The First Description of Monozygotic Twin Females Discordant for the Craniofrontonasal Syndrome Phenotype and the Report of Four Novel Pathogenic Variants in the EFNB1 Gene

Background: Craniofrontonasal syndrome (CFNS) is a rare X-linked disorder that results from pathogenic variants in the EFNB1 gene. The syndrome paradoxically presents with greater severity of the symptoms in heterozygous females than hemizygous males. Results: Our primary �nding was the description of monozygotic twins, i.e., patients 5 & 6, discordant for the CFNS phenotype. Intriguingly, patient 5 presented classical CFNS gestalt, whereas patient 6 manifested only very subtle craniofacial features, not resembling CFNS. Besides, we have expanded the mutational spectrum of the EFNB1 gene through reporting four novel pathogenic variants – p.(Trp12*), p.(Cys64Phe), p.(Tyr73Metfs*86), p. (Glu210*). All those alterations were found applying either targeted NGS of a custom gene panel or PCR followed by Sanger sequencing and evaluated using in silico predictors. Lastly, we have also expanded the CFNS phenotypic spectrum by describing in patient 3 novel features of the syndrome, such as bi�d hallux, bicornuate uterus, and abnormal right ovary segmented into six parts Conclusions: We have described the unreported so far differences of the clinical phenotype in the monozygotic twin patients 5 & 6 harbouring an identical p.(Glu210*) variant located in the EFNB1 gene. With our �nding, we have pointed to an unusual phenomenon of mildly affected females with CFNS, who may not manifest features suggestive of the syndrome. Consequently, this study may be valuable for geneticists consulting patients with craniofacial disorders.

Wieacker and Wieland in 2005 explained the above paradox as a cellular interference, which assumes that due to a random X-inactivation, heterozygous females are uniquely mosaic and therefore have both functional and nonfunctional ephrin-B1, a protein which is encoded by the EFNB1 gene (Wieland et al., 2004;Wieacker and Wieland, 2005).These two ephrin-B1 forms' coexistence affects the adhesion and sorting of cells, disrupting normal embryological development (6, 7).Further reports describing more severely affected males, who all were mosaic for deleterious variants in the EFNB1, strengthen the hypothesis about the described pathomechanism's biological relevance (8).However, the precise molecular explanation for this phenomenon remains not yet fully understood (7).

Cohort description
We recruited four sporadic female individuals (patients 1-4) and one familial case consisting of two female individuals (patient 5 & 6), out of whom all but one, i.e. patient 6, presented phenotypic features suggestive of CFNS.

Clinical report
We recruited six female cases who, out of whom all but one, i.e. patient 6, presented with phenotypic characteristics suggestive for CFNS.The comparison of all clinical features noted in our cohort was outlined in Table 1, whereas the photographic documentation of facial dysmorphism and hand and foot anomalies was shown in Supplementary Fig. 1.Our study's primary nding was the description of monozygotic twins, i.e., patients 5 & 6, discordant for the CFNS phenotype.Patient 5 presented with classical CFNS dysmorphia such as corse facial feature, CS and syndactyly (Fig. 1a-d), whereas patient 6 manifested only very subtle craniofacial features not resembling CFNS (Fig. 1e).Patient 3 presented some novel for CFNS clinical features such as a bi d hallux (Supplementary Fig. 1d) bicornuate uterus, abnormal right ovary segmented into six parts by ve serpentine-like constrictions, with the largest ovary part of 1.5 cm, while the smallest one of 0.5 cm in diameter.To our best knowledge, this phenotype has not been reported in CFNS yet.She also showed congenital horizontal nystagmus, alternating divergent strabismus, defects of the genitourinary system, including horseshoe kidney, and not noted in CFNS thus far.

Patient 5
Patient 5 was a girl born in the 38 th week of gestation from 4 th pregnancy twined to unrelated healthy parents (Figure 1f).The pregnancy history was unknown because patient 5 was adopted.Her body mass was 2370 g (<3 rd percentile), length 49 cm (<3 rd percentile), Apgar score was 8-10-10 at 1',3' and 5'.She was referred for dysmorphic evaluation at 1 st month of age.She had a coarse face, plagiocephaly, CS, micrognathia, a small anterior fontanel, signi cant hypertelorism, bilateral epicanthal folds, bilateral low-set ears, at nasal bridge, anteverted nares, and a midline crease of the nasal tip.Brachydactyly, syndactyly of toes and longitudinal ridging of a nger-and toenails were also observed.On examination at the age of 5.5 months, she presented with a weight of 6110 g (<3 rd centile) and head circumference of 37.8 cm (<3 rd centile).

Patient 6
Patient 6 was a girl born in the 38 th week of gestation from 4 th pregnancy twined to unrelated healthy parents (Figure 1f).The pregnancy history was unknown because patient 6 was adopted.2a).The presence of both alterations was con rmed by Sanger sequencing.Patients 2-4 were screened before the advent of the NGS method.Thus the molecular diagnosis was achieved by Sanger sequencing on gDNA isolated from peripheral blood lymphocytes, which revealed the presence of the following three heterozygous alterations out of which two were novel -c.191G>T p.(Cys64Phe), c.216del p.
(Tyr73Metfs*86).In contrast, one variant has been previously reported c.451G>A p.(Gly151Ser) (HGMD no: CM041297) (Figure 2b).The family history of patient 5 showed that she has a twin sister who, despite the lack of typical CFNS symptoms, underwent targeted PCR and Sanger sequencing.We evaluated the pathogenicity of missense variants in silico applying multiple online prediction tools including Polyphen-2, SIFT, CADD, MutationTaster and other resources such as DANN, FATHMM-MKL, LRT, BayesDel addAF, BayesDel noAF, GERP, PhyloP100, PhastCons integrated into either VarSome online tool or Alamut ® Visual software product.The classi cation of all variants was performed following the American College of Medical Genetics and Genomics (ACMG) guidelines (Table 2).Applying SWISS-MODEL, we have visualized in 3D both wild type and mutated missense alterations in the ephrin-B1, i.e. p.(Cys64Phe) and p.(Gly151Ser) (13).

Zygosity analysis
The monozigosity of twin patient 5 & 6 was con rmed based on an analysis of 33 STR markers localized on 13, 18, 21, X and Y chromosomes.

Face2Gene analysis
The craniofacial phenotype of patient 6 was assessed using Face2Gene online available tool.Among the suggested 30 different syndromes, CFNS was not listed by the algorithm.However, the rst ve proposed diagnoses were as follows -Cornelia de Lange syndrome, Costello syndrome, Megalencephaly-Capillary Malformation-Polymicrogyria Syndrome, Alpha-Thalassemia/mental Retardation Syndrome and CHARGE syndrome.On the contrary, the phenotype of patient 5 was correctly identi ed as CFNS (listed as second).

Discussion
Although monozygotic twins originate from a single zygote and share the same genetic material and similar intrauterine environment, they occasionally may show discordant phenotypes of monozygotic disorder.The differences in clinical phenotype can be explained through at least several mechanisms such as epigenetic factors, an asymmetric split of the embryo, discordant cell differentiation or abnormalities in placental blood ow (14)(15)(16).Surprisingly, here we have diagnosed monozygotic twin patients, i. e. patient 5 & 6, who presented with highly variable severity of the CFNS symptoms.Both individuals carried the same p.(Glu210*) pathogenic EFNB1 variant and identical germline genetic information.In patient 5, we noted a typical female presentation of CFNS (Table 1, Fig. 1a-d).In contrast, in patient 6, we only detected mild facial anomalies unsuggestive for CFNS, including anteverted nares, depressed nasal bridge, low-set ears, coarse facial features, micrognathia and short neck (Table 1 and Fig. 1e).Besides, the craniofacial phenotype of patient 6 was analyzed using Face2Gene, which did not match CFNS among the possible dysmorphological diagnoses.
Mild clinical features in female individuals with CFNS are rather unusual.As mentioned before, CFNS inherits paradoxically and presents more severe clinical symptoms in females, who harbour the heterozygous EFNB1 variants in comparison to hemizygous males.Furthermore, rarely reported mosaic male individuals are more severely affected than their hemizygous counterparts.This is because other ephrin family members can presumably substitute the complete lack of ephrin-B1 in purely hemizygous males (3,7,17).In the medical literature, we have found merely one description of mildly affected CFNS female patient.Twigg et al. reported a familial case (family no.3217) heterozygous for a missense pathogenic variant p.(Pro54Leu), in which one of the affected females had minimal clinical manifestations of CFNS.However, this patient was shown to have a lower mutation level in the hair roots and buccal swab (2).In our case, we were unable to check for the mosaicism in mesoderm or ectoderm-derived cell lines, although the level for the causative variant in blood cells reached 50% of reads, being unsuggestive of somatic mosaicism.
Except for mosaicism in other than blood cells, one may suspect the presence of additional modi ers of the phenotype, including epigenetic factors (18-20).To check whether the variable severity of CFNS in both twin females resulted from skewed X chromosome inactivation, we performed XCI testing.We hypothesized that similar to male patients who show minimal CFNS symptoms, our mildly affected twin sister may have a highly preferential expression of the EFNB1 from a single gene copy, resembling its status in hemizygosity.To our surprise, we demonstrated unequal XCI in the severely affected twin patient 5 (84% vs. 16%), and almost random X inactivation in the mildly affected twin patient 6 (46% vs. 54%).Our nding, therefore, suggests that skewed X inactivation cannot account for the mild presentation of CFNS in one of our twin sisters and probably other mildly affected female individuals.Recently, another research group did not nd evidence for preferential XCI or a distinct correlation between XCI ratios in a group of familial X-linked hypohidrotic ectodermal dysplasia patients showing variable disease manifestation (21).Hence our result strengthens the above conclusion regarding the presence of additional yet undetected modifying factors resulting in discordant phenotype in X-linked disorders.
The second result of our study is the expansion of the EFNB1 gene mutational spectrum.Here, we have described three additional novel variants located in the EFNB1 gene -p.(Trp12*),p.(Tyr73Metfs*86), p.(Glu210*) and consequently increased the total number of CFNS-associated pathogenic variants to 123.All newly identi ed alterations were found applying either targeted NGS of a custom gene panel or PCR followed by Sanger sequencing.Subsequently, we evaluated the pathogenicity of the detected variants using in silico predictors (Table 2).We have also compared the phenotypic presentation of the six CFNS-affected individuals, all of whom were females (Table 1).Lastly, we have also broadened the phenotypic spectrum of CFNS syndrome, as we reported new features present in patient 3, such as a bi d hallux, bicornuate uterus and abnormal right ovary segmented into six parts.

Conclusions
First of all, we have pointed to an unusual phenomenon of mildly affected females with CFNS, who may not manifest features suggestive of the syndrome.As a consequence, this study may be valuable for clinical geneticists consulting patients with craniofacial disorders and who potentially may overlook such individuals.Second, we excluded skewed XCI pattern as a cause of discordant phenotype in monozygotic twins described here.Our study strengthens the recent conclusion regarding the presence of additional yet undetected modifying factors resulting in X-linked disorders' discordant phenotype.Finally, we have expanded the mutational spectrum of the EFNB1 gene by reporting three other novel pathogenic variants causing CFNS.

Methods
Targeted NGS We designed and applied the custom On-Demand AmpliSeq (ThermoFisher Scienti c) panel targeting 37 genes related to craniofacial disorders (9,10).We constructed the barcoded gDNA libraries according to the manufacturer's sample preparation protocol (Ion AmpliSeq Library Kit 2.0; On-Demand Panels) and subsequently sequenced them on the Ion Torrent S5 platform using the Ion 530™ or 540™ Chip.
PCR and Sanger sequencing PCR followed by Sanger sequencing was used to validate variants detected through targeted NGS (patient 1 & 5) and screen the coding sequence of the EFNB1 gene (patients 2-4).Besides, we performed targeted Sanger sequencing in the twin sister of patient 6 (targeted analysis of exon 4).We designed speci c primers (Supplementary Table 1) using Primer3 tool v. 0.4.0.The PCR reactions and PCR product puri cations were carried out following standard protocols.Next, Sanger sequencing was performed on an automated sequencer Applied Biosystems Prism 3700 DNA Analyzer using dye-terminator chemistry kit v.3, ABI 3130XL.Finally, the  3 of the left foot diagnosed in patient 5 (d).The family pedigree (f).The analysis of relatices was not possible because twins were adopted.Supplementary Files

Figures
Figures
* -the symptom cannot be assessed (the patient too young)

Table 2
The overview of missense and nonsense variants found in the EFNB1 gene through MutationTaster, Varsome online tools (obtained on 2 th November 2020) and Alamut ® Visual software (obtained on 10 th November 2020)