Учебники / Genetic Hearing Loss Willems 2004

Figure 1 Intercellular junctions and pathways across cellular sheets. (A) Schematic drawing of intestinal epithelial cells. Di erent types of intercellular junctions, including desmosomes, adherens junctions, gap junctions, and tight junctions (TJs), are illustrated. The junctional complex, which is located at the most apical region of lateral membranes, is circled. (B) Distinct pathways across cellular sheets: a schematic representation of the transcellular versus the paracellular route. (Reprinted by permission from Nature Reviews Molecular Cell Biology (8), copyright 2001, Macmillan Magazines Ltd.).

cells and thus set up a semipermeable barrier that prevents or reduces paracellular di usion (‘‘barrier function’’) (16). The actual barrier capacity of TJs can be determined by measurements of the electrical resistance across the epithelium (17). Barrier function of TJs was demonstrated for relatively large molecules, such as hemoglobin (3), as well as for smaller molecules, such as the colloidal tracer lanthanum hydroxide (18). TJs partly block the paracellular transport of water and small electrolytes. Depending on the functional requirements of an epithelium, there may be small or large amounts of water and small solutes flowing passively through the TJ (17). The paracellular permeability of di erent epithelia was found to correlate with the number of TJ strands along the apical-basal axis (19). The morphological pattern of the strands also varies among tissues; however, the physiological correlate of these ultrastructural di erences is yet unknown.

B.Tight Junction Proteins

TJ strands are composed of at least three types of membrane-spanning proteins: occludin (20), di erent members of the claudin family (21), and junction adhesion molecule (JAM) (22,23). These strand-associated membrane proteins interact with the actin-based cytoskeleton (24), as well as with mem- brane-associated proteins that function as adapters and signaling proteins (reviewed in Refs. 25,26).

1.Occludin

Occludin was the first identified TJ protein (20). It is exclusively localized to TJs of both epithelial and endothelial cells (20). It spans the membrane four times with cytoplasmic amino and carboxy termini and forms two extracellular loops that are composed mostly of glycine and tyrosine (20). Initially occludin was thought to be the main TJ sealing protein. Several lines of evidence supported this hypothesis: Occludin expression induced adhesion in suspended cell assays (27); disruption of occludin interactions by the addition of peptides corresponding to its extracellular loops resulted in a drop in transepithelial electrical resistance (28,29); mutation or overexpression of occludin in cultured cells a ected permeability properties (30–32). However, several studies suggested that occludin might not be absolutely required for TJ formation. Some cell types have TJs although they express only low levels of occludin (33,34). Expression of a carboxy terminus truncated occludin or addition of peptides corresponding to occludin extracellular loops disrupted occludin localization in TJs in cultured cells, but did not a ect gross TJ morphology (29,30). Moreover, occludin-null embryonic stem cells di erentiated into epithelial cells and formed well-developed TJ structures (35).

Occludin-null mice demonstrated a wide variety of histological abnormalities, including the gut, brain, testis, salivary glands, and bone, but TJs appeared una ected and the epithelial barrier function was normal. This phenotype suggested that TJs and occludin have a more complex function than previously assumed (36).

2.The Claudin Family

The claudins are a family of more than 20 genes encoding TJ proteins. Claudin 1 and 2 were discovered first, based on their cofractionation with occludin from isolated chicken liver junctions (37). Both of these 22-kD proteins have no sequence similarity to occludin, but they share the same topology of four transmembrane domains with two extracellular loops, and cytoplasmic amino and carboxy termini (37). Immunofluorescence and immunoelectron microscopy revealed that claudin 1 and 2 were both targeted to and incorporated into the TJ strand itself (37). When each of these proteins was transfected into cells that lack TJs, they were highly concentrated at cell-cell contact sites. In freeze-fracture replicas of these contact sites, well-developed networks of strands were identified that were similar to TJ strand networks in vivo (38). Although occludin was also concentrated at cell contact sites when transfected into cells, it created only a small number of short strands. However, when cotransfected with claudin 1, occludin was incorporated into well-developed claudin 1–based strands (38). These findings suggested that claudins are mainly responsible for TJ strand formation, and that occludin is an accessory protein in the TJ, although it might have additional functions that are not yet understood (38).

After the discovery of claudin 1 and 2, additional family members were identified through amino acid sequence similarity searches (39), and to date more than 20 known claudin paralogs have been identified in humans (Table 1). All members of the claudin family share the same membrane topology and other structural features, as indicated by the observation that eight di erent claudins can bind to the three submembrane proteins ZO-1, -2, and -3 through their carboxy terminus cytosolic domain (40). Similar to claudin 1 and 2, other claudins are also concentrated at preexisting TJs when transfected into MDCK cells (39). Northern blotting showed that the tissue distribution pattern varies significantly among di erent claudin family members, and that many tissues express multiple claudin species (39). These findings suggested that multiple claudin family members are involved in the formation of TJ strands in a tissue-dependent manner. When claudin 1, 2, and 3 were coexpressed in mouse L fibroblasts in di erent combinations, di erent claudins were copolymerized into individual TJ strands (heteropolymers) (Fig. 3A), and when two transfected clones of mouse L fibroblasts singly

Figure 3 Illustration of interactions between di erent claudins within and between TJ strands. Claudins can form homopolymers or heteropolymers (A), which can interact with other homoand heteropolymers (B–D) (18,38). Various claudin combinations increase both the structural and the functional diversity of TJs in di erent tissues, and might explain the di erence in transepithelial resistance found between ‘‘tight’’ and ‘‘leaky’’ epithelia. (Based on Ref. 21.)

expressing claudin 1, 2, or 3 were cocultured it was found that claudin 1- or claudin 2–based strands (homopolymers) laterally associated with claudin 3– based strands but not with each other (Fig. 3C) (41). When L fibroblasts transfectants singly expressing claudin 1 were cocultured with transfectants coexpressing claudin 1 and 2, claudin 1 homopolymers laterally associated with claudin 1 and 2 heteropolymers to form paired strands (Fig. 3D) (21). Since the claudin family includes more than 20 members, a very large number of combinations are possible within and between TJ strands in di erent tissues (Fig. 3). This mode of assembly of claudins increases both the structural and the functional diversity of TJs in di erent tissues (41), and might explain the 100,000-fold di erence in transepithelial resistance found between ‘‘tight’’ and ‘‘leaky’’ epithelia (19).

Convincing support for the function of the claudins in creating the TJ physiological barrier came from several sources. Clostridium perfringens enterotoxin removes specific claudins from TJ strands, causing TJ disintegration and reduction of the TJ barrier function (42). Additional evidence came from the phenotypes of humans and animals with mutations in specific claudin genes (Table 1). Mutations in human CLDN16 cause renal hypomagnesemia with hypercalciuria and nephrocalcinosis, due to a defect in paracellular resorption of Mg2+ in the thick ascending limb of Henle in the kidney (43). A null mutation in bovine Cldn16 causes chronic interstitial nephritis, which is characterized by failures in selective filtration and absorption in surface renal epithelium, due to dysfunction of paracellular renal transport systems (44). In addition, Cldn11-null mice demonstrate both neurological and reproductive deficits, due to the absence of TJs in the central nervous system myelin and between Sertoli cells (45). Finally, we demonstrated that mutations of CLDN14 in humans cause profound, congenital, recessive deafness DFNB29, and hypothesized a failure to maintain the electrochemical gradient between the endolymph and its surrounding tissues in the inner ear’s organ of Corti (46).

Taken together, the evidence strongly suggests that claudins are the primary proteins responsible for the physiological and structural paracellular barrier function of TJs (21). Yet the discovery that at least one of the family members, claudin 1, is not restricted to TJs in the rat epididymis raises the possibility that claudins might have additional roles, such as cell-cell adhesion (47). Claudins were also suggested to participate in morphogenesis, based on their strong and specific expression in vertebrate primordia (48).

II.CLAUDIN 14

A.Mutations in the Human CLDN14 Gene Cause Autosomal Recessive Deafness DFNB29

The DFNB29 locus on chromosome 21q22.1 was defined by two large consanguineous Pakistani families segregating profound, congenital, recessive deafness (46). DFNB29-a ected individuals in these families show no signs of vestibular dysfunction or any other symptoms beside deafness. The DFNB29 2.3-Mb interval included three genes: CLDN14 (claudin 14), KIAA0136, a gene of unknown function, and CHAF1B (chromatin assembly factor 1B-p60 subunit) (49), and excluded the DFNB10 locus, which was later found to encode a novel serine protease, TMPRSS3 (50,51). Mutations of gap junction proteins encoded by GJB2 (Cx26) and GJB3 (Cx31) are significant causes of deafness (52–55). It was thus hypothesized that other proteins with functions

important for inner ear intercellular junctions might be essential for hearing. This hypothesis made CLDN14, encoding a member of the claudin family of TJ proteins, an excellent candidate gene for DFNB29.

Comparison of CLDN14 cDNA sequence to the genomic sequence of human chromosome 21 indicated that the CLDN14 gene has three exons, with the entire protein of 239 amino acid residues encoded in exon 3 (46). Sequencing of exon 3 led to the identification of two CLDN14 mutations that cosegregated with deafness in both DFNB29 families (46). One of the families cosegregated 398delT, a single nucleotide deletion within codon Met133, located in the third transmembrane domain. This frameshift mutation is predicted to cause premature translation termination 69 nucleotides later, after the incorporation of 23 incorrect amino acids and the loss of almost half of the predicted claudin 14 protein (46). The second family cosegregated a missense mutation, V85D (aspartic acid substituted for valine), due to a transversion of T to A at position 254. Valine 85 is conserved among 12 of 20 claudins, while isoleucine is present among five claudins, and the remaining three claudins have either a cysteine or a proline at this position (46). Aspartic acid at position 85 is predicted to a ect hydrophobicity and disrupt the predicted secondary structure in the second transmembrane domain (46). Neither one of the two CLDN14 mutations was detected in 300 normal control chromosomes from Pakistani individuals, indicating that these two variants are not common polymorphisms in the Pakistani population (46). Among 100 Pakistani families that could support statistically significant linkage of recessive nonsyndromic deafness (56), only these two families segregated deafness that was linked to CLDN14 (46). Therefore, mutations of CLDN14 probably account for a small portion of recessive nonsyndromic deafness in Pakistan. Finding two mutant alleles of CLDN14 cosegregating with recessive deafness in two consanguineous families demonstrates the significant role of claudin 14 in the cochlea and its importance in the hearing process.

B.Cldn14 Knockout Mice Are Deaf: A Mouse Model for Autosomal Recessive Deafness DFNB29

To explore the role of claudin 14 in the inner ear and in other organs and tissues a targeted deletion of Cldn14 was used to create Cldn14-null mice, on a C57BL6/129SVj mixed background. Cldn14-null homozygous mice are viable, healthy, and fertile. To evaluate their hearing, Cldn14-null homozygous and heterozygous mice and their wild-type (WT) littermates underwent auditory brainstem response (ABR) analyses at 4 weeks of age. Responses to 50-As duration clicks, and 8-, 16-, and 32-kHz tone bursts were recorded. Thresholds were determined in 5-dB steps of decreasing stimulus intensity,

until waveforms lost reproducible morphology. Cldn14-null homozygous mice were found to be deaf, while thresholds of their heterozygous and homozygous WT littermates were indistinguishable (T. Ben-Yosef, personal communication, 2003). Cldn14 knockout mice are therefore a valuable model for studying the pathophysiology of autosomal recessive deafness DFNB29.

C.CLDN14 Expression

Northern blot analysis demonstrated CLDN14 expression in human kidney and liver (46). CLDN14 is probably expressed in much lower levels in additional human and mouse tissues, including brain, pancreas, spleen, skin, colon, and testis, as indicated by RT-PCR (46) and by the presence of several CLDN14/Cldn14 expressed sequence tags (ESTs) in the NCBI EST database (http://www.ncbi.nlm.nih.gov/dbEST/index.html), but the physiological significance of these findings is not clear. The CLDN14 transcript is alternatively spliced, as indicated by the finding of two splicing isoforms in human liver cDNA, with and without exon 2, and by Northern blot analysis showing a shorter transcript in human kidney (46). Although RNA stability or rate of translation may di er, these splice isoforms do not alter the amino acid sequence of claudin 14, which is encoded entirely by exon 3. The functional distinctions, if any, between these splice isoforms remain to be explored.

The developmental profile and cell-specific expression pattern of Cldn14 in the mouse inner ear was investigated using in situ hybridization and immunocytochemistry to detect Cldn14 mRNA and protein, respectively. In the inner ear no expression was detected by in situ hybridization at embryonic days 15 or 17, but it was detected at postnatal days 4 and 8 (46). At postnatal day 4, Cldn14 expression was apically located in the inner and outer hair cell region of the entire organ of Corti. At postnatal day 8, the highest Cldn14 expression was detected in the supporting cells of the organ of Corti, including the pillar, Deiters’, and inner sulcus cells (46).

The cochlea of the inner ear has two compartments with di erent ionic compositions. The perilymph of the scala vestibuli and scala tympani has low K+ and high Na+ concentrations, similar to cerebrospinal fluid (2). The ionic composition of the endolymph of the scala media is similar to that of an intracellular microenvironment, which has high K+ and low Na+ concentrations (reviewed in Ref. 1). This large K+ gradient contributes to an 80– 100-mV endocochlear potential, attributed in part to Na+ - K+-ATPase activity in the stria vascularis (57–61). This electrochemical gradient is critical for the depolarization of sensory hair cells, increasing the sensitivity of the mechanically activated transduction channels located at the top of stereocilia (62,63). The postnatal onset and rise of the endocochlear potential is presumably dependent upon the development of the paracellular barrier