As a base for human transcriptome and functional genomics, we created the "full-length long Japan" (FLJ) collection of sequenced human cDNAs. We determined the entire sequence of 21,243 selected clones and found that 14,490 cDNAs (10,897 clusters) were unique to the FLJ collection. About half of them (5,416) seemed to be protein-coding. Of those, 1,999 clusters had not been predicted by computational methods. The distribution of GC content of nonpredicted cDNAs had a peak at ∼58% compared with a peak at ∼42%for predicted cDNAs. Thus, there seems to be a slight bias against GC-rich transcripts in current gene prediction procedures. The rest of the cDNAs unique to the FLJ collection (5,481) contained no obvious open reading frames (ORFs) and thus are candidate noncoding RNAs. About one-fourth of them (1,378) showed a clear pattern of splicing. The distribution of GC content of noncoding cDNAs was narrow and had a peak at ∼42%, relatively low compared with that of protein-coding cDNAs.

No abstract

We isolated two mouse cDNA clones which show significant similarities with human angiotensin-converting enzyme-related carboxypeptidase (ACE2). The cDNAs were 2746 and 1995 bp in length and seemed to arise from the same gene by alternative splicing. The longer cDNA encoded a 798-amino acid protein containing the sequence motif conserved among zinc metallopeptidases. Mouse ACE2 showed 83% identity with human ACE2. Northern blot analysis revealed that 2.8- and 2.0-kb transcripts were expressed mainly in the kidney and the lungs. The mouse ACE2 gene was mapped to chromosome X 70.5 cM.