Can bioreactors replace animals?
Three case studies on how to use bioreactors to test, grow, and scale tissue-engineered therapies
10 minute read
In 2021, 820,000 animals were used for research in the United States. According to reports compiled by the USDA as part of the Animal Welfare Act. 16% (130,000) of those 820,000 animals were pigs. In all honesty, I thought that percentage would be a lot higher; pigs are extremely anatomically similar to humans, and because of those similarities, they are used to evaluate new treatments for a myriad of diseases. You may remember last year’s highly publicized heart transplant that was grown inside a “designer pig.” But that’s beside the point. The fact is that 130,000 pigs are used in research every year.
Ultimately, that number pales in comparison to the 128 million pigs raised for slaughter in 2023 in the United States alone. But, as scientists, we have the responsibility and the power to reduce the unnecessary use of animals in research, to keep driving that 130,000 to zero.
Many of you are familiar with the 3Rs—replacement, reduction, and refinement. These principles urge us to avoid using animals in research whenever possible, reduce the number of animals used to collect data, and refine procedures in order to minimize distress and maximize welfare.
Until a few weeks ago I was less familiar with the 6Rs: robustness, registration, and reporting, in addition to replacement, reduction, and refinement. Strech and Dirnagl argue in their 2019 paper on the topic that the addition of these three new Rs is simple: Even if an animal study follows the 3Rs principle, it can fail to deliver high quality results. “Studies can be robust but reported in a biased or otherwise inappropriate way. Alternatively, they can be appropriately reported but not robust. Both scenarios compromise the value of the study.”
Kyra Smith, a PhD student in the Gottardi Bioengineering and Biomaterials Lab at the University of Pennsylvania, believes that studies can compromise the 6R principles in many ways, even when authors simply don’t report negative data. “That leads to somebody else doing the exact same animal study just to fail. That’s more animal lives,” Smith explains. “And it’s the same for robustness. If you do an animal study and you don't have your good controls, you don't know what you need to do going in, you don't have the good endpoints or good readouts, then it's a waste.”
So how do we satisfy all 6Rs? Smith believes we use well-designed bioreactors more often.
Bioreactors! Some of us love them. Some of us make peace with the fact we need to, in all likelihood, be using them. You probably already know a great deal about them. At least I thought I did. I knew they came in many different forms. Huge 1,000 liter stir tanks used to cultivate mass quantities of cells. Well-plates capable of stretching cellular monolayers. Flexible tubes and chambers perfused with liquid culture media to grow three dimensional tissue. I even found a study that built a “humanoid robot to mechanically stress cells grown in soft bioreactors.” Which is to say, the researchers built a biomimetic shoulder joint to better grow tendon from human fibroblasts. It’s wild stuff!
What I didn’t know was just how powerful bioreactors are for testing, growing, and scaling tissue-engineered therapies.
Using a custom-designed perfusion bioreactor, in which culture media continually flows through a series of cell culture chambers, Smith can simulate osteoarthritis, studying the crosstalk between cartilage and bone as the disease progresses. Smith does this by fixing a tissue-engineered construct with separate cartilaginous and boney regions into the custom device. Each region can be isolated from the other using separate input and output streams.
She explains, “We can give inflammatory signals to only the cartilage layer and not the bone layer and see how that affects the system. Because that’s more representative of osteoarthritis, where the cartilage is getting signals from the synovium, the synovium doesn't touch the bone, but the bone still sees its own degradation and metabolism.” You could do the same studies using osteochondral explants, too.
Smith’s bioreactor model adds needed complexity to the traditional study of explants in a well-plate, where the entire tissue is subject to the same environmental conditions, no matter what conditions each region would have experienced in the body.
Smith is hopeful that this bioreactor model of osteoarthritis “can provide strong preliminary data so that you go into your animal study having all the knowledge you need to have the most efficient animal study possible.”
Smith is hopeful that this bioreactor model of osteoarthritis “can provide strong preliminary data so that you go into your animal study having all the knowledge you need to have the most efficient animal study possible.”
So, bioreactors can be powerful tools for gathering preliminary data, for modeling diseases like osteoarthritis, for reducing the number of animals on which we need to test medical therapies. But researchers like Smith have just begun scratching the surface of what we can use them to model.
It’s a well-established phenomenon that people are constantly moving. That this movement—bending, walking, running, jumping, even sitting—loads our joints. Often people injure themselves through mechanical overloading; think skiers jumping out of helicopters, tearing their ACLs or postal workers lifting boxes day after day, herniating their intervertebral discs.
Dr. Hagar Kenawy, a recently-minted PhD from the Chahine Lab at Columbia University, explains that there’s a limited body of research showing this loading and overloading of the intervertebral disc “can lead to inflammation, which potentially could lead to lower back pain.” It’s incredibly difficult to study this relationship between loading and inflammation in human patients. And would be impossible to study in the lab without a bioreactor.
To better tease out this relationship in the lab, she uses mechanically-loading bioreactors. Specifically, Dr. Kenawy applies cycles of compression to mouse intervertebral discs and then quantifies the molecules that the disc cells secrete.
Unlike a traditional animal experiment, Kenawy can program her bioreactor to compress the discs in different patterns, for increasing durations of time, and with varying magnitudes of force, simulating both healthy and injurious loads. She can then compare her results to unloaded controls to better understand how disc cells respond to compression in many different forms.
The end goal for Kenawy? In addition to the identification of many inflammatory pathways that affect disc health, Kenway also identified molecules secreted by the disc cells with the potential to be anabolic. She hopes “we can use the loading [or these anabolic factors] to mitigate downstream inflammation and to condition cells to enter a healing phenotype.” Perhaps we could one day treat low back pain by implanting these conditioned cells directly into patients.
Kenawy hopes “we can use the loading [or these anabolic factors] to mitigate downstream inflammation and to condition cells to enter a healing phenotype.” Perhaps we could one day treat low back pain by implanting these conditioned cells directly into patients.
Testing tissue is one thing, but what about growing tissue? After all, Fleshy Futures is all about tissue engineering the 21st century, right?
Let’s go back to Kyra Smith’s perfusion bioreactor. Smith not only uses her bioreactor to study more complicated tissue crosstalk but she also uses it to grow cartilage and bone. More specifically, she uses the bioreactor to grow a gradient of cartilage and bone tissues that seamlessly spans the osteochondral interface. With the same device, she can both study osteoarthritis and grow its potential treatments. A powerful one-two sucker punch.
For tissue growth, Smith uses the perfusion system to deliver separate cocktails of growth factors to the top and bottom regions of a PLLA scaffold. These scaffolds have a gradient of pores that increase in size as you move down, with the idea being that the small pores activate chondrogenesis via “mesenchymal condensation.” This separation of culture media ideally stimulates osteogenesis near the bottom of the scaffold and chondrogenesis near the top, leading to the growth of tissue with a gradient of properties.
After using her bioreactor to acquire a plethora of preliminary data, Smith was able to design a thorough animal study that appropriately embodied the 6Rs. Smith implanted her engineered scaffolds in rabbit knees to determine whether or not they can repair defects that span the osteochondral interface. Her biggest question? Can these scaffolds adequately activate the healing capabilities of infiltrating cells and heal osteochondral defects on their own? Unfortunately, early signs point to no. It turns out scaffolds were much more effective at healing cartilage defects when preseeded with mesenchymal stromal cells and allowed to mature in the dual-phase bioreactor before implantation.
Bioreactors seem to be enabling the growth of avascular cartilaginous tissues for Smith. But what about large, vascularized tissues? Can we use bioreactors to grow those? What about tissues with more complex hierarchies?
These questions of pattern and organization keep Dr. John Yuen up at night. A former PhD student in the Kaplan Lab at Tufts and the current Entrepreneur-in-Residence and Director of Technology at Tuft’s Cellular Agriculture Commercialization Lab, Yuen knows a thing or two about growing tissue in bioreactors. Or attempting to grow tissue in bioreactors, anyway.
During his PhD, Dr. Yuen worked on applications for cultured meat, a field aiming to reduce our dependence on industrial animal agriculture, struggling to culture vascularized fat tissue in the lab. I’ll let him explain the endeavor in his own words:
“If you want to make cultivated meat, it has to be a big chunk. The contemporary limitation is you can't make something bigger than one millimeter without the blood vessels, because the [fat and muscle] cells will die for lack of nutrition and waste removal. We were trying for years and years to go the perfusion bioreactor route, to send the culture media into the middle of the tissue so the cells in the middle don’t die.
We were trying to work in the blood vessels, regardless of whether it's low-cost enough for meat production, but just to make it happen. Then one day I looked at the picture of adipose tissue under the microscope and essentially, at least from a food taste/texture perspective, it’s just big spheres of fat cells packed together and aggregated. So we thought: ‘Maybe we can just grow those spheres of fat cells individually, eliminate the nutrient diffusion issue, and pack them together at the end of cell culture instead of growing them already packed together where you need the blood vessels.’
So we tried it, and it seems to work. The moment you take the individual fat cells and pack them together visually from the naked eye, it suddenly looks like fat. Maybe liposuction is the most similar thing.
We would like to think that the significance of this paper is that it opens a door to a whole new approach of producing fat tissue for food, and it’s food specific because once you pack the cells together, the fat cells will die because they don’t have the blood vessels. For food, its analogous to when you sacrifice the animal, and the cells die. It’s a similar process in that way, and therefore okay for our application.”
For Yuen, the solution to his bioreactor problems was a relatively simple one. Forgo the bioreactor entirely. Forgo the headaches that come with a dynamic, and sometimes demanding system. Forgo the need to pattern and grow vasculature for tissue en-masse. Trade all those complications for the production of smaller components that can then be layered together into the product. It’s a simple idea, probably pretty efficient, and definitely works with some existing manufacturing infrastructure. And I find that approach to be quite inspiring.
“We would like to think that the significance of this paper is that it opens a door to a whole new approach of producing fat tissue for food. We can just grow spheres of fat cells individually, instead of growing them already packed together where you need the blood vessels,” Yuen explains.
In a failing effort not to sound too cheesy, let me sum this all up. Initially, I set out to learn more about what bioreactors are, and these three brilliant PhDs (and soon to be PhDs) ended up teaching me less about what bioreactors are and more about the incredible things bioreactors can do.
Kenawy showed me that mechanically-loaded bioreactors can help us answer questions that would be nearly impossible to ask using traditional benchtop or animal studies.
Smith reminded me that there’s still an overwhelming need to refine our experimental methods, especially when they involve working with animals. Needs that when filled by novel approaches—like new bioreactors—can open up the door to better answer complicated questions such as tissue crosstalk.
And Yuen opened my eyes to the idea that solutions to complicated problems of scale and manufacturing may already be within our reach. That they may be inherently simple. And that we just need to open our minds and look a little harder under the microscope to see them.
Yes, bioreactors are powerful tools, but they might not be the correct solution to every tissue engineering problem. But for the engineering problems we need their help solving, why don’t more people use them? More on that next time.