Genome Economy

The Human Genome Project's discovery1 that the human body runs on an instruction manual of a mere 35,000 or so genes--compared to the worm's 19,000, the fruit fly's 13,000, and the tiny mustard relative Arabidopsis thaliana's 25,000--placed humanity on an even playing field with these other, supposedly simpler, organisms. It was a humbling experience, but humility quickly gave way to awe with the realization that the human genome might encode 100,000 to 200,000 proteins. Scientists base this number on the analysis of DNA sequences--called expressed sequence tags, or ESTs--that are reverse-transcribed from mRNAs. The question is, where is the information for all those extra proteins? The disparity between gene and protein number challenges the classic 'one gene/one polypeptide' paradigm that has governed molecular biology for decades. Still, some researchers decided long ago that the paradigm was simplistic--to them news of the paucity of protein-encoding genes in the human genome wasn't surprising. "It seems that gene number is not as important as gene interactions," says Xiangyin Kong, an investigator at the Shanghai Research Center of Biotechnology at the Chinese Academy of Sciences. The human genome sequence, particularly its repetitive nature, also rejuvenated interest in genetic mechanisms that underlie the evolution of biological complexity. Eukaryotes, with their myriad cell organelles and other compartments, are far more complex than the evolutionarily much older bacteria and archaeans. How does a large eukaryotic organism like a plant or animal extract the most information from its genome? Recent research reveals a number of strategies for genome economy, and some are indeed surprising. Introns, Exons, and Alternate Splicing Many genes in multicellular eukaryotes are discontinuous, comprised of exon sequences that encode the actual protein, and introns, whose transcripts are spliced out from a much longer initial mRNA. This discovery spawned the idea that a 'split' gene can encode more than one protein by splicing together alternate exon combinations.2 In effect, the cell shuffles exons, assembling a winning protein hand from assorted cards in its genetic deck. At least 10,000 human genes mix and match exons that encode specific protein regions, or domains.3 Even exons from different genes can come together. Tissue plasminogen activator, for example, is cobbled together from gene parts that encode plasminogen, epidermal growth factor and fibronectin.4 The human genome displays considerable internal redundancy--genes comprising families and superfamilies--so the number of sequences that are truly different is much lower than the total. Throughout the human genome, various genes employ exons that encode the same or similar protein domains; many genes are duplicated entirely. This repetition seems to be a hallmark of humans and plants, occurring for example to a greater extent in the human genome than in those of the nematode worm Caenorhabditis elegans, the fruit fly Drosophila melanogaster, or the yeast Saccharomyces cerevisiae. "Human genes have more protein domains, and these domains establish a complex interaction net," Kong says. Gene duplication aside, the extent and significance of exon shuffling in plants such as Arabidopsis remains elusive, according to molecular biologist Chris Somerville of the Carnegie Institution at Stanford University, because plant researchers have limited evidence. According to molecular biologist Jeffrey Palmer of Indiana University, "only a few cases of exon shuffling are known in plants, or more generally, outside of animals." A team led by Tatsuo Kakimoto of Japan's Osaka University recently reported alternatively spliced messages from an Arabidopsis gene encoding a receptor for cytokinin hormones.5 Alternate splicing also occurs in an mRNA transcript generated from two former mitochondrial genes now housed in the nucleus of rice.6 Complicating the case for alternative splicing in plants is the fact that researchers don't have as good a picture of the plant protein complement, or proteome. Says biology professor Scott Poethig of the University of Pennsylvania, "The bottom line is that the number of proteins that Arabidopsis makes is still unknown." An Intron, and Something More Elevated levels of prostate specific antigen, or PSA, are associated with increased risk of prostate cancer. The gene for PSA has five exons and four introns, but it also encodes a second, different protein, called PSA-linked molecule (PSA-LM). Both PSA and PSA-LM share the same signal sequence, which directs the proteins through the cell's secretory network of organelles. However, the rest of the PSA-LM gene is part of the fourth intron of the PSA gene.7 Studies of PSA and PSA-LM levels in healthy and cancerous prostate tissue suggest that they might function antagonistically in the body. "PSA-LM is abundant in normal prostate tissue and low when there is cancer, and PSA is just the opposite," explains Raveh Gill-More, director of corporate collaborations at Compugen Ltd., a genome firm based in Tel Aviv. PSA-LM turned up in screens of ESTs unique to prostate tissue. Gill-More thinks that PSA-LM escaped detection because it is relatively scarce in the serum of men with prostate cancer and because it does not share sequences with other genes in databases. PSA-LM is secreted in serum, semen, and urine. Its mRNA is transcribed with the same promoter (the RNA polymerase binding site at the start of a DNA sequence) as PSA, and Compugen researchers have raised antibodies against it, says Gill-More. His colleagues hope to further characterize the expression of PSA and its mate, and perhaps improve the current diagnostic test for prostate cancer by following levels of both proteins. "The combination can be a better marker than either alone," Gill-More believes. And the existence of PSA-LM reveals one way in which a gene can be more than the sum of its parts. "These are two proteins encoded by the same gene, with not even one amino acid [sequence] in common," he adds. Plant scientists are also searching for genes within genes, but the small size of most plant introns makes this a difficult endeavor. Says Stanford's Somerville, "unlike animal introns, plant introns are very small--100 to 5,000 bases--so the opportunity for genes in introns is much lower than in animals with big genomes." Palmer agrees, but he cautions that small introns are probably the case in most organisms. Says Palmer, "It's vertebrates, and a subclass of fly introns, that are exceptionally large when viewed from the perspectives of all eukaryotes--we have to watch that anthropocentric trap we all fall into from time to time." Penn's Poethig has another perspective: "We are working on a maize gene that has a 20 kilobase intron and a 10 kilobase intron, so [small intron size] is not universally true" in plants. Researchers know of at least one gene within an intron in rice. A functional rps14 gene encoding a mitochondrial ribosomal protein sits within an intron of another gene, for succinic dehydrogenase.6 Both genes reside in the nuclear genome, having moved 'horizontally' out of mitochondrial DNA. Palmer points to genes within mobile introns still contained in mitochondria and chloroplasts. However, "they encode factors involved in the metabolism of the intron itself, such as homing endonucleases, reverse transcriptases, and splicing factors." One Transcript, Two Proteins Just as PSA-LM might have evaded detection because it is rare, so too has a tooth protein been overshadowed by another polypeptide encoded by the same gene8, 9. In the inherited condition dentinogenesis imperfecta, teeth are discolored, oddly shaped, and have peeling enamel. The problem stems from a protein unique to dentin, the bone-like substance beneath the enamel that forms the bulk of the tooth. Dentin is a complex mixture of extracellular matrix proteins, of which 90 percent is collagen. It shares several of its proteins with bone, but two of them, called dentin phosophoprotein (DPP) and dentin sialoprotein (DSP), occur only in teeth. DPP is abundant, accounting for 50 percent of the noncollagen protein in dentin, and is the site of known mutations causing dentinogenesis imperfecta. DSP, on the other hand, accounts for only 5-8 percent of the noncollagen protein in dentin, which explains why researchers discovered it just a few years ago. Both proteins derive from a single mRNA transcript of a gene encoding DSPP, or dentin sialophosphoprotein, that maps to an area of chromosome 4 that contains several tooth-associated genes. The transcript is translated into an initial precursor protein, dentin sialophosphoprotein, which then is cleaved to yield DPP and DSP. Often when genes that encode two proteins of similar function map to the same chromosomal site, as DPP and DSP do, their amino acid sequences are very similar. It's evidence of gene duplication and subsequent divergence, a common evolutionary mechanism. But that's not the case for DPP and DSP. "They are not similar at the protein level, and there is no evidence that they arose from gene duplication," says Shanghai's Kong. "Two such proteins encoded by one gene is one mechanism of extracting maximal information from a genome." Other genes that encode as yet undiscovered tag-along proteins could help account for the low numbers of genes in genomes. When Antisense Really Makes Sense Antisense sequences are so named because they occur on the untranscribed or 'sense' side of a gene's double helix. However, like the phrase 'junk DNA,' calling a sequence 'antisense' might just reflect arrogance and ignorance on the part of scientists. Several years ago, researchers began wondering if antisense sequences regulate gene activity by binding to the complementary mRNAs encoded on the helix's sense side. They now know that antisense sequences can actually encode proteins. Consider the gene for neurofibromin, the tumor suppressor protein that is absent or inactivated in neurofibromatosis type 1 (NF1), an inherited illness that causes 'cafe-au-lait' spots on the skin and tumors beneath the skin. Within an intron of the neurofibromin gene, but encoded on the antisense strand of the DNA, are instructions for three other proteins. One, oligodendrocyte-myelin glycoprotein, might also control cell proliferation, although its connection to NF1 isn't clear. The other two intron-encoded genes are homologs of a mouse gene that causes myeloid leukemia. The mouse gene also acts as an integration site for certain viruses.10 Antisense transcripts seem to be important in plants too. A team led by David Baulcombe, head of the Sainsbury Laboratory at the John Innes Centre in Norwich, England, has been studying short pieces of antisense RNA in plants transformed with foreign genes and those responding to virus infection.11 The antisense molecules might play a role in a defense response called post transcriptional gene suppression, or PTGS, which inactivates invading genes. Baulcombe hypothesizes that an enzyme transcribing from an RNA template rather than from DNA synthesizes the antisense sequences. But does antisense play a role in uninfected, untransformed plants? Cautions Baulcombe, "Whether PTGS is a regulatory process in plant development is a controversial topic right now. Some people think it is because some PTGS mutants do not grow well. However, it is possible that the poor growth is a secondary consequence of having lost the defense systems." Trans-Splicing mRNA From Both DNA Strands Discovering the gene trio sequestered in the neurofibromin intron discredited the long-held idea that antisense sequences lack information. In recent weeks, it has also become apparent that organisms can use a variant antisense strategy: both sides of the DNA double helix can act as sense or coding strands to generate a single mRNA for a single protein. A gene in Drosophila that controls chromatin structure in early development, called modifier of mdg4,12 transcribes RNA from four exons on one strand and from two exons in the opposite orientation on the opposing strand. A small region of overlap allows the two mRNAs to hybridize, after which a kind of cut-and-paste operation links all of the transcribed exons into a single mRNA strand. Trans-splicing, as this operation is called, has been seen before in vitro as well as in viruses, and probably in mitochondrial and chloroplast genes too. "Cases that described nuclear genes found very low amounts of the trans-spliced product, and people never had a clear idea of whether the trans-spliced RNA was important for the cell, or just an accident," explains Victor Corces, a professor in the department of biology at Johns Hopkins University. The trans-spliced gene in the fly, however, is not an artifact, and it is vital. The gene actually can use its parts in different combinations to yield four proteins, according to Corces. The discovery has broad implications. "This mechanism can be used to make proteins with new functions by combining domains from two different proteins. This issue is interesting from an evolutionary point of view, and as a way to encode more proteins from smaller genomes," he adds. Corces wonders, however, how common this route to protein production might be. "I think it will happen every time you have one gene within an intron of a second one, and the transcription of one is opposite to the second. This arrangement is not unusual," he says. Investigators at The Institute for Genomic Research in Rockville, Md., are designing a computer program to detect this organization in genomes, Corces adds. "With respect to the several cases of trans-splicing known in plant mitochondrial genomes," notes Palmer, "I would bet, just based on how many are known and how readily plant mitochondrial genomes rearrange, that there are cases of different parts of the same trans-spliced gene being transcribed off different strands. You'd have to look at the sequenced mitochondrial genes of Arabidopsis and now sugar beet, however, to know this for sure." Overlapping Genes Perhaps one of the most efficient ways to economize on genetic information is to simply overlap genes--that is, start a second gene partway through an adjacent, upstream sequence. A research team led by Susan Douglas, senior investigator at the Institute for Marine Sciences in Halifax, Nova Scotia, recently sequenced the DNA in the nucleomorph of a cryptomonad alga called Guillardia.13 The nucleomorph is a remnant nucleus that survived internalization and enslavement of an ancestral red alga by another, nonphotosynthetic eukaryote--a form of endosymbiosis. The alga's chloroplast allowed the hybrid organism to make its own food, while its nucleus withered, losing many genes and sending others to the host nucleus. The nucleomorph, though petite, still functions with a highly slimmed down, compact genome of 551,000 base pairs on just three chromosomes. The DNA encodes 464 putative proteins, more than half of which have been identified. How did Guillardia pack information for these proteins into a supersmall genome? For one thing, it eliminated much of the noncoding sequences that intervene between genes in humans and other complex eukaryotes. Guillardia's genes are tightly spaced--one gene per 977 base pairs compared to one per 4,300 or so in humans and about the same in Arabidopsis. Guillardia also has few introns. Perhaps most fascinating to Douglas, though, were the 44 genes that overlap. Most of the overlaps occur in genes transcribed from opposite strands of the helix, but some are located on a single DNA strand. In other words, the genes share nucleotide sequence on the same sense strand. Overlapping genes are common in bacteria but far rarer in eukaryotes. Douglas thinks gene organization in Guillardia is sort of a "hybrid between eukaryotic and prokarytic." Along with overlapping genes, Guillardia's nucleomorph has distinctly eukaryotic features, like adding poly A tails to mRNA. According to Douglas, her work "shows how important it is to look at oddities in nature because they're full of surprises." With a peek at the first few eukaryotic genomes, researchers have new insight into the instructions needed to assemble nature's more complex organisms. Over the next few years, the list of sequenced genomes should grow by leaps and bounds. As they fill in gaps and compare notes, scientists are undoubtedly in for even many more surprises. Ricki Lewis (rickilewis@nasw.org) and Barry A. Palevitz (palevitz@dogwood.botany.uga.edu) are contributing editors for The Scientist. References 1. E. Russo, "Negotiating the human genome," The Scientist, 15[4]:16, Feb. 19, 2001. 2. W. Gilbert, "Why genes in pieces?" Nature, 271:501, 1978. 3. International Human Genome Sequencing Consortium, "Initial sequencing and analysis of the human genome," Nature, 409:860-921, Feb. 15, 2001. 4. T. Ny et al., "The structure of the human tissue-type plasminogen activator gene: Correlation of intron and exon structures to functional and structural domains," Proceedings of the National Academy of Sciences (PNAS), 81:5355-9, 1984. 5. T. Inoue et al., "Identification of CRE1 as a cytokinin receptor from Arabidopsis," Nature, 409:1060-3, Feb. 22, 2001. 6. N. Kubo et al., "A single nuclear transcript encoding mitochondrial RPS14 and SDHB of rice is processed by alternative splicing: Common use of the same mitochondrial targeting signal for different proteins," PNAS, 96:9207-11, 1999. 7. N. Heuze et al., "Molecular cloning and expression of an alternative hKLK3 transcript coding for a variant protein of prostate-specific antigen," Cancer Research, 59:2820-4, 1999. 8.S. Xiao et al., "Dentinogenesis imperfecta 1 with or without progressive hearing loss is associated with distinct mutations in DSPP," Nature Genetics, 27:201-4, February 2001. 9. X. Zhang et al., "DSPP mutation in dentinogenesis imperfecta Shields type II," Nature Genetics, 27:151-2, February 2001. 10. M.R. Wallace et al., "Type 1 neurofibromatosis gene: identification of a large transcript disrupted in three NF1 patients," Science, 249:181-6, 1990. 11. A.J. Hamilton, D.C. Baulcombe, "A species of small antisense RNA in posttranscriptional gene silencing in plants," Science, 286:950-2, 1999. 12. M. Labrador et al., "Protein encoding by both DNA strands," Nature, 209:1000, Feb. 22, 2001. 13. S. Douglas et al., "The highly reduced genome of an enslaved algal nucleus," Nature, 410:1091-6, April 26, 2001.

Genome Economy

Interested in reading more?

Become a Member of