Nice basic guide to the three new-ish tools for rewriting Genes
February 11, 2014 3:27 PM   Subscribe

Zinc-finger-nucleases, TALENs, and CRISPR, oh my! The three tools, especially the last one, CRISPR, make rewriting Genes doable. Now the "fun" begins.
posted by aleph (14 comments total) 17 users marked this as a favorite
High-frequency off-target mutagenesis induced by CRISPR-Cas nucleases in human cells

If you are using these tools for anything significant, factor in the cost of sequencing a genome or three unless you're generating a perfect control at the same time.
posted by benzenedream at 3:46 PM on February 11, 2014 [2 favorites]

Its still kind of weird seeing CRISPR being presented outside of tiny phage conferences.
posted by Blasdelb at 3:48 PM on February 11, 2014 [4 favorites]

DNA normally compresses itself in a very compact way around histone proteins, something like a bunch of hair in a scrunchie. The different ways that DNA can "scrunch up" helps decide how some cells in your body make hair, other cells make heart muscle, nerve tissues, and so on. Our lab uses TALENs to excise specific and interesting regions of DNA, to see how the scrunchiness and functionality of the cell changes. Generally, TALENs are useful for cutting DNA in a very specific way. The excision is made on one of DNA's two strands. The idea is that TALENs make cuts that leave long hanging ends that bind to each other in a very accurate and stable fashion, like making such a clean cut of the ends of a pair of shoelaces that we're able to quickly and cleanly retie the cut ends.
posted by Blazecock Pileon at 3:53 PM on February 11, 2014 [1 favorite]

It's been possible (though expensive) to synthesize more or less arbitrarily-long sequences of DNA for a while now. Like Blazecock says, the idea is to build the sequence in parts (making long strands de novo all in one go is still beyond our capabioities, as far as I know) with complementary single-strand overhangs that will want to bind to each other when you let the fragments associate. We've already built a few entire genomes.that way, albeit as far as I know they've all been teeny tiny viral ones. Big enough for genes, though.

The difficulty has always been in getting it into the existing genome in a way that will enable the cell to actually use it. You have to unpackage the existing chromatin (supercoiled strands of DNA wrapped around histones and supported by other proteins), cut it open, insert your sequence, and then repackage the chromatin. You also have to make sure that your gene has a functioning promoter and activation/inactivation pathway, so that it will turn on and off when it needs to. All of that is a big pain in the butt, and turns out to be much more complicated than just sticking a bunch of nucleotides together. (Bacteria are comparatively easy because they can use plasmids, but eukaryotic cells like ours don't work that way.)

Traditionally we've done this with engineered viruses, some of which have naturally-evolved machinery for making functional modifications to host-cell genomes. It's still far from perfected, but it can be done to a degree. It's thought thay quite a bit of our own genomes are actually "fossilized" viral sequences, which were inserted sometime in the deep past but then stopped working for whatever reason (maybe just error, maybe deliberate cell defense) and continued to be replicated because it was selectively neutral. There's generally no significant cost associated with preserving noncoding DNA.

Anyway, my point is just that making genes is not so hard. The hard part is making them work.
posted by Scientist at 4:30 PM on February 11, 2014 [1 favorite]

In the cattle world there's some ongoing discussion about using TALENs to knockout the horned gene in cloned embryos made from elite bulls, which could move us from mostly horned dairy cows to mostly polled (hornless) cows in one generation. Why does this matter? Right now, we routinely dehorn heifers for the safety of the farmers and other animals using chemicals, which is messy and painful. It's not as sexy as inserting a new gene, but we have some use cases for that too, in the sense of repairing genes with loss-of-function mutations. Exciting times!
posted by wintermind at 6:42 PM on February 11, 2014 [2 favorites]

My friend Kevin wrote a wonderful (technical) review on CRISPR-Cas9 possibilities here. This tool will revolutionize genetics, biochemistry, selection based approaches to many many problems, you name it.

It's a Big Fucking Deal.

My bet is for the Nobel to go to this work in 4-5 yrs.
posted by lalochezia at 8:54 PM on February 11, 2014 [1 favorite]

I'm a bit confused about the therapeutic application of this. If someone has a problematic mutation, it exists in every cell in their body.

Even if the negative side effects of that mutation manifest only in certain tissues, isn't that still millions or billions of cells that would need to be edit in order to modify an adult animal?
posted by Alex404 at 2:04 AM on February 12, 2014

Alex, I'm pretty sure the only cells we care about are the ones that are expressing the relevant gene(s): All your non Hemoglobin expressing cells can have the broken copies that cause sickle cell anemia, but if we fix your bone marrow, you are no longer going to suffer symptoms. Oh, and we'll have to have at your gametes to prevent you from passing on the disease. That too..
posted by fFish at 3:19 AM on February 12, 2014

Alex404, if you do the editing in embryos then you can repair all cells, but that kind of engineering disturbs people when talking about human embryos. So, we'll do it in cows and pigs instead.
posted by wintermind at 5:39 AM on February 12, 2014

We've already built a few entire genomes.that way, albeit as far as I know they've all been teeny tiny viral ones.

A group led by Craig Venter did this in bacteria a few years back. It's a neat project but not very scalable at the moment.

MIT Technology Review makes a lot more sense when you think of it as promotional material for the university and its professors.
posted by euphorb at 7:04 AM on February 12, 2014

Very cute, but still has a ways to go. As Scientist above said, rewriting genes really isn't that hard.

To perform truly successful gene therapy one must:

1) rewrite the right gene(s) 1.5) in a way that you know for sure results in the gene working correctly 2) preferably in the right cells 3) in such a way that the patient's body will maintain it instead of one having to redo it every X months 4) and that doesn't kill the patient.

1) was solved years ago, with varying degrees of ease/efficacy. The *rest* are still the really hard parts, and I don't see this technology addressing it at all (which the Nature piece addresses). This is a VERY cool way to amp up 1), which is interesting and important, but I'm not celebrating just yet.

(Reading the Nature Perspectives article now - has someone who got farther figured out whether this will disrupt the methlylation/chromatin packing?)

/sorry sorry scientists are pessimists by nature and profession
posted by aperturescientist at 10:45 AM on February 12, 2014 [1 favorite]

To get gene editing in a live person in a reasonably safe manner you would expect either:
(a) very well characterized, reproducible off target sites that are validated to be harmless
(b) extremely low (~1e-06 offtarget events per onsite event) semi-random activity

It remains to be seen whether either of these goals is obtainable. That said, all of the above technologies are far superior to massive doses of lentriviruses disrupting 100s-1000s of sites per cell.
posted by benzenedream at 12:24 PM on February 12, 2014

Therapeutic uses is still far far off, unless maybe doing a modified adoptive transfer therapy where you swap a human designed/selected antibody into ex vivo peripheral blood B cells, activate them, then put them back inside the patient. But you could just infuse the patient with recombinant antibody in the first place.

The major impact I suspect will be, paradoxically, in the generation of transgenic animals. CRISPR/cas9 has much higher efficiencies than traditional homologous recombination.

As for selection, yeah, all the early hype has been "Hey lookee, we can do this! What can you do with this new toy/discovery too?" What you do is to give it some homologous donor DNA and induce the homology directed repair pathway and get recombination of something into the cut/lesion/deletion sight. Put something like a selectable marker for FACsorting or express a surface tag not normally expressed (some of the CDs from immune cells are great, especially if you kill/truncate their signal transduction domains). Select/isolate to grow up and screen a few colonies of clonal cells.

For actual application, the ideal would be to use an adenovirus since they have incredibly high transduction efficiencies, don't integrate and randomly mess around with the genome, and are diluted out through cell division or through degradation in about a week so the guide strand sequence that you feed the cas9 complex has enough time to do its intended thing to a lot of cells but not so much time/concentration as to make lesions at lower-efficiency non-/less-specific sites at frequencies worth worrying about. Especially if you design/choose good guide sequences.

At adeno efficiencies, for some applications, you might be able to get away with polyclonal pools of edited cells.

But working with adeno can be hard.

But whole genome sequencing is getting increasingly affordable.

There are other research uses for cas9; the disadvantages of lentivirus are moot for knockdown libraries. sgRNA/cas9 lesioning libraries are going to be much more effective than siRNA knockdown libraries. There's a couple of articles last month from the Lander and Zhang labs.
posted by porpoise at 10:16 PM on February 13, 2014

fFish - "but if we fix your bone marrow"

Ah, much better than my stupid B cell antibody off the cuff idea. There are lots of disorders where a certain myeloid lineage has one tiny problem in some pathway somewhere. Many of them are childhood disorders where life expectancy is very small. Harvest haematopoietic stem cells (much easier now that GM-CSF stem cell mobilization is routine after having been shown to be safe and as-good/better than marrow aspiration), edit them, reintroduce into patient. For some disorders myeloablative therapy (chemo/radiation) might not even be needed.

Offspring can be screened pre-implantation or maybe by the time offspring are going to reproduce even more efficacious technologies will have been developed.
posted by porpoise at 10:29 PM on February 13, 2014

« Older Every government is a liar. That's a prima facie...   |   The Sixth Extinction Newer »

This thread has been archived and is closed to new comments