First, a quick refresh on Genia's proposed sequencing scheme, shown in Figure 1 from the paper included below. A polymerase sits atop a nanopore and extends a primer, using nucleotides which are tagged with a base-specific tag. As nucleotides are incorporated, the tag is cleaved and then captured by the nanopore, generating a base-specific signal. Tags are engineered to generate distinctive signals, simplifying (or so argue the Genia team) the problem over trying to detect native nucleotides. The system offers the hope of long reads with significantly lower error rates than have been achieved with other (PacBio, Oxford Nanopore) single molecule systems, with potentially the ability (like PacBio) to detect base modifications by the change in polymerase kinetics.
The current paper focuses on the polymerase engineering, generating a functional complex of phi29 polymerase sitting atop an alpha-hemolysin pore. The challenge is that the pore is a heptamer of subunits but the functional system needs precisely one polymerase and that polymerase should be positioned correctly so that the probability of a released tag entering the pore is maximized.
As a first step, the group developed a procedure in which a one subunit is tagged with a His tag, while the other six bear Strep-tags. This consists of mixing the different tagged versions in the appropriate ratio, then purifying complexes containing a single His-tagged subunit using ion exchange chromatography. A chromatogram in the supporting information (S3, shown below).
The other key bit of engineering is to dock a polymerase onto this complex. For this, the His-tagged subunit also contains another tag, the SpyTag, between the original C-terminus and the His-tag. The phi29 polymerase is engineered with a C-terminal SpyCatcher domain. I confess prior ignorance to SpyTag/SpyCatcher system, a nifty gift from nature. SpyCatcher is an immunoglobulin-like domain lifted from a bacterial pilin system which will specifically bind to SpyTag, which is followed by a covalent isopeptide bond between the two catalyzed by the SpyCatcher domain. The order of operations for building the complex here is to first form and purify the heptameric pore complexes with only one subunit tagged with His and SpyTag, then bind that to the polymerase via the SpyCatcher system. Linkers selected computationally are used to tie the SpyCatcher domain to the polymerase. The distance from the presumed tag exit site and the entrance to the pore may be as small as 46 Angstroms.
Once all this hardware is assembled, the pore still works. In Figure 3 below, the upper left quadrant shows the signal with just a pore present. Upper right adds in polymerase but no nucleotides; the authors note that the background signal increases, perhaps because the bound polymerase is changing the background gating characteristics of the pore. Lower left shows pore and tagged nucleotides but no polymerase; a new pattern of very short events is seen. Lower right is the Proof-of-Concept; with the entire system in place and incorporation of their derivatized G nucleotide, specific and prolonged signals are seen from the system. The ability of the complexed polymerase to function was also shown by using it to catalyze rolling circle amplification (Figure S4)
Interestingly, despite the effort to generate four distinct tags, the A and G tags are not well separated. The authors state that while their mean blockade levels are not distinguishable, the A has a characteristic 2-level signature, which I think refers to how most (but note not always) the A signal drops a bit more just before recovering.
I've enlarged the G and A signals from Figure 4 above to give a better shot at seeing the quirk in the A signal
Fed into a classification algorithm, the system achieves 98-99% success at callings bases if the true base is G, C or T -- but A remains troublesome with 14.4% of As incorrectly called as Gs, as shown in the confusion matrix displayed as Table 1 (and reproduced below).
Finally, the whole bit is put together to do some actual sequencing -- albeit with any homopolymers squashed to single bases. One A was missed for unknown reasons, but this shows promise.
Once all this hardware is assembled, the pore still works. In Figure 3 below, the upper left quadrant shows the signal with just a pore present. Upper right adds in polymerase but no nucleotides; the authors note that the background signal increases, perhaps because the bound polymerase is changing the background gating characteristics of the pore. Lower left shows pore and tagged nucleotides but no polymerase; a new pattern of very short events is seen. Lower right is the Proof-of-Concept; with the entire system in place and incorporation of their derivatized G nucleotide, specific and prolonged signals are seen from the system. The ability of the complexed polymerase to function was also shown by using it to catalyze rolling circle amplification (Figure S4)
Interestingly, despite the effort to generate four distinct tags, the A and G tags are not well separated. The authors state that while their mean blockade levels are not distinguishable, the A has a characteristic 2-level signature, which I think refers to how most (but note not always) the A signal drops a bit more just before recovering.
I've enlarged the G and A signals from Figure 4 above to give a better shot at seeing the quirk in the A signal
Finally, the whole bit is put together to do some actual sequencing -- albeit with any homopolymers squashed to single bases. One A was missed for unknown reasons, but this shows promise.
.
Obviously a next question is when can I get my hands on this? That really has two parts: a technical side and a business/legal side.
On the technical side, the Genia team (a term which includes their academic collaborators) appears to have demonstrated each piece of the system. While the papers each show data from a single monitoring electrode, both mention Genia's arrayed system (listed as 264 individually addressable electrodes in this paper). But having a laboratory setup is different than having shippable instruments and consumables. As the Oxford Nanopore team sometimes says, this involves shipping soap bubbles around the world. Now that Genia is owned by Roche, there is certainly a lot of muscle and expertise in this space, but Roche/Genia hasn't yet set a date for beta testing let along commercial launch. Perhaps that will be communicated soon.
On the business/legal side, there is the spectre of patent lawsuits. Oxford has frequently suggested they have a very strong patent position covering a wide range of methods for using nanopores for DNA sequencing, not just for the "strand sequencing" method they have successfully launched for hte MinION device. If Oxford isn't bluffing, then at some point they'll send off a lawsuit, and I may again steel myself to the unpleasant task of reading patents. Roche is a big player with extremely deep pockets and prior experience with such suits, but if Oxford really does have blocking IP that could stall launching. Or, it could be that even in the case Oxford has strong IP that an accommodation could be reached, by which a royalty is paid to Oxford on all products using Genia technology. Resolution of this question can only come after each side shows its hand with legal documents and then the process of litigating in the courts or settling outside is followed.
Keith, are you wiling to take a guess at how long it might be to get your hands on an early access device?
ReplyDeleteSimilar approaches have been taken before, I'd say it's somewhat similar to the Illumina/ONT 'exonuclease sequencing' in the fact that something gets clipped off a DNA base and transits through the pore to create the signal. Do you have any comments on how well the covalent attachment is likely to affect kinetics/dynamics - does PacBio offer some glimpse into the types of bias this method might have?
As with the first ONT MinION I'm sure this system can be improved, but it's unclear if Roche will give Genia that much time and money! Especially given the IP landscape...I'd like to see you write an article on that front!
So my feeling of briefly skimming the paper was that it add further details, but not much in terms of development over the previous publication.
ReplyDeleteDon't know about the IP position, it seems like a very different system to anything ONT have suggested.
My sense is that there is nothing in ONT's patent portfolio that can touch Genia. If there is a problem it is with PacB and phosphate labeled nucleotides.
ReplyDeleteAs an editorial note I am suspecting that if ONT sues Roche they will be labeled as a laudable case of David taking on Goliath and the calls to let the companies duke it out in the marketplace will be few and far between. Funny how that was not the case with the mspA dispute with Illumina. And how this and practically every other blog failed to discuss the outcome of that dispute in any detail, or cover the fact that ONT seems to have essentially admitted to infringing on the patent in question with everything up to and including the R7 chemistry
The patent in the ILMN suit was only valid in the US. Thrown out everywhere else. nowhere, including in the legal documents, can I find anything resembling an admission of infringement. As you must know, settling a case can present a better risk\reward profile than taking it all the way, isn't that at the heart of the US legal system? Perhaps, a REasonable person may also view the settlement as an admission by ILMN of ONTs counter case/defence.
ReplyDelete