top | item 13794582

(no title)

jasonkolb | 9 years ago

Is this using CRISPR or some other method of editing the DNA?

discuss

order

jfarlow|9 years ago

This treatment does not use CRISPR. It involves the hollowed out and repurposed lentivirus (similar in kind to HIV)[1]. The virus keeps it's own viral 'insert-into-DNA' machinery, but is stripped of its replication machinery as well as the code for its physical shell. The insert-into-DNA machinery is further hijacked so it can insert nothing but the DNA that encodes for the new mutated hemoglobin (HBB [T87Q][2]) that, when the patient's cell reads that new DNA will produce a new version of the protein. That insertion machinery with its new payload is loaded into a viral shell in a lab somewhere (again, its replication machinery and the code for new shells has been gutted).

When this new virus is given to the patients it does indeed infect the patients' cells. It uses its viral machinery to insert itself into the genome of the patient, more or less randomly - and that is not ideal. CRISPR systems are much newer and are being worked on right now, but the technologies you see in use in this article predate CRISPR. CRISPR will only speed up what was here a monumental (and slightly more risky) effort.

Regardless of how the code gets inserted into the genome (lentiviral in this case, CRISPR likely in future therapies), the instruction set to produce the new protein is not only capable of doing the job of the broken hemoglobin, but actually enables the broken hemoglobin to regain some of its function, likely coming very close to actually curing the patient.

In computer terms, someone with sickle-cell disease has a typo in the source code that leads to a buffer overflow error in the oxygen transport module. We found stuxnet can inject live code into a running OS, so we stripped it of its payload, it's ability to replicate but kept its injection capabilities and gave it our hot-fix as its payload. Our patch will be injected into billions of running nodes, inserting a ~500 line patch randomly into the each node's memory stack (yes, that's scary - but if a few of the nodes (cells) go down, it's not a horrible problem, and the current price of doing the patch at all... CRISPR can help here in future versions). That new code provides not only an alternative oxygen transport package, but this new package, so long as its running on >20% of the nodes, actually forces the original code's memory to periodically flush, thus de facto correcting the original typo's overflow bug - allowing both the new package and the old (kinda bug-fixed) package to both now be useful oxygen transport code. No more bug = cure.

The patients had a code regression, and we're applying not just a 1.0 fix, but a very real 1.1 patch on the human hemoglobin instruction set (randomly, into live code, on millions if not billions of cells, using modified HIV technology).

Diagram of the patch's entire injected instruction set (8.5kb): https://www.ncbi.nlm.nih.gov/core/lw/2.0/html/tileshop_pmc/t...

[1] Lentivirus: https://en.wikipedia.org/wiki/Lentivirus

[2] The 'updated' hemoglobin: https://serotiny.bio/notes/proteins/hbb/

fleetingmoments|9 years ago

Applying the patch directly on production, with no backups, too (!)

akanet|9 years ago

This was a great little writeup, and I appreciate you taking the time to make it.

ak39|9 years ago

Thank you for this brilliant and detailed explanation showing the differences between CRIPR and the virus method! I haven't seen anyone explain the concept of viruses in the language of buffer overflows! Delightful.

reimertz|9 years ago

Thank you for applying this to a mental model I can relate to. If there was an equivalent of Reddit gold, I would give you one.

splike|9 years ago

CRISPR is typically delivered via a lentivirus.