Discussion
To date, five patients have been described to carry homozygousTDP2 variants. A homozygous splice-site variant in TDP2(c.425+1G>A) has been reported in three Irish siblings and in an additional patient from the United States who shared the same haplotype, suggesting a founder effect (Gómez-Herreros et al., 2014; Zagnoli-Vieira et al., 2018). Recently, a homozygous nonsense variant in the exon 3 of TDP2(c.400C>T; p.Arg134Ter) has been detected in a 11-year-old Italian girl (Ciaccio et al., 2019).
In this study we identified a novel homozygous spice-site variant disrupting the correct splicing of TDP2 exon 5. By using immunofluorescent detection of γH2AX as an indirect measure of DSBs, we demonstrated that the novel TDP2 c.636+3_636+6del splice-site variant (as the c.425+1G>A) profoundly impacts on the nuclear DSB repair after treatment with etoposide. Although the functional effect of the c.400C>T (p.Arg134Ter) variant was not established, it is expected to lead to nonsense-mediated mRNA decay, as in the case of the c.425+1G>A and the c.636+3_636+6del splice-site variants.
The five previous and our current patients share several clinical features, mainly craniofacial and neurological (Table 1). Particularly, 5 out of 7 patients with TDP2 inactivation showed speech impairment, which may be considered a recurrent feature in SCAR23 along with drug-resistant epileptic seizures and intellectual disability – shared by all described cases, progressive ataxia (6/7), and craniofacial dysmorphisms (6/7) (Table 1). We previously observed neutropenia in a 6-year-old child carrying the c.425+1G>A variant (Zagnoli-Vieira et al., 2018). Interestingly, TDP2 is highly expressed in the hemolymphoid system and interacts with CD40, a member of the tumour necrosis factor (TNF) receptor family that plays a critical role in several immunological processes (Pype et al., 2000), and mutations of CD40 gene have been associated to an autosomal recessive form of immunodeficiency with hyper IgM and neutropenia (OMIM #606843; Ferrari et al., 2001). Nevertheless, haematological manifestations have not been reported in other patients and do not seem to be a pathognomonic signature for SCAR23, at least based on current observations in patients.
Our patients manifested eye anomalies, particularly early-onset bilateral cataracts in the proposita, a feature never described in SCAR23. The extended in silico panel for cataract (Table S2) was negative, a finding against a potential multilocus inheritance. Interestingly, the exon 5 of TDP2 is predicted to contain a 21-bp transcription factor binding site for PAX6 (chr6:24,654,475-24,654,495, GRCh38/hg38; V$PAX6_01, Transfac Matrix Database v.7.0), whose deficiency is associated with distinct ocular phenotypes including cataract (OMIM #106210). Although more functional studies are needed to support this hypothesis, TDP2 could be involved in the gene regulatory circuitry of PAX6/SOX2 that orchestrates lens development and neurogenesis.
TDP2 has been recognized as a key player for neural homeostasis because of its role in preventing DNA damage led by abortive TOP2 activity. Curiously, biallelic inactivation of TDP1 causes spinocerebellar ataxia with axonal neuropathy (SCAN1; OMIM #607250) (Takashima et al., 2002), further supporting the central role of topoisomerases and their partner tyrosyl-DNA phosphodiesterases in neurodevelopment. Moreover, Stoll et al. (2013) found a significant enrichment of a rare 240-kb chromosome 22q11.22 microdeletion spanning TOP3B (encoding the DNA topoisomerase IIIβ) in Finnish patients affected by schizophrenia and cognitive impairment, whereas mutation of TOP1(encoding the DNA topoisomerase I) has been implicated in autism spectrum disorders (King et al., 2013; McKinnon, 2016). Interestingly, a small inherited duplication in 6p22.3 (37.89 kb) joining the first noncoding exon of KIAA0319with the 3′ end of the neighbouring gene TDP2 (exons 4–7) was identified in an Irish patient with autism spectrum disorder (ASD) (Holt et al., 2012). In Decipher, another inherited heterozygous duplication of KIAA0319 -TDP2 with similar size (33.43 kb) was detected in a female patient (#281899) with delayed speech and infantile spasms. Although several studies implicated KIAA0319 in dyslexia susceptibility (Francks et al., 2004; Cope et al., 2005; Harold et al., 2006; Paracchini et al., 2006; Dennis et al., 2009; Eicher et al., 2014), a condition frequently associated with ASD, we cannot rule out the involvement of TDP2 , a gene that “keeps the brain healthy” (McKinnon, 2014), as susceptibility factor in the development of schizophrenia when combined with other genetic and environmental factors. In this regard, we detected the c.636+3_636+6del variant at the heterozygous state in the DNA of the probands’ cousin suffering from schizophrenia. Although the variant was inherited from the father, with no apparent neurobehavioral abnormalities, a possible involvement ofTDP2 may not be completely ruled out and merits further investigations.
Apart from their role in normal brain function, topoisomerases, when altered and mis-expressed, may play a critical role in chromosome instability (CIN), a hallmark of carcinogenesis (Liu et al., 2019). Clinically, a large arsenal of topoisomerase inhibitors has been used in personalized cancer therapy to suppress DNA replication of tumour cells (Chen et al., 2015). Several studies showed that TDP2-dependent repair of DSBs protects cells from genome instability and chromosome translocations (Gómez-Herreros et al., 2013, 2017; Canela et al., 2019). Moreover, TDP2 has been suggested as a promising target for anticancer therapy, further underlying its possible role in carcinogenesis. According to the COSMIC database, 94 different variants in TDP2 , mostly missense substitutions (76.7%), have been detected in 103 out of 35,902 tumour samples (0.28%) tested from different anatomical regions, with an overall enrichment for the large intestine (28/2331, 1.2%). These data raise the question of whether SCAR23 patients and TDP2mutation carriers may be more prone to develop tumours and need specific oncological follow-up. Although more observations are needed, we excluded the presence of a second coding hit in the tumour sample of the patients’ mother, weakening a possible involvement of TDP2 as a tumour suppressor gene in colorectal carcinogenesis. In this regard, it should be also considered that other redundant mechanisms are available to repair TOP2-induced DSBs, such as the Mre11-dependent homologous recombination (Hoa et al., 2016).
In summary, our results strongly suggest that biallelic loss-of-function variants in TDP2 are associated with SCAR23. Whether the accumulation of other classes of variants, such as missense substitutions in crucial TDP2 functional domains, especially the C-terminal exonuclease-endonuclease-phosphatase (EEP) catalytic domain (aa 117-351), might be associated with distinct phenotypes or degrees of severity needs further investigations.