Probing an Unexpected Role of LSD1 in Cancer
By Mike Vineyard, PhD Candidate in Chemistry and Chemical Biology at Harvard University
The average adult human is made up of roughly 30 trillion cells. The cell is the fundamental building block of life. Upon looking in the mirror we can see that we are made up of many different types of tissues; we have eyes, skin, blood, and brain tissues all composed of different cell types. Despite this diverse collection of cell types, each cell in the body contains the same DNA, the same instructions to create life. A current goal of the field of biology is to better understand just how each of these cells know what to become despite starting with the same instructions. Various diseases including several cancers like acute myeloid leukemia have been discovered to have origins in the dysregulation of some of the processes that govern “cellular differentiation” or the process of cells becoming specialized cell types (from a less specialized cell type) to complete the various parts of an organism.
The questions that have plagued biologists for decades have been explored through the lens of a field called “epigenomics.” It has been deduced that while each cell type in the human body has the same DNA that encodes the same genes, each can distinguish their cellular identities through differences in the “expression” or reading and production of the proteins encoded by these genes. Put more simply, production of gene products can be turned on or off, up or down, through changes in the regulation of how the DNA is read.1
How does this process of turning on and off genes, work? The DNA that encodes your genes is wrapped around molecular “beads” called nucleosomes; nucleosomes are made up of proteins called histones. This complex of DNA and protein, “chromatin” is commonly imagined as “beads on a string.” Chromatin folds on itself to form the more commonly recognized chromosomes. This folding arrangement is how we fit two lengthwise meters of DNA into each of our 30 trillion cells; a minor miracle in itself.1
Histone proteins can be chemically modified by machinery in the cell; these modifications can influence the way in which chromatin folds, making the associated DNA more or less “readable.” One can think of this as the mechanism by which the production of gene products, as mentioned above is attenuated.2
The cellular machinery that interacts with chromatin and performs these modifications can be thought of as “epigenomic regulators.” They influence gene expression by adding, taking away, or “reading” chemical modifications on histone proteins. These modifications are known as post translational modifications or PTMs. Whether PTMs are causative for changes in gene expression or simply correlative is currently contested.2 Given their relationship to gene expression, these regulatory proteins have become attractive therapeutic targets in disease.
One can imagine a world of precision medicine in which changes in gene expression can be measured in disease against unhealthy tissues and be used as biomarkers for successful, predictive treatment. Before such a vision might be accomplished, a more fundamental understanding of the role that these regulators play in disease biology is required. Many of these regulators, in addition to their catalytic activity (the activity which effect PTMs and thereby influence chromatin conformation), host secondary functions.3 For example, an epigenetic writer may also function as a reader for another mark in some contexts or as scaffold for other proteins like transcription factors, proteins that are known to greatly influence gene expression and are implicated as drivers of various cell differentiation processes.
This concept of multidisciplinary function by epigenomic regulators has been explored recently in a study from the lab of Professor Brian Liau here at Harvard University and as a member of the Liau Lab, this work is the current focus of my thesis.
We used a genetic screening technology called CRISPR-scanning, recently developed at Cold Spring Harbor Laboratories. This screening technology makes use of CRISPR cutting, reviewed here[link], followed by the formation of relatively random mutations created by the cell’s DNA repair machinery. Upon CRISPR mutagenesis, what you are left with is a pool of cells harboring random mutations across the coding sequence of a protein that you’re interested in. In our case, we were examining the role of lysine-specific demethylase one (LSD1) in acute myeloid leukemia (AML). LSD1 demethylates H3K4Me1 and H3K4Me2, meaning it removes one or two methyl groups from lysine 4 of the histone H3 tail. Methylation at that site is typically associated with active gene transcription and open chromatin states; in these contexts LSD1 is seen as a transcriptional co-repressor.4
LSD1 has been shown to be important in some cancers like AML and small cell lung cancer. Additionally, researchers have found some success in using small molecules to modulates the enzyme acitivty of LSD1. These small molecules generally fall into a class of molecules known as “suicide inactivators.” This means that when the drug binds, it does so irreversibly and it actually uses the enzyme’s own activity to perform this binding (hence the name, suicide inactivator).
While these molecules have been found to be effective in blocking cancer growth through inhibition of LSD1, there is still a lack of understanding between the events of binding of LSD1 with a small molecule inactivator and blocking cancer growth. Put another way, it’s not entirely clear why inhibiting LSD1 is important in cancer; the initial notion assumed by the field was that reverting the methylation state of the LSD1 substrate, H3K4 effects a change in gene expression that prevents the cancer from proliferating. Finally, it was unclear to researchers why LSD1 was important (as demonstrated by both pharmacological inhibition as well as genetic knockout) in some cancers but not others.
To better understand the role of LSD1 in cancer, we set out to use the aforementioned CRISPR-scanning technology. After generating the pool of LSD1 mutants in cells, we applied the selection pressure of a small molecule inhibitor of LSD1. Part of the workflow of CRISPR-scanning involves barcoding the cells; this allows a researcher to track growth through next generation sequencing. We identified several barcodes in our pool, which after 8 weeks displayed 5-15 fold higher growth than non-mutagenized controls. We individually validated these results by performing the CRISPR mutagenesis in a non-pooled setting and isolated single cell clones of drug-resistant AML mutants for further study. We genotyped these clones to identify the sequences generated by the CRISPR mutagenesis that gave rise to drug resistance. We then recombinantly expressed, purified, and tested the corresponding proteins in a demethylase activity assay. Interestingly, these drug resistant mutants proved to be enzyme-dead!
The fact that these mutants were enzyme dead came as quite the surprise; the previously assumed notion of how these inhibitors work in cancer relies on the idea that the demethylase activity is targeted and destroyed with a suicide inhibitor. If cells can continue to proliferate in the absence of LSD1 enzyme activity, this tells us that the activity is indeed dispensable in this AML model. We then had to ask the question: if the activity is dispensable, what then is the drug doing?
We hypothesized that, contrary to the previous notion of the enzyme activity being important in cancer, it is actually the catalytic site of the enzyme, which is important. Further, we postulated that the catalytic site, which is known to play host to various proteins, maintains a critical protein-protein interaction in AML cells necessary for growth.
Our lab and others tested this hypothesis. Through a series of chemical genetic experiments in which we vary both the structure of the small molecule probe as well as the peptide sequence (a “bump-hole” experiment), we definitely demonstrate that GFI1/GFI1B interacts with the catalytic site of LSD1 and that maintenance of this interaction is necessary for AML cell proliferation.
Making use of the clones isolated from our experiments, we were also able to study the differences in gene expression with RNA-seq as well as chromatin localization of the various proteins in question using ChIP-seq in the presence and absence of the suppressor, relative to the wild type AML cells. The role of LSD1 in AML was explored using CRISPR-suppressor scanning, a technology that is amenable to generating hypotheses.4 Highlighted in this piece is idea that chromatin regulators can host less obvious scaffolding functions in addition to their enzymatic function and in some disease contexts, this scaffolding function may prove essential; this emphasizes the complex role that chromatin regulators play in influencing gene expression.
References
The average adult human is made up of roughly 30 trillion cells. The cell is the fundamental building block of life. Upon looking in the mirror we can see that we are made up of many different types of tissues; we have eyes, skin, blood, and brain tissues all composed of different cell types. Despite this diverse collection of cell types, each cell in the body contains the same DNA, the same instructions to create life. A current goal of the field of biology is to better understand just how each of these cells know what to become despite starting with the same instructions. Various diseases including several cancers like acute myeloid leukemia have been discovered to have origins in the dysregulation of some of the processes that govern “cellular differentiation” or the process of cells becoming specialized cell types (from a less specialized cell type) to complete the various parts of an organism.
The questions that have plagued biologists for decades have been explored through the lens of a field called “epigenomics.” It has been deduced that while each cell type in the human body has the same DNA that encodes the same genes, each can distinguish their cellular identities through differences in the “expression” or reading and production of the proteins encoded by these genes. Put more simply, production of gene products can be turned on or off, up or down, through changes in the regulation of how the DNA is read.1
How does this process of turning on and off genes, work? The DNA that encodes your genes is wrapped around molecular “beads” called nucleosomes; nucleosomes are made up of proteins called histones. This complex of DNA and protein, “chromatin” is commonly imagined as “beads on a string.” Chromatin folds on itself to form the more commonly recognized chromosomes. This folding arrangement is how we fit two lengthwise meters of DNA into each of our 30 trillion cells; a minor miracle in itself.1
Histone proteins can be chemically modified by machinery in the cell; these modifications can influence the way in which chromatin folds, making the associated DNA more or less “readable.” One can think of this as the mechanism by which the production of gene products, as mentioned above is attenuated.2
The cellular machinery that interacts with chromatin and performs these modifications can be thought of as “epigenomic regulators.” They influence gene expression by adding, taking away, or “reading” chemical modifications on histone proteins. These modifications are known as post translational modifications or PTMs. Whether PTMs are causative for changes in gene expression or simply correlative is currently contested.2 Given their relationship to gene expression, these regulatory proteins have become attractive therapeutic targets in disease.
One can imagine a world of precision medicine in which changes in gene expression can be measured in disease against unhealthy tissues and be used as biomarkers for successful, predictive treatment. Before such a vision might be accomplished, a more fundamental understanding of the role that these regulators play in disease biology is required. Many of these regulators, in addition to their catalytic activity (the activity which effect PTMs and thereby influence chromatin conformation), host secondary functions.3 For example, an epigenetic writer may also function as a reader for another mark in some contexts or as scaffold for other proteins like transcription factors, proteins that are known to greatly influence gene expression and are implicated as drivers of various cell differentiation processes.
This concept of multidisciplinary function by epigenomic regulators has been explored recently in a study from the lab of Professor Brian Liau here at Harvard University and as a member of the Liau Lab, this work is the current focus of my thesis.
We used a genetic screening technology called CRISPR-scanning, recently developed at Cold Spring Harbor Laboratories. This screening technology makes use of CRISPR cutting, reviewed here[link], followed by the formation of relatively random mutations created by the cell’s DNA repair machinery. Upon CRISPR mutagenesis, what you are left with is a pool of cells harboring random mutations across the coding sequence of a protein that you’re interested in. In our case, we were examining the role of lysine-specific demethylase one (LSD1) in acute myeloid leukemia (AML). LSD1 demethylates H3K4Me1 and H3K4Me2, meaning it removes one or two methyl groups from lysine 4 of the histone H3 tail. Methylation at that site is typically associated with active gene transcription and open chromatin states; in these contexts LSD1 is seen as a transcriptional co-repressor.4
LSD1 has been shown to be important in some cancers like AML and small cell lung cancer. Additionally, researchers have found some success in using small molecules to modulates the enzyme acitivty of LSD1. These small molecules generally fall into a class of molecules known as “suicide inactivators.” This means that when the drug binds, it does so irreversibly and it actually uses the enzyme’s own activity to perform this binding (hence the name, suicide inactivator).
While these molecules have been found to be effective in blocking cancer growth through inhibition of LSD1, there is still a lack of understanding between the events of binding of LSD1 with a small molecule inactivator and blocking cancer growth. Put another way, it’s not entirely clear why inhibiting LSD1 is important in cancer; the initial notion assumed by the field was that reverting the methylation state of the LSD1 substrate, H3K4 effects a change in gene expression that prevents the cancer from proliferating. Finally, it was unclear to researchers why LSD1 was important (as demonstrated by both pharmacological inhibition as well as genetic knockout) in some cancers but not others.
To better understand the role of LSD1 in cancer, we set out to use the aforementioned CRISPR-scanning technology. After generating the pool of LSD1 mutants in cells, we applied the selection pressure of a small molecule inhibitor of LSD1. Part of the workflow of CRISPR-scanning involves barcoding the cells; this allows a researcher to track growth through next generation sequencing. We identified several barcodes in our pool, which after 8 weeks displayed 5-15 fold higher growth than non-mutagenized controls. We individually validated these results by performing the CRISPR mutagenesis in a non-pooled setting and isolated single cell clones of drug-resistant AML mutants for further study. We genotyped these clones to identify the sequences generated by the CRISPR mutagenesis that gave rise to drug resistance. We then recombinantly expressed, purified, and tested the corresponding proteins in a demethylase activity assay. Interestingly, these drug resistant mutants proved to be enzyme-dead!
The fact that these mutants were enzyme dead came as quite the surprise; the previously assumed notion of how these inhibitors work in cancer relies on the idea that the demethylase activity is targeted and destroyed with a suicide inhibitor. If cells can continue to proliferate in the absence of LSD1 enzyme activity, this tells us that the activity is indeed dispensable in this AML model. We then had to ask the question: if the activity is dispensable, what then is the drug doing?
We hypothesized that, contrary to the previous notion of the enzyme activity being important in cancer, it is actually the catalytic site of the enzyme, which is important. Further, we postulated that the catalytic site, which is known to play host to various proteins, maintains a critical protein-protein interaction in AML cells necessary for growth.
Our lab and others tested this hypothesis. Through a series of chemical genetic experiments in which we vary both the structure of the small molecule probe as well as the peptide sequence (a “bump-hole” experiment), we definitely demonstrate that GFI1/GFI1B interacts with the catalytic site of LSD1 and that maintenance of this interaction is necessary for AML cell proliferation.
Making use of the clones isolated from our experiments, we were also able to study the differences in gene expression with RNA-seq as well as chromatin localization of the various proteins in question using ChIP-seq in the presence and absence of the suppressor, relative to the wild type AML cells. The role of LSD1 in AML was explored using CRISPR-suppressor scanning, a technology that is amenable to generating hypotheses.4 Highlighted in this piece is idea that chromatin regulators can host less obvious scaffolding functions in addition to their enzymatic function and in some disease contexts, this scaffolding function may prove essential; this emphasizes the complex role that chromatin regulators play in influencing gene expression.
References
- Dawson, M.A., 2017. The cancer epigenome: concepts, challenges, and therapeutic opportunities. Science, 355(6330), pp.1147-1152.
- Henikoff, S. and Shilatifard, A., 2011. Histone modification: cause or cog?. Trends in Genetics, 27(10), pp.389-396.
- Komor, A.C., Badran, A.H. and Liu, D.R., 2017. CRISPR-based technologies for the manipulation of eukaryotic genomes. Cell, 168(1-2), pp.20-36.
- Shi, J., Wang, E., Milazzo, J.P., Wang, Z., Kinney, J.B. and Vakoc, C.R., 2015. Discovery of cancer drug targets by CRISPR-Cas9 screening of protein domains. Nature biotechnology, 33(6), p.661.