11394–11407 Nucleic Acids Research, 2020, Vol. 48, No. 20 Published online 14 October 2020 doi: 10.1093/nar/gkaa837 The role of ZFP57 and additional KRAB-zinc finger proteins in the maintenance of human imprinted methylation and multi-locus imprinting disturbances Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Ana Monteagudo-Sánchez1 , Jose Ramon Hernandez Mora1 , Carlos Simon2,3 , Adam Burton4 , Jair Tenorio5,6 , Pablo Lapunzina5,6,7 , Stephen Clark8 , Manel Esteller9,10,11,12 , Gavin Kelsey 8,13 , Juan Pedro López-Siguero14 , Guiomar Perez de Nanclares15 , Maria-Elena Torres-Padilla4 and David Monk 1,16,* 1 Imprinting and Cancer group, Bellvitge Institute for Biomedical Research, Gran via, L’Hospitalet de Llobregat, Barcelona, Spain, 2 Department of Obstetrics and Gynecology, Valencia University and INCLIVA, Valencia, Spain, 3 Department of Obstetrics and Gynecology, BIDMC, Harvard University, Boston, MA, USA, 4 Institute of Epigenetics and Stem Cells, Helmholtz Zentrum München, München, Germany, 5 Medical and Molecular Genetics Institute, University Hospital La Paz, Madrid, Spain, 6 CIBERER, Centro de Investigacion Biomedica en Red de Enfermedades Raras, ISCIII, Madrid, Spain, 7 ITHACA, European Reference Network on Rare Congenital Malformations and Rare Intellectual Disability, 8 Epigenetics Programme, The Babraham Institute, Babraham, Cambridge, UK, 9 Josep Carreras Leukeamia Research Institute, Can Ruti, Cami de les Escoles, Badalona, Barcelona, Spain, 10 Department of Physiological Sciences II, School of Medicine, University of Barcelona, Barcelona, Catalonia, Spain, 11 Institucio Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Catalonia, Spain, 12 Centro de Investigacion Biomedica en Red Cancer (CIBERONC), Madrid, Spain, 13 Centre for Trophoblast Research, University of Cambridge, UK, 14 Servicio de Endocrinologı́a Pediátrica, Hospital Carlos de Haya, Málaga, Spain, 15 (Epi)Genetics Laboratory, BioAraba Research Health Institute, Araba University Hospital, Vitoria-Gasteiz, Alava, Spain and 16 Biomedical Research Centre, University of East Anglia, Norwich Research Park, Norwich, UK Received June 09, 2020; Revised September 10, 2020; Editorial Decision September 17, 2020; Accepted October 13, 2020 ABSTRACT only expressed following embryonic-genome activa- tion, implying that other KRAB-zinc finger proteins Genomic imprinting is an epigenetic process reg- (KZNFs) recruit KAP1 prior to blastocyst formation. ulated by germline-derived DNA methylation that Furthermore, we uncover ZNF202 and ZNF445 as ad- is resistant to embryonic reprogramming, resulting ditional KZNFs likely to recruit KAP1 to imprinted in parental origin-specific monoallelic gene expres- loci during reprogramming in the absence of ZFP57. sion. A subset of individuals affected by imprinting Together, these data confirm the perplexing link be- disorders (IDs) displays multi-locus imprinting dis- tween KZFPs and imprint maintenance and highlight turbances (MLID), which may result from aberrant the differences between mouse and humans in this establishment of imprinted differentially methylated respect. regions (DMRs) in gametes or their maintenance in early embryogenesis. Here we investigated the extent of MLID in a family harbouring a ZFP57 truncating INTRODUCTION variant and characterize the interactions between hu- Genomic imprinting in mammals refers to a highly- man ZFP57 and the KAP1 co-repressor complex. By coordinated epigenetic mechanism that leads to parental ectopically targeting ZFP57 to reprogrammed loci in allele-specific gene expression (1). Imprinted genes are mouse embryos using a dCas9 approach, we confirm known to play critical roles in fetal development, placenta that ZFP57 recruitment is sufficient to protect oocyte- and brain function, which when perturbed, result in im- derived methylation from reprogramming. Expres- printing disorders (IDs) (2). The characteristic monoal- sion profiling in human pre-implantation embryos lelic expression of imprinted genes is largely conferred and oocytes reveals that unlike in mice, ZFP57 is by allele-specific DNA methylation, referred to as differ- * To whom correspondence should be addressed. Tel: +44 1603 591 546; Email: d.monk@uea.ac.uk  C The Author(s) 2020. Published by Oxford University Press on behalf of Nucleic Acids Research. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. Nucleic Acids Research, 2020, Vol. 48, No. 20 11395 entially methylated regions (DMRs) that act as imprint- whether ZFP57 can directly regulate those loci has not ing control regions (ICRs). The differential DNA methy- been addressed. lation at the majority of ICRs is inherited from the ga- In this study, we investigate the stability of mosaic hy- metes and maintained in pre-implantation embryos dur- pomethylation at imprinted DMRs in a family with a ing global epigenetic reprogramming (3). The resulting ZFP57 truncating mutation and characterize ZFP57-KAP1 germline-derived DMRs (gDMRs) in general are ubiqui- functional interactions in humans. Furthermore, we iden- Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 tously present in all tissues later in development, but a sub- tify additional KZFPs that are expressed at the appropri- set have a tissue-specific profile, often restricted to the pla- ate time during human development that have the ability to centa (4,5). In addition to gDMRs, a variety of mechanisms recruit the KAP1. Together, these results shed light on the ensure monoallelic expression of imprinted transcripts, in- species-specific function of ZFP57 in regulating imprinting cluding somatically acquired DMRs (also known as sec- maintenance during development. ondary DMRs) and allele-specific histone modifications, including H3K9me3 and H3K27me3 (6,7). Intriguingly, a MATERIALS AND METHODS number of tissue-specific imprinted genes, whose temporal imprinting is limited to cleavage stage embryos and pla- Ethics centa, appear to utilize non-canonical mechanisms not as- The family with the ZFP57p.E123* mutation provided in- sociated with gDMRs but reliant on histone modifications formed consent to be involved in this study. The eth- (8–10). ical approval for this aspect of the work was granted Gene encoding Kruppel-associated (KRAB)-zinc finger by the Bellvitge Institute of Biomedical Research Ethics proteins (KZFPs), which are involved in the establishment Committee (PR096/10), the Basque Ethics Committee and maintenance of repressive chromatin states, are highly (IRB#PI2013214 and PI2017018) and Hospital Universi- abundant in the mammalian genome (11). KZFPs have tario La Paz (IRB-HULP PI446). Collecting placenta sam- classically been linked to the silencing of transposable ples was approved by the Bellvitge Institute of Biomedical elements, with underlying sequence-specificity attributed Research Ethics Committee (PR006/08). The use of sur- to unique combinations of zinc-finger that target different plus human oocytes and embryos for this study was eval- retrotransposon families (12). Once recruited to a sequence, uated and approved by the scientific and ethic committee of transcriptional repression is mediated by the recruitment the Instituto Valenciano de Infertilidad (IVI) (1310-FIVI- of the KAP1 co-repressor complex, which induces the local 131-CS), Centro de Medicina Regenerativa de Barcelona deposition and maintenance of H3K9me3 (13). KZFPs (CMRB), the National Committee for Human Reproduc- have been shown not only to recognise transposable el- tion (CNRHA) and the Regional Health Departments for ements, but an increasing number of these proteins also Valencia and Catalyuna (10/2017). Ethical approval for the bind to single-copy regions in the genome. One of the using hES cells was granted by the Bellvitge Institute for most-studied members in this respect is ZFP57, which in Biomedical Research Ethics Committee (PR096/10) and addition to Intracisternal A Particle (IAP) Long Terminal Comites Éticos de Investigación Clı́nica (CEIC) del Centro Repeats (LTR) retrotransposons, also binds to imprinted de Medicina Regenerativa de Barcelona-CMR[B] (28/2012) gDMRs (14–17). Because ZFP57 is highly abundant in and complied with the legal guidelines outlined by the Gen- mouse oocytes and recognises methylated CpG dinu- eralitat de Catalunya and the Conselleria de Salut. Mouse cleotides in the TGCCme GC hexamer consensus motif, this work was approved and performed under the authoriza- KZFP is largely responsible for maintaining methylation tion of the Upper Bavarian authorities. Animal husbandry at the majority of gDMRs during mouse pre-implantation and breeding were conducted according to the institutional reprogramming (15,18,19). The high oocyte expression guidelines for the care and the use of laboratory animals. of Zfp57 is consistent with it being a maternal-zygotic effect gene in mice. When expression is depleted in zygotes Patient samples derived from heterozygous females (Zfp57 –/– offspring from Zfp57+/– mothers) the severity of both methylation A family with siblings diagnosed with TNDM who were re- defect and penetrance of the phenotype are less severe ferred to the paediatric endocrinology at Carlos de Haya than in homozygous nulls from Zfp57 –/– mothers (16). Hospital (Malaga, Spain) were used in this study. The pre- Recessive mutations in the human orthologue causes viously identified mutation was confirmed in proband II.2 transient neonatal diabetes mellitus (TNDM) associated using standard PCR and sequencing as described by Court with multi-locus methylation disturbance (MLID) (20), and colleagues (23). Two separate cohorts of MLID patients a phenomenon in which loss-of-methylation (LOM) were screened for mutations, the first comprising of mainly at numerous imprinted DMRs is observed in addition Beckwith–Wiedemann syndrome patients were from the to the disease-associated loci (2). To date, 14 families Spanish overgrowth registry (n = 28), while patients with have been described with either missense or truncating Pseudohypoparathyroidism type 1B (PHP1B; n = 10) and mutations in ZFP57 (21,22), all have a notably similar TNDM (n = 3) were from the BioAraba Research Health methylation signature; complete hypomethylation of the Institute. A collection of 15 human leukocyte and buccal PLAGL1 DMR (PLAGL1:alt-TSS-DMR) and variable swabs were collected from volunteers at PEBC-IDIBELL combinations of mosaic hypomethylation at GRB10 and used as controls to determine the normal range of (GRB10:alt-TSS-DMR), PEG3 (PEG3:TSS-DMR), methylation at imprinted DMRs. Twelve human metaphase MEST (MEST:alt-TSS-DMR), NAP1L5 (NAP1L5:TSS- II oocytes and a single surplus human blastocyst were ob- DMR) and GNAS (GNAS-AS1:TSS-DMR). However, tained from the Instituto Valenciano de Infertilidad (FIVI) 11396 Nucleic Acids Research, 2020, Vol. 48, No. 20 (Valencia, Spain). Vitrified oocytes were thawed using a Pyrosequencing analysis for methylation quantification kit for de-vitrification (Kitazato Valencia, Spain) following Approximately 50 ng of bisulphite converted DNA was manufacturer’s instructions. used for PCR prior to pyrosequencing. Standard bisul- phite PCR was used to amplify the imprinted DMRs with Single cell lymphoblastoid colony expansion the exception that one primer was biotinylated (see Sup- Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 plementary Table S1 for primer sequences). In all cases Isolated clonal cell lines were generated by dilution plating the entire biotinylated PCR product (diluted to 40 ␮l) of a polyclonal lymphoblastoid cell as previously described was mixed with 38 ␮l of Binding buffer and 2 ␮l (10 (24). Transformation with Epstein–Barr virus (EBV) was mg/ml) streptavidin-coated polystyrene beads. After wash- performed on a fresh blood sample from patient II.1. Iso- ing in 70% ethanol, DNA was denaturated with 50 ␮l 0.5M lated B-cells within the sample were immortalized with the NaOH. The single-stranded DNA was hybridized to 40- supernatant of the EVB producer cell line B95.8. Surviving pmol sequencing primers dissolved in 11 ␮l annealing buffer colonies were propagated to a density sufficient for DNA at 80◦ C. For sequencing, forward primers were designed isolation. to the complementary strand. The pyrosequencing reaction was carried out on a PyroMark Q96 instrument. The peak Methylation array hybridisation heights were determined using Pyro Q-CpG1.0.9 software (Biotage). We generated methylation datasets for affected family mem- bers with the ZFP57p.E123* mutation using the Illumina Infinium HumanMethylation450 (HM450k) and Infinium Nested multiplex bisulphite PCR HumanMethylationEPIC (EPIC) BeadChip arrays. Bisul- We employed a multiplex nested PCR approach to max- phite conversion of 600 ng of DNA was performed accord- imise data generation from individual mouse embryos ing to the manufacturer’s recommendations for the Illu- which were subject to direct bisulphite conversion. The pro- mina Infinium Assay (EZ DNA methylation kit, ZYMO, tocol has been described elsewhere (27); essentially two sets Orange, CA, USA). The bisulphite-converted DNA was of primers (outer and inner pairs) were designed to each lo- used for hybridisation following the Illumina Infinium HD cus and robustly optimized to ensure efficient amplification methylation protocol at genomic facilities at IDIBELL of both methylated and unmethylated strands at a single an- (Barcelona, Spain). In addition to these samples, we also nealing temperature without contamination or the forma- utilized our control leukocyte datasets from GSE103738 tion of primer dimers (primers detailed in Supplementary and GSE149572 (25). Table S1). For the multiplex step, outer primers (four sep- arate pairs targeting different loci) were co-amplified in the Statistical analyses of the Illumina arrays first reaction using Immolase Taq polymerase (Bioline) for 45 cycles. The multiplex PCR was carried in a total volume Raw data was normalised using the default control probes of 97 ␮l, including the 30 ␮l of embryo converted DNA as in BeadStudio (version 2011.1) Infinium HD), subtracting a template. We included two controls: blank with no sam- signal background and correcting for interpolate variation. ple and 10 ng of BS-converted DNA as a positive control. Probes with a detection P-value >0.01, containing SNPs The nested second round amplifications utilized the inter- within the integration or extension bases as well as those nal primer pairs specific for each region with individual with potential cross-reaction due to multiple sequences ho- PCRs performed on aliquots of the first-round product. The mologies were discarded. Probes lacking signal values in nested reaction was performed in a total volume of 25 ␮l one or more of the analysed samples were discarded also. reaction using Immolase Taq polymerase (Bioline). These For the analysis of known imprinted domains, probes map- were amplified for 45 cycles and the resulting amplicons ping to the DMRs identified by Court et al. and Hernandez used for pyrosequencing or sub-cloned into the pGEM-T Mora et al. were directly examined (25,26). easy vector. Allele-Specific bisulphite PCR Single colony restriction enzyme-based analysis of methyla- tion For standard bisulphite conversion, we used the EZ DNA Methylation-Gold kit (ZYMO, CA, USA) according to To analyse the methylation state of four different regions in the manufacturer’s instructions. Approximately 2 ␮l of the lymphoblastoid single colonies we performed a proto- bisulphite-converted DNA was used in each amplifica- col similar to the one described by Cheow and colleagues tion reaction catalysed by Immolase Taq polymerase (Bi- (28). After collecting the DNA of each of the clonal lym- oline). The resulting PCR product was sub-cloned into phoblastoid population, the DNA was digested with the the pGEM-T easy vector (Promega) for sequencing (for methylation-sensitive enzyme BstUI for 2 h at 37◦ C in a primer sequences see Supplementary Table S1). The EZ thermocycler. The digested DNA was subjected to ampli- DNA Methylation-Direct kit (ZYMO), in which cell lysates fication by PCR using primers flanking a restriction site in are subject to direct bisulphite conversion, was used for the the region of interest, and the methylation state was identi- methylation profiling of individual blastocysts and pooled fied depending on the size of the amplicon, larger amplicons oocytes. Nested bisulphite PCR was required to obtain am- were methylated and small amplicon resulted irrespective of plicons for sub-cloning due to the small amounts of avail- methylation status (primer sequences are detailed in Supple- able samples. mentary Table S1). Nucleic Acids Research, 2020, Vol. 48, No. 20 11397 Genotyping and imprinting analysis with BC100 buffer (10 mM Tris pH 7.8, 0.5 mM EDTA, 0.1 mM phenylmethylsulfonyl fluoride, 0.1 mM dithiothreitol Potential SNPs in ZFP57 exons and hap-ASM were iden- (DTT), 10% glycerol, 100 mM KCl) and three times with tified in the UCSC hg19 browser and genotyped by PCR BC500 buffer (500 mM KCl). Then, proteins were eluted and direct sequencing in hES cell lines. Sequence traces were with 0.2 M Glycine pH 2.3. interrogated using Sequencher v4.6 (Gene Codes Corpo- Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 ration, MI) to distinguish heterozygous and homozygous samples. Heterozygous sample cell lines were analyzed for Western blotting allelic expression using RT-PCR and bisulphite PCR, incor- Protein extracts and co-IPs were mixed with 5× SDS load- porating the polymorphism within the final PCR amplicon ing buffer (1 M Tris pH 7.8, 87% glycerol, 10% SDS and (see Supplementary Table S1 for primer sequences). 0.1 g of bromophenol blue), and samples were boiled for 5 min at 95◦ C. Samples were loaded on a gel along with 5 ␮l of Quantitative Real-Time-PCR pre-stained PageRuler protein ladder (Fermentas) and run Expression of ZFP57 isoforms, ZNF202 and ZNF445 at 25 mA. The gels were blotted with transfer buffer (Tris transcripts were analyzed using a fluorochrome (SYBR® 0.25M, glycine 1.92 M and 1% SDS) on to nitrocellulose Green)-based quantitative real-time RT-PCR assay and membrane in a blotting chamber (Bio-Rad) for 45 min at normalized against RPL19. cDNA was synthesized as pre- 0.40 A. Successful transfer of proteins was confirmed by viously described (29). All assays were run in triplicate in 96- Ponceau (Sigma) staining. Then, membrane was blocked well or 384-well plates using Real-Time-PCR System (Ap- with 5% low fat milk (Merck) diluted in PBS-Tween for 1 plied Biosystems). Only samples with two or more valid h at room temperature. The membrane was subsequently readings per triplicate were included in our analysis. Disso- washed three times in PBS–Tween and incubated with the ciation curves were generated at the end of each reaction to primary antibody, in 5% low fat milk overnight at 4◦ C on rule out the presence of primer-dimers or unexpected DNA a horizontal agitator. Afterwards, membrane was washed species in the reaction. Results were analyzed with the SDS three times for 5 min with PBS-Tween, then the correspond- 2.3 software (Applied Biosystems). ing peroxidase-conjugated secondary antibody was applied in 5% low fat milk dilution 1:5000 for 1 h at room temper- ature. The membrane was washed with PBS–Tween three Antibodies times more and placed on a developing machine. The primary antibodies used included ␣-HA (Sigma H6900; co-IP 5 ␮g; WB 1:1000), ␣-FLAG (Sigma 180, 5 ␮g; Histone peptide array WB 1:1000 ChIP 5 ␮g), ␣-myc (Sant Cruz Biotech 9E10, 5 ␮g; WB 1:5000), ␣-V5 (Bio-Rad MCA1360, co-IP 5 ␮g; WB MODified™ Histone peptide Array (Active Motif) contain- 1:1000) and ␣-B actin (Sigma, A3854, WB 1:1000). The sec- ing 384 different histone modification combinations was ondary antibodies used were ␣-mouse (Sigma A9044) and used. The array was blocked overnight in 5% low fat milk ␣-rabbit (Sigma A0545) (Merck) diluted in TBS–Tween solution, followed by three washes for 5 min with TBS–Tween and incubated overnight in an orbital shaker with 6 ␮M of the corresponding pu- Transfection, protein extraction and immunoprecipitation rified protein in TBS–Tween. The array was washed three All cell lines were grown in DMEM (Life Technologies) sup- times in TBS–Tween and incubated with ␣-HA primary an- plemented by 10% fetal bovine serum (Life Technologies). tibody, in 5% low fat milk for 3 h on a horizontal blocker at Cells were routinely tested for mycoplasm infection with 4◦ C. Following three more washes in TBS-Tween, ␣-rabbit PCR- detection kit (Biotools-Biot. & Med). All transfec- peroxidase-conjugated secondary antibody was applied in tions for mammalian over-expression were carried out using 5% low fat milk dilution 1:2500 for 1 h at room tempera- PEI reagent following supplier’s recommendations. Trans- ture. Following three last washes in TBS–Tween the array fections with Cas9 utilized OptiMEM media (Gibco) and filters were develop and quantified. Lipofectamine 3000 (ThermoFisher) using recommended protocols. dCas9-fusion plasmids, in vitro transcription, zygote microin- Nuclear and cytoplasm extract were prepared for jections and embryo culture protein extraction to perform western blot and co- immunoprecipitation assays. Cell were pelleted and washed The dCas9-Zfp57[KRAB]-T2A-GFP fusion construct was twice with cold PBS, and proteins were extracted according generated using the pLV hUbC-dCas9-VP64-T2A-GFP to the Dignam protocol: first extraction for the cytoplasmic plasmid backbone (30). The NheI sites were used to re- soluble fraction was made with Buffer A (10 mM Tris pH move VP64 and introduce mouse ZFP57 (amino acids 13– 7.8; 10 mM KCl; 1.5 mM MgCl2 ) and second extraction for 55 encoding the KRAB domain). We also generated the the nuclear soluble fraction was made with Buffer C (10 mM control plasmid, dCas9-control, with no insert (Nhe1 self- Tris pH 7.8; 0.42 M NaCl; 1.5 mM MgCl2 ; 0.2 mM EDTA; ligation). In vitro transcription (IVT) was utilized to gener- 25% glycerol). Extracts were kept at −80◦ C until use. ate full-length RNA for zygote injections using the mMES- For immunoprecipitation experiments, cell extracts were SAGE mMACHINE® T7 Ultra Kit (Thermofisher). The incubated with either ␣-HA resin or ␣-FLAG, ␣-V5 or ␣- IVT template was the PCR product generated using the Myc tag antibody crosslinked to protein G-magnetic beads dCas9-Zfp57[KRAB]-T2A-GFP construct amplified with (ThermoFisher), overnight. Beads were washed three times a forward primer encompassing the T7 RNA polymerase 11398 Nucleic Acids Research, 2020, Vol. 48, No. 20 promoter upstream of the dCas9 sequence and a reverse the additional clinical features observed in II.1 (Supplemen- primer with a stop codon and polyA tail designed to the tary Table S2). Upon molecular investigation this child was end of the GFP (primers detailed in Supplementary Ta- shown to have the same homozygous deletion (Figure 1). ble S1). The in vitro transcription, DNAseI treatment and Using the EPIC methylation arrays and bisulphite PCR LiCl clean-up were performed following the suppliers rec- with pyrosequencing, we confirmed a hypomethylated sig- ommendations. The resulting IVT RNA was resuspended nature similar to the sibling at ubiquitous imprinted DMRs Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 in TE and quantified by spectrophotometer (Nanodrop). (Spearman’s correlation rho = 0.85, P = 0), with the ex- To ensure IVT products were full-length without degra- ception of additional hypomethylation at the PEG3 DMR. dation, an aliquot of IVT RNA was visualized following (Figure 1; Supplementary Figure S1A; Supplementary Ta- gel electrophoresis on a 1.5% agarose gels prepared under ble S3). MOPS buffer/formaldehyde denaturing conditions. Em- To further characterise the stability of the imprinting bryos were collected from 5 to 7 week-old F1 (C57BL/6J defects in the two children and understand the potential × CBA/H) super-ovulated females crossed with F1 males. role of ZFP57p.E123* in tissue-specific mosaicism, we hy- Superovulation was induced by intraperitoneal injection of bridised buccal-derived DNA and a second leukocyte sam- pregnant mare serum gonadotropin (PMSG, Intervet, 5 ple from II.1 to the EPIC arrays. When comparing the two IU) and human chorionic gonadotropin (hCG, Intervet, blood-derived samples we observed that the hypomethyla- 7.5 IU) 46–48 h later. Early zygotes were collected at 18h tion signature is stable over time (Spearman’s correlation hCG injection and microinjected with 1–2pl of 200 ng/␮l rho = 0.89, P = 0), with the PLAGL1 DMR being the dCas9-Zfp57[KRAB]-T2A-GFP mRNA, 20 ng/␮l of cr- most severely affected consistent with this being the disease- RNA and 20 ng/␮l of tracrRNA. Embryos were cultured causing locus. The two sibling samples differed from the in K-modified simplex optimized medium microdrops un- unaffected mothers sample in their imprinted methylation der oil at 37◦ C, 5% CO2 for 78 h until the blastocyst stage. profile (Spearman’s correlation rho = 0.53, P = 0). Further- more, the hypomethylation also occurred in buccal samples, Oocyte and embryo RNA-seq analysis with the same imprinted DMRs affected to a similar extent (Spearman’s correlation rho = 0.86, P = 0) (Supplementary The single-cell (sc) RNA-seq libraries were generated us- Figure S1A). ing the Nextera XT kits on previously amplified cDNA Hypomethylation associated with MLID is often par- (https://doi.org/10.1038/nprot.2014.006). After sequencing, tial, with distinct loci subject to varying degrees of LOM. reads were trimmed using Trim Galore, with a minimal To understand this mosaic pattern and why some regions ‘phred’ score for base of 13 and a minimum allowed read are only partially affected, we generated a lymphoblastoid length of 35 bases (as recommended by https://doi.org/10. cell line following Epstein-Barr virus (EBV) transformation 1016/j.cell.2015.10.039). Remaining reads were mapped by from the second blood sample of II.1. Initial characteri- Hisat2 against the GRCh38 human genome. Reads were as- sation upon EBV transformation showed a stable methy- signed to transcripts present in the 94 Ensembl annotation lation pattern indicating that the immortalization process version of the GRCh38 human genome by using the Rsub- had not induced additional changes at imprinted gDMRs read R package (https://doi.org/10.1093/nar/gkt214). Cells (Supplementary Table S3). We subsequently utilized this with <0.75 million or >10 million assigned reads were fil- line to generate 28 expanded colonies (<500 cells) derived tered out, as well as cells with >7500 detected transcripts. from single cells, which we used to characterize imprinted Transcripts per million (‘TPM’) normalisation was also per- methylation using a modified SCRAM technique (28) (Fig- formed. Expression plots for pseudo-merged individual em- ure 2A, B). This revealed PLAGL1 LOM in 93% (26/28) of bryos were created using in-house R scripts. lines, whilst GRB10 hypomethylation was observed in 61% (17/28) of the lines and in 35% (10/28) and 25% (7/28) for KZNF ChIP-seq data analysis GNAS-XL and NAP1L5, respectively (Figure 2C). These re- sults are consistent with the detected levels of mosaicism Public ChIP data was obtained from GEO data series observed using the Infinium arrays. Interestingly, while the GSE78099 (12). Files with peaks processed score signals lines varied with respect to hypomethylation at GRB10, were retrieved and the data plotted using in-house R scripts. GNAS and NAP1L5 varied, we never observed hypomethy- lation of these loci independent of PLAGL1, and lines with RESULTS normal PLAGL1 also had normal methylation at the other DMRs. Together these results indicate that the methylation Characterization of methylation defects in TNDM-MLID signature associated with recessive mutations of ZFP57 in siblings with homozygous ZFP57p.E123* humans differ significantly from the equivalent zygotic mu- We previously reported a child with TNDM (II.1), who tations in mouse. we found was homozygous for a one base-pair deletion (p.E123*) in ZFP57, inheriting the mutations from both Determining the histone modification binding preference for heterozygous parents. Methylation profiling at diagnosis the ZFP57-KAP1 co-repressor complex revealed severe hypomethylation at PLAGL1 and mosaic LOM at GRB10, NAP1L5 and GNAS-XL DMRs in II.1, One possible explanation why only selected gDMRs are af- but not in the carrier mother (23) (Figure 1). Ten years fected when ZFP57 is mutated could be due to preferen- after the first child, a second sibling (II.2) was born with tial binding to genomic regions with specific patterns of hi- comparable TNDM and birthweight, but lacked several of stone modifications. The TGCCme GC consensus motif is Nucleic Acids Research, 2020, Vol. 48, No. 20 11399 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 1. Methylation profiling in a family with TNDM-MLID caused by ZFP57 mutation. (A) Family pedigree with transmission of the ZFP57p.E123* and examples of the sequence electropherograms showing the mutation in I.1, II.1 and control samples. (B) Analysis of 36 ubiquitously imprinted DMRs using Infinium HumanMethylationEPIC BeadChip arrays. Methylation differences are color coded according to severity, with blue and red representing hypo- and hypermethylation, respectively. present in the human genome >100 000 times, with ∼62% modomain containing member of the KAP1 co-repressor of these mapping to loci with >75% methylation, yet ChIP- complex, from transiently transfected cells and exposed the seq data reveals ∼3000 binding sites (12), of which only se- proteins separately to MODified™ histone peptide arrays. lected gDMRs are affected when the gene is mutated. To These experiments conclusively revealed that ZFP57 is un- determine whether ZFP57 recognises any post-translation able to directly bind to any histone modification, consistent histone modification, adding an additional layer of the with the absence of chromo-domains, whilst HP1␥ robustly complexity to the recognition of the binding motifs within binds to H3K9me2 and H3K9me3 (Supplementary Figure gDMRs, we purified HA-ZFP57 and HP1␥ -HA, a chro- S2). This suggests that in addition to direct sequence recog- 11400 Nucleic Acids Research, 2020, Vol. 48, No. 20 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 2. Characterising the level of mosaic methylation through analysis of single cell-derived clones. (A) Schematic overview of the modified SCRAM technique. (B) Examples of the resulting PCRs for representative colonies. (C) Methylation results for all colonies analysed with the colony ID indicated and the investigated imprinted DMR labelled on the left. nition, the ZFP57-KAP1 co-repressor complex associates To determine the effect of the ZFP57p.E123* on the recruit- with these repressive histone marks, which decorate most ment of the KAP1 co-repressor complex, we generated a gDMRs, via the recruitment by HP1␥ . truncated version of ZFP57 (FLAG-ZFP57znf ) which was transiently over-expressed with KAP1-HA. We observed a direct interaction between KAP1 and the ZFP57 mu- tant lacking all zinc fingers, consistent with the presence Human ZFP57 interacts with DNMT3 and KAP1 co- of the intact KRAB domain (Figure 3B). This suggests repressor complex that the LOM we observe in our family is due to a fail- Previous studies have shown that mouse ZFP57 physi- ure of ZFP57p.E123* to binding to DNA. Furthermore, us- cally interacts with DNMT1, DNMT3A/B and KAP1 in ing site-directed mutagenesis we generated a previous pub- mouse embryonic stem cells (14,15). Since mouse and hu- lished single-base substitution that produces the p.R228H man ZFP57 only share 37% amino-acid identify, we wanted amino acid change (20). This mutation maps to the third to address whether the human orthologue was able to in- zinc-finger and has been shown to reduce the ability to teract with the KAP1 co-repressor complex. To test this bind methylated DNA (18). This variant also interacts with physical interaction, we generated full-length ZFP57 epi- KAP1 indicating that the MLID in this patient is not due to tope tagged inducible over-expressing cell lines based on a failure in recruiting the KAP1 co-repressor complex and the T-REx 293 system to overcome the lack of endogenous is consistent with the observation that the KRAB domain expression in cell lines. In ZFP57-myc and KAP1-HA ex- is the primary mediator of KAP1 interactions (Figure 3B). pressing cells, antibodies against the HA epitope were used This data points towards a reduced recognition for methy- to pull-down KAP1 associated proteins and myc antibody lated consensus sequences as the potential mechanism be- was used to probe the immunoprecipitated material (Figure hind MLID. 3A). ZFP57-myc was detected when it was over-expressed with KAP1-HA confirming that these proteins interact ZFP57-KRAB recruitment is sufficient to protect non- as observed in mice. Similar co-IP experiments revealed imprinted germline-derived allelic methylation from epige- that HP1␥ -FLAG also interacts physically with ZFP57- netic reprogramming HA in our T-REx 293 cells. Furthermore, FLAG-tagged DNMT3A and 3B could also interact with ZFP57-HA, We next sought to determine if ZFP57 binding alone is although no co-immunoprecipitation was observed with sufficient to recruit the KAP1 co-repressor complex and DNMT1 indicating that ZFP57 maintains allelic methyla- protect non-imprinted oocyte-derived methylation during tion by recruiting the de novo methyltransferases (Figure pre-implantation development. We hypothesised that the 3A). tethering of ZFP57 to genomic locations that are nor- Nucleic Acids Research, 2020, Vol. 48, No. 20 11401 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 3. Characterizing in vivo protein-protein interactions for ZFP57 and member of the KAP1 co-repressor complex. (A) Western blotting of co- immunoprecipitated material from tetracycline induced ZFP57-HA and transiently transfected interacting partners including epitope tagged KAP1, HP1␥ , DNMT3A, DNMT3B and DNMT1. (B) Co-immunoprecipitation and western blotting for transiently transfected FLAG-ZFP57znf with KAP1-HA and HA-ZFP57p.R228H with KAP1-V5. mally methylated in oocytes but reprogrammed very early vidual blastocysts were subsequently used for methylation in pre-implantation development would be sufficient to analysis. We quantified the methylation status of the three recruit KAP1 co-repressor complex and maintain allelic targeted regions by utilizing a nested-multiplex bisulphite DNA methylation. To address this, we generated a dCas9- PCR approach coupled with locus-specific pyrosequencing. Zfp57 fusion construct that lacked the zinc-fingers, but To control for amplification bias that may have occurred retained the full KRAB domain in the -C terminus, en- because of the high number of PCR cycles, we included suring that any DNA targeting was solely due to ribonu- primers for the Airn imprinted gDMR into the nested- cleoprotein complex (Figure 4). We validated that our multiplex reaction. Following pyrosequencing we observed dCas9-Zfp57[KRAB]-T2A-GFP construct could indeed di- that two loci, Sgk1 and Serpinb6a, retained methylation in rectly interact with mouse KAP1-HA in transfected mouse a strand-specific manner upon ZFP57 targeting, suggest- NIH3T3 cells based on co-IP (Supplementary Figure S3). ing maintenance of oocyte-derived methylation in half of To identify specific loci to target, we first interrogated the injected embryos (Figure 4B, C). Two embryos from the mouse oocytes and embryo RRBS datasets (31) for re- Tsc22d1 experimental group also had elevated methylation gions that were highly methylated in oocytes, lack methy- levels compared to controls (Figure 4B). Interestingly, the lation in sperm and progressively demethylate during pre- injected embryos that had the highest levels of methylation implantation development. We discarded any genes that were the same, suggesting that the amount of ribonucleo- presented with a lethal phenotype when knocked-out (ac- protein complex was sufficient to ensure successful target- cording to the Mouse Genome Informatics database) as ing to their intended loci and recruitment of the KAP1 co- our manipulations may alter gene dosage in the early em- repressor complex to non-imprinted loci not normally tar- bryos. Finally, we selected three different loci, Serpinb6a, geted by ZFP57. Sgk1 and Tsc22d1, to which we designed two crRNAs on opposing DNA strands. The specificity of crRNA targeting Alternative ZFP57 promoter usage originates from a hap- was assessed using T7 endonuclease Surveyor assays follow- ASM interval ing transfection of Cas9 and tracrRNA into NIH3T3 cells. All crRNAs were able to target the corresponding regions Mutations of mouse Zfp57 cause the resulting phenotypes and, reassuringly, multiplexing did not affect the target ef- via maternal-zygotic effect and are directly associated with ficiency at individual loci (Supplementary Figure S3). Sub- high oocyte Zfp57 expression, whereas the phenotypes in sequently, in vitro transcribed dCas9-Zfp57[KRAB]-T2A- humans are only apparent following recessive transmission GFP mRNA, multiplexed crRNAs and tracrRNA were mi- and the inheritance of two mutated alleles. To determine croinjected into mouse zygotes. To ensure that any effect if this species-specific difference in inheritance pattern is we observed was due to the recruitment of the KRAB do- linked to the developmental timing of expression, we de- main of the dCas9-ZFP57, we also generated a control con- termined ZFP57 expression in our RNA-seq datasets from struct lacking the KRAB effector (dCas9-control) which human oocytes, cleavage embryos and blastocysts. This re- we also compared to a non-injected control group (Figure vealed that human ZFP57 is not expressed in oocytes and 4A). To control microinjections, we took advantage of the is only detectable at the blastocyst stage (Supplementary ATTOTM 550 fluorochrome present on the tracrRNA. Em- Table S4). Interrogation of the aligned sequences revealed bryos which did not display red fluorescence were discarded alternative promoter usage due to the identification of a from subsequent analyses, while positive embryos were de- unique first exon (NM 001366333.2). This ∼90 bp 5’UTR veloped in vitro for 78 h until the blastocyst stage. Indi- sequence is located 4 kb upstream of the first exon of the 11402 Nucleic Acids Research, 2020, Vol. 48, No. 20 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 4. Characterizing the ability of ZFP57 to protect non-imprinted oocyte-derived methylation from epigenetic reprogramming. (A) Schematic repre- sentation of the dCas9-Zfp57[KRAB]-T2A-GFP construct and the microinjection strategy for mouse zygotes (B) The results of nested-multiplex bisulphite PCR coupled with locus-specific pyrosequencing for the maternally methylated imprinted control Airn and the targeted loci Serpinb6a, Sgk1 and Tsc22d1. (C) Confirmation of strand-specific methylation by bisulphite PCR and sub-cloning for the Serpinb6a interval. Each circle represents a single CpG dinu- cleotide on a DNA strand, a methylated cytosine (•) or an unmethylated cytosine (). most prevalent isoform (BC157878) in an interval previ- gDMRs during the cleavage stages of development which ously reported to show haplotype-dependent allele-specific coincided with epigenetic reprogramming. To determine if methylation (hap-ASM) in T-cells (32) (Figure 5A). We other KZFPs could be involved in this process we ques- confirm methylation asymmetry between the two alleles in tioned whether any of those with available high-resolution a heterozygous hES cell line (Val11B), which resulted in KZFP ChIP-seq maps, could bind to ubiquitously im- monoallelic expression of ZFP57, but methylation is ab- printed DMR. Recently ChIP-seq for 223 KZPs (12), repre- sent in gametes and term placenta (Figure 5B, C). Quanti- senting ∼30% of all KZFPs in the human genome has been tative RT-PCR with exon-spanning primers revealed a sim- reported. The analysis exposed 49 KZFPs binding to dif- ilar expression profile for both isoforms, indicative of co- ferent imprinted DMRs, with 25 binding to multiple loci expression in most tissues, being most abundant in spine, (Figure 6A; Supplementary Figure S5 and S6). Focusing heart, hES cells and placenta (Supplementary Figure S4). on the 25 potential candidates that can bind to multiple Fascinatingly, the isoform originating from the alternative imprinted loci, we next filtered candidates based on their promoter results in skipping of exon 2 and a significantly expressed profile, which we limited to before EGA and the truncated KRAB domain (Figure 5A). expression of ZFP57. For this purpose, we re-interrogated our RNA-seq datasets. Only two KZFPs were highly abun- dant in oocytes and expressed during the first cleavage divi- Temporal expression profiling identifies ZNF202 and sions; ZNF202 and ZNF445 (Figure 6B). Quantitative RT- ZNF445 potential imprint factors PCR for these two factors revealed ubiquitously expression Our RNA-seq profiling revealed that human ZFP57 is only in most tissues (Figure 6C). Both genes possess SCAN mo- detectable following embryonic-genome activation (EGA) tifs in the N-terminal domain upstream of their KRAB do- suggesting that other factors may recruit the KAP1 co- mains, with ZNF202 having eight zinc fingers and ZNF445 repressor complex to maintain methylation at imprinted fourteen. Using transient overexpression, ZNF202-FLAG Nucleic Acids Research, 2020, Vol. 48, No. 20 11403 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 5. Alternative promoter usage and the ZFP57 hap-ASM. (A) Schematic overview of the ZFP57 gene, the location of two transcripts observed in blastocyst -derived RNA-seq datasets and the position of the hap-ASM. Grey exonic regions indicate the KRAB domain sequences. (B) Characterization of the allelic expression in Val11B hES cell line and methylation overlapping the alternative ZFP57 promoter heterozygous for SNP rs3129054 by bisulphite PCR followed by sub-cloning. Each circle represents a single CpG dinucleotide on a DNA strand, a methylated cytosine (•) or an unmethylated cytosine (). (C) Characterisation of monoallelic expression in the Val11B hES cell line. and ZNF445-HA both interact with KAP1 in co-IP ex- wide locations. When comparing these imprinted loci to periments confirming their ability to recruit the KAP1 co- those bound by ZFP57, we found that 77% (10/13) are in repressor complex (Figure 6D). common, suggesting these two proteins could work in uni- Additional analysis of the high-resolution KZFP ChIP- son to protect the imprinted regions during early develop- seq maps revealed that ZNF202 binds to four imprinted ment. DMRs (FAM50B, PLAGL1, KCNQ1OT1 and L3MBTL1), as well as >7000 addition loci throughout the human genome (12). This implies that ZNF202 has no partic- Screening MLID cohorts for ZNF202 and ZNF445 muta- ular selectivity for imprinted domains. On the contrary, tions ZNF445 binds to 13 different imprinted DMRs (DI- Since very few mutations in epigenetic factors have been di- RAS3, ZDBF2, MEST, PEG13, H19, KCNQ1OT1, MEG3- rectly associated with MLID, we performed a direct screen DLK1 IG-DMR, MEG3, NNAT, GNAS-NESP, GNAS- for pathogenic variants that could act via maternal-effect in AS1, GNAS-XL and SNU13) out of a total of 638 genome- ID patients with Beckwith-Wiedemann syndrome (BWS), 11404 Nucleic Acids Research, 2020, Vol. 48, No. 20 Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 Figure 6. Identification of candidate KZFPs binding to imprinted DMRs. (A) Graphical representation of interactions between KZFP and multiple imprinted DMRs. White squares indicate a lack of binding, whereas black squares represent a positive ChIP-seq peak located within the defined imprinted DMR sequence. (B) RNA-seq expression profile for ZNF202, ZNF445 and ZFP57 during human pre-implantation development. (C) Quantitative RT- PCR showing the distribution of expression for the same three KZFPs across different tissues. (D) Western blotting of co-immunoprecipitations between FLAG-ZNF202 and HA-ZNF445 with epitope tagged KAP1. Nucleic Acids Research, 2020, Vol. 48, No. 20 11405 Pseudohypoparathyroidism type 1B (PHP1B) and TNDM, ing pre-implantation reprogramming, which is largely de- along with their mothers. Amplification of ZNF202 and pendent upon the sequence-specific recruitment of the co- ZNF445 exons and direct Sanger sequencing of the result- repressor complex, which is consistent with our observa- ing amplicons identified one heterozygous ZNF445 vari- tions that ZFP57 interacts with the DNMT3A/B and not ant (rs143141433) in the mother of a child with PHP1B DNMT1. Interestingly, both the ZFP57p.E123* deletion and MLID that was not present in 200 controls. This rare mis- the non-synonymous p.R228H base substitution retain the Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 sense variant (1000 genomes 36/5008) results in a p.E957G ability to interact with KAP1, suggesting that if present in amino acid change that was predicted to be benign follow- the cell of patients, these mutated forms would bind KAP1 ing Polyphen analysis. but would not recruit the co-repressor complex to imprinted gDMRs. Scrutiny of ZFP57 expression in blastocysts iden- tified an alternative splice isoform that would result in the DISCUSSION translation of a protein lacking the KRAB domain, and Multi-locus imprinting disturbances have been observed for therefore would not have the capacity to recruit KAP1, de- most IDs with varying frequency, often mirroring the inci- spite binding to target sequences. Similar isoform struc- dence of epimutation for each individual disease. Despite in- ture has been noted for several additional KZFPs, includ- tense efforts, the underlying genetic causes have rarely been ing ZNF74, ZNF468 and ZNF268 (37–39), with competi- identified, with mutations in ZFP57 and members of the tive repressive action occurring between isoforms, with the SubCortical Maternal Complex (SCMC) being recurrently KRAB-containing version responsible for silencing tran- observed in different patient cohorts (20,33–35). Intrigu- scription. If this antagonism occurs for ZFP57 in the TE ingly, mutations in ZFP57 are associated with TNDM and and placenta, it may result in widespread LOM account- a highly predictable methylation signature (20–23), whereas ing for the polymorphic nature placenta-specific imprinted there is considerable variability in both clinical presentation gDMRs in humans. More work is required to follow-up on and imprinted loci affected in cases with underlying SCMC this hypothesis. involvement. Curiously ChIP-seq data generated in HEK 293 cells by In contrast to mouse where ZFP57 maintains allelic Imbeault and colleagues reveals that human ZFP57 binds methylation at the majority of gDMRs (14–16), humans are to several more imprinted gDMRs than are affected in the less dependent on this KZFP. Using extremely rare sam- TNDM MLID probands (12). One possible reason for such ples from a family harbouring a recessive ZFP57 mutation a discrepancy could be that starker LOM only occurs fol- we investigated the influence of this KZFP on imprinted lowing maternal-zygotic transmission, such as in mouse, gDMR methylation. We show that the truncating mutation since most cases in humans are zygotic mutants. However, leads to absolute hypomethylation of PLAGL1 DMR and this is unlikely, since a TNDM MLID case with a homozy- varying degrees of LOM at several additional imprinted loci gous ZFP57 mutation was born to a homozygous mother, which is reproducible, not only between tissues of an af- who herself presented with TNDM at birth. The child had fected individuals, but stable over time. Furthermore, the af- no evidence of a more severe phenotype nor was the de- fected sibling also had similar aberrant methylation profiles. gree of her LOM more severe than other individuals with The imprinted gDMRs associated with NAP1L5, GRB10 ZFP57 mutations (21). An alternative explanation is that and GNAS presented incomplete LOM in these subjects, there is compensation by other KZFPs that cooperate to which we were able to show was due to cell mosaicism rather maintain imprinted gDMRs and that the affected loci in than allelic partial demethylation. ZFP57-mutated patients would not be bound by the ad- Studies in mouse models, both in vivo and in mES cells, ditional KZFPs. Such cooperation between KZFPs is also have shown that ZFP57 binds to the methylated allele at supported by the fact that ZFP57 is not expressed dur- gDMRs and recruits the scaffold protein KAP1, the obliga- ing the critical time just after fertilization in humans, im- tory histone methyltransferase SETDB1 and DNA methyl- plying other KZFPs are necessary to initially recruit the transferases to deposit H3K9me3 and cytosine methyla- KAP1 co-repressor complex to imprinted gDMRs. Since tion (5mC) at target sequences (13,15). However, it is un- our oocyte and embryo RNA-seq datasets confirmed this known whether human ZFP57 recruits these and other notion, we bioinformatically analysed the publicly avail- components of the co-repressor complex. Using co-IPs, we able high-resolution KZFP ChIP-seq maps to identify addi- show that human ZFP57 does indeed interact with KAP1, tion candidates that bind to imprinted DMRs. Whilst many DNMT3A/B and HP1␥ . These observations endorse the KZFPs bind to these genomic intervals, very few are as- model centred around a sequence-specific KZFP recruiting sociated with multiple imprint loci. Furthermore, when we KAP1, which in turn encourages a self-perpetuating loop determined the correct temporal expression in our human in which SETDB1 deposits H3K9me3 that is recognised RNA-seq datasets, only two KZFPs were highly abundant by HP1. Furthermore, since ZFP57 has a strong prefer- in oocytes with continued expression during cleavage stages ence for methylated DNA the inclusion of DNMT3s within of development, ZNF202 and ZNF445 respectively. Since the co-repressor complex safeguards the underlying 5mC. ZNF202 only bound to four imprinted DMRs and promis- In general, methylation is heritably maintained in somatic cuously binds throughout the genome (12), we focused our cells by the Uhrf1/Dnmt1 complex that recognizes hemi- attentions on ZNF445. methylated DNA and methylates the newly synthesized un- Recently Takahashi and co-workers identified ZNF445 as methylated daughter strand, protecting against methylation a primary regulator of genomic imprinting in mouse (40). dilution during each round of DNA replication (36). This Although maternal-zygotic deletions of Zfp445 in mouse form of maintenance is different to that occurring dur- embryos did not affected imprinted DMRs or allelic ex- 11406 Nucleic Acids Research, 2020, Vol. 48, No. 20 pression, double homozygous zygotic mutations for Zfp445 regarding native protein extraction used with the MODi- and Zfp57 gave rise to more remarkable phenotypes associ- fied™ histone peptide arrays. ated with dramatic LOM than either Zfp445 or Zfp57 mu- tations alone. This led the authors to conclude that murine FUNDING imprinted DMRs depend on ZFP445 and ZFP57 to differ- ent degrees and that these KZFPs act in unison to main- Spanish Ministry of Economy and Competitiveness Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 tain allele-specific imprints in early mouse development. (MINECO) [BFU2014-53093-R, BFU2017-85571-R to Our human expression data suggests that rather than act- D.M.], co-funded with the European Union Regional ing together during the epigenetic reprogramming window, Development Fund (FEDER); Research in the GPdN these two KZFPs separately recruit the KAP1 co-repressor laboratory is funded partially funded by Instituto de complex, ZNF445 initially, followed by ZFP57, since they Salud Carlos III (Institute of Health Carlos III) of the are not co-expressed. We speculate that pathogenic muta- Spanish Ministry of Economy and Competitiveness, tions in ZNF445 would not be compatible with life, since co-financed by the European Regional Development a large number of imprinted gDMRs could present with Fund [PI16/00073]; Department of Health of the Basque LOM during early development. However, since individu- Government [GV2017/111040]; P.L. and J.T. received als with biallelic ZFP57 mutations survive (maybe with an funds from the National Agency of Research (ISCIII- ascertainment bias for those with the DMR signature re- FEDER) [PI15/01481]; Work in G.K.’s lab is supported ported), as do mice carrying targeted Znf445 deletions, mu- by grants from the UK Medical Research Council tations in ZNF445 may have no clinical consequence. To de- (MRC) [MR/K011332/1, MR/S000437/1]; Biotechnol- cipher whether mutations of ZNF445 are associated with ogy and Biological Sciences Research Council (BBSRC) developmental phenotypes it may be more appropriate to [BBS/E/B/000C0423]; A.M.S. is a recipient of an FPI PhD screen women with idiopathic infertility or recurrent mis- studentship from MINECO [BES-2015-072547]; EMBO carriage in the future. short-term fellowship to visit in M-ET-P laboratory. In conclusion, our studies have identified developmen- Funding for open access charge: Internal UEA funding. tal and epigenetic discrepancies between the role of ZFP57 Conflict of interest statement. None declared. in mouse and humans which are due to difference in the timing of expression during pre-implantation devel- REFERENCES opment. Screening for additional KZFPs expressed dur- 1. Monk,D. (2015) Germline-derived DNA methylation and early ing this crucial developmental window identified ZNF445, embryo epigenetic reprogramming: the selected survival of imprints. which works concurrently with ZFP57 to recruit the KAP1 Int. J. Biochem. Cell Biol., 67, 128–138. co-repressor complex in mouse, but unaccompanied imme- 2. Monk,D., Mackay,D., J.G.,ET., Maher,E.R. and Riccio,A. (2019) Genomic imprinting disorders: lessons on how genome, epigenome diately following fertilization in humans. Our analysis also and environment interact. Nat. Rev. Genet., 20, 235–248. identified other KZFPs with the capacity to bind individual 3. Ferguson-Smith,A. C. (2011) Genomic imprinting: the emergence of imprinted gDMRs, suggesting that they may be candidates an epigenetic paradigm. Nat Rev Genet., 12, 565–575. for isolated LOM or milder cases of MLID observed in IDs. 4. Hanna,C.W., Peñaherrera,M.S., Saadeh,H., Andrews,S., McFadden,D.E., Kelsey,G. and Robinson,W.P. (2016) Pervasive polymorphic imprinted methylation in the human placenta. Genome DATA AVAILABILTY Res., 26, 756–767. 5. Monk,D. (2015) Genomic imprinting in the human placenta. Am. J. The coordinates of human imprinted gDMRs used in Obstet. Gynecol., 213, S152–S162. this are available at www.humanimprints.net. The IIIu- 6. John,R.M. and Lefebvre,L. (2011) Developmental regulation of somatic imprints. Differentiation, 81, 270–280. mina Infinium HumanMethylation450 and HumanMethy- 7. Sanli,I. and Feil,R. (2015) Chromatin mechanisms in the lationEPIC BeadChip array data has been deposited in developmental control of imprinted gene expression. Int. J. Biochem. the GEO repository and assigned the accession number Cell Biol., 67, 139–147. GSE149568. The RNA-seq expression data corresponding 8. Inoue,A., Jiang,L., Lu,F., Suzuki,T. and Zhang,Y. (2017) Maternal to the oocyte and embryo datasets can be found at NCBI- H3K27me3 controls DNA methylation-independent imprinting. Nature, 547, 419–424. BioProject under the accession number PRJNA630371. 9. Hanna,C.W., Pérez-Palacios,R., Gahurova,L., Schubert,M., Krueger,F., Biggins,L., Andrews,S., Colomé-Tatché,M., Bourc’his,D., Dean,W. et al. (2019) Endogenous retroviral insertions SUPPLEMENTARY DATA drive non-canonical imprinting in extra-embryonic tissues. Genome Biol., 20, 225. Supplementary Data are available at NAR Online. 10. Chen,Z., Yin,Q., Inoue,A., Zhang,C. and Zhang,Y. (2019) Allelic H3K27me3 to allelic DNA methylation switch maintains noncanonical imprinting in extraembryonic cells. Sci. Adv., 5, ACKNOWLEDGEMENTS eaay7246. 11. Urrutia,R. (2003) KRAB-containing zinc-finger repressor proteins. We would like to thank the patients and family members Genome Biol., 4, 231. that participated in this study and the clinical staff who 12. Imbeault,M., Helleboid,P.Y. and Trono,D. (2017) KRAB zinc-finger contributed to sample collection. We are grateful to Di- proteins contribute to the evolution of gene regulatory networks. dier Trono for the mouse KAP1-HA plasmid, and to mem- Nature, 543, 550–554. bers of the PEBC-IDIBELL, especially Charles Arribas 13. Schultz,D.C., Ayyanathan,K., Negorev,D., Maul,G.G. and Rauscher,F.J. 3rd. (2002) SETDB1: a novel KAP-1-associated histone for performing the methylation array hybridizations, Fer- H3, lysine 9-specific methyltransferase that contributes to nando Setien Baranda for help growing colonies from single HP1-mediated silencing of euchromatic genes by KRAB zinc-finger lymphoblastoid cells and Laia Bosch-Presegué for advice proteins. Genes Dev., 16, 919–932. Nucleic Acids Research, 2020, Vol. 48, No. 20 11407 14. Li,X., Ito,M., Zhou,F., Youngson,N., Zuo,X., Leder,P. and suggests a germline methylation-independent mechanism of Ferguson-Smith,A.C. (2008) A maternal-zygotic effect gene, Zfp57, establishment. Genome Res., 24, 554–569. maintains both maternal and paternal imprints. Dev. Cell, 15, 27. Sanchez-Delgado,M., Court,F., Vidal,E., Medrano,J., 547–557. Monteagudo-Sánchez,A., Martin-Trujillo,A., Tayama,C., 15. Quenneville,S., Verde,G., Corsinotti,A., Kapopoulou,A., Iglesias-Platas,I., Kondova,I., Bontrop,R. et al. (2016) Human Jakobsson,J., Offner,S., Baglivo,I., Pedone,P.V., Grimaldi,G., Oocyte-Derived methylation differences persist in the placenta Riccio,A. et al. (2011) In embryonic stem cells, ZFP57/KAP1 revealing widespread transient imprinting. PLos Genet., 12, e1006427. Downloaded from https://academic.oup.com/nar/article/48/20/11394/5923428 by FUNDACIO BIBLIOTECA JOSEP LAPORTE user on 11 February 2021 recognize a methylated hexanucleotide to affect chromatin and DNA 28. Cheow,L.F., Quake,S.R., Burkholder,W.F. and Messerschmidt,D.M. methylation of imprinting control regions. Mol. Cell, 44, 361–372. (2015) Multiplexed locus-specific analysis of DNA methylation in 16. Takahashi,N., Gray,D., Strogantsev,R., Noon,A., Delahaye,C., single cells. Nat. Protoc., 10, 619–631. Skarnes,W.C., Tate,P.H. and Ferguson-Smith,A.C. (2015) ZFP57 and 29. López-Abad,M., Iglesias-Platas,I. and Monk,D. (2016) Epigenetic the targeted maintenance of postfertilization genomic imprints. Cold characterization of CDKN1C in placenta samples from Spring Harb. Symp. Quant. Biol., 80, 177–187. non-syndromic intrauterine growth restriction. Front. Genet., 7, 62. 17. Shi,H., Strogantsev,R., Takahashi,N., Kazachenka,A., 30. Kabadi,A.M., Ousterout,D.G., Hilton,I.B. and Gersbach,C.A. (2014) Lorincz,M.C., Hemberger,M. and Ferguson-Smith,A.C. (2019) Multiplex CRISPR/Cas9-based genome engineering from a single ZFP57 regulation of transposable elements and gene expression lentiviral vector. Nucleic. Acids. Res., 42, e147. within and beyond imprinted domains. Epigenetics Chromatin, 12, 49. 31. Smith,Z.D., Chan,M.M., Mikkelsen,T.S., Gu,H., Gnirke,A., 18. Baglivo,I., Esposito,S., De Cesare,L., Sparago,A., Anvar,Z., Riso,V., Regev,A. and Meissner,A. (2012) A unique regulatory phase of DNA Cammisa,M., Fattorusso,R., Grimaldi,G., Riccio,A. et al. (2013) methylation in the early mammalian embryo. Nature, 484, 339–344. Genetic and epigenetic mutations affect the DNA binding capability 32. Do,C., Lang,C.F., Lin,J., Darbary,H., Krupska,I., Gaba,A., of human ZFP57 in transient neonatal diabetes type 1. FEBS Lett., Petukhova,L., Vonsattel,J.P., Gallagher,M.P., Goland,R.S. et al. 587, 1474–1481. (2016) Mechanisms and disease Associations of haplotype-dependent 19. Strogantsev,R., Krueger,F., Yamazawa,K., Shi,H., Gould,P., allele-specific DNA methylation. Am. J. Hum. Genet., 98, 934–955. Goldman-Roberts,M., McEwen,K., Sun,B., Pedersen,R. and 33. Docherty,L.E., Rezwan,F.I., Poole,R.L., Turner,C.L., Kivuva,E., Ferguson-Smith,A.C. (2015) Allele-specific binding of ZFP57 in the Maher,E.R., Smithson,S.F., Hamilton-Shield,J.P., Patalan,M., epigenetic regulation of imprinted and non-imprinted monoallelic Gizewska,M. et al. (2015) Mutations in NLRP5 are associated with expression. Genome Biol., 16, 112. reproductive wastage and multilocus imprinting disorders in humans. 20. Mackay,D.J., Callaway,J.L., Marks,S.M., White,H.E., Acerini,C.L., Nat. Commun., 6, 8086. Boonen,S.E., Dayanikli,P., Firth,H.V., Goodship,J.A., Haemers,A.P. 34. Begemann,M., Rezwan,F.I., Beygo,J., Docherty,L.E., Kolarova,J., et al. (2008) Hypomethylation of multiple imprinted loci in Schroeder,C., Buiting,K., Chokkalingam,K., Degenhardt,F., individuals with transient neonatal diabetes is associated with Wakeling,E.L. et al. (2018) Maternal variants in NLRP and other mutations in ZFP57. Nat. Genet., 40, 949–951. maternal effect proteins are associated with multilocus imprinting 21. Boonen,S.E., Mackay,D.J., Hahnemann,J.M., Docherty,L., disturbance in offspring. J. Med. Genet., 55, 497–504. Grønskov,K., Lehmann,A., Larsen,L.G., Haemers,A.P., 35. Sparago,A., Verma,A., Patricelli,M.G., Pignata,L., Russo,S., Kockaerts,Y., Doomsm,L. et al. (2013) Transient neonatal diabetes, Calzari,L., De Francesco,N., Del Prete,R., Palumbo,O., Carella,M. ZFP57, and hypomethylation of multiple imprinted loci: a detailed et al. (2019) The phenotypic variations of multi-locus imprinting follow-up. Diabetes Care., 36, 505–512. disturbances associated with maternal-effect variants of NLRP5 22. Touati,A., Errea-Dorronsoro,J., Nouri,S., Halleb,Y., Pereda,A., range from overt imprinting disorder to apparently healthy Mahdhaoui,N., Ghith,A., Saad,A., Perez de Nanclares,G. and phenotype. Clin Epigenetics, 11, 190. H’mida Ben Brahim,D. (2019) Transient neonatal diabetes mellitus 36. Sharif,J., Muto,M., Takebayashi,S., Suetake,I., Iwamatsu,A., and hypomethylation at additional imprinted loci: novel ZFP57 Endo,T., Shinga,J., Mizutani-Koseki,T.T., Okamura,K. et al. (2007) mutation and review on the literature. Acta Diabetol., 56, 301–307. The SRA protein Np95 mediate epigenetic inheritance by recruiting 23. Court,F., Martin-Trujillo,A., Romanelli,V., Garin,I., Dnm1 to methylated DNA. Nature, 450, 908–912. Iglesias-Platas,I., Salafsky,I., Guitart,M., Perez de Nanclares,G., 37. Côté,F., Boisvert,F.M., Grondin,B., Bazinet,M., Goodyer,C.G., Lapunzina,P. and Monk,D. (2013) Genome-wide allelic methylation Bazett-Jones,D.P. and Aubry,M. (2001) Alternative promoter usage analysis reveals disease-specific susceptibility to multiple methylation and splicing of ZNF74 multifinger gene produce protein isoforms defects in imprinting syndromes. Hum. Mutat., 34, 595–602. with a different repressor activity and nuclear partitioning. DNA Cell 24. Sánchez,A.M., Delgado,M.S., Ciurana,S.G., Medrano,J., Biol., 20, 159–173. Poo-Llanillo,M.E., Ishida,M., Moore,G.E., Platas,I.I., Simón,C. and 38. Sun,L., Gu,S., Li,N., Zheng,D., Sun,Y., Li,D., Ji,C., Ying,K., Xie,Y. Monk,D. (2018) Epigenetic symmetry of DLGAP2: pre-implantation and Mao,Y. (2005) A novel zinc finger gene ZNF468 with two maternal methylation switches to a random monoallelic profile in co-expressional splice variants, ZNF468.1 and ZNF468.2. Biochem. somatic tissues. OBM Genetics, 2, 026. Genet., 43, 271–286. 25. Hernandez Mora,J.R., Tayama,C., Sánchez-Delgado,M., 39. Shao,H., Zhu,C., Zhao,Z., Guo,M., Qiu,H., Liu,H., Wang,D., Monteagudo-Sánchez,A., Hata,K., Ogata,T., Medrano,J., Xue,L., Gao,L., Sun,C. et al. (2006) KRAB-containing zinc finger Poo-Llanillo,M.E., Simón,C., Moran,S. et al. (2018) Characterization gene ZNF268 encodes multiple alternatively spliced isoforms that of parent-of-origin methylation using the Illumina Infinium contain transcription regulatory domains. Int. J. Mol. Med., 18, MethylationEPIC array platform. Epigenomics, 10, 941–954. 457–463. 26. Court,F., Tayama,C., Romanelli,V., Martin-Trujillo,A., 40. Takahashi,N., Coluccio,A., Thorball,C.W., Planet,E., Shi,H., Iglesias-Platas,I., Okamura,K., Sugahara,N., Simón,C., Moore,H., Offner,S., Turelli,P., Imbeault,M., Ferguson-Smith,A.C. and Harness,J.V. et al. (2014) Genome-wide parent-of-origin DNA Trono,D. (2019) ZNF445 is a primary regulator of genomic methylation analysis reveals the intricacies of human imprinting and imprinting. Genes Dev., 33, 49–54.