"An important milestone has been passed by SBH, a new-generation technique of large-scale DNA sequencing." That'show the co-inventor of SBH, molecular biologist RadomirCrkvenjakov, welcomes the results of a double-blind test of thetechnique, which pitted his team at Argonne NationalLaboratory against that of Leroy Hood of the CaliforniaInstitute of Technology.
SBH stands for "sequencing by hybridization," an approach,Crkvenjakov told BioWorld, that "has a theoretical potential ofaccelerating large-scale sequencing by one or two orders ofmagnitude over present rates." SBH owes this promise, headded, "to the ease of automation and miniaturization of theconstituent biochemical steps."
As a holder of U.S. Patent No. 5,202,231, "Method of Sequencingof Genomes by Hybridization of Oligonucleotide Probes,"Crkvenjakov feels free to make these claims. Issued April 13,the patent protects SBH, with the Belgrade institute as assignee.Hyseq Inc., incorporated last year in Illinois to exploit theinvention, holds an option on licensing the patent and is nowseeking private capitalization.
Crkvenjakov's confidence rests on the results of winning thechallenge against Hood's molecular biologists at CalTech,"validating the SBH concept by sequencing without error threeDNA fragments totaling 343 base pairs, using octamers --oligonucleotides eight bases long."
SBH, Crkvenjakov explained, uses short, synthesized stretchesof DNA as probes to determine the presence of these sequencesin unknown DNA, and an optical scanner to identifycomplementary hybrids of probes with target DNA.Computational approaches then assemble the long sequencefrom lists of the short overlapping sequences found to becontained within it.
SBH has been germinating in the minds of nine Yugoslavscientists for the past six years at the Belgrade Institute forMolecular Genetics. In February 1991, Argonne's Biological andMedical Research Division invited this group of South Slavsequencers to visit the U.S. and put its SBH sequencingtechnique to the test. If successful, permanent positionsawaited some of them.
The "milestone" came last Friday, when Science published thedesign and results of their pivotal challenge from CalTech,titled "DNA Sequence Determination by Hybridization: AStrategy for Efficient Large-Scale Sequencing," withCrkvenjakov as a lead author.
The blind test involved homologous DNA fragments cloned intoan eight-kilobase vector. CalTech set the baseline data bysequencing, using traditional methods, three 2-kb insertscontaining related variable gene segments, with 92 to 94percent similarity. These came from four primate T-cellreceptor loci -- one human, two rhesus monkey and one gorilla.
Hood and his co-workers sent Argonne the first three of thesefour samples, in number-coded tubes, along with scrambledlists of octamer probes totaling 272 oligomers, including thegorilla DNA. The Yugoslavs, by means of SBH, were toresequence homologous 116-base-pair regions of these clones.
As Science describes the test conditions, instead of 65,635, (thenumber of possible permutations of an octamer's eight-basepairs times four bases) a reduced probe set, derived bytraditional sequencing, was synthesized. It had about twice asmany non-matching as matching probes in relation to each ofthe 116-bp targets.
Science informed its readers that "the probe informationderived from a fourth similar sequence (obtained from gorillaDNA, whose sample was not sent) was included in the final272-oligomer probe list. This information made the test morechallenging by increasing the number of complete sequencesthat could be obtained from the list."
Compounding this all's-fair complexity, Crkvenjakov toldBioWorld, "the material from CalTech contained two mislabeledsamples, in which sequences and codes did not correspond. Onesequence was right, the other wrong by 50 percent. When thecode labels were switched, the error disappeared.
"We hybridized 164 of the 272 probes provided, andcompletely reconstructed three sequences, out of 10,000possible. When the sequences of the three unknown DNAsegments of 116 or 111 bp from human and rhesus monkeysobtained by SBH were compared with those derived fromCalTech's gel-based traditional methods, they proved to beidentical."
This validation consolidated the job status at Argonne ofCrkvenjakov and his colleagues. They are now pursuing furtherR&D of SBH, in particular, miniaturizing the probes ontomicrochips, as a long-range goal.
A few hundred base pairs sent in to pinch-hit for the 65,536 inan octamer, Crkvenjakov explained, "by truthfully sequencinggive high statistical probability that the rest will do the same."SBH relies on such probabilistics more than strict biochemicalbean counting, he allows. "But the universe of biologicalsequencing is much narrower than the laws of chemistry," hesaid. "Most of the genes we are discovering in man are presentin lower organisms, constrained by evolution. Thus, subhumanprimate genomes don't differ from the human genecomplement by more than 10 percent."
As for the actual sequencing rationale, he explained that "anylinear sequence is an assembly of overlapping, shortersubsequences. Sequencing by hybridization (SBH) is based onthe use of oligonucleotide hybridization to determine the set ofconstituent subsequences (such as octamers) present in a DNAfragment."
Leroy Hood, chairman of molecular biology at the University ofWashington, Seattle, sees SBH as "a powerful, integrated, globalapproach to DNA fingerprinting for medical genetics." By wayof example, he told BioWorld: "If you have an array, say, ofoctamers, you can determine all of the 8-mer sequences thatexist in a given gene, such as cystic fibrosis, and readily focusdown on the differences between normal and mutant ones.Essentially, a single analysis can, hopefully, pick up a singlemutation across 250,000 base pairs of gene sequence."
Beyond the bedside, Hood perceives SBH as greatly increasingthe accuracy and efficiency, while decreasing the cost, of theHuman Genome Project (HGP) for sequencing all of the genes inall of the human chromosome complements. "The chips giveone the opportunity to very rapidly map DNA clones, DNAfragments, so this will facilitate enormously the creation ofoverlapping physical maps," he said.
But that's not all. "The second benefit for the HGP is that thesechips should allow one to take DNA from stretches of sequencethat have been done by conventional sequencing techniquesand quickly check them for errors."
"The success of the HGP," said Crkvenjakov, "will depend onwhether DNA sequencing approaches can greatly increasethroughput (at least 100-fold more than the current value of104 [10,000] base pairs per day per machine), and decreasecost."
And the success of SBH, Hood pointed out, depends on thesolution of "major technical problems. These include the cheap,inexpensive and flexible synthesis of those chips forminiaturization. Also, devising techniques that permit you tocarry out the hybridization analyses in a totally reproduciblefashion, to distinguish any perfect match from single-base ormore mismatches."
Companies are beginning to enter the futuristic field of large-scale gene sequencing, both Hood and Crkvenjakov observed.Leading the way is Affymax of Palo Alto, Calif., andAmsterdam, which has a major branch working on thistechnology. Beckman Instruments of Fullerton, Calif., isexpressing strong interest.
"These commercial efforts," Hood predicted, "plus the academicefforts, will push the technology right along to where, in Iwould guess, two or three years, we'll have chips that canreally begin to do some of these things."
-- David N. Leff Science Editor
(c) 1997 American Health Consultants. All rights reserved.