This is part three of our four part series on Octant's multiplexing capabilities. To read more on multiplexing for drug discovery, check out our posts on Cellular Intelligence, broad target scans, and deep mutational scans.
The human body isn’t a test tube, but much of industry still builds drugs like it is. For example, two of the most dominant trends over the past 50 years of drug discovery, high-throughput chemical screening and structure-based drug design focus on targeting molecules to proteins isolated from their cellular environment. But for many diseases, it is the protein’s complex mechanistic interactions within that cellular environment causing the disease. Developing better drugs to modulate cellular mechanisms like signaling, trafficking, and transcription requires a better ability to do robust drug discovery in living cells.
Developing drugs in living cell systems has historically been difficult for several reasons, one of which is the lack of precision high-throughput reporters (sensors that transmit information about what is happening in a particular cellular process - often in response to drug treatment). Traditional methods for building these reporters are an expensive and manual optimization of the thousands of variables in their design.
In this blog post, we’ll describe how we use multiplexing to rapidly develop better cell activity reporters for a new type of direct-to-biology drug discovery. In contrast to traditional methods, Octant’s multiplexing technology enables us to quickly design, build, test, and learn from hundreds of optimizations at a time, resulting in a novel and better understanding of what is happening in cells.
Biological reporters are intricately designed modifications to living cells that report out on what is happening in a cell’s core processes. For example, they might detect changes in a key receptor protein’s signaling activity or a change in how a particular protein is being folded and trafficked by the cell. We measure these activities by building and installing genetic circuits that cause the cell to express unique DNA barcodes when a particular cellular event occurs (e.g. the pathway is activated). We then screen thousands of compounds against our engineered reporter cells, counting the resulting barcodes with next-generation sequencing. This enables a reliable proxy for an increase or decrease in the measured activity of the cellular mechanism caused by a change to the cellular environment (e.g., drugging the cell with a compound). With more and better reporters for drug screening, we can better understand the mechanistic role a target protein plays in a disease, and measure subtle changes in how different drug candidates affect that activity.
At Octant, we’ve built a robust set of next-generation multiplexable toolkits to manipulate mammalian cells with speed and precision. If we want to build a high definition reporter for a specific mechanism, we’ll quickly build and test an entire library of hundreds of reporter constructs for that mechanism, each with a slightly different but unique set of parameters such as linkers, promoters, or response elements. We are able to do this all at the same time in the same mixtures by labeling and tracking our genetic circuits with DNA barcodes.
One of the difficulties of building large numbers of reporters is verifying them to make sure you’ve built what you think you have. Our in-house sequencing platform OCTOPUS enables us to rapidly verify full genetic sequences for thousands of reporter constructs at a time. OCTOPUS offers next-day results, giving us quick turnaround time to iterate on our experiments and take exponentially more bets on optimizing our construct designs.
We can then integrate these reporters into cell lines and test them all in one carefully controlled, multiplexed experiment to select with precision for the genetic constructs that most closely model the biological activity we are seeking to measure. This multiplexed synthetic biology approach dramatically increases the chances of finding an optimal parameter set, with less effort, optimizing complex genetic circuits where interactions are often unintuitive and difficult to predict. Once we’ve identified a sensitive high-fidelity reporter, we use it in our drug discovery assays for identifying and optimizing compounds. For an example from our Retinitis Pigmentosa program, check out the scientific deep dive of this blog post.
With the power to make an impact similar to the scale of the computing revolution, the power of scaled-up and parallelized design-build-test-learn cycles is hard to overstate. For example, in this earlier blog post years ago we described how we performed a massively parallel reporter assay (MPRA) to quickly test 80 different reporters under 20 different conditions. This helped us to improve our multiplexed GPCR reporter infrastructure in a single experiment in a few weeks.
Our technology has dramatically improved since then. Last year, in collaboration with the English Lab at the University of Utah, we developed a toolkit to expedite the development of reporter assays for drug discovery programs, conducting an MPRA on the order of 6,000 unique reporters in the process. With this streamlined approach, Octonauts can now identify new reporter candidates in under a week by investigating a broad range of cellular mechanisms at a massive scale.
We hope we’ve given you a flavor for how powerful multiplexing can be. Parallelizing biological engineering gives us the tools to build better reporters to better understand what is happening in cells, to build better drugs. Below, we dive deeper into the science of the above-referenced MPRA work we did with the English Lab, for those who want a more technical understanding of the multiplexed reporter development we do at Octant.
In collaboration with the English Lab, we built an MPRA platform that can sample 6,000 different transcriptional reporters in a single experiment to identify new or optimal reporters for any given cellular pathway within a week. To do this, we tested over 300 different transcription factor binding motifs from the JASPAR database with several different contexts, periodicities, and promoters, as well as created a ~6,000-member library of reporter candidates that we transfected into HEK293 cells to assess responses to a variety of broad stimuli (e.g. bovine pituitary extract, nucleotide mixes).
To design our library, we extracted top nucleotide sequence motifs obtained from a HT-SELEX (Jolma et. al, 2013, Cell). For a given TRE, we took one motif from the HT-SELEX and repeated it four times. The four motifs were spaced such that they faced alternating sides of the DNA helix (e.g., up, down, up, down), and linked by a randomly generated spacer set. Having two different spacer sets lets us be fairly confident that the responses observed are due to protein interactions with the motif array and not the randomized sequences. We then varied the distance of the motif array to the minimal promoter to test different periodicities.
While synthesized on the oligo chip, we tested each TRE with one of 3 promoters:
Next, we transfected the TRE-MPRA library into human-derived cells and subjected them to a variety of conditions. Here, we observed significant responses to forskolin from our CRE control and our newly designed TREs based on CREB motifs. We were even able to measure subtle differences in fold change caused by different periodicities, spacer sets, and minimal promoters. Several TREs were observed with significant responses, many based on binding motifs of PAX1, XBP1, MafB, and ATF7. We recapitulated expected biological interactions and detected significant cAMP responses from several new reporter candidates, validating our library, assay, and analysis framework.
We then transferred the TRE-MPRA library to the English Lab to further characterize. Across the ten conditions tested, ~70% of reporters showed differential transcriptional output in at least one condition relative to negative controls, while ~13% were altered in over half of the conditions.
Next, the English Lab validated its ability to detect transcriptional responses triggered by both exogenous and endogenous GPCR activation. Using the TRE-MPRA, the team was able to identify the predominant signaling mechanisms of three clinically relevant GPCRs (PAR1, GPR91, and MRGPRX2) and identify novel reporters to read out on their activities. The resulting identified optimal architectures were often surprising and not always immediately intuitive for scientists attempting to perform rational design. For example, they learned that observable activation of the TRE unit AP1 by HTR2A, PAR1, GPR91, and MRGPRX2 agonism relied exclusively on pairing with the minCMV promoter; other minimal promoters led to inactive reporters. Without being able to exhaustively test thousands of combinations of elements of synthetic promoters through the TRE-MPRA it would have been very easy to miss out on this reporter architecture and land on a less sensitive reporter, or even no reporter at all.
For a closer and more interactive look at how the reporters behave under different conditions, check out this dashboard developed by Sam Himes at the English Lab.
Together, the results of these experiments demonstrate that with our TRE-MPRA platform, we can screen thousands of synthetic promoters to not only identify constructs sensitive to small molecules, but also discover ones that report on GPCR signaling. This toolkit benefits the wider drug discovery community by offering a streamlined approach to impartially investigate targets across a broad range of cellular signaling pathways, mapping signaling profiles and identifying new reporters for these targets. It can be especially valuable for “orphaned” targets whose signaling pathways still remain undetermined. With a thorough protocol and well-documented code provided, any future Octonaut can put this library to use, substantially reducing lead times for reporter development and keeping Octant focused on drug hunting.
This project would not have been possible without the collaboration, hard work, and dedication of the talented researchers in the English Lab. For the full story, check out our preprint here.
High-throughput screening (HTS) is a method used in drug discovery to rapidly test thousands to millions of chemical compounds for biological activity against a target. This process involves using automated equipment to conduct a large number of experiments in a relatively short time frame. This approach allows researchers to quickly identify potential drug candidates and optimize lead compounds for further development.
A massively parallel reporter assay (MPRA) is a functional genomics technique utilizing next-generation sequencing-based readouts of barcoded reporters to assess the activity of thousands of reporter candidates in a single experiment.
Structure-based drug design (SBDD) is a method that uses the three-dimensional structure of a protein target to identify or design new chemical compounds that can bind to the target and modulate its activity. This approach leverages computational chemistry tools and molecular modeling techniques to optimize the interaction between the chemical compound and the target protein.
.jpg)
