November 15, 2015 (Vol. 35, No. 20)
Richard A. A. Stein M.D., Ph.D.
Whereas ChIP-Seq Maps the Genomic Sites Targeted By Proteins, Chem-Seq Maps the Sites Targeted By Small Molecules
The use of small molecules to achieve biological effects is as old as human history. Over time, studies that used small molecules have advanced our knowledge about biological systems, provided the framework for new therapeutic compounds, and forged new disciplines.
One such discipline, chemical genetics, emerged in the 1990s, a combination of small-molecule screening and traditional genetics. In chemical genetics, small molecules are used instead of mutations to perturb the functions of genes or proteins, thereby generating clues to gene-phenotype relationships.
Subsequent advances made it possible to characterize genetic interactions at a larger scale, to a totalizing extent, in fact. And the new genetics, like classical genetics before it, was integrated with chemistry. The result: chemical genomics. While chemical genomics is no older than genomics itself, the new hybrid discipline is already informing diverse pursuits.
“Chemical genomics is becoming increasingly popular for performing small-molecule screens in order to find new chemicals that can be used for therapeutic or other applications,” says Eric D. Brown, Ph.D., professor of biochemistry and biomedical sciences at McMaster University and holder of the Canada Research Chair in Microbial Chemical Biology.
Most screens that seek to identify novel therapeutic products follow one of two paradigms. Biochemical screens seek to find a compound that hits a specific target, but this approach often fails to yield chemical matter suitable for perturbing cells or organisms.
“Alternatively,” notes Dr. Brown, “it is possible to perform a phenotypic screen to find a chemical compound that elicits a desired cellular phenotype.” In the latter approach, even though knowing the molecular mechanism of action is not a requirement, finding the target emerges as one of the greatest bottlenecks. “This is because what usually comes out from an initial screen is not a perfect molecule,” explains Dr. Brown. “The initial compound would benefit from additional chemistry to make it more potent.”
Using a chemical genomics approach, Dr. Brown and colleagues recently identified a new antimicrobial agent that belongs to a novel chemical class and has a new mechanism of action. This compound showed activity against multidrug-resistant Pseudomonas aeruginosa by inhibiting LolA, one of the five bacterial glycoproteins that transport lipoproteins to the outer plasma membrane.
“Even though one can identify many chemical genetic interactions, the big challenge is trying to infer what the experiment is trying to reveal about the way the compound is interacting with the organism,” concludes Dr. Brown. “The solution is lots of data, lots of different kinds of compounds, and comparative analyses.”
“Chemical genomics approaches that rely heavily on biophysical mechanisms are not yet sufficiently high throughput,” complains Martha J. Larsen, director of high-throughput screening at the Center for Chemical Genomics (CCG), University of Michigan. The CCG collaborates with investigators who seek to understand the cellular targets of small molecules with biomedical importance.
Some of the approaches used to characterize these molecules involve the identification of a small compound that interacts with a known phenotype, a strategy known as forward chemical genomics. Another set of approaches, one constituting a strategy called reverse chemical genomics, relies on the initial identification of a small molecule, whose effects on the phenotype are subsequently studied.
Forward chemical genomics and reverse chemical genomics are not mutually exclusive, explains Larsen: “There are situations when we need to approach screens from both directions to address a biological question.”
One of the challenging aspects of chemical genomics screens is using compounds that are label-free. “Those assays are still not quite at the stage where we can reconstitute a screen to look at large numbers of small molecules, and we are still very limited by the technology,” advises Larsen. Even when the molecules can be labeled, the high cost of many of the reagents can be prohibitive, particularly for academic users. “And assays that use very small volumes would usually require expensive equipment,” adds Larsen.
An experimental approach has recently emerged at the interface between chemical genomics and chromatin biology. Called Chem-Seq, this approach is being used to better gauge the abilities of small molecules to modulate gene expression at a global scale.
“Chem-Seq is very much like ChIP-Seq,” says Sheng Ding, Ph.D., a senior investigator at the Gladstone Institute of Cardiovascular Disease and a professor of pharmaceutical chemistry at the University of California, San Francisco. “But, instead of using antibodies, it uses small-molecule affinity probes to pull down DNA-binding protein and examine where the small molecule binds across the genome.”
Using Chem-Seq, Dr. Ding and collaborators recently unveiled the genome-wide binding sites and the mechanism of action of a small molecule, SD70, which was initially found during the genotypic screening of a chemical library. This compound was initially shown to inhibit translocation events induced by dihydrotestosterone and genotoxic stress in a human prostate adenocarcinoma cell line, and has emerged as a potential anticancer therapeutic agent.
“If there is a small molecule that binds to DNA-binding proteins, Chem-Seq helps identify where the molecule acts across the genome,” explains Dr. Ding. “It also helps to clarify mechanisms and develop a global picture of the interaction,” Dr. Ding and collaborators revealed that SD70 indirectly co-localized with the androgen receptor on regulatory enhancers and inhibited the KDM4C demethylase, causing, as a result, elevated dimethylated H3K9 levels in gene promoter and enhancer regions.
“Even though Chem-Seq is very powerful in terms of examining drug effects across the genome, on the practical side it has a few limitations,” cautions Dr. Ding. One of these stems from the fact that, in order to capture the interaction between a small molecule and genome by Chem-Seq, the small-molecule target has to be a DNA-binding protein or part of a transcriptional complex. Additionally, unlike ChIP-Seq, which uses antibodies to characterize the DNA-binding proteins, Chem-Seq relies on affinity probes whose synthesis can be intricate and time-consuming.
“The small molecule has to first be immobilized with a linker, and this requires chemical synthesis,” informs Dr. Ding. Structure-activity analyses are required to identify a permissive linker position that does not affect biological activity of the small molecule. “Synthesizing the affinity probe is, in itself, a process that requires significant chemistry expertise,” continues Dr. Ding.
Another challenge is that the interaction between small chemicals and chromatin varies in different cell types, or even in the same cell type under different conditions, depending on the transcriptional and epigenetic landscapes. “These factors affect DNA-binding proteins,” says Dr. Ding, “and they could alter the binding pattern of small molecules across the genome.”
For example, when a genomic region contains many sites that have undergone DNA methylation, a kind of epigenetic change, the result can be highly condensed chromatin, and many genes can become inactivated, altering the interactions that may occur between small chemicals and chromatin. “A small molecule might not be able to bind to such a region,” notes Dr. Ding. “There fore, it is important to look at different cell types, and to do so under different signaling conditions, to understand how these small molecules would affect transcriptional loci across the genome.”
Probing for Phenotypes
“We started working on Chem-Seq at a time when next-generation sequencing was not yet widely available,” says Raphaël Rodriguez, Ph.D., team leader in chemical biology at the Institut Curie Research Center. “Instead of genome-wide sequencing on the pulled-down DNA, we used PCR to validate the targets.”
In an initial study that illustrated the possibility of recognizing nucleic acid elements in a structure-dependent manner using a small molecule, Dr. Rodriguez and colleagues prepared a library of biotinylated analogues derived from pyridostatin, a small molecule that activates the DNA damage response and induces telomere dysfunction. Using the pyridostatin analogues, Dr. Rodriguez and colleagues were able to selectively isolate G-quadruplex-containing nucleic acid fragments in vitro, and showed that the compounds targeted telomeres in vivo.
“When we did the first experiment to pull down nucleic acids with a small molecule, we realized that we were probing for interactions,” explains Dr. Rodriguez. “But ideally we wanted to probe for phenotypes.”
Subsequently, in another study, Dr. Rodriguez and Kyle M. Miller, Ph.D., assistant professor at the University of Texas at Austin, used high-throughput sequencing to examine the genomic binding sites of pyridostatin and described its subcellular localization using high-resolution microscopy. In this study, pyridostatin was shown to induce DNA damage during the G1 and G2 phases of the cell cycle by transcription-dependent mechanisms and during the S phase of the cell cycle in cells undergoing replication.
High-throughput sequencing and the analysis of the DNA associated with gamma-H2AX revealed that a covalently labeled version of pyridostatin targeted genes containing sequences that adopted a G-quadruplex conformation. And cellular localization studies revealed that the molecule co-localized with a DNA helicase that can bind to G-quadruplexes.
“In this second study, instead of pulling out the small molecule and its target, we pulled down the DNA damage that was associated with the target,” emphasizes Dr. Rodriguez. This experimental strategy opened the framework for a new approach to capture functional interactions between DNA and small molecules.
“One of the challenges in chemical genomics is the difficulty in identifying all the targets of a small molecule, and this is where techniques such as Chem-Seq would become useful, to visualize where small molecules bind chromatin and how that affects gene expression changes,” says Paul E. Brennan, Ph.D., a principal investigator of medicinal chemistry, Structural Genomics Consortium, Oxford University.
Recently, Dr. Brennan and colleagues reported that LP99, a quinolone-fused lactam, is a potent and selective inhibitor of the BRD7 and BRD9 bromodomains that are part of BAF and PBAF human SWI/SNF chromatin-remodeling complexes. This compound was able, in vitro and in human cell cultures, to inhibit the association between these two bromodomains and acetylated histones.
“This work was a close collaboration with Darren Dixon, Ph.D., a professor of organic chemistry at Oxford University,” notes Dr. Brennan. “Such collaborations will allow us to solve medicinal chemistry problems and, in time, target more proteins and more difficult proteins.”
Dr. Brennan and colleagues used LP99, the first selective BRD7/9 bromodomain inhibitor, to reveal that BRD7/9 proteins regulate the secretion of proinflammatory cytokines. They also showed that the BRD7/9 small molecule inhibitors could exert effects similar to the ones seen with IL-6-neutralizing antibodies.
“We think LP99 is a clean BRD7/9 inhibitor, but the development of techniques such as Chem-Seq will help clarify what other roles small molecules such as LP99 play in cells,” concludes Dr. Brennan. “In addition, they will generate an unbiased view of how they are modulating the transcriptome.”
“Chem-Seq emerges as a great way to understand where the interactions between small molecules and DNA-binding proteins are happening,” says Terry S. Furey, Ph.D., associate professor of genetics at the University of North Carolina at Chapel Hill. As there are certain parallels between Chem-Seq and ChIP-Seq, previous obstacles that were encountered in ChIP-Seq experiments might provide insights into addressing potential bottlenecks during Chem-Seq studies, such as the ones associated with data analysis.
In a recent study, Dr. Furey and colleagues used ChIP-Seq and mRNA-Seq to survey the chromatin structure and to qualitatively and quantitatively characterize the chromatin accessibility of the genomic regions that bind androgen receptor. This work revealed that androgen receptor binding induces genome-wide transcriptional changes, and that even though most androgen receptor binding sites are sensitive to DNAse I, a substantial proportion of the sites are low in the DNAase-Seq signal.
In vivo evidence pointed toward binding occurring to both full- and half-site recognition motifs, underscoring the complexity of the interaction between the androgen receptor and chromatin. “Chem-Seq is very similar to ChIP-Seq and, from a data analysis standpoint, I believe that all the analysis techniques used for other functional genomics assays, such as ChIP-Seq, could be repurposed for this one,” insists Dr. Furey.
Role of Epigenetics in Tumor Mutations
“There is a general interest in understanding epigenetic and chromatin landscapes and in probing the physical organization of the genome,” says Shamil R. Sunyaev, Ph.D., a professor of medicine at Harvard Medical School. “But an even bigger drive is to find functional genomic elements that regulate transcription.”
A major effort in Dr. Sunyaev’s lab is focusing on understanding the interplay between epigenetic changes and mutations in malignant tumors. In a study that quantitated the contribution of epigenetic and gene expression changes to mutation patterns in several cancer types, Dr. Sunyaev and colleagues revealed that a large percentage of the variance in mutation rates among the cancer genomes can be explained by factors such as replication timing and chromatin accessibility to the replication machinery.
“There is a lot of need in this area, and understanding epigenomic landscapes, especially at the single cell level, is hugely desirable,” explains Dr. Sunyaev. Due to the fact that the epigenetic landscapes vary across different cell types, the association between epigenetic landscapes and the mutational landscape of tumors opens a promising avenue to identify the cell of origin of malignant tumors.
“Several disciplines that are beyond the traditional field of chromatin organization are benefiting from the data,” informs Dr. Sunyaev. These include developmental biology, genetics, mutagenesis, and chemical genomics.
“There are lot of things that we cannot probe,” Dr. Sunyaev points out. “If Chem-Seq would come along, as it often happens with many other new technologies, we would like to push it through applications that were not accessible using previous approaches.”
Small Molecules, Metabolites Can’t Take the GC-MS Heat
Gas chromatography mass spectrometry (GC-MS) has a thermal degradation problem, and it is more pervasive than many scientists might have guessed. True, the heat that GC-MS uses to vaporize samples has been known to alter the structures of specific molecules. But now a metabolomics investigation has shown that GC-MS can skew molecular profiles, altering the structures of as many as 40% of the molecules of interest.
This negative result, unlike many negative results, has been given due prominence. It appeared online October 4 in the journal Analytical Chemistry, in an article entitled, “Thermal Degradation of Small Molecules: A Global Metabolomic Investigation.”
“We found that even relatively low temperatures used in GC-MS can have a detrimental effect on small molecule analysis,” said study senior author Gary Siuzdak, Ph.D., senior director of TSRI’s Scripps Center for Metabolomics and professor of chemistry and molecular and computational biology.
In the TSRI study, a set of small molecule standards and, separately, human plasma metabolites were heated, at different temperatures and for different durations, under GC-MS-like sample conditions. Then the samples were analyzed by liquid chromatography coupled to electron spray ionization mass spectrometry (LC-MS), a technique that does not subject molecules to thermal degradation.
It turned out that small molecules transformed and even disappeared during the experiment, throwing into question the nature of the data being generated by GC-MS.
“[Heating] at an elevated temperature of 100°C had an appreciable effect on both the un-derivatized and derivatized molecules, and heating at 250°C created substantial changes in the profile,” wrote the authors of the Analytical Chemistry article. “For example, over 40% of the molecular peaks were altered in the plasma metabolite analysis after heating (250°C, 300s) with a significant formation of upregulated, degradation and transformation products.”
The study’s results, though disconcerting, needn’t be demoralizing. If GC-MS has limitations, it may be just as well to know what they are, so that they can be accommodated. “Fortunately, these problems can be overcome with the use of standards in GC-MS as well as using newer, ambient temperature mass spectrometry technologies,” explained Dr. Siuzdak. “[Our] report will likely stimulate more scientists to move to these less destructive alternatives.”