We studied the molecular characteristics of three naturally occurring variants in the human apolipoprotein B (apoB) signal peptide, their frequencies in non-insulin-dependent diabetic and random populations, and their association with several measures of lipid and carbohydrate metabolism. In a random sample of 197 French whites, there were two common alleles, 5'beta-SP-24 and 5'beta-SP-27, with frequencies of 0.35 and 0.65, respectively. In a random sample of 181 Mexican Americans, there was an additional allele, 5'beta-SP-29, with a frequency of 0.03. DNA sequence analysis indicated that the signal peptide alleles consisted of the following: 5'beta-SP-29 encoded 29 amino acids in the signal peptide containing two copies of the sequence CTG GCG CTG encoding Leu-Ala-Leu and a consecutive run of eight Leu-encoding condons; 5'beta-SP-27 encoded 27 amino acids with a run of only six Leu condons; 5'beta-SP-24 encoded 24 amino acids and contained a single copy of CTG GCG CTG and a run of six Leu codons. In the sample of French whites, average apoAl and glucose levels were significantly different among signal peptide genotypes. 5'beta-SP-24/24 homozygotes had higher apoAl levels than the two other signal peptide genotypes (1.59 vs. 1.42 g/L, respectively). Heterozygous 5'beta-SP-24/27 individuals had the highest glucose levels. In the random sample of Mexican Americans, average glucose levels were also significantly different among signal peptide genotypes. However, the rank order of average glucose levels was not the same between the two samples. In the sample of Mexican Americans, glucose levels were significantly elevated (6.14 mM) in the 5'beta-SP-24/24 homozygotes relative to the other genotypes. Correspondingly, average glycosylated hemoglobin levels were increased and C-peptide levels were decreased in homozygous 5'beta-SP-24/24 individuals. There were no significant differences in the frequency of the three signal peptide alleles between the random sample of Mexican Americans and a sample of 203 non-insulin-dependent diabetic Mexican Americans. The cause of the association (e.g., chance, linkage-phase disequilibrium, causation) between signal peptide-length variation and plasma glucose levels is not known.
