Skip to Main Content

"Off-target Activity of Targeted Therapies Undergoing Clinical Trials in Cancer"

 .

"Off-target Activity of Targeted Therapies Undergoing Clinical Trials in Cancer"

February 02, 2022

Yale Cancer Center Grand Rounds | February 1, 2022

Presentation by: Dr. Jason Sheltzer

ID
7397

Transcript

  • 00:00And I want to introduce Jason Shelter.
  • 00:06Jason is an assistant professor of
  • 00:09surgery and received his PhD from MIT,
  • 00:12where he worked in the laboratory
  • 00:14of Doctor Angelika Amon in the Koch
  • 00:17Institute for Cancer Research.
  • 00:18After completing his PhD,
  • 00:20he established his own research
  • 00:22group as an independent fellow at
  • 00:24the Cold Spring Harbor Laboratory.
  • 00:26The Shelter Lab is broadly interested
  • 00:29in understanding the genomic changes
  • 00:31that drive cancer progression,
  • 00:32particularly aneuploidy,
  • 00:33which is found in more than
  • 00:3590% of human tumors.
  • 00:37Additionally,
  • 00:38they're working to identify genomic
  • 00:40alterations that create druggable
  • 00:42therapeutic vulnerabilities and cancer.
  • 00:44They have recently discovered
  • 00:46the first ever selective
  • 00:48inhibitor of the kinase CDK 11,
  • 00:50and developing CDK 11 inhibition as a new
  • 00:53strategy to treat malignancies without.
  • 00:56Further delay Jason all yours.
  • 01:01Thanks so much for the kind introduction,
  • 01:04so I'm very excited to be able to share
  • 01:06with you today some research my lab
  • 01:08has done about off target activity of
  • 01:11cancer drugs undergoing clinical trials.
  • 01:13These are my disclosures and this
  • 01:15project really comes from a journal
  • 01:17article that I read a few years ago
  • 01:19that had a statistic in it that I
  • 01:21found to be just absolutely shocking.
  • 01:23If you look at all drugs that enter
  • 01:26clinical testing and oncology,
  • 01:2897% of drug indication pairs that
  • 01:31enter clinical trials fail during that
  • 01:34testing and don't end up receiving
  • 01:36FDA approval and this 97% failure
  • 01:39rate for oncology drugs is the
  • 01:41highest of any field of medicine.
  • 01:43So more cancer drugs fail than
  • 01:45psychiatric drugs or endocrinology,
  • 01:47drugs, or infectious disease,
  • 01:49drugs or anything else.
  • 01:51And if you look at the proximate causes
  • 01:54for trial failure, the most common,
  • 01:57immediate causes that drugs run into
  • 01:59are toxicity and limited efficacy.
  • 02:01That is,
  • 02:02the drugs have too many side
  • 02:03effects for patients to safely take.
  • 02:05Or maybe the patients can safely take them,
  • 02:07but they have limited efficacy and they
  • 02:10don't actually shrink the patient's tumor.
  • 02:12And while these are kind of the.
  • 02:13Proximate causes for oncology
  • 02:15drug trial failure.
  • 02:17I think the underlying reasons
  • 02:19why so many drugs run into these
  • 02:21problems isn't very well understood,
  • 02:24and today I'm going to share some
  • 02:26evidence from my lab towards one
  • 02:28potential explanation for this high
  • 02:30failure rate and the hypothesis that
  • 02:32I'm going to argue for is that many
  • 02:35drugs are entering clinical testing and
  • 02:37oncology with an incorrect understanding
  • 02:40of their mechanism of action,
  • 02:42and I think this mischaracterization
  • 02:44of cancer drugs.
  • 02:45Maybe one factor by no means the only factor,
  • 02:48but one factor that contributes
  • 02:50to this extremely high failure
  • 02:53rate for oncology therapies.
  • 02:55So in my lab,
  • 02:56we've been interested in using
  • 02:58genetic approaches to investigate
  • 03:00the mechanisms of action of
  • 03:02different experimental cancer drugs,
  • 03:04and by searching through the current
  • 03:06literature and looking on clinicaltrials.gov,
  • 03:08we put together a list of 12
  • 03:11different drugs targeting 7 different
  • 03:13cancer related proteins that we
  • 03:15were interested in studying.
  • 03:16These drugs have been used in
  • 03:19about 30 different clinical trials
  • 03:21targeting several 100 cancer patients.
  • 03:23So six of these proteins are reported
  • 03:26to be cancer genetic dependencies.
  • 03:29That is the function of these proteins
  • 03:31is reported to be essential for the
  • 03:33growth and proliferation of cancer cells.
  • 03:36For instance,
  • 03:36pack four is a kinase.
  • 03:38It's been reported that Pack 4
  • 03:41kinase activity is essential for
  • 03:43the growth of colon cancer.
  • 03:44Lung cancer, breast cancer,
  • 03:46and a few other cancer types.
  • 03:48And because of that genetic data
  • 03:51concerning pack four that motivated.
  • 03:54Wiser to develop a small molecule
  • 03:56pack for inhibitor
  • 03:59PF 3758309 which they then
  • 04:01entered into clinical testing.
  • 04:03Caspase 3 is a little bit different.
  • 04:04I'm going to talk about Caspase 3 separately,
  • 04:07so we were interested in testing the
  • 04:09mechanism of action of these drugs
  • 04:11and seeing whether they killed cancer
  • 04:13cells through the inhibition of these
  • 04:15proteins and as a first step towards
  • 04:17this process we wanted to confirm that
  • 04:19the proteins these drugs were targeting
  • 04:22were truly cancer genetic dependencies.
  • 04:24That is, they were essential.
  • 04:25For cancer growth and to investigate this,
  • 04:28we set up a crisper competition
  • 04:29assay to see what happened when
  • 04:31we knocked these jeans out.
  • 04:33To do this CRISPR assay,
  • 04:35we transduced cancer cell lines
  • 04:37with cast 9 and then we transduced
  • 04:40them a second time with a guide
  • 04:42RNA coexpressed along with GFP.
  • 04:45This would then create a mixed
  • 04:47population of GFP positive cells
  • 04:49that had the guide RNA and caused
  • 04:51mutations in the target gene and then
  • 04:54UN transduced non fluorescent cells.
  • 04:56We then measure the percentage
  • 04:58of GFP cells over time.
  • 05:00If the percent of GFP positive
  • 05:02cells decreases overtime,
  • 05:03that tells us that whatever gene
  • 05:05the guide RNA is knocking out,
  • 05:07it must be required for cancer growth
  • 05:09because the GFP positive cells are dying.
  • 05:12In contrast,
  • 05:13if the percent of GFP positive
  • 05:15cells stays about the same,
  • 05:16then that's evidence that whatever
  • 05:18this guide RNA is targeting,
  • 05:20it isn't important for cancer
  • 05:22growth because these GFP positive
  • 05:24cells can grow just fine.
  • 05:25So that's what the assay looked like we
  • 05:28designed and cloned multiple guide RNA's.
  • 05:30Targeting each of the putative
  • 05:32cancer genetic dependencies we were
  • 05:34interested in studying and then we
  • 05:36did a bunch of competition assays
  • 05:38and this is what one of these
  • 05:40competition assays looks like.
  • 05:42So here we're in MD AMB 231 sells
  • 05:45a triple negative breast cancer
  • 05:46cell line as negative controls.
  • 05:48We have guide RNA's targeting
  • 05:50nonessential loci. Rosa 26 and eight.
  • 05:53The S1 guide RNA's targeting.
  • 05:55These genes exhibit no drop out.
  • 05:58As positive controls we have guide RNA's,
  • 06:00targeting the essential replication genes,
  • 06:03or PA3 and PC,
  • 06:04and a guide RNA's targeting these genes,
  • 06:07which are required for DNA replication.
  • 06:09Drop out between 50 fold and 200
  • 06:12fold over 5 passages in culture.
  • 06:15We then looked at the effects of guide RNA's,
  • 06:17targeting each of the putative
  • 06:19cancer genetic dependencies that
  • 06:20we were interested in studying,
  • 06:22and we were really astounded when the
  • 06:25guide RNA's targeting these cancer drug
  • 06:28targets exhibited no dropout whatsoever.
  • 06:30These guide RNAs behaved exactly
  • 06:32the same as guide RNA's,
  • 06:33targeting known non essential
  • 06:35genes like Rosa 26 and a VS1.
  • 06:39This was incredibly surprising to us
  • 06:41because right now there are patients who
  • 06:43are receiving anti htac 6 therapy and.
  • 06:45Anti milk therapy and anti Kim.
  • 06:47One therapy based on the belief that these
  • 06:50proteins are required for cancer growth,
  • 06:53but this experiment suggests that in
  • 06:56these experimental conditions in this
  • 06:58cell line we can eliminate these genes
  • 07:01without any effect on cancer whatsoever.
  • 07:04So this is what it looked
  • 07:05like in one cell line.
  • 07:06We ended up repeating this faceing 32
  • 07:09different cancer cell lines from more
  • 07:10than a dozen different cancer types,
  • 07:12and in each of these experiments
  • 07:14we got the same result.
  • 07:15There is no drop out of the guide RNA's
  • 07:17targeting these drug targets and there
  • 07:19was no evidence that any of these genes
  • 07:22were actually dependency in any cancer type.
  • 07:24So this made us take a step back and think,
  • 07:27well, is there something that
  • 07:28could be going wrong in this assay?
  • 07:29Could we be you know doing
  • 07:31something incorrect here?
  • 07:32And so we thought.
  • 07:34Well, maybe with CRISPR.
  • 07:36We're generating heterozygous
  • 07:37mutations but not homozygous mutations.
  • 07:40You know, maybe we're we're
  • 07:42introducing mutations into these genes,
  • 07:43but we're not really knocking
  • 07:45out the total protein.
  • 07:47So we thought OK,
  • 07:48instead of doing this population
  • 07:50based approach,
  • 07:51let's make single cell Dr Knockout
  • 07:53clones and be as sure as humanly
  • 07:55possible that we were really
  • 07:57eliminating 100% of the target protein.
  • 08:00So we did that.
  • 08:01We used A2 CRISPR guide RNA strategy
  • 08:03where we designed to guide RNA targeting
  • 08:06an upstream exon into downstream
  • 08:08exon so that we could physically
  • 08:10cut the gene out of the genome.
  • 08:12And there would be no protein left.
  • 08:14So we sorted a single cells
  • 08:16that were double positive.
  • 08:18That picked up both guide RNA's
  • 08:20that we transduced in and then we
  • 08:23verified target knockout using
  • 08:25two independent antibodies.
  • 08:26So for instance 1 gene we were
  • 08:29interested in studying as math K14.
  • 08:31This is the gene that encodes
  • 08:33the kinase P38 alpha.
  • 08:34We generated knockout clones and we
  • 08:37verified complete target knockout
  • 08:38using one antibody and then verified
  • 08:41it again using a second antibody.
  • 08:43So that we could be,
  • 08:44you know,
  • 08:45as sure as physically possible that
  • 08:47we had truly eliminated all trace of
  • 08:50these putative cancer drivers from the cell.
  • 08:53However,
  • 08:53when we tested the fitness effects
  • 08:55of these knockout clones,
  • 08:57we got exactly the same result
  • 08:58that we got from the competition
  • 09:01assays knocking out these putative
  • 09:03cancer genetic dependencies had
  • 09:05no effect on cancer growth.
  • 09:07So here, for instance,
  • 09:08is a proliferation assay in
  • 09:10a Melanoma cell line.
  • 09:11We have three map K14 knockout
  • 09:14clones and then two control rows
  • 09:16of 26 clones and these map K14
  • 09:19knockout clones grow exactly as
  • 09:20well as the rows of 26 control.
  • 09:23Jones,
  • 09:23we could also put these cells in soft
  • 09:26Agar challenge their clonogenic ability.
  • 09:28We saw no difference in
  • 09:31Clonogenic ability either.
  • 09:32These knockout cells grew just fine.
  • 09:35So to sum up,
  • 09:36a whole bunch of data that I
  • 09:37don't have time to show you.
  • 09:39We ended up eliminating all six
  • 09:41different cancer driver genes
  • 09:42that we were studying in at least
  • 09:45three different cancer types each,
  • 09:46and there was no fitness effect
  • 09:49whatsoever that we could discuss.
  • 09:51So this was a really strange
  • 09:53finding to us and it made us try
  • 09:54to figure out what was going on.
  • 09:56So we were looking at the
  • 09:57targets of 12 different
  • 09:59anti cancer drugs in various stages
  • 10:00of clinical development and we
  • 10:02looked at these drug targets with
  • 10:04multiple different CRISPR techniques.
  • 10:05We did CRISPR competition assays.
  • 10:08We made CRISPR knockouts,
  • 10:10but concordantly.
  • 10:11They both showed that we could eliminate
  • 10:13these jeans without a detrimental
  • 10:15effect on cancer proliferation.
  • 10:17This then raised the question well,
  • 10:19why were these genes believed to be
  • 10:22cancer essential in the 1st place?
  • 10:24And when we looked into the
  • 10:26literature on these genes,
  • 10:27we found the two main lines of evidence
  • 10:29had identified these genes as cancer,
  • 10:31essential initially.
  • 10:32The first line of evidence identifying
  • 10:35these genes as cancer essential were
  • 10:38experiments done using RNA interference.
  • 10:40The second line of evidence were
  • 10:42experiments done using small molecule drugs,
  • 10:45many of which had then gone
  • 10:47on to enter clinical testing.
  • 10:49So we wanted to see if we could
  • 10:51backtrack a little and understand why
  • 10:52we had come to such a different result
  • 10:55than these previous experiments done.
  • 10:57Using RNA I and small molecule drugs.
  • 11:01So I'll first show you what we
  • 11:02learned when we looked at some
  • 11:04of the prior RNA I experiments.
  • 11:06So this is an RNA I experiment
  • 11:08published in the literature a few
  • 11:10years ago that had identified the
  • 11:12kinase pack for as essential for
  • 11:14the growth of colon cancer cells.
  • 11:16In this experiment,
  • 11:17the investigators took SI
  • 11:19RNA's targeting pack four.
  • 11:21They introduced them into HCT 116,
  • 11:23colon cancer cells,
  • 11:25and they found that the SI RNAs decreased
  • 11:28colon cancer cell survival data like
  • 11:31this motivated Pfizer to enter a pack
  • 11:33for inhibitor into clinical trials.
  • 11:36We had found no fitness effect when we
  • 11:38had knocked out packed 4 using CRISPR,
  • 11:40so we wanted to see if we could
  • 11:42recapitulate this result that
  • 11:44had been published using RNA I.
  • 11:46Two of these SI RNA constructs were
  • 11:48commercially available and we had
  • 11:50HCT 116 cells growing in my lab,
  • 11:52so we purchased these siren's
  • 11:54from this prior publication and
  • 11:56then tested them in our cells.
  • 11:58We transfected these siren's,
  • 12:00the same from the prior publication
  • 12:03into HCT 116 cells,
  • 12:04and we could confirm by Western blot.
  • 12:06These SI RNAs decrease protein
  • 12:09expression as expected and we did
  • 12:12a self survival assay and we could
  • 12:14confirm that they killed HCT 116.
  • 12:16Colon cancer cells exactly
  • 12:18as had been reported.
  • 12:20However, using CRISPR,
  • 12:21we were also able to generate a
  • 12:24pack for knockout clone in this
  • 12:26exact same cancer cell line.
  • 12:28So here we had a pack for knockout clone.
  • 12:30You can see there's no pack for
  • 12:32expression in either the control
  • 12:34or the knockdown condition.
  • 12:36And then when we did a self
  • 12:38survival assay on these cells,
  • 12:40we found that transfecting the
  • 12:42pack 4 knockout cells with pack 4
  • 12:45targeting SI RNA had exactly the same
  • 12:48detrimental impact on colon cancer
  • 12:50survival as it did in the pack for
  • 12:53expressing Rosa 26 control cells.
  • 12:55So these packed 4 targeting SI RNAs
  • 12:58are killing colon cancer cells,
  • 13:00but their ability to kill colon
  • 13:02cancer cells is entirely
  • 13:03independent of the expression
  • 13:05of pack four because they're.
  • 13:06Exactly as lethal in the control cells
  • 13:09expressing pack four as they are in the pack.
  • 13:124 knockout clones that we
  • 13:14generated using crisper.
  • 13:15So this prior experiment was
  • 13:17was totally reproducible.
  • 13:18These sirens killed colon cancer cells,
  • 13:21but just the interpretation was wrong
  • 13:24because the toxicity of these Sir nase,
  • 13:26is just entirely independent
  • 13:28of pack for expression,
  • 13:30and this seems to be commonly
  • 13:32the case where we test SIRN as
  • 13:35and SH RNA's in the literature.
  • 13:37Over CRISPR derived knockout clones.
  • 13:39The SI and SH RNA's may kill cancer cells,
  • 13:42but it's just independent of the expression
  • 13:45of the gene that they were designed against.
  • 13:48The next thing that we wanted to
  • 13:49figure out was what was going on
  • 13:51with the small molecule drugs,
  • 13:52many of which had then gone on to enter
  • 13:55clinical testing and I'll show you
  • 13:56what happened with one of those drugs.
  • 13:59So pack one is a drug that was
  • 14:01described with few years ago in a
  • 14:03paper in nature chemical biology.
  • 14:05It was developed as a Caspase 3
  • 14:08activator compound so the apoptosis
  • 14:10enzyme caspase 3 is normally present
  • 14:12in an inactive procaspase state in
  • 14:15the cell and pack one was developed
  • 14:18to catalyze the conversion of caspase
  • 14:203 from its inactive procaspase
  • 14:22state to its active caspase 3 state,
  • 14:25at which point it would then
  • 14:27kill cancer cells in this drug.
  • 14:29Has been entered into three
  • 14:31different clinical trials.
  • 14:32However,
  • 14:32this mechanism of action was worked
  • 14:35out based on in vitro biochemistry
  • 14:37and no one had described a mutation
  • 14:40in Caspase 3 that conferred resistance
  • 14:42to it or had assessed the effects of
  • 14:45this drug in a Caspase 3 knockout cell.
  • 14:48So using CRISPR we generated multiple
  • 14:51Caspase 3 knockout clones and then
  • 14:53we did a dose response curve.
  • 14:55Examining the viability of wildtype
  • 14:57and Caspase 3 knockout clones in
  • 15:00different concentrations of pack one.
  • 15:03So this is what it looked like
  • 15:04for two control clones,
  • 15:052 clones expressing Arosa 26 guide RNA pack,
  • 15:09one is a potent anti cancer agent.
  • 15:12You can see it has an IC50 value
  • 15:14of around one or two micromolar.
  • 15:16However,
  • 15:16when we did the same assay in the Caspase
  • 15:203 knockout clones that we generated,
  • 15:22we ended up getting exactly
  • 15:24the same drug curve.
  • 15:25This drug is exactly as potent in caspase
  • 15:283 knockout clones as it is in caspase
  • 15:313 expressing Rosa 26 control clones.
  • 15:34It has an IC50 value of 1 to 2 micromolar,
  • 15:38regardless of whether these
  • 15:39cells express caspase 3,
  • 15:41so this drug,
  • 15:42which entered clinical trials as
  • 15:44a caspase 3 activating compound.
  • 15:46Its anti cancer activity actually
  • 15:49comes from something entirely
  • 15:51independent of caspase 3 and this
  • 15:53is actually the case for many
  • 15:54of the drugs that we studied.
  • 15:56So to show you a few more examples,
  • 15:58HDK 6 is a histone deacetylase
  • 16:01Celgene has developed each DAC.
  • 16:036 inhibitors sitter in a statin
  • 16:05richelain ISTAT.
  • 16:06We knocked out HDK 6 but we saw no change
  • 16:09in sensitivity to these
  • 16:11putative HDK 6 inhibitors.
  • 16:13Milk is a cancer related
  • 16:15kinase uncle therapy.
  • 16:16Science is developed this drug,
  • 16:18Novartis, developed this drug.
  • 16:20We use CRISPR to knockout milk.
  • 16:22We saw no change in sensitivity to
  • 16:25these milk inhibitory compounds.
  • 16:28So to sum up a whole bunch of data
  • 16:29that I don't have time to show you,
  • 16:31we found that target knockouts conferred
  • 16:33no resistance for 12 different
  • 16:35cancer drugs that we were studying.
  • 16:37We made these knockouts and did
  • 16:39these tests in at least three
  • 16:41different cancer types each,
  • 16:42so this kind of leaves us in an odd position.
  • 16:46We were studying 12 different preclinical
  • 16:48or clinical anti cancer drugs and
  • 16:50in each of these cases we found that
  • 16:53the reported mechanism of action.
  • 16:55Was actually incorrect.
  • 16:56This then raised the question well if
  • 16:59these drugs are killing cancer cells at
  • 17:01nanomolar or low micromolar potency,
  • 17:04how is it they actually work?
  • 17:05What is it they're actually targeting?
  • 17:07We wanted to see if we could figure
  • 17:08out how they were actually functioning.
  • 17:11We've had the best success so
  • 17:13far with one drug called O TS964.
  • 17:16This is what the drug looks like.
  • 17:18It was described in a paper in science
  • 17:20Translational Medicine a few years
  • 17:22ago as an inhibitor of a kinase called PBK,
  • 17:24which is also called Pop K in the literature,
  • 17:27but using CRISPR.
  • 17:28We knocked out PVK and we
  • 17:30saw no effect whatsoever.
  • 17:32On sensitivity to this compound
  • 17:34telling us that this drug O TS964
  • 17:37must have some other cellular target.
  • 17:40To see if we could figure out
  • 17:41what this drug was actually doing,
  • 17:43we used a genetic based approach
  • 17:46for this approach.
  • 17:47We took highly mutagenized
  • 17:49colon cancer cells, HCT 116.
  • 17:51They have a very high mutation rate
  • 17:53because they're microsatellite unstable
  • 17:55and then we expose these drugs to
  • 17:58a nearly lethal concentration of
  • 17:59O TS96 four such that about 99.9%
  • 18:03of cells on the plate were killed.
  • 18:06However,
  • 18:06there were a few stragglers that
  • 18:08remained when we cut these cells
  • 18:09in the drug for a period of weeks
  • 18:11until these cells were able to grow
  • 18:13and form little micro colonies.
  • 18:15We then subjected these cells to
  • 18:17whole exome sequencing and when we
  • 18:20did sequencing on the resistant clones,
  • 18:22what we were hoping to see was a
  • 18:24mutation that blocked whatever it was.
  • 18:27This drug was actually targeting.
  • 18:29Maybe these cells could survive a
  • 18:31lethal treatment because they had
  • 18:33some mutation preventing drug binding
  • 18:35to whatever O TS96 or was actually doing.
  • 18:38So when we did whole exome sequencing
  • 18:40on these clones,
  • 18:41we were really excited to see that
  • 18:43every clone that we looked at had
  • 18:46the same mutation in it.
  • 18:47Every drug resistant clone had a
  • 18:49mutation in the cyclin dependent kinase,
  • 18:51CDK 11.
  • 18:52They had a glycine to serine substitution,
  • 18:55right smack dab in the middle of
  • 18:57the CDK 11 kinase domain.
  • 18:59So this immediately suggested to
  • 19:01us that maybe this drug,
  • 19:02which had been developed as a PDK inhibitor,
  • 19:05was actually functioning through
  • 19:07inhibition of CDK 11.
  • 19:09Instead, one potential limitation
  • 19:11to this is that, well,
  • 19:13there actually isn't a precedent for this.
  • 19:14CDK 11 hasn't been previously dropped,
  • 19:18so we wanted to see if this mutation
  • 19:20actually had anything to do with
  • 19:22sensitivity to OTS 964 in order to do that,
  • 19:26we wanted to see whether this mutation
  • 19:28that we discovered in the resistance.
  • 19:29Jones was actually sufficient to
  • 19:32confer resistance to OTS 964.
  • 19:34To test this,
  • 19:35we used a CRISPR knockin strategy
  • 19:37where we introduced this glycine
  • 19:39to serine substitution that we
  • 19:41recovered in drug resistant cells.
  • 19:44We knocked it into drug naive
  • 19:46cancer cells and then tested its
  • 19:48effects on on O TS964 sensitivity.
  • 19:50This is what it looked like.
  • 19:53Here we have four different
  • 19:54cancer cell lines treated with a
  • 19:56lethal concentration of O TS964,
  • 19:58with a negative control guide RNA.
  • 20:00Or if we just cut in the CDK 11 gene,
  • 20:03we have no cancer cell viability.
  • 20:05But if we introduce a repair template that
  • 20:08includes the glycine to serine substitution,
  • 20:11then we can restore viability
  • 20:12in the presence of an otherwise
  • 20:14lethal concentration of O TS964.
  • 20:17So this tells us that this mutation is
  • 20:19in fact both necessary and sufficient
  • 20:22for resistance to this compound.
  • 20:24We then followed this up with
  • 20:26some biochemical assays.
  • 20:27We confirmed that O TS964 inhibits CDK 11.
  • 20:31With an IC50 value of around 40
  • 20:34to 50 animal or in vitro,
  • 20:36and we did a cell based target engagement
  • 20:39assay using mass spectrometry,
  • 20:41we found that 100 animal or
  • 20:43treatment with O TS964.
  • 20:45It didn't bind to hundreds of other
  • 20:48cellular kinases, but it bound.
  • 20:50It caused about 70% of binding site
  • 20:53occlusion for CDK 11, and only CDK 11.
  • 20:56So from this work we concluded that
  • 20:59by profiling a mischaracterized
  • 21:01anti cancer agent we were actually
  • 21:03able to serendipitously discover the
  • 21:06first selective inhibitor of CDK 11.
  • 21:10So to sum up what I told you so far,
  • 21:13we're kind of operating in a space
  • 21:15in which the vast majority of new
  • 21:17therapies that get tested in human
  • 21:19patients in oncology don't end up working,
  • 21:21and we put together a collection
  • 21:23of these drugs to study.
  • 21:24And one thing that we found while
  • 21:26studying them is that many of these
  • 21:29drugs have actually been designed
  • 21:30to target proteins that have no
  • 21:32detectable role in cancer growth.
  • 21:34Furthermore,
  • 21:35while these drugs do kill cancer cells,
  • 21:38they largely kill cancer cells
  • 21:39through off target effects rather
  • 21:41than through the target that they
  • 21:43were initially designed against,
  • 21:45and I think that this can increase
  • 21:46the burden of side effects and the
  • 21:48decrease the efficacy when some
  • 21:50of these drugs are actually used.
  • 21:52We don't truly understand how
  • 21:54they're working or where their anti
  • 21:56cancer activity comes from.
  • 21:58Think this conclusion has a number of
  • 22:00important considerations and caveats though.
  • 22:02For instance, there could be
  • 22:04unrecognized cell type specificity.
  • 22:06We did these competitions
  • 22:08in 32 cancer cell lines.
  • 22:10We generated knockout clones
  • 22:11in three cancer types each,
  • 22:13but it was,
  • 22:13you know,
  • 22:14physically, impossible for us to test
  • 22:16every subtype of leukemia or every subtype.
  • 22:20Kidney cancer in existence,
  • 22:21and so we can't fully recognize
  • 22:23rule out some unrecognized cell
  • 22:24type specificity that hasn't been
  • 22:26reported in the literature on these.
  • 22:28Targets. Secondly,
  • 22:30we specifically tested the hypothesis
  • 22:33that these proteins are required
  • 22:35for cell autonomous cancer growth,
  • 22:38that is, cells going from you know,
  • 22:39one cancer cell to 2:00 to 4:00 to 8:00,
  • 22:41and so on, and this had been reported for
  • 22:43each of the drugs that we had studied.
  • 22:45However, if it turned out that,
  • 22:47say, pack four had some role in
  • 22:50angiogenesis or in immune evasion,
  • 22:53or some other non cell autonomous process,
  • 22:55that wouldn't be ruled out for
  • 22:57the cell autonomous proliferation
  • 22:59focused assays that we've done.
  • 23:01I think a third important consideration
  • 23:03is while our data suggests that
  • 23:05these drugs are promiscuous and
  • 23:06may have multiple targets in the
  • 23:08cell just because a cancer drug is
  • 23:10promiscuous doesn't necessarily mean
  • 23:11that it will fail in the clinic.
  • 23:14There are a number of drugs like sunitinib,
  • 23:16Serafin, IB which do have multiple
  • 23:19targets in the cell.
  • 23:21And so,
  • 23:22just because something is promiscuous
  • 23:23doesn't necessarily mean that it will fail.
  • 23:25However,
  • 23:26I think that if our goal in cancer
  • 23:29biology is to kind of reach a plateau
  • 23:31of targeted precision medicine where
  • 23:33you sequence a patient's tumor,
  • 23:35you identify the mutations and
  • 23:38amplifications and alterations and
  • 23:39then design a drug cocktail based
  • 23:41on that particular genetic profile
  • 23:43in order to get to that level.
  • 23:45I think we need to have a really good
  • 23:47understanding of what drugs do and how
  • 23:50their anti cancer activity actually arises.
  • 23:52And what we'd suggest is that
  • 23:54pre clinical genetic validation,
  • 23:56particularly using CRISPR instead of RNA.
  • 23:58I may help us get genetic insight into
  • 24:01how anti cancer drugs work and may
  • 24:03decrease the number of investigational
  • 24:05drugs that enter clinical trials,
  • 24:07but end up failing during clinical testing.
  • 24:10So this is work that was done by my group.
  • 24:12In particular,
  • 24:13two really talented students
  • 24:14and Lynn and Chris Giuliano.
  • 24:16I'd like to acknowledge the
  • 24:18funding and thank you so much,
  • 24:19I'd be happy to answer any
  • 24:20questions that you have.
  • 24:25Thanks very much.
  • 24:26I thought that was really great.
  • 24:28I think you know one of the one of the
  • 24:32things we're all aware of is that when
  • 24:35we combine drugs that the toxicity
  • 24:37goes way up and you know of course,
  • 24:40much of the reason for that is
  • 24:42that many of these drugs are
  • 24:44promiscuous and are doing much
  • 24:46more than what we need them to do.
  • 24:49There's a there was a question a minute ago.
  • 24:53Uh oh, so the from from Jeffrey Townsend.
  • 24:58How were the original 12 drugs selected
  • 25:01and assembled for investigation?
  • 25:03Yep, so I didn't have time to discuss
  • 25:06that extensively in this talk,
  • 25:08but what we were interested in
  • 25:11our underlying hypothesis is that
  • 25:13the gold standard for knowing a
  • 25:15cancer drugs mechanism of action is
  • 25:17the identification of a mutation
  • 25:19that confers resistance to it.
  • 25:20The classic example here is Gleevec
  • 25:22and the mutations in BCR ABL.
  • 25:24Set block, Liebeck activity and our
  • 25:26thinking was that drugs that lacked
  • 25:28that level of genetic validation
  • 25:30were less likely to be acting
  • 25:32through an on target mechanism.
  • 25:34So we selected drugs that
  • 25:35specifically did not have that level
  • 25:37of genetic evidence behind them.
  • 25:41And from from Mike Hurwitz.
  • 25:45Sort of along that line.
  • 25:46Do you find it striking that every
  • 25:48single one of your targets was wrong?
  • 25:50Yeah, so for the sake of time,
  • 25:53yeah, for the sake of time,
  • 25:55I focused on the ones that were
  • 25:57where we discovered that the
  • 25:59mechanism of action was incorrect.
  • 26:02However, we did have a few examples
  • 26:04where we could validate it,
  • 26:06and I'm just trying to here.
  • 26:08I'm going to show just
  • 26:10one example of that now.
  • 26:12So this is not Lynn 3A.
  • 26:15This is a drug that's been reported to
  • 26:18function through P53 activation blocks.
  • 26:20The interaction between MDM two
  • 26:22and P53 we generated P53 knockout
  • 26:26clones using crisper and when we
  • 26:28did this drug sensitivity curve
  • 26:30we found that a nutlin has no
  • 26:33effect on the P53 knockout clones,
  • 26:35while it kills the P53 expressing
  • 26:37Rosa 26 control phones.
  • 26:39So in general so this is.
  • 26:41What we would expect for a drug that
  • 26:43acts for an on target activity.
  • 26:45You know a huge delta between the
  • 26:47target knockouts and the target,
  • 26:48expressing control clones,
  • 26:50and we found a few examples of this.
  • 26:54OK,
  • 26:54and I think this is the last
  • 26:57question from from Karen Anderson.
  • 27:00Did you make the searing mutant of CDK
  • 27:0211 and show that the inhibitor was no
  • 27:04longer effective in biochemical assays?
  • 27:07So we have been doing the
  • 27:09biochemical assays through ACR, oh,
  • 27:10at the moment, we are not skilled
  • 27:12in in vitro biochemistry ourselves,
  • 27:15and so we've just done it with
  • 27:18the the through the CR out and
  • 27:20we'd be glad to to launch the
  • 27:22collaboration to investigate that,
  • 27:23because I think that would be very powerful.
  • 27:26Well, I want to thank both Jason and Kurt.
  • 27:30It makes me proud to have these kinds
  • 27:33of presentations on my first day here.
  • 27:35So thank you very, very much
  • 27:38and we'll see you all next week.