key: cord-0290754-e40rr195
authors: Nitoiu, Alexandra; Nabeel-Shah, Syed; Farhangmehr, Shaghayegh; Pu, Shuye; Braunschweig, Ulrich; Blencowe, Benjamin J.; Greenblatt, Jack F.
title: KRAB Zinc Finger protein Znf684 interacts with Nxf1 to regulate mRNA export
date: 2021-10-01
journal: bioRxiv
DOI: 10.1101/2021.09.29.462476
sha: a5adf622f48b59bc769e4636de959444bbee2f09
doc_id: 290754
cord_uid: e40rr195

Cys2His2 (C2H2) type zinc finger (ZnF) proteins constitute a large class of proteins that are generally considered to be DNA-binding transcription factors. Using affinity purification followed by mass spectrometry, as well as reciprocal co-immunoprecipitation experiments, we determined that the C2H2-ZnF protein Znf684 interacts physically with several proteins involved in mRNA export, including Nxf1 and Alyref. We utilized individual nucleotide resolution cross-linking immunoprecipitation followed by high throughput sequencing (iCLIP-seq) experiments to show that Znf684 binds directly to specific mRNAs in vivo and has an RNA-binding profile similar to those of Nxf1 and Alyref, suggesting a role in mRNA export regulation. Immunofluorescence microscopy (IF) experiments revealed that Znf684 localizes to both the nucleus and cytoplasm. Using cellular fractionation experiments, we demonstrate that overexpression of Znf684 negatively impacts the export of SMAD3 and other target mRNAs. Taken together, our results suggest that Znf684 regulates the export of a subset of transcripts through physical interactions with Nxf1 and specific target mRNAs.

the KRAB domain, resulting in a decrease in the protein length by approximately 30 residues ( Figure 1A ). However, the number and placement of the zinc finger domains remains evolutionarily conserved. This observation suggests that Znf684 might be functionally divergent in species lacking the KRAB domain, perhaps not functioning to repress gene expression. Of note, certain species, such as mice and ray-finned fish, appear to have lost the ZNF684 gene entirely.

We next examined the Znf684 expression profile across various tissues and cell lines using publicly available RNA sequencing (RNA-seq) data. In general, Znf684 appears to be weakly expressed across the tissues and cell lines we examined (Supplementary figure 1A) . However, Znf684 is significantly upregulated during embryonic development at stages corresponding to 8C and morula (Supplementary figure 1B) . These observations suggest that Znf684 expression is tightly regulated across tissues and during embryonic development.

To begin characterizing Znf684, we analyzed our previously published affinity purification followed by mass spectrometry (AP-MS) data generated in human embryonic kidney (HEK293) cells inducibly expressing GFP-tagged variant of Znf684 (21). We scored Znf684 AP-MS data against numerous control purifications using Significance Analysis of INTeractome express (SAINTexpress) (22). Application of SAINTexpress indicated that Znf684 binds to mRNA nuclear export factors such as Nxf1 and Pabpn1, as well as to mitochondrial ribosomes and other proteins (False Discovery Rate [FDR] <0.01; Figure 1B , Supplementary Table S1). This suggested that Znf684 might have roles in mRNA export regulation and mitochondrial translation.

To validate the interaction with mRNA export factors, we performed coimmunoprecipitation (co-IPs) experiments using whole cell extracts prepared from cells expressing either free GFP or GFP-Znf684 and probed the resulting Western blots with antibody against Nxf1. Cell lysates were treated with a promiscuous nuclease (Benzonase) to reduce any indirect RNA (or DNA)-mediated interactions, ensuring that only physical protein-protein interactions were captured in co-IPs. Our results show that Nxf1 is present only in the GFP -Znf684 immunoprecipitated samples and not in the free GFP ones ( Figure 1C ). To further validate these findings, we performed reciprocal Co-IP experiments and observed that endogenous Nxf1 was able to pull down GFP-Znf684 ( Figure 1D ). We next asked whether or not Nxf1 could also interact with Alyref and PABPC1, which have also been shown to function in the mRNA export pathway (2, 23). As shown in figure 1E, Znf684 successfully pulled down both Alyref and PABPC1. These results confirmed the mass spectrometry data and showed that Znf684 interacts physically with mRNA export-related proteins. Given that Znf684 interacts with both Nxf1 and components of the TREX complex, it became likely that Znf684 plays a role in mRNA export regulation.

To begin understanding the mechanism of Znf684's function in mRNA export, we studied Znf684's subcellular localization. Using publicly available immunofluorescence (IF) data from the Human Protein Atlas (www.proteinatlas.org), we observed that endogenous Znf684 localizes to both the nucleus and cytoplasm in several human cancer cell lines, including Cervical cancer (SiHa), Bone Osteosarcoma Epithelial (U2OS), and Rhabdomyosarcoma (RH30) cells (Supplementary Figure S2) . We also performed IFs in HEK293 cells overexpressing either GFP-Znf684 or free GFP. Our results showed that Znf684 localizes to both the nucleus and cytoplasm, consistent with its localization in human cancer cells (Figure 2A ). Since the cytoplasmic signal appeared as distinct puncta, bearing resemblance to mitochondria, and Znf684 interacts with many mitochondrial ribosomal proteins, we examined whether Znf684 localizes to mitochondria by costaining cells with the mitochondrial marker Mitotracker. We found that Znf684 does localize to mitochondria, in addition to its presence in the nucleus ( Figure 2B ).

To investigate whether Znf684 binds directly to RNA, we performed crosslinking and immunoprecipitation (CLIP), followed by gel electrophoresis and autoradiography. Our data shows that GFP-Znf684 cross-linked robustly to RNA in UV-irradiated cells, while GFP alone did not crosslink and thus did not produce any observable radioactive signal ( Figure 3A ). Since the strength of the radioactive signal decreased in response to over-digestion with RNase I, we concluded that Znf684 binds directly to RNA ( Figure 3A ).

To identify which RNAs are bound by Znf684, we performed individual-nucleotide resolution UV crosslinking and immunoprecipitation (iCLIP) followed by high throughput sequencing (iCLIP-seq) experiments. iCLIP-seq peak distribution analysis indicated that the majority of peaks, ~57% (22,291/39,111 peaks), are located in the 3'UTRs of protein-coding genes (Supplementary Table S2 ; Figure 3B ). ~35% of the total peaks were found in the coding regions of the target mRNAs, whereas only ~8% were found in the 5'UTRs ( Figure 3B ). Consistent with these observations, our metagene plots show that Znf684 binds to target mRNAs with a marked preference for 3' UTRs ( Figure 3C ). We then performed motif analysis and found that the two most enriched motifs are guanosine (G)-rich sequences, both of which are very similar at the last three positions ( Figure 3D ). Collectively, these results suggest that Znf684 preferentially binds to the 3'UTRs of its target mRNAs, with a preference for G-rich sequences.

We next compared the iCLIP peak distribution with Znf684 chromatin occupancy. By utilizing our previously published chromatin immunoprecipitation combined with high throughput sequencing (ChIP-seq) data (13, 21), we identified 266 significant peaks for Znf684. These peaks were found across repeat elements (Supplementary Table S3 ) and did not overlap with iCLIP-seq peaks. Although Znf684 exhibits distinct DNA-and RNA-binding profiles, G-rich sequences appeared to be the preferred sites for binding in each case (Supplementary Figure S3 ).

To examine if Znf684 targets mRNAs with a specific function, we performed enrichment analysis using the Kyoto Encyclopedia of Genes and Genomes (KEGG). We found that Znf684 target mRNAs were significantly enriched in genes related to cellular macromolecule biosynthetic processes and organelle organization (FDR  0.05) ( Figure 3E ). Furthermore, regulation of gene expression and RNA processing were also over-represented in Znf684 targets ( Figure 3E ). These results suggest that Znf684 binds to functionally important transcripts in vivo.

To further examine the functional link of Znf684 with mRNA export factors, we analyzed publicly available iCLIP-seq data for Nxf1 and Alyref (2). Our analysis indicated that both factors had RNA-binding profiles similar to that of Znf684, as they displayed a preference for binding to 3' UTRs (Supplementary Figure S4A) . Although Nxf1 and Alyref have been reported to bind RNA without any sequence preference (24, 25), our motif analysis identified moderately enriched Grich sequence motifs for these proteins, similar to that of Znf684 (Supplementary Figure S4B ).

These observations are consistent with the idea that Znf684 functions in conjunction with mRNA export factors, including Nxf1 and Alyref, to regulate the export of target mRNAs.

To further investigate Znf684's localization in the cell and its role in mRNA export, we separated nuclear and cytosolic fractions using cells overexpressing either GFP-Znf684 or free GFP. We overexpressed Znf684 in HEK293 cells and confirmed approximately twofold overexpression by reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) ( Figure 4A ). Importantly, the expression levels of Nxf1 and Alyref remained unchanged upon Znf684 overexpression ( Figure 4A ). To rule out the possibility of cross-contamination between nuclear and cytosolic compartments, Western blotting experiments were performed, and the isolated nuclear and cytoplasmic fractions were probed with the nuclear U170k and cytoplasmic GAPDH markers. As shown in Figure 4B , we did not detect any cross-contamination between the cellular fractions (also see Supplementary Figure S5A ). Furthermore, we also probed the Western blot with an anti-GFP antibody and found that Znf684 was present in both the cytoplasmic and nuclear fractions, consistent with our findings based on IF experiments ( Figure 4B ).

Next, we compared the relative mRNA levels of four Znf684-targets, as identified by the iCLIP-seq data, in the nuclear and cytoplasmic fractions from GFP-Znf684 and GFP overexpressing cells by RT-qCPR experiments ( Figure 4C ; Supplementary Figure S5B ). Our results show that the relative abundance of SMAD3, VEGFA, DDX3X and AKT1 transcripts in the nucleus was significantly (p<0.05) increased in the Znf684-overexpressing cells compared to the controls ( Figure 4D ). This indicates that fewer SMAD3, VEGFA, DDX3X and AKT1 mRNA transcripts reach the cytoplasm when Znf684 is overexpressed. These results indicate that Znf684 may play an inhibitory role in mRNA nuclear export ( Figure 4E ). The observed increase in the nuclear abundance of certain transcripts could also be related to transcription upregulation and/or mRNA degradation in the cytoplasm. Since Znf684 belongs to the KRAB family of C2H2-ZnF transcription factors, we investigated whether the expression levels of target genes were affected in cells overexpressing GFP-Znf684. Our results indicate that mRNA transcript levels decreased significantly (* p<0.05, ** p<0.01) upon Znf684 overexpression for AKT1, FHIT and VEGFA and SMAD3 (Supplementary Figure S5C) . Thus, the observed increase in the nuclear transcript levels in Znf684-overexpressing cells is unlikely due to transcription upregulation. Further studies-including cellular fractionation followed by RNA sequencing and mRNA stability analyses-are currently under way to investigate the genome-wide effects of Znf684 overexpression on mRNA export.

Next, we examined potential links between Znf684 and cancer. Using archived patient RNA-seq data from 'The Cancer Genome Atlas' (TCGA), we examined the relationship of Znf684 expression levels to cancer diagnosis and prognosis (26). Our analysis revealed that Znf684 expression is significantly altered in 9 of the 31 cancer types we analyzed. Only three cancer types Further investigation using Kaplan-Meier curves (26) revealed that patients that have low expression of Znf684 exhibited significantly worse overall survival among those who were diagnosed with KIRC (P < 0.001) ( Figure 5B ). These results suggest that lower than normal expression level of Znf684 may be a predictor for worse overall survival for patients with KIRC.

We suggest that continuing to study Znf684 might yield valuable insights for the development of cancer diagnostics or anti-cancer therapeutics.

mRNA export regulation plays an important role in the modulation of eukaryotic gene expression (7). In this study, we found evidence that Znf684 functions in mRNA nuclear export through a physical interaction with Nxf1. This idea is supported by its interactions with other nuclear export factors, similarities in RNA-binding profiles with those factors, and effect on mRNA export into the cytoplasm when overexpressed.

The NXF1:NXT1 complex is a major export factor for bulk mRNAs (7). Additional factors, such as SR proteins, have also been shown to participate in mRNA export (27). Increasingly, it is becoming apparent that the nuclear export of specific classes of mRNAs can be selective (5, 28-30). In this context, we have recently shown that MKRN2, an RNA-binding E3 ubiquitin ligase, specifically binds to mRNAs that contain CU-rich regions in the 3′ UTR and regulates their export (31). Results presented in the current study point toward a role for Znf684 as an adapter for Nxf1 to regulate the export of specific mRNAs. Although many of the mechanistic details remain elusive, our observation that Znf684 overexpression inhibits export from the nucleus of certain mRNA transcripts suggests that it may have an inhibitory role in the nuclear export of a subset of mRNAs. Further experiments are necessary to explicitly elucidate detailed aspects of the mechanism by which Znf684 influences mRNA export. It also remains to be seen whether the DNA-or RNA-binding ability of Znf684 has a role in the observed expression changes. To this end, correlational analyses of iCLIP-seq, ChIP-seq, and knock-down RNA-seq data for Znf684 are under way.

Mitochondrial translation is another potential topic for future research on Znf684, as our AP-MS data showed that Znf684 interacts with many mitochondrial ribosomal proteins.

Consistently, our IF experiments using Mitotracker showed that Znf684 localizes to the mitochondria. In line with Znf684 having a function in mitochondrial biology, the KEGG pathway analysis of our iCLIP data indicated that Znf684 binds to a subset of mRNAs that are involved in organelle organization. Together, these results point towards Znf684 not only having a role in the mRNA nuclear export pathway, but also having some function in the mitochondria. This function could be related to mitochondrial gene translation, given the association with mitochondrial ribosomal proteins. 

Co-IPs were performed as described previously (37, 38). Cell pellets were lysed in lysis buffer (140mM NaCl, 10mM Tris pH 7.6-8.0, 1% Triton X-100, 0.1% sodium deoxy-cholate, 1mM EDTA) containing protease inhibitors. Cell extracts were incubated with 75 units of Benzonase (Sigma E1014) for 30 minutes in the cold room with end-to-end rotation. Cell debris was separated by microcentrifuging at 15,000g for 30 minutes at 4°C. The supernatant was incubated with 2.5μg of GFP antibody (Life Technologies G10362) overnight, and then 25μL protein G beads were added and incubated for an additional 2 hours with end-to-end rotation at 4ºC. The beads were separated using a magnetic rack, and then the supernatant was removed. Next, the beads were resuspended in 1mL lysis buffer containing an additional 2% NP40 and 1% Triton X and rotated for 5 minutes in a cold room. This step was repeated three times. The samples were then boiled in SDS gel sample buffer. Samples were resolved using 10% SDS-PAGE and transferred to a PVDF membrane using a Gel Transfer Cell. Primary antibodies were used at 1:5000 dilution, and secondary antibodies were used at 1:10,000. Blots were developed using ECL Western Blotting Substrate.

The AP-MS procedure was performed essentially as described (21, 39). Briefly, two independent batches of ∼20x10 6 cells were grown representing two biological replicates.

Expression of the tagged protein was induced using doxycycline 24 h prior to harvesting. HEK293 cell pellets were lysed in high-salt NP-40 lysis buffer (10 mM Tris-HCl pH 8.0, 420 mM NaCl, 0.1% NP-40, plus protease/phosphatase inhibitors) with three freeze-thaw cycles. The lysate was sonicated and treated with Benzonase for 30 min at 4⁰C with end-to-end rotation. The cell lysate was centrifuged to pellet any cellular debris. We immunoprecipitated GFP-tagged Znf684 with 1 μg anti-GFP antibody (G10362, Life Technologies) overnight followed by a 2-hour incubation with Protein G Dynabeads (Invitrogen). The beads were washed 3 times with buffer (10mM TRIS-HCl, pH7.9, 420mM NaCl, 0.1% NP-40) and twice with buffer without detergent (10mM TRIS-HCl, pH7.9, 420mM NaCl). The immunoprecipitated proteins were eluted with NH4OH and lyophilized. Proteins for MS analysis were prepared by in-solution trypsin digestion. Briefly, protein pellets were resuspended in 44uL of 50mM NH4HCO3, reduced with 100mM TCEP-HCL, alkylated with 500mM iodoacetamide, and digested with 1 μg of trypsin overnight at 37°C. 

Individual nucleotide resolution UV crosslinking and immunoprecipitation (iCLIP) was performed as previously described (42) with the modifications detailed in our previous report (37).

Briefly, cells were grown in 15cm culture plates and were UV cross-linked with 0.4J/cm2 at 254nm in a Stratalinker 1800 after induction with Doxycycline (1μg/ml) for 24 hours. Cells were lysed in 2mL of iCLIP lysis buffer. 1mL of lysate was incubated with 4 μL Turbo DNase (Life Technologies catalogue number AM2238) and 20 μL RNase I (1:250; Ambion catalogue number AM2294) for 5 minutes at 37°C to digest the genomic DNA and obtain RNA fragments within an optimal size range. GFP-Znf684 was immunoprecipitated using 5 μg of anti-GFP antibody (Life Technologies G10362). A total of 2% input material was obtained for size-matched control libraries (SMI) prior to the IPs. Following stringent washes with iCLIP high salt buffer and dephosphorylation with T4 polynucleotide kinase, on-bead-ligation of pre-adenylated adaptors to the 3'-ends of RNAs was performed using the enhanced CLIP ligation method (43). The immunoprecipitated RNA was 5'-end-labeled with 32P using T4 polynucleotide kinase (New England Biolabs catalogue number M0201L), separated using 4-12% BisTris-PAGE, and transferred to a nitrocellulose membrane (Protran). For the input sample, the membrane was cut to match the size of the IP material. RNA was recovered by digesting proteins using proteinase K (Thermo Fisher catalogue number 25530049) and subsequently reverse transcribed into cDNA.

The cDNA was size-selected (low: 70 to 85 nt, middle: 85 to 110 nt, and high: 110 to 180 nt), circularized to add an adaptor to the 5'-end, linearized, and then PCR amplified using AccuPrime SuperMix I (Thermo Fisher catalog number 12344040). The final PCR libraries were purified on PCR purification columns (QIAGEN), and the eluted DNA was mixed at a ratio of 1:5:5 from the low, middle, and high fractions and submitted for sequencing on an Illumina HiSeq2500 to generate single-end 51 nucleotide reads with 40M read depth.

Fractionation was performed as described (31). Cell pellets were resuspended in 1X PBS, then 10% of the cells were separated and pelleted by microcentrifuging at 400g. The remaining 90% was kept as the total fraction. The 10% cells were resuspended in 1X  buffer (150mM potassium acetate, 5mM magnesium acetate, and 20mM HEPES pH 7.4), and 1X -plus buffer (1X  buffer with the following added before use: 1mM DTT, 0.01% sodium deoxycholate, 0.25% Triton X, and protease inhibitor cocktail (Roche catalogue number 05892791001)). The mixture was iced for 3 minutes, and microcentrifuged for 5 minutes at 400g. The resulting supernatant served as the cytoplasmic fraction. The nuclear and endoplasmic reticulum (ER) fractions remained in the pellet, which was resuspended with 1X  buffer and 1X  plus buffer, then microcentrifuged for 5 minutes at 400g. The supernatant containing the ER was removed, and the pellet was washed with 1X  buffer. The resulting pellet containing the nuclear fraction was resuspended in 1X  buffer. Samples were resolved using 10% SDS PAGE and probed with nuclear (U170K, Abcam ab83306) marker antibodies at 1:1000 dilution and cytoplasmic (GAPDH, Abcam ab8245) marker antibodies at 1:10,000 dilution.

IFs were performed as described (37). GFP-Znf684 and GFP expressing HEK293 cells were seeded on poly-L-lysine coated and acid-washed coverslips. Protein expression was induced using 1μg/ml doxycycline for 24 hours. Cells were washed three times with PBS, then fixed in 4% Paraformaldeyde for 15 minutes. Cells were subsequently permeabilized with 0.2% Triton X-100 in PBS for 5 minutes and incubated with block solution (1% goat serum, 1% BSA, 0.5% Tween-20 in PBS) for 1 hour. GFP antibody was used for staining at 1:100 concentration in block solution for 2 hours at room temperature (RT). Cells were incubated with Goat anti-mouse secondary antibody and Hoescht stain in block solution for 1 hour at room temperature. Cells were fixed in Dako Fluorescence Mounting Medium (S3023). Imaging was performed the next day using a Zeiss confocal spinning disc AxioObserverZ1 microscope equipped with an Axiocam 506 camera using Zen software. A single focal plane was imaged.

RT-qPCR was performed as described (31). Total RNA was extracted using Trizol Reagent For qPCR, the following program was used: 40 cycles of 95°C for 15 s and 60°C for 30 s, with a final cycle of 95 °C for 15 s and then 60 °C. Actin was used as loading control. The data was analyzed using the ∆∆CT method as described previously (44). Primer sequences are provided in Supplementary Table S2 . 

C2H2 zinc finger proteins greatly expand the human regulatory lexicon

Evolution of C2H2-zinc finger genes and subfamilies in mammals: species-specific duplication and loss of clusters, genes and effector domains

KRAB zinc finger proteins

C2H2 Zinc Finger Proteins: The Largest but Poorly Explored Family of Higher Eukaryotic Transcription Factors

Keep your fingers off my DNA: Protein-protein interactions mediated by C2H2 zinc finger domains

Sequences encoding C2H2 zinc fingers inhibit polyadenylation and mRNA export in human cells

Transcription factor trapping by RNA in gene regulatory elements

We would like to thank Nujhat Ahmed, Giovanni Livingston Burke, and Kristie Ng for their experimental help and advice. Ernest Radovani is gratefully acknowledged for helpful discussions and assistance with MS analyses. Guoqing Zhong is acknowledged for her support in tissue culture. We would also like to thank Edyta Marcon for her help in obtaining materials and administrative support.