Targeted Comparative Sequencing Illuminates Vertebrate Evolution

Image: Courtesy of Elliott Marguiles PIPS ON PARADE: Researchers used a MultiPipMaker to show the alignments, expressed as percent identity plots, between a human reference sequence and several other species. This is a 20 kb region surrounding exon 2 of the MET gene. Gap-free alignable segments are represented as horizontal lines along the human reference sequence; the line's height represents the identity of that alignment. Aristotle envisioned humanity as the pinnacle of a "Great Chain of Being," a parade of organisms that evolved toward a state of undefined perfection. He might be surprised to learn that to understand how humans came to be, science needs to consult the lesser beasts, particularly their DNA, for genomes carry the ultimate fossils, the living records of shared ancestries, written in a simple four-letter alphabet. "Comparative genome analyses will be immensely important in identifying the protein-encoding elements in the human genome, and in providing insights into genome evolution," says Eric Green, chief of the National Human Genome Research Institute's (NHGRI) Genome Technology Branch at the recent Genome Sequencing and Analysis Conference. Green is also director of the Division of Intramural Research and the National Institutes of Health's Intramural Sequencing Center (NISC). Even more interesting may be the DNA sequences that do not encode protein, yet persist in modern genomes.1 The triumph of these sequences over natural selection suggests a vital regulatory function. A process called targeted comparative sequencing promises to provide some answers. Researchers are aligning sections of genomes from numerous species and are discovering noncoding assemblies among the protein-encoding, conserved sequences. At the forefront is the NISC's Comparative Sequencing Program, which is continually adding species as researchers probe a portion of human chromosome 7 for corresponding sequences, or orthologs. CLUES IN COMPARISONS "The question is, can we use sequence conservation information to identify functional elements?" posed Elliott Margulies, a research fellow on the program at a meeting of the American Society for Human Genetics. Targeted comparative sequencing is faster than deciphering entire genomes and provides different information, depending upon the species selected. Comparing human DNA sequences to orthologs in other primates, for example, can highlight how humans differ from them. Comparing more distant relatives reveals the basic toolkit for vertebrate life. For the NISC lab, the idea for targeted comparative sequencing grew out of Green's and his colleagues' longstanding efforts in analyzing human chromosome 7, the home for genes that cause cystic fibrosis, Williams syndrome, and more. "Human-mouse comparisons fed the fire to pursue the comparative study of additional species," Margulies relates. "We found that tools developed in the lab for comparative mapping between human and mouse could be easily extended to other species." Thus, by 2000, the lab's dissection of the human chromosome had evolved into a multispecies comparative mapping project. At the meetings, Green and Margulies presented data from 18 species. TOOLS NEEDED: BACS AND PIPS Targeted comparative sequencing requires BAC libraries and computational methods to align sequences and determine the extent of the match. A bacterial artificial chromosome, or BAC, is a cloning vector that can hold up to 200 kilobases (kb) of DNA. A genome that has been sheared and distributed into BAC clones constitutes a library. BAC clones are small enough to permit detailed peeks at specific chromosomal landscapes. "BAC libraries are best for comparative, targeted sequencing because it is not cost-effective to perform whole genome shotgun sequencing if only limited chromosomal regions are the focus of species comparisons," says Pieter de Jong, principal investigator at BACP(phage)AC Resources at the Children's Hospital Oakland Research Institute, one of four centers comprising NIH's BAC Resource Network. Investigators search BAC libraries for known highly conserved regions and then sequence the selected clones. The next stage is computational. The NHGRI program uses a Web server called PipMaker (http://bio.cse.psu.edu).2 A pip is a percent identity plot, a way of comparing orthologous DNA sequences. Developer Webb Miller, a computer science and engineering professor at Pennsylvania State University, explains: "PipMaker compares two or more genomic sequences of a megabase or less in length, and returns information about the resulting alignments in a variety of formats, ranging from graphical summaries of the matching regions to nucleotide-level details." A pip has a panel of dots that marks the degree of identity, usually between 50% and 100%, along a DNA sequence between two species. When plots are displayed in the order of whole-organism complexity, the molecular evidence for evolution is overwhelming. At the top, closest to the reference human sequence, the rain of dots is a downpour, diminishing to a drizzle down the chart as the phylogenetic distance from human increases. PipMaker differs from gene analysis tools, which detect protein-encoding portions of the genome, in that it charts sequence similarities without regard to function,3 such as by searching for open reading frames or expressed sequence tags. "PipMaker can locate protein-coding segments, but in well-annotated regions, the alignments are also used to study noncoding DNA, such as distinguishing regulatory sites from neutrally evolving regions," Miller says. Probing aligned sequences for conserved noncoding regions is also more expedient than experimentation that would involve inserting human genes into mice and then inferring conservation if the rodent expresses its transgenic baggage.4 Image: Courtesy of Elliott Marguiles COMPARE AND CONTRAST: This ideogram of human chromosome 7 shows the relative position of one of several targeted genomic regions that the NIH Intramural Sequencing Center is investigating. Depicted is a 1.5 Mb region at 7q31 containing 10 known genes, including CFTR, the gene mutated in patients with cystic fibrosis. A NOAH'S ARK OF BAC LIBRARIES As with the fossil record, more diverse BAC libraries will reveal more of evolution. Green calls the effort to expand the libraries "the Noah's ark approach." Simply put, if a sequence persists relatively unchanged from human to horse to hippo to hedgehog, it must do something. However, points out Margulies, species (such as humans and chimpanzees) may share a sequence because insufficient time has elapsed for mutations to have accrued. Consulting an eclectic menu of organisms makes this time concern less problematic. Deciphering the environs of the cystic fibrosis gene cftr demonstrates how sequence alignments can showcase evolution and discern distinctions. A 20 kb stretch corresponds in 94% of its DNA bases between human and baboon. A retroviral sequence found in humans but not baboons suggests that it wormed its way into the human genome after humans parted ways with a shared ancestor. For this 20 kb region, humans have less in common with cows, in which exons but not introns are conserved; even less with a mouse; and far less with the pufferfish Fugu rubripes. Alignments covering 1.5 megabases that include the cftr gene are even more telling: Three percent of the sequence is highly conserved. The species lineup is more precise than Aristotle's Great Chain of Being--cow, rat, and mouse are closer to the human sequence than platypus, which is closer than Fugu and chicken. Of the conserved sequence, 37% is protein-encoding and 63% is noncoding, says Margulies. Conserved noncoding sequences pepper this region, including the introns of cftr and wnt2, a gene associated with susceptibility to autism. Perhaps these findings will stimulate a closer look at introns, the sequences formerly known as junk. A complication of targeted comparative sequencing is that some genome regions diverge faster than others because mutation rates and selective pressures vary. Consider the homeobox (hox) genes, which control body plan in many species, and the locus control region (lcr) that oversees synthesis of beta-globin. The hox genes are ancient and remarkably similar in Fugu, shark, and chicken, whereas lcr has evolved faster and may be unique to mammals. BRING ON THE BACS The chromosome 7 dissection initially used a dozen BAC libraries of the usual suspects--rat, dog, cat, cow, pig, and zebrafish among them. But several dozen species will soon join the list. Awards for those candidates submitted to NIH in October are still top secret, says the NHGRI's Jane Peterson. Those species new to the NSF roster were selected for their evolutionary broadness and nonstatus positions as model organisms.5 The list is big on reptiles, which are poorly represented so far. At NISC, BAC libraries for vervet and squirrel monkeys, macaque, galago, and lemur will join the better-studied primates in revealing, by comparison, what makes humans characteristically human. Equally welcome additions are the marsupials and monotremes; as the most primitive mammals, these pouched and egg-laying cousins could be the missing links between placental mammals and other vertebrates. The evolutionary tales hidden in the human genome will yield to targeted comparative sequencing in a piecemeal fashion. For now, the key to ferreting out the stories within orthologous sequences is to amass BAC libraries. Fundamental to this process, however, is uncovering the best use of this wealth of genomic diversity in illuminating evolution. For example, perhaps only eight of the 32 vertebrates under scrutiny in the chromosome 7 analysis are sufficient to unveil conserved, noncoding, regulatory regions. "We expect different subsets of species to be important for different biological questions," sums up Margulies. And so, the evolution of targeted comparative sequencing may revolutionize the model species concept, in that the organisms that reveal the past are not chosen for their ease of study, but for the meanings hidden in selected portions of their genomes. Ricki Lewis (rickilewis@nasw.org) is a contributing editor. References 1. J.W. Thomas, J.W. Touchman, "Vertebrate genome sequencing: building a backbone for comparative genomics," Trends in Genetics, 18:104-8, 2002. 2. S. Schwartz et al., "PipMaker-A web server for aligning two genomic DNA sequences," Genome Research, 10:577-86, 2000. 3. R.C. Hardison, "Conserved noncoding sequences are reliable guides to regulatory elements," Trends in Genetics, 16:369-72, 2000. 4. G.G. Loots et al., "Identification of a coordinate regulator of interleukins 4, 13, and 5 by cross-species sequence comparisons." Science, 288:136-40, 2000. 5. J. Couzin, "NSF's ark draws alligators, algae, and wasps," Science, 297:1638-9, Sept. 6, 2002.

Targeted Comparative Sequencing Illuminates Vertebrate Evolution

Interested in reading more?

Become a Member of