Localisation of pseudohypoaldosteronism genes to chromosome 16p12.2-13.11 and 12p13.1-pter by homozygosity mapping
Localisation of pseudohypoaldosteronism genes to chromosome 16p12.2-13.11 and 12p13.1-pter by homozygosity mappingSandra S. Strautnieks1,+, Richard J. Thompson1,+, Aaron Hanukoglu2, Michael J. Dillon3, Israel Hanukoglu4, Ursula Kuhnle5, Jonathan Seckl6, R. Mark Gardiner1 and Eddie Chung1,*
1Department of Paediatrics, University College London Medical School, The Rayne Institute, 5 University Street, London WC1E 6JJ, UK, 2Department of Paediatrics, E. Wolfson Hospital, and Tel-Aviv University, Sackler School of Medicine, Tel-Aviv, Israel, 3Division of Medicine, Institute of Child Health and Renal Unit, Great Ormond Street Hospital for Children NHS Trust, 30 Guilford Street, London WC1N 1EH, UK, 4Research Institute, College of Judea and Samaria, Ariel 44820, and Institute of Endocrinology, Sourasky Medical Centre, Tel-Aviv, Israel, 5University of Munich, Children's Hospital, D-80337 Munich, Germany and 6Department of Medicine, The University of Edinburgh, Western General Hospital, Edinburgh, Scotland EH4 2XU, UK
Received November 14, 1995;Revised and Accepted November 28, 1995
Pseudohypoaldosteronism type 1 (PHA1, OMIM 264350) is a rare Mendelian disorder characterised by end-organ unresponsiveness to mineralocorticoids. Most steroid hormone insensitivity syndromes arise from mutations in the corresponding receptor, but available genetic evidence is against involvement of the mineralocorticoid receptor gene, MLR, in PHA1. A complete genome scan for PHA1 genes was undertaken using homozygosity mapping in 11 consanguineous families. Conclusive evidence of linkage with heterogeneity was obtained with a maximum two-locus admixture lod score of 9.9. The disease locus mapped to chromosome 16p12.2-13.11 in six families and to 12p13.1-pter in the other five families. The two chromosomal regions harbour genes for subunits of the amiloride-sensitive epithelial sodium channel: SCNN1B and SCNN1G on 16p and SCNN1A on 12p. Liddle's syndrome of hypertension and pseudoaldosteronism has been shown to arise from mutations in SCNN1B and SCNN1G. These results strongly suggest that PHA1 and Liddle's syndrome are allelic variants caused by mutations in genes encoding subunits of this sodium channel. These genes are of broad biological interest both in relation to sodium and water homeostasis in mammals and by virtue of their homology to the mec genes of Caenorhabditis elegans involved in mechanosensitivity and neuronal degeneration.
Pseudohypoaldosteronism type 1 (PHA1, OMIM 264350) is an uncommon inherited disorder characterised by target-organ unresponsiveness to mineralocorticoids. Since the first report by Cheek and Perry in 1958 (1 ) over 100 cases have been reported. Marked elevation of serum aldosterone levels is present in all cases, and is associated with salt-wasting, hyponatraemia, hyperkalaemia and increased plasma renin activity. Clinical expression of the disease varies from severely affected infants who may die to apparently asymptomatic individuals (2 ).
Familial and sporadic cases have been reported. Inheritance is Mendelian and may be either autosomal dominant or recessive. The inheritance pattern appears to correspond to whether the hormonal insensitivity is renal or multisystem (3 ). Multisystem PHA1 is more severe with salt loss from all aldosterone sensitive end organs including salivary glands, sweat glands, colon and kidney, and is usually inherited as an autosomal recessive trait. The renal form of PHA1 is characterised by renal salt-wasting only and is usually inherited as an autosomal dominant trait (3 ). The majority of reported cases falls into one of these categories (2 ,3 ). There is a high incidence of consanguinuity in autosomal recessive PHA1.
The molecular basis of PHA1 is unknown (4 ). By analogy with other end-organ hormone insensitivity syndromes a defect at the level of the human mineralocorticoid receptor (hMR) has long been suspected, and the observation of reduced or absent binding sites for tritiated aldosterone in lymphocytes of patients with PHA1 (5 ) supported this hypothesis. The molecular cloning of cDNA encoding hMR (6 ) and localisation of the gene for hMR (MLR) to chromosome 4q31.1-31.2 has allowed direct genetic approaches to evaluating the hypothesis that PHA1 arises from mutations in MLR. Both direct sequence analysis of hMR cDNA (7 -9 ), and linkage analysis using three simple sequence length polymorphisms spanning MLR (10 ) have excluded MLR as the disease locus in some families including those analysed in this study. A genome scan for regions of homozygosity was therefore undertaken in 11 consanguineous PHA1 families in order to map the disease locus or loci. Homozygosity mapping is a powerful strategy for mapping rare recessive traits in children of consanguineous marriages (11 ,12 ), and has been successfully applied to a number of diseases (13 -15 ).
Evidence of genetic heterogeneity was obtained with families mapping either to chromosome 16p or chromosome 12p. Genes encoding subunits of the amiloride-sensitive epithelial sodium channel, SCNN1G, SCNN1B and SCNN1A map to these chromosomal regions (16 ,17 ). This strongly suggests that PHA1 is allelic to Liddle's syndrome of hypertension and pseudoaldosteronism, which has recently been shown to arise from mutations in SCNN1G and SCNN1B (18 ,19 ).
Affected individuals only were typed initially using about 200 microsatellite loci covering the autosomes at intervals of approximately 15 cM. Loci displaying apparent excess homozygosity (homozygous in four or more unrelated affected individuals) were further investigated by analysis of parents and subsequently by analysis of additional adjacent loci if indicated. A total of 341 marker loci were analysed. Multipoint analysis using the MAPMAKER/HOMOZ program revealed no evidence of linkage assuming genetic homogeneity. However direct inspection of raw allele data and the admixture test for locus heterogeneity applied using the HOMOG program revealed evidence of linkage to two chromosomal regions: 16p12.2-13.11 and 12p13.1-pter.
Six families showed evidence of linkage to 16p12.2-13.11. Using HOMOG analysis, a maximum admixture lod score of 4.4 was obtained with data from 15 loci in this region. Haplotypes and pedigrees of the six linked families are shown in Figure 1 a. Genetic and physical map data for these loci are shown in Figure 2 a. Regions of homozygosity ranged from less than 7 cM to 73 cM in affected individuals. The common region of homozygosity in these six families encompassed loci D16S412, D16S403 and D16S417.
These observations demonstrate that autosomal recessive pseudohypoaldosteronism type 1 is genetically heterogeneous and provide conclusive evidence for the existence of loci causing this disease on chromosome 16p12.2-13.11 and 12p13.1-pter. They represent the first application of homozygosity mapping to the direct elucidation of locus heterogeneity within such a limited resource of inbred pedigrees, and the first extension of the `positional candidate' approach to the implication of a multimeric candidate channel on account of the mapping of the corresponding disease loci to chromosomal regions encoding its separate subunits.
The genome screen for shared regions of homozygosity proved sufficiently powerful to detect unequivocal linkage despite the presence of locus heterogeneity. Highly significant statistical support for this conclusion was obtained. HOMOG3R analysis produced a likelihood ratio of 8.2 × 109 in favour of locus heterogeneity with all families accounted for by linkage to either 16p or 12p. It is recognised that linkage analysis in consanguineous families provides results which are highly sensitive to both disease and marker allele frequencies (20 ). Although allele frequencies for the loci analysed have been determined for particular family groups such as the CEPH family collection, the relevant frequencies are of course those for the population from which the disease families were drawn. The families analysed are from several quite different population backgrounds making it unlikely that a single frequency for any allele is applicable with any degree of certainty. A stringent approach was therefore adopted by setting a lower bound of 0.10 for all marker allele frequencies when multipoint analysis was carried out using MAPMAKER/HOMOZ.
High resolution mapping by this approach involves the identification of a common region of homozygosity amongst the unrelated affected individuals. On chromosome 16p this is defined by the telomeric locus D16S499 (Family 013) and the centromeric locus D16S420 (Family 003), separated by a genetic distance of about 8 cM. The latter boundary is of particular interest, as it excludes the candidate gene SCNN1B in family 003 if the assumption of a common ancestral disease allele is correct. If however this assumption is incorrect, the heterozygosity at [beta]ENaCGT-1 and D16S420 does not of course exclude this candidate gene. With either assumption SCNN1G remains a candidate, although no recombinations have previously been reported between SCNN1B and SCNN1G which are known to be physically within a 400 kb region. On chromosome 12p the common region of homozygosity is defined by the telomeric locus D12S93 (Family 009) and the centromeric locus D12S77 (Family 009 and 011), a genetic distance of 10 cM. Clearly, this region might have been substantially smaller had allele data at D12S356 been available in Family 009 or Family 004.
Linkage disequilibrium mapping may also allow refinement of the genetic localisation in a rare recessive disease if a founder effect is apparent. Unfortunately, the number of disease chromosomes available in this family resource is too small to allow allelic association to be identified unless association happens to be with a very rare marker allele. It is noteworthy that four out of six affected individuals are homozygous for the 181 allele at [beta]ENaCGT-1, but this allele is not significantly less frequent on normal chromosomes in this sample.
Following definitive exclusion of MLR, the strongest candidate gene for this disorder, a systematic genome search was undertaken without high priority being assigned to regions harbouring the several additional existing candidate genes. It was therefore interesting to find that PHA1 could be assigned to two quite separate chromosomal regions each of which harbour genes for sub-units of the amiloride-sensitive epithelial sodium channel. The recent identification of mutations in the genes encoding the [beta] and [gamma] subunits of this channel as the cause of Liddle's syndrome of pseudoaldosteronism (18 ,19 ), confirmed its identity as the functional mineralocorticoid-sensitive epithelial sodium channel in humans, and prompted the suggestion that mutations rendering this channel non-functional or unresponsive to aldosterone could cause PHA1.
These mapping data substantially strengthen what is already a quite plausible case for dysfunction at this channel as the cause of mineralocorticoid insensitivity in PHA1. Although receptor gene mutations have been identified in the majority of steroid hormone insensitivity syndromes, most notably those involving thyroxine, cortisol, androgens and Vitamin D, non-receptor defects are well recognised (21 -23 ). Abundant evidence from physiological studies indicate that the effects of mineralocorticoids on sodium exchange, especially in the kidneys, are exerted via a pathway that includes the channel encoded by SCNN1B, SCNN1G and SCNN1A (24 -26 ). The clinical features of Liddle's syndrome comprise in many respects a `mirror-image' of PHA1, indeed it is also designated pseudoaldosteronism. The underlying mutations which have been described in [beta]ENaC and [gamma]ENaC all cause truncation of the cytoplasmic carboxyl terminus of these subunits and cause constitutive activation of the channel. It is not unreasonable to propose that mutations elsewhere in these two genes, and in the [alpha]ENaC gene on chromosome 12p, may render the channel non-functional or unresponsive to mineralocorticoids. Although the localisation of PHA1 to two quite separate chromosomal regions each of which harbours genes for subunits of this channel could be coincidental, the mapping data in combination with the existing functional evidence make the suggestion that Liddle's syndome and PHA1 are allelic variants compelling.
Direct proof of this hypothesis requires of course the demonstration of mutations correlating with the disease state and work is in progress to screen these genes for mutations in affected individuals in these families. As discussed above, the close proximity of SCNN1B and SCNN1G render both plausible candidate genes in all chromosome 16p families with the possible exception of family 003. It is also conceivable that different alleles at these loci may account for sporadic cases of PHA1 and for cases in families displaying a dominant pattern of inheritance.
Elucidation of the molecular basis of pseudohypoaldosteronism is likely to be of broad biological interest. If, as anticipated, mutations are identified in these genes, analysis of their functional consequences will provide insights into the molecular mechanisms involved in the control of the amiloride-sensitive epithelial sodium channel by mineralocorticoids which is central to sodium and water homeostasis in mammals. Moreover, the close homology of the epithelial sodium channel to members of a family of Caenorhabditis elegans genes involved in mechanosensitivity and, when mutated, neuronal degeneration suggests that important processes related to cell volume control may also be illuminated (27 ).
Ascertainment of families was carried out through collaboration with the physicians who had reported cases of PHA1 in the literature and British paediatric endocrinologists. The 11 pedigrees included in this study are shown in Figure 1 . All parents are consanguineous being first cousins in nine of the 11 families.
Diagnosis was made according to standard clinical and biochemical criteria. All affected individuals had documented raised urinary sodium in the presence of hyponatraemia, hyperkalaemia, increased plasma renin activities, elevated serum aldosterone levels, and normal renal and adrenal function. Probands in seven of the 11 families (family 002, 003, 008, 009, 011, 012 and 013) have documented evidence of generalised involvement, i.e. raised sodium in sweat, saliva, or stool. In the other four families, the spectrum of organ involvement other than renal has not been fully evaluated. Details of families 003, 008, 009, and 012 have been reported previously (see 10 for references).
Genomic DNA was extracted from white cells by standard methods. DNA was amplified by the polymerase chain reaction (PCR) using fluorescently labelled primers or dUTPs (Applied Biosystems). The marker loci were selected from published genetic maps. Primers for a set of 200 polymorphic marker loci were supplied by the HGMP UK (28 ). Primer sequences are available from GDB. Additional loci on 16p and 12p were identified from published genetic and consortium maps (28 -31 ) except for [beta]ENaCGT-1, an intragenic polymorphism in SCNN1B (18 ), and D12S889 (32 ). The genes SCNN1Band SCNN1G lie within 400 kb of each other in the same YAC clone which also contains marker D16S420 (17 ). The same cosmid clone contained both D12S889 and SCNN1A which are within 40 kb of each other (32 ).
PCR was performed in 96 well microtitre plates (Hybaid). Each well contained 20-50 ng of genomic DNA; 1.5 mM MgCl2; 1 * reaction buffer (Advances Biotechnologies, UK); 200 µM each of dGTP, dATP, dTTP and dCTP; 50 ng of each primer and 0.2 U of Red Hot DNA polymerase (Advance Biotechnologies, UK), in a total volume of 15 µl. Thirty cycles were performed in a thermocycler (Hybaid OmnigeneTM). Alleles were separated on a 6% polyacrylamide electrophoresis gel for 3-4 h at a rate limiting voltage of 1000 V using a model 373A DNA sequencer (Applied Biosystems). Analysis of the alleles sizes was carried out by GENESCANTM672 (Ver 1.2) and Genotyper softwares (Applied Biosystems) using GENESCANTM-500 TAMRA or GENESCANTM-2500 ROX as size standards.
Linkage analysis was carried out assuming fully penetrant autosomal recessive inheritance with a disease allele frequency of 0.001. Multipoint analysis was performed using the MAPMAKER/HOMOZ program (20 ), which uses an algorithm based on Hidden Markov Models. This allows very rapid calculation of multipoint lod scores in small inbred families. This has previously been impractical because of the huge computational time required. The analysis on 16p was performed from D16S405 to D16S298 using the marker loci and genetic distances shown in Figure 2 a. The analysis on 12p was performed over the region shown in Figure 2 using the marker loci and genetic distances indicated. Allele frequencies were obtained from GDB with frequencies restricted to not less than 0.10. Homogeneity testing was carried out using the HOMOG and HOMOG3R programs (33 ,34 ). The HOMOG3R program calculates log likelihoods under the assumption that in a proportion ([alpha]1) of families a trait is linked to locus 1 and in a proportion of other families ([alpha]2) it is linked to locus 2, and the two loci are located in different regions of the genome such that the trait is never linked to both loci.
This study was made possible by the cooperation of patients and their families and the collaborating doctors including Dr M. Donaldson (Glasgow, UK), Dr D. M. Johnston (Peterborough, UK), Dr J. Kirk (Birmingham, UK), Dr P. G. F. Swift (Leicester, UK) and Dr E. Vamos (Brussels, Belgium). We are grateful to the Wellcome Trust, Action Research, the UK-Israel Science and Technology Research Fund and the Medical Research Council (UK) for financial support. The work of Drs A. and I. Hanukoglu is supported by a grant from the Israel Ministry of Health. R.T. is a Wellcome Trust Medical Graduate Training Fellow and J.S. is a Wellcome Trust Senior Research Fellow in Clinical Science. We thank the HGMP UK for supplying a set of 200 screening markers and Dr P. Marynen for primers of D12S889.
3 Hanukoglu, A. (1991) Type 1 pseudohypoaldosteronism includes two clinically and genetically distinct entities with either renal or multiple target organ defects. J. Clin. Endocrinol. Metab.73, 936-944.MEDLINE Abstract
4 Corvol, P. (1994) Editorial: the enigma of pseudohypoaldosteronism. J. Clin. Endocrinol. Metab.79, 25-26.
5 Kuhnle, U., Nielsen, M.D., Tietze, H.-U., Schroeter, C.H., Schlamp, D., Bosson, D., Knorr, D. and Armanini, D. (1990) Pseudohypoaldosteronism in eight families: different forms of inheritance are evident for various genetic defects. J. Clin. Endocrinol. Metab.70, 638-641.MEDLINE Abstract
6 Arriza, J.L., Weinberger, C., Cerelli, G., Glaser, T.M., B.L., H., Housman, D.E. and Evans, R.M. (1987) Cloning of the human mineralocorticoid receptor complementary DNA: structural and functional kinship with the glucocorticoid receptor. Science237, 268-275.MEDLINE Abstract
7 Zennaro, M.-C., Borensztein, P., Jeunemaitre, X., Armanini, D. and Soubrier, F. (1994) No alteration in the primary structure of the mineralocorticoid receptor in a family with pseudohypoaldosteronism. J. Clin. Endocrinol. Metab.79, 32-38.MEDLINE Abstract
8 Komesaroff, P.A., Verity, K. and Fuller, P.J. (1994) Pseudohypoaldosteronism: molecular characterisation of the mineralocorticoid receptor. J. Clin. Endocrinol. Metab.79, 27-31.MEDLINE Abstract
9 Arai, K., Tsigos, C., Suzuki, Y., Listwak, S., Zachman, K., Zangeneh, F., Rapaport, R., Chanoine, J.-P. and Chrousos, G.P. (1995) No apparent mineralocorticoid receptor defect in a series of sporadic cases of pseudohypoaldosteronism. J. Clin. Endocrinol. Metab.80, 814-817.MEDLINE Abstract
10 Chung, E., Hanukoglu, A., Rees, M., Thompson, R., Dillon, M., Hanukoglu, I., Bistritzer, T., Kuhnle, U., Seckl, J. and Gardiner, R.M. (1995) Exclusion of the locus for autosomal recessive pseudohypoaldosteronism Type 1 from the mineralocorticoid receptor gene region on human chromosome 4q by linkage analysis. J. Clin. Endocrinol. Metab.80, 3341-3345.MEDLINE Abstract
11 Smith, C. (1953) The detection of linkage in human genetics. J. Royal Stat. Soc. B.15, 153-184.
12 Lander, E.S. and Botstein, D. (1987) Homozygosity mapping: a way to map human recessive traits with the DNA of inbred children. Science236, 1567-1570.MEDLINE Abstract
13 Ben Hamida, C.B., Doerflinger, N., Belal, S., Linder, C., Reutenauer, L., Dib, C., Gyapay, G., Vignal, A., Le Paslier, D., Cohen, D., Pandolfo, M., Mokini, V., Novelli, G., Hentati, F., Ben Hamida, M., Mandel, J.-L. and Koenig, M. (1993) Localization of Friedreich ataxia phenotype with selective vitamin E deficiency to chromosome 8q by homozygosity mapping. Nature Genet.5, 195-200.MEDLINE Abstract
14 Houwen, R.H.J., Barharloo, S., Blankenship, K., Racymaekers, P., Juyn, J., Sandkuijl, L.A. and Freimer, N.B. (1994) Genome screening by searching for shared segments: mapping a gene for benign recurrent intrahepatic cholestasis. Nature Genet.8, 380-386.
15 Pollak, M.R., Wu Chou, Y.-H., Cerda, J.J., Steinmann, B., La Du, B., Seidman, J.G. and Seidman, C.E. (1993) Homozygosity mapping of the gene for alkaptonuria to chromosome 3q2. Nature Genet.5, 201-204.MEDLINE Abstract
16 Meisler, M.H., Barrow, L.L., Canessa, C.M. and Rossier, B.C. (1994) SCNN1, an epithelial cell sodium channel gene in the conserved linkage group on mouse chromosome 6 and human chromosome 12. Genomics24, 185-186.MEDLINE Abstract
17 Voilley, N., Bassilana, F., Mignon, C., Merscher, S., Mattei, M.-G., Carle, G.F., Lazdunski, M. and Barbry, P. (1995) Cloning, chromosomal localisaiton and physical linkage of the [beta] and [gamma] subunits (SCNN1B and SCNN1G) of the human epithelial amiloride-sensitive sodium channel. Genomics28, 560-565.MEDLINE Abstract
18 Shimkets, R.A., Warnock, D.G., Bositis, C.M., Nelson-Williams, C., Hansson, J.H., Schambelan, M., Gill Jr., J.R., Ulick, S., Milora, R.V., Findling, J.W., Canessa, C.M., Rossier, B.C. and Lifton, R.P. (1994) Liddle's Syndrome: heritable human hypertension caused by mutations in the [beta] subunit of the epithelial sodium channel. Cell79, 407-414.MEDLINE Abstract
19 Hansson, J.H., Nelson-Williams, C., Suzuki, H., Schild, L., Shimkets, R., Lu, Y., Canessa, C., Iwasaki, T., Rossier, B. and Lifton, R.P. (1995) Hypertension caused by a truncated epithelial sodium channel [gamma] subunit: genetic heterogeneity of Liddle syndrome. Nature Genet.11, 76-82.MEDLINE Abstract
20 Kruglyak, L., Daly, M.J. and Lander, E.S. (1995) Rapid multipoint linkage analysis of recessive traits in nuclear families including homozygosity mapping. Am. J. Hum. Genet.56, 519-527.MEDLINE Abstract
21 Andersson, S., Berman, S.M., Jenkins, E.P. & Russell, S.W. (1991) Deletion of steroid 5[alpha]-reductase 2 gene in male pseudohermaphroditism. Nature354, 159-161.MEDLINE Abstract
22 Weber, A. and Clark, A.J.L. (1994) Mutations of the ACTH receptor gene are only one cause of familial glucocorticoid deficiency. Hum. Mol. Genet.3, 585-588.MEDLINE Abstract
23 Wilson, L.C. and Trembath, R.C. (1995) Albright's hereditary osteodystrophy. J. Med. Genet.31, 779-784.
24 Duchatelle, P., Ohara, A., Ling, B.N., Kemendy, A.E., Kokko, K.E., Matsumoto, P.S. and Eaton, D.C. (1992) Regulation of renal epithelial sodium channels. Mol. Cell. Biochem.114, 27-34.MEDLINE Abstract
25 Linguelglia, E., Renard, S., Waldmann, R., Voilley, N., Chapigny, G., Plass, H., Lazdunski, M. and Barbry, P. (1994) Different homologous subunits of the amiloride-sensitive Na+ channel are differently regulated by aldosterone. J. Biol. Chem.269, 13736-13739.
26 Rossier, B.C., Canessa, C.M., Schild, L. and Horisberger, J.-D. (1994) Epithelial sodium channels. Curr. Opin. Nephrol. Hypertens.3, 487-496.MEDLINE Abstract
27 Hong, K. and Driscoll, M. (1994) A transmembrane domain of the putative channel subunit mec-4 influences mechanotransduction and neurodegeneration in C. elegans. Nature367, 412-413.
29 Kucherlapati, R., Craig, I. and Marynen, P. (1994) Report of the second international workshop on human chromosome 12 mapping 1994. Cytogenet.Cell Genet.67, 245-276.MEDLINE Abstract
30 Kozman, H.M., Keith, T.P., Donis-Keller, H., White, R.L., Weissenbach, J., Dean, M., Vergnaud, G., Kidd, K., Gusella, J., Royle, N.J., Sutherland, G.R. and Mulley, J.C. (1995) The CEPH consortium linkage map of human chromosome 16. Genomics25, 44-58.MEDLINE Abstract
31 Gyapay, G., Morissette, J., Vignal, A., Dib, C., Fizames, C., Millasseau, P., Marc, S., Bernardi, G., Lathrop, M. and Weissenbach, J. (1994) The 1993-94 Genethon human genetic linkage map. Nature Genet.7(Special issue), 246-338.
32 Baens, M., Aerssens, J., van Zand, K., Van den Berghe, H. and Marynen, P. (1995) Isolation and regional assignment of human chromosome 12p cDNAs. Genomics29, 44-52.MEDLINE Abstract
33 Ott, J. (1986) Linkage probability and its approximate confidence interval under possible heterogeneity. Genet. Epidemiol. (Suppl 1), 251-257.
34 Ott, J. (1991) Analysis of Human Genetic Linkage, revised edition. Johns Hopkins University Press, Baltimore.
*To whom correspondence should be addressed
+The first two authors contributed equally to this work
This page is maintained by OUP admin. Last updated Thu Oct 31 15:21:55 GMT 1996. Part of the OUP Journals World Wide Web service.Copyright Oxford University Press, 1996
