Germline ATM variants predispose to melanoma: a joint analysis across the GenoMEL and MelaNostrum consortia.

PURPOSE: Ataxia-Telangiectasia Mutated (ATM) has been implicated in the risk of several cancers, but establishing a causal relationship is often challenging. Although ATM single-nucleotide polymorphisms have been linked to melanoma, few functional alleles have been identified. Therefore, ATM impact on melanoma predisposition is unclear. METHODS: From 22 American, Australian, and European sites, we collected 2,104 familial, multiple primary (MPM), and sporadic melanoma cases who underwent ATM genotyping via panel, exome, or genome sequencing, and compared the allele frequency (AF) of selected ATM variants classified as loss-of-function (LOF) and variants of uncertain significance (VUS) between this cohort and the gnomAD non-Finnish European (NFE) data set. RESULTS: LOF variants were more represented in our study cohort than in gnomAD NFE, both in all (AF = 0.005 and 0.002, OR = 2.6, 95% CI = 1.56-4.11, p < 0.01), and familial + MPM cases (AF = 0.0054 and 0.002, OR = 2.97, p < 0.01). Similarly, VUS were enriched in all (AF = 0.046 and 0.033, OR = 1.41, 95% CI = 1.6-5.09, p < 0.01) and familial + MPM cases (AF = 0.053 and 0.033, OR = 1.63, p < 0.01). In a case-control comparison of two centers that provided 1,446 controls, LOF and VUS were enriched in familial + MPM cases (p = 0.027, p = 0.018). CONCLUSION: This study, describing the largest multicenter melanoma cohort investigated for ATM germline variants, supports the role of ATM as a melanoma predisposition gene, with LOF variants suggesting a moderate-risk.

Prevalence of monogenic diabetes in the population-based Norwegian Childhood Diabetes Registry.

AIMS/HYPOTHESIS: Monogenic diabetes (MD) might be misdiagnosed as type 1 diabetes. The prevalence of MD among children with apparent type 1 diabetes has not been established. Our aim was to estimate the prevalence of common forms of MD in childhood diabetes. METHODS: We investigated 2,756 children aged 0-14 years with newly diagnosed diabetes who had been recruited to the nationwide population-based Norwegian Childhood Diabetes Registry (NCDR), from July 2002 to March 2012. Completeness of ascertainment was 91%. Children diagnosed with diabetes who were under12 months of age were screened for mutations in KCNJ11, ABCC8 and INS. Children without GAD and protein tyrosine phosphatase-like protein antibodies were screened in two ways. Those who had a parent with diabetes were screened for mutations in HNF1A, HNF4A, INS and MT-TL1. Children with HbA1c <7.5% (<58 mmol/mol) and no insulin requirement were screened for mutations in GCK. Finally, we searched the Norwegian MODY Registry for children with genetically verified MD. RESULTS: We identified 15 children harbouring a mutation in HNF1A, nine with one in GCK, four with one in KCNJ11, one child with a mutation in INS and none with a mutation in MT-TL1. The minimum prevalence of MD in the NCDR was therefore 1.1%. By searching the Norwegian MODY Registry, we found 24 children with glucokinase-MODY, 15 of whom were not present in the NCDR. We estimated the minimum prevalence of MD among Norwegian children to be 3.1/100,000. CONCLUSIONS/INTERPRETATION: This is the first prevalence study of the common forms of MD in a nationwide, population-based registry of childhood diabetes. We found that 1.1% of patients in the Norwegian Childhood Diabetes Registry had MD.

Exocrine pancreatic function in hepatocyte nuclear factor 1ß-maturity-onset diabetes of the young (HNF1B-MODY) is only moderately reduced: compensatory hypersecretion from a hypoplastic pancreas.

OBJECTIVES: To examine the exocrine pancreatic function in carriers of the hepatocyte nuclear factor 1ß gene (HNF1B) mutation by direct testing. METHODS: Patients with HNF1B mutations and control subjects were assessed using rapid endoscopic secretin tests and secretin-stimulated magnetic resonance imaging. Seven patients and 25 controls underwent endoscopy, while eight patients and 20 controls had magnetic resonance imaging. Ductal function was assessed according to peak bicarbonate concentrations and acinar function was assessed according to peak digestive enzyme activities in secretin-stimulated duodenal juice. The association of pancreatic exocrine function and diabetes status with pancreatic gland volume was examined. RESULTS: The mean increase in secretin-stimulated duodenal fluid was smaller in patients than controls (4.0 vs 6.4 ml/min; P = 0.003). We found lower ductal function in patients than controls (median peak bicarbonate concentration: 73 vs 116 mEq/L; P < 0.001) and lower acinar function (median peak lipase activity: 6.4 vs 33.5 kU/ml; P = 0.01; median peak elastase activity: 0.056 vs 0.130 U/ml; P = 0.01). Pancreatic fluid volume outputs correlated significantly with pancreatic gland volumes (r² = 0.71, P = 0.008) in patients. The total fluid output to pancreatic gland volume ratios were higher in patients than controls (4.5 vs 1.3 ml/cm³; P = 0.03), suggesting compensatory hypersecretion in the remaining gland. CONCLUSION: Carriers of the HNF1B mutation have lower exocrine pancreatic function involving both ductal and acinar cells. Compensatory hypersecretion suggests that the small pancreas of HNF1B mutation carriers is attributable to hypoplasia, not atrophy.

To test, or not to test: time for a MODY calculator?

To test, or not to test, that is often the question in diabetes genetics. This is why the paper of Shields et al in the current issue of Diabetologia is so warmly welcomed. MODY is the most common form of monogenic diabetes. Nevertheless, the optimal way of identifying MODY families still poses a challenge both for researchers and clinicians. Hattersley's group in Exeter, UK, have developed an easy-to-use MODY prediction model that can help to identify cases appropriate for genetic testing. By answering eight simple questions on the internet ( www.diabetesgenes.org/content/mody-probability-calculator ), the doctor receives a positive predictive value in return: the probability that the patient has MODY. Thus, the classical binary (yes/no) assessment provided by clinical diagnostic criteria has been substituted by a more rational, quantitative estimate. The model appears to discriminate well between MODY and type 1 and type 2 diabetes when diabetes is diagnosed before the age of 35 years. However, the performance of the MODY probability calculator should now be validated in other settings than where it was developed-and, as always, there is room for some improvements and modifications.

The spectrum of ABCC8 mutations in Norwegian patients with congenital hyperinsulinism of infancy.

Potassium channels in the plasma membrane of the pancreatic beta cells are critical in maintaining glucose homeostasis by responding to ATP and coupling metabolic changes to insulin secretion. These channels consist of subunits denoted the sulfonylurea receptor SUR1 and the inwardly rectifying ion channel KIR6.2, which are encoded by the genes ABCC8 and KCNJ11, respectively. Activating mutations in the subunit genes can result in monogenic diabetes, whereas inactivating mutations are the most common cause of congenital hyperinsulinism of infancy (CHI). Twenty-six Norwegian probands with CHI were analyzed for alterations in ABCC8 and KCNJ11. Fifteen probands (58%) had mutations in the ABCC8 gene. Nine patients were homozygous or compound heterozygous for the mutations, indicating diffuse pancreatic disease. In five patients, heterozygous and paternally inherited mutations were found, suggesting focal disease. One patient had a de novo mutation likely to cause a milder, dominant form of CHI. Altogether, 16 different ABCC8 mutations (including the novel alterations W231R, C267X, IVS6-3C>G, I462V, Q917X and T1531A) were identified. The mutations IVS10+1G>T, R1493W and V21D occurred in five, three and two families, respectively. KCNJ11 mutations were not found in any patients. Based on our mutation screening, we estimate the minimum birth prevalence of ABCC8-CHI in Norway to 1:70,000 during the past decade. Our results considerably extend the knowledge of the molecular genetics behind CHI in Scandinavia.

Lymphomas diagnosed in Uganda during the HIV/AIDS pandemic.

BACKGROUND: There are numerous reports from different countries documenting a change in frequency and profile of lymphomas after the onset of the HIV/AIDS pandemic. In Uganda little is known concerning the distribution of lymphoma subtypes diagnosed at the Department of Pathology, Makerere University College of Health Sciences during this period. OBJECTIVE: To examine the frequency and diagnostic profile of lymphomas diagnosed in Uganda in the HIV/AIDS era. DESIGN: Retrospective study. SETTING: Department of Pathology, Makerere University College of Health Sciences, Kampala, Uganda. SUBJECTS: One thousand and thirteen patients diagnosed with lymphomas in the period 1980-1989. RESULTS: The most common type of non-Hodgkin lymphoma was Burkitt lymphoma (36%). The frequencies of lymphocytic and histiocytic types were 34.5% and 8.2% respectively. CONCLUSION: There was a decrease in histopathologically diagnosed lymphomas in Uganda in the period 1980-1989. Burkitt lymphoma continues to be the most common subtype diagnosed, some major lymphoma subtypes like T-cell and follicular lymphomas were not reported in the country in the HIV/AIDS era.

Lack of pancreatic body and tail in HNF1B mutation carriers.

AIMS: Hepatocyte nuclear factor 1B (HNF1B) gene mutation carriers have a systemic disease characterized by congenital malformations in the urogenital tract, diabetes mellitus of maturity-onset diabetes of the young type and dysfunction of the liver and exocrine pancreas. We aimed to investigate pancreatic structure and exocrine function in carriers of HNF1B mutations. METHODS: We studied five subjects from two families with the previously reported mutation R137_K161del and the novel mutation F148L in HNF1B. All patients underwent computed tomography (CT) and magnetic resonance cholangiopancreatography (MRCP). We measured faecal elastase and serum vitamins D and E. RESULTS: One of the mutation carriers reported abdominal symptoms. All five subjects had faecal elastase deficiency, three had vitamin D deficiency and two had vitamin E deficiency. Neither CT nor MRCP depicted tissue corresponding to the pancreatic body and tail in the five mutation carriers, indicating agenesis of the dorsal pancreas. The head of the pancreas was slightly atrophic but had normal X-ray attenuation at CT in all patients. CONCLUSIONS: Agenesis of the pancreatic body and tail and pancreatic exocrine dysfunction are parts of the phenotype in HNF1B mutation carriers. This strengthens the evidence for a critical role of HNF1B in development and differentiation of at least the dorsal pancreas.

Prevalence of HNF1A (MODY3) mutations in a Norwegian population (the HUNT2 Study).

AIMS: Previous reports have indicated that maturity-onset diabetes of the young (MODY) caused by hepatocyte nuclear factor 1A (HNF1A) mutations (MODY3) is the most common MODY subtype in Northern Europe, but population-based prevalence estimates are lacking. We sought to determine the prevalence of HNF1A-MODY in diabetic subjects of a defined Norwegian population (the HUNT2 Study). METHODS: Of the 1972 diabetic HUNT2 subjects, we identified a subgroup of 43 suspected MODY cases based on information on family history, disease onset and anti-glutamic acid decarboxylase autoantibody status. These cases were considered a discovery group for HNF1A mutations and underwent full DNA sequencing. Subsequently, the entire cohort of diabetic HUNT2 subjects was screened for three selected HNF1A mutations. Possible founder effects were examined using the Norwegian MODY Registry. RESULTS: Three subjects from the discovery group harboured HNF1A mutations. Two subjects had the previously described R229Q mutation, one had a novel S6N alteration, whereas the HNF1A hot-spot mutation P291fsinsC was not identified. Genotyping the cohort of diabetic HUNT2 subjects identified five additional R229Q-positive subjects. Microsatellite analysis performed for all R229Q-positive probands of the Norwegian MODY Registry and those found in the HUNT2 population revealed that 17 of 18 (94%) had genotypes consistent with a common haplotype. CONCLUSIONS: Clinical MODY criteria were fulfilled in 2.2% of diabetic HUNT2 subjects. The minimum prevalence of HNF1A-MODY among diabetic HUNT2 subjects was 0.4%. Because of founder effects, registry-based prevalence studies probably need to be very large and they should also include prospectively collected phenotypes and extensive mutation screening to establish the true prevalence of MODY.

Wolcott-Rallison syndrome with 3-hydroxydicarboxylic aciduria and lethal outcome.

Wolcott-Rallison syndrome (WRS) (OMIM 226980) is a rare, autosomal recessive disorder with infancy-onset diabetes mellitus, multiple epiphyseal dysplasia, osteopenia, mental retardation or developmental delay, and hepatic and renal dysfunction as main clinical findings. Patients with WRS have mutations in the EIF2AK3 gene, which encodes the pancreatic eukaryotic translation initiation factor 2-alpha kinase 3. We report a female patient who developed insulin-requiring diabetes at 2.5 months of age. Multiple epiphyseal dysplasia was diagnosed at age 2 years. At age 5.5 years she developed a Reye-like syndrome with hypoketotic hypoglycaemia and renal and hepatic insufficiency and died. A partial autopsy showed fat infiltration in the liver and kidneys. Examination of urine by gas chromatography and mass spectrometry showed large amounts of C(6)-dicarboxylic acid (adipic acid), 3-hydroxy-C(8)-dicarboxylic acid, 3-hydroxy-C(10)-dicarboxylic acid, and 3-hydroxydecenedioic acid. Acetoacetate and 3-hydroxybutyrate were absent. The findings suggested a metabolic block in mitochondrial fatty acid oxidation, but lack of material precluded enzyme analyses. The clinical diagnosis of WRS was suggested in retrospect, and confirmed by sequencing of DNA extracted from stored autopsy material. The patient was compound heterozygous for the novel EIF2AK3 mutations c.1694_1695delAT (Y565X) and c.3044T > C (F1015S). Our data suggest that disruption of the EIF2AK3 gene may lead to defective mitochondrial fatty acid oxidation and hypoglycaemia, thus adding to the heterogeneous phenotype of WRS.

Clinicopathological characteristics and non-adhesive organ culture of insulinomas.

BACKGROUND AND AIMS: Insulinoma is a very rare type of islet cell tumour, but nevertheless the most common endocrine tumour of the pancreas. We aimed at reviewing our clinical experience with this tumour type and to assess whether organ culture could be obtained from surgically resected insulinoma material. MATERIAL AND METHODS: All patients with insulinomas (6 men and 10 women) referred to Haukeland University Hospital between 1986 and 2006 were included in the study. Median age of onset was 53 years (range 21-74). Biochemical diagnosis was established during a 72 h fast test. Imaging and localization of the tumours were performed with intra-operative ultrasonography, endoscopic ultrasonography, CT-scan and/or transcutaneous ultrasonography. For six patients, organ cultures were set up from tumour tissue fragments. RESULTS: The annual incidence of insulinoma was 0.8 per million. The patients generally presented with non-specific, episodic symptoms, which often were mistaken for cardiovascular, neurological or diabetic disease and in some cases delayed the diagnosis with several years. Two patients had diabetes prior to the diagnosis of insulinoma. Patient weight gain was probably due to increased food intake, compensating for the hypoglycemia. Intra-operative ultrasonography detected all tumours correctly, whereas 73% were detected by endoscopic ultrasonography and 38% by CT scan. Five insulinomas were located in the head, eight in the body and three in the tail of the pancreas. All were removed by open-access surgery, eleven cases by resection and five by enucleation. One tumour was malignant with liver metastases and two patients had tumours defined as borderline. Insulinoma tissue fragments developed into spheroids during the first week of culturing and insulin secretion into the media was demonstrated. CONCLUSIONS: Insulinomas are rare and diagnostically challenging tumours. Intra-operative ultrasonography was superior to other imaging modalities to locate the lesion. In organ culture, insulinomas readily form spheroids which may be used to yield insight into beta-cell biology.

The HNF1A mutant Ala180Val: Clinical challenges in determining causality of a rare HNF1A variant in familial diabetes.

AIMS: Heterozygous mutations in hepatocyte nuclear factor-1A (HNF1A) cause maturity-onset diabetes of the young type 3 (MODY3). Our aim was to compare two families with suspected dominantly inherited diabetes and a new HNF1A variant of unknown clinical significance. METHODS: The HNF1A gene was sequenced in two independently recruited families from the Norwegian MODY Registry. Both familes were phenotyped clinically and biochemically. Microsatellite markers around and within the HNF1A locus were used for haplotyping. Chromosomal linkage analysis was performed in one family, and whole-exome sequencing was undertaken in two affected family members from each family. Transactivation activity, DNA binding and nuclear localization of wild type and mutant HNF-1A were assessed. RESULTS: The novel HNF1A variant c.539C>T (p.Ala180Val) was found in both families. The variant fully co-segregated with diabetes in one family. In the other family, two subjects with diabetes mellitus and one with normal glucose levels were homozygous variant carriers. Chromosomal linkage of diabetes to the HNF1A locus or to other genomic regions could not be established. The protein functional studies did not reveal significant differences between wild type and variant HNF-1A. In each family, whole-exome sequencing failed to identify any other variant that could explain the disease. CONCLUSIONS: The HNF1A variant p.Ala180Val does not seem to cause MODY3, although it may confer risk for type 2 diabetes mellitus. Our data demonstrate challenges in causality evaluation of rare variants detected in known diabetes genes.

Conserved sequence patterns in phages Mu and lambda DNA.

The genetic maps of bacteriophages Mu and lambda can be aligned with respect to the functions of their genes. We were interested to ascertain whether the congruence of gene order is reflected at the nucleotide sequence level. A sliding window analysis of sequences from the early regions of both phages revealed a substantial degree of similarity. Equally high scores, however, were found when the early region of Mu was compared to the late region of lambda and in self-comparisons of either Mu or lambda. Hence, the similarity is due to a common pattern of nucleotides rather than to sequence similarities between functionally related genes. Employing degenerated scoring matrices we could show that primarily adenine and thymine residues contribute to the high scores and that a specific clustering of these residues is the basis for the conserved pattern. Since such a similarity was not observed with control sequences of other phages. Escherichia coli or eukaryotic viruses, the data support the notion that Mu and lambda have diverged from a common phage module. In general, our approach could offer a simple and sensitive way to trace distant relationships.

Sequence analysis of the zebrafish hox-B5/B6 region.

The region between the zebrafish homeobox genes hox-B5 and hox-B6 was sequenced, and searched for consensus binding sites of retinoic acid receptors and other transcription factors. A continuous sequence of 7.2 kb covering the zebrafish hox-B5/B6 genes was then compared to the corresponding region of the mouse Hox-B complex. Except for the open reading frames, the only highly conserved regions that could be found were stretches extending 0.3 kb upstream from the initiation codons. Within the conserved upstream regions of hox-B5/B6, we identified a common 10 bp sequence, which is also present close to the initiation codons of several other Hox genes and which therefore may be implicated in the control of their expression.

Structure and early embryonic expression of the zebrafish engrailed-2 gene.

The Drosophila homeobox gene engrailed (en) is needed for correct embryonic development, and related sequences are active during vertebrate embryogenesis. Here we report the protein coding sequence and embryonic expression pattern of the zebrafish engrailed-2 gene (eng-2) which is directly homologous to En-2 in mice and Xenopus. The predicted zebrafish Eng-2 protein shares 65% overall identity to its Xenopus counterpart. In addition to the highly conserved homeodomain region, sequence conservation is present within three short stretches in the N-terminal region. The embryonic expression of the eng-2 gene was analysed by in situ hybridization to whole-mount embryos and tissue sections. Transcripts are first detected in two lateral bands at the 10-h stage, when epiboly is completed. Within the next 2 h of development, these two bands migrate and fuse at the midline. By the time the neural keel becomes visible (11-12 h), a transverse stripe of eng-2 expressing cells is seen at the presumptive midbrain-hindbrain boundary. Later this stripe becomes significantly compressed along the AP axis, and in 24-h embryos eng-2 transcripts are detected mainly in the posterior midbrain. In the hindbrain, eng-2 expression seems restricted to the primordium of the cerebellum. A second site of activity was observed in each somite where specific myotomal cells, the muscle pioneers, express eng-2. Our observations are discussed in relation to early regionalization of the central nervous system (CNS) and the generation of morphological borders.

The zebrafish homeobox gene hox[zf-114]: primary structure, expression pattern and evolutionary aspects.

It is gradually becoming accepted that vertebrate homeobox genes, like their counterparts in Drosophila, are crucial for normal development of the embryo. Most vertebrate homeoboxes reported so far are related to the Drosophila Antennapedia (Antp) sequence, and here we describe hox[zf-114], a novel Antp-like homeobox gene from the zebrafish. The sequence of the hox[zf-114] homeodomain indicates that this gene could be a member of a subfamily defined by the mouse Hox-1.5/-2.7/-4.1 genes. However, the evolutionary origin of hox[zf-114] is unclear and, based on the putative protein sequence, we conclude that it is not directly homologous to Hox-1.5, Hox-2.7 or Hox-4.1, or to other known mammalian homeobox genes. Nevertheless, as revealed by in situ hybridization, hox[zf-114] exhibits a spatial expression pattern typical for vertebrate Antp-like homeobox genes. Transcripts are detected in the posterior hindbrain, where a sharp anterior border of expression is observed, and throughout the spinal cord. The hox[zf-114] gene is also active in a region that gives rise to the pectoral fins. These findings suggest a role for hox[zf-114] in anteroposterior patterning of the neural tube and in pectoral fin development.

Genomic structure and sequence analysis of a human inositol polyphosphate 1-phosphatase gene (INPP1).

Lithium remains the most widely used long-term treatment for bipolar affective disorder, but the molecular mechanisms underlying its therapeutic efficacy have not been fully elucidated. Two enzymes involved in the phospholipase C signalling system, namely the myo-inositol monophosphatase (IMPase) and the inositol polyphosphate 1-phosphatase (IPPase), have been postulated as targets for the therapeutic action of lithium in manic-depressive illness. Intriguingly, Drosophila mutants lacking IPPase activity display a defect in synaptic transmission, and this alteration could be phenocopied by lithium exposure. We recently demonstrated the presence of several polymorphisms in the IPPase-encoding inositol polyphosphate 1-phosphatase gene (INPP1) cDNA and suggested that polymorphic variants of the human IPPase might be associated with the striking difference in lithium response among bipolar patients. We report the genomic structure and organization of the INPP1 gene on chromosome 2q32. Based on DNA sequencing of the entire genomic region containing INPP1, we found that the gene consists of six exons and spans more than 25 kb. Expression analysis showed that INPP1 is present as a 1.9 kb mRNA transcript in all organs and tissues examined, including the central nervous system. The level of expression varies, with at least a fourfold higher transcript level in testis compared with other tissues with high expression. A highly polymorphic dinucleotide repeat, (CA)18-25, with an observed heterozygosity of 0.86 was detected immediately downstream of the gene. The present sequence information will be used to further investigate the possible role of the INPP1 gene in lithium-treated bipolar illness.

Polymorphisms in CYP2D6 duplication-negative individuals with the ultrarapid metabolizer phenotype: a role for the CYP2D6*35 allele in ultrarapid metabolism?

Ultrarapid drug metabolism mediated by CYP2D6 is associated with inheritance of alleles with duplicated or amplified functional CYP2D6 genes. However, genotyping for duplicated CYP2D6 alleles only explains a fraction (10-30%) of the ultrarapid metabolizer phenotypes observed in Caucasian populations. Using a sample of CYP2D6 duplication-negative ultrarapid metabolizer subjects and selected control subjects with extensive metabolism, we examined parts of the CYP2D7 pseudogene, and the promoter region and 5'-coding sequence of CYP2D6 for polymorphisms possibly associated with the ultrarapid metabolizer phenotype. In an initial screening of 17 subjects (13 ultrarapid metabolizers and four extensive metabolizers), we identified three DNA variants in the 5'-end of the CYP2D7 pseudogene and 29 variants in the 5'-end of the CYP2D6 gene. Five variants were then selected for examination in a larger sample of subjects having the ultrarapid metabolizer (n = 27) or extensive metabolizer phenotype (n = 77). Subsequent statistical analyses of allele, genotype and estimated haplotype distributions showed that the 31A allele of the 31G > A (Val(II)Met) polymorphism was significantly more frequent in ultrarapid metabolizer subjects than in extensive metabolizer subjects (P = 0.04). Also, estimation of haplotype frequencies suggested that one of the haplotypes with the 31A variant was significantly more frequent among the ultrarapid metabolizers compared with the extensive metabolizers (P = 0.03). The average metabolic ratio was significantly lower in subjects possessing the 31A allele compared with subjects homozygous for the 31G allele (P = 0.02). We also observed a nonsignificant over-representation of the G-allele of a - 1584 C > G promoter polymorphism in the ultrarapid metabolizer group. Since our results are based on a relatively low number of subjects, further studies on larger samples and functional analyses of the polymorphisms detected are necessary to determine the role of the 31G > A and - 1584C > 6 variants in CYP2D6 duplication-negative ultrarapid metabolizer subjects.

Molecular cloning and characterization of homeo-box-containing genes from Atlantic salmon.

As the most primitive group among vertebrates, fish might serve as a model system when studying the genetic regulation of embryogenesis in higher animals. To identify genes important for early development, we have constructed a genomic library from Atlantic salmon (Salmo salar) and screened it with homeobox-containing probes from Drosophila melanogaster. Five different salmon homeoboxes were isolated. Two of these were located in the same clone, separated by only 7.5 kb. This demonstrates the presence of clustered homeobox genes in fish. The two clustered homeoboxes were sequenced and shown to be closely related to the ANT-C/BX-C class of Drosophila, being about 80% homologous to the Ultrabithorax gene (Ubx) homeobox. One of the clustered genes appears to be the salmon equivalent of the mouse Hox-2.1 gene, indicating that some of the vertebrate homeobox-containing genes are conserved in evolution. A more diverged homeobox that shares only 60% homology with Ubx, was also sequenced. In analogy to Drosophila, therefore, the salmon genome contains more than one class of homeoboxes. In addition, Northern-blot experiments demonstrated that two of the homeobox genes are expressed in salmon embryos, suggesting their importance for proper development.

Structure and neural expression of a zebrafish homeobox sequence.

A genomic library of zebrafish was constructed and screened with homeobox-containing probes. One of the positive clones contains a transcribed region which shares extensive sequence homology with the murine Hox-1.4 and Hox-2.6 genes and the human HHO.c13 gene. Characterization of this zebrafish homologue (ZF-13) with respect to expression demonstrated that it is transcribed during embryogenesis where a major RNA species of 2.5 kb and a minor transcript of 4.6 kb are detected. The highest concentration of both transcripts was found in embryos at the stage of somite formation. By in situ hybridization the spatial localization of expression was analysed in hatching embryos. Hybridization signals were mainly detected throughout the neural tube and in the brain. A small amount of RNA derived from ZF-13 was localized in differentiated muscle cells. Our results suggest that homeobox genes of distantly related vertebrate species are very similar with respect to structure and function.

A zebrafish homologue of the murine Hox-2.1 gene.

Homeobox-containing sequences were isolated from a genomic library of zebrafish (Brachydanio rerio). A lambda clone containing two homeobox cross-hybridizing regions was characterized. DNA sequencing of one of these regions (ZF-21) revealed that it contains a homeobox closely related to the Antennapedia class of Drosophila homeobox sequences. Moreover, the deduced amino acid sequence of the C-terminal end (81 residues including the homeobox) is identical to the corresponding part of the murine Hox-2.1 protein. Similar to Hox-2.1, a ZF-21 derived transcript of 2.3 kb is present in embryos at the somite forming stages.