When it comes to mimicking human disease or predicting the human body’s response to candidate drugs, traditional laboratory animal models are woefully inadequate. New technologies—3D cell culturing, human induced pluripotent stem cells, and gene editing—are leading to new solutions for replacing, refining, and reducing animal models.
Six years ago, Ping Yeh’s oncologist told him his Hodgkin’s lymphoma was resistant to the standard chemotherapy regimen and he would need a more potent, seven-drug cocktail. The treatment knocked his cancer into remission, but afterward Yeh needed an ultrasound to check whether the treatment, which can be cardiotoxic in some patients, had damaged his heart.
“The treatment could have cured me. Or cured me and killed me,” says Yeh, a nanotechnologist. “It was a pretty scary experience and it planted the seed for me to figure out if there was a better way to test for drug safety.” That seed sprouted into Minneapolis-based StemoniX, which Yeh cofounded in 2014 for the purpose of combining advances in engineering, manufacturing, and human stem cells to develop drug screening and testing platforms with more relevance to human physiology.
Fortunately, Yeh’s heart was spared. Unfortunately, patients like Yeh encounter toxic or ineffective drugs all too often, because the animal models used to test drugs before they go into patients are imperfect in many ways. Of all the drugs that enter clinical trials, only about 10% go on to be approved (1). The other 90% fail during trials—for reasons ranging from off-target, undesirable effects and problematic dosing, to low or no efficacy, and the worst-case scenario, toxicity.
“The closer you can get to mimicking the human situation, the better the research is going to be in [terms of] understanding the fundamental pathology of disease and also in predicting patient efficacy and toxicity of drug therapies,” says Richard Eglen, vice president and general manager of Corning Life Sciences in Boston, Massachusetts.
However, the challenge will be to design models that hold significant advantages over current approaches. That means producing models that give robust, reproducible data, that are predictive of human biology, and that will not greatly increase costs. StemoniX, Corning, and several other companies are coming up with innovative ways to meet that challenge.
Researchers would like to develop MSCs—also known as adult stem cells—into individualized, cell-based patient therapies, such as replacements for faulty insulin-producing
islet cells in diabetes.
Imposing structure on cultures
Culturing methods have continued to improve, and are now yielding 3D cell cultures, spheroids, and even more complex organoids that more accurately reflect human tissues. Corning has led the way in providing researchers with the best surfaces on which to grow these cultures. Now, the company has added a 1,536-well spheroid plate that can work at the highest levels of automated screening. The rounded-bottom wells, which are coated with an ultralow attachment surface, encourage plated cells to aggregate with each other and form a sphere.
“That is going to enable researchers to do ultra-high-throughput screening of up to 100,000 compounds per day,” says Eglen. The plates have small well volumes and therefore produce small spheroids that range from 500 to 2,000 cells, but they give reproducible responses that are linear relative to cell number.
A group at the Scripps Research Institute in La Jolla, California, has used the plates with pancreatic tumor spheroids to identify an inhibitor of those cells’ RAS oncogene mutation. A parallel 2D-cell-culture–based screen did not turn up the same inhibitor (2). Corning’s own research using known liver toxins on human liver spheroids shows they can be used to screen compounds for liver toxicity. “What everyone would like is the ability to better predict the metabolism of drugs before putting them into the patient,” says Eglen.
Corning has developed another technology to support the 3D growth of mesenchymal stem cells (MSCs). Researchers would like to develop MSCs—also known as adult stem cells—into individualized, cell-based patient therapies, such as replacements for faulty insulin-producing islet cells in diabetes, for instance. But it’s been a challenge growing enough MSCs under pristine conditions to then administer the cells back to the patient. Corning’s digestible microcarrier technology uses an inert,
nonanimal-derived polymer to grow the cells in a 3D orientation in solution on the microcarrier’s surface. Then, by adding an enzyme, the microcarrier can be dissolved away to leave just the MSCs for isolation.
Human cell mass production
StemoniX offers “off-the-shelf” structured 2D microHeart and microBrain plates, as well as microBrain 3D spheroid plates, for high-throughput screening that offers precise measurement of human tissue responses. The company chose to focus on brain and heart cells in part because neurotoxicity and cardiotoxicity are the top two reasons why drugs fail for safety reasons during clinical trials.
The microHeart cultures are grown along microchannels to form sarcomeric unit structures like those found in heart muscle—complete with physical markers, correct ion-channel formation, and a unidirectional contraction.
The microBrain 2D and 3D cells are a mix of astrocytes and neurons that form synapses as well as being rudimentary neural networks. The 2D culture is key for measuring visual changes in the cells, such as neural projections. The 3D spheroid cultures are harder to visualize, but they exhibit spontaneous, synchronous neural firing, which can be quantitatively screened for drug responses that influence that firing. Similarly, the cardiomyocyte contractions can be quantitatively assayed for drug responses, such as arrhythmias.
“No model’s perfect,” acknowledges Yeh, but he finds StemoniX’s heart and brain models extremely useful, “because they are human cells that are structured correctly, easily measurable, and give more predictive and reproducible data at a fraction of the cost of transgenic animals.”
Micropatterning allows researchers to add sophistication to cultures, for example, by manipulating signaling molecules or growth factors in specific patterns or gradients, but with more experimental control to influence signaling than in whole-animal models.
While micropatterning techniques have been available for use with 2D cultures, the current techniques are crude when it comes to designing gradients or layering two or more proteins. One such technique, microcontact printing, uses a stamp dipped into protein solution to imprint it onto a culture surface, but lacks precision for printing specific quantities or aligned proteins. To address those limitations, a trio of academic researchers at France’s National Center for Scientific Research (CNRS) dreamt up PRIMO, a contactless and maskless technique for micropatterning, and started up Alvéole.
PRIMO is based on light-induced molecular adsorption of proteins (LIMAP) and combines a UV illumination system with a photoactive reagent. The PRIMO box can be attached to an inverted microscope. It contains a UV laser and a micromirror device projector that can project any image file from a computer onto the surface of culture plates through the microscope’s objective lens, with a resolution of 1.2 μm.
Glass slides, coverslips, plastic plates, and also hydrogels can first be treated with polyethylene glycol (PEG) as an antifouling agent, then the PLPP photoinitiator reagent is added in solution. When combined with the projected UV light in the pattern of choice, the PLPP degrades the PEG, leaving a “hole” to be filled by the protein(s) of choice. PRIMO’s Leonardo software controls the movements of the motorized microscope stage for aligning or shifting patterns. Also, using the software’s 256 grayscale levels to vary the UV intensity allows researchers to control how much PEG is degraded and therefore how much protein adheres to the surface.
For 3D applications, protein micropatterns laid down on a coverslip can be transferred to the surface of a hydrogel, since illuminating hydrogels directly with UV light can change their rigidity. Researchers have also used PRIMO to coat the bottom and sides of a polydimethylsiloxane polymer microwell to grow hepatocytes in a particular 3D configuration. Another group used PRIMO to grow human induced pluripotent stem cells (hiPSCs) into astrocytes oriented along thin, straight lines to measure microtubule growth dynamics in a patient with the neurodegenerative disease Rett syndrome (3).
“Using PRIMO, we helped that group try different shapes and widths of lines in just a few weeks—with other micropatterning techniques it would have taken months,” says Marie-Charlotte Manus, operational marketing manager at Alvéole in Paris.
The system is not well suited to high-throughput imaging or screening yet, says Manus, although the company is working on a version to use with multiwell plates. For now, individual researchers can use PRIMO to get closer to the in vivo microenvironment by optimizing the signal patterns that their cell cultures experience.
More precise CRISPR models
Gene editing technology is bringing a whole new level of genetic manipulation to human cell cultures, whether they are transformed cell lines, patient-derived primary cultures, or iPSCs. While the research world is not morally ready to consider making knockout humans yet, technologies like CRISPR/Cas9 allow researchers to knock out specific genes quickly from human cell cultures and then screen for desired properties.
“In the postgenome era, we can interrogate biological systems in ways we have never thought about before,” says Benedict Cross, head of functional genomic screening for Horizon Discovery in Cambridge, United Kingdom. “Taking a whole-genome, systems approach is not easy to do in a reasonably controlled way in whole animals.”
Horizon’s CRISPR Knockout Screening platform can knock out every single gene, or suites of genes related to a specific process, such as apoptosis, in a population of human cells, and then measure how those altered cells respond to drug compounds. It can be used to identify novel drug targets, to find genes related to drug sensitivity or resistance, or to select patients who are the best candidates for clinical trials.
Each CRISPR knockout guide RNA encodes a barcode unique to the gene it is knocking out, and all the cells are kept together as a mixed population of knockouts, whether they are grown in 2D monolayers, 3D spheroids, or in suspension. One set of cells is subjected to an experimental treatment, such as a drug, and another set is kept as a control. The treated cells that are enriched for the desired activity above the control cells are isolated and their barcodes sequenced to reveal their missing gene.
“It’s quite precise and quantitative at that point,” says Cross. The higher the frequency of a gene’s barcode, the more likely the inhibition of that gene leads to the desired biological activity. “It’s a phenotypic screen with a whole-genome level analysis.”
Horizon has also tweaked the CRISPR/Cas9 system to create two other screening platforms, CRISPR Interference (CRISPRi) and CRISPR Activation (CRISPRa). CRISPRi knocks gene message levels down, but not entirely out, and CRISPRa amps up the targeted gene’s expression.
All three CRISPR screening platforms can be used in parallel to study knockouts, loss-of-function, and gain-of-function of the same genes. “Interrogating a biological process is massively improved when you are able to look in both directions, to find things that activate and inhibit the pathway endogenously,” says Cross.
Through its acquisition of Sigma Advanced Genetic Engineering (SAGE) in 2014, Horizon Discovery has also deployed CRISPR technology to make improved and “humanized” rat models for drug discovery and safety testing.
As Kevin Forbes, manager of Horizon Discovery’s in vivo R&D group in St. Louis, Missouri, explains, the rise of mouse models in drug discovery was tied to the ease of making genetic knockouts through embryonic stem cell (ESC) technology. However, prior to that, rats were the preferred models because their larger size allows for multiple blood draws and longer-term compound studies, and because their prosocial behavior and slower metabolisms are more analogous to human biology.
Horizon can now make custom gene-edited rat, mouse, and rabbit models, with modifications ranging from small point mutations to megabase deletions, including knockout–knockin models in which the animal gene is replaced with the human version of a gene. Horizon also has “off-the-shelf” rat knockout models for use in toxicology, oncology, and cardiovascular disease studies as well as models for research into neurological disorders, including autism, Parkinson’s disease, and Alzheimer’s disease.
One rat model allows the study of drug absorption, distribution, metabolism, excretion, and toxicology (ADME/Tox), because it has been gene-edited to remove the rat versions of three nuclear receptors, PXR, CAR, AHR, and the major drug-metabolizing enzyme cytochrome P450 CA4. The receptors act as xenobiotic sensors and turn on the cytochrome P450 genes to metabolize “foreign” drugs. In these ADME/Tox rats, the human versions of those four genes were edited back in. Those rats, in theory, should “see” novel drugs in the same way that human small-intestine and liver cells would detect drugs to set their metabolism in motion.
Forbes describes CRISPR/Cas9 as a more efficient and faster process for making complex transgenic models such as the ADME/Tox rat when compared to ESC-based transgenic methods, estimating that CRISPR shaves off as much as a year from the process.
These “humanized” rats will soon provide a whole-animal solution for preclinical drug safety testing that is much more advantageous than testing on human liver cells in vitro. For example, cell cultures won’t catch tumors formed elsewhere in the body, nor pick up cardiac or cognitive problems caused
Of course, drug development may never eliminate the use of animals, because they represent the biological complexity of integrated organ systems, Yeh notes. But these new technologies should greatly reduce both numbers and types of animals needed for drug testing in the future.
As models improve in their “humanness,” drug developers should be able to better predict how compounds will act in the human body—avoiding the problems that doom so many candidate drugs and harm patients like Yeh. Inspired by his recent close call, he keeps working to build better models: “If a model has enough of the right parts, connected correctly, then we’ll continue to improve the functionality.”
1. M. Hay et al., Nat. Biotech. 32, 40–51 (2014).
2. S. Kota et al., Oncogene 37, 4372–4384 (2018); doi: 10.1038/s41388-018-0257-5.
3. C. Delépine et al., Hum. Mol. Genetics. 25, 146–157 (2016); doi: 10.1093/hmg/ddv464.
Newly offered instrumentation, apparatus, and laboratory materials of interest to researchers in all disciplines in academic, industrial, and governmental organizations are featured in this space. Emphasis is given to purpose, chief characteristics, and availability of products and materials. Endorsement by Science or AAAS of any products or materials mentioned is not implied. Additional information may be obtained from the manufacturer or supplier.