Demonstration of de novo chemotaxis in E. coli using a real-time, quantitative, and digital-like approach

Chemotaxis is the movement of an organism in response to an external chemical stimulus. This system enables bacteria to sense their immediate environment and adapt to changes in its chemical composition. Bacterial chemotaxis is mediated by chemoreceptors, membrane proteins that bind an effector and transduce the signal to the downstream proteins. From a synthetic biology perspective, the natural chemotactic repertoire is of little use since bacterial chemoreceptors have evolved to sense specific ligands that either benefit or harm the cell. Here we demonstrate that using a combined computational design approach together with a quantitative, real-time, and digital detection approach, we can rapidly design, manufacture, and characterize a synthetic chemoreceptor in E. coli for histamine (a ligand for which there are no known chemoreceptors). First, we employed a computational protocol that uses the Rosetta bioinformatics software together with high threshold filters to design mutational variants to the native Tar ligand binding domain that target histamine. Second, we tested different ligand-chemoreceptors pairs with a novel chemotaxis assay, based on optical reflectance interferometry of porous silicon (PSi) optical transducers, enabling label-free quantification of chemotaxis by monitoring real-time changes in the optical readout (expressed as the effective optical thickness, EOT). We found that different ligands can be characterized by an individual set of fingerprints in our assay. Namely, a binary, digital-like response in EOT change (i.e. positive or negative) that differentiates between attractants and repellants, the amplitude of change of EOT response, and the rate by which steady state in EOT change is reached. Using this assay, we were able to positively identify and characterize a single mutational chemoreceptor variant for histamine that mediated chemotaxis comparably to the natural Tar-aspartate system. Our results demonstrate the possibility of not only expanding the natural chemotaxis repertoire, but also provide a new quantitative assay by which to characterize the efficacy of the chemotactic response.


44
Bacterial chemotaxis is the movement of bacteria in response to an external chemical stimulus. 45 This system enables bacteria to sense their immediate environment and quickly adapt to changes in its 46 chemical composition, moving away from repellents or towards attractants, thus assuring their survival. 47 Furthermore, it is characterized by specific responses, high sensitivity, and a dynamic range (1, 2). 48 Chemotaxis is facilitated by methyl-accepting chemotaxis protein (MCP or the chemoreceptor), which is 49 typically a transmembrane protein that binds the effector through the periplasmic ligand binding site and

62
Bacterial chemotaxis is characterized by simplicity of design, plasticity, and broad range of 63 chemical detection repertoire that is highly conserved across the prokaryotic kingdom (5, 6). In particular, 64 E. coli's chemotaxis system consists of only five chemoreceptors: Tsr (taxis to serine and repellents), Tar Triticum vulgaris (termed as wheat germ agglutinin or WGA), and acetonitrile were purchased from 98 Sigma-Aldrich, Israel. Absolute ethanol, potassium and phosphate salts, dimethyl sulfoxide (DMSO), and 99 diisopropylethylamine (DIEA) were supplied by Merck, Germany. Acetic acid, NaCl, and MgSO 4 were 100 supplied by Bio-Lab Ltd, Israel. Bacto agar and tryptone were supplied by BD Biosciences, USA. Low Technion -Israel Institute of Technology) and mechanically diced into 10 mm by 5 mm chips by an 112 automated dicing saw (Disco, Japan). Lectin from tritium vulgaris, commonly referred to as wheat germ 113 agglutinin (WGA), was immobilized on the Si diffraction gratings as previously described.(12) Briefly, 114 the chips were thermally oxidized in ambient air at 800 °C for 1 hour in a Lindberg/Blue Split-Hinge 115 furnace (Thermo Scientific, USA), followed by silanization in 2% DETAS (50% ethanol in ddH 2 O, 116 acidified with 0.6% acetic acid) for 1 h. Chips were washed with ethanol and dried under a stream of 117 nitrogen. Carboxylation was performed by incubation in a solution comprised of 1% succinic anhydride 118 (in acetonitrile with 4% DIEA) for 3 h followed by washing with ethanol and drying with nitrogen.

119
Amine activation was promoted by incubation in 129 mM DIC and 87 mM NHS constituted in 120 acetonitrile for 1 h. After rinsing with ethanol, chips were stored at 4 °C overnight in WGA solution (1 121 mg/mL, 10% DMSO) to promote lectin immobilization onto the Si diffraction grating.

Bacterial strains and plasmids
Derivative of E. coli RP437 strain UU1250 (Δtsr-7028 Δ(tar-tap)5201 Δtrg-100 Δaer-1, kan r ) (13) was 124 used as a parental strain to construct all the chemotactic strains in this work, and were obtained from Dr.

144
To construct the Histamine-Tar variant, a protocol presented by Moretti et al (14) was followed. The 145 output of the protocol is a library of variants, ranging from dozens to thousands of protein PDB files, 146 depending on the parameters of the design. Each variant is a mutated version of the input protein such that 147 the mutations increase affinity to a specific ligand of our choice. By using this protocol to design a 148 histamine sensitive Tar, a library of 870 mutated Tar receptors was obtained. Each design consisted of 3-5 iterations of the protocol for optimal results. Next, a filtering process was performed using the parameters 150 presented in table S1 Table. 151 The filtering process resulted in 11 variants, which were predicted as energetically favorable to bind 152 histamine. S1 Fig presents a sequence comparison between the designed strains and the wildtype Tar.

153
These were constructed through mutating two places in the native Tar chemoreceptor using two sets of 154 forward and reverse primers, as shown in S2 Tar-EGFP chimera was constructed using a flexible linker 5'-ggtagcggcagcggtagc-3', sequence which 158 was obtained from iGEM 2016 parts catalog (Bba J18921) to an EGFP sequence from the iGEM catalog 159 (Bba E0040). PCR reaction to the chemoreceptors plasmid and the EGFP sequence part with an overlap 160 region of the linker was performed, using PCR primers for EGFP sequence and chemoreceptor plasmid, 161 as shown in S2 Table. 162 Soft agar swarming assay 163 Swarming assay was carried out based on the Alders assay (15) (i.e. using 0.5% (w/v) soft agar plates). 5-164 cm Petri dishes were filled with 5 mL soft agar and 5 mL of BA (bio-assay buffer) or TB based media, for 165 PctA-Tar chimera and Histamine-Tar varianst, respectively. The plates were cooled down to room 166 temperature in order to solidify for at least 1 h. The tested strains were grown overnight at 30°C at the 167 appropriate media. Using a sterile tip, a drop of bacterial suspension was placed in the middle of the 168 solidified agar containing the chemo-attractant.

169
Microscope repellent chemotaxis drop assay 170 An overnight culture grown at 30 °C was diluted 1:50 and grown to an optical density (OD) value of ~ 1.

171
The correlation between bacteria concentration and OD 600 measurement was determined empirically (1 172 OD 600 = 8 x 10 8 cells/mL). Cells were re-suspended in motility buffer and placed on a microscope slide (8 173 µL)(. A 2 µL drop of repellent was added to the bacteria a ratio of 1:5 (v/v), respectively, after validating 174 that the final repellent concentration is not lethal for the bacteria. The bacterial response was then recorded for 15 min using Nikon Eclipse Ti microscope (magnification 100x), taking a frame every 50 176 ms. 178 An overnight culture grown at 30 °C was diluted 1:50 and grown to an OD 600 value of ~1. A 40 µL 179 suspension of bacteria in motility buffer was placed into a commercial sticky-Slide I Luer (Ibidi,180 Germany), which is specifically designed for perfusion applications. For microscopy studies, the sticky- washed with motility buffer solution to remove cells that were loosely adhered to the chip surface. Lastly, 202 the chemo-effector was delivered in motility buffer for approximately 30 min. Zero-order diffraction was 203 measured as a function of reflectance intensity versus wavelength. The frequency of the reflected light 204 can be described using the following equation (16, 17):

Microscope chemotaxis chip assay
Where $ is the phase delay between the source beam and the reflected one, is wavelength, is the 207 depth of the Si pores, and $ is the refractive index of the medium filling the pores. Through frequency 208 analysis using application of a fast Fourier transform (11)

215
where EOT 0 is the EOT measured at t = 0, the time when the chemo-effector was introduced.

218
Characterization of the native Tar functionality 219 In order to build a generic chemotactic capacity expansion toolkit, we first needed to test our rapid 220 or digital detection approach on the natural Tar chemoreceptor. To do so, we reintroduced the tar gene to 221 E. coli strain UU1250 that lacks all native chemoreceptors. This strain was first subjected to several 222 chemotaxis assays, using aspartate to test the attractant response and 1( to test the repellent response 223 (3).

224
Attractant response of the Tar receptor was initially tested using the chip microscope assay (see  Table 1). Figure  Results for a typical chemotaxis assay are presented in Figure 2C in terms of the recorded EOT value, which is ascribed to higher refractive index of the solution (e.g., when chemoattractant was added).

285
Notably, while the preparation of the chip to chemotaxis experiments is time consuming (~ 7 hours), a 286 positive chemotactic response is recorded within approximately 10 minutes.

287
To observe differences in chemotactic responses, three different bacterial strains were tested 288 (Table 1), and ΔEOT measurements are plotted in Figure 2D. First, the response of the bacterial strain 289 expressing the Tar chemoreceptor was studied upon the introduction of its natural attractant, aspartate (20 290 mM in motility buffer). In this case, a rapid decrease in the EOT (of ~4 nm) is observed within less than 5 291 min, after which a continuous moderate reduction is recorded (Fig. 2D, yellow trace). This behavior is 292 ascribed to the decreasing population of cells within the chip, as the bacteria respond to the chemo-293 attractant gradient in the solution and actively move towards it (6). An opposite trend is obtained when 294 the ΔZras strain, which expresses all four natural chemoreceptors, is exposed to the chemo-repellent 295 solution of benzoate (50 mM in motility buffer). A rapid increase in the EOT is observed within 5 min, 296 followed by a stabilization of the EOT signal around a constant value within the next hour (Fig. 2D,   297 orange trace). The response of the UU1250 strain, which lacks all chemoreceptors, to benzoate was also 298 tested as control. Because this strain lacks the ability to move towards the chemo-effector, most bacteria 299 remained inside the pores, resulting in small amplitude fluctuations of ΔEOT around the baseline EOT 300 value (Fig. 2D,  To confirm the functionality of the chimeric receptor, a swarming plate assay was carried out. In To further characterize the PctA-Tar synthetic chemoreceptor, we tested its response to the L-341 amino acid Alanine attractant using our real time chemotaxis assay (Fig. 3C). The graph presents the immune response(24), which in some cases can be life threatening. To construct a histamine-sensitive 363 receptor, we used the Rosetta software suite, a collection of algorithms and programs for macromolecular 364 modeling and design. For the generation of new ligand-binding domains we followed the protocol 365 described by Moretti et al (14). Briefly, the program is given two inputs, A PDB file containing the we ran 3-5 iterations of the protocol, to increase the chance of a successful hit. Subsequently, the virtual 371 library was subjected to a filtering process designed to eliminate variants which Rosetta predicts will not 372 bind the ligand correctly, this is done according to the parameters which are specific per design (see Table   373 S1). Using this algorithm, we managed to generate a library of mutated Tar ligand binding domains that 374 were computational predicted to bind histamine and activate the chemotaxis pathway in response to it.

375
Out of the 870 suggested mutations, only 11 variants passed the filtering process, meaning, they were 376 predicted to bind histamine. As expected, all mutations are located near the natural binding pocket for 377 aspartate as can be seen in Fig. 4A  The bacterial chemotaxis system has been studied for several decades due to its wide range of 416 applications. In recent years, with the advance of technology, knowledge of the inner working of this system has led to attempts to harness it for targeted usage via different approaches, such as: rational 418 design of chimeras and directed evolution (10 ,8    The energies of the two sides of the interface, possibly normalized by the number of residues S1 Table. Filtering parameters-Filtering parameters used according to Rosetta protocol.