Experimental drugs face steep odds. According to the Pharmaceutical Research and Manufacturers of America (PhRMA), out of every 5,000 to 10,000 compounds that enter the discovery and development pipeline, just 5 enter clinical trials, and only 1 receives approval. Other sources suggest that 12–16% of drugs that reach clinical trials eventually reach the market.
Besides having a high attrition rate, the discovery and development process is lengthy and costly. “On average, it takes 10–15 years for a new medicine to complete the journey from initial discovery to the marketplace,” PhRMA notes. “The average R&D investment for each new medicine is $1.2 billion, including the cost of failures.”
These figures would be less daunting if discovery and development resources were devoted to only the most promising experimental drugs. But how might one assess an experimental drug’s promise? One way is to consider the drug’s target.
Perhaps the most interesting experimental drugs are the ones that hit strategic targets, that is, targets that correspond to molecules, pathways, and mechanisms that shape physiological and pathological processes. To identify strategic targets, scientists are exploiting various approaches, including synthetic lethal screens, single-cell analyses, mechanism-of-action studies, and next-generation CRISPR systems.
Synthetic lethal screens suggest new treatments for kidney cancer
“We have performed synthetic lethal screens to identify additional targets in cells that lack the VHL gene,” says William G. Kaelin, Jr, MD, a 2019 Nobel laureate in Physiology or Medicine and the Sidney Farber Professor of Medicine at the Dana-Farber Cancer Institute and Brigham and Women’s Hospital. At the 18th Annual Discovery on Target event, which will be held virtually in September 2020, Kaelin will highlight insights emerging from recent studies on the von Hippel–Lindau tumor suppressor (pVHL), a protein that can guide target discovery for undruggable oncoproteins.
The VHL gene is inactivated genetically, by mutations, or epigenetically, by hypermethylation, in almost 90% of the patients with clear cell renal cell carcinoma, which is the most common kidney cancer. In the presence of mutant pVHL, or in response to decreased pVHL levels, HIF-2α accumulates and transcriptionally activates oncogenic genes.
In a study that performed synthetic lethal screens, Kaelin and colleagues reported that cells lacking pVHL become hyperdependent on the cyclin-dependent kinase CDK4 and its paralog CDK6. “We think this relationship is very robust because we also detected it in Drosophila cells,” explains Kaelin. This finding catalyzed research efforts in Kaelin’s group, in collaboration with clinicians, to identify and test CDK4/6 inhibitors in kidney cancer.
Another experimental strategy that investigators in Kaelin’s group are pursuing to treat cancer involves modulating the stability of undruggable cellular proteins linked to specific neoplasms. “A number of people,” Kaelin observes, “have come to appreciate that one can identify chemicals that will directly or indirectly target undruggable proteins for degradation.”
A noteworthy example is that of thalidomide-like molecules, which bind to the cereblon ubiquitin ligase complex and alter its substrate specificity, targeting two Ikaros family zinc finger protein transcription factors, IKZF1 and IKZF3, for degradation. Kaelin and colleagues recently identified a 25-mer derived from IKZF3 and showed, with in vitro and rodent experiments, that it can target heterologous proteins for destruction with thalidomide-like molecules. This can be useful for target validation studies.
Efforts to identify new protein degraders could benefit from new perspectives. “We are particularly interested in developing screens that are based on positive rather than negative selection,” Kaelin states. Many screens for degraders measure loss of proteins, which can be the result of cellular toxicity. In other words, a compound that looks like a specific degrader in such an assay might actually be a poison. “In contrast, in the technologies that we developed, degradation of the protein of interest leads to a fitness advantage,” Kaelin stresses.
Negative selection screens, also known as loss-of-function screens, seek to identify genes whose deletion is detrimental to the cell, and the data can subsequently be quantitated with various approaches. They are pervasive in cancer biology and cancer pharmacology.
“As we continue to identify and validate therapeutic targets, it is important to become a little more sophisticated and rigorous,” maintains Kaelin, “and that includes being mindful of some of the ways in which one can get misled when using negative selection or ‘down’ assays, for example, by failing to exclude off-target effects.”
Single-cell approaches find antibody leads for immuno-oncology
“One of the reasons we perform single-cell approaches is they allow us to pick up subtle differences that would be missed with more conventional tools,” says David S. Johnson, PhD, CEO and co-founder of GigaGen, a company that specializes in microfluidics and molecular genomics. Two GigaGen platforms, Surge and Magnify, can facilitate the development of polyclonal antibody therapies for transplant rejection, infectious diseases, and immunodeficiency disorders. The platforms can also help identify monoclonal antibody leads for checkpoint-resistant cancers.
The Surge platform, based on a combined droplet microfluidics, genomics, and protein library engineering approach, generates complete antibody repertoires from single B lymphocytes, while simultaneously maintaining the native pairing of the antibody light and heavy chains. This native pairing is crucial for the development of antibodies with optimal function and with higher sensitivity and specificity.
The Magnify technology profiles transcriptomes from tens of thousands of individual cells from the tumor microenvironment of drug responders and nonresponders and helps identify rare targets. If it is a target that can be addressed with antibodies, GigaGen can put it into the Surge platform in search of the best therapeutic lead. “Surge and Magnify work very well together,” asserts Johnson. “We use them in parallel.”
A lead product at GigaGen is an antibody against CTLA-4, the immune checkpoint inhibitor targeted by ipilimumab, a monoclonal antibody approved by the FDA for the treatment of melanoma. Ipilimumab, which is currently in clinical trials for other malignancies, activates the immune system by targeting CTLA-4 on T cells. However, its clinical utility is hindered by dose-dependent and cumulative autoimmune adverse events thought to occur due to its checkpoint inhibition activity.
“We wanted to find a drug that maintains the regulatory T-cell killing capability but does not have the classical checkpoint inhibition activity, so that we could see the efficacy of ipilimumab without its toxic side effects,” explains Johnson. This effort led to the development of GIGA-564, the first molecule at GigaGen that is being moved into clinical studies for oncology applications.
“GIGA-564 is a very specific anti-CTLA4 product that depletes regulatory T cells in the tumor microenvironment but lacks checkpoint inhibition,” says Johnson. “This makes it a candidate for stronger efficacy and better safety in the clinic.”
Another clinically validated GigaGen product is a bispecific molecule that contains an anti-PD-L1 antibody linked to an anti-transforming growth factor-β (TGF-β) antibody. “On one side, anti-PD-L1 activates T cells,” notes Johnson. “On another side, anti-TGF-β enhances the infiltration of T cells and other immune cells into the tumor microenvironment.”
Scientists at GigaGen found that the two antibodies in this molecule act synergistically. “This is another example of when the Surge platform was really helpful in identifying hundreds of candidates on each side,” Johnson points out. “We were able to put them together in interesting ways to develop a unique molecule with unique properties.”
The two antibodies and the bispecific molecule were tested in mouse models, and their effects were analyzed using the Magnify platform. “We found transcriptional pathways that became activated very specifically in the anti-PD-L1 and the anti-TGF-β arms of the study,” details Johnson. Some of the genes and pathways that were identified participate in the reprogramming of the extracellular matrix. Targeting them promises to increase T-cell infiltration into the tumor microenvironment.
One of the challenges when targets identified in many of these screens are prioritized involves identifying the most promising candidate from many potential leads. “Studies with single cells generate hundreds of candidates, and one has to somehow find ways to narrow down that list,” maintains Johnson. “But single-cell analysis also helps with that part, since you can pick up more subtle differences in the tumor microenvironment in vivo studies, making these studies more sensitive, which in some cases can be a kind of efficacy signal that helps to focus on the most promising candidates.”
Targeting the pathological process behind neurodegenerative disease
A distinct targeting approach is being followed at Vaccinex. “We are not targeting the mutation that triggers the disease, but the pathological process that is triggered by the accumulation of misfolded proteins,” says Maurice Zauderer, PhD, the company’s CEO.
The lead antibody developed at Vaccinex, pepinemab, is specific for semaphorin 4D, a transmembrane glycoprotein that was first identified as a molecule involved in repulsive guidance during axonal growth. Semaphorin 4D regulates the cytoskeleton in several cell types.
In addition to giving cells their shape, the cytoskeleton allows cells to change shape. When cells move, projections that attach to the extracellular matrix pull the cell toward a certain direction. This process is influenced by semaphorin 4D, which induces collapse of the cytoskeleton and is capable of reversibly immobilizing cells.
The ability of cells to extend projections toward other cells is very important in the brain, where it supports interactions between neurons and glial cells. “We thought this is an area that could be relevant for neurological diseases,” notes Zauderer. Scientists at Vaccinex became interested in whether this process is involved in the pathogenesis of Huntington’s disease, a devastating neurodegenerative condition that currently lacks disease-modifying therapies.
The biology of astrocytes attracted considerable interest in this context. “Astrocytes,” Zauderer explains, “are key inflammatory cells in the brain that become activated during disease progression, when they abandon their normal physiological functions and secrete inflammatory molecules that recruit other inflammatory cells, such as microglia.”
In initial experiments, scientists at Vaccinex revealed that semaphorin D is upregulated in neurons during disease progression, a finding that was recapitulated using transgenic mice. The receptor for semaphorin 4D is Plexin-B1, a molecule abundantly expressed on astrocytes. As a result of binding semaphorin 4D, the cytoskeleton of the astrocytes partially collapses. The astrocytes pull back their characteristic extensions, and they change their gene expression profiles to secrete inflammatory molecules.
“We tried to use a blocking antibody to semaphorin 4D to prevent the signal from activating the astrocytes and to preserve their normal functions,” relates Zauderer. Studies at Vaccinex revealed that treatment with semaphorin 4D blocking antibody showed benefits in animal models and supported the initiation of human clinical trials.
In a Phase II study on Huntington’s disease, scientists at Vaccinex treated participants with pepinemab or a placebo for six months. In addition to taking clinical and cognitive measurements, the scientists used fluorodeoxyglucose imaging to assess glucose transport.
One of the functions of astrocytes is to completely cover the brain capillaries with their foot processes and contribute to the uptake of glucose, the main energy source of the brain, from circulation. These foot processes express glucose transporters, which are downregulated in astrocytes that switch to the inflammatory state. “We wanted to know if blocking semaphorin 4D signaling, and preventing the astrocyte transition to the inflammatory state, would alter glucose transport in the patients who received the drug,” says Zauderer.
Although the placebo group showed the expected disease-associated decline in fluorodeoxyglucose uptake, indicating decreased glucose transport, the fluorodeoxyglucose signal increased in the pepinemab arm of the study, indicating enhanced glucose transport. Based on these findings, scientists at Vaccinex initiated a larger, 18-month study that finished enrolling patients in December 2018, and they plan to unblind the data in September 2020.
The working hypothesis at Vaccinex is that the same pathological process is relevant for other neurodegenerative diseases, but it is triggered by different factors, such as a mutation in the HTT gene in Huntington’s disease, α-synuclein in Parkinson’s disease, or β-amyloid in Alzheimer’s disease. “These insults activate a common pathogenic pathway,” asserts Zauderer, “and this is the origin of disease pathology that we are trying to address.”
Adapting CRISPR to enhance drug discovery
“Our work involves characterizing and developing new CRISPR systems with properties that can be important and exciting for applications such as target discovery and validation,” says Trevor Martin, PhD, co-founder and CEO of Mammoth Biosciences. A major effort at Mammoth involves interrogating metagenomic sequences that are generated from the sequencing of microbial DNA from environmental samples, followed by the use of machine learning algorithms to identify new CRISPR systems that are subsequently tested in the laboratory.
Using this approach, scientists at Mammoth identified novel Cas nucleases with shorter or alternative PAM (protospacer adjacent motif) sequence requirements. The PAM sequence is the short DNA sequence a few nucleotides downstream of the DNA region targeted for cleavage by the Cas nucleases, and the discovery of novel PAM sequences promises tools with more flexible targeting.
Other strategies used at Mammoth to expand the CRISPR toolkit include identifying Cas enzymes with unique properties, such as increased temperature stability or smaller size, which enable new applications and cellular delivery options. “Other nucleases that we identified have higher binding affinities, which means that we expect them to not have that many off-target effects,” reports Martin.
The research strategy at Mammoth Biosciences leverages natural diversity and combines it with engineering, omics, and molecular genetics tools to develop entirely new systems. “We are big believers in the potential of the next-generation CRISPR systems,” Martin declares, adding that Mammoth is adapting them to further enable next-generation therapies.
Mammoth recently announced a collaboration and license agreement with Horizon Discovery Group to facilitate the engineering of a new generation of CHO cells that will assist in producing biotherapeutics, including therapeutic antibodies. Marin says, “We are very interested in working with partners who can leverage our systems for interesting applications.”