Jonathan D. Licht, M.D. Professor, School of Medicine (Hematology/Oncology), Northwestern University Robert H. Lurie Comprehensive Cancer Center, Northwestern University
Friday morning session: Targeting Molecular Mechanisms for Personalized Therapeutics, Beckman Auditorium, 9:05 AM -- 9:50 AM
Abstract: Abnormalities in the epigenetic regulation of chromatin structure and function can lead to aberrant gene expression and cancer development. Epigenetic regulators can be affected by chromosomal rearrangement, point mutations, deletions and amplifications in cancer. The result may be changes in the function of specific transcription factors and pathways or more global, genome wide changes in chromatin modification and gene expression. Consequently, epigenetic therapies aim to restore normal chromatin modification patterns through the inhibition of various components of the epigenetic machinery. Specific transcriptional targeting of hematological malignancies was achieved through the treatment of acute promyleocytic leukemia. More recently specific therapies have been developed against Bcl6 in lymphoma and MLL proteins in leukemia. Histone deacetylase and DNA methyltransferase inhibitors represent the first putative global epigenetic therapies, however these agents have pleiotropic effects and it remains unclear how they lead to therapeutic responses. More recently, drugs that inhibit histone methyltransferases were developed, perhaps representing more specific agents. I will review emerging epigenetic targets in cancer and present recent models of promising epigenetic therapies.
Speaker bio: Dr. Licht is the Johanna Dobe Professor of Medicine, Chief, Division of Hematology/Oncology and an Associate Director (Clinical Sciences) of the Robert H. Lurie Comprehensive Cancer Center of Northwestern University. For over 20 years, Dr. Licht's laboratory has studied aberrant transcriptional regulation as a cause of hematological malignancy. This has included the molecular biology of acute promyelocytic leukemia and transcriptional mechanisms and targets of the PLZF transcription factor. He has studied chromatin changes and gene expression mediated by the MMSET protein overexpressed in a subset of multiple myeloma, as well as aberrant gene and miRNA regulation in myeloproliferative neoplasms, and most recently chromatin changes mediated by EZH2 in lymphoma. Dr. Licht is the Principal Investigator of a Leukemia Society Specialized Center of Excellence grant on chromatin mechanisms in hematological malignancies and Senior Scientific Leader of the Northwestern University NCI-Funded Physical Sciences Oncology Center, also centered on aberrant chromatin regulation. Dr. Licht is a Senior Editor of Clinical Cancer Research and serves on the editorial boards of Oncogene, Clinical Epigenetics and Cancer Biology and Therapy. Dr. Licht is a member of the Basic Sciences Committee of the NCI Board of Scientific Counselors.
Good morning, so my name is Katie Whalen, I'm a third year graduate student from the Department of Chemistry.
And this morning I have the great honor of introducing Dr. Jonathan Licht.
Dr. Jonathan Licht received his MD from Columbia University.
He currently serves as the joint adjunct professor of medicine in the chief division of hematology and oncology at Northwestern University.
He also serves as the associate director of the Robert H. Fogarty Comprehensive Cancer Center at Northwestern University,
as well as the principal investigator of the Leukemia Society Specialized Center of Excellence Grant.
In addition to this, he is the senior scientific leader of Northwestern University at the high-funded Physical Sciences and Oncology Center.
So on top of all those obligations, Dr. Licht has been running successful labs for over 20 years.
His lab studies amaranth transcriptional regulation as a cause of hematological illnesses.
They focus on the molecular biology of acute promyelocytic leukemia, as well as the transcriptional mechanisms of the PLZM transcription factor.
They're also interested in chromatin changes in gene expression mediated by the MF-SET protein and myeloma.
The title of this talk today is Understanding and Targeting Transcription.
Thank you very much for that kind introduction.
And I'm delighted to be here.
I want to compliment the organizers for your enthusiasm, for the great program you put together, and for the very interesting posters that we saw.
Lastly, I want to compliment all the grad students and postdocs who put up those posters.
I really enjoyed visiting with them.
So here we go.
So I'd like to start off with some background and some vignettes about aberrant gene regulation and hematologic malignancies.
And then we'll go into some of our own work, if the computer will cooperate.
I have some research support from Epizyme, and now my computer is quitting because of that.
I don't know.
Yeah, I'd say.
I don't know what it's doing.
Okay.
So we'd like to talk a little bit about epigenetics.
And like all of you, when I don't know anything, I go to Wikipedia.
And here's the definition.
It's a term biology used to refer to changes in gene expression stable over rounds of cell division and across generations, but do not involve changes in the underlying DNA sequence.
Another definition by Adrian Byrd is the structural adaption of chromosomal regions so as to register, signal, or perpetuate altered activity states.
So the fact is that we have one genome that's present in all of our cells, but only some genes are in one cell and other genes are on in another cell.
How does this happen?
It happens by the regulation of chromatin.
The DNA protein complex, also including RNA, which is a unit of gene regulation.
This was discovered 120 years ago by chemists like people in this room who discovered dyes that could stain cells.
And people looked, and they saw something that was colorful and densely colored.
That was chromatin.
Two types of chromatin.
U-chromatin, less densely packed, and these are expressed genes.
And heterochromatin, which are densely packed and not expressed.
The principal component of chromatin, of course, is the positively charged histone protein attracted to negatively charged DNA.
Now, there are a variety of transcriptional activation and transcriptional repression mechanisms.
I'll just touch on these very briefly.
One major concept is there are multi-protein machines, often megadaltons in size, that work in a concerted effort to turn on transcription.
With components such as CPP, P300, and PCAF.
The mediator complex, which is a multi-protein complex that decorates the RNA polymerase II and helps guide RNA polymerase II to the proper start sites of transcription.
And the SWI switch remodeling complex, which actually physically moves chromatin out of the way to allow gene transcription to occur.
Here are some of the counterparts of turning off gene transcription.
These include SYN3A complexes, SYN3A being a major cofactor that binds many transcription factors and brings them to a variety of complexes, including notably histone deacetylases.
The co-rest is a different complex, but has some of the same histone deacetylases.
Histone deacetylases, as you can see, are amongst all these complexes.
The NERD complex is a chromatin remodeling complex that helps move chromatin closer together, packs it in, doesn't allow it to be transcribed.
And the polycomb complex is a prominent complex in gene silencing, which has an important enzymatic activity encoded by the protein EZH2, enhancers S2.
And we'll talk about that in a few more moments.
The chromatin remodeling machinery actually uses ATP to hydrolyze ATP to allow DNA to loop around and move around on the nucleosomes seen here.
So nucleosomes can slide along the DNA, moving them out of the way of transcription start sites or, in the case of the NERD complex, bringing the chromatin onto promoters and shutting them down.
But this is a DNA-dependent process.
Also from Wikipedia, the fount of all knowledge, here's the so-called histone code.
What's been clear from the work of my colleague David Alice and many others over the past decade is that histones, the chromatin consists of DNA wrapped around a histone octamer.
But there are these tails that stick out from the histone octamer, which can be modified on multiple residues.
Histones can be acetylated, methylated, ubiquitylated, phosphorylated, proline isomylated, and other modifications as well.
These show the major sites of modification, the enzymes that can modify these sites.
So you can see here that there are literally thousands of different combinations of histone modifications.
Collectively, these are known as a histone code.
The histone code is written by enzymes.
For today, I'm going to focus on histone lysine methylation.
Histone lysine methylation is, histone lysine are very versatile residues.
Histone tail lysine residues can be acetylated, methylated, and when they're methylated, they can be mono-, di-, or trimethylated.
Each one of those methylations can mean something different to the cell.
This shows that there are evolutionarily conserved classes of histone lysine methyltransferases
that can put on marks that are associated with gene activation, such as histone 3-lysine 4-methylation.
As you can see here, these enzymes are conserved across evolution, and there are abnormalities of these histone methylation marks in disease.
Seen over here, histone 3-lysine 4-activation, histone 3-lysine 27-gene repression, and so forth down the line.
Now, these marks are written.
They're histone lysine methylations or acetylations that are there.
Are they just sort of changing the physical chemistry of chromatin?
Well, that may be part of it, but more likely, it's that these types of histone modifications are read by specific modules.
For example, the chromo domain specifically recognizes histone 3-lysine 9 trimethylation,
whereas the WDR5 domain recognizes histone 3-lysine 4-methylation.
So when a histone is modified on this residue, histone 3-lysine 9 tends not to be methylated on this residue,
so it will bring in a complex containing the chromo domain proteins.
And very often, the histone modifying enzymes themselves have a domain that recognizes the very modification they make.
For example, the WDR5 is part of an MLL, Mixed Lineage Leukemia, transcription complex.
MLL itself enzymatically makes the H3K4 mark.
WDR5 helps attract more molecules to chromatin, allowing the MLL to spread along the chromatin,
making histone 3-lysine 4 marks all along the chromatin in a self-reinforcing pattern.
So here's some rules, oversimplified rules, but rules nonetheless.
Gene activation generally associated with histone acetylation, histone 3-lysine 4 trimethylation.
Histone 3-lysine 9 and 27 are not methylated.
They tend to be acetylated, whereas repressed genes, you lose generally histone acetylation.
You lose this histone 3-lysine 4 methylation, and you gain K9 and K27 methylation.
Now, there are exceptions to these, but this is generally true.
Now, you can imagine that this is a very, you know, very complicated set of processes,
have to be well-coordinated across time, across space, across development, across differentiation.
And it comes then to no surprise that these types of processes can be disrupted in many forms of cancer.
Here's a couple examples.
Transcription factors can be affected in cancer.
For example, MLL fusion proteins, the MLL histone methyltransferases,
fuse to aberrant partners in many cases acute myeloid leukemia.
You can have overexpression transcription factors.
For example, the ERG transcription factor, a sequence-specific DNA binding factor,
is rearranged and overexpressed in over 85% of cases of prostate cancer.
So you can't really study prostate cancer without knowing something about what's going on here.
We've studied, for many years, acute promyelocyte leukemia.
I'll show you a vignette about this, which is associated with a fusion between the retinoic acid receptor,
a critical nuclear receptor that's activated by retinoic acid, fused to a heterologous protein, PML.
So these basic transcription factors can be shuffled, deleted, overexpressed, mutated in cancer.
Co-factors that bridge between the basic sequence-specific transcription factors and the general machinery can be mutated as well.
Recent sequencing of the genome, as discussed by Dr. Weinstock yesterday,
in many forms of cancer, including lymphoma, have shown CBP and P300 mutations.
This would lead to a deficit of histone acetylation in these cases of lymphoma.
How that causes disease, we don't yet know.
But it's very striking from the recent genome sequences that have come out over the past two years
how frequently chromatin regulatory machinery is mutated in human cancer,
including mutations in the chromatin remodeling factors, such as ISY mutations in rhabdoid tumors,
certain types of connective tissue tumors, and the ARID1A mutation,
which was initially found, I believe, in ovarian cancer, but was now found almost everywhere.
Very important for chromatin remodeling and seems to be almost a ubiquitous mutation.
Intriguingly, in a paper in Nature two months ago, there's three types of histone 3s, histone 3.1, 2, and 3.
Histone 3.3 itself is mutated in human cancer.
That's the basic chromatin itself.
So you have a barren nucleosome with a barren histone 3.3.
It can no longer be histone 3 lysine 27 methylated or acetylated.
It's that 27 residue that's mutated.
So every bit of this machinery can be affected in human cancer.
Can we target this?
Can we somehow affect these processes for the treatment of human cancer?
Well, we're already doing it in the case of acute promyelocytic leukemia.
The retinoic acid receptor is a bistable switch.
It can be decorated with co-activator or co-repressor proteins.
When present on a promoter in the absence of retinoic acid or at low levels of retinoic acid found in our bodies,
it oscillates between a co-repressor-bound state and a co-activator-bound state.
This will control gene expression and sort of act as a rheostat to encourage or discourage the differentiation of fully mature granulocytes from hematopoietic stem cells.
So this is really a tightly balanced process.
It's interesting that the retinoic acid receptor is not required for white blood cell differentiation but acts as a volume control.
Well, what happens in acute promyelocytic leukemia is the PML-RAR protein is formed due to a chromosomal translocation,
and this seems to be stuck in the off state.
It's bound to co-repressors even at physiologic levels of retinoic acid.
It doesn't oscillate into that activated state.
Only when you add pharmacologic doses of retinoic acid does the co-repressor leave, the co-activator come on, and gene transcription ensues.
But you need to add 100-fold more retinoic acid, 10 to the minus 6th molar rather than 10 to the minus 8th molar, to get this reaction to go.
But this reaction does go.
And acute promyelocytic leukemia is one form of acute leukemia generally found in adults,
which is now about 90% curable in large part due to this type of transcription therapy.
I might add, though, you say, oh, this is a great example of bench-to-bedside.
It actually worked the other way around.
Impaired trials in Shanghai, China, by Zhu Chen, who's now the health minister of China,
and his mentor, Yang Zhu, showed that retinoic acid happened to cure these patients.
In the meantime, Janet Rowley found 25 years ago that there's a consistent chromosomal translocation 15-17 in acute promyelocytic leukemia.
And several groups knew that the retinoic acid receptor was on chromosome 17.
So they put together a retinoic acid response in these patients, but no other leukemia patients.
The 15-17, they asked, is the retinoic acid receptor affected?
That's how they found it.
It was actually bedside-to-bench research that showed how this worked.
But a very interesting translational research nonetheless.
Here's another example of how you might be able to target transcription in hematologic malignancy.
Diffuse large B-cell lymphoma, seen here.
This is a normal set of B-cells in a germal center, B-cells being educated, exposed to antigen, proliferating.
This is a very privileged site here.
This is a site where DNA damage and apoptosis is suppressed.
You want that because these B-cells are undergoing rapid proliferation,
and they're mutating their immunoglobulin genes to select for antibody diversity.
If you allow DNA damage and apoptosis to go on, the cells would die.
So they're suppressed in large part by a protein called BCL6,
which is apparently rearranged and overexpressed in lymphoma.
It normally goes on in the germal center stage and goes off as B-cells further differentiate.
Here it's stuck in the on position, encouraging the formation of diffuse large B-cell lymphoma.
Now, starting over a decade ago,
Ari Melnick, when he was a postdoc in my lab, worked with structural biologists at the Ontario Cancer Institute
to understand how does BCL6 actually work as a transcription factor.
We knew it repressed transcription, and we knew it had very tight binding to a cofactor known as NCOR,
and a related cofactor known as SMART.
We actually co-crystallized BCL6 bound to NCOR and SMART,
and found it bound to a unique groove around the side and the bottom of a repression domain called the BTB domain.
So we thought that if we took this little bit of SMART or NCOR and linked it to a decoy peptide,
we might be able to break up this normal co-repressor-repressor interaction,
break up BCL6 repressive function,
turn back on apoptosis and other processes in the germal center,
and kill lymphoma cells.
The idea here is that BCL6 is decorated by a co-repressor complex that turns off transcription.
If you add a peptide that blocks this,
then this co-repressor complex falls off and genes are turned back on.
So, in fact, that all worked exactly as predicted.
Here's a simplified diagram showing four lymphoma cell lines with high-level expression BCL6,
all killed by this specific peptide.
But four lymphoma cell lines with low or barely detectable BCL6 expression are not affected at all.
So this is a way of drugging the so-called undruggable.
We actually formed a peptide that's in IND-enabling studies now to treat this form of lymphoma.
Dr. Melnyk has gone on to form a small molecule, a 500 Dalton or so molecule,
that can mimic the peptide and also block BCL6 function as a form of specific therapy for hematologic malignancy.
So the lesson here is that we understand some of the basic biology of these types of transcription factors,
how they work, their transcriptional mechanisms,
and if we can solve it at the atomic level, maybe we can come up with a therapeutic agent.
Another process that's frequently abnormal in hematologic and other types of malignancies is DNA methylation.
DNA methylation of CpG correlates with gene silencing.
The CpG dinucleotide is relatively infrequent in the genome,
but common in the—oh, here we go.
Can I get another battery for this? Thanks. Is there more batteries for this? Thank you very much.
CpGs are present in promoters in the 5' region of many genes.
These methylated CpGs can be stably inherited with cell division.
This is a form of epigenetic marking, and it's the ultimate epigenetic mark.
Once a gene is heavily methylated on CpGs, it tends to be shut off pretty much for good.
There's aberrant methylation in cancer.
It turns out that repetitive sequences found near the centrosome, found throughout the genome,
even in places like endogenous retroviruses, these are hypomethylated in cancer.
But specific genes are hypermethylated in cancer,
and there seems to be some kind of Darwinian selection to hypermethylate those genes that will suppress cell growth.
The genes that are tumor suppressors are methylated, shut off, allowing cell growth to occur.
But there's even more.
In the past two years, it's been found that DNA methyltransferase itself is mutated in a significant fracture,
in this case of acute myeloid leukemia, perhaps leading to really global demethylation problems.
This methylation can be started, for example, by a transcription factor complex.
For example, we found that PLZF binds to histone deacetylases.
The histone methyltransferase G9A, which methylates chromatin,
and it turns out that methylated chromatin on histone 3-lysine 9 binds a protein called heterochromatin protein 1,
which can recognize DNA methyltransferase, which can then methylate DNA.
In this way, a transcription factor can potentially lead to the hypermethylation of DNA.
In fact, methylated DNA is recognized by specific methyl DNA binding proteins, such as methylated binding protein, MVP.
This then can bring in histone deacetylases, more molecules of G9A, which methylate histone 3-lysine 9,
bring in more HP1, bring in more DNA methyltransferase, and again, perpetuating the reaction.
This is a recurrent theme, as I mentioned before, of self-reinforcing.
Once the reaction gets going, it tends to stay going.
DNA methylation is inherited after replication.
There are two types of DNA methyltransferases.
A novel DNA methyltransferase that takes unmethylated DNA and methylates it.
Those are DNA methyltransferases 3A and 3B, 3A being mutated in human cancer, 3B never so far.
There's a methyltransferase that recognizes hemimethylated DNA and remethylates the daughter strand.
This is DNMT3-1.
Actually, there's a structural biologic basis for this.
One of my colleagues, Dintra Patel, has actually figured out how these methyltransferases recognize
methylated versus hemimethylated DNA and the very specific structural mechanisms.
Methylation can go awry in cancer.
I mentioned many tumors are hypomethylated, but specific genes are hypermethylated.
Really, there's been a use of anti-methylation agents in hematologic malignancies,
but we've sort of backed into it.
A lot of this work was done in Mount Sinai, where it was for many years.
Again, it's a little bit of a bedside to bench kind of thing.
It was known that these agents hypomethylate for many years.
It was also known that if you give some of these hypomethylate agents at high dose,
it's just conventional chemotherapy.
It just kills cells.
The idea was to give these drugs at a lower dose that was well-tolerated,
and we tried to see whether or not this was associated with changes in DNA methylation.
The syndrome that we use these types of hypomethylate agents is called
the myelodysplastic syndrome, which is a pre-leukemia.
Patients have a barren clone within the bone marrow that takes over the bone marrow,
and it forms blood cell elements quite poorly.
The patients end up presenting with low white cells, red cells, and platelets.
In many cases, these patients will progress on to acute leukemia.
This shows that there's a whole panel of genes that are apparently hypomethylate,
shown in red here, compared to normal hematopoietic cells.
If you treat patients with these demethylating agents,
you can find that there's quite good demethylation of these genes.
However, we found that there's basically no good correlation between specific genes,
such as tumor suppressor genes that are hypomethylated in leukemia and MDS.
We found patients who responded had no change in methylation,
and other patients who had great demethylation and no response whatsoever.
We do not yet know whether or not this demethylation of specific genes
is a marker that the drug got into the patient,
or actually a mechanism by which the patient responds.
But this is in current clinical therapy.
We use it routinely for acute myeloid leukemia of the elderly,
and for patients at all ages with myelodysplastic syndrome.
Are we targeting transcription? Maybe.
But are we affecting DNA methylation? Definitely.
Is that related to response? I don't know.
These are some background to some ways that we're targeting transcription
in hematologic malignancies.
I'd next like to go into a current interest of mine,
which is the problem of global chromatin anomalies in lymphoid malignancies, particularly.
Lymphoid and myeloid, but we're going to focus more on lymphoid.
As I said to you, there are mutations in chromatin regulators of all sorts
in hematologic malignancies.
One theme that seems to be a recurring theme is a focus
on the histone-3-lysine-27 methylation site.
This is made by the ECH2 protein.
It's an enzymatic component of the polycomb repressor complex number two.
It's lost or mutated in many forms of myeloid malignancy,
myelodysplastic syndrome, and acute myeloid leukemia.
Other components of this multiprotein complex can be mutated as well,
such as these components.
This is associated in this work from Nature Genetics two years ago
with loss of the histone-3-lysine-27 mark.
These are two leukemic cell lines without ECH2 mutations.
These two have ECH2 mutations.
This immunoblock shows a significant loss of histone-3-lysine-27 methylation.
Now, that doesn't mean just one gene here, one gene there.
That means all across the genome.
This must have some very widespread consequences on genome integrity
and gene expression.
The histone demethylase, UTX, which oxidizes histone-methylated residues
using alpha-ketoglutarate as a cofactor,
will oxidize and then demethylate the histone-methyl mark.
UTX is found on the X chromosome, and it's specific for H3K27.
UTX is mutated in many forms of hematologic malignancy.
In this tantalizing immunoblock that we're following up by mass spec studies,
we found that these multiple myeloma cell lines,
which represent a late B-cell malignancy,
myeloma cells are basically protein factories that make immunoglobulin,
and patients have many symptoms due to, in fact, excess immunoglobulin.
These cell lines with UTX mutation tend to have higher K27 levels
than these cell lines without the UTX mutations.
So now we're really confused.
We have diseases where K27 methylation is down
and others when the K27 methylation is up.
Why is this occurring? Don't know yet.
It gets even more complicated with the finding two years ago,
again from these exome sequencing, this is discussed by Dr. Weinstock,
of recurrent mutations of ECH2 in diffuse large B-cell lymphoma,
particularly a type called follicular and germal center type lymphoma.
These were recurrent mutations in almost one single spot
and always found in a heterozygous condition.
It turns out if you model, this is not the kerosene structure of ECH2,
but another histone methyltransferase.
Turns out the histone tail fits into this groove right here,
into this sort of cage.
It turns out that there's a tyrosine right here
that's mutated in ECH2, creating a lot of space in here.
And now instead of this histone methyltransferase
doing mostly mono and dimethylation,
it has enough room to do trimethylation.
The problem is as you add a methyl group onto the histone tail,
it gets theoretically more and more difficult.
Okay, so we won't back up. Okay.
It gets more and more difficult to add more histone methyl marks.
This one's got plenty of room, so it becomes a trimethylase.
That's not normal.
So in fact, if you take a look by either immunoblot
or in this case mass spectroscopy seen right here,
these two cell lines with the ECH2 mutation
have a dramatically increased histone 3-lysine 27 trimethyl content.
That's by immunoblot, and this is mass spectroscopy
seeing a decrease of K27 dimethylation
and an increase in K27 trimethylation.
And it's not subtle. It's about a three-, four-fold change.
Again, this means that chromatin all throughout the genome
is going to be changed.
What happens in terms of gene expression
and what happens in terms of phenotype?
We've modeled this by taking a mouse B-cell cell line,
and we overexpress either wild-type ECH2
or these two different mutant forms.
Now you can overexpress as much wild-type ECH2 as you want in these cells.
You can make quite a bit of it,
but you do nothing to the K27 trimethylation,
so overexpression doesn't cause this.
But expressing either of these point mutants found in patients,
you get that depletion of K27 dimethylation
and an increase in K27 trimethylation.
These are other ECH2 components which are unchanged in expression.
So does this have a biological correlate?
It does, in that these cells don't grow very well in soft agar,
but if we add the mutant ECH2,
they readily form colonies in soft agar.
I don't think I added it today,
but we've more recently found that this is associated
with a dramatic increase in K27 trimethylation
across many genes by the technique known as ChIP-seq,
and those genes are almost always repressed,
and there's about 1,000 of them.
So that also begs another question.
We have this global change in K27 methylation.
It's everywhere, but not really everywhere.
It seems to be localized to some spots more than others.
We don't know the rules for that.
And despite the fact we have K27 trimethylation in many, many places,
we're only getting maybe 5%, 10% of genes affected in expression.
We don't know the rules for this.
This is not normal transcription.
This is aberrant transcription,
and the rules that we learn sort of,
oh, here's a transcription factor, and it binds a cofactor,
and it binds a histone modifier, and it turns on or off a gene,
those are normal rules.
These are pathophysiological rules.
We're hoping to try to understand what they are
so we know what genes and pathways are affected.
One thing that's very promising is that the AACR meeting this past week,
there were two companies that are on the verge of developing ECH2 inhibitors,
and these will soon be available to the research community,
I believe, as tool compounds.
And so the question is,
will those ECH2 inhibitors reverse this colony formation assay?
Will it reverse malignant behavior in models as they're developed?
So this is yet another way that we can target aberrant transcription regulation.
In the last third of the talk,
I'll talk about epigenetic disorder in multiple myeloma.
Multiple myeloma is a malignancy of very late B cells, as I mentioned,
and its molecular pathogenesis was quite obscure
until, I'd say, about 12, 13 years ago,
with discovery by investigators in France and the United States
of recurrent chromosomal translocations
that affect about five or six entities
where there are recurrent chromosomal translocations
that link the immunoglobulin promoter enhancer to these genes,
cyclin D1, D3, CMAF, MAFB, and notably the one we study, MMCET.
These other cases probably have...
they do have point mutations and a variety of other disorders,
but these categories here form distinct clinical
and pathological subsets of multiple myeloma.
So the MMCET also represents a global chromatin anomaly syndrome.
MMCET is a large histone methyltransferase protein
with many domains involved in chromatin recognition.
These PhD, PWWP domains, HMG domains, PhD domains,
we think that these bind to DNA and to chromatin.
The specificity of it is yet to be completely figured out.
The CET domain is the histone methyltransferase domain.
In multiple myeloma, due to the chromosomal rearrangement,
MMCET, which is normally expressed in every cell at a low basal level,
is overexpressed by, we would estimate, at least tenfold.
These are myeloma cell lines without the translocation.
Here's MMCET.
Here's the overexpression with the 414 translocation.
You see some different sizes of the protein due to the fact
that the immunoglobulin promoter can invade the MMCET gene,
either 5' to the entire coding sequence
or a little bit internal to the 5' end,
leading to making a transcript that will encode
a little less than the full-length protein
due to cryptic ATGs within the coding sequence.
But in all cases, that CET domain is maintained.
Is this biologically important for the myeloma cells?
We and several other groups have shown by shRNA and siRNA
that knocking down MMCET expression level
will cause these cells to stop growing.
They'll undergo apoptosis.
They'll undergo G1 arrest.
So MMCET is a growth stimulator for these types of myeloma cells.
What mark does MMCET make?
What's its chromatin signature?
Using a radioactive histone methyltransferase assay,
we take MMCET made in bacteria,
in this case recombinant histones,
acidacylmethionine as a methyl donor,
we were able to show that MMCET will catalyze transfer
from SAM to histone, seen here.
Positive control is G9A, which mentioned briefly,
is a histone 3 lysine 9 methyltransferase,
and there it is methylating histone 3.
At least in vitro, MMCET is more promiscuous,
can methylate histone 3 and histone 4.
There's a mixture of 3 and 4,
and it seems to methylate both of them.
How do we figure out what it really does in vivo?
For this, we used two systems,
one of which was developed by a gentleman named Josh Loring
at Johns Hopkins.
He very cleverly knocked out the rearranged MMCET allele
in multiple myeloma cells by somatic homologous recombination.
He inserted a cassette, including stop codons
and transcription termination sites,
into the rearranged MMCET allele,
leading to a drop of MMCET RNA levels,
and as I'll show you in a moment, protein levels as well.
He also targeted the non-translocated allele,
and you can see that the non-translocated allele
is expressing a low level of the protein
so that these levels were still high.
Here in MMCET knockout cells,
MMCET levels are way down,
and you can see a deficit of K36 methylation,
and again, with our theme of K27,
a surfeit of K27 methylation.
So this represents another global chromatin abnormality.
This complicates the slide.
Don't look at it.
Look at it again.
These are multiple myeloma cell lines from 414 patients.
They all have overexpressed MMCET,
some different sizes, as I explained to you before,
but they all have a surfeit of K36 dimethylation
and a deficit of K27 methylation.
These are the cell lines without the translocation.
They have high K27, low 36,
and it's completely switched over here.
So again, this is a profound switch in chromatin.
I think I'll just skip this part here.
Can we reverse this switch?
Is MMCET really causing the switch?
So we take our knockout cell lines,
and we added back MMCET wild type or some point mutants,
and let me just skip to the data.
So this is a knockout cell line that has low K36 and high K27.
We add back MMCET.
We gain K36, die and try in this case.
We lose K27.
This mutant over here is interesting
because it seems super active, just like the EZH2 mutant.
It does more K36 trimethylation,
and you have a deficit of mono and dimethylation,
but it'll still get rid of K27 methylation.
This one is completely inactive.
So these mutants are very useful.
We have a super active and an inactive mutant.
So then we could ask,
is this histomethylation activity important for growth control?
So we take our knockout cells,
and we add back the dead mutant,
and the cells grow pretty poorly.
We add back a wild type MMCET.
We stimulate growth.
If we add back the super active mutant,
we also stimulate growth.
So the histomethyltransferase inactive mutant
does not fully stimulate cell growth.
To get full stimulation cell growth,
we need the histomethyltransferase activity.
But things even get a little more complicated.
We made one mutant of MMCET in this PhD domain number 4,
which has an intermediate phenotype, intermediate growth,
and where's wild type MMCET?
There's the K36 on, K27 off.
When you delete PhD domain number 4,
you are turning on K36,
but you're not completely turning off K27.
There's less than over here, but not as little as over there.
So there's two processes going on.
Actually, by mass spectroscopy analysis,
we found that when MMCET's high in a cell,
we get both active K36 methylation
and active K27 demethylation.
And why is that?
Because MMCET can actually bind to the demethylase
known as Jumanji D3.
So we think we attract it to chromatin,
removes the K27 mark, and adds the K36 mark.
So we call this HMT's gone wild.
So this complex here will demethylate on K27
and methylate on K36,
and we have a global switch of chromatin.
What does this do in terms of gene expression?
Well, we can do gene expression arrays,
and more recently, RNA-seq,
and look for genes turned on and turned off.
And some of the right types of genes
are being affected by MMCET.
Genes involved in apoptosis, DNA repair,
and chromatin packaging and remodeling.
And we actually find that when we look at genes
regulated by wild-type MMCET,
we see some repressed genes,
but a lot of genes really at the high end here
activated 20- to 60-fold.
If we put MMCET without the HMT domain into a cell,
we only get 2- to 4-fold regulation.
So MMCET really seems to boost transcriptional activation.
And here's an example of an MMCET-regulated gene.
It's an adhesion gene known as JAM2.
It may be involved in some of the malignant behavior
of these cells.
When MMCET's knocked down, JAM2 goes down.
In the knockout cell, it goes down.
If we add back MMCET, JAM2 goes up.
The active unit turns it on.
The dead unit does not.
Again, saying that histomethylation is important
for gene expression.
Now, we've done genome-wide binding
for MMCET across the genome,
but I want to share with you a caveat
about some of these programs
that will say where is a protein bound in a promoter.
This looked great here.
It's bound in introns and intergenic regions.
This is a ChIP-seq, and the program identified peaks.
But I want to say have a caveat
about peaks that are called by these programs,
because when you look at the raw reads here,
this is just MMCET-bound everywhere.
And what we think is happening
is we've just filled up the cup so high,
and this protein is so chromatin-bound,
it's just sort of binding everywhere.
But we don't know exactly why it turns on only some genes
but not others.
We still don't know the answer.
So we need new rules for a barren histone code modification.
There's a global increase of K36,
but only about 10% genes are changed.
And some of these genes that are bound by MMCET
actually were pressed and not activated,
so we still don't know the exact rules here.
We think there are MMCET targets
relevant to cancer development.
Which ones are actually involved by pathogenesis?
Here, we've actually triangulated
between downloading a number of patient data sets.
There are specific patients with the 414 translocation.
They have a barren gene regulation.
We found that in our system,
these genes that are turned on in those patients
are regulated by MMCET in our cellular systems,
and we're knocking down those genes systematically
in myeloma cell lines
to see if they're critical for myeloma cell growth.
I'm going to skip this part here
and just go to one final thing right here.
If it lets me.
Can we affect MMCET?
Can we make inhibitors of MMCET
to target it for hematologic malignancy in other cancers?
Well, there are histomethyltransferase inhibitors.
One's called Bix1294.
It was an inhibitor of G9A,
but it does have some effect on MMCET here,
but it's at the micromolar level.
We really need to get to nanomolar compounds.
It's intriguing, though, that this HMT inhibitor
killed MMCET-positive cells
better than MMCET-negative cells,
so we're motivated to keep on going for this.
We've developed, with Milan Merkzich,
a chemist at Northwestern,
a mass spec-based screen for MMCET inhibitors.
Here we take a histone tail
and we incubate it with MMCET
and let this label transfer occur,
let methylation occur.
But we do it on histone tail substrates,
very small, about 10 amino acid peptides,
which are then immobilized onto gold spots.
So we have this 96 or 384-well gold-spotted plate
that then goes in to a mass spectrometer,
and we can actually detect the methylation shift
of one methylation here,
a little bit of dimethylation,
but mostly monomethylation.
So we put this into this high-throughput mass spec machine
and we look for this 14-Dalton mass to charge shift.
And in our first 10,000,
we got about a half dozen hits out of the first 10,000.
We've tested about three of those.
We were able to get fresh batches from the chemical supplier.
That's the one thing about this screening,
you don't even really want to go back to the original plates.
You want to say, they say it's compound X,
get a fresh batch of compound X,
because unless you want to start going back to original plates
and figuring out what was in there
and do analytical chemistry,
you have to take it that it was compound X.
So the first three didn't work out,
but we're continuing.
So the last thing I'll say in motivation is
MMSET is not just affected in myeloma.
This is data from Oncomine
showing that MMSET as well as EZH2
are highly over-regulated in prostate cancer.
Recent data from Arul Shanayan's group
says that EZH2 actually turns on MMSET
in many advanced tumors.
And our recent data shows that MMSET actually affects,
when you put MMSET into prostate cells,
it encourages migration and invasion
seen right here in this assay.
So actually, some of the effects of EZH2 in prostate cancer
may actually be MMSET effects.
Hence, targeting MMSET or EZH2 or both in prostate cancer
might be a viable therapeutic target.
So I'm going to move to conclusions
and say that I think I've given you an overview
of some of the disorders that occur in histone
and DNA methylation in hematological lymphocytes.
K27 is very intriguing
because many of the problems,
too much K27, too little K27,
or K27 itself being mutated,
is a recurring theme in cancer.
And this is an intriguing activity
we're going to be studying for years to come.
We think that HMT activity is important
for altering cell growth in cancer
and work I couldn't have time to talk about today,
it may affect other processes
such as DNA repair and replication.
And we think that HMT inhibitors
may represent a future therapy for many forms of cancer.
So I'd like to thank members of my laboratory
who did this work
and a variety of collaborators across the country
and our funding organizations seen here.
Thank you very much for your attention.
APPLAUSE
Excellent talk, Dr. Hurley. Thank you very much.
Any other type of questions?
Eric, can you go ahead?
I'm just thinking about early diagnosis
for multiple myeloma.
If anybody looked at, like, for instance,
MGUS or MMSET or translocations.
MMSEC is found in MGUS.
The overexpression is found in MGUS,
so it may be a very early event.
And this remains a very big mystery.
MGUS is the prodromal form of,
for others who may not know,
is the prodromal form of multiple myeloma.
Patients, particularly as they get into
their seventh and eighth and ninth decades,
more and more normal people,
people who are otherwise healthy,
have a hyperproliferation of a barren clone
of immunoglobulin-producing cells.
They have an inbarrent immunoglobulin production
in the bloodstream.
Those patients convert to multiple myeloma
at the rate of about 1% to 2% per year.
What causes that transition, no one knows.
One of my colleagues,
we just had a forum at this at the AACR meeting,
one of my colleagues at the National Cancer Institute
is obtaining specimens from patients serially
at diagnosis, and patients have agreed
to repeat bone marrow biopsies,
which is quite brave,
to determine what are the genetic changes
that are accumulated by things such as exome sequencing
as these patients go along,
and to see what mutations may occur
in addition to MMSET
to stimulate the full-blown syndrome,
characterized by even higher levels
of immunoglobulin-producing cell production
and tumors within bones and other complications.
It is an early lesion.
What causes it to go from early,
sort of smoldering to full-blown disease,
no one knows.
I really enjoyed your talk,
because you really focused on
what was known and what was unknown.
You focused on chromatin-wide changes
in chromatin marks,
but genome-wide, I should say.
What's the balance between gene-specific changes
and genome-wide changes
in terms of regulation?
I think it depends upon
the pathogenic lesion.
For example,
PML-RAR in acute promyelocytic leukemia,
the original thought was that
PML-RAR binds to retinoic acid receptor normal targets
and just shuts them off.
That seems at least partially true,
but PML-RAR actually binds
to a wider complement of sites
than the normal retinoid receptor.
So it's repressing
the normal retinoid gas receptor sites
at a local level,
so it won't cause a global chromatin change.
It'll cause local chromatin changes,
but actually spreads a little bit beyond
the usual local changes to others.
MLL fusion proteins cause
chromatin changes at specific genes.
MLL fusions are attracted to specific genes
such as the Hox genes.
On those Hox genes,
you have an increase of
histone-3-lysine-79 methylation,
not globally, but just on those Hox genes,
which are big drivers of malignant cell proliferation.
If you give the H3K79 methyltransferase inhibitors,
you shut off those Hox genes,
but not other genes,
and cell proliferation goes down.
I think it depends upon the inciting lesion.
Yes, David.
John, you showed really nice data
that if you modify the epigenome,
say, by over-expression,
you can change the phenotype of cells,
including prostate cells,
but my assumption is that
that's only in a very small minority
of the cells that are transduced,
and so my question is,
is there, in fact,
in a larger set of cells,
induction of apoptosis
or some other phenotype,
and do you have any thoughts about
what might make one cell versus another
that's genetically homogeneous,
permissive of the transplanted phenotype?
That's really an excellent question
because we found, in fact,
that it's been very hard,
in some cases,
to make stable transductants
within them so that cells don't tolerate it,
so those were stable clones,
so you're asking,
we made stable clones,
and what happened to the ones
that didn't survive,
and why did some cells survive
and why others didn't?
That's a great question.
Again, something I don't know,
but a great question.
All right.
If there's no further questions,
please join me in thanking our panel.