ZNF354C is a transcriptional repressor that inhibits endothelial angiogenic sprouting

Zinc finger proteins (ZNF) are a large group of transcription factors with diverse functions. We recently discovered that endothelial cells harbour a specific mechanism to limit the action of ZNF354C, whose function in endothelial cells is unknown. Given that ZNF354C has so far only been studied in bone and tumour, its function was determined in endothelial cells. ZNF354C is expressed in vascular cells and localises to the nucleus and cytoplasm. Overexpression of ZNF354C in human endothelial cells results in a marked inhibition of endothelial sprouting. RNA-sequencing of human microvascular endothelial cells with and without overexpression of ZNF354C revealed that the protein is a potent transcriptional repressor. ZNF354C contains an active KRAB domain which mediates this suppression as shown by mutagenesis analysis. ZNF354C interacts with dsDNA, TRIM28 and histones, as observed by proximity ligation and immunoprecipitation. Moreover, chromatin immunoprecipitation revealed that the ZNF binds to specific endothelial-relevant target-gene promoters. ZNF354C suppresses these genes as shown by CRISPR/Cas knockout and RNAi. Inhibition of endothelial sprouting by ZNF354C is dependent on the amino acids DV and MLE of the KRAB domain. These results demonstrate that ZNF354C is a repressive transcription factor which acts through a KRAB domain to inhibit endothelial angiogenic sprouting.

ZNF354C overexpression leads to the transcriptional repression of many genes. To identify the impact of ZNF354C on gene expression, mRNA-sequencing was performed in HMEC-1 (Fig. 2a, Sup. Table 1). As compared to control transfected cells, ZNF354C overexpression mainly decreased gene expression (Fig. 2b). In detail, 167 genes were significantly (padj.) downregulated with a log2 fold change (logFC) of more than 0.5, whereas only 7 genes were upregulated (Sup. Fig. 2a,b). A screen for the presence of the ZNF354C motifs in the promoters of the differentially regulated genes (DEGs) revealed that more than 60% of these contain the sequence 5′-NNCCAC-3′ within the 0-500 bp upstream of their transcriptional target site (Fig. 2c). RT-qPCR after overexpression of ZNF354C confirmed the RNA-seq results (Fig. 2d). Gene set enrichment analysis of the downregulated genes (logFC > 0.3) revealed "ZNFs", primary cilium and microtubule assembly/ organisation processes to be enriched in the ZNF354C regulated genes (Fig. 2e,f). This suggests that ZNF354C acts as a direct transcriptional repressor for the majority of the DEGs in endothelial cells.

Specific mutations of the KRAB-A box abolish the repressive effects of ZNF354C on endothelial cells.
The KRAB domain has been shown to be critical for transcriptional repression in KRAB-ZNFs and mutations in the KRAB-A box disrupt transcriptional repression [16][17][18] (Fig. 3a). The critical amino acids of the KRAB-A box are also present in ZNF354C (Fig. 3b,c). To determine whether these amino acids of the A-Box are required for the function of ZNF354C, three pCMV6-ZNF354C mutant constructs were generated (DV → AA, EW → AA, and MLE → KKK) by site-directed mutagenesis (Fig. 3a). Overexpression of the MLE → KKK mutant in HMEC-1 resulted in higher target gene expression than the WT plasmid (dotted line). The DV → AA as well as the EW → AA only partially increased target gene expression ( Fig. 3d-f). This suggests that the MLE sequence mediates target gene transcriptional repression in ZNF354C.
ZNF354C co-precipitates on specific target gene promoters and interacts with dsDNA, TRIM28 and Histone 3. KRAB-ZNFs bind to DNA with the zinc finger domains and many interact with Tripartite Motif Containing 28 (TRIM28, also known as KRAB-associated protein 1 or TIF1B) via their KRAB domain to mediate transcriptional repression 17,19 . Using proximity ligation assays, an interaction of ZNF354C with dsDNA and TRIM28 was also observed in endothelial cells (Fig. 4a,b). Immunoprecipitation with antibodies targeting dsDNA, Histone 3 or the Histone 3 modification lysine 4 trimethylation (H3K4me3) confirmed these findings and also demonstrated that ZNF354C interacts with chromatin (Fig. 4c). Overexpression of ZNF354C in HMEC-1 followed by chromatin immunoprecipitation revealed that ZNF354C indeed binds promoter sequences close to the transcriptional start site of the selected target genes Regulator Of G Protein Signaling 5 (RGS5), Protein Associated With Topoisomerase II Homolog 2 (PATL2), Huntingtin associated protein 1 (HAP1) and the long non-coding RNA Negative Regulator Of Antiviral Response (NRAV) (Fig. 4d). Importantly, RGS5 is relevant in the cardiovascular system as it promotes tumour growth 20 . No binding was observed for the negative control ADAM Metallopeptidase with Thrombospondin Type 1 Motif 1 (ADAMTS1) (Fig. 4d). Thus, ZNF354C regulates transcriptional repression through direct DNA binding. LentiCRISPRv2-mediated knockout or siRNA-mediated depletion of ZNF354C increases target gene expression. In order to identify the physiological target genes of ZNF354C in endothelial cells, loss of function experiments were performed. Genes were either knocked out by CRISPR/Cas9 or downregulated with siRNAs. LentiCRISPRv2 with 6 different guide RNAs (gRNAs) targeting genomic sequences 5′ and 3′ of the transcriptional start site of ZNF354C were used to produce a knockout of ZNF354C in HMEC-1 (Fig. 5a). After puromycin selection, genotyping revealed a positive outcome for most of the gRNA combinations  . 5b); however, only a combination of all gRNAs strongly reduced ZNF354C protein and mRNA expression (Fig. 5c,d). HMEC-1 subjected to CRISPR-mediated knockout had an approximately twofold higher expression of the selected target genes, among them RGS5 (Fig. 5e).
To substantiate these results, knockdown of ZNF354C was performed with two different siRNAs in HMEC-1, HUVECs and HAoSMCs. Surprisingly, although ZNF354C mRNA expression levels strongly decreased, no obvious effect was detected on protein level after 48 h (Sup. Fig. 3a,b). Application of the protein synthesis  www.nature.com/scientificreports/ inhibitor cycloheximide revealed that ZNF354C is a highly stable protein with a low turnover rate whose nuclear expression only starts to decrease after 48 h of translational inhibition (Sup. Fig. 3c,d). Nonetheless, knockdown of ZNF354C in HMEC-1 for 72 h revealed decreased protein levels ( Fig. 5f) and demonstrated increased expression levels of the target genes ( Fig. 5g), supporting the concept that ZNF354C is a transcriptional repressor of these genes.
KRAB-A box mutations restore cellular angiogenic sprouting. Finally, we investigated whether the ZNF354C A-Box KRAB domain mutants have a dominant negative effect on endothelial angiogenic sprouting. Compared to the overexpression of the WT plasmid, the MLE → KKK mutant failed to affect endothelial cell sprouting potential ( Fig. 6a-g). A similar effect was observed for the DV → AA mutant, whereas the EW → AA mutant acted as the WT plasmid. None of the constructs, however, induced a higher degree of sprouting than the negative transfection control. This indicates that either the mutants do not exhibit a dominant negative effect or that the endogenous levels of ZNF354C are too low to impact endothelial angiogenic sprouting.

Discussion
The present study demonstrates that the KRAB-containing C2H2-type zinc finger protein ZNF354C is an important transcription factor that facilitates endothelial cell function. Although ZNF354C is localised to both the nucleus and the cytoplasm, it interacted strongly with dsDNA, TRIM28 and the histone environment.  16 . Their inhibitory function is mediated by specific amino acids within the KRAB-A box [16][17][18] . These amino acids in ZNF354C are highly conserved to other ZNFs and to other mammalian homologues and thus are candidates for the repressive function of ZNF354C. Substitution of the amino acids DV → AA or MLE → KKK not only dramatically reduced the repressive activity of the well-known KOX1 KRAB domain 16 , it also prevented the binding of TRIM28 to the KRAB domain of KOX1 17 . Although the EW sequence has been linked to repression 16 , its role in ZNF354C was minimal. Substitution of MLE → KKK within the KRAB domain of ZNF354C not only abolished the repressive activity but also the inhibition on angiogenic sprouting-substitution of DV → AA was important for angiogenesis but not for transcriptional repression in our settings. This argues that the amino acids MLE within KRAB are probably the most important, potentially mediating the effect through co-repressor binding, as is the case with TRIM28.
In this study, strong differences between the overexpression of ZNF354C in HMEC-1, HUVEC and HAoSMC were observed. Although the overexpression was strongly detectable in HMEC-1, it was very much weaker in HUVECs and not detectable in HAoSMC. Most likely, this is a conseqence of the different transfection efficiencies between the cells. Whereas cell lines as HMEC-1 are usually fairly responsive to transfection, particular HAoSMC are hard to transfect.
The RNA-seq analysis not only confirmed ZNF354C as a transcriptional repressor, it further led to the identification of target genes of ZNF354C. PATL2 is important for mammalian oocyte maturation whose removal causes oocyte meiotic deficiency in humans 21 . HECT and RLD domain containing E3 ubiquitin protein ligase 5 (HERC5) is an E3 ubiquitin ligase which is upregulated by inflammatory cytokines in endothelial cells 22 . HAP1 is involved in intracellular trafficking and linked to Huntington disease 23 . The long non-coding RNA NRAV has been identified to be important for influenza A virus replication through suppression of interferon-stimulated www.nature.com/scientificreports/ gene transcription 24 . GO analysis revealed that many DEGs are part of cilia/microtubule associated processes, e.g. many Dynein Axonemal Heavy Chain proteins were repressed after ZNF354C overexpression but due to their low expression pattern in the raw RNA-Seq data, these proteins were not further studied. Also genes relevant for the cardiovascular system were suppressed by ZNF354C: Cardiomyopathy Associated 5 (CMYA5) and Actinin Alpha 2 (ACTN2) are differentially expressed in the myocardium during progression of hypertrophy 25 . Loss of murine RGS5 reduces tumour angiogenesis 20 . This may explain why, after ZNF354C overexpression, sprouting angiogenesis was inhibited.
Taken together, these findings suggest that ZNF354C is important for endothelial cell function by acting as a repressive transcription factor dependent on specific amino acids within its KRAB domain.

Methods
Cell culture. All cells were cultured in a humidified atmosphere of 5% CO2 at 37 °C. Cells were cultured similarly as described in 26

RNA isolation, reverse transcription and RT-qPCR. Total RNA Isolation was performed with the
Quick-RNA MiniPrep kit (Zymo Research, Freiburg, Germany) according to the manufacturer's protocol. For reverse transcription, SuperScript III Reverse Transcriptase (Thermo Fisher) and oligo(dT)23 together with random hexamer primers (Sigma) were used. CDNA amplification with the oligonucleotides indicated (Table 1) was performed with RT-qPCR using ITaq Universal SYBR Green Supermix and ROX as reference dye (Bio-Rad, #1725125) in an AriaMX cycler (Agilent). Relative expression of human target genes was normalised to b-actin (beta-actin) or 18S ribosomal RNA. Both expressions were analysed with the delta-delta Ct method using the AriaMX qPCR software (Agilent).

Protein isolation, western analysis and nuclear-cytoplasmic extraction. HUVECs were washed
in Hanks solution (Applichem). After cell lysis with Triton X-100 buffer (20 mM Tris/Cl pH 7.5, 150 mM NaCl, 10 mM NaPPi, 20 mM NaF, 1% Triton, 2 mM Orthovanadat, 10 nM Okadaic Acid, protein-inhibitor mix, 40 µg/ ml Phenylmethylsulfonylfluorid), the extract was centrifuged (10 min, 16,000×g, 4 °C). The protein concentration of the supernatant was determined with the Bradford assay. The cell extract was boiled in Laemmli buffer Immunoprecipitation. Nuclei isolation of 10 7 HMEC-1 per sample was performed as described above with buffer A (10 mM HEPES pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.75% Nonidet, 2 mM Orthovanadat, 10 nM Okadaic Acid, protein-inhibitor mix, 40 µg/ml Phenylmethylsulfonylfluorid) followed by buffer C (20 mM HEPES pH 7.9, 0.4 M NaCl, 1 mM EDTA, 1 mM EGTA, 2 mM Orthovanadat, 10 nM Okadaic Acid, protein-inhibitor mix, 40 µg/ml Phenylmethylsulfonylfluorid), each step with 15 min incubation at 4 °C. After pre-clearing of the supernatant with 20 µL DiaMag Protein A and Protein G beads (Diagenode), the lysed nuclei were incubated with H3-pan (#C15200011, Diagenode), H3K4me3 (#C15410003, Diagenode) or dsDNA [35I9 DNA] (#ab27156, Abcam) antibodies overnight at 4 °C. The samples were then incubated with 50 µl of total DiaMag Protein A and Protein G (Diagenode) beads for 2 h at 4 °C, followed by 3 washing steps in buffer C. Prior to elution, beads were put into a new Eppendorf tube, boiled in Laemmli buffer and western analyses was performed. mRNA-sequencing, bioinformatics and data deposition. For RNA-seq analysis, total RNA was isolated from HMEC-1 with Quick-RNA MiniPrep kit (Zymo Research, Freiburg, Germany) according to the manufacturer's protocol. RNA and library preparation integrity were verified with LabChip Gx Touch 24 (Perkin Elmer). 2 µg of total RNA was used as input for VAHTS Stranded mRNA-seq Library preparation following manufacturer's protocol (Vazyme). Sequencing was performed on NextSeq500 instrument (Illumina) using v2 chemistry, resulting in an average of 21Mio reads per library with 1 × 75 bp single end setup. The resulting raw reads were assessed for quality, adapter content and duplication rates with FastQC 27 . Trimmomatic version 0.39 was employed to trim reads after a quality drop below a mean of Q20 in a window of 5 nucleotides 28 . Only reads between 30 and 150 nucleotides were cleared for further analyses. Trimmed and filtered reads were aligned against the Ensembl human genome version hg38 (GRCh38) using STAR 2.6.1d with the parameter "-outFil-terMismatchNoverLmax 0.1" to increase the maximum ratio of mismatches to mapped length to 10% 29 . The number of reads aligning to genes was counted with featureCounts 1.6.5 tool from the Subread package 30 . Only Table 1. List of primers for qRT-PCR.
Chromatin immunoprecipitation (ChIP). HMEC-1 were electroporated with pCMV6-entry (#PS100001, Origene) or pCMV6-ZNF354C (NM_014594, #RC209312, Origene) and incubated for 24 h. Crosslinking and isolation of nuclei was performed with the truCHIP Chromatin Shearing Kit (Covaris, USA) according to the manufacturer's protocol. ChIP was performed as described in 35 . Samples were incubated overnight at 4 °C with a monoclonal anti-Flag M2 antibody (F3165-0.2MG, Sigma-Aldrich). 5% of the samples served as input. After elution, the DNA was purified with the QiaQuick PCR purification kit (Qiagen, Hilden, Germany) and subjected to qPCR analysis. As a negative control during qPCR, oligonucleotides targeting the GAPDH promoter were used. The primers are listed in Table 2. Table 2. List of promoter primers for ChIP.

RGS5
CAG TAG TGC CTG TAG CAG AG  GCC AAT CCA GAG CCT TAG AG   PATL2  GCT CAC ATG ACC CGG TGA AG  GGA AAG TCC TGG GTA GAT CC   HAP1  ATG ACC CCA GCT CAC CAA GG  CCA GGC AAA GAC TGA GGA CA   NRAV  AAA TAA GGC AGC GAG GAC AC  GTA GCG ACG GTA TCT CTA GC   ADAMTS1  TTC GGT TGG AGA ACG CAG TCC GAA GGT GGA GAA GTG GGG TGA G