Light-Controlled Mammalian Cells and Their Therapeutic Applications in Synthetic Biology

Abstract

The ability to remote control the expression of therapeutic genes in mammalian cells in order to treat disease is a central goal of synthetic biology-inspired therapeutic strategies. Furthermore, optogenetics, a combination of light and genetic sciences, provides an unprecedented ability to use light for precise control of various cellular activities with high spatiotemporal resolution. Recent work to combine optogenetics and therapeutic synthetic biology has led to the engineering of light-controllable designer cells, whose behavior can be regulated precisely and noninvasively. This Review focuses mainly on non-neural optogenetic systems, which are often used in synthetic biology, and their applications in genetic programing of mammalian cells. Here, a brief overview of the optogenetic tool kit that is available to build light-sensitive mammalian cells is provided. Then, recently developed strategies for the control of designer cells with specific biological functions are summarized. Recent translational applications of optogenetically engineered cells are also highlighted, ranging from in vitro basic research to in vivo light-controlled gene therapy. Finally, current bottlenecks, possible solutions, and future prospects for optogenetics in synthetic biology are discussed.

Keywords: cell engineering; mammalian cells; optogenetics; synthetic biology.

Figure 1

Opsin‐based photoreceptors: a) natural G‐protein‐coupled (GPC) photoreceptors such as melanopsin or animal rhodopsin signal through endogenous pathways. The specificity of signaling is determined by the different alpha subunits in the trimeric G‐protein complex. Gαq subunit–dependent signaling is linked to increased levels of inositol‐3‐phosphate (IP3) and diacylglycerol (DAG), whereas Gαt signaling mainly results in lowered cGMP levels that trigger subsequent signal amplification steps in photoreceptor cells of the mammalian retina. The intracellular binding loops for the trimeric G‐protein were exchanged to reroute Gαt to Gαs or Gαq signaling in engineered versions of bovine rhodopsin. These membrane‐bound GPC photoreceptors are retinal‐dependent. b) Exposure to light leads to activation of the bound retinal conformer and a subsequent change in conformation that is translated to structural changes in the apoprotein. While microbes incorporate mainly all‐trans‐retinal that is activated to 13‐cis‐retinal in melanopsin, mammals use 11‐cis‐retinal that is transformed into its all‐trans‐conformer upon illumination. c) Besides GPCRs, there are various GPCR‐like photoreceptors with high structural and amino‐acid sequence homology that do not act via G‐proteins. These receptors enable transport of protons or ions across the membrane. Examples of this class of receptors are the light‐driven proton pumps microbial rhodopsin and halorhodopsin. Photoswitchable ion channels enable passive transport of cations across the membrane. Chimeric photoreceptors combine features of GPCRs and microbial rhodopsin and can trigger both G‐protein‐dependent signaling and changes in membrane potential.

Figure 2

Nonopsin photoactivatable proteins: a) naturally occurring and engineered photoactivatable proteins can be activated by light of different wavelengths ranging from ultraviolet to infrared. Different mechanisms of activation are known: association, dissociation, uncaging, or unhinging as well as direct photoactivation of enzymes. For example, UV‐B light triggers a stress response in plants which is mediated by dimerization of UVR8 proteins. LOV domains from AsLOV2 and AtLOV2 are stimulated by blue light, causing the then‐unhinged C‐terminal alpha helix Jα to bend outward, activating a kinase domain in the natural protein. The CRY2 and CIB1 pair mediates responsiveness to blue light in plants via a blue light–induced heterodimerization mechanism. CRY2 also has a tendency to cluster into homo‐oligomers upon activation. In the Magnet system, engineered variants of the VVD domains are either positively or negatively charged. Upon exposure to blue light, they only bind to each other, but do not form homodimers. The Dronpa145N mutant of the fluorescent protein Dronpa was found to form homotetramers in the dark that dissociate on exposure to cyan light, just like the wild‐type protein, but at lower concentrations. UV light reverses this reaction. The photoreceptor protein PhyB binds to its partner PIF6. Far‐red light–induced dissociation of the complex reverses this interaction. The bacteriophytochrome BphP1 forms homodimers in the dark, but forms heterodimers with PpsR2 upon exposure to near‐infrared light. This transition is reversible and the backward reaction can be boosted by exposure to red light. Soluble guanylate cyclases are a class of photoactivatable enzymes that are activated by blue light through their BLUF (blue light sensor using FAD) domain. b) Photosensors can be activated by a specific wavelength in the UV–vis spectrum. While shorter wavelengths have higher energy, they also show greater phototoxicity, damaging DNA and perturbing pathways that involve chromophore‐binding proteins. Additionally, tissue penetration is limited for blue light compared to red light. In the retina as well as in many microorganisms, there are opsins for every section of the visible spectrum. However, most light‐inducible systems available for molecular biology are activated exclusively in the blue and green ranges.

Figure 3

Applications for gene regulation and cell signaling: a) to modulate protein levels and localization directly and noninvasively with light, AsLOV2 can be fused to a transcription factor linked via a nuclear localization signal. This construct unfolds upon exposure to blue light, making the NLS accessible to import factors that mediate translocation into the nucleus, where the transcription factor (TF) can act on its promoter to increase gene expression. In another strategy, AsLOV2 is used as a photoswitchable degradation tag for a N‐terminally fused cargo protein. Again, blue light–induced unhinging is used to reveal the C‐terminal tetrapeptide degradation tag RRRG. Subsequently, the entire protein is degraded by proteases. In order to build a secretion system that can be precharged and timed, dimers of UVR8 were fused to a cargo protein targeted for secretion and tethered to the ER membrane. In the dark, the UVR8 tandem dimers form large clusters that effectively block transport from the ER to the Golgi. Upon exposure to UV light, these homo‐oligomers dissociate, allowing transport to the Golgi. b) In addition to regulating protein levels, photosensitive protein switches can be used to directly interfere with endogenous signaling pathways. To this end, AsLOV2 was fused to Rac1, masking the binding site of Rac1. Upon light induction, the cognate binding partner Pac1 interacts with Rac1 and the respective signaling cascade is triggered, leading to morphological changes in the cell with high spatiotemporal resolution. The OptoFGFR system was used to build a whole array of inducible switches for the mitogen‐activated protein kinase/extracellular signal‐regulated kinase (MAPK/ERK) pathway by fusing engineered CRY2PHR, VfAU1–LOV, CPH1s‐o, or the cobalamin binding domain of Thermus thermophilus (TtCBD or CarH) to the intracellular tyrosine kinase domain of the FGFR1 receptor of human or mouse. The resulting chimeric receptors are sensitive to blue, green, and red light, resulting in an ON‐type switch, except for the TtCBD‐based receptor which showed OFF‐switch behavior. In another example, overexpression of BLUF domain–coupled photoactivatable guanylate cyclases (BLUF–PACs) in mammalian cells was used to modulate second messenger levels (e.g., cAMP) in response to blue light.

Figure 4

Photoregulatable Cas9 and TALE: the versatile genome‐editing system CRISPR/Cas9 provides highly precise binding to specific sequences in genomic or plasmid DNA. a) Recently, the catalytic moiety of the system, the SpCas9 enzyme, was engineered to be light sensitive by applying the heterodimerization and uncaging approach. To engineer a photoactivatable dimerizing variant of SpCas9, called paCas9, wild‐type SpCas9 was split in two parts that were fused to either pMag or nMag tags. Upon exposure to blue light, the pMag and nMag tags dimerize to reconstitute functional paCas9 that is able to bind to and cleave DNA. A different strategy was used to block the DNA‐binding site of SpCas9 via a Dronpa dimer that was fused to the wild‐type enzyme. Here, Dronpa dissociates upon stimulation with blue light, uncovering the binding domain and allowing the functional enzyme to work on the DNA. This fusion construct was termed photoswitchable Cas9 (psCas9). b) A catalytically dead version of SpCas9 (dCas9) was used to engineer a light‐inducible gene expression system based on CRY2–CIB1 interaction. In this example, CIB1 was fused to dCas9, and CRY2 was used to bind a transactivator to the DNA‐binding complex, while another system termed light‐inducible transcriptional effector (LITE) used engineered TALE proteins fused to CRY2 to bind CIB1 fused to the orthogonal transactivator VP64. While paCas9 and psCas9 are both blocked from binding to DNA, the dCas9 as well as the TALE component of these photoswitchable transcription systems resides on the DNA in the dark. Exposure to blue light recruits CRY2–VP64 to dCas9–CIB1 and CIB1–VP64 to TALE–CRY2, triggering assembly of the transcription initiation complex. c) In the epiLITE system, the aforementioned system was used to recruit inactivating epigenetic effectors (e.g., deacetylases, methylases) to a defined region on the genome, resulting in reduction of gene expression from the targeted locus.

Figure 5

Optotherapeutic strategies: several optogenetic implants have already been tested in mice and proved to be functional to treat diabetes as well as erectile dysfunction. a) An implant harboring synthetic mammalian designer cells equipped with blue light–induced melanopsin signals through the protein kinase C (PKC) signaling pathway. Activation of melanopsin results in influx of Ca²⁺ ions from outside of the cells and release of calcium from the ER. Ultimately, this leads to activation of the phosphatase calcineurin, which dephosphorylates NFAT and triggers expression of insulin from a synthetic NFAT‐responsive promoter. b) To tackle the delicate issue of erectile dysfunction, a blue light–inducible soluble guanylate cyclase (sGC) of bacterial origin was engineered to favor GTP over ATP as a substrate. 1) To evaluate the performance of the system more easily in cell culture, cAMP levels were read out by the cAMP‐responsive CRE pathway. 2) Activation of the enzyme also directly leads to increased cGMP levels, triggering relaxation of smooth muscle cells. To avoid side effects, this system was not used as a cell‐based therapy, but was integrated directly into smooth muscle cells in the endothelium of the corpora cavernosum of rats. c) Two other approaches capitalized on the soluble c‐di‐GMP synthase BphS, which can be directly activated by far‐red light. In one system, c‐di‐GMP activated a chimeric transcription factor based on the bacterial protein BldD that dimerizes and binds to the DNA in a c‐di‐GMP‐dependent manner and is fused to the transactivators VP64 and p65. To avoid overproduction of c‐di‐GMP and overactivation of the system, YhjH was coexpressed; it hydrolyzes c‐di‐GMP to form inactive pGpG. In another study, c‐di‐GMP was used as an inducer of the endogenous STING pathway, which activates the transcription factor interferon regulatory factor 3 (IRF3) through phosphorylation by TANK‐inding kinase 1 (TBK1). Activated IRF3 binds to IRF3‐binding sites on a synthetic promoter and triggers insulin expression.

Figure 6

State of the art of in vivo optogenetics: applications of optogenetics in vivo have to deal with the issue of tissue penetration of light and potential phototoxicity. Many optogenetic systems are dependent on high‐energy blue and green light, which does not penetrate deeply into tissues. Constructs that rely on red and near‐infrared light, such as the Bph1/PpsR2 or PhyB/PIF6 system, can be used in deep‐layer implants, because tissue penetration of light generally increases with wavelength. However, to expand the selection of optogenetic tools available for more advanced and multiplexed circuits, several solutions have been introduced by the scientific community. To enable activation of blue and green light–dependent circuits, upconverting nanoparticles (UCNPs) can be used. These nanoparticles collect energy from more than one photon and emit a single photon with a shorter wavelength. A straightforward approach to circumvent the tissue barrier is to directly equip the implant with its own light source that can be powered wirelessly via electromagnetic induction. With the advent of very small and efficient micro LEDs (µLEDs), this strategy is becoming increasingly feasible. Luminopsin is another tool that uses chemical energy stored in small molecules to drive light‐powered reactions. This is achieved by the localization of luciferases in the vicinity of the photoactivatable protein. Luciferases convert the exogenously provided high‐energy substrate coelenterazine into the low‐energy product coelenteramide while simultaneously transferring the light‐energy produced by the reaction to a nearby chromophore in a photosensitive protein in a process called bioluminescence resonance energy transfer (BRET).

Figure 7

Optogenetics in translational mammalian synthetic biology: the use of light‐controlled mammalian designer cells is gaining broad acceptance, with numerous applications appearing every year. Optogenetic therapy can lead to precise and specific treatment with reduced side effects in humans and in animal models. Additionally, light‐controllable production of pharmaceuticals reduces the need for external inducers that might cause problems in downstream processing. The use of light‐controlled signaling pathways for the implementation of ligand‐free, cost‐effective, and efficient screening systems can promote drug development.