NSRD loci 
A disease locus could be identified and mapped using linkage analysis, that is based on the following principles. If two loci, A (A1 + A2) and B (B1 + B2), are on different chromosomes they will segregate independently and there is a 50% chance that one child receive allele B1 or B2.   
If the loci are syntenic, that is if they lie on the same chromosome, then they might be expected always to segregate together, with no recombinants. A crossover, if it occurs between the positions of the two loci, will create a recombinant.  
The possibility of the occurrence of a crossover is directly dependent on the distance of the two loci. Higher is the distance higher is the possibility that a crossover will occur, and thus a recombinational event.  
Recombination fractions is a measure of the distance between two loci and also define the genetic distance. To map a disease locus we need genetic markers which should be sufficiently polymorphic to give a reasonable chance that a randomly person will be heterozygous.  
When a polymorphic DNA marker is used, families can be selected for linkage analysis because they have an interesting disease or because they have a good structure for mapping, with a reasonable hope that family members will not all be homozygous (and hence uninformative) for the marker. The standard tools for linkage analysis are now microsatellites, which are DNA markers characterized by (CA)n repeats. They are numerous and spaced across the entire human genome.  
Using such DNA markers, linkage studies revealed the presence of a marked genetic heterogeneity for NSRD, with 17 different loci so far identified.  
In particular, linkage analyses using highly polymorphic microsatellite markers in two consanguineous families from Tunisia showed a 2-point LOD score of 9.88 at q = 0.01 with the marker D13S175 located on the long arm of chromosome 13. The disorder was referred as NSRD1 and the gene symbol DFNB1 was used.  
Additional loci have been rapidly mapped. DFNB2 was identified always in consanguineous families from Tunisia to chromosomes 11q13.5(4). DFNB3 on 17p11.2-q12 was also linked to deafness in consanguineous families from a remote village in Bali(5).  
A fourth deafness locus, DFNB4, described in Middle Eastern Druze individuals, was positioned at 7q31(6). Three additional loci, DFNB5, DFNB6 and DFNB7, mapped to 14q12, 3p14-p21 and 9q13-q21, respectively, by studying multiple inbred families from India(7,8,9). DFNB8 was mapped to the distal arm of chromosome 21 in a family with NSRD from Pakistan(10); DFNB9, located at 2p22-23, was described in a consanguineous family living in an isolated region of Lebanon(11).  
DFNB10 was recently located in a 12 cM region near the telomere of chromosome 21 in a large inbred Palestinian family(12) and DFNB12 to 10q21-22 in a consanguineous Sunni family from Syria(13). Finally, several additional loci (DFNB11, DFNB13, DFNB14, DFNB15, DFNB16, and DFNB17) have been recently reported to the Hearing Loss Homepage on the Web.  A description of NSRD loci is given in Table 1. 

Linkage studies: the role of DFNB1, DFNB2 and DFNB4 in Cucasians 
In addition to Tunisian patients, DFNB1 was shown to be responsible for NSRD in a highly inbred Bedouin family(14). Linkage to markers D13S175, D13S143, and D13S115 on chromosome 13 has also been shown in 9 out of 18 New Zealand and 1 Australian non-consanguineous kindreds with NSRD(15).  
This finding suggest that the DFNB1 locus may make an important contribution to autosomal recessive neurosensory deafness in the Caucasian population.  
Finally, DFNB1 has also been demonstrated to be responsible for hearing impairment in one large family of a group of 27 of Pakistani origin with NSRD(16). In a recent study we performed a genetic linkage study with four microsatellite markers linked to DFNB1 in a total of 48 independent Mediterranean families, of which 30 and 18 were of Italian and Spanish descent, respectively(17). A maximum two-point LOD score of 7.28 was found with marker D13S115 at a recombination frequency of 0.1.  
Significant LOD scores were also obtained for D13S143, D13S292 and D13S175. Genetic heterogeneity was confirmed using the HOMOG program which indicated absence of linkage to DFNB1 in approximately 40% of the sample.  
This study clearly demonstrated that DFNB1 plays an important role in 80% of Mediterranean families with NSRD. Furthermore, results from multipoint analysis predicted that the DFNB1 gene should has been localized between markers D13S175 and D13S115 which are separated by approximately 14.2 cM. Successively, six polymorphic markers, three linked to DFNB2, on chromosome 11, and three others to DFNB4, on chromosome 7, were analyzed in those families certainly not linked to DFNB1(18).   
Positive LOD Scores were detected with 6 families out of 48 analyzed (.12) with markers linked to DFNB4.  
The HOMOG test showed that there was linkage with heterogeneity. These findings suggested that a second locus (DFNB4) could also play an important role in our population.  
Finally, two families out of 48 (.04) showed the segregation of the alleles of markers associated to DFNB2 locus within affected and normal members of the family. In this case, the number of families most likely associated to DFNB2 was too low to reach a statistical significance and to get positive LOD Scores.  
Nevertheless, the finding of two families segregating with DFNB2 but not with the other NSRD loci so far investigated was a good suggestion for a possible implication of DFNB2 in a minority of our deaf cases.  
Considering that approximately 80% of our patients was associated to DFNB1 and that most likely an additional proportion of about 15% of cases could be associated to DFNB4 and DFNB2, it results that, despite the great genetic heterogeneity present in recessive deafness, three genes could account together for at least 95% of all cases of NSRD in our population.  
This finding has severe important implications for the genetic counselling and the prevention of such a common disease.  

Identification of connexin-26 (GJB2) as the DFNB1 gene 
After the demonstration that a large proportion of NSRD cases in our patient population was linked to DFNB1(17), the candidate region to contain the gene (~14.2 cM) was further analyzed using additional informative microsatellite markers in an attempt to narrow the interval.  
Several recombinational events were observed which further refined the region to approximately 5 cM flanked by markers D13S141 and D13S232.  
Efforts then focused on defining DFNB1 candidate genes mapping to this approximate chromosomal region which were also expressed in human cochlear cells. Previous studies showed connexin 26 expression in cochlear cells, and genomic mapping data placed this gene within the interval defined by our linkage studies.  
Mammalian connexin genes contain the complete coding region in a single exon of approximately 800 nucleotides, greatly facilitating analysis of the entire protein region using a rapid PCR-based approach employing patients' genomic DNA.  
A single-base deletion (35 del G) located 227 nucleotides (nt) downstream to the mRNA cap site, 29 nt 3' to the 1st nt of the chain-initiating ATG codon was detected in 34 of 54 NSRD chromosomes from our Italian patients (19).   
Three of 6 unrelated Spanish and 1 Israeli patients with NSRD linked to chromosome 13 were also homozygous for the G deletion.   
This mutation was not present in 50 normal chromosomes.   
The deletion results in a frameshift leading to a UGA stop codon two residues downstream.  
An example of the deletion segregating within a deafness family is reported in Figure 1. Some additional mutations, always leading to a truncated not functioning GJB2 protein, have been also detected (Gasparini, unpublished data). 

Other NSRD Genes 
Mutations in patients affected by NSRD have also been detected in the myosin VIIA (MYO7A) gene, on human chromosome 11q13 (DFNB2)(20,21).  
Myosin VIIA is expressed in the sterocilia of the hair cells and is the intracellular anchor for contact between each sterocilium (22).  
Connexin 26 (GJB2) function 
Connexin 26 is a member of a large family of proteins involved in formation of gap junctions which allow the direct transfer of small molecules and ions between neighboring cells.  
Each cell contributes half of the gap junction, and is composed of the oligomeric assembly of connexons, which are composed of an hexamer of an integral membrane protein named connexins(23,24).  
A schematic example of connexons is given in Figure 2. Gap junctions are rare between mammalian neurons, but are common in non-neural cells, such glia, epithelial cells and smooth and cardiac muscle cells.  
There are more than 11 different types of connexins, each with tissue specificity and distinct physiological properties to the gap junction channels(24,25).   
At least two other connexin genes have been implicated in human diseases. Point mutations in the connexin-32 gene are associated with X-linked Charcot-Marie-Tooth neuropathy(26), and mutations in the connexin-43 gene are found in patients with heterotaxia and heart malformations.   
The involvement of gap junctions in the intracellular responses to sound was suggested more than 20 years ago (27,28). Within the auditory organ, gap junctions are located between the outer hair cells and supporting cells (including melanocytes), providing a morphological basis for the occurrence of intracellular responses to sound in supporting cells and for electric coupling of receptor cells.  
The endothelium of the scala media of the cochlea is involved in the production of a receptor response to the auditory stimulus and is separated from the endolymphatic space by tight junctions in the marginal cell layer, which is coupled by gap junctions.   
Very recently, immunohistochemical and ultrastructural analysis of connexin-26 in the rat cochlea showed that gap junctions in both epithelial and connective tissue cells are involved in recycling endolymphatic potassium ions through the sensory cells during the transduction of voltage gating in these channels. The identification of mutations in the GJB2 gene clearly confirm the crucial role of gap junctions in auditory transduction(19,29). 
Most of the mutations detected in the GJB2 gene lead to complete absence of connexin-26.(19,29) Considering that GJB2 is expressed in several tissues but its loss produce only deafness, it must be assumed that other connexins could substitute for connexin-26.  
Connexin-26 and connexin-32 can form heterodimers with functional gap junction channels (30,31).  
However, mutations in the connexin-32 gene cause the Charcot-Marie-Tooth neuropathy(26) while mutations in the connexin-26 gene only cause deafness (19,29). These findings suggest that the expression of connexin-26 in the cochlea is essential for audition and that other connexins can not compensate for the loss of connexin-26 in the auditory epithelial cells. 

Perspectives 
Until few years ago genetic deafness was a “mare magnum” in which the absence of knowledge was the main feature. Successively, several loci have been described and the presence of genetic heterogeneity, previously only hypothesized, was clearly demonstrated. In few years, our work allowed us to demonstrate the presence of a major locus (i.e. DFNB1) in NSRD, and then to identify the DFNB1 gene (GJB2), confirming its important role in determining the most common form of genetic deafness.  
Moreover, the identification of a very common mutation, within GJB2 gene, in our population greatly facilitates molecular diagnosis and genetic counseling of congenital deafness.  
In addition the identification of the mutation early after birth should help to inform families and hopefully permit faster treatment of the affected child.   

Paolo Gasparini, Paolo Fortina*, Xavier Estivill**, Salvatore Melchionda, Lucio Vigliaroli, Leopoldo Zelante  

Servizio di Genetica Medica e Divisione di ORL, IRCCS-Osp.CSS,  
S.Giovanni Rotondo (FG), Italy  
(*) Dept. of Pediatrics, Univ. of Pennsylvania, School of Medicine, The Children's Hospital  
of Philadelphia, PA, USA  
(**) Dept. of Molecular Genetics, IRO, Crta, Castelldefels Km 2,7, Hospitalet,  
Barcelona, Spain