Skip to Content

MELK Is Not A Cancer Target. Surprise!

The Maternal Embryonic Leucine zipper Kinase (MELK) is definitely an interesting enzyme. It’s been implicated in a number of cancer cell lines, and it also has important roles in the normal cell cycle, in embryogenesis, and other functions. It’s one of those proteins that’s found all across mammalian (and non-mammalian) species, with a lot of sequence conservation, but its roles in different species seem to vary more than you’d think.

The drug industry is interested because of the oncology connection – aggressive breast cancer cell lines and others are reported to have enhanced MELK activity, and their proliferation can apparently be slowed by siRNA knockdown. It’s also been implicated in glioblastoma, colon cancer, ovarian cancer and others. Several small-molecule inhibitors have been reported, and there are reports that they may ultimately work through a p53-based mechanism downstream. The one that made it to the clinic first is OTSSP167, from OncoTherapy Science in Japan, although its progress has not been as fast as anticipated, from the looks of it.

There was a report last year that the compound could well be hitting other kinase targets at the concentrations that it reaches in vivo, and that these might be responsible for some of its effects. That has certainly happened several times in the kinase-inhibitor world, although to be sure it happened more in earlier years. But now there’s a paper out that calls into question the whole idea of inhibiting MELK in the first place.

A group at Cold Spring Harbor has found that knocking MELK down (update – to be more accurate, completely knocking it out) via CRISPR treatment in a whole list of different cancer cell lines has no effect on their growth. Considering the earlier siRNA work, that’s something to sit down and think about. What’s more, these modified cell lines are still sensitive to the OTS compound, which certainly lends force to the earlier reports of off-target effects. The off-target effects aren’t just important, apparently – they’re the only thing the compound has to offer.

So at the very least, MELK monotherapy doesn’t look promising. But hey, the compound in the clinic isn’t a MELK monotherapy agent anyway, so it has that going for it. This is an embarrassing situation for OncoTherapy Science, I’d have to say, and doesn’t make their web site’s presentation on the compound look very robust. On the other hand, though, this is a very good example of target (de)validation through CRISPR, and I congratulate the Sheltzer group at Cold Spring for doing it. (Here’s an article with some background on the work).

It certainly raises a number of questions – for one, what was going on with those earlier siRNA experiments? That technique certainly can have some off-target effects of its own, and in this case those may have been the whole reason for the effects on cell growth. Another thing to consider is that there’s an X-ray structure of the OTS compound bound to MELK, which just goes to show you that that says nothing at all about what’s happening in vivo. The whole MELK field is going to have to figure out what’s going on, because I’d be willing to bet that many groups would have thought that CRISPR-ing it out would have been lethal all by itself. But apparently not, so it’s back to the drawing board for everyone.

23 comments on “MELK Is Not A Cancer Target. Surprise!”

  1. BK says:

    Depending on what part of the country you reside, MELK is also something you drink…

    1. PP says:

      drink plenty of MELK! now with vitamin R

  2. In vivo veritas says:

    So a germ-link knockout has no effects, but siRNA does? I’m not familiar with this field or target, but in mine, we would assume that there was a developmental compensatory effect in the CRISPR animal. Unless I’m misunderstanding, this would not deter many pharma groups……

    1. SW says:

      They did their CRISPR experiments in cultured cell line, not whole animals so it’s hard to see developmental compensatory effect playing a role.

      1. In vivo veritas says:

        Ahh…. thanks. I guess my biases (see user name) and this quote : “I’d be willing to bet that many groups would have thought that CRISPR-ing it out would have been lethal all by itself” were enough to make me think they were doing this in whole critters.
        Thanks for the clarification!

  3. anoano says:

    You should propose the verb “to CRISPR” to the Oxford dictionary!
    For the compound, MELK by itself didn’t have effect, but could the MELK activity + side effects be more potent than the just the side activity? The MELK activity would have a role, but not as a single target.

  4. anonymous says:

    This whole thing is a classic case of blind leading the blind!

  5. Mark Thorson says:

    Any idea what % of marketed drugs have putative mechanisms of action that are flatly wrong?

    1. Diver Dude says:

      To a first approximation? All of them.

  6. OldLabRat says:

    As far as I can tell, the OncoTherapy compound might be a better pan-kinome inhibitor than staurosporine, so ascribing the in vivo activity to something ( or several somethings) other than MELK inhibition makes sense.

  7. HTSguy says:

    Just for clarity here: the authors at Cold Spring Harbor did not “knock down” (ie reduce the expression levels of”) MELK. They used CRISPR-Cas9 to make a frameshift null mutation in the gene (one flavor of what is commonly called a knockout. Large deletions are a more definitive kind of knockout, but hard to make with CRISPR-Cas9). This removed any detectable expression of MELK.

    1. Barry says:

      Surely if the CRISPR-induced frame-shift creates a stop-codon, that’s a credible knockout? How is a large deletion mutant more credible?

      1. anon electrochemist says:

        Our knowledge of mRNA processing in eukaryotes is still pretty limited. Remember that 95% of our multiexonic genes use alternative splicing.

        1. Crispr says:

          Their knockingout efficiency (illustrated in Figure 1 and supplements) is amazing. CRISPR/Cas9 introduces a double-strand cut at targeted site, which is then followed by error-prone non-homologous recombination-mediated repair. The end is that you get a mixture of mutations: mis-sense, introduction of early stop codon, in-frame mutations, etc. Therefore, theoretically one cannot achieve complete knockout following lenti-crispr infection.

          Even more interestingly, they picked out sg sequence that is very close to the stop-codon. So if an early stop-codon is introduced in a uniform manner, there would be tiny deletions, and the protein should be still there on western blot ??

  8. kingly says:

    For what it’s worth, there is a 2016 PLoS paper that shows that shRNA knockdown of MELK does not promote MELK as a target (in the same cell line where OTSSP167 is potent).

    As has been said, this is a case where a company attributes a specific target to a compound that is highly promiscuous. It’s like saying ponatinib is an ABL inhibitor (along with half of the kinome).

  9. milkshaken says:

    do you remember c-Met kinase? So many companies jumped off that cliff. Pfizer shelved their broad non-selective c-Met series from SUGEN, only to re-discover them in LaJolla as ALK inhibitors and re-purposed the lead compound without additional medchem work as Crizotinib. A billion USD a year drug was born…

  10. gippgig says:

    There could always be translational frameshifting going on in MELK under certain circumstances in certain cells. Undetectable levels might also be sufficient to have an effect. That’s why a deletion is the gold standard for a knockout.

  11. hn says:

    I’ve long felt there needs to be more research done with broad spectrum kinase inhibitors. Some show surprisingly little toxicity.

    1. milkshaken says:

      Done! Both Sunitinib and Crizotinib are broad-spectrum cancer compounds. Sutent hits one third of human kinome with IC50 below 100nM

      I happened to work on glaucoma project – with Sutent as initial lead (it promotes retinal ganglion cell survival, completely unrelated to angiogenesis)

      1. hn says:

        With inhibitors hitting dozens, if not hundreds, of kinases, I would expect to see some useful phenotypes especially against different cancers. How extensively have drugs like sunitinib been tested against rare cancers?

  12. LiqC says:

    The OTSSP compound has a 2,6-dichlorophenol in it, which I rarely see in drug candidates (I don’t look at them as much). Is there anything alarming about chlorophenols in general? Dioxin/agent orange comes to mind. My hunch is no, as dioxins were a small impurity in a cpd produced as a mixture of isomers on very large scale.

    1. milkshaken says:

      triclosan is a polychlorinated phenol, and used to be ubiquitous in consumer products, including toothpaste – since it is a great preservative (while the “antimicrobial” could be advertised on the product). Chlorinated phenols smell really bad, and are persistent in environment

  13. Barry says:

    Med. Chemists spend a lot of effort trying to get rid of phenols in drug candidates, sometimes in vain. A phenol that makes three H-bond contacts (two as acceptor, one as donor) may be irreplaceable without a lot of loss in binding affinity/specificity. But the bias is that a phenolic drug will suffer oxidation (potentially to a toxic quinone) and rapidelimination (whether as the sulfate or glucuronide) faster than a bio-isostere. And plenty of drugs do fine even as phenols (and some even as quinones)

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.