Mutations in eight CS patients
Essentially two genes have been associated with CS, namely ERCC6 in 68% and ERCC8 in 32% of patients (6). The situation is possibly reversed in Tunisia and Arab countries, where ERCC8 mutations seem to be more frequent (5, 13, 14, 17, 18). The present study expands the clinical spectrum and increases the relevance of two mutations in the CSA subtype. These genetic defects seem to be specific to the Tunisian and North African population, as they have not been reported elsewhere, at least to date. Indeed, since the first description of CS by Dr. Cockayne in 1936, only eight patients have been reported in the Tunisian population: two siblings have one of the mutation described here (c.598_600delinsAA) (12) (13), two other siblings have a private mutation (c.400-2A > G) in ERCC8 (18), and four more CS patients have been clinically and biochemically characterized but the mutations have not been identified (19, 20) .
In six patients of our Cohort, Sanger sequencing identified a recurrent ERCC8 mutation, namely the homozygous mutation c. 598_600delinsAA p.Tyr200Lysfs*12, which was previously identified in two Tunisian siblings (12, 13). Indels are the second most common class of mutation in the human genome (21), and often involve domains with repetitive sequences (22). ERCC8 encodes a 44 kDa protein, CSA that contains 7 WD40 domains each of which is constituted by several WD repeats [tryptophan (Trp, W), aspartic acid (Asp, D)]. The c. 598_600delinsAA variation in ERCC8 patients could lead to a nonsense-mediated mRNA decay (NMD). In detail, the alteration of the fourth evolutionarily conserved amino-acid residue in the WD4 repeated motif may result in a premature stop codon after 12 aminoacids (AA). The WD motifs are required for the construction of the beta-propeller structure, which is important for protein complex formation and interactions of CSA with transcription and repair factors DDB1, RNA polymerase II, TFIIH (10, 23).
The relatively larger proportion of ERCC8 defects, and in particular the c.598_600delinsAA mutation, in Tunisian patients can be attributed to a founder effect. Further investigations including haplotype analysis are required to verify whether this is the case. Interestingly, one of the six patients has Algerian ancestries suggesting that this variation is a founder mutation in North Africa.
Furthermore, via targeted gene sequencing, we detected in two patients (CSEA1 and CS1EA2) a variation that has not been previously reported in the Tunisian population, i.e. c.843 + 1G > C. This homozygous mutation leads to the abolition of the consensus donor splice site in intron 9, generating a novel splice event, and leading to exon 9 skipping in the ERCC8 gene and the emergence of a premature stop codon. The donor splice mutation detected in patients from the CS1 family leads to a shorter protein lacking the last two WD40 domains, which may affect the function of this protein. This variant co-segregated in the CS1 family members, an additional argument indicating that this variant is causal of the CS disorder in these patients.
The c.843 + 1G > C variation has been already described in a CS patient from Lebanon (14) but the conclusions on the consequence of this variant on the transcript differ in our study. Indeed, Chelby et al suggested that intron 9 was present, because PCR failed to amplify a fragment that contained this exon. However, in our understanding the primers used in this PCR were located exactly in exon 9. In this case the reason for lack of amplification was the absence of this exon, in agreement with our findings. Moreover, the presence of intron 9 was not further demonstrated. In addition, the amplification obtained with primers 1 englobing exon 9 to intron 9 could be due to contaminating DNA acting as a competitor in the PCR reaction (24), if samples were not treated with DNase before RT-PCR, as we did. Moreover, this transcript could be poorly expressed and therefore was not detected in the previous study. This mutation ultimately results in the same consequences as the c.843 + 2T > G and c.843 + 5G > C variations that have been also suggested to alter donor splice site and lead to a premature stop codon p.Ala240Glyfs*8 (12, 25) .
Remarkable clinical features and lack of clinical photosensitivity
Each of the reported cases in the present study displays each distinct clinical features. It is worth to note that some patients (CS1EA1, CS11 and CS16) suffered from intra-uterine growth retardation. This clinical feature is more frequently associated with the severe form of CS-type II, which is usually linked to mutation in ERCC6. Conversely, in our study all patients had the CS-A form (mutation of ERCC8), which is normally less frequent (14% of cases) (16, 26). Other clinical manifestations as microcephaly and ataxia at birth are not unique for CS and have been described also in mitochondria-associated diseases, which makes the CS diagnosis more difficult at early stages.
Previous studies reported CS patients that do not present clinical photosensitivity, as in Tunisian, Turkish, Italian, and Moroccan populations (5, 19, 27, 28). Therefore, cutaneous photosensitivity was classified as a minor criterion in the diagnosis of CS that appears only in about 75% of patients and was not correlated with the type of genetic defect in the TCR-NER pathway. Our data, with the two siblings from the CS1 family (mutation c.843 + 1G > C), as well as the CS11 patient mutation c.598_600delinsAA not displaying clinical photosensitivity confirm that this defect is not an essential criterium for CS. The absence of clinical photosensitivity required assessing the repair of UV-induced DNA damage by TC-NER in primary fibroblasts from these patients. Indeed, since 1977 it has been shown that fibroblasts from CS patients had increased sensitivity to UV irradiation (29). Conventional methods to assess TC-NER include RRS following UV damage that is impaired in CS (30), and unscheduled DNA synthesis that is not affected in these patients whereas it is in XP patients (31). When clinical photosensitivity is identified in CS, it remains very moderate compared to other forms of genodermatosis related to defects of the NER system.
Comparison between conventional mild phenotype CS patients and CS patients who did not show photosensitivity displayed no difference in RRS, which was low in both groups as compared to controls. This indicates that photosensitivity, even if not clinically visible, is present at the cellular level in CS. RRS following UV damage remains useful to confirm the diagnostic and is complementary to genetic investigations.
This further substantiate that Cockayne syndrome may not be accounted to the defective NER system alone. Indeed, variations in ERCC6 and ERCC8 genes were also associated with the UV sensitive syndrome (UVSS), a milder form clinically characterized by mild cutaneous symptoms. In UVSS patients, reduced RRS after UV radiations was also observed, indicating that despite TC-NER was impaired this defect did not lead to neurodegeneration or premature ageing typical of CS.
Lack of association between CS and clinical photosensitivity in some patients suggests that other or additional mechanisms than the DNA repair defect are involved in the etiology of CS. In this context, CS exhibit altered mitochondrial metabolism and an accumulation of oxidative stress at the cellular level (32, 33). CSA and CSB are indeed multifunctional proteins that are involved in several processes in addition to DNA repair (34, 35).
Heterogeneous clinical features in patients with the same mutation and siblings
CS is a clinically heterogeneous disease and is caused by a large number of distinct mutations in ERCC6 or ERCC8 (5, 36). For comparison, other monogenic diseases, for instance the Hutchinson-Guilford progeria syndrome (HGPS) is due to a single point mutation that blocks the physiological processing of the Laminin A protein (37). Conversely, 38 pathogenic variants have been described just for ERCC8/CSA and which concern totally 84 CS patients (36). Since genotype/phenotype correlation remains elusive, relevant information may originate from the study and the assessment of their clinical symptoms in multiple patients and, when available, siblings carrying the same mutation. However, this situation is rather infrequent and only three other cases of siblings as well as patients carrying the same mutations (13, 38) have been described in CS. The present study that reports a detailed clinical characterization of six patients, including two siblings that carry the same mutation as well as two siblings carrying a second mutation, represents a powerful data set to address this question.
The six patients carrying the c.598_600delinsAA mutation shared common characteristics: early age symptoms [0–24 months], prenatal abnormalities as microcephaly, cerebellar hypoplasia, olighydramnios, and lower post-natal weight and height. They also displayed different combinations (presence/absence) of other defects like normal or low birth weight and height, ataxia, cataracts, dental abnormalities, hypomyelination, cerebellar atrophy, etc. Importantly, within this group the two CS6 siblings displayed remarkable phenotypic differences, like post-natal height, independent walking, dental abnormalities, and cryptorchidism.
Two siblings from the CS1 family (mutation c.843 + 1G > C) presented high levels of transaminase which are commonly observed in other CS patients reflecting a possible mild liver damage (4, 39). Moreover, the younger of the two patients displayed severe symptoms like the emergence of cataracts at an early age. Indeed, the presence of cataracts is normally associated with a worst probability of survival and death before the age of 7 for CS patients (40). Only one of the two siblings (CS1EA1, a male) showed prenatal microcephaly olighydramnios and cataracts. Conversely, only the other sibling (CS1EA2, a female) showed bird-like nose dysmorphism, limb spasticity, ataxia, hair and dental abnormalities, cerebellar atrophy. These clinical differences in the context of the same mutation and, in the case of siblings also of comparable genetic background, underscore the large heterogeneity of CS clinical symptoms that is difficult to reconcile with a simple genotype/phenotype alteration, and the reason of which remains obscure.
It is important to note that the clinical heterogeneity of patients that share the same recurrent mutation increases the difficulty for clinicians to confirm the clinical diagnosis of this disease, and may generate confusion with pathologies that display related symptoms like those linked to mitochondrial etiopathology such as mitochondrial cytopathies. Moreover, the clinical heterogeneity in CS may represent a further challenge for treatments, which have not been developed for CS to date.
Characteristics of the Tunisian cohort
We reported six patients with the same homozygous variation, including one of Algerian origin. This mutation was previously observed in two other Tunisian patients (41), which suggests that it is a founder mutation in the region. The CS6 siblings were born from a consanguineous marriage. Although the CS1 siblings were born from a non-consanguineous marriage, the emergence of the homozygous mutation, and thereby of CS, is likely due to the high rate of endogamy in this region. In Tunisia, the high rate of endogamy contributes to the increased risk (96.64%) of recessive diseases in isolated communities even without consanguinity (42).
The two siblings of the CS1 family harbor the same genetic variation as in a previously reported Lebanese patient, who also displayed a severe CS phenotype (14). North Africa's abundant prehistoric and historic cultural heritage has added to the diversity of the genetic pool of its population nowadays (43). This pool originates from a combination of Middle Eastern, Sub- Saharan Africa and Western European genetic components. For instance, the two Tunisian CS1 patients described here share a variant with the Lebanese patient born from Druze parents, possibly dating back to a common ancestry.