Colleagues of mine have begun making extensive use of 454 sequencing.
This is one of the largest (only?) commercial services for pyrosequencing-based sequence analysis of DNA. Pyrosequencing is one particular type of “sequencing by synthesis”, a cheaper, faster, and more parallelizable method for DNA sequencing than the traditional Sanger technique.
The technique works by exposing a growing DNA strand to deoxyribonucleotide triphosphates one-at-a-time. If the next base of the template is T, the next base of the newly synthesized strand should be an A. When dATP is exposed to the strand, it is incorporated into the new strand, which grows longer by at least one base pair. In the process, pyrophosphate is released, and other enzymes in the reaction mix convert the pyrophosphate into light. This is the signal detected in 454 sequencing. In contrast, when the strand is exposed to dCTP, dGTP, or dTTP, no incorporation is possible, and no light signal is generated.
One problem with sequencing by synthesis is homopolymers. Suppose the template has a stretch of Ts: TTTTTTTTAG, for example. When the reaction mix is exposed to dATP, incorporation and strand synthesis can happen all the way until the next non-T base. This generates a stronger light signal than incorporation of just one base. For short stretches of homopolymer, the strength of the light signal can be used to estimate how many bases were incorporated. But after a point, the noise overwhelms the signal, and it can be very difficult to tell CGTTTTTTTTAG from CGTTTTTTTTTAG, for example.
Last night, I was reading few papers in the history of developing pyrosequencing technology. What puzzled me is that right from the beginning, the early exponents of sequencing-by-synthesis seemed to have anticipated this problem and developed solutions. The most common approach was to put a cleavable tag on the 3′ position of the dNTP. Since polymerases require a 3′ hydroxyl group, and extra non-hydroxyl shizzle hanging off the 3′ end of a dNTP would mean that after incorporation of a single base, no new bases could be incorporated until the extra shizzle was removed. 3′ tags cleavable by light, reducing agents, or palladium, among others, were developed by various teams.
What happened to this technology? I don’t know why it isn’t used in 454 sequencing. Several possibilities:
1. Licensing issues. Are cleavable 3′-tagged dNTPs covered by intellectual property not available to 454 or Roche?
2. Problems with cleavage. As described in various publications, cleavable tags seemed to work well in the laboratory, but I wonder about the accessibility of cleavage reagents to the picotiter plates and emulsion bubbles used for modern 454 techniques. Does the lasers used for photocleavage of some tags effectively reach all parts of the bead surfaces and emulsion bubbles inside each picotiter plate well? Maybe platinum or reducing agent-based tags require reagents that are too expensive or diffusive poorly into the emulsion bubbles?
3. Other reasons?
It seems like better handling of homopolymers would be a great improvement of 454 sequencing technology. What don’t I know about cleavable 3′ dNTP tag technology which makes it unsuitable for fixing the homopolymer problem?