iPSC generation and characterization

The study focuses on a young female CdLS case characterized by a heterozygous missense point-mutation [nucleotide 5483(G/A)] of the NIPBL gene. This mutation causes a single amino-acid substitution [amino-acid 1828 (Arg/Gln)] in the corresponding protein-product. Noticeably, the ClinVar section of the NIH National Library of Medicine (https://www.ncbi.nlm.nih.gov/clinvar/variation/159154/) contains the description of four other CdLS cases characterized by an identical mutation [(NM_133433.4(NIPBL): c.5483G > A (p.Arg1828Gln)] of the NIPBL gene. In three of the cases, this specific NIPBL mutation is annotated as the CdLS pathogenic mutation.

In a first set of experiments, we generated 3 distinct iPSC lines from the peripheral blood leukocytes of the CdLS patient and the two healthy parents, using a Sendai-virus vector based reprogramming approach. As indicated by whole-genome DNA-sequencing (WGS) analyses, the CdLS-derived iPSC line (MNCdLS1) maintains the allelic 5483(G/A) point-mutation of the NIPBL gene (Fig. 1a). The WGS analysis did not reveal any other unexpected mutations in the MNCdLS1 line. The point-mutation is absent in the iPSC lines generated from the two healthy parents (data not shown). Our shallow WGS analyses confirm that the established CdLS-derived as well as the father-derived (MNFa1) and the mother-derived (MNMo1) iPSC lines are characterized by a normal karyotype (Fig. 1b). The MNCdLS1 iPSCs were used to generate isogenic cell-lines, such as the G12 cell-line, in which the NIPBL gene mutation was corrected with a CRISPR/CAS9 approach (Supplementary Figure S1) [15]. The stemness of the established iPSC lines was monitored periodically by immunofluorescence analysis of the OCT4A, TRA1-60, SOX2 and LIN28 stem-cell markers (Fig. 1c).

Fig. 1

Characterization of the CdLS and control iPSC lines a WGS (Whole Genome Sequencing) was conducted with DNA extracted from the patient derived MNCdLS1 iPSCs. The mutation (nucleotide 5483) of the NIPBL gene on chromosome 5p13.2 is indicated in a box. b Karyotype analysis was performed using the WGS data generated from the DNA of MNCdLS1, MNMo1 and MNFa1 iPSCs. c Immunofluorescence analysis was performed on MNCdLS1 iPSCs. Stem cell marker proteins were determined with anti-OCT4A (Cell Signaling #2750), anti-TRA1-60 (Miltenyi Biotech 130–100-347), anti-SOX2 (Cell Signaling #4194) and anti-LIN28 (Cell Signaling #3694) antibodies. d The panel shows the levels of NIPBL and cohesin complex proteins expressed in MNCdLS1 and G12 cells. Soluble and insoluble nuclear proteins were isolated from the MNCdLS1 and G12 iPSC lines. Protein extracts (20 µg of protein/sample) were separated by 4–15% gradient polyacrylamide gel electrophoresis and transferred to nitrocellulose membranes. We used the primary anti-NIPBL (Betyl Biotech A301-779A), anti-SMC1A (Cell Signaling #6892), anti-SMC3 (Cell Signaling #5696), anti-RAD21 (Abcam ab992), anti-MAU2 (Abcam ab183033), anti-HistoneH3 (Upstate 06–599) and anti-TOPO isomerase 2β (BD Transduction Laboratories 611,492) antibodies. For the detection of the NIPBL protein, soluble protein extracts of the 293 T cells (293TN) were used as positive controls. The molecular weight of the detected proteins is shown on the right. Histone H3 (H3, 17 KDa) and TOPO isomerase 2β (TOPO2β,180 KDa) were used as loading controls for the insoluble and soluble nuclear fractions, respectively

In a second set of studies, we evaluated whether the NIPBL gene-mutation altered the levels of the NIPBL protein and other components of the cohesin complex. To this purpose, we prepared soluble and insoluble nuclear extracts from the MNCdLS1 as well as the mutation-corrected G12 iPSC lines and we subjected them to Western-Blot analyses. In both the soluble and insoluble (DNA-bound) extracts of the MNCdLS1 and G12 cells, we observed similar levels of the NIPBL protein, the cohesin complex components, SMC1A, SMC3 and RAD21, as well as the NIPBL partner protein, MAU2 (Fig. 1d). This was consistent with the transcriptomic results, which demonstrated that the MNCdLS1 and G12 iPSC lines expressed similar levels of the NIPBL, SMC1A, SMC3, RAD21 and MAU2 mRNAs as well as the HDAC8 and ANKRD11 transcripts, which encode two other cohesin-complex related proteins (Supplementary Figure S1b).

Differentiation of iPSCs into the hepatic cell lineage

CdLS is a disease characterized by significant deficits in the development of numerous tissues, which are likely to be the consequence of a defective differentiation of the embryonic cells along various histological lineages. Thus, we evaluated the effects exerted by the above mentioned NIPBL mutation on the in vitro differentiation of iPSCs along the hepatocytic lineage. We selected this lineage as there are two reports in murine models indicating that complete deletion [18] and haploinsufficiency [19] of the NIPBL gene in the hepatic tissue cause a dysregulation of chromatin reorganization involving the formation of Topologically-Associated-Domains (TADs). The hepatic cell lineage differentiation process of iPSC cultures requires at least 22 days and consists of 3 steps: 1) the generation of definitive endoderm cells at day 6 (D6); 2) the transformation of definitive endoderm cells into hepatoblasts at day 9 (D9); 3) the progression of hepatoblasts into a population of mature hepatic cells at day 22 (D22). Thus, we performed comparative studies on the MNCdLS1 and G12 cell-lines, which we induced to mature along the hepatic lineage.

In a first set of studies, we conducted single-cell RNA-seq analyses on MNCdLS1 and G12 iPSCs at D6, D9 and D22 (Supplementary Table S1 for the differential expression analysis data). In both MNCdLS1 and G12 cultures, the single-cell RNA-seq data demonstrated that undifferentiated iPSCs (D0), definitive endoderm cells (D6), hepatoblasts (D9) and mature hepatic cells (D22) separated into 4 distinct clusters (Fig. 2a; Supplementary Figure S2a–e). D22 cells (Fig. 2b) could be further distinguished into 4 populations according to a number of gene signatures available in published liver atlases [20,21,22,23,24,25]: (1) mature hepatocytes; (2) cholangiocytes; (3) hepatic stellate cells; (4) proliferating cells (small fraction). The mRNA levels of stage-specific differentiation markers [26] were measured in each cluster (Fig. 2c). Similar amounts of the NANOG, POU5F1 (also known as OCT4A) and SOX2 transcripts, which code for stem-cell markers, were detectable in the clusters of undifferentiated MNCdLS1 and G12 iPSCs. By converse, an almost identical content of the two other LIN28A and PODXL stem-cell marker mRNAs was measurable not only in the clusters of undifferentiated MNCdLS1 and G12 iPSCs, but also in the clusters of definitive endoderm cells at D6 and hepatoblasts at D9. Similar expression levels of the definitive endoderm markers, HHEX, FOXA2, CXCR4, SOX17 and KIT mRNAs were observed in the D6 clusters of MNCdLS1 and G12 cells. In addition, comparable amounts of the hepatoblast marker, HNF4A, were detectable in the D6, D9 and the D22 clusters derived from both MNCdLS1 and G12 iPSCs. In contrast, the other hepatoblast marker, PROX1, was detectable only in the D22 hepatocytes and cholangiocytes derived from the G12 iPSCs. Finally, the expression profiles of the AFP and ALB transcripts encoding two mature hepatocyte markers, were different. Indeed, the AFP mRNA was expressed only in the D22 clusters of MNCdLS1 and G12 derived hepatocytes and the levels of this transcript were similar in the two clusters. In contrast, significant amounts of the ALB mRNA were observed only in the NIPBL-mutation corrected G12 hepatocytes, cholangiocytes, hepatic stellate and proliferating cells at D22. To validate our single-cell RNA-seq results, we performed real-time PCR analyses on the transcripts (POU5F1, HHEX, FOXA2, CXCR4, HNF4A, PROX1, AFP and ALB) coding for some of the markers mentioned above (Supplementary Figure S3). Taken together, the PCR results confirmed the data obtained with the single-cell RNA-seq methodology.

Fig. 2

Single-cell RNA-sequencing of MNCdLS1 and G12 iPSC cultures undergoing differentiation along the hepatocyte lineage a Single-cell RNA-seq studies were performed on undifferentiated MNCdLS1 and G12 iPSC cultures (iPSCs) and following differentiation of the cultures at day 6 (D6), day 9 (D9) as well as day 22 (D22). The UMAP (Uniform Manifold Approximation and Projection) plots show the distribution and annotation of the cells based on the gene expression profiles of the undifferentiated iPSCs (pink dots) as well as the differentiated counterparts at day 6 (definitive endodermal cells, green dots), at day 9 (hepatoblasts, yellow dots) and day 22 (hepatocytes, violet dots; cholangiocytes, red dots; hepatic stellate cells, blue dots; proliferating cells, light blue dots). The cell cultures were obtained from the patient (MNCdLS1, left) and the mutation-corrected isogenic cell line (G12, right). b Enrichment of the cell type-specific signatures (hepatoblasts, upper left; hepatocytes, upper right; cholangiocytes, lower left; hepatic stellate cells, lower right) confirms cluster annotation. c The violin plots display the expression levels of indicative genes in MNCdLS1 (red) and G12 (blue) cultures across the various differentiation stages

In a second set of studies, we evaluated whether expression of some of the biomarker mRNAs (NANOG, FOXA2, HNF4A, PROX1, AFP and ALB), which we considered, resulted in the synthesis of the corresponding protein (Fig. 3a). To this purpose, we performed immunofluorescence experiments in MNCdLS1 and G12 cells at different stages along the hepatocyte differentiation pathway. Representative immunofluorescence images, showing the expression of specific marker proteins (NANOG, FOXA2, HNF4 and AFP) in G12 cells [25, 26], are presented in Supplementary Figure S4a. At D9, the endodermal FOXA2 protein was equally expressed in MNCdLS1 and G12 cells (Fig. 3a). By converse, at D22, the levels of the hepatoblast PROX1 protein were significantly lower in MNCdLS1 than G12 cells (Fig. 3a). These observations are consistent with the mRNA levels which we determined by single-cell RNA-seq and PCR analyses. However, the data obtained by immunofluorescence analysis were not always in line with the transcriptomic data. Indeed, at D22, the number of HNF4A positive MNCdLS1 cells was significantly lower than the number of HNF4A positive G12 cells, which is different from what we observed at the mRNA level (Fig. 2c). Similarly, almost all the HNF4A positive MNCdLS1 and G12 cells were also AFP positive, while the number of HNF4A and ALB co-expressing cells was significantly lower in MNCdLS1-derived hepatocyte-like cultures than the G12 counterparts (Fig. 3b, right graphs). Taken together, the immunofluorescence results support the idea that the differentiation deficits observed in patient-derived iPSCs tend to become evident during the late stages of hepatocyte maturation.

Fig. 3

Immunofluorescence analyses on differentiated MNCdLS1 and G12 iPSCs Immunofluorescence analyses were performed on differentiated MNCdLS1 and G12 iPSCs at day 9 and day 22, as indicated. Fluorescent images were captured under the microscope (Magnification × 200). We used the following antibodies: anti-FOXA2 (Invitrogen 701,698), anti-PROX1 (Invitrogen PA5-85,552), anti-HNF4A (Invitrogen MA1-199), anti-AFP (anti-AFPα; Invitrogen PA5-16,658) and anti-ALB (anti-albumin; Invitrogen PA5-111,010). a Representative immunofluorescence images obtained with anti-FOXA2 and anti-PROX1 antibodies are illustrated. b Representative immunofluorescence images obtained with combinations of anti-HNF4A and anti-AFP or combinations of anti-HNF4A and anti-ALB antibodies are illustrated. The number of HNF4A positive cells (magenta-fluorescence) and the number of nuclei (blu-fluorescence; DAPI) were counted in 10 separate fields (approximately 1000 cells/field) and the percentage of HNF4A positive nuclei relative to the total number of nuclei was calculated (upper right bar-graph). The percentage of HNF4A positive cells (magenta-fluorescence) in AFP positive cells (green-fluorescence) is illustrated by the middle right graph. The percentage of HNF4A positive cells (magenta-fluorescence) in ALB positive cells (green-fluorescence) is shown in the bottom right graph. ** Significantly lower in MNCdLS1 relative to G12 cell cultures (p < 0.01; unpaired t-test). c CYP3A4 enzymatic activity was measured in hepatocyte differentiated cells at D25 of differentiation with or without induction with dexamethasone (50 µM) for 24 h. Following exposure of cells to the luciferin substrate for 1 h, cell supernatants were used for CYP3A4 enzymatic assays. In parallel cells were detached from the plates and the number of hepatocytes (30–40 µm in diameter) was determined and used for the normalization of the results. ** Significantly lower in MNCdLS1 relative to G12 cell cultures (p < 0.01; unpaired t-test)

To define whether the iPSC derived mature hepatocytes were functionally active, we determined CYP3A4 enzymatic activity in the MNCdLS1 and G12 derived hepatocytes at D25 following exposure of the cultures to dexamethasone (50 µM) or vehicle (DMSO). Indeed, it is known that hepatic CYP3A4 is inducible by dexamethasone at both the transcriptional and protein levels [27]. In basal conditions, the cell lysates and supernatants of the G12 hepatocyte cell populations were endowed with measurable and equivalent amounts of CYP3A4 activity, while the patient-derived counterparts contained barely detectable amounts of the enzyme. Given this last result, we tried to induce CYP3A4 activity by the addition of dexamethasone (50 µM). However, we did not observe a significant induction of the CYP3A4 mRNA and the corresponding protein (Fig. 3b and Supplementary Figure S4b). This might be the consequence of the fact that the original hepatocyte medium used from Day 9 up to Day 25 contained a low, albeit significant, amount of dexamethasone (10 nM). The limited induction of CYP3A4 activity by dexamethasone may be due to the immaturity of the hepatocytes generated in our cultures. CYP3A4 activity was normalized for the number of mature hepatocytes (30–40 micron in diameter) and the results obtained are presented in Fig. 3b. Differentiated iPSCs at D22 are capable of producing the cholangiocyte specific TM4SF1 and the hepatocyte specific FABP1 transcripts. Noticeably, expression of the two last markers was similar in MNCdLS1 and G12 cultures (Supplementary Figure S4b).

iPSCs and liver cells transcriptomic analyses

To identify Differentially Expressed Genes (DEGs), we conducted comparative RNA-seq studies in MNCdLS1 and control G12/MNMo1/MNFa1 derived iPSCs as well as the corresponding hepatocyte-like cells at D22 (Supplementary Tables S2, S3).

We performed initial experiments in undifferentiated iPSCs to identify DEGs in MNCdLS1 and G12 cultures. Subsequently, we performed further analyses to select the genes commonly and significantly up- or down-regulated in MNCdLS1 iPSCs relative to the G12, MNMo1 and MNFa1 counterparts. Using an FDR (False Discovery Rate) value < 0.01, we identified the same number of common genes (116 genes), which were significantly up-regulated and down-regulated in MNCdLS1 relative to the three corresponding controls (Supplementary Table S2). If the analysis was restricted to DEGs with a Log2-Fold-Change (Log2FC) value > 1.0 or < -1.0, only 10 and 42 common DEGs turned out to be significantly up-regulated (Log2FC ≥ 1, FDR ≤ 0.01) and down-regulated (Log2FC ≤ -1, FDR ≤ 0.01), respectively (Fig. 4a and Supplementary Table S4). The up-regulated genes consisted of 6 protein-coding genes (CHCHD2, ERVH48-1, FOXL2, IER3, NLRP7 and PRAC1), 3 pseudogenes (PAICSP7, ROCK1P1 and SLC6A10P) and one antisense-RNA coding gene (PAX8-AS1). Except for 3 genes (C14orf39, LOC400499 and LOC441666), all the down-regulated genes encode functional protein products, including zinc finger proteins (ZNF208, ZNF257, ZNF385D, ZNF676, ZNF728 and ZNF736), solute transporters (SLC16A14 and SLC44A5), cell adhesion molecules (CDH11, NRCAM, NCAM1 and PWWP3B) and a neurotransmitter receptor (GRIN2A). Noticeably, some of the down-regulated genes (DPP6, GRIK2 and RBFOX1) are involved in neural development/differentiation and neurological diseases.

Fig. 4

Transcriptomic analysis of iPSCs and hepatocyte-like differentiated cells RNA prepared from undifferentiated a and differentiated b MNCdLS1, G12, MNMo1 and MNFa1 cultures were used for RNA-seq analysis. The heat maps show the up-regulated DEGs (red) and down-regulated DEGs (blue) in MNCdLS1 relative to all control cells (G12, MNMo1 and MNFa1). We selected DEGs characterized by FDR threshold values < 0.01 and a Log2FC > 1.0 for the upregulated DEGs and < -1.0 for the down-regulated DEGs

In a second set of studies, we conducted additional RNA-seq experiments in the MNCdLS1, G12, MNMo1 and MNFa1 derived populations of hepatic cells at D22. Using a threshold 0.01 FDR-value, we identified 276 and 220 common genes, which were up- and down-regulated, respectively, in MNCdLS1 derived hepatic cells relative to the G12, MNMo1 and MNFa1 control counterparts (Supplementary Table S5). Forty of the up-regulated genes presented with a Log2FC value ≥ 1.0 and 93 of the down-regulated genes presented with a Log2FC value ≤ -1.0 (Fig. 4b). Twenty-seven of the up-regulated genes are protein-coding, 8 genes are transcribed into anti-sense RNAs (CDKN2B-AS1, FLG-AS1, PAX8-AS1 and PSG8-AS1) and Long-Intergenic-Non-Protein-Coding RNAs (LINC00578, LINC00862, LINC01239 and LINC01435), while 5 of them are pseudogenes (FOLH1, HERC2PS, PAICSP7, SLC6A10P and TDGF1P3). Thus, approximately one third of the up-regulated genes do not code for functional proteins. This is consistent with the data reported for the liver of NIPBL-deficient mice [18], where a significant number of illegitimate transcripts are generated due to a dysregulation of chromatin reorganization involving the formation of Topologically-Associated-Domains (TADs). As for the common 93 genes, which are down-regulated in MNCdLS1 derived hepatocytic cell populations, 92 of them (with the exception of LOC441666) encode functional proteins. Some of these genes code for cell-adhesion molecules (CDH12, CTNND2, PCDHs and VCAM1), solute carrier proteins (SLC16A9, SLC22A31 and SLC38A4), zinc finger proteins (ZNF208, ZNF257 and ZNF736), semaphorins (SEMA3F and SEMA6B) and the protocadherin-beta proteins (PCDHB9-16) whose gene-cluster locates on chromosome 5q31.38. Interestingly, the brain of NIPBL-deficient mouse embryos and the NeuN positive nuclei in the brain of CdLS patients contain reduced levels of several PCDHB transcripts [28, 29]. Finally, many of the genes down-regulated in MNCdLS1 derived hepatocyte-like cells code for proteins related to neurological disorders, such as epilepsy (CTNND2, KCNJ3, PKNOX2, PLCL1, PTRRZ1 and SEMA6B), schizophrenia (GPR153), craniofacial deafness hand syndrome (CDH12), spinocerebellar ataxia (DAB1), intellectual developmental disorders (DPP6) and leukoencephalopathy (BEND5).

In a last set of analyses, we focused on genes which were up- or down-regulated in CdLS-derived iPSCs and maintained this altered expression during hepatic cell differentiation. Only 3 protein-non-coding genes (PAICSP7, PAX-AS1 and SLC6A10P) were up-regulated, while one gene encoding a Long-Intergenic-non-protein-Coding RNA (LOC441666) and 4 protein-coding genes (DPP6, ZNF208, ZNF257 and ZNF736) were down-regulated in patient-derived iPSCs and hepatocytic cell populations (D22).

The decrease of the DPP6 mRNA observed in undifferentiated and differentiated MNCdLS1 iPSCs upon RNA-seq analysis was confirmed with the use of a specific real-time PCR assay (Fig. 5a). In human pancreatic adenocarcinoma [30] and neuronally differentiated murine P19 cells [31], the DPP6 gene is silenced as a consequence of the DNA-methylation of the corresponding promoter region [32]. As native and differentiated MNCdLS1 iPSCs show significantly lower levels of the DPP6 transcript relative to the control counterparts, we evaluated whether this was due to DNA-methylation of the DPP6 promoter. To define the DNA-methylation level of the DPP6 promoter, the corresponding genomic DNA region of MNCdLS1 and G12 iPSCs was chemically modified with bisulfite (Fig. 5b). The bisulfite treated MNCdLS1 DNA fragment turned out to be sensitive to BstU1 digestion, suggesting that the DPP6 promoter region was already DNA-methylated and it was not modified by bisulfite treatment. In contrast, the promoter region derived from the G12 iPSCs was resistant to BstU1 digestion, indicating that bisulfite treatment converted the un-methylated cytosines of the DPP6 promoter into uracil residues. Thus, DPP6 silencing in MNCdLS1 iPSCs seemed to be due to methylation of the promoter region. DNA-methylation of the DPP6 promoter by DNMT3B (DNA Methyl-Transferase Type 3B) causes silencing of the DPP6 gene [30]. Consistent with this, DNMT3B mRNA levels are significantly increased in our patient derived iPSCs relative to the control counterparts [Log2FC (FDR): MNCdLS1 vs G12 = 0.7395 (5.80E-29); MNCdLS1 vs MNMo1 = 0.88332 (8.60E-42); MNCdLS1 vs MNFa1 = 0.42825 (8.12E-10)]. In MNCdLS1 iPSCs, the DPP6 promoter region is silenced by an epigenetic mechanism involving DNA-methylation.

Fig. 5

Chromatin accessibility of the genomic region surrounding the DPP6 gene on Chr7q36.1 a DPP6 mRNA levels were measured in MNCdLS1, G12, MNMo1 and MNFa1 cells at each stage of hepatic cell differentiation by Real Time PCR using a specific DPP6 Taqman probe (Applied Biosystems 4,448,192, Hs00736294ml). The 18S Taqman probe (Hs99999901_s1) was used as a reference. The results were obtained from 3 biological replicates/experimental group and the experiments were repeated at least twice with similar results. *Significantly higher (p < 0.05; unpaired t-test) relative to MNCdLS1 cells; **Significantly higher (p < 0.01; unpaired t-test) relative to MNCdLS1. b Genomic DNA was extracted from iPSCs, MNCdLS1, and G12, and treated with bisulfite using the EZ DNA methylation Kit (Zymo Research, Irvine, CA). Subsequently the DNA fragment (149 bp) containing the promoter region of the DPP6 gene (nucleotides –252 to -104 from the transcription initiation site of the DPP6 gene variant 2 NM_001936) was amplified by PCR. The resultant DNA fragment was digested with BstU1. The DNA fragments derived from MNCdLS1 and G12 in the presence ( +) or absence (-) of BstU1 were analyzed on a 12% polyacrylamide gel (left figure). The DNA sequence of the promoter region is presented on the right and the sequence of the deoxy-oligonucleotide primers used for PCR amplification are underlined. The DNA sequence of the DPP6 mRNA is shown in bold italics. The transcription initiation site is shown in green, while the translation initiation signal (ATG) is underlined. The recognition sites of the BstU1 enzyme are marked in red. c ATAC-seq analysis was performed on genomic DNA prepared from undifferentiated (MNCdLS1, G12, MNMo1 and MNFa1) and hepatocyte differentiated iPSCs at day 9 (MNCdLS1 and G12) and at day 22 (MNCdLS1, G12, MNMo1 and MNFa1). Tagmented DNA fragments are shown as vertical lines and the corresponding genes are indicated. The DPP6 gene is underlined in red. The results were obtained from three biological replicates/experimental group, and the experiments were repeated at least twice with similar results

ATAC-seq analyses of iPSCs and liver cells

Since NIPBL is required for cohesin loading and cohesin is a major regulator of the 3D genome organization, we complemented the transcriptomic results with chromatin-accessibility data, which we obtained with an ATAC-seq approach. The objective of these studies was to identify the genes whose expression is altered in CdLS-derived iPSCs and maintained in the differentiated hepatic cell lineages via modifications of chromatin-accessibility. Although tagmented-DNA fragments do not necessarily represent transcriptionally active genomic regions, the results of our ATAC-seq analyses suggest that MNCdLS1 iPSCs contain several transcriptionally inactive genes, which map to closed-chromatin regions, as shown in the case of the DPP6 gene and the ZNF gene cluster (Fig. 5c and Supplementary Figure S5). Indeed, the decrease in DPP6 mRNA associates with alterations in chromatin-accessibility in the region of chromosome 7q36.2 containing the DPP6 gene. In undifferentiated and D9/D22 differentiated MNCdLS1 iPSCs, the ATAC-seq analysis indicates that a closed-chromatin region surrounds the DPP6 gene. By converse, in G12 iPSCs, the gene locates inside an open-chromatin region, as demonstrated by the frequency of tagmented DNA fragments. We observed similar chromatin-accessibility differences in the DPP6 gene between MNCdLS1 iPSCs and MNMo1/MNFa1 control iPSCs (Fig. 5c). As for the ZNF gene cluster, containing ZNF208, ZNF257, ZNF676 as well as ZNF728, the levels of the corresponding transcripts are significantly lower in MNCdLS1 relative to G12, MNMo1 and MNFa1 iPSCs. In addition, the expression of ZNF208 and ZNF257 mRNAs decreases also in differentiated MNCdLS1 iPSCs at D22 (Fig. 4 and Supplementary Table S2). In both undifferentiated and D9 differentiated MNCdLS1, iPSCs are covered by chromatin, while the G12 control counterparts do not show chromatin coverage (Supplementary Figure S5, Supplementary Tables S6, S7, S8).

To compare the global alterations of chromatin organization and mRNA expression in undifferentiated and differentiated (D22) MNCdLS1, G12, MNMo1 and MNFa1 iPSCs, we combined the ATAC-seq and RNA-seq results, presenting them as circular plots (Fig. 6). In undifferentiated iPSCs, the circular plots show significant chromatin-accessibility and gene-expression differences between patient-derived and control cells. A comparison of the MNCdLS1 and G12 results indicates a lower number of peaks relative to what is observed following comparison of the MNCdLS1 with the MNMo1 and MNFa1 data. The circular plots generated at D22 demonstrate that many of the chromatin-accessibility and gene-expression differences observed in CdLS derived iPSCs relative to the control counterparts are maintained in hepatic cell lineages. Overall, our data support the idea that the NIPBL-mutation of our patient is responsible for the perturbations in gene-expression and chromatin-accessibility observed in CdLS iPSCs.

Fig. 6

Circular plots of the integrated RNA-seq and ATAC-seq analyses The global transcriptional profiles are integrated with the chromatin-accessibility results and are presented under the form of circular plots. Inner circles represent the Log2FC values of the transcriptomic data comparisons shown above the circular plots. The middle circles represent the Log2FC (± 1·0) of the comparisons performed on the ATAC-sec data. The outer circles indicate the chromosome positions. The upper circular plots are derived from the data obtained on undifferentiated iPSCs, while the lower ones are obtained from the results at day 22 (D22)

To obtain information regarding the characteristics of the genomic regions whose levels of chromatin-accessibility are modified in undifferentiated and differentiated MNCdLS1 iPSCs relative to the control counterparts, we evaluated the human genome annotations. Supplementary Figure S6 contains the annotations defining the types of genomic regions whose chromatin accessibility is altered in MNCdLS1-derived iPSCs and the hepatic cell populations. A series of comparisons (MNCdLS1 vs G12, MNCdLS1 vs MNFa1 and MNCdLS1 vs MNMo1) indicate that the profiles of the various genomic regions, which are changed in patient-derived iPSCs and D22 differentiated cells, are similar. Relative to the genome of control undifferentiated iPSCs, the distal intergenic regions of the MNCdLS1 counterpart are the most frequently altered (36–42%). Other frequently modified regions are intronic sequences (including first introns, 33–40%) and promoter regions (11–19%). The major differences between iPSCs and D22 cells are observed in the percentage of promoter regions. For instance, in the MNCdLS1 vs. G12 comparison, the percentage of promoter regions increases to 38% in the former type of cultures.

ChIP-Seq analyses on iPSCs

To get insights into the mechanisms underlying the transcriptomic differences observed in the MNCdLS1 and control iPSCs, we focused our attention on chromatin organization, as the ATAC-seq results suggested that MNCdLS1 cells were characterized by specific abnormalities in the closed and open chromatin regions. Thus, we performed ChIP-Seq studies on the iPSCs derived from the CdLS patient and the respective controls using specific antibodies targeting RAD21 (cohesin complex component), CTCF (loop extrusion regulator) and H3K4me3 (transcriptionally active histone mark). The number of DNA fragments precipitated with the RAD21-targeting antibody was higher in MNCdLS1 relative to that of the corresponding G12 cells. By converse, the number of the DNA fragments recognized by the H3K4me3 and the CTCF specific antibodies was not significantly altered in MNCdLS1 and G12 iPSCs (Supplementary Tables S9, S10).

Subsequently, we considered the two specific regions of chromosome7q36.2 and chromosome 19p12 containing the DPP6 gene (Fig. 7a) and the ZNF gene cluster (Fig. 7b), respectively. As expected, the profile of the chromatin bound DNA fragments recognized by the anti-H3K4me3 antibody was similar to the profile detected by ATAC-seq analysis. In both genomic regions, the number and intensity of the chromatin bound DNA fragments specifically precipitated with the anti-H3K4me3 antibody were reduced in the MNCdLS1 derived iPSCs relative to the G12 counterparts. The results suggest that the decreased levels of DPP6 and ZNF transcripts observed in MNCdLS1 cells are due to an increased coverage of the corresponding genomic regions by chromatin. When the ChIP-Seq analyses were performed with the anti-RAD21 antibodies, the number of DNA fragments was substantially increased in the patient derived iPSCs. When the ChIP-Seq studies were performed with an antibody targeting the CTCF transcriptional regulator, however, we observed that the number and the intensity of the DNA fragments in the genomic regions containing the DPP6 gene and the ZNF gene cluster decreased in the MNCdLS1 iPSCs relative to the G12 counterparts (Fig. 7a–b). This observation suggests that there is a reduced number of cohesin complex proteins which are bound to the chromosome and correctly stabilized by CTCF in MNCdLS1 iPSCs relative to the control iPSCs.

Fig. 7

Epigenomic profiling of the DPP6 gene, ZNF gene cluster and PCDHB genes in iPSCs We performed ATAC-Seq (Assay for Transposase-Accessible Chromatin) and ChIP-Seq (Sequencing following Chromatin Immunoprecipitation) analyses for the genomic regions recognized by H3K4me3, RAD21 and CTCF proteins. The DPP6 gene, the ZNF and the PCDHB cluster regions with the corresponding epigenomic tracks are shown in a, b and c, respectively. In the case of the H3K4me3, RAD21 and CTCF genes, the tracks corresponding to the MNCdLS1 and G12 iPSCs are displayed. Following immunoprecipitation with anti-H3K4me3, anti-RAD21 and anti-CTCF antibodies, reverse cross-linked DNA fragments were isolated and sequenced. The MNCdLS1 and G12 peak callings are shown with MACS3 for ChIP-Seq and Genrich for ATAC-Seq. The genomic tracks (MNCdLS1 = red; G12 = blue) are visualized using the respective bigWig files. All plots were generated using the R package "Trackplot".

The augmented chromosome interaction of RAD21 protein in MNCdLS1 iPSCs seems to be a general phenomenon, being observed in the entire genome. In contrast, the alteration of CTCF interaction occurs only in the genomic regions (DPP6 gene and ZNF gene cluster) where transcription is strictly regulated by chromatin reorganization. The concept is exemplified by the results obtained with the PCDHB gene-cluster, whose mRNAs and proteins are not expressed in undifferentiated iPSC cultures and appear only upon differentiation into hepatocyte-like cells. In fact, ATAC and ChIP-Seq analyses performed in undifferentiated MNCdLS1 and G12 iPSCs demonstrate that the genomic region containing the PCDHB gene cluster (chromosome 5q31) present with an increase in RAD21 chromatin binding (Fig. 7c). Consistent with the fact that PCDHB proteins are expressed only in differentiated hepatic cells, we observed no difference in the number and intensity of the DNA fragments generated with the use of the anti-H3K4me3 antibody in MNCdLS1 and G12 iPSCs (Fig. 7c). In addition, MNCdLS1 and G12 iPSCs were characterized by a similar pattern of DNA fragments when the anti-CTCF antibody was used. This last observation indicates no sign of CTCF binding dysregulation in MNCdLS1 and G12 iPSCs. Similar results were obtained, when we performed ChIP-Seq studies on chromosome 5p13.2 (NIPBL gene) and chromosome 1q32.3 (PROX1 gene) (Supplementary Figure S7a-b). In conclusion, the ChIP-Seq data are consistent with the idea that the CTCF-binding dysregulation observed in MNCdLS1 iPSCs occurs only in the genomic regions which are tightly and directly regulated by chromatin reorganization in control iPSCs.