Design and optimization of a cell-free atrazine biosensor

Recent advances in cell-free synthetic biology have spurred the development of in vitro molecular diagnostics that serve as effective alternatives to whole-cell biosensors. However, cell-free sensors for detecting manmade organic water contaminants such as pesticides are sparse, partially because few characterized natural biological sensors can directly detect such pollutants. Here, we present a platform for the cell-free detection of one critical water contaminant, atrazine, by combining a previously characterized cyanuric acid biosensor with a reconstituted atrazine-to-cyanuric acid metabolic pathway composed of several protein-enriched bacterial extracts mixed in a one pot reaction. Our cell-free sensor detects atrazine within an hour of incubation at an activation ratio superior to previously reported whole-cell atrazine sensors. We also show that the response characteristics of the atrazine sensor can be tuned by manipulating the component ratios of the cell-free reaction mixture. Our approach of utilizing multiple metabolic steps, encoded in protein-enriched cell-free extracts, to convert a target of interest into a molecule that can be sensed by a transcription factor is modularly designed, which should enable this work to serve as an effective proof-of-concept for rapid field-deployable detection of complex organic water contaminants.


35
Cell-free gene expression (CFE) has recently emerged as a powerful strategy for rapid, 36 field-deployable diagnostics for nucleic acids 1-5 and chemical contaminants. 6-9 One reason for 37 this success is that CFE reactions minimize many of the constraints of whole-cell sensors, 38 including mass transfer barriers, cytotoxicity, genetic instability, plasmid loss, and the need for 39 biocontainment. 8,10 In addition, CFE reactions can be stabilized through freeze-drying and then 40 are activated upon rehydration, enabling the biosensors to be used outside the laboratory at the 41 point of sampling in the field. 1 However, previously reported cell-free biosensors have so far 42 been limited to detecting either nucleic acids 2,3 or chemical contaminants that can be directly 43 sensed with well-characterized allosteric protein transcription factors or riboswitches. 8,9,[11][12][13] 44 Here, we expand the ability of cell-free biosensors to detect complex organic molecules 45 by developing a combined metabolism and biosensing strategy. Our strategy is motivated by the 46 observation that the space of known natural transcription factors may be insufficient to directly 47 detect organic molecules of analytical interest, especially those that are man-made and 48 relatively new to natural environments. On the other hand, a wealth of metabolic biochemistry 49 often exists that could convert a target molecule of interest into a compound that can be directly 50 sensed by a transcription factor. Thus, a range of new CFE-based diagnostics could be 51 developed by combining in vitro metabolic conversion with natural transcriptional biosensors.

52
Recently, such a strategy was validated in CFE reactions using a simple one-enzyme pathway 53 where the enzyme, transcription factor, and reporter are encoded on separate plasmids. In that 54 work, cocaine and hippuric acid were catabolized in vitro to make benzoic acid, which is sensed 55 by the allosteric transcription factor BenR. 7,14,15 However, the metabolic pathways tested were 56 short -containing only a single enzyme -and converged to a simple and abundant analyte 57 detected by a native E. coli transcription factor. A more general approach would be necessary 58 for detecting xenobiotic, or new-to-nature, analytes.

59
Specifically, in this study, we develop a strategy for multi-enzymatic metabolic 60 biosensing of atrazine-one of the most commonly detected herbicides in American surface 61 water, and a suspected endocrine-disrupting compound. 16

70
Here, we combine that CYA cell-free sensor with a reconstituted cell-free metabolic pathway 71 that converts atrazine to CYA through three steps in a single pot reaction. Due to the burden 72 imposed by synthesizing several proteins in situ in a single batch CFE reaction, we developed 73 an extract mixing strategy, where individual extracts enriched with a single enzyme or 74 transcription factor are combined to reconstitute the complete biosensing reaction. This modular 75 approach allows the system to be optimized by simply searching over the ratios of each distinct 76 enriched extract. Using this approach, we developed a sensor capable of discriminating high 77 concentrations of atrazine (10-100 µM) within an hour of incubation. We anticipate that our 78 combined metabolism and biosensing strategy for detecting atrazine will be broadly applicable 79 for the rapid cell-free detection of pesticides and other water contaminants, as well as

98
To design our cell-free atrazine sensor, we took inspiration from Pseudomonas sp. strain 99 ADP-1, which metabolizes atrazine into cyanuric acid through a three-enzyme pathway encoded 100 on the atzABC operon ( Figure 1A). 21,22 We hypothesized that by co-expressing each of these

126
When the biosensor was challenged with each of these compounds, we observed weak 127 activation only in response to propazine, which has a more similar chemical structure to atrazine 128 than melamine ( Figure 1C). This result is consistent with previous observations that ADP-1's 129 atrazine chlorohydrolase, AtzA, which actually evolved from the melamine deaminase TriA in

164
We then aimed to identify the ratio of AtzA-:AtzB-:AtzC-:AtzR-enriched:unenriched 165 extracts that gave the highest fold induction (defined as fluorescence in the presence of 100 µM 166 atrazine / fluorescence in the absence of atrazine) for the sensor. We first determined the 167 optimal fraction of the transcription factor AtzR in the sensing reaction by only detecting the 168 downstream analyte, cyanuric acid. The greatest fold induction was observed at a 5% AtzR-169 enriched extract and 95% unenriched extract, a ratio that minimizes leak and maximizes ON 170 state, likely because that mixture also has the greatest amount of the unenriched extract ( Figure 2A). Next, we iteratively optimized the ratios of AtzA-, AtzB-, and AtzC-enriched 172 extracts in the reaction, starting from an assumption of 10% dosage for each sensor ( Figure   173   2B-D). Surprisingly, we observed low sensitivity of the atrazine response to perturbations in the 174 concentrations of these enzyme-enriched extracts, at least in the range of 1-10% of the total 175 extract composition. As before, though, no response to atrazine could be observed if any of the 176 enzyme-enriched extracts was individually left out. At each condition, we chose the extract ratio 177 that gave the highest fold induction. Using this iterative, coarse-grained optimization, we 178 obtained an optimal response at 5% AtzR, 10% AtzA, 2% AtzB, and 20% AtzC-enriched 179 extracts, with the balance (63%) made up by the unenriched extract.

180
Having established an optimal ratio of each extract in our sensor, we measured its dose 181 response to atrazine by the observed sfGFP fluorescence after 4 hours of reaction (Figure 2E).

182
The calculated limit of detection, defined as the concentration of atrazine that yielded a

265
The exception to this was the AtzB-enriched extract; since we could not successfully transform 266 the purified plasmid into the knockout strain, this extract was instead prepared using the 267 Rosetta2 (DE3) pLysS strain. For each transformed strain, 1 L culture of 2X YT + P (16 g 268 tryptone, 10 g yeast extract, 5 g NaCl, 7 g potassium phosphate dibasic, 3 g potassium 269 phosphate monobasic) was inoculated from a saturated overnight culture of the chassis strain in