence similarity. The length of each exon and protein domain is proportional to its actual length. Global alignment was performed using the Needleman-Wunsch algorithm. RING represents RING domain. R363Q and R365Q found in Familial leukodystrophy (Sferra et al., 2021) may affect all isoforms. (b) Comparison of RNF220 isoform expression in various mouse tissues by western blot (9-week-old, male). (c) Identification of each isoform in HEK293 with siRNAs targeting isoform-specific regions illustrated in (a). (d) Amino acid sequence and cDNA alignment of the ΔN-RNF220 N-terminal region across species. The annotated AUG start codon and the 2ndary translation start codon are indicated with their Kozak similarity score (Gleason et al., 2022). Nuclear localization signal (NLS) predicted by cNLS Mapper (Kosugi et al., 2009) is marked. Identification of isoform 4a or 4b was performed using HEK293 cells transfected with the corresponding cDNAs. (e) Immunostaining of RNF220 isoform 1, 4a, and 4b in HEK293 cells after transfection of the cDNAs. Tubulin (orange), RNF220 (green), and DAPI (blue). Scale bar, 10 µm. (f) Immunoprecipitation (IP) analysis of interaction between RNF220 isoform 1, 4a, and 4b and WDR5 in HEK293 cells.

While studying the molecular role of RNF220, we unexpectedly found that the predominant isoform working in most tissues is not isoform 1, but rather a N-terminal deleted 308 amino acids isoform (ΔN-RNF220; isoform 3 in mouse, isoform 4 in human). Notably, ΔN-RNF220 is critical for muscle differentiation in myoblast cells and disappears upon terminal differentiation similar to isoform 1 in neural tissues as previously reported. Isoform 1 and ΔN-RNF220 exhibit distinct tissue distribution, subcellular localization, subnuclear organization, and protein-protein interactions. Given that the prior studies on RNF220 functions have been limited to analysis of the 566 amino acids isoform 1, our findings underscore the need to investigate the short isoform, ΔN-RNF220, to fully understand the biological functions of RNF220 throughout the body.

MATERIALS AND METHODS

Mouse Tissues

The protocols of animal experiments were reviewed and approved by the Institutional Animal Care and Use Committee of Sookmyung Women's University. Tissues were dissected from a 9-week-old male mouse and used for protein extraction followed by Western blotting.

Cells

HEK293 cells were purchased from Korean Cell Line Bank and human embryonic stem cells (hESCs) (H9) were purchased from Wicell Research Institute. C2C12s were obtained from the American Type Culture Collection.

Plasmids and siRNAs

Each isoforms’ EGFP-tagged RNF220 expression vector and non–tagged expression vector were cloned using hRNF220-WT (iso 1, kindly gifted by Cheol-Hee Kim and Seunghee Lee) (Kim et al., 2018) and Q5 Site-Directed Mutagenesis Kit (NEB) according to the manufacturer's protocol. FLAG-WDR5 plasmid was obtained from Addgene (plasmid #59974). siRNAs were purchased from IDT. The isoform 4-targeting siRNA sequence was custom-designed in-house to specifically target the exon 5a region of RNF220.

Further information is available in the Supplementary Information.

RESULTS

Four human RNF220 isoform polypeptides were annotated based on the RefSeq database (https://www.ncbi.nlm.nih.gov/refseq/). Detailed information on the transcriptional variants and their corresponding 4 peptide isoforms is summarized in S Table 1. Fifteen exons of the isoform 1 RNF220 transcript (NM_018150.4) were annotated with 2 alternative transcription start sites leading to isoforms initiating in exons 2a or 5a. The red indicated in Figure 1a is not in the isoform 1 peptide and the yellow is only included in isoform 4. Mouse has 4 annotated RNF220 isoforms, 1 short distinct and 3 corresponding to human isoforms. Interestingly, zebrafish and fruit fly express only a single, short isoform which corresponds to the human isoform 4 (hereafter this isoform will be referred as N-terminal deleted RNF220, ΔN-RNF220) (Fig. 1a, S Table 2) suggesting that these long-overlooked short isoform of RNF220 may mediate the most conserved functions of this gene in evolution. We examined the expression pattern of RNF220 in mouse tissues using 2 antibodies, Ab1 and Ab2, that detect the N- and C-terminal regions, respectively (Fig. 1b, S Fig. 1). Using Ab2 and less specific Ab1, we found that putative isoform 1 is expressed only in the brain, whereas putative ΔN-RNF220 is expressed in all the examined tissues. A partial correlation was observed between the protein and mRNA levels of each isoform, which indicates that both transcriptional and post-transcriptional regulation contribute to their tissue-specific expression (S Fig. 2, Fig. 1b). A human embryonic kidney cell line, HEK293, also expressed both isoform 1-like and ΔN-RNF220-like bands. These are depleted by siRNAs selectively targeting exons of each isoform (Fig. 1c). Based on molecular weight and siRNA-mediated suppression, we conclude that the isoforms primarily expressed in these cells and tissues are isoform 1 (63 kDa) and ΔN-RNF220 (35 kDa).

Using public ChIP-seq data (Carninci et al., 2006, Forrest et al., 2014), we sought to identify which RNF220 TSSs (transcription start sites) for different isoforms might be utilized in various human tissues based on H3K4me3, which is a hallmark of active TSS chromatin (Guenther et al., 2007). The peaks of H3K4me3 enrichment were observed that mapped to multiple RNF220 TSSs in all tissues examined (S Fig. 3). The ΔN-RNF220 TSS showed the most highly enriched peak of H3K4me3 in all tissues examined, except the hippocampus, where the TSS for iso 1/3 was stronger. These findings align with the mouse immunoblotting results (Fig. 1b) and suggest that ΔN-RNF220 is more active than any other isoform in most tissues, both in humans and mice.

We noticed that the ΔN-RNF220 was detected as 2 bands. We found that the mRNA for ΔN-RNF220 may have a 2ndary start codon at M38, supported by another potential Kozak sequence which is conserved in evolution. This suggests the smaller band (4b) might be translated from this 2ndary start codon (Fig. 1d). Interestingly, using the 2ndary start codon may partially disrupt a putative nuclear localization signal. Overexpressed isoforms 1 and 4a are stained in nuclear and localized to nuclear speckle-like structures, forming a distinctive ring pattern, consistent with previous reports (Sferra et al., 2021). In contrast, isoform 4b displayed a completely different nuclear staining pattern along with partial cytoplasmic distribution (Fig. 1e). Similar results were observed in live imaging of cells transfected with these EGFP-fused RNF220 isoforms (S Fig. 4). We investigated interactions with WDR5, a nuclear epigenetic scaffold protein known to associate with RNF220 isoform 1, to determine if isoforms 1, 4a, and 4b have distinct biological roles. Interestingly, WDR5 interacts strongly with RNF220 isoform 1 and isoform 4a, but to a lesser extent with isoform 4b (Fig. 1f). This difference may correlate with the distinct subcellular and subnuclear localization of these isoforms. These findings suggest that isoform 4b exhibits distinguished biological roles from those of isoform 1 and 4a based on the subnuclear structure and subcellular localization.

Given the reported roles of RNF220 isoform 1 in murine neural stem cells (NSCs) (Ma et al., 2019, Zhang et al., 2020) in mice, we analyzed the expression of RNF220 isoforms in NSCs derived from hESCs (Fig. 2a). While both isoform 1 and ΔN-RNF220 were expressed in hESCs, differentiation into NSCs resulted in increased isoform 1 and decreased ΔN-RNF220. A unique splicing pattern at exon 7 generated isoform 1, whereas other isoforms arose from alternative splicing (illustrated in Figs. 1a and 2b). The changes of isoforms during hESCs differentiation into NSCs may be driven by isoform-specific transcription.

Fig. 2.

Expression of RNF220 isoforms during NSC differentiation and muscle differentiation. (a) Expression of the isoforms during NSC differentiation from hESC was tested with all isoforms targeting Ab2. Only isoforms 1 and 4 were detected. NIM, neural induction medium; NSMM, neural stem cell maintenance medium. (b) RT-PCR of exon 7/8 alternative spliced mRNA for isoforms (pF1R1; 146 bp-iso 1 and 224 bp-iso 4, pF2R1; 165 bp-iso 4). Isoform mRNA expression was assessed by qRT-PCR using primers specific to exon 2 (isoform 1) and alternatively spliced exon 7 (pF2/pR1; isoform 4). Mean ± SD. ***P < 0.001. (c) Expression of RNF220 during C2C12 myoblast differentiation (left). Differentiation of C2C12 myoblasts, after transduced with lentivirus for control shRNA (shNC) or shRNF220 targeting all isoforms (#3), followed by Western blotting and qRT-PCR. Mean ± SD. ***P < 0.001. (d) C2C12 transduced with lentivirus for control (shNC), targeting all isoforms (#3) or targeting isoform 1 only (#5), was differentiated. After 3 days, cells were fixed and stained with anti-MHC and DAPI. Scale bar, 50 µm. (e) Quantification results for experiments as shown in (d). Statistical analysis was performed for the comparisons of multinucleated cells (4 or more nuclei). Differentiation index and fusion index were calculated. ***P < 0.001. Data are means ± SEM.

Notably, the muscle exhibits minimal expression of ΔN-RNF220, which is unusual compared to other tissues (Fig. 1b). Given that RNF220 is essential for proliferation and stemness across many tissues’ cancers (Deng et al., 2023, Yan et al., 2021) and its loss leads to NSC exit from the cell cycle and promoted differentiation (Zhang et al., 2020), we hypothesized that the low expression of ΔN-RNF220 in muscle is linked to its very low proportion of proliferating stem cells in muscle (Marchok and Herrmann, 1967, Snow, 1977). To determine if RNF220 performs analogous roles in muscle differentiation as observed in neural differentiation, we examined its expression during the differentiation of C2C12 murine myoblasts. Unexpectedly, cultured C2C12 cells strongly expressed ΔN-RNF220, but its expression decreased during differentiation into myotubes as expression of the sarcomere protein myosin heavy chain (MHC) increased (Fig. 2c). shRNA knockdown of all forms of RNF220 in C2C12 cells reduced expression of MyoD, the basic helix-loop-helix transcription factor responsible for muscle cell lineage determination, both before and during differentiation. The isoform-specific qRT-PCR showed that the mRNA for ΔN-RNF220 is expressed in C2C12. Loss of ΔN-RNF220 decreased the expression of both myogenic differentiation factor myogenin and MHC during differentiation. Reduced cell fusion to form myotubes and impaired differentiation were observed when all RNF220 isoforms were targeted, while an shRNA specifically targeting isoform 1 failed to reveal the same effect (Fig. 2d, e and S Fig. 5), and it was rescued by reintroduction of isoform 4a or 4b (S Fig. 6).

Overall, we conclude that it is ΔN-RNF220, rather than the full-length isoform 1, that is required for MyoD expression and subsequent differentiation into myotubes in C2C12 myoblasts. These findings further suggest that isoforms with the presence and absence of the N-terminal half may not only exhibit tissue-specific expression but also perform distinct biological functions.

DISCUSSION

Recent studies have expanded the roles of RNF220 beyond neuronal tissues to other tissues and processes, such as cancer and immunity. In all of these studies, isoform 1 has been regarded as the primary functional molecule expressed from the RNF220 locus, and the function was investigated. Similarly, RNF220 knockout mice lacking exon 2, which eliminates isoform 1 but may not impact ΔN-RNF220, have been used in most studies to unveil the many novel biological functions of the RNF220 locus (except 1 study targeted exon 7; Kim et al., 2018). Our findings reveal distinct expression patterns and functions for a short and N-terminally truncated RNF220, ΔN-RNF220. ΔN-RNF220 is strongly expressed across most tissues, whereas isoform 1 is confined primarily to neuronal tissues in humans and mice. ΔN-RNF220 appears crucial for myoblast differentiation. The dual translation initiation also modulates the subcellular localization of isoform 4a/b differently from isoform 1 and binding affinity to WDR5, underscoring their unique functional attributes. Familial leukodystrophy is caused by R363Q and R365Q mutations (Sferra et al., 2021) affecting all isoforms. Neurological symptoms likely stem from abnormal isoform 1 due to its tissue-specific expression, while ΔN-RNF220, with broader expression, may drive systemic pathologies such as cardiomyopathy. In skin fibroblasts, where isoform 1 mRNA was undetectable (Sferra et al., 2021), mutant isoform 4a or 4b likely caused nuclear membrane abnormalities. The unique cytosolic and subnuclear structure of isoform 4b may further contribute to these phenotypes.

ΔN-RNF220 is strongly expressed at the myoblast stage but disappears as myocyte differentiation progresses. The loss of RNF220 during differentiation suggests that nondividing, terminally differentiated muscle cells do not express RNF220. For cells to maintain their differentiation potential as myoblasts do, functional RNF220 appears to be necessary. The absence of long isoforms in zebrafish and fruit flies supports the idea that ΔN-RNF220 represents the original form of RNF220. Notably, this was also essential for nervous system patterning in zebrafish (Ma et al., 2020b) and Drosophila (Sferra et al., 2021), resembling the function of isoform 1 in humans.

Our results indicate that previous studies on RNF220, which primarily focused on isoform 1 expressed mostly in neuronal tissues, may have overlooked broader functional roles of ΔN-RNF220 across various tissues. Thus, future investigations should carefully consider that ΔN-RNF220 may play an important role in multiple tissues through its own function.

Author Contributions

Young-Hyun Go: Resources, Investigation. Hyuk-Jin Cha: Resources. Donald J. Wolfgeher: Data curation, Investigation. Sojung Ha: Formal analysis, Investigation. Jee Won Kim: Resources, Investigation. Woo-Young Kim: Writing – review & editing, Writing – original draft, Supervision, Funding acquisition, Conceptualization. Gyu-Un Bae: Resources, Funding acquisition. Stephen J. Kron: Writing – review & editing, Resources, Funding acquisition. SeokGyeong Choi: Writing – review & editing, Writing – original draft, Visualization, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of Competing Interests

S.J.K. is a founder and owner of OncoSenescence, Riptide Therapeutics, and Oligo Foundry. The other authors declare no competing interests.

Appendix A. Supplemental material

Supplementary material

.