442 0168-9525/98/$ – see front matter. Published by Elsevier Science. PII: S0168-9525(98)01553-4 L ETTER Sequencing of multiple complete genomes of bacteria and archaea makes it possible to perform systematic, genome-scale comparisons that aim to delineate the genomic complement of a particular phenotype. Recently, the first genome of a hyperthermophilic bacterium, Aquifex aeolicus, has been sequenced 1 . Previous studies based on rRNA and aminoacyl-tRNA analysis had suggested a very early divergence of Aquifex from the rest of the bacteria 2,3 . Aquifex is exceptional among bacteria in that it occupies the hyperthermophilic niche otherwise dominated by archaea 2 . In the published analysis of the Aquifex genome, it has been concluded that the genome sequence yielded ‘only a few specific indications of thermophily’ 1 . With three genomes of extreme thermophilic archaea (Methanococcus jannaschii, Methanobacterium thermo- autotrophicum and Archaeoglobus fulgidus) currently available 4–6 , we reasoned that a detailed comparison of the Aquifex and archaeal genomes could reveal genome-scale adaptations for thermophily. The protein sequences encoded in all complete bacterial genomes were compared with the non- redundant protein sequence database using the gapped BLAST program 7 , and a phylogenetic breakdown was automatically produced using the TAX_COLLECTOR program (Ref. 8, and D.R. Walker, unpublished). The results show that the fraction of Aquifex gene products that have archaeal proteins as clear best hits is by far greater than for each of the other bacteria (Table 1). Taking the fraction of ‘archaeal’ genes in Bacillus subtilis (Table 1) as a conservative estimate for the random expectation in a bacterial genome and using the normal approximation of the binomial distribution, it could be estimated that the excess of ‘archaeal’ genes in Aquifex could not be explained by a random fluctuation, with p<<10 -10 . A reciprocal comparison showed that, for proteins encoded in each of the three archaeal genomes, Aquifex proteins are the best hits significantly more frequently than proteins from other bacteria, even those with genomes 2–3 times larger than the Aquifex genome, such as Synechocystis sp. or B. subtilis (Table 2). In a complementary analysis, bacterial proteins were compared with Evidence for massive gene exchange between archaeal and bacterial hyperthermophiles TABLE 1. ‘Archaeal’ genes in bacterial genomes Bacterial species a Reliable best hits to archaeal proteins b Aquifex aeolicus 246 (16.2%) Bacillus subtilis 207 (5.0%) Synechocystis sp. 126 (4.0%) Borrelia burgdorferi 45 (3.6%) Escherichia coli 99 (2.3%) a The data on Haemophilus influenzae, Helicobacter pylori (Proteobacteria), Mycoplasma genitalium and Mycoplasma pneumoniae (Gram-positive bacteria) are not shown because, in these species, the majority of the best hits are to homologs from larger genomes within the same phylogenetic lineages, namely E. coli and B. subtilis, respectively. b All database hits with associated expectation (e) values <10 -3 were analyzed; a ‘reliable best hit’ was registered when the e-value with an archaeal protein was lower than that with any bacterial or eukaryotic protein by at least a factor of 100. M EETING R EPORT TIG NOVEMBER 1998 VOL. 14 NO. 11 releasing Sir proteins from the Ku70p–Ku80p telomerase complex (David Shore, Univ. of Geneva, Switzerland). Cdc13p protein binds single-stranded DNA at the Ku70p–Ku80p telomerase complex (Vicki Lundblad, Baylor, USA). Nuclear organization of telomeres is important with telomeres located at the nuclear periphery (Sussan Gasser, ISREC, Switzerland). Target- ting DNA to the periphery using a ER–Golgi anchoring signal can pro- duce silencing (Rolf Sternglanz, SUNY, USA). Hence, any gene brought to the nuclear periphery will be silenced by the Sir protein com- plex. In summary, the importance of chromatin structure was evident in all sessions. Yeast origins, cen- tromeres and telomeres bind elegant multiprotein complexes that act as regulatory machines to change chromatin structure and to allow important cellular processes to occur. Further reading 1 Dutta, A. and Bell, S.P. (1997) Annu. Rev. Cell Dev. Biol. 13, 293–332 2 Pluta, A.F. et al. (1995) Science 270, 1591–1594 3 Loo, S. and Rine, J. (1995) Annu. Rev. Cell Dev. Biol. 11, 519–548 4 Smith, J.S. and Boeke, J.D. (1997) Genes Dev. 11, 241–254 5 Weaver, D.T. (1995) Trends Genet. 11, 388–392 Robert A. Sclafani robert.sclafani@uchsc.edu Department of Biochemistry and Molecular Genetics, University of Colorado Health Sciences Center, 4200 E. Ninth Avenue, Denver, CO 80262, USA.