Richard Gayle
Welcome to the Machine* November 19, 1999
*for a musical interlude, try this. Needs quicktime. works best with headphones.
The nucleotide chemistry of Marv Caruthers resulted in machines that synthesized the oligonucleotides that have made PCR, sequencing and most of our DNA manipulations possible. I did my post-doc in Marv’s lab and I am probably one of the few people here who can claim to have made oligonucleotides by hand!! All the reactions were done on sintered glass funnels, of which we had 7 which worked. It took about 10 hours to make a 14-mer. This was the practical limit since by the end of the day exposure to volatiles made you as loopy as frat boy on a Friday night. Needless to say, the first ABI machines that automated this were a godsend and my liver is eternally thankful.
The third bit of technology ABI has been most valuable in providing has been DNA sequencing. Let me tell you how it used to be (I love telling war stories!). Although Sanger dideoxy sequencing was developed by the late '70s, it was hampered by the inability to easily get the short primers needed. A technique had been developed by Maxam and Gilbert using chemical cleavage of the DNA. Besides being laborious and inconsistent, this approach required the use of reagents such as DMS and hydrazine (normally used in rocket fuel). I also used another technique that was incredibly labor-intensive: two-dimensional electrophoresis/homochromatography. The first dimension of separation was done on a cellulose-acetate strip in a huge electrophoresis apparatus with an extremely flammable buffer (I'm not sure if it was Varsol or not but the Halon fire suppressant system and double interlock doors were very intimidating). And it could only do about 12 bases at a time. Needless to say, the automation of primer synthesis and subsequent automation of DNA sequencing has been a key factor in the molecular biology revolution.
ABI, now called PEBiosystems, has a new technology for sequencing, capillary electrophoresis, which may revolutionize the genome sequencing efforts. Interestingly, as with protein microsequencing and oligonucleotide synthesis, this technology is being associated with a standout researcher. However, the reasons he stands out stem as much from his contrary, gadfly approach to the way things are done as they do from his research. This scientist is Craig Venter.
Venter first startled the scientific community when, while at the NIH, he proposed sequencing cDNAs (Expressed Sequence Tags or ESTs, a term I believe he coined) instead of genomic sequences. That is, he wanted to only sequence the messenger RNA that was translated into proteins, not the genes themselves on the chromosome. Since large amounts of the chromosome do not code for anything (that we know of),this would get at the important parts (i.e. the coding regions) quickly. Since the Human Genome Project was still pretty tenuous politically at that point, people were worried. There was concern about losing focus on the reasons for genomic sequencing. Venter ignored this and, through him, the NIH started submitting patent applications containing 100s of ESTs. Although horrifying to a large part of the scientific community and being ultimately unsuccessful, it did serve to galvanize a lot of attention to this approach. Nowadays, ESTs are an extremely useful part of our arsenal of techniques.
Since leaving the NIH, Venter has concentrated on sequencing large quantities of randomly sized DNA pieces (shotgun sequencing) and then using computers to find overlapping sequences, allowing contiguous regions (contigs) of DNA to be fit together. The capillary electrophoresis equipment promises to increase the throughput. Shotgun sequencing contrasts with the accepted approach which creates physical maps of a chromosome first, then clones and sequences the accurately mapped pieces of DNA. The pieces can be put together based on the physical map. Shotgun sequencing is like putting a jigsaw puzzle together by examining each piece to see if a knob it possesses fits properly into another piece's hole and makes a nice picture. The approach the HGP is taking uses the picture on the jigsaw box to help determine where a particular piece fits, then matching it up with other pieces that fit in the same area. Which approach works best will, of course, be determined by whether an accurate picture of the puzzle exists or not, and just how fast you can do the comparisons. Venter's approach rapidly generates sequences but takes a lot of time assembling the complete sequence. The HGP takes its time constructing the maps but can be sure it will get a complete sequence eventually.
Venter claims that his new enterprise, Celera, using new sequencers from PEBiosystems, will complete the sequencing of the Human Genome by 2000, instead of 2003, the HGP endpoint. His shotgun approach can generate a lot of sequence and they are sequencing quite a lot more than just the human genome. They are now in the process of depositing the sequence for Drosophila. I'm not sure if there is an organism he will say "No" to.
One of the difficulties with a shotgun approach is orienting the overlaps or contigs. A physical map can help, since there are landmarks to help align different sequences. Shotgun sequencing does not wait for mapping so it has no landmarks, or so we would believe. Two recent papers, one in Science and one in Nature Genetics, deal with a novel technology which has tremendous potential in the generation of genomic maps. The fact that Venter's name is on these papers indicates that he does recognize the importance of this technology for his sequencing effort. The technology is called optical mapping.
This technology was developed in the lab of David Schwarz, now at the UW Biotechnology Center in Madison, Wisconsin. Using fluoroscopy they can visualize a single DNA molecule on an optical microscope. They have bound phage DNA, BACs, YACs etc. onto activated glass slides underneath coverslips. They then add restriction enzymes under the slide and clip the bound DNA molecules. Staining with the fluorophore YO-YO-1 allows the DNA molecules to be visualized. They map the DNA by simply measuring the distance between restriction enzyme cuts. Software programs then take all the data and construct contigs. Restriction maps without subcloning.
The paper in Science describes the shotgun mapping of the genome from Deinococcus radiodurans,a bacterium with a single 2.6 megabase circular chromosome and two smaller chromosomes, 415 and 176 kilobases in size. The DNA of this bacterium was isolated, placed on a microscope and cleaved with NheI. Here is a great figure, which shows one of these molecules as it traversed 6 microscope fields. Examining 157 molecules at an average depth of 35, generated enough data to allow a complete cleavage map of the 2.6 megabase chromosome to be manually constructed. This took several months. They also describe the development of a program, Gentig, which was able to replicate the same map in 20 to 30 minutes!!
The Nature Genetics paper is a jaw dropping tour de force for anyone who learned molecular biology on Sanger's knee. Conventional mapping of Plasmodium falciparum (the parasite responsible for malaria) has been hampered by several things. This organism has 14 linkage groups, presumably chromosomes but 5 of them (the blob chromosomes) can not be physically separated by electrophoresis. This makes the construction of sequence specific YACs difficult. This paper describes shotgun approaches to optically map the Plasmodium genome. But they do not separate out the chromosomes in any fashion before putting them on the activated glass slide. I repeat, they simply use the genomic DNA mixture containing ALL 14 chromosomes of Plasmodium and map them all at once!!
This paper is a molecular biologist's dream. Grow up Plasmodium.isolate the genomic DNA, mount it to a derivatized glass slide, digest with restriction enzymes and exam under a microscope. No cloning at all!!! Just use a microscope to collect data!!! Simply follow the DNA strand along, measuring the distance between cleavage sites. Use software to align the contigs. No work for me - the computer does it all! Pure shotgun mapping without ANY cloning! Generate detailed maps in a few MONTHS! Once the data were generated, the software assembled the contigs in only 20 minutes.
And the data that were generated were actually useful. They found 14 linkage groups, ranging in size from 676 kilobases to 3.4 megabases. They were able to resolve the restriction maps for all 5 blob chromosomes. Comparison with the fully sequenced chromosome 2 revealed only the absence of a 600 base pair BamHI fragment from the map. They accurately mapped 24 megabases using 2 restriction enzymes in less than 6 months.
The landmarks that a restriction map provides substantially accelerates the speed with which contigs can be aligned. It is no surprise to me that Venter has attached himself to this technology. Construction of chromosome specific clones and the generation of physical maps are two of the most time-consuming aspects of genomic sequencing efforts. Venter now has technology to circumvent both of these bottlenecks and can do everything via shotgun approaches. No need to clean anything up, just automate the data acquisition and let computers do it all. Welcome to the Machine.