Cell cultures

The first batch of the N. siki cell line, named batch1 (Fig. 1A), was thawed from a stock at the University of Bari (Uniba, Italy), kept in culture for 21 days (9 passages) and used to prepare strand sequencing (Strand-seq) libraries for 96 cells, metaphase preparation (prep1A) for FISH analysis (Mercuri et al. 2022), and genomic DNA for WGS, as shown in Figure 1B and Supplemental Table S1.

Figure 1.

Geographical origin and experimental workflow for the different batches of the Nomascus siki cell line analyzed in the present study. As shown in panel A, all three batches originated from the University of Cambridge: batch1 (purple) was directly transferred to Uniba (Italy); batch2 (green) and batch3 (red) were first shipped to UTAD (Portugal) and subsequently to Uniba for further processing and analysis. Panel B illustrates the processing of all the batches at Uniba (Italy), specifying culturing timelines, with the number of passages indicated in parentheses, the production of frozen aliquots, and metaphase preparations for each batch, along with the sequencing techniques applied to them.

Open in new tabDownload PowerPointLink to figure

The second cell line batch (batch2), derived from the same N. siki individual, stocked at the Universidade de Trás-os-Montes e Alto Douro (UTAD; Portugal), was thawed at Uniba and kept in culture for 7 days (three passages). A first metaphase preparation (prep2A) was obtained at the end of the culture, and one aliquot of the cell line was frozen (aliquot 2al1).

Later, aliquot 2al1 was thawed and used for Strand-seq library preparation, which led to the sequencing of 99 single cells. The culture was further kept for an additional 22 days (10 passages), after which a second metaphase preparation (prep2B) was produced, and a second aliquot (2al2) was frozen. Following sequencing, aliquot 2al2 was thawed and cultured for 10 days (five passages) to obtain a metaphase preparation (prep2C).

All the metaphase preparations were used to perform FISH experiments, as described in the following paragraphs.

A third stock aliquot of the same cell line from Portugal (batch2.1) was cultured at Uniba to replicate the results. The culture was maintained for 33 days (16 passages), and two metaphase preparations were obtained from this third batch, one after 1 week of culture (prep2.1A; three passages) and the other after 33 days of culture (prep2.1B; 16 passages). Aliquot 2.1al1 was frozen after 7 days of culture (three passages) and cultured again for 14 days (six passages) to obtain the last metaphase preparation (prep2.1C) and genomic DNA.

Finally, a last aliquot from a different batch (batch3) of the same cell line from the UTAD in Portugal was thawed at Uniba and kept in culture for 7 days (three passages) to obtain a metaphase preparation (prep3A) and the genomic DNA (Fig. 1B; Supplemental Table S1).

All cell line batches originate from a common source: the University of Cambridge (Cambridge, United Kingdom). However, detailed information regarding their initial handling and precise transfer timelines to Uniba and UTAD is limited (Fig. 1A).

Single-cell template strand sequencing to identify structural rearrangements

Out of 99 single cells sequenced from N. siki (aliquot 2al1, batch2), 28 cells passed quality control and were used for further analyses. First, based on their orientations, we assigned each sequenced template strand to either a Crick (C; “+”) or a Watson (W; “–”) state. Then, to select the informative chromosomes for composite file construction, we identified those in either a WW or CC state (i.e., displaying template strands in the same orientation on both homologs) per cell. During this process, we observed that Chr 14 showed a different pattern for the p- and q-arms, with an increased read depth in only one of the two arms per cell. Similarly, Chromosome 22 was noted to behave nonhomogeneously across cells, with the majority of them having a partial switch from WW to WC or from CC to WC in the terminal region of the q-arm.

We then grouped the WW and CC chromosomes of 22 cells with an increased read depth on the q-arm of Chr 14 into a composite file (Fig. 2A) and those from six cells with increased read depth on the p-arm into a separate composite file and analyzed them separately. All the cells belonging to the first group (q-arm increased read-depth) showed a 17 Mb heterozygous inversion on the q-arm of Chromosome 22, whereas no such inversion was detected in the second group (p-arm increase). No additional structural differences were observed between the two composite files.

Figure 2.

Detection and validation of heterozygous inversions. (A) Composite file for the Chromosome 14 iso-q subpopulation from batch2, built using the GGSC Nleu3.0/nomLeu3 reference genome. Orange and blue bins represent Watson (W) and Crick (C) strand inheritance, respectively, and pink ones represent centromeres. Chromosome 14 exhibits increased read depth and a clear p/q switch pattern, consistent with the presence of an isochromosome of the q-arm. Red squares highlight two of the detected heterozygous inversions. (B) Experimental validation by interphase FISH using locus-specific BAC probes (Supplemental Table S2). The schematic on the left shows the probe design and expected signal patterns for the directly oriented chromosomes. On the right, representative FISH images from nuclei in preparations prep2A (Exp3) and prep2B (Exp4) confirm the presence of the predicted heterozygous inversions (dir/inv configuration).

Open in new tabDownload PowerPointLink to figure

To further interpret these findings, we compared the composite files from batch2 with the one previously generated by Mercuri and colleagues (2022) from batch1, which was based on 62 informative cells from the same N. siki individual. That analysis had identified 22 inversions larger than 500 kb, including a 17 Mb heterozygous inversion on the terminal q-arm of Chromosome 22 (Chr 22a: 103,740,071–122,121,000, GGSC Nleu3.0/nomLeu3), consistently present in all sequenced cells (Fig. 3). This was confirmed to be the same event detected in batch2.

Figure 3.

Circular plot displaying the map of chromosomal rearrangements detected by Strand-seq. Each concentric ring represents the read-depth profile (copy number [CN]) across chromosomes for a specific batch: batch1 (innermost ring, green inversions), iso-q subpopulation of batch2 (middle ring, red inversions), and iso-p subpopulation of batch2 (outermost ring, blue inversions). Yellow segments indicate regions with a CN of two, and orange segments mark regions with CN = 3, such as Chromosome 14 (trisomy) and the isochromosome arms in iso-p and iso-q subpopulations. Inversion events are shown as colored bars (green for batch1, red for iso-q batch2, and blue for iso-p batch2): Inversions plotted outside the corresponding ring indicate heterozygous events, and those plotted inside the ring denote homozygous inversions.

Open in new tabDownload PowerPointLink to figure

Read-depth analysis of Chr 14 in batch1 revealed consistent trisomy across all informative cells. In contrast, batch2 displayed increased read depth restricted to either the p-arm or the q-arm, indicating potential subclonal variation. When comparing the inversion profiles detected by Strand-seq in batch1 with those in the two batch2 subgroups (defined by p- or q-arm read-depth increases on Chr 14), no additional structural differences were observed (Fig. 3).

Characterization of structural variants

iso-p, iso-q, trisomy, and Chromosome 22 inversion characterization

To investigate the increased read-depth on Chr 14 observed by Strand-seq in batch2 subpopulations, we performed a dedicated FISH experiment (Exp6) (Table 1; Supplemental Table S2). This consisted of a cohybridization with probes specific for the short (p) and long (q) arms of Chromosome 14 (Supplemental Table S2) and was performed on metaphase preparations from batch1 (prep1A), batch2 (prep2A, prep2B, prep2C), batch2.1 (prep2.1A), and batch3 (prep3A).

The results showed that batch1 and batch3 were trisomic for Chromosome 14, whereas in batch2 and batch2.1, metaphases contained either an isochromosome of the p-arm (iso-p) or of the q-arm (iso-q), in addition to the two normal homologs. This finding explains the arm-specific read-depth increases seen in Strand-seq and suggests the presence of distinct subclonal populations within the cultures.

To further clarify the relationship between the inversion on Chromosome 22 and the structural status of Chromosome 14, we designed a combined FISH experiment (Exp7) by cohybridizing the probes from both Exp5 and Exp6 on the same metaphases. This experiment was carried out on prep2B, prep2C, prep2.1A, prep2.1B, and prep2.1C (Fig. 4; Supplemental Fig. S1). In line with the Strand-seq results, the heterozygous inversion on Chromosome 22 was observed only in metaphases carrying the iso-q (Fig. 4A; Supplemental Fig. S1). Conversely, the inversion was never detected in metaphases in which the iso-p was present (Fig. 4B; Supplemental Fig. S1). These findings support a clear association between the presence of the inversion in Chromosome 22 and a specific subpopulation defined by the structure of Chromosome 14.

Figure 4.

Validation of isochromosomes and the Chromosome 22 inversion (inv22) by FISH. Cohybridization on metaphases from prep2.1A of BAC clones CH271-162A04 (red) and CH271-113K11 (blue) on Chromosome 14 and CH271-74O21 (red) and CH271-119L09 (green) on Chromosome 22. A wild-type Chromosome 14 exhibits one red hybridization signal on 14p and one blue signal on 14q. i(14p) shows two red signals on the same chromosome, and i(14q) shows two blue signals on the same chromosome. Similarly, the wild-type Chromosome 22 displays a centromere red–green signal pattern, whereas an inverted Chromosome 22 shows a centromere green–red signal pattern. This experiment confirms the presence of a supernumerary isochromosome i(14p)/i(14q) and demonstrates the presence of the inversion on Chromosome 22 in a heterozygous state but only in metaphases in which i(14q) is present. (A) FISH image of a metaphase with i(14q) and inversion on Chromosome 22 in a heterozygous state. (B) FISH image of metaphase with i(14p) and no Chromosome 22 inversion.

Open in new tabDownload PowerPointLink to figure

To clarify how such rearrangements might have functionally impacted the cell line, we performed an enrichment analysis on the human orthologs of the genes detected on the gibbon reference genome (GGSC Nleu3.0/nomLeu3) in the rearranged regions. We detected 136 genes in the inverted region on Chromosome 22, none of which are disrupted by the breakpoints; 386 genes on the p-arm of Chromosome 14; and 530 genes on the q-arm (Supplemental Table S3). Enrichment analysis, performed on the human orthologs using the GENE2FUNC module of FUMA, revealed a significant overrepresentation of genes involved in cell cycle regulation and progression (Supplemental Fig. S2A), along with a significant enrichment of genes that have been implicated in various tumor types when upregulated among those located on the p-arm of Chromosome 14 (Supplemental Fig. S2B).

Batch comparison at the SNP level

To assess the differences between the batches at a single-nucleotide level, we performed SNP-calling on WGS short-read data from batch1, batch2.1, and batch3. We obtained 14,373,821 calls, which were reduced to 10,843,361 by retaining only SNPs with filtered depth (DP) > 10 and genotype quality (GQ) > 20 in all the samples (see Methods). Among these, 10,651,171 SNPs displayed identical genotypes across the three batches: 6,252,888 were heterozygous (Ref/Alt) and 4,398,283 were homozygous for the alternative allele (Alt/Alt). Only 22 SNPs showed a fully divergent genotype among the three samples, in which each batch carried a different allelic state (one Ref/Ref, one Ref/Alt, one Alt/Alt).

For the remaining 192,168 SNPs, we performed pairwise comparisons to assess the number of identical genotypes shared between each pair. Batch1 and batch2.1 shared 55,613 identical SNPs, of which 1890 were homozygous for the reference allele, 24,490 were heterozygous, and 29,233 were homozygous for the alternative allele. The comparison between batch1 and batch3 revealed 64,133 shared SNPs, including 2212 homozygous for the reference allele, 28,281 heterozygous, and 33,640 homozygous for the alternative allele. Finally, batch2.1 and batch3 exhibited the highest number of identical SNPs, with 72,422 in total, comprising 2116 homozygous for the reference allele, 35,583 heterozygous, and 34,723 homozygous for the alternative allele. These results correspond to a shared SNP proportion of 98.74% between batch1 and batch2.1, 98.82% between batch1 and batch3, and 98.90% between batch2.1 and batch3 (Supplemental Table S4).

To further confirm the genetic identity of the batches, identity-by-descent analysis was performed using PLINK v1.9, showing a proportion of identical alleles of approximately 0.95 in all pairwise comparisons. Relationship inference using KING v2.2.6 supported these findings, with kinship coefficient ranging from 0.4845 to 0.4859 across all pairs, consistent with all three batches originating from the same individual (Supplemental Table S4).

Cell culture

The EBV-transformed lymphoblastoid N. siki cell lines were cultured in RPMI medium supplemented with 16% FBS, 1% L-glutamine, and 1% penicillin–streptomycin in a 5% CO₂ humidified incubator at 37°C. At Uniba, growth was monitored daily by means of inverted microscopy, and cells were split 1:2 every 2 days. At UTAD, cells were cultured for two passages following the same procedure.

Strand sequencing and inversion identification

Strand-seq libraries were obtained from the cell culture following the protocol described by Sanders and colleagues (Falconer et al. 2012; Sanders et al. 2017) and sequenced on NextSeq 500 (MID-mode, 75 bp paired-end protocol). We demultiplexed according to the barcodes and converted raw data to FASTQ using bcl2fastq v2.19.0.316. Reads were then aligned to the gibbon reference genome (GGSC Nleu3.0/nomLeu3) with BWA-MEM (version 0.7.17) (Li 2013). The obtained SAM files were converted to BAM files and sorted with SAMtools view and SAMtools sort (version 0.7.17) (Li et al. 2009), and duplicate reads were marked using Picard MarkDuplicates (version 2.20.4-SNAPSHOT; https://broadinstitute.github.io/picard/). Strand-seq reads for each cell were counted and plotted with mosaicatcher v.0.3.1 (Weber et al. 2023) to assess the sequencing quality. Cells that had low coverage, had incomplete BrdU incorporation, or had already started the second round of BrdU incorporation were excluded during the manual quality evaluation.

To detect germline inversions, we merged single-cell data in a directional pseudobulk file, called a composite file, as previously described (Maggiolini et al. 2020; Mercuri et al. 2022). To do so, we used an in-house R script to identify all the chromosomes inherited in a CC (Crick; “+”) or WW (Watson; “–”) state among all the cells and merged those bulks into two files with SAMtools merge (Li et al. 2009). Then, we filtered both BAM files by SAMtools view (Li et al. 2009) to keep only the first read of each proper pair with a mapping quality greater than 10 and reverse-complemented all the reads of the WW chromosomes by converting the WW BAM file to BED file (BEDTools bamtobed), manually changing the mapping strand, and converting the BED file back to BAM file (BedToBam) (Quinlan and Hall 2010). The latter BAM file was merged with the one containing all the reads of the chromosomes inherited in a CC state with SAMtools merge (Li et al. 2009). The obtained composite BAM file was then sorted and indexed by using SAMtools sort and index (Li et al. 2009) and used as input for a breakpointR run (Porubsky et al. 2020), intending to identify inversion breakpoints and obtain a BED file loadable into the UCSC Genome Browser (Kent et al. 2002; Raney et al. 2024). Circular plots of the detected rearrangements in composite files were generated using the ggbio (v1.52.0) package (Yin et al. 2012) in R (v4.4.1) (R Core Team 2021).

SNP analysis

Genomic DNA was extracted from the cell cultures using FastPure blood/cell/tissue/bacteria dna isolation mini kit (Vazyme) and sequenced using NovaSeq X with a coverage of 30×.

Reads were then aligned to the gibbon reference genome (GGSC Nleu3.0/nomLeu3) with BWA-MEM (version 0.7.17) (Li 2013). The obtained SAM files were converted to BAM files and sorted with SAMtools view and SAMtools sort (version 0.7.17) (Li et al. 2009), and duplicate reads were marked using Picard MarkDuplicates (version 2.20.4-SNAPSHOT; https://broadinstitute.github.io/picard/). Germline short variants were called using GATK best practices workflows (Van der Auwera et al. 2013). We kept only SNPs with a DP between 10 and 75 and a GQ greater than 20. We then compared the state of all the retained SNPs in the sequenced samples using an R custom script.

Identity-by-descent analysis was performed using the option –genome from PLINK v1.9 (Chang et al. 2015) and the relationship inference using KING v2.2.6 (Manichaikul et al. 2010).

FISH

Three-color FISH experiments were conducted using human (n = 4) and gibbon BAC clones (n = 10) as detailed in Supplemental Table S2. The clones were directly labeled by nick-translation with Cy3-dUTP (PerkinElmer), Cy5-dUTP (PerkinElmer), and fluorescein-dUTP (Enzo), following the method described by Lichter et al. (1990) with minor adjustments. Hybridization took place at 37°C in a solution containing 2 × SSC, 50% formamide, 10% dextran sulfate, and 3 mg of sonicated salmon sperm DNA, totaling 10 mL. Posthybridization washes were performed at 60°C in 0.1 × SSC (three times for high stringency hybridizations using species-specific clones) or at 42°C in 2 × SSC (four times each for cross-species hybridizations). Nuclei and metaphases were stained with DAPI. Digital images were captured using a Leica DMRXA2 epifluorescence microscope with a cooled CCD camera (Princeton Instruments). Fluorescence signals from DAPI, Cy3, Cy5, and fluorescein were detected with specific filters, recorded as grayscale images, and later pseudocolored and merged using Adobe Photoshop. Interphase nuclei FISH involved three-color FISH with two probes within the predicted inversion and one reference probe outside, whereas metaphase chromosome FISH used two probes within the inverted region.

Enrichment analysis

Gibbon genes located within the 17 Mb inversion on Chromosome 22, as well as on the p- and q-arms of Chromosome 14, were extracted from the curated subset of the RefSeq track in the UCSC Genome Browser for the GGSC Nleu3.0/nomLeu3 reference genome (Pruitt et al. 2014). Human orthologs of these genes were analyzed using the GENE2FUNC module of FUMA, which annotates genes in a biological context by mapping associated diseases and known drug targets based on OMIM and DrugBank identifiers, assesses tissue-specific expression patterns using GTEx v6 RNA-seq data, and evaluates functional enrichment of gene sets using a hypergeometric test against curated biological pathways and functional categories derived from MSigDB and WikiPathways (Watanabe et al. 2017).