The propositus was diagnosed with FA at the age of 7 based on the positive chromosomal DEB test. The subject presented mild and isolated thrombocytopenia since 1.5 years of age, and at the time of diagnosis, his peripheral blood (PB) smear revealed white blood cells (WBCs) 6.7 × 10⁹/l, neutrophils 2.5 × 10⁹/l, hemoglobin (Hb) 13.3 g/dL, mean corpuscular volume (MCV) 95 fL, and platelets 113 × 10⁹/l (Table 1). Hematopoietic cellularity on BM trephine was 40%, and no dysplastic features were visible on BM cellular morphology. Moreover, chromosomal analysis showed a normal karyotype (46, XY) of BM cells. P1 was in the 10th percentile for height and weight, and was investigated through abdominal ultrasound, echocardiography, thyroid ultrasound, brain magnetic resonance imaging, audiometry, and spirometry, without appreciating any malformation or abnormality. Both bone age and density scan were in the normal range as well.

The patient entered the monitoring plan in use at the IRCCS Istituto “G. Gaslini” (Genoa, Italy). Over follow-up, he showed a moderate decline of the overall blood count, especially affecting the platelet population. Currently, 5 years after diagnosis and at the age of 11, WBCs are 4.6 × 10⁹/l, neutrophils 1.2 × 10⁹/l, Hb 12.3 g/dL, MCV 103 fL, and platelets 50 × 10⁹/l (Table 1). Hematopoietic cellularity on BM trephine is 25%–28%, and BM cells still have a normal karyotype without dysplastic signs. Based on this finding, although hematopoietic stem cell (HSC) donor search was initiated either in the family or in the international registries, HSC has not been planned yet. The research was approved by the Ethics Review Boards at the institutions that enrolled the patients. According to the Declaration of Helsinki, all the family members or their legal guardians signed written informed consent for the analysis.

P1’s lymphoblastoid cell lines (LCLs) were generated from primary lymphocytes isolated from peripheral blood (Bottega et al., 2018). LCLs and HEK293 cells were grown in RPMI 1640 1X (L-glutamine) and DMEM 1X (Euroclone, Pero, IT) media, respectively, with 10% fetal bovine serum (FBS) and 1% penicillin–streptomycin at 37°C in a humidified atmosphere containing 5% CO₂. Following DNA, RNA, or protein extraction, cell cultures were transferred to RPMI or DMEM with 10% DMSO and then frozen in liquid nitrogen for storage.

Genomic DNA was extracted from patient’s peripheral blood and LCLs. FA genes were analyzed using the Ion PGM system for next-generation sequencing (Life Technologies, Carlsbad, CA), as described in De Rocco et al. (2014). FANCA deletion was identified analyzing Ion PGM row data, as described in Nicchia et al. (2015). For Sanger sequencing, PCR was carried out using the KAPA2G Fast HotStart ReadyMix (KAPA Biosystems, Wilmington, MA) and the primer pair 28F 5′-GTT​GAT​GGT​CTG​TTT​CCA​CC-3′ and 28R 5′-ACC​CTA​GAC​TCG​AGA​CGA-3′ for the detection of c.2778 + 83C > G and c.2778 + 112T > C variants. PCR products were purified with ExoSAP-IT (Applied Biosystems, Foster City, CA) and sequenced using the ABI PRISM sequencer (Applied Biosystems, Foster City, CA). Nucleotide numbering reflects the FANCA cDNA with +1 corresponding to the A of the ATG translation initiation codon in the reference sequence (RefSeq NM_000135). Variants identified were searched in the following annotation databases: Single-Nucleotide Polymorphism Database (dbSNP; http://www.hgmd.cf.ac.uk/ac/index.php), Genome Aggregation Database (gnomAD; https://gnomad.broadinstitute.org), and Human Gene Mutation Database (HGMD; http://www.hgmd.cf.ac.uk/ac/index.php). The in silico analyses of splicing mutations were carried out using Splice Site Prediction by Neural Network (NNSplice; http://www.fruitfly.org/seq_tools/splice.html).

Total RNA was extracted from the patient’s LCLs and peripheral blood using the High Pure RNA Isolation Kit (Roche, Basel, CH) and PAXgene Blood RNA Kit IVD (QIAGEN, Hilden, DE), respectively, according to the manufacturer’s instructions. cDNA was synthetized using the Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, CH), and RT-PCRs were prepared with KAPA2G Fast HotStart (KAPA Biosystems, Wilmington, MA). To verify the deletion from exons 11 to 14, RT-PCR primers 9F 5′-TTG​ATG​TAC​TGC​AGA​GAA​TGC-3′ and 16R 5′-GTG​TCT​TGG​CCA​ATG​AGA​TG-3’ were used. To assess the c.2778 + 83C>G effect, RT-PCR primers 27F 5′-AGC​TGC​TTA​TCT​CCA​GGC​C-3, 28R_+83 5′-CAA​ACC​CTA​GAC​TCA​GGA​CG-3,′ and 30R 5′-GGA​AAT​CCA​TCA​GTG​CGT​TG-3′3′ were designed within the predicted intronic retention and used in Savino et al. (1997) to characterize the variant. RT-PCR amplicon extraction from 3% agarose gel was performed using the QIAquick Gel Extraction Kit (QIAGEN, Hilden, DE) in line with the manufacturer’s instructions. To increase the performance and purify the extraction, the products obtained were further subjected to classical PCR with the same primer pair used for RT-PCR. All RT-PCR amplicons were purified and sequenced, as reported previously.

A 533-bp cassette that contains FANCA exon 28 with portions of introns 27 and 28 was created by the amplification of cDNA obtained by patients’ PB or LCLs in order to obtain all the investigated haplotypes. Amplification primers (clonF 5′-ATCATCCATATGGAATGTGGGGTTGTG-3′ and clonR 5′-TGTTGTCATATGGTCAAGATTCCAATC-3′) were designed with an NdeI restriction site (underlined) at the end in order to allow the cloning into the α-globin minigene under the control of the α-globin promoter and the SV40 enhancer (Mattioli et al., 2014).

HEK293 cells seeded in a six-well plate were transfected with 3 ug of each minigene plasmid using Lipofectamine 3000 (Thermo Fisher Scientific, Waltham, MA).

For the analysis of the splicing isoforms, the primers α2F (5′-CTT​CAA​GCT​CCT​AAG​CCA​CTG-3′) and β2R (5′-CAC​CAG​GAA​GTT​GGT​TAA​ATC-3′) were used.

Protein whole-cell extracts were prepared using RIPA buffer (Upstate Biotechnology, Lake Placid, NY). In brief, the cells were washed with phosphate-buffered saline and resuspended in RIPA buffer containing protease and phosphatase inhibitors (Roche, Basel, CH) and benzonase (Novagen, Madison, WI) at 25 U/mL. The samples were lysed and quantified with the Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA), according to the manufacturer’s instructions. Protein extracts were then loaded on 6% SDS-PAGE gel and blotted with the following antibodies: anti-FANCA (rabbit, A301-980A, Bethyl, 1:2000), anti-FANCD2 (rabbit, Abcam, ab2187; 1:2000), anti-vinculin (rabbit, Abcam, ab155120; 1:5000) as the loading control, and anti-rabbit (goat, Bethyl, A120-101P, 1:2500). Immunodetection was operated with Immobilon Crescendo Western HRP Substrate (Merck Millipore, Burlington, MA) and Pierce ECL Western blotting Substrate (Thermo Fisher Scientific, Waltham, MA) to reveal FANCA, and FANCD2 and vinculin proteins, respectively. The densitometric analysis of the anti-FANCA antibody signal was performed via ImageJ software, based on the use of vinculin for data normalization and the comparison with both positive and negative controls’ results. To analyze FANCD2 monoubiquitination, the cells were treated with 2 mM hydroxyurea (HU) (Sigma-Aldrich, St. Louis, MO) for 18 h prior to sample collection.

LCL survival under MMC treatment (0, 3, 10, 33, 100, and 333 nM) was tested after a few cell passages (Time 0) and a month after the establishment of the cultures (Time 1), as previously described in Bottega et al. (2019).

Targeted next-generation sequencing (t-NGS) of patient’s DNA from peripheral blood (PB) revealed two single-nucleotide heterozygous substitutions (c.2778 + 83C > G and c.2778 + 112T > C) in intron 28 of the FANCA gene, both confirmed in the proband (II-1, P1) and his mother (I-2) by Sanger sequencing, indicating they are in cis (Figures 1A, B). The c.2778 + 83C > G transversion is reported in GnomAD (“uncertain significance”, MAF: 0.000004012) and in HGMD (“disease causing”) and, according to NNSplice software, predicted to generate a strong cryptic donor splice site (GT) at position 84–85 of FANCA intron 28, favoring its recognition over the physiological splicing site (SS) (Table 2). The 2778 + 112T > C variant is annotated in GnomAD (MAF: 0.0002739) and dbSNP (rs140073727), and is likely not to affect splicing, not even when combined with c.2778 + 83C > G (Table 2). On the other allele, NGS data allowed us to postulate an intragenic deletion of FANCA of paternal origin encompassing exons 11–14 (Figure 1A), and subsequently confirmed by MLPA analysis (data not shown) and RT-PCR on total PB RNA (Figure 1C, D).

To determine the c.2778 + 83C > G splicing effect, we performed RT-PCR on the total RNA from a LCL established from the proband (P1) (Figure 2A, B). As a control, we used the LCL RNA of another FANCA patient (P2), compound heterozygote for a large intragenic deletion (exons 11–31 deletion; data not shown) and the same c.2778 + 83C > G mutation (Savino et al., 1997), leading to an alternative transcript with the retention of the first 83 nucleotides of intron 28 (Figure 2A; Table 2). In line with the literature (Savino et al., 1997), in P2, we observed four bands. The fastest of 403 bp represents the wild-type band, suggesting the production of a small amount of normal transcript. The product of 486 bp (403 + 83) is due to retention of the first 83 nucleotides of intron 28. The remaining two are spliced products (521 bp and 604 bp) with an insertion of 118 bp (alternatively processed exon 29a with an unknown biological effect) between exons 29 and 30 (Figure 2B).

In P1, instead, we revealed only two bands corresponding to the WT amplicon (403 bp) and a novel faint product (559 bp) not detectable in P2 and in the control (Figure 2B), later identified by Sanger sequencing as another alternative spliced form of FANCA with 156 additional bp (another alternatively cleaved exon 28a with uncertain biological significance) between exons 28 and 29. Of note, the analysis did not reveal any product with the partial intronic retention. To further confirm these data, we carried out a RT-PCR using a reverse primer specifically designed to anneal over the 83-bp insertion of intron 28 (Figure 2C). As in the previous analysis, the expected amplicon (263 bp) was detected in only P2’s sample (Figure 2C), suggesting a different c.2778 + 83C > G splicing impact on P1.

In trying to elucidate whether any mutational event could have restored the physiological splicing pattern in P1, we sequenced exon 28 and its flanking regions from the genomic DNA of both patients’ cultured cells. However, in P2, we ascertained the c.2778 + 83C>G mutation and the 2778 + 112T>C variant, both in an homozygous condition, and therefore, on the same haplotype, in P1, we unveiled a third alteration, c.2778 + 86insT, a de novo thymine insertion (Figure 2D) not found in the genomic DNA from neither the patient’s PB and BM cells nor from his parents’ PB (data not shown). We confirmed the presence of c.2778 + 86insT also in a second P1’s LCL established 2 years apart (data not shown), suggesting its onset in vivo.

According to NNSplice software, the combination of c.2778 + 83C>G and c.2778 + 86insT on the same allele was predicted to decrease the strength of the cryptic splice site at the intronic position 83 (Table 2). Thus, to further investigate the contribution of c.2778 + 83C>, c.2778 + 112T>C, and c.2778 + 86insT variants on pre-mRNA processing, we generated a pTBNdeI hybrid minigene, where we subcloned FANCA exon 28 with adjacent portions of intron 27 and the WT or mutagenized forms of intron 28 (Pagani et al., 2003) (Figure 3). Owing to the amplification of the genomic DNA from patients’ PB and/or LCL for the creation of the FANCA cassette, the analysis allowed us to ascertain the cis arrangement of c.2778 + 83C>G and c.2778 + 86insT in P1, as determined by Sanger sequencing on the subcloned fragments. We also examined the effect of c.2778 + 40G>T, an unreported, apparently neutral de novo variant, found in heterozygosis in a FANCA patient (P3) carrying c.2778 + 83C>G and c.2778 + 112T>C in the homozygous status (data not published). Minigenes expressing the four different haplotypes of intron 28 (Figure 3; Table 2) were transiently transfected in HEK293 cells, and the pattern of splicing was analyzed by RT-PCR amplification with specific primers. Transfections confirmed the expression of two isoforms corresponding to the transcripts generated by the competition of the physiological (TCACgtaggt; 507 bp) and cryptic (gcacgtaggt; 590 bp) donor splice sites at different extents. Indeed, the wild-type GCTT haplotype was mainly associated with normal splice processing, with intron 28 removal in almost 90% of transcripts (Figure 3). On the contrary, analysis of the GGTC (P1’s PB and P2’s LCL) and TGTC (P3’s LCL, differing at position c.2778 + 40) mutant haplotypes showed a significant increase (almost 70%) in the splicing isoform with the 83-bp intronic retention due to the strong recognition of the cryptic splice site (gcaGgtaggt). The identical pattern shared by the two aberrant haplotypes GGTC and TGTC indicated that the c.2778 + 40G > T variant did not affect the alternative splicing process. Interestingly, the GGTTC haplotype (P1’s LCL), carrying c.2778 + 86insT, restored almost completely the splicing of intron 28, providing an explanation to the RT-PCR data observed in P1’s LCL (Figure 2C).

In order to confirm the compensatory effect of c.2778 + 86insT, we compared P1 and P2’s LCL cellular phenotypes. Since both the affected individuals were compound heterozygous for c.2773 + 83C > G and a large FANCA deletion of exons 11–14 in P1 and exons 11–31 in P2, only the allele with the intronic mutation was expected to contribute to any protein production. We detected FANCA protein in both P1 and P2’s LCLs, albeit at lower levels in comparison to the WT control (Figure 4A). Indeed, the densitometric analysis of immunoblot bands allowed us to assess a reduction to 38.6% and a relevant decrease to 15.4% of protein expression in P1 and P2, respectively (Figure 4A). These findings were consistent with the role of c.2778 + 86insT in restoring, even partially, the activity of the physiological donor splice site.

Subsequently, we evaluated whether the expression level of FANCA could guarantee FANCD2 monoubiquitination. We observed detectable levels of monoubiquitinated protein in both untreated and treated cells with HU in P1 but not in P2 (Figure 4A), suggesting that about 40% of FANCA was enough to preserve this post-translational modification. Conversely, when FANCA expression was significantly reduced as in P2 (Figure 4A), no FANCD2 monoubiquitination was detected.

Assuming that P1’s cells used to establish the LCL were a mosaic for the c.2778 + 86insT variant and those carrying it could have some proliferative advantages, we eventually examined the proficiency of patients’ LCLs to survive different concentrations of MMC (3–100 nM) (Figure 4B). We, thus, thawed one of the original vials stored after immortalization and carried out the test after a few cell passages (Time 0) and 1 month of culture (Time 1). P2’s cells showed a MMC sensitivity comparable to that of the FANCA−/− control at both time points. On the contrary, P1’s LCL had a survival rate intermediate between that of WT and FANCA−/− cells at Time 0, and a completely restored phenotype at Time 1 (Figure 4B), further corroborating the compensatory role of c.2778 + 86insT.