In the search for diabetes genes, the combined approaches of positiona
l cloning with random markers and subsequent evaluation of candidate g
enes mapping to areas of interest win be increasingly used. For islet
candidate genes of unknown function, expressed trinucleotide (triplet)
repeats represent a unique subset. It is unlikely that abnormal expan
sion of expressed islet triplet repeats would be a major cause of diab
etes, yet the triplet repeats are frequently polymorphic and can thus
be used to map the genes in the human genome. In this study, a human i
slet cDNA library was screened with (CGG)(7) and (GAG)(7), and 23 trip
let repeats were isolated. Sequencing revealed four known and six nove
l islet genes containing 4-15 triplet repeats. The four known cDNAs in
cluded ferritin, the major iron-binding protein in cells; HSGSA2R, a f
ull-length clone of the alpha-subunit of the G-regulatory protein; HUM
SATB1A, a DNA-binding protein expressed predominantly in thymus; and H
UMPPAPRO, a ribosomal protein. The triplet repeats in ferritin and HUM
PPAPRO were found to be monomorphic. Characterization of the six uniqu
e novel expressed islet triplet cDNAs revealed that they were 0.6-1.5
kb in size, contained 4-15 triplet repeats, and were expressed in isle
ts and all other tissues examined. Four of the novel clones, CGG-isl 1
0, CGG-isl 11, CAG-isl 6, and CAG-isl 7, were mapped to human chromoso
mes 19, 16, 12, and 3, respectively, via somatic cell hybrids. One isl
et cDNA, CAG-isl 7, contained a repeat that was highly polymorphic, wi
th 14 alleles (4-18 triplets) in African-Americans (heterozygosity = 0
.86) and 6 alleles (heterozygosity = 0.77) in whites, Northern analysi
s indicated that the mRNA was abundant in pancreatic islets. A putativ
e full-length clone contained an open reading frame encoding 213 amino
acids with a variable number of alanines (4-18) within the COOH-termi
nal. The gene was uniquely mapped with odds >1,000:1 on chromosome 3p
in Centre d'Etude du Polymorphisme Humain pedigrees. There were no dif
ferences in CAG-isl 7 allele frequencies between African-American pati
ents with NIDDM (n = 108) and control subjects (n = 116), nor was expa
nsion above 18 repeats noted. Linkage analysis in 14 nonglucokinase ma
turity-onset diabetes of the young pedigrees showed a cumulative logar
ithm of odds score of -33.19 at theta = 0.00. Abnormal expansion was n
ot observed in 20 IDDM patients with one NIDDM parent. While these dat
a suggest no major role for CAG-isl 7 in diabetes, at least four of th
e six novel islet triplet genes are coexpressed in pancreatic islets a
nd neural tissue, and these genes can now be considered as candidates
for diabetes and/or neuropsychiatric diseases.