14 Min Read



Production of therapeutic proteins typically involves culture of genetically engineered hosts, such as Chinese hamster ovary (CHO) cells. Expressed proteins then are harvested, purified, filtered, formulated, and ultimately, filled into vials to be packaged as products for administration by injection or intravenous infusion. Biopharmaceutical manufacturers and industry suppliers continue to improve the productivity of their expression systems by applying rational approaches for single-gene overexpression, knockout, or knockdown (1). But as I learned from Martin Fussenegger (professor of biotechnology and bioengineering in the Department of Biosystems Science and Engineering at ETH Zurich as well as professor of life sciences at the University of Basel, both in Switzerland), advances in synthetic biology now enable pharmaceutical scientists to investigate different production and treatment paradigms, including means for tunable in vivo expression of therapeutic proteins.

Much of Fussenegger’s research explores strategies for equipping mammalian somatic cells with gene circuits, modified or wholly synthetic gene networks that biochemically associate a given stimulus with a desired cellular output, such as initiation of therapeutic-protein secretion (2–5). Such networks enable novel drug-development strategies. For instance, circuits can be integrated into human cells during ex vivo engineering. The resulting “designer cells” can be encapsulated within semipermeable polymers to protect against patient immune responses while permitting import and export of key molecules. Upon implantation into patient tissues, the encapsulated cells will perform their therapeutic and/or diagnostic functions in response to physiological conditions.

In October 2023, I spoke with Fussenegger about the utility of gene circuits and designer cells for protein expression. He highlighted different available approaches for circuit design, including open- and closed-loop systems. The latter strategy provides mechanisms for dynamic, continuous control of physiological conditions in a patient’s body. Thus, he explained, closed-loop circuits hold particular promise for treating neurological, psychiatric, and metabolic disorders against which current therapies have short-lived efficacy and high potential for side effects from poor dose control.

High-Order Control of Gene Expression
How would you describe gene circuits relative to other gene-modification approaches? What kinds of components do gene circuits have, and what can such systems do as a result? Before the establishment of synthetic biology as a field, researchers investigated what they called gene-control systems. Today, we call them gene switches. Perhaps the earliest such system was one responsive to tetracycline (6). Over time, research into gene-control mechanisms gradually moved into the fields of small-molecule drugs, food additives, volatile agents, and now optogenetics (7–9) and electrogenetics (10, 11). For me, the term gene circuit connotes multiple gene switches or feedback loops. Circuits involve higher-order control than what a single gene switch can provide.

One example is the synthetic mammalian-cell oscillator that ETH Zurich colleagues and I introduced in 2009 (12). The oscillator encodes for both a positive and a time-delayed negative-feedback loop, providing mammalian cells with an autoregulated mechanism for sense–antisense transcription control.

Initially, gene circuits were based on open-loop controls, which require external signal initiators to turn cell outputs on or off. Such inducers have included chemicals, heat, light, electricity, and so on (13–15). Today, researchers also can create circuits with closed feedback loops, in which cells provisioned with gene circuits detect specific biomarkers — e.g., peptides or small molecules associated with a metabolic disease — and respond with precise dosing of a therapeutic, perhaps a needed enzyme or other such protein. For either approach, several logical functions are possible, including and, or, and not gates as well as half and full adders (16).

Our laboratory has developed several circuits that work on a closed-loop principle, including one for treatment of gouty arthritis (17). Cells are programmed to sense uric acid in real time and put out urate-degrading enzymes accordingly. Other examples include a sensor for blood fatty-acid content (18). That circuit is linked to release of satiety hormones, which help patients to feel full and stop eating until their blood fatty-acid levels decrease again. A similar mechanism is used with a diabetes circuit that our team developed (19). It enables constant monitoring of a person’s blood glucose. If programmed cells detect that levels are too high, then they are triggered to release insulin, which brings blood sugar levels down.

I like that closed-loop approaches work according to homeostasis, which our bodies naturally try to achieve to maintain our health. Engineered gene circuits could be used to replace broken circuits in a patient’s body or to help correct metabolic disorders. In that way, gene circuits could become advantageous when developing cell-based therapies.

Could gene circuits be applied during biomanufacturing — e.g., during development of host-cell lines that express therapeutic antibodies ex vivo? Or are gene circuits best suited for in vivo protein expression? I am unsure about how biomanufacturers might want to use gene circuits. But therapeutic proteins often impose burdens on cell health during culture. So, for instance, if you notice that product proteins are impairing cell growth or creating slightly cytotoxic conditions in the culture environment, then you might integrate gene circuits into your production cell line to fine-tune expression levels. Titrating product proteins in that way could help to produce biopharmaceuticals in optimal ranges. That said, I believe that regulators are likely to become familiar with gene circuits through designer cells and related therapies that are coming through research and development. I understand that biomanufacturers can be reluctant to present new approaches to health authorities because of the strict regulations. So advancement is likely to come with more novel treatment approaches.

Design Considerations
How do you approach gene-circuit design? What kinds of factors need consideration? For closed-loop systems, you need first to identify an appropriate sensor. Gene-circuit function depends on a cell’s ability to detect given biomarkers in real time. Thus, you must consider mechanisms for accurate detection. The same goes for sensitivity. You need to ensure that detection is sensitive enough that outputs will achieve homeostatic concentrations. Outputs should be nontoxic and produced or released within physiological ranges.

As an analogy, consider how much work our bodies perform to measure glucose levels. Our bodies give several opportunities to sense glucose. Taste receptors in our tongues tell us whether foods are sweet. However, sugar levels are much higher on our tongues than in our blood when we taste something. Thus, our taste receptors are insufficient for measuring blood glucose. We have similar receptors in the gut that scan dietary intake for sugar molecules and prepare the digestive system accordingly by triggering insulin release. There, too, sugar levels are much higher than they are in the bloodstream. The challenge for the body is to ensure that its sensor inputs (detected glucose levels on the tongue or in the gut) and outputs (insulin release) are calibrated to keep conditions in a physiological range. If your body produces too much insulin, then you become hypoglycemic.

In much the same way, sensing and dosing (based on sensor sensitivity) are the most important considerations for gene-circuit design. A couple strategies are available for tuning sensor sensitivity. Depending on what sensor type is used — e.g., a transcription factor or a G-protein–coupled receptor (GPCR) — we can adjust the amount of sensor that we establish in the cell line. The more receptor or transcription factor that we add, the more sensitive the system will be. Thus, we can tune signal-link escape by overexpressing critical components.

In our laboratory, we often rely on fine-tuning target promoters with the enhancer models, distant to minimal promoters, and other such components. We have tricks to determine whether a sensor system is too “leaky,” or insufficiently sensitive. In principle, the sensor can be mutated, but we try to keep it as intact as possible. Closed feedback loops are helpful in this context because their outputs are self-adjusting.

Many of BPI’s readers work with CHO cell lines. Do cell-line–specific concerns arise during gene-circuit design? Each cell line has distinctive host proteins, which can be expressed in abundant or limited levels. But in principle, gene circuits can be designed and integrated into almost any cell line. Our laboratory often works with human embryonic kidney (HEK293) cells and mesenchymal stem cells (MSCs), although we routinely test gene switches in lines that are standard for biopharmaceutical manufacturing — e.g., CHO and baby hamster kidney (BHK) cells. HEK systems happen to be ideal for our applications because they generate all of the G proteins, signaling cascades, and kinases that we need. Depending on the application, work with CHO cells might involve supplementation or overexpression of G proteins.

What aspects of gene-circuit design present the largest obstacles? As we discussed, the first and most difficult challenge is to achieve the right sensor sensitivity so that outputs will fall into physiological ranges. The second biggest concern is to address leakiness in a system’s basal gene expression. In your house, if you flick a simple light switch, then the light should turn off rather than dim. Biology is not digital; it usually does not result in all-or-nothing outcomes. I often tell my students and colleagues that biological systems such as gene circuits will show 5–10% variation from expected outcomes. Slight differences are to be expected because of the systems’ molecular composition and because of interference from epigenetic factors.

A debate has emerged in the research community about whether to use orthogonal controls to program cellular behavior. Generally, I do not like the idea of external controls for gene circuits. Having multiple strange components in a patient’s body could result in side effects. I would rather design an endogenous system and accept risks for some level of interference. After all, that is how our bodies work. Our cellular processes experience all kinds of interference, but our bodies manage.

Issues with leakiness typically involve basal gene expression being too high. We like for our circuits to effect a basal-expression switch of five- to 10-fold. We do not need expression rates to be substantially lower or higher than that. Natural metabolic processes make do with even twofold switches. Although we do not need to be exact about leakiness, tightness is of the essence. Below a certain threshold, genetic control becomes nearly impossible. On the other hand, leakiness can deregulate your system over time. For example, if you are scoring a cancer biomarker over time, persistent leakiness will cause the circuit to be triggered without measuring the biomarker.

What synthetic-biology tools does your laboratory use to design gene circuits, and what kinds of technologies or techniques would enhance such work? We use a breadth of standard genetic-engineering techniques. Our studies mostly involve empirical approaches — educated guesses and some trial and error. Although some organizations are implementing machine-learning (ML) algorithms to help with discovery efforts, we have yet to use those technologies. Sometimes, such tools seem unnecessary. Nature works by subtle differentiation and gradual selection; often, we can operate the same way with our experiments.

An interesting problem has emerged, however. Circuits have become so complex that they can be difficult to establish in human cells. As my team discovered when working on the mammalian-cell oscillator, it is increasingly difficult to have full adders and all your other different components operating in one cell. You have limited control over expression of a circuit’s individual components. So where can you put them in a given chromosome, and how susceptible will they be to interference?

Such questions raise others: Must circuits be designed to operate inside a chromosome, or could they operate outside of it? I would love to have access to more-robust plasmid systems for human cells. In that way, a gene circuit could be delivered to a cell but preserved nicely on a distinct genetic entity that does not tamper with the host-cell chromosomes.

Problems with genetic crosstalk also require new solutions. When working with transcription processes, there is always potential for crosstalk among components. Our team has tried to dispatch transcriptional work to different cell types, which then communicate with each other. But that approach involves significant complexity, which is difficult to control. All told, your choices are limited when it comes to translational potential because you cannot introduce too many modifications to a chromosome.

What advantages might circuit-based cell therapies have over more conventional biopharmaceuticals? Why do you believe that researchers should persist in developing gene-circuit–based cell therapeutics? Gene-circuit–enabled therapeutics and standard biologics could deliver the same kinds of products, so the conceptual difference is not significant in that sense; using gene circuits, you just produce a biopharmaceutical in a patient’s body instead of a bioreactor. Clearly, cell-based therapies are expensive to produce and have critical regulatory issues to address, so much work remains to get such therapies to patients. But the dynamic nature of circuit-based therapies could be especially helpful for treating complex metabolic diseases. Rather than delivering a bulk of therapeutic proteins, cells bearing gene circuits could enable precise control of dosing depending on current patient conditions.

Consider conditions requiring hormonal control: neurodegenerative disease, generalized anxiety, depression, and other such disorders for which small-molecule approaches are ineffective or nonoptimal. The dynamic control of circuit-based therapeutics could win out over current treatments in cases that involve interfacing with metabolism.

The same can be said for therapies based on chimeric antigen receptor (CAR) T cells. Currently, CAR-T products offer little control in vivo: Once administered, the cells seek out and destroy cancer. But CAR-T activity can generate side effects such as cytokine storms. Researchers know that greater dose control is needed. Thus, next-generation CAR-T treatments could include genetic mechanisms for dose control — e.g., an and gate that prompts cells to initiate tumor destruction only in the presence of two or three specific biomarkers. Some such strategy could help to fine-tune cytokine release and associated processes. Without dynamic control of dosing, a biopharmaceutical works basically the same way as pills taken at regular intervals.

1 Fischer S, Handrick R, Otte K. The Art of CHO Cell Engineering: A Comprehensive Retrospect and Future Perspectives. Biotechnol. Adv. 33(8) 2015: 1878–1896; https://doi.org/10.1016/j.biotechadv.2015.10.015.

2 Kojima R, Aubel D, Fussenegger M. Building Sophisticated Sensors of Extracellular Cues That Enable Mammalian Cells To Work as ‘‘Doctors’’ in the Body. Cell Mol. Life Sci. 77, 2020: 3567–3581; https://doi.org/10.1007/s00018-020-03486-y.

3 Scheller L, Fussenegger M. From Synthetic biology to Human Therapy: Engineered Mammalian Cells. Curr. Opin. Biotechnol. 58, 2019: 108–116; https://doi.org/10.1016/j.copbio.2019.02.023.

4 Heng BC, Aubel D, Fussenegger M. Prosthetic Gene Networks As an Alternative to Standard Pharmacotherapies for Metabolic Disorders. Curr. Opin. Biotechnol. 35, 2015: 37–45; https://doi.org/10.1016/j.copbio.2015.01.010.

5 Mahameed M, Fussenegger M. Engineering Autonomous Closed-Loop Designer Cells for Disease Therapy. iScience 25, 2022: 103834; https://doi.org/10.1016/j.isci.2022.103834.

6 Gossen M, Bujard H. Tight Control of Gene Expression in Mammalian Cells By Tetracycline-Responsive Promoters. Proc. Nat. Acad. Sci. USA 89(12) 1992: 5547–5551; https://doi.org/10.1073/pnas.89.12.5547.

7 Müller K, et al. Optogenetics for Gene Expression in Mammalian Cells. Biol. Chem. 396, 2015: 145–152; https://doi.org/10.1515/hsz-2014-0199.

8 Joshi J, Rubart M, Zhu W. Optogenetics: Background, Methodological Advances and Potential Applications for Cardiovascular Research and Medicine. Front. Bioeng. Biotechnol. 7, 2019: 466; https://doi.org/10.3389/fbioe.2019.00466.

9 Emiliani V, et al. Optogenetics for Light Control of Biological Systems. Nature Rev. Meth. Prim. 2, 2022: 55; https://doi.org/10.1038/s43586-022-00136-4.

10 Mansouri M, Fussenegger M. Electrogenetics: Bridging Synthetic Biology and Electronics To Remotely Control the Behavior of Mammalian Designer Cells. Curr. Opin. Chem. Biol. 68, 2022: 102151; https://doi.org/10.1016/j.cbpa.2022.102151.

11 Brier ML, Dordick JS. Remote Activation of Cellular Signaling. Science 368(6494) 2020: 936–937; https://doi.org/10.1126/science.abb9122.

12 Tigges M, et al. A Tunable Synthetic Mammalian Oscillator. Nature 457(7227) 2009: 309–312; https://doi.org/10.1038/nature07616.

13 Doshi A, et al. Small-Molecule Inducible Transcriptional Control in Mammalian Cells. Crit. Rev. Biotechnol. 40, 2020: 1808583; https://doi.org/10.1080/07388551.2020.1808583.

14 Stefanov BA, et al. Genetically Encoded Protein Thermometer Enables Precise Electrothermal Control of Transgene Expression. Adv. Sci. 8(21) 2021: e2101813; https://doi.org/10.1002/advs.202101813.

15 Krawczyk K, et al. Electrogenetic Cellular Insulin Release for Real Time Glycemic Control in Type 1 Diabetic Mice. Science 368, 2020: 993–1001; https://doi.org/10.1126/science.aau7187.

16 Wong A, et al. Layering Genetic Circuits To Build a Single Cell, Bacterial Half Adder. BMC Biol. 13, 2015: 40; https://doi.org/10.1186/s12915-015-0146-0.

17 Kemmer C, et al. Self-Sufficient Control of Urate Homeostasis in Mice By a Synthetic Circuit. Nature Biotechnol. 28, 2010: 355–360; https://doi.org/10.1038/nbt.1617.

18 Ye H, et al. Self-Adjusting Synthetic Gene Circuit for Correcting Insulin Resistance. Nature Biomed. Eng. 1, 2017: 0005; https://doi.org/10.1038/s41551-016-0005.

19 Rössger K, et al. A Closed-Loop Synthetic Gene Circuit for the Treatment of Diet-Induced Obesity in Mice. Nature Comm. 4, 2013: 2825; https://doi.org/10.1038/ncomms3825.

Brian Gazaille, PhD, is managing editor of BioProcess International; 1-212-600-3594; [email protected].

You May Also Like