Abstract
Prader-Willi syndrome (PWS) is characterized by neonatal hypotonia, developmental delay and hyperphagia/obesity and is caused by the absence of paternal contribution to chromosome 15q11-q13. Using induced pluripotent stem cell (iPSC) models of PWS, we previously discovered an epigenetic complex that is comprised of the zinc-finger protein ZNF274 and the SET domain bifurcated 1 (SETDB1) histone H3 lysine 9 (H3K9) methyltransferase and that silences the maternal alleles at the PWS locus. Here, we have knocked out ZNF274 and rescued the expression of silent maternal alleles in neurons derived from PWS iPSC lines, without affecting DNA methylation at the PWS-Imprinting Center (PWS-IC). This suggests that the ZNF274 complex is a separate imprinting mark that represses maternal PWS gene expression in neurons and is a potential target for future therapeutic applications to rescue the PWS phenotype.
Introduction
PWS (MIM 176270), a neurobehavioral disorder of genomic imprinting, is characterized by neonatal hypotonia and failure to thrive during infancy and, subsequently, by obesity, hyperphagia, cognitive disability and behavioral abnormalities. PWS is caused by the absence of a normal paternal contribution to chromosome 15q11-q13, most commonly due to a large deletion (LD) of the ∼5000kb imprinted region or to maternal uniparental disomy (mUPD) of chromosome 15 (1,2). Maternally inherited deletion of the 15q11-q13 imprinted region or paternal UPD causes Angelman syndrome (AS), a disorder characterized by seizures, movement difficulty, cognitive disability and failure to speak (3–5).
The germline imprint of 15q11-q13 is a differentially methylated CG-rich segment, termed the PWS-IC (6–9), located in the first exon of the gene encoding small nuclear ribonucleoprotein polypeptide N (SNRPN). SNRPN is a bicistronic transcript that also encodes SNURF (referred to henceforth as SNRPN in this report). In the brain, a long non-coding RNA (lncRNA) initiates at upstream (U) exons of SNRPN (Fig. 1A), extends >600kb distally to overlap UBE3A, and silences the paternal UBE3A allele via an antisense-mediated mechanism (10–18). The PWS lncRNA serves as the host gene (HG) to several box C/D class small nucleolar RNAs including the SNORD116 and SNORD115 clusters (10,18). There are 30 copies of SNORD116 in the cluster which has been subdivided into three groups based on DNA sequence similarity (19), Group 1 (SNOG1, SNORD116 1–9), Group 2, [SNOG2, SNORD116 10–24 and Group 3 (SNOG3, SNORD116 25–30)]. Moreover, it has been recently shown that loss of SNORD116 in both human iPSC deletion and mouse models of PWS have a deficiency of prohormone convertase PC1 that may potentially be associated with the neuroendocrine dysfunction in PWS (20,21). These results suggest an association between SNORD116 deletion and PWS.
CRISPR/Cas9-mediated Knock-Out of ZNF274 reduces H3K9me3 and activates SNORD116 expression in PWS iPSCs. (A) Simplified genetic and allelic expression map of 15q11.2-q13. Active and inactive (repressed) transcripts are denoted by open and closed boxes, respectively. Arrows indicate the direction of transcription. A solid black line represents paternal SNURF/SNRPN transcripts expressed in most cell types, whereas a dashed black line indicates neuron-specific transcripts, including upstream exons of SNRPN and UBE3A-ATS. UBE3A is maternally expressed in neurons whereas other genes are only paternally expressed in all cell types. The PWS-IC is denoted by the black (methylated)/white (un-methylated) circle. Orange dashes under the SNORD116 cluster represent the six ZNF274 binding sites within the SNORD116s classified as Group 1(19) (SNOG1-BS1 to SNOG1-BS6). (B) ChIP assay for ZNF274 in iPSCs. Here and in subsequent figures, PWS patient lines are shown in black, their corresponding ZNF274 KO lines in green, control (CTRL) cell lines in blue and AS, here used as a negative control, in white. Here and in subsequent figures, quantification of ChIP was performed and calculated as percent input for each sample. Binding at ZNF180, a previously reported ZNF274 binding site associated with high levels of H3K9me3 signal, was used as a positive control and, for each line, all other binding sites were normalized to this one. The PWS parental line was set as 1 for each panel and relative normalization to this positive sample was done for each cell line. A minimum of two biological replicates per cell line were performed. Significance was calculated using two-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the three UPD KOs to PWS UPD. Here and in subsequent figures, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001. (C) ChIP assay for the repressive histone modification H3K9me3 in iPSCs. The same color code as in Figure 1B is used for AS, CTRLs, PWS and ZNF274 KO cell lines.
Using our PWS and AS iPSC models, we have previously reported that ZNF274 binds to six sites in the maternal copy of the SNORD116 cluster and, associated in a complex with the histone methyltransferase SETDB1, mediates the deposition of the repressive H3K9me3 chromatin mark on the maternal allele (22–24). Given its putative function in tethering the epigenetic silencing complex to the maternal allele, we have targeted ZNF274 using CRISPR/Cas9 technology and followed the impact of the knockout on maternal allele expression in PWS-specific iPSCs through the process of neuronal differentiation.
Results
Generation and characterization of ZNF274 KO lines
In addition to our previously characterized PWS LD iPSCs (25), we generated iPSC lines from a PWS UPD patient (Supplementary Material, Fig. S1B and Table S1D). We performed CRISPR/Cas9-mediated knockout of ZNF274 in PWS-specific iPSCs in order to determine the impact of ZNF274 depletion on H3K9me3 accumulation at the SNORD116 locus. The genetic alterations induced by ZNF274 knockout (ZNF274 KO) are summarized in Supplementary Material, Table S1C for the PWS LD ZNF274 KO lines (LD KO1 and LD KO2) and UPD ZNF274 KO lines (UPD KO1, UPD KO2 and UPD KO3). Karyotypic analysis of the engineered iPSC lines showed no detectable abnormalities (Supplementary Material, Table S1D). We performed routine testing for pluripotency (Supplementary Material, Fig. S1B), and we did not observe any sequence changes within the top potential off-target loci (Supplementary Material, Table S1E).
CRISPR/Cas9-mediated knock out of ZNF274 depletes H3K9me3 at the SNORD116 locus in PWS-specific iPSCs
We performed Chromatin ImmunoPrecipitation (ChIP) on the PWS LD and UPD iPSCs and their derivative ZNF274 KO clones as well as from iPSCs from control individuals (CTRL1 and CTRL2) (22,25–27) and an AS patient with a large deletion of the maternal chromosome 15q11-q13(25) and observed a complete absence of ZNF274 binding to all PWS LD and UPD ZNF274 KO clones at all the 12 known ZNF274 binding sites tested, demonstrating an efficient ZNF274 KO (Fig. 1B and Supplementary Material, Fig. S2A). This absence of ZNF274 binding is associated with a marked reduction of H3K9me3 at the six ZNF274 binding sites (BSs) (SNOG1-BS1 to SNOG1-BS6) in all PWS LD and UPD ZNF274 KO clones, demonstrating the efficient disruption of ZNF274 function at the SNORD116 locus in our KO iPSC lines. Although we observed a complete absence of the ZNF274 protein at all the known ZNF274 BSs we tested, the level of H3K9me3 was not reduced at the ZNF274 BS in the ZNF180.
3’UTR that we used as our reference. The likely explanation is that ZNF274 BSs are most often shared with a second zinc finger protein (ZNF75D) (28). In fact, 89.1% of ZNF274 binding sites in KRAB domain-containing zinc finger protein genes are shared with ZNF75D (28). For example, the 3’UTR of ZNF180 and ZNF554 are two target regions bound by both ZNF274 and ZNF75D and we observe little or no effect on H3K9me3 levels at those sites (Supplementary Material, Fig. S2B). The 3’-ends of ZNF781 and ZNF90, on the other hand, are two of the rare sites that are bound only by ZNF274 and, consistent with this observation, we detect a more marked reduction of H3K9me3 for those two BSs (Supplementary Material, Fig. S2B).
The levels of H3K9me3 are also reduced in the PWS LD and UPD ZNF274 KO clones at SNOG2 and SNOG3, which are SNORD116 class group 2 and 3 loci (19) located downstream of SNOG1-BS1 to SNOG1-BS6 (Supplementary Material, Fig. S2C). The spread of H3K9me3 deposition at the maternal-specific SNOG1 ZNF274 BSs is consistent with our previous observation (22). We observed a concomitant activation of expression of the SNORD116 host gene Group1 (116HGG1) transcript in our ZNF274 KO PWS LD and UPD iPSCs (Supplementary Material, Fig. S2D). These results suggest that ZNF274 KO in PWS iPSCs reduces H3K9me3, leading to chromatin de-condensation and partial transcriptional activation of 116HGG1 within the PWS locus.
Impact of ZNF274 KO on SNRPN activation in PWS-specific iPSCs
We also investigated the mechanism by which ZNF274 KO activates 116HGG1 expression by examining the regulatory elements of the PWS lncRNA. For this, we analysed SNRPN transcripts driven by the major promoter in exon 1 and by its alternative upstream exon promoters, U1B and U1A, that drive PWS lncRNA predominantly in brain (11,12,14). SNRPN transcripts driven by U1B and U1A skip exon 1 and splice into exon 2 (8,29–31). We assayed for RNAs that splice from exon U4 to exon 2 since the U4 internal exon is included in most SNRPN U1B and U1A transcripts (11,12,32). We detected ZNF274 KO-mediated activation of U4/exon 2 (Fig. 2A) but not of exon 1/exon 2 SNRPN transcripts (Fig. 2B), a finding that suggests that the ZNF274 complex represses maternal PWS transcripts through its action on the U1B and U1A promoters rather than the major SNRPN exon 1 promoter. Consistent with this suggestion, we detect only partial activation of the SNRPN exon 3–4 coding transcript (Fig. 2C) in the absence of SNRPN major exon 1 promoter usage (Fig. 2B). To further understand the impact of ZNF274 KO on the regulation of SNRPN and PWS, we examined DNA methylation in the PWS-IC, which is contained within SNRPN exon 1 (Fig. 2D). In PWS LD iPSCs and their ZNF274 KO derivatives, we observe almost identical levels of CpG methylation at the maternal PWS-IC (Fig. 2D). These findings are not only consistent with our observation that the maternal SNRPN exon 1 promoter is not activated by ZNF274 KO but suggest, importantly, that ZNF274 is an epigenetic regulator of chromosome 15q11-q13 imprinting that acts independently of the PWS-IC.
Effect of ZNF274 KO on SNRPN expression and PWS-IC methylation in iPSCs. (A) Gene expression of the SNRPN U exons (U4/ex2), (B)SNRPN major promoter (ex1/2) and (C)SNRPN mRNA (ex3/4) in iPSCs. The same color code as in Figure 1B is used. Here and in subsequent figures, gene expression was assessed using the comparative CT method, GAPDH was used as an endogenous control. Data were normalized to CTRL1 for each panel and plotted as the mean with Standard Deviation (SD). A minimum of 3 biological replicates per cell line were performed. Significance was calculated using one-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the three UPD KOs to PWS UPD. (D) DNA methylation level at the PWS-IC in iPSCs was evaluated using a quantitative restriction endonuclease assay (EpiMark 5hmC and 5mC Analysis Kit) that measures the relative levels of 5-methylcytosine (5mC), 5-hydroxymethylcytosine (5hmC), and unmodified cytosine (C). As expected, CTRL iPSCs show approximately equal levels of 5mC and C at the PWS-IC whereas PWS LD and AS iPSCs display, respectively, almost complete (5mC) and almost no (C) methylation. Although a slight shift from 5mC to 5hmC is apparent, there is almost complete methylation (5mC) of the maternal PWS-IC in the ZNF274 KO PWS iPSCs. A minimum of 2 biological replicates per cell line were performed. Significance was calculated using two-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD.
ZNF274 KO restores maternal gene expression in PWS neurons
Given that the SNRPN exon U1B and U1A are active mainly in the brain (11,12,14), we derived neural progenitor cells (NPCs) and neurons from our iPSC lines to further understand the mechanism of the rescue of maternal PWS transcripts by ZNF274 KO (Fig. 3). Although ZNF274 KO increases the expression of maternal 116HGG1 by ≥100X in PWS iPSCs (Supplementary Material, Fig. S2D), the levels attained are much lower than those in CTRL iPSC lines for 116HGG1 (Fig. 3A). However, a more robust activation is observed upon neural differentiation with 116HGG1 expression almost reaching control levels in ZNF274 KO NPCs and attaining or surpassing these in rescued neurons (Fig. 3B and C). ZNF274 KO in PWS LD and UPD lines results in a marked increase of 116HGG1 expression relative to CTRL lines after 4 weeks of differentiation (Fig. 3B) and restores normal levels of expression in neurons after 10 weeks of differentiation (Fig. 3C).
ZNF274 KO-mediated activation of maternal 116HGG1 transcripts during in vitro neurogenesis. Gene expression of the SNORD116 Host Gene Group I (116HGG1) in each cell line through the differentiation process: in (A) iPSCs (n = 3 minimum), (B) 4-week-old neural precursor cells (NPCs) (n = 1 minimum) and (C) mature 10-week-old neurons (n = 2 minimum). The same color code as in Figure 1B is used. Data were normalized to CTRL1 or CTRL2 for each panel and plotted as the mean with Standard Deviation (SD). Significance was calculated using one-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the three UPD KOs to PWS UPD.
We then examined the expression of transcripts across chromosome 15q11-q13 in iPSC-derived neurons from PWS LD and UPD, their ZNF274 KO derivatives, CTRLs and the AS LD. Like 116HGG1, other transcripts (SNORD116–1 and IPW) within the PWS locus, are expressed in ZNF274 KO neurons at the same level as those in CTRL neurons (Fig. 4 and Supplementary Material, Table S2). Similar levels of ZNF274 KO-mediated reactivation of maternal neuronal transcripts are also detected downstream of the PWS locus (SNORD115–1, its 115HG host gene and the antisense overlapping UBE3A, UBE3A-ATS) (Fig. 4 and Supplementary Material, Table S2). While ZNF274 KO activates neuronal UBE3A-ATS to normal levels, a concomitant decrease in UBE3A expression is not observed at least relative to normal control UBE3A mRNA levels (Fig. 4 and Supplementary Material, Table S2).
ZNF274 KO activates transcription in PWS neurons across chromosome 15q11.2-q13. Gene expression of 17 transcripts across the 15q11.2-q13 region in mature 10-week-old neurons. qRT-PCR data was combined from two normal cell lines (CTRLs) and ZNF274 KO from each parental line: (A) LD KOs and (B) UPD KOs. The same color code as in Figure 1B is used. Significance was calculated using two-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the combined LD KOs to PWS LD and the combined UPD KOs to PWS UPD.
The SNRPN U4/exon 2 transcripts are completely rescued by ZNF274 KO in neurons while SNRPN transcripts utilizing exon 1 remain silent and exon 3/4 transcripts are partially activated (Figs 4, 5A–C and Supplementary Material, Table S2). These results are consistent with our hypothesis that the ZNF274 complex regulates PWS transcripts via the SNRPN upstream promoters. In support of this, higher levels of SNRPN U4/exon 2 expression are attained in neurons and NPCs upon ZNF274 KO than in iPSCs (Fig. 6 and Supplementary Material, Table S2) in accord with the reports that the U1B and U1A promoters are highly active in the brain (11,12,14). While ZNF274 KO in PWS LD and UPD neurons activates robust expression of most maternal PWS transcripts (Figs 4, 5A and C, 6 and Supplementary Material, Table S2), there is no change in 5mC levels at the PWS-IC (Fig. 5D), a finding consistent with our observation that ZNF274 KO does not activate SNRPN exon 1 transcription (Figs 4, 5B and Supplementary Material, Table S2) and with our contention that ZNF274 binding to the maternal PWS locus mediates silencing by a mechanism that is independent of the PWS-IC. Our hypothesis is consistent with reports that deletion of the PWS-IC does not result in the loss of imprinted expression in brain (33,34).
ZNF274 KO activates SNRPN upstream promoters in PWS iPSC-derived neurons without decreasing DNA methylation of the PWS-IC. (A) Gene expression of the SNRPN U exons (U4/ex2), (B) SNRPN major promoter (ex1/2) and (C) SNRPN transcript body (ex3/4) in neurons. The same color code as in Figure 1B is used. Data were normalized to CTRL1 or CTRL2 for each panel and plotted as the mean with Standard Deviation (SD). A minimum of 2 biological replicates per cell line were performed. Significance was calculated using one-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the three UPD KOs to PWS UPD. (D) DNA methylation level at the PWS-IC in mature 10-week-old neurons evaluated as in Figure 2B. As expected, CTRL neurons show approximately equal levels of 5mC and C at the PWS-IC whereas PWS (LD and UPD) and AS neurons display, respectively, almost complete (5mC) and almost no (C) methylation. There is almost complete methylation (5mC) of the maternal PWS-IC in the ZNF274 KO PWS neurons. A minimum of 2 biological replicates per cell line were performed. Significance was calculated using two-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the UPD KO to PWS UPD.
ZNF274 KO activates SNRPN upstream promoters during neuronal differentiation of PWS iPSCs. Gene expression of the SNRPN U exons (U4/ex2) in iPSCs, NPCs and neurons. The same color code as in Figure 1B is used. Data were normalized to CTRL1 or CTRL2 neurons for each panel and plotted as the mean with Standard Deviation (SD). A minimum of 3, 1, and 2 biological replicates per cell line for iPSCs, NPCs, and neurons, respectively, were performed. Significance was calculated using one-way analysis of variance (ANOVA) test with a Dunnett post-test to compare the two LD KOs to PWS LD and the three UPD KOs to PWS UPD.
Further upstream in the imprinted PWS region, we detect expression of MAGEL2 and MKRN3 in PWS LD and UPD neurons and an up-regulation of both upon ZNF274 KO (Fig. 4 and Supplementary Material, Table S2). NDN is not detected in PWS LD and UPD neurons and is not activated by ZNF274 KO (Fig. 4 and Supplementary Material, Table S2). The latter result would suggest that, like SNRPN exon 1 but not other PWS transcripts, NDN imprinting is regulated by the PWS-IC, consistent with a mouse model in which deletion of the maternal PWS-IC activated Ndn expression (35). CYFIP1 and CHRNA7, genes outside the 15q11-q13 imprinted region, are expressed in neurons derived from all iPSC lines. The mRNA levels of both genes are the same in PWS LD and UPD, and are not affected in their ZNF274 KO derivatives (Fig. 4 and Supplementary Material, Table S2). Our results suggest a critical role for ZNF274-mediated repression of most neuronal transcripts within but not outside of the imprinted chromosome 15q11-q13 region.
Discussion
The findings summarized here indicate that maternally inherited silent PWS transcripts are activated by CRISPR-mediated knockout of ZNF274. Loss of ZNF274 results in a reduction of H3K9me3 within the PWS locus (Fig. 1B and C and Supplementary Material, Fig. S2) and activates expression of maternal transcripts in PWS iPSCs, NPCs and neurons (Figs 1–6 and Supplementary Material, Table S2). Expression of maternal transcripts induced by ZNF274 KO in PWS neurons attains normal levels and robust activation is observed not only within the PWS locus but also throughout the chromosome 15q11-q13 imprinted region (Figs 3–6 and Supplementary Material, Table S2). Two PWS maternal mRNAs that are not rescued by ZNF274 KO are the SNRPN transcript driven by the exon 1 promoter and NDN (Figs 2B, 4, 5B and Supplementary Material, Table S2). The most likely explanation in the case of SNRPN is that ZNF274 KO does not alter CpG methylation of the maternal PWS-IC (Figs 2D and 5D) and, hence, does not activate the major SNRPN exon 1 promoter. How NDN expression is not rescued by ZNF274 KO is less clear although it has been reported in mouse models that the PWS-IC directly regulates Ndn (35). The expression of both MAGEL2 and MKRN3 are up regulated in PWS LD and UPD neurons by ZNF274 KO (Fig. 4 and Supplementary Material, Table S2) suggesting that ZNF274 binding to the SNORD116 cluster contributes to silencing of maternal alleles by a mechanism that remains to be clarified (Fig. 7).
Model of ZNF274-mediated silencing at the PWS locus. Open and closed boxes denote expressed and silenced 15q11.2-q13 genes, respectively, for the CTRL and PWS ZNF274 KO lines in iPSCs and neurons. The closed (methylated-5mC)/open (un-methylated-C) circles denote the PWS-IC. Arrows indicate the transcription start sites of SNRPN (we did not add arrows to other genes for clarity in the figure); thickness is relative to the degree of expression. Genes names are denoted sequentially between the CTRL and PWS ZNF274 KO lines. Red lollipops represent H3K9me3 signal. ZNF274 is denoted by the magenta ellipse. Colored segment over SNORD116 represents a putative regulatory element acting in cis on the SNRPN upstream exons. Yellow circles represent potential brain specific transcription factors activating the SNRPN upstream exons.
We did not detect a decrease in the levels of UBE3A despite robust activation of UBE3A-ATS (Fig. 4 and Supplementary Material, Table S2). One explanation for this is that UBE3A-ATS-mediated silencing of UBE3A may not be detectable due to the relative immaturity of the neurons differentiated from the iPSCs (25). Another possibility is that the level of expression of the maternal UBE3A mRNA is intrinsically higher than that of the paternal allele and thus more resistant to antisense-mediated silencing. In this regard, UBE3A expression in both PWS LD and UPD iPSC-derived neurons is increased relative to CTRLs (Fig. 4 and Supplementary Material, Table S2).
In a recent report, the activation of maternal transcripts in human PWS fibroblasts and a mouse model of PWS was demonstrated by using novel compounds that target histone methyltransferase G9a (36). Consistent with our findings, the activation of maternal PWS RNAs via G9a inhibition was associated with reduced levels of H3K9me3 and H3K9me2 at the SNORD116 locus as well as reduced levels of H3K9me2 at the PWS-IC, without affecting DNA methylation levels at the PWS-IC (36). We show here that, at least in humans, the ZNF274/SETDB1 complex is also required for H3K9me3-mediated silencing of maternal chromosome 15q11-q13 transcripts. While it remains to be determined if the G9a- and ZNF274/SETDB1-histone methylation are mutually independent or complimentary, there appear to be mechanistic differences. For example, NDN and SNRPN exon 1 are activated by G9a inhibition (36) but not by ZNF274 KO (Figs 2B, 4, 5B and Supplementary Material, Table S2). This difference could be explained by our observation that the ZNF274/SETDB1 complex specifically regulates brain-specific PWS lncRNA promoters whereas for Kim and colleagues, H3K9me2 reduction at the PWS-IC is responsible for NDN and SNRPN exon 1 expression, independently of the cell type.
We posit that the ZNF274 complex represses a cis-acting regulatory element that is required for initiating transcription from the SNRPN U1B and U1A promoters (Fig. 7). The regulatory element repressed by the ZNF274-complex could be an enhancer that activates the SNRPN U1B/U1A promoters. Alternatively, the element could be 116HG lncRNA cloud that functions to regulate the transcription of other genes (37). In our model (Fig. 7), a low level of expression of transcripts driven by the SNRPN upstream promoters in ZNF274 KO iPSCs is upregulated upon neuronal differentiation by brain-specific transcription factors. The activation of normally silent maternal PWS neuronal transcripts in our stem cell knockout model suggests that ZNF274 is a potential target for future therapeutic application in PWS. Our data (Fig. 1B, Supplementary Material, Fig. S2A and B) are consistent with the observation that ZNF274 acts in concert with other ZNF proteins to deposit H3K9me3 at genomic target sites (28). Although ZNF274 KO would be expected to result in complete loss of H3K9me3 at only about 10% of its target sites (28), genome-wide ZNF274 KO does not represent a viable strategy since ZNF274 binding does regulate potentially crucial functions outside the PWS locus (38). With this in mind, we are currently characterizing the ZNF274 BS over SNORD116 in order to specifically block or deplete ZNF274 binding at the PWS locus to reactivate maternal transcripts, offering a potential therapeutic treatment for PWS.
Materials and Methods
Culture conditions of iPSCs and neuronal differentiation
iPSCs were grown on irradiated mouse embryonic fibroblasts and fed daily with conventional hESC medium composed of DMEM-F12 supplemented with knock-out serum replacer, nonessential amino acids, L-glutamine, β-mercaptoethanol, and basic FGF. iPSCs were cultured in a humid incubator at 37 °C with 5% CO2 and manually passaged once a week (25).
Neuronal differentiation of iPSCs was performed using a monolayer differentiation protocol (39,40) with some modifications (25,26). Briefly, iPSC colonies were cultured in hESC medium for 24h before switching to N2B27 medium. Cells were fed every other day with N2B27 medium containing Neurobasal Medium, 2% B-27 supplement, 2mM L-glutamine, 1% Insulin-transferrin-selenium, 1% N2 supplement, 0.5% Pen-strep and was supplemented with fresh noggin at 500 ng/ml. After three weeks of neural differentiation, neural progenitors were plated on tissue culture plates coated with poly-ornithine/laminin. The neural differentiation medium consisted of Neurobasal Medium, B-27 supplement, nonessential amino acids, and L-glutamine, and was supplemented with 1 μM ascorbic acid, 200 μM cyclic adenosine monophosphate, 10 ng/ml brain-derived neurotrophic factor, and 10 ng/ml glial-derived neurotrophic factor. Unless otherwise specified, cells were harvested once neural cultures reached at least 10 weeks of age.
Lentiviral production, transduction, and clone screening
sgRNAs were designed using a web-based CRISPR design tool and cloned into lentiCRISPR (Addgene Plasmid 49535 and 52961) and lentiGuidePuro (Addgene Plasmid 52963) using our standard protocol (41–43). Lentiviral particles were made by transfecting 293FT cells with 2nd generation packaging systems using Lipofectamine 2000 (Life Technologies). Prior to transduction, iPSCs were treated with 10 μM ROCK inhibitor, Y-27632, overnight. The next day, iPSCs were singlized using Accutase (Millipore) and transduced with lentivirus in suspension in the presence of 8 μg/ml polybrene in a low-attachment dish for 2 h. Then, the iPSCs/lentivirus mixture were diluted 1: 1 in hESC medium and plated on puromycin-resistant (DR4) MEF feeders at a low density, supplemented with 10 μM ROCK inhibitor, Y-27632, overnight. Attached cells were cultured in hESC medium for an additional 72 h before starting drug selection using puromycin at 0.5 μg/ml during the first week and at 1 μg/ml during the second week. Puromycin-resistant iPSC colonies were individually picked into a new feeder well and screened for indels by performing PCR on genomic DNA and sequencing. The pluripotency of gene edited iPSCs was validated by immunocytochemistry using mouse anti-human stage specific embryonic antigen 4 (SSEA4) and rabbit anti-human OCT3/4, both from Molecular Probes, as previously described (41). Karyotyping and Affymetrix HD 6.0 array were performed by the Genetics and Genomics Division of the UCONN Stem Cell Core. Twenty G-banded metaphase cells from each iPSC line were examined to generate a karyotype for each line.
RNA isolation and RT reaction
RNA was isolated from cells using RNA-Bee (Tel Test, Inc.). Samples were DNase-treated as needed with Amplification Grade DNaseI (Invitrogen) at 37 °C for 45 min, and cDNA was synthesized using the High Capacity cDNA Reverse Transcription Kit (Life Technologies) according to the manufacturer's instructions.
RT-qPCR and expression arrays
For single gene expression assay, expression levels of target genes were examined using TaqMan Gene Expression Assays (Applied Biosystems) on the Step One Plus (ThermoFisher Scientific) or on the BioRAD CFX96 Real Time PCR system (Biorad). An amount of RT reaction corresponding to 30ng of RNA was used in a volume of 20ul per reaction. Reactions were performed in technical duplicates or triplicates and the GAPDH Endogenous Control TaqMan Assay was used as an endogenous control, following the manufacturer’s protocol. Relative quantity (RQ) value was calculated as 2−ΔΔCt using the normal cell lines CTRL1 or CTRL2 as the calibrator sample.
We designed custom-formatted Taqman low-density arrays (TLDA) with 48 target genes, including two housekeeping genes, allowing for 8 samples (including CTRL2) per card. All primer/probe sets are inventoried in Supplementary Material, Table S2a. Gene expression assays were supplied by Applied Biosystems. For TLDA analysis, 400ng of DNAse-treated RNA was used per RT reaction, according to the manufacturer’s directions. cDNA sample, equivalent to 150ng effective RNA, ribonuclease-free water, and PCR master mix were loaded into each TLDA-card fill port. The samples were distributed on the plate by centrifugation. Real-time PCR was performed on the 7900HT or ViiA7 Real-Time PCR systems (Applied Biosystems). We systematically loaded the same CTRL2 sample as our calibrator or Internal Positive Control (IPC) into each card and used the Thermo Fisher Cloud interface to analyse the data with the IPC settings to normalize Cq values across the different plates.
Chromatin immunoprecipitations
ChIP assays were performed as described before (22,44,45). The antibodies anti-ZNF274 (Abnova, Cat# H00010782-M01) and anti-trimethyl histone H3 (Lys9) (H3K9me3; Millipore, Cat# 07–442) were used. Quantification of ChIPs was performed using SYBR Green quantitative PCR. PCR primers used to amplify the purified DNA can be found in Supplementary Material, Table S3. The enrichment of the DNA was calculated as percent input, as described (45). Normal rabbit IgG was used for the isotype controls and showed no enrichment. Data were presented as means with SD and represent the average of at least two biological replicates from independent cultures.
Detection of 5hmC levels
Percentages of 5-methylcytosine (5mC), 5-hydroxymethylcytosine (5hmC) and unmodified cytosine (C) in DNA were assessed using the EpiMark 5-hmC and 5-mC Analysis Kit (New England Biolabs, catalog # E3317S) (22,46,47). qPCR primers used in these assays are denoted in Supplementary Material, Table S3. Reported values represent the average of at least two independent experiments, each analysed in triplicate by quantitative PCR. Data were presented as means and SD of independent experiments.
Statistical tests
Statistical analysis was carried out using Prism software (GraphPad). For each condition shown, averaged values from a minimum of two biological replicates from independent cultures were calculated and the resulting standard deviation (SD) was reported in the error bars. Unless otherwise specified, for each experiment, averaged values for each sample were compared with that of the parental PWS cell line of the same genotype (PWS LD or PWS UPD) and the significance for each un-manipulated vs. KO pair was calculated using the one- or two-way analysis of variance (ANOVA) with the Dunnett post-test.
Supplementary Material
Supplementary Material is available at HMG online.
Acknowledgements
We thank David S. Rosenblatt, Gail Dunbar and Daniel J Driscoll for patient clinical evaluation and information, and for providing skin biopsies/fibroblasts. We thank the Jackson Laboratory for Genomic Medicine and the UCONN Health Clinical Research Core for use of their real-time PCR instruments, the UCONN Health Molecular Core and Antoine David, a student in the Magistère de Génétique Graduate Program at Université Paris Diderot, Sorbonne Paris Cité, for their assistance. We also thank Justin Cotney and members of the lab for their helpful discussions and Sourour Mansour and Stormy Chamberlain for critical reading of the manuscript. The contents in this work are solely the responsibility of the authors and do not necessarily represent the official views of the state of Connecticut.
Conflict of Interest statement. The authors declare no competing financial interests.
Funding
Foundation for Prader-Willi Research, CT Regenerative Medicine Fund, Cascade fellowship (M. L.)
Web Resources
UCSC Human Genome Browser, http://genome.ucsc.edu/cgi-bin/hgGateway Web-based CRISPR design tool, http://crispr.mit.edu TIDE: method for easy quantitative assessment of genome editing, https://tide.nki.nl/ CRISP-ID: Detecting CRISPR mediated indels by Sanger sequencing, http://crispid.gbiomed.kuleuven.be/ RoadMap Epigenomics, http://egg2.wustl.edu/roadmap/web_portal/imputed.html#imp_sig. All websites were last accessed on December 12, 2017.
References
Author notes
Maéva Langouët and Heather R. Glatt-Deeley contributed equally to this work.
