Invited Editorial Priming the Search for HOX Mutations

The article by Kosaki et al. (p. 50) illustrates the power of the ever-expanding DNA sequence databases in greatly accelerating the development of tools for genomic mutation discovery. In this case the authors have developed coding sequence PCR primer sets for a large, highly conserved group of transcription factors, the HOX genes, whose full mutation spectrum in humans has not been elucidated. Due to their broad and important biological roles, it is anticipated that the availability of these sets will facilitate the identification of mutations or polymorphisms in HOX genes in a wide variety of phenotypes. HOX gene mutations in humans associated with specific defects have been described for only three of the 39 known genes, HOXD13, HOXA13, and HOXA11. Except for the newly described mutation in human HOXA11, this subject has been reviewed (Innis, ‘97; Veraksa et al., ‘00; Goodman and Scambler, ‘01). All of these human syndromes are associated with heterozygous HOX mutations that are inherited in an autosomal dominant pattern with almost complete penetrance. Two spontaneous coding sequence Hox mouse mutants have been identified (Mortlock et al., ‘96; Johnson et al., ‘98), and many of the Hox genes have been “knocked out” in mice via homologous recombination in embryonic stem cells. Mice carrying heterozygous, engineered null mutations for these genes often exhibit incomplete penetrance, which has been attributed in large part to functional overlap of HOX proteins to the growth and allocation of mesenchyme (Davis and Capecchi, ‘96; Fromental-Ramain et al., ‘96; Zakany et al., ‘97; Greer et al., ‘00). Stochastic variables and background genetic effects may also play a role and phenotypic differences with humans may reflect the nature of the mutations or variation between species. A brief glance at the human conditions caused by these mutations, as well as the available mouse models illustrates the future of HOX mutation searches. HOXD13 mutations in synpolydactyly involve an expansion of an endogenous homopolymeric alanine repeat with increasing severity related to increased expansion length, and mice heterozygous for Hoxd13 alanine expansions are very similar (Muragaki et al., ‘96; Akarsu et al., ‘96; Goodman et al., ‘97; Johnson et al., ‘98). A different, and distinct, limb malformation occurs with human intragenic HOXD13 deletions, suggesting that HOXD13 alanine expansions act through a gain-of-function (Goodman et al., ‘98). Hoxd13 engineered null mice exhibit malformations distinct from human intragenic HOXD13 deletion, suggesting that the roles of the genes in mice may differ in comparison to humans (Dolle et al., ‘93; Davis and Capecchi, ‘96). It is also possible that differences in this comparison result from some remaining function of mutant human HOXD13 proteins or unforeseen experimental consequences of engineered null alleles in mice. Sporadic or familial point mutations and alanine tract expansion of HOXA13 have been reported in hand-foot-genital syndrome (Mortlock and Innis, ‘97; Goodman et al., ‘00). A constitutional deletion of 7p14p15 causing a loss of the entire HOXA cluster including HOXA13 results in a phenotype similar to hand-footgenital syndrome in addition to other anomalies (Devriendt et al., ‘99). Therefore, except for a patient with a missense mutation in the homeodomain, which may lead to a gain-of-function (Goodman et al., 2000), haploinsufficiency of HOXA13 function is sufficient to observe hand-foot-genital syndrome. Similar defects to human HOXA13 haploinsufficiency were observed in the mouse mutant Hypodactyly, however, this mutation is a simultaneous loss and gain of function and is more severe than the engineered Hoxa13 knockout (Mortlock et al., ‘96; Fromental-Ramain et al., ‘96; Post and Innis, ‘99; Post et al., ‘00). A unique combination of radio-ulnar synostosis and amegakaryocytic thrombocytopenia has been reported for heterozygous HOXA11 homeodomain mutations in humans (Thompson and Nguyen, ‘00). Whether or not there are other skeletal manifestations in these patients remains to be determined, however, mice with engineered null alleles of Hoxa11 have a dissimilar skeletal phenotype and have not been reported to have thrombocytopenia (Small and Potter, ‘93). These data suggest that the human mutation may not be simply a null allele or that the function of HOXA11 in humans may be different compared to mice. How does this information impact human HOX mutation search strategies and interpretation? First,