Here, we present a protocol for harnessing the natural transformability of the edible algae Arthrospira platensis (common name: spirulina) to generate strains that express heterologous proteins. We describe the preparation of plasmids and the steps to grow A. platensis. We then detail the transformation and passage of the strains, followed by genomic DNA extraction and genotyping to assess integration of the gene of interest. This simple transformation protocol can be applied to genome manipulation of edible algae. For complete details on the use and execution of this protocol, please refer to Jester et al. (2022).1.

The use of the edible photosynthetic cyanobacterium Arthrospira platensis (spirulina) as a biomanufacturing platform has been limited by a lack of genetic tools. Here we report genetic engineering methods for stable, high-level expression of bioactive proteins in spirulina, including large-scale, indoor cultivation and downstream processing methods. Following targeted integration of exogenous genes into the spirulina chromosome (chr), encoded protein biopharmaceuticals can represent as much as 15% of total biomass, require no purification before oral delivery and are stable without refrigeration and protected during gastric transit when encapsulated within dry spirulina. Oral delivery of a spirulina-expressed antibody targeting campylobacter—a major cause of infant mortality in the developing world—prevents disease in mice, and a phase 1 clinical trial demonstrated safety for human administration. Spirulina provides an advantageous system for the manufacture of orally delivered therapeutic proteins by combining the safety of a food-based production host with the accessible genetic manipulation and high productivity of microbial platforms.

ABSTRACTArthrospira platensis(commonly known as spirulina) is a photosynthetic cyanobacterium1. It is a highly nutritious food that has been consumed for decades in the US, and even longer by indigenous cultures2. Its widespread use as a safe food source and proven scalability have driven frequent attempts to convert it into a biomanufacturing platform. But these were repeatedly frustrated by spirulina’s genetic intractability. We report here efficient and versatile genetic engineering methodology for spirulina that allows stable expression of bioactive protein therapeutics at high levels. We further describe large-scale, indoor cultivation and downstream processing methods appropriate for the manufacturing of biopharmaceuticals in spirulina. The potential of the platform is illustrated by pre-clinical development and human testing of an orally delivered antibody therapeutic against campylobacter, a major cause of infant mortality in the developing world and a growing antibiotic resistance threat3,4. This integrated development and manufacturing platform blends the safety of food-based biotechnology with the ease of genetic manipulation, rapid growth rates and high productivity characteristic of microbial platforms. These features combine for exceptionally low-cost production of biopharmaceuticals to address medical needs that are unfeasible with current biotechnology platforms.

Drugs which independently inhibit a shared target or pathway can have synthetic activities that result in multiplicative instead of merely additive potencies. This characteristic of drug combinations can be quantified by expressing the potency of the combination as if it were a single agent. We show that by optimizing this quantity we can prospectively design drug cocktails with apparent potencies that far exceed any of its individual components. We illustrate the power of this approach, which is based on statistical design of experiments to select optimal drug combinations, and response surface methodology to determine optimal drug ratios, by building a drug cocktail comprised of three antibodies for treating C. difficile infection that is almost 1000-fold more potent than the current, clinically approved antibody monotherapy. High synthetic activities do not require unusual drug interactions, and therefore may be achievable much more readily than generally appreciated. One-Sentence Summary A development pathway is described for designing antibody cocktails with potencies that far exceed what is achievable with single antibodies

Biosensors are important components of many synthetic biology and metabolic engineering applications. Here, we report a second generation of Saccharomyces cerevisiae digoxigenin and progesterone biosensors based on destabilized dimeric ligand-binding domains that undergo ligand-induced stabilization. The biosensors, comprising one ligand-binding domain monomer fused to a DNA-binding domain and another fused to a transcriptional activation domain, activate reporter gene expression in response to steroid binding and receptor dimerization. The introduction of a destabilizing mutation to the dimer interface increased biosensor dynamic range by an order of magnitude. Computational redesign of the dimer interface and functional selections were used to create heterodimeric pairs with further improved dynamic range. A heterodimeric biosensor built from the digoxigenin and progesterone ligand-binding domains functioned as a synthetic "AND"-gate, with 20-fold stronger response to the two ligands in combination than to either one alone. We also identified mutations that increase the sensitivity or selectivity of the biosensors to chemically similar ligands. These dimerizing biosensors provide additional flexibility for the construction of logic gates and other applications.

Biosensors for small molecules can be used in applications that range from metabolic engineering to orthogonal control of transcription. Here, we produce biosensors based on a ligand-binding domain (LBD) by using a method that, in principle, can be applied to any target molecule. The LBD is fused to either a fluorescent protein or a transcriptional activator and is destabilized by mutation such that the fusion accumulates only in cells containing the target ligand. We illustrate the power of this method by developing biosensors for digoxin and progesterone. Addition of ligand to yeast, mammalian, or plant cells expressing a biosensor activates transcription with a dynamic range of up to ~100-fold. We use the biosensors to improve the biotransformation of pregnenolone to progesterone in yeast and to regulate CRISPR activity in mammalian cells. This work provides a general methodology to develop biosensors for a broad range of molecules in eukaryotes.

Article Figures and data Abstract eLife digest Introduction Results Discussion Materials and methods References Decision letter Author response Article and author information Metrics Abstract Biosensors for small molecules can be used in applications that range from metabolic engineering to orthogonal control of transcription. Here, we produce biosensors based on a ligand-binding domain (LBD) by using a method that, in principle, can be applied to any target molecule. The LBD is fused to either a fluorescent protein or a transcriptional activator and is destabilized by mutation such that the fusion accumulates only in cells containing the target ligand. We illustrate the power of this method by developing biosensors for digoxin and progesterone. Addition of ligand to yeast, mammalian, or plant cells expressing a biosensor activates transcription with a dynamic range of up to ~100-fold. We use the biosensors to improve the biotransformation of pregnenolone to progesterone in yeast and to regulate CRISPR activity in mammalian cells. This work provides a general methodology to develop biosensors for a broad range of molecules in eukaryotes. https://doi.org/10.7554/eLife.10606.001 eLife digest Small molecules play essential roles in organisms, and so methods to sense these molecules within living cells could have wide-ranging uses in both biology and biotechnology. However, current methods for making new "biosensors" are limited and only a narrow range of small molecules can be detected. One approach to biosensor design in yeast and other eukaryotic organisms uses proteins called ligand-binding domains, which bind to small molecules. Here, Feng, Jester, Tinberg, Mandell et al. have developed a new method to make biosensors from ligand-binding domains that could, in principle, be applied to any target small molecule. The new method involves taking a ligand-binding domain that is either engineered or occurs in nature and linking it to something that can be readily detected, such as a protein that fluoresces or that controls gene expression. This combined biosensor protein is then engineered, via mutations, such that it is unstable unless it binds to the small molecule. This means that, in the absence of the small molecule, these proteins are destroyed inside living cells. However, the binding of a target molecule to one of these proteins protects it from degradation, which allows the signal to be detected. Feng, Jester, Tinberg, Mandell et al. use this method to create biosensors for a human hormone called progesterone and a drug called digoxin, which is used to treat heart disease. Further experiments used the biosensors to optimize the production of progesterone in yeast and to regulate the activity of a gene editing protein called Cas9 in human cells. The biosensors can be also used to produce long-term environmental sensors in plant cells. This approach makes it possible to produce a wide variety of biosensors for different organisms. The next step is to continue to explore the ability of various proteins to be converted into biosensors, and to find out how easy it is to transfer a biosensor produced in one species to another. https://doi.org/10.7554/eLife.10606.002 Introduction Biosensors capable of sensing and responding to small molecules in vivo have wide-ranging applications in biological research and biotechnology, including metabolic pathway regulation (Zhang et al., 2012), biosynthetic pathway optimization (Raman et al., 2014; Tang and Cirino, 2011), metabolite concentration measurement and imaging (Paige et al., 2012), environmental toxin detection (Gil et al., 2000), and small molecule-triggered therapeutic response (Ye et al., 2013). Despite such broad utility, no single strategy for the construction of biosensors has proven sufficiently generalizable to gain widespread use. Current methods typically couple binding of a small molecule to a single output signal, and use a limited repertoire of natural protein- (Tang et al., 2013) or nucleic acid aptamer-binding (Yang et al., 2013) domains, which narrows the scope of small molecules that can be detected. A general solution to small molecule biosensing should be adaptable to a range of small molecules and responses. A promising approach to biosensor design in eukaryotes uses conditionally stable ligand-binding domains (LBDs) (Banaszynski et al., 2006; Tucker and Fields, 2001). In the absence of a cognate ligand, these proteins are degraded by the ubiquitin proteasome system (Egeler et al., 2011). Binding of the ligand stabilizes the LBD and prevents its degradation. Fusing the destabilized LBD to a suitable reporter protein, such as an enzyme, fluorescent protein, or transcription factor, renders the fusion conditionally stable and generates sensor response (Figure 1a). Naturally-occurring LBDs can be engineered to be conditionally stable (Banaszynski et al., 2006; Miyazaki et al., 2012; Iwamoto et al., 2010), making it possible in principle to convert them into biosensors for target ligands. Designed LBDs can also be used, especially in cases for which natural binding proteins do not exist or lack sufficient specificity or bioorthogonality. Figure 1 with 2 supplements see all Download asset Open asset A general method for construction of biosensors for small molecules. (a) Modular biosensor construction from a conditionally destabilized LBD and a genetically fused reporter. The reporter is degraded in the absence but not in the presence of the target small molecule. (b) yEGFP fluorescence of digoxin LBD-GFP biosensors upon addition of 250 μM digoxin or DMSO vehicle. (c) yEGFP fluorescence of progesterone LBD-GFP biosensors upon addition of 50 μM progesterone or DMSO vehicle. (d) Positions of conditionally destabilizing mutations of DIG0 mapped to the crystal structure of the digoxin LBD (PDB ID 4J9A). Residues are shown as colored spheres and key interactions highlighted in insets. In b-c, fold activation is shown above brackets, (-) indicates cells lacking biosensor constructs, and error bars represent the standard error of the mean (s.e.m.) of three biological replicates. https://doi.org/10.7554/eLife.10606.003 Here, we convert a single designed LBD scaffold into multiple highly specific biosensors for the clinically relevant steroids digoxin and progesterone. We engineer LBDs fused with fluorescent reporters to be conditionally stable in the budding yeast Saccharomyces cerevisiae. Attaching these conditionally-stabilized LBDs to transcription factors (TFs) yields biosensors that respond to their target ligands with greater signal induction than observed with fusions to fluorescent proteins. We use TF-biosensors to improve the biosynthetic yield of progesterone in yeast. The biosensors retain function when ported directly into mammalian cells, with up to 100-fold activation over background, allowing us to develop a method for tight control of CRISPR/Cas9 genome editing. The biosensors also show up to 50-fold activation by ligand in Arabidopsis thaliana. The method presented here enables the rapid development of eukaryotic biosensors from natural and designed binding domains. Results Fluorescent biosensors built from engineered LBDs LBDs intended for biosensor development should recognize their targets with high affinity and specificity. We began with the computationally-designed binding domain DIG10.3 (Tinberg et al., 2013), hereafter DIG0, which binds the plant steroid glycoside digoxin and its aglycone digoxigenin with picomolar affinities. Introduction of three rationally-designed binding site mutations into DIG0 resulted in a progesterone binder (PRO0) with nanomolar affinity (Tinberg et al., 2013). We constructed genetic fusions of DIG0 and PRO0 to a yeast-enhanced GFP (yEGFP, LBD-biosensors DIG0-GFP and PRO0-GFP) and constitutively expressed them in S. cerevisiae (Supplementary file 1). The fusions showed little change in fluorescence in response to digoxin or progesterone, respectively (Figure 1b,c and Figure 1—figure supplement 1). Work by Wandless and co-workers has shown that mutagenesis of LBDs can be used to identify variants that are stable only in the presence of a target ligand (Banaszynski et al., 2006). We randomly mutagenized the LBDs of DIG0-GFP and PRO0-GFP by error-prone PCR and subjected libraries of 105 (Gil et al., 2000) integrants to multiple rounds of FACS, sorting alternately for high fluorescence in the presence of the ligand and low fluorescence in its absence. We isolated LBD variants having greater than 5-fold activation by cognate ligand (Figure 1b,c and Figure 1—figure supplement 1). By making additional variants that contain only one of the up to four mutations found in the progesterone biosensors, we showed that some mutations are additive, while others predominately contribute to sensitivity (Figure 1—figure supplement 2a). Many of the conditionally-destabilizing mutations identified in DIG0 involve residues that participate in key dimer interface interactions (Figure 1d). The conditionally-destabilizing mutations of PRO0 are located throughout the protein (Figure 1—figure supplement 2b–d); the DIG0 interface mutations also rendered PRO0-GFP conditionally stable on binding progesterone (Figure 1—figure supplement 2e). TF-biosensors amplify ligand-dependent responses To improve the dynamic range and utility of the biosensors, we built conditionally-stable LBD-transcription factor fusions (TF-biosensors) by placing an LBD between an N-terminal DNA-binding domain (DBD) and a C-terminal transcriptional activation domain (TAD, Figure 2a). The use of TFs serves to amplify biosensor response and allows for ligand-dependent control of gene expression (Shoulders et al., 2013; Beerli et al., 2000; Louvion et al., 1993). Our initial constructs used the DBD of Gal4, the destabilized LBD mutant DIG1 (E83V), and either VP16 or VP64 as a TAD to drive the expression of yEGFP under the control of a GAL1 promoter. The dynamic range of TF-biosensor activity was maximal when the biosensor was expressed using a weak promoter and weak activation domain, because of lower yEGFP expression in the absence of ligand (Figure 2—figure supplement 1a,b). Figure 2 with 2 supplements see all Download asset Open asset Ligand-dependent transcriptional activation. (a) TF-biosensor construction from a conditionally destabilized LBD, a DNA-binding domain and a transcriptional activation (TAD) domain. (b) Positions of conditionally destabilizing mutations of Gal4 mapped to a computational model of Gal4-DIG0 homodimer. Residues are shown as colored spheres and key interactions are highlighted in insets. The TAD is not shown. (c) Concentration dependence of response to digoxin for digoxin TF-biosensors driving yEGFP expression. (d) Concentration dependence of response to progesterone for progesterone TF-biosensors driving yEGFP expression. (e) Time dependence of response to 250 μM digoxin for digoxin TF-biosensors. (f) Time dependence of response to 50 μM progesterone for progesterone TF-biosensors. (g) Time-dependent response to withdrawal of 250 μM digoxin for digoxin TF-biosensors. (h) Time-dependent response to withdrawal of 50 μM progesterone for progesterone TF-biosensors. In c-f, (-) indicates cells lacking biosensor plasmids and error bars represent s.e.m. of three biological replicates. Marker symbols in e and g are the same as in c. Marker symbols in f and h are the same as in d. https://doi.org/10.7554/eLife.10606.006 We chose Gal4-DIG1-VP16 (hereafter G-DIG1-V) for further TF-biosensor development because it has both a large dynamic range and maximal activation by ligand. A FACS-based screen of an error-prone PCR library of G-DIG0-V, G-DIG1-V, and G-DIG2-V variants identified mutations L77F and R60S in the Gal4 dimer interface (hereafter GL77F, GR60S) that further increased TF-biosensor response by lowering background activity in the absence of ligand (Figure 2b and Figure 2—figure supplement 1c). Although these Gal4 mutations were identified by screening the libraries of digoxin-dependent TF-biosensors, they also increased progesterone-dependent activation of the G-PRO-V series of biosensors, indicating a shared mechanism of conditional stability in both systems (Figure 2—figure supplement 1d). Combining mutations in Gal4 and DIG0 or PRO0 led to activations of up to 60-fold by cognate ligand, a ten-fold improvement over the most responsive LBD-biosensors (Figure 2c,d and Figure 2—figure supplement 2a) and a dynamic range that has been challenging to achieve with stability-based biosensors in yeast (Rakhit et al., 2011). The TF-biosensors were also rapidly activated, showing a five-fold increase in signal after 1 hr of incubation with ligand and full activation after ~14 hr (Figure 2e,f and Figure 2—figure supplement 2b). In contrast to the LBD-biosensors, the TF-biosensors exhibited a broad range of fluorescence levels across single cells, as well as a population of nonfluorescent cells in the presence of ligand (Figure 2—figure supplement 2). We used FACS to isolate cells from the nonfluorescent population and found those cells to be inviable, possibly indicating plasmid loss or toxicity from biosensor activation. Upon withdrawal of ligand, strains expressing TF-biosensors rapidly exhibited reduction in signal, reaching half of their maximum yEGFP fluorescence after approximately 5 hr and nearly undetectable fluorescence after 10–15 hr (Figure 2g,h). The response of the TF-biosensors to the withdrawal of ligand is likely much faster than observed by fluorescence, as the reduction in fluorescence signal is dependent on both the degradation of the TF-biosensors as well as the degradation and dilution of previously expressed yEGFP. TF-biosensors are tunable and modular An attractive feature of the TF-biosensors is that the constituent parts – the DBD/promoter pair, the LBD, the TAD, the reporter, and the yeast strain – are modular, such that the system can be modified for additional applications. To demonstrate tunability, we replaced the DBD of G-DIG1-V with the bacterial repressor LexA and replaced the Gal4 DNA-binding sites in the GAL1 promoter with those for LexA. LexA-based TF-biosensors with DIG1 and a weak TAD (B42) showed a strong response to digoxin (nearly 40-fold) only when the promoter-driving reporter expression contained LexA-binding sites (Figure 3a). These results demonstrate that the biosensors can function with different combinations of DBDs and TADs, which could produce diverse behaviors and permit their use in eukaryotes requiring different promoters. Furthermore, the reporter gene can be swapped with an auxotrophic marker gene to enable growth selections. The TF-biosensors drove the expression of the HIS3 reporter most effectively when steroid was added to the growth media, as assessed by the growth of a histidine auxotrophic strain in media lacking histidine (Figure 3b,d). Fusion of the Matα2 degron to the biosensor improved dynamic range by reducing the growth of yeast in the absence of ligand. Finally, the yeast strain could be modified to improve biosensor sensitivity toward target ligands by the deletion of the gene for a multidrug efflux pump (Ernst et al., 2005), thereby increasing ligand retention (Figure 3c–d). Figure 3 Download asset Open asset Tuning TF-biosensors for different contexts. (a) The TAD and DBD of the TF-biosensor and the corresponding binding site for the DBD in the reporter promoter can be swapped for a different application. Expression of a plasmid-borne luciferase reporter was driven by TF-biosensors containing either a LexA or Gal4 DBD and either a VP16 or B42 TAD. Promoters for the reporter contained DNA-binding sites for either Gal4 or LexA. (b) TF-biosensors were transformed into the yeast strain PJ69-4a and tested for growth on this minimal media containing 1 mM 3-aminotriazole (3-AT) and the indicated steroid. To determine the effect of including an additional destabilization domain, the degron from Matα2 was cloned into one of four positions. (c) G-DIG1-V biosensor response to digoxigenin in yEGFP reporter strain PyE1 either with or without a deletion to the ORF of PDR5. (d) Ligand and TF-biosensor-dependent growth on this media in yeast strains containing deleted ORFs for efflux-related transcription factors (PDR1 and PDR3) or ABC transporter proteins (YOR1, PDR5, SNQ2). In a and c, error bars represent s.e.m. of three biological replicates. https://doi.org/10.7554/eLife.10606.009 TF-biosensors enable a selection in yeast to improve the bioproduction of a small molecule Improving bioproduction requires the ability to detect how modifications to the regulation and composition of production pathways affect product titers. Current product detection methods such as mass spectrometry or colorimetric assays are low-throughput and are not scalable or generalizable. LBD- and TF-biosensors could be coupled with fluorescent reporters to enable high throughput library screening or to selectable genes to permit rapid evolution of biosynthetic pathways (Tang and Cirino, 2011; Dietrich et al., 2010; Chou and Keasling, 2013). Yeast-based platforms have been developed for the biosynthesis of pharmaceutically relevant steroids, such as progesterone and hydrocortisone (Duport et al., 1998; Szczebara et al., 2003). A key step in the production of both steroids is the conversion of pregnenolone to progesterone by the enzyme 3β-hydroxysteroid dehydrogenase (3β-HSD). We aimed to use a progesterone biosensor to detect and improve this transformation. An important feature of biosensors intended for pathway engineering is their ability to detect a product with minimal activation by substrate or other related chemicals. TF-biosensors built from PRO1 showed the greatest dynamic range and selectivity for progesterone over pregnenolone when driving yEGFP expression or when coupled with a HIS3 reporter assay (Figure 4a,b and Figure 4—figure supplement 1a). We investigated whether this sensor could be used to detect the in vivo conversion of pregnenolone to progesterone by episomally-expressed 3β-HSD (Figure 4c). Using GL77F-PRO1-V driving a yEGFP reporter, we could detect progesterone production, with biosensor response greatest when 3β-HSD was expressed from a high copy number plasmid and from a strong promoter (Figure 4d). Figure 4 with 1 supplement see all Download asset Open asset Application of biosensors to metabolic engineering in yeast. (a) Fold activation of GL77F-PRO1-V by a panel of steroids in yEGFP reporter strain PyE1. Data are represented as mean ± SEM. (b) Growth of degron-G-PRO1-V in HIS3 reporter strain PJ69-4a is stimulated by progesterone but not pregnenolone. (c) Schematic for directed evolution of 3β-HSD using TF-biosensors for the conversion of pregnenolone to progesterone. (d) Fold activation of GL77F-PRO1-V by a panel of plasmids expressing wild-type 3β-HSD under varying promoter strengths in yEGFP reporter strain PyE1 when incubated in 50 μM pregnenolone. Data for plasmids containing CEN/ARS and 2 μ (2 micron) origins are shown. (e) Fold activation of GL77F-PRO1-V by a panel of evolved 3β-HSD mutants expressed under the TDH3 promoter on a CEN/ARS plasmid and incubated in 50 μM pregnenolone. (f) Progesterone titer in 1 OD of cells produced by strains expressing 3β-HSD mutants. Progesterone became toxic at levels of 100 μM and above, leading to substantial cell death. β-estradiol and hydrocortisone were not soluble in yeast growth media at levels above 25 μM. In a and d-f, data are presented as mean ± s.e.m. of three biological replicates. In d and e, (-) indicates cells lacking 3β-HSD. *indicates significance with a threshold of p < 0.05 using 2-tailed Student's t-test. https://doi.org/10.7554/eLife.10606.010 We then sought to use the biosensor to improve this enzymatic transformation. To select for improved progesterone production, we required a growth assay in which wild-type 3β-HSD could no longer complement histidine auxotrophy when the yeast were grown on plates supplemented with pregnenolone. To this end, the selection stringency was tuned by adding the His3 inhibitor 3-aminotriazole (Figure 4—figure supplement 1e). We mutagenized the 3β-HSD coding sequence using error-prone PCR and screened colonies that survived the HIS3 selection for their yEGFP activation by pregnenolone. By transforming evolved 3β-HSD mutations into a fresh host background, we showed that the mutations in the enzyme, and not off-target plasmid or host escape mutations, were responsible for increased biosensor response (Figure 4e). Two of the mutants, 3β-HSD N139D and 3β-HSD F67Y, were assayed for progesterone production using gas chromatography and mass spectrometry and were found to produce two-fold more progesterone per OD than cells bearing the wild-type enzyme (Figure 4f). Yeast-derived biosensors port directly to mammalian cells and can be used to tightly regulate CRISPR/Cas9 genome editing Yeast is an attractive platform for engineering in vivo biosensors because of its rapid doubling time and tractable genetics. If yeast-derived biosensors function in more complex eukaryotes, the design-build-test cycle in those organisms could be rapidly accelerated. We first assessed the portability of yeast TF-biosensors to mammalian cells. Single constructs containing digoxin and progesterone TF-biosensors with the greatest dynamic ranges (without codon optimization) were stably integrated into human K562 cells using PiggyBac transposition. We characterized the dynamics of the TF-biosensors in human cells by dose response and time course assays similar to the yeast experiments (Figure 5a–d). As with yeast, the human cells demonstrated greater sensitivity to digoxin, with fluorescence activation increasing up to 100 nM of cognate ligand for digoxin biosensors and 1 mM for progesterone biosensors. We observed >100-fold activation for the most sensitive progesterone biosensor GL77F-PRO1-V. The increase in mammalian dynamic range over yeast may arise from more aggressive degradation of destabilized biosensors or greater accumulation of target-stabilized biosensors or reporters resulting from larger cell sizes and slower doubling times. The time course data show that fluorescence increased four-fold within 4 hr of target introduction and rose logarithmically for 24–48 hr. We next assessed whether these biosensors could drive more complex mammalian phenotypes. The CRISPR/Cas9 system has proved to be an invaluable tool for genome editing (Mali et al., 2013; DiCarlo et al., 2013; Gratz et al., 2013; Hwang, 2013). Despite the high programmability and specificity of Cas9-mediated gene editing achieved to date, unchecked Cas9 activity can lead to off-target mutations and cytotoxicity (Fu et al., 2013; Mali et al., 2013; Pattanayak et al., 2013). Further, it may be desirable to tightly regulate Cas9 activity such that gene editing occurs only under defined conditions. To facilitate inducible gene editing, we fused human codon-optimized versions of the DIG3 and PRO1 LBDs to the N-terminus of Cas9 from S. pyogenes. We integrated this construct into a reporter cell line containing an EGFP variant with a premature stop codon that renders it non-functional. Upon separate stable integration of the DIG-Cas9 and PRO-Cas9 fusions, we transfected a guide RNA targeting the premature stop codon as well as a donor oligonucleotide containing the sequence to restore EGFP activity via homologous recombination. After a 48-hr incubation period, we observed an ~18-fold increase in GFP positive cells with digoxigenin relative to the mock control (Figure 5e). Figure 5 Download asset Open asset Activation of biosensors in mammalian cells and regulation of CRISPR/Cas9 activity. (a) Concentration dependence of response to digoxin for constructs containing digoxin TF-biosensors and Gal4 UAS-E1b-EGFP reporter individually integrated into K562 cells. GR60S,L77F-PRO1-V serves as a digoxin insensitive control. (b) Concentration dependence of response to progesterone for constructs containing progesterone TF-biosensors and Gal4 UAS-E1b-EGFP reporter individually integrated into K562 cells. GR60S-DIG1-V serves as a progesterone insensitive control. (c) Time dependence of response to 100 nM digoxin for constructs containing digoxin TF-biosensors and Gal4 UAS-E1b-EGFP reporter individually integrated into K562 cells. GR60S,L77F-PRO1-V serves as a digoxin insensitive control. (d) Time dependence of response to 25 μM progesterone for constructs containing progesterone TF-biosensors and Gal4 UAS-E1b-EGFP reporter individually integrated into K562 cells. GR60S-DIG1-V serves as a progesterone insensitive control. (e) DIG3 and PRO1 fused to the N-terminus of S. pyogenes Cas9 were integrated into a K562 cell line containing a broken EGFP. EGFP function is restored upon transfection of a guide RNA and donor oligonucleotide with matching sequence in the presence of active Cas9. The data are presented as mean fluorescence ± s.e.m. of three biological replicates. https://doi.org/10.7554/eLife.10606.012 Environmental detection in the plant Arabidopsis thaliana To assess generalizability of these sensors to multicellular organisms, we engineered G-DIG1-V to function as an environmental biosensor in plants. The DIG1 sequence was codon optimized for expression in Arabidopsis thaliana. We tested biosensor fusions to two different degrons, Matα2 from yeast and DREB2a from Arabidopsis (Sakuma et al., 2006), and we used the VP16 and VP64 variants as the TAD. We initially tested the G-DIG1-TAD variants with a transient expression assay using Arabidopsis protoplasts and a reporter gene consisting of firefly luciferase under the control of a Gal4-activated plant promoter (pUAS::Luc). The biosensor containing the Matα2 degron and VP16 TAD showed the highest fold activation of luciferase in the presence of digoxigenin (Figure 6—figure supplement 1a). We next inserted the genes encoding G-DIG1-V-Matα2 and the Gal4-activated pUAS::Luc into a plant transformation vector and stably transformed them into Arabidopsis plants. Primary transgenic plants were screened in vivo for digoxigenin-dependent luciferase production, and responsive plants were allowed to set seed for further testing. Second generation transgenic plants (T1, heterozygous) were tested for digoxin- or digoxigenin-dependent induction of luciferase expression. After 42 hr, we observed 30-50-fold induction of luciferase activity in digoxin-treated plants compared to the uninduced control (Figure 6). Both digoxin and digoxigenin were capable of inducing the biosensor. Digoxigenin-dependent luciferase induction was observed in multiple independent transgenic T1 lines (Figure 6—figure supplement 1b), and a rising dose response to digoxigenin was observed in the transgenic plants (Figure 6—figure supplement 1c). The specificity of the digoxigenin biosensor in plants parallels that in yeast cells (Figure 6—figure supplement 1d). Figure 6 with 1 supplement see all Download asset Open asset Application of biosensors in plants. (a) Activation of luciferase expression in transgenic Arabidopsis plants containing the G-DIG1-V biosensor in the absence (left) or presence (right) of 100 μM digoxin. Luciferase expression levels are false colored according to scale to the right. Relative luciferase units corresponding to 1 min of image pixel integration (to avoid saturating pixels) are shown above each individual plant. (b) Brightfield image of plants shown in a. https://doi.org/10.7554/eLife.10606.013 Discussion In vivo biosensors for small molecules enable the regulation and detection of cellular responses to endogenous metabolites and exogenous chemicals. Here, we show that LBDs can be conditionally destabilized to create biosensors that function in yeast, mammalian cells, and plants, and we demonstrate the use of these biosensors in metabolic engineering and genome editing applications. While this method requires a high-affinity ligand-binding domain as a starting point, nearly all small molecules of interest have a natural protein interactor. Furthermore, the use of de novo-designed binders opens the possibility of generating biosensors for ligands with unsuitable or unknown binding proteins. By incorporating standard mutagenesis and screening, our method constitutes a simple platform for sensor development that can be applied to many areas of biotechnology. These sensors act either at the level of post-translational control over protein function or at the level of transcription (Figure 7a), and they can be tuned by altering any of their components (Figure 7b) or by modifying efflux of the target ligand in the host organism. These tunable features should make the biosensors useful in many different cellular and environmental contexts. Figure 7 Download asset Open asset Schematic of biosensor platform. (a) Biosensors for small molecules are modularly constructed by replacing the LBD with proteins possessing altered substrate preferences. (b) Activity of the biosensor can be tuned by 1) introducing destabilizing mutations (red Xs), 2) adding a degron, 3) altering the strength of the TAD or DNA binding affinity of the TF, 4) changes in the number of TF-binding sites or sequences, and 5) titrating 3-aminotriazole, an inhibitor of His3. (c) Yeast provide a genetically tractable chassis for biosensor development befor