ARTICLE

https://doi.org/10.1038/s41467-021-25538-y OPEN

Programmable receptors enable bacterial

biosensors to detect pathological biomarkers in
clinical samples
Hung-Ju Chang 1, Ana Zúñiga 1, Ismael Conejero1,2,3, Peter L. Voyvodic1, Jerome Gracy1,
Elena Fajardo-Ruiz 1, Martin Cohen-Gonsaud1, Guillaume Cambray 1, Georges-Philippe Pageaux4,
Magdalena Meszaros 4, Lucy Meunier4 & Jerome Bonnet 1 ✉
1234567890():,;

Bacterial biosensors, or bactosensors, are promising agents for medical and environmental
diagnostics. However, the lack of scalable frameworks to systematically program ligand
detection limits their applications. Here we show how novel, clinically relevant sensing
modalities can be introduced into bactosensors in a modular fashion. To do so, we have
leveraged a synthetic receptor platform, termed EMeRALD (Engineered Modularized
Receptors Activated via Ligand-induced Dimerization) which supports the modular assembly
of sensing modules onto a high-performance, generic signaling scaffold controlling gene
expression in E. coli. We apply EMeRALD to detect bile salts, a biomarker of liver dysfunction,
by repurposing sensing modules from enteropathogenic Vibrio species. We improve the
sensitivity and lower the limit-of-detection of the sensing module by directed evolution. We
then engineer a colorimetric bactosensor detecting pathological bile salt levels in serum from
patients having undergone liver transplant, providing an output detectable by the naked-eye.
The EMeRALD technology enables functional exploration of natural sensing modules and
rapid engineering of synthetic receptors for diagnostics, environmental monitoring, and
control of therapeutic microbes.

1 Centre de Biologie Structurale (CBS), INSERM U1054, CNRS UMR5048, University of Montpellier, Montpellier, France. 2 Neuropsychiatry: Epidemiological

and Clinical Research, Inserm Unit 1061, Montpellier, France. 3 Department of Psychiatry, CHU Nimes, University of Montpellier, Montpellier, France.
4 Department of Hepatogastroenterology, Hepatology and Liver Transplantation Unit, Saint Eloi Hospital, University of Montpellier, Montpellier, France.
✉email: jerome.bonnet@inserm.fr

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

E
arly disease detection and monitoring of chronic patholo- However, current methods for in vitro diagnostics are only
gies help reduce mortality and improve patients’ quality of available in hospitals and testing laboratories, limiting the fre-
life1,2. In that context, in vitro diagnostic technologies play quency of monitoring for patients. In addition, most biomarkers
key roles at different stages of the healthcare chain3. However, appear at late disease stages (when significant cellular damage has
many diagnostic technologies require heavy equipments, are already occurred) and lack specificity26. As liver diseases are
expensive, and necessitate trained personnel, limiting testing to chronic, evolutive pathologies, patients would benefit from
centralized facilities such as hospitals. Yet, as exemplified by monitoring devices that enable rapid and simple assessment of
glucose monitoring for diabetes, field-deployable diagnostic liver function with high sensitivity and specificity.
devices can tremendously improve patient healthcare, follow-up, An alternative biomarker of liver dysfunction is the presence of
and self-reliance4,5. Robust, scalable, and cost-effective biosensing bile salts in serum. Bile salts are key components of bile which are
technologies for field-deployable diagnostics have thus been critical for digestion in which they help absorption of fat27.
under intense research interest over the past decade6. Interestingly, serum bile salts have emerged as a general bio-
Bacteria must sense and respond to myriad chemical and marker of liver disease, and are the gold-standard diagnostic
physical signals to survive and reproduce, and are thus ideal method for pregnancy cholangitis28,29. Several studies have also
candidates for engineering biosensors. Whole-cell biosensors pointed to bile salts as a general biomarker of interest for early
(WCB) are genetically modified living cells that detect molecules cirrhosis, hepatitis, drug-induced liver injury. In addition, bile
of interest, generally using a transcription factor regulated by the salts provide a highly specific and dynamic assessment of liver
ligand of interest, and activating transcription of a reporter gene7. function. For example, serum bile salts are highly correlated with
While explored since the dawn of genetic engineering8, recent the obstruction state of the bile duct and rapidly decrease when
advances in synthetic biology have improved whole-cell biosensor biliary stenting is performed30. Furthermore, specific bile salt
robustness, sensitivity, signal-to-noise ratio, and signal-processing profiles may be associated with particular liver diseases31–33. Bile
capabilities, supporting their use in complex media like waste- salts thus represent an ideal and specific biomarker for diagnosis
water and clinical samples9–11. With these new capabilities, and dynamic monitoring of liver disease. Yet, as for other bio-
whole-cell biosensors have the potential to provide miniaturized, markers, current detection methods for bile salts based on
field-deployable, diagnostic devices capable of multiplexed enzymatic assays34 are only performed in a centralized fashion
detection and sophisticated computation10,11. As a self- and cannot discriminate between different bile salts classes.
manufacturing and biodegradable platform, biology provides a In this work we take advantage of the natural capacity of
cost-effective and environmentally friendly alternative to tradi- enteropathogenic bacteria to detect bile salts upon arrival into the
tional diagnostic methods. The self-manufacturing nature of gut to activate their virulence pathways35,36. We build EMeRALD
biology also offers a unique advantage for low-resource settings receptor detecting bile salts in E. coli by rewiring bile salt-sensing
and remote, highly constrained conditions, such as those found in modules from Vibrio cholerae and Vibrio parahaemolyticus. As
space exploration12. Despite all these advantages, the scope of the synthetic receptor operates in a surrogate, non-pathogenic
application for bacterial biosensors is limited by the difficulty to host, we perform directed evolution of the sensing module and
generate novel sensors detecting biomarkers of interest. Although improve its limit-of-detection (LOD) and sensitivity. Finally, we
significant progress has been recently made13, scalable platforms optimize a colorimetric version of the system to operate in clinical
to rapidly generate new receptors are needed (Fig. 1a). samples. The resulting bactosensor can detect pathological levels
In order to address this challenge, we recently designed a of bile salts in serum from patients having undergone liver
synthetic receptor platform termed EMeRALD (Engineered transplant and provides a signal detectable with the naked eye.
Modularized Receptors Activated via Ligand-induced Our work highlights the flexibility and modularity of the
Dimerization)14 (Fig. 1b). EMeRALD receptors are derived EMeRALD receptor platform for rapid characterization and
from membrane-bound one-component systems, which are engineering of novel sensing capabilities in whole-cell biosensors.
bitopic proteins with a typical architecture of a cytoplasmic DNA-
binding domain (DBD), a juxtamembrane linker, a transmem-
brane region, and a periplasmic ligand-binding domain (LBD). Results
Direct fusion between the LBD and the DBD provides a simple Engineering of synthetic bile salt receptors in E. coli using the
yet efficient solution to transduce incoming signals into a tran- EMeRALD platform. Enteropathogenic bacteria such as Vibrio
scriptional output15,16. The EMeRALD platform operates in cholerae or Vibrio parahaemolyticus cause acute intestinal infec-
Escherichia coli and uses the DBD from the CadC pH sensor tions mediated by toxin secretion37. These pathogens use bile salts
which is inactive in its monomeric state17. Ligand-induced as an intestinal location signal to activate their virulence path-
dimerization of the LBD triggers dimerization of the cytoplasmic ways. Bile salt sensing is mainly under the control of inner
DBD and transcriptional activation18. This straightforward membrane sensor/cofactor couples TcpP–TcpH for V. cholerae35
mechanism offers the potential to modularize receptor sensing and VtrA–VtrC for V. parahaemolyticus38. As using pathogens as
and signaling by domain swapping. We previously built a syn- biosensors would involve significant host-specific regulation
thetic receptor responding to caffeine by using a nanobody for correction38,39 and biosafety containment issues, an alternative
this ligand as LBD14. In this work we attempted to leverage the strategy is to rewire pathogen-sensing modules of interest into a
EMeRALD platform to detect pathological biomarkers. As a pilot modular receptor platform operating in a surrogate host (Fig. 1a).
application, we aimed to detect bile salts, a biomarker of liver To engineer an EMeRALD bile salt receptor in E. coli, we fused
dysfunction. the V. cholerae TcpP bile-salt-sensing module and its transmem-
Liver disease includes dozens of pathologies such as cirrhosis, brane region (TM) to the DNA-binding domain of CadC (Fig. 1c).
hepatitis, liver cancer, hepatobiliary problems such as cholangitis, As a reporter, we placed superfolder green fluorescent protein
and drug-induced liver injury19–23. Liver disease is a global (sfGFP)40 under the control of the CadC target promoter,
healthcare burden accounting for two million deaths per year, pCadBA. We also expressed the cofactor protein TcpH,
and impacts the quality of life of millions of people worldwide24. previously described to protect TcpP from proteolysis by the V.
Liver is the second most common transplanted organ, but only cholerae RseP protease once dimerized in response to bile salts41.
10% of needs are met. Currently, liver disease diagnostics and We first confirmed using inducible gene expression systems that
monitoring is performed by assessing a panel of biomarkers25. co-expression of the TcpH cofactor was necessary for TcpP

2 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y ARTICLE

a
Rewiring Engineering &
Diagnostics Drug discovery
sensing of interest Repurposing

Probiotic
control

Bacteria with Modularized

sensing of interest synthetic receptor platform Applications

b EMeRALD platform
Ligand
Swappable
Ligand - Binding Domain

Periplasm

Cytosol
Synthetic receptor scaffold

™
Transmembrane region

Juxtamembrane linker

CadC DNA Binding Domain DBD

CadC-LBD Reporter
PCadBA

Bile salts TcpH TcpP Bile salts

Gall
TcpH / TcpP
bladder Liver

Enteropathogenic Rewiring
Bile salts bacteria
ToxT
Virulence
activation
by bile salts

toxT ctxAB Reporter

V. cholerae Cholera Toxin E. coli

Virulence Pathway EMeRALD Platform

pathogenic bacteria non -pathogenic bacteria

Fig. 1 Design principles and architecture of EMeRALD-based bacterial sensors. a General strategy of rewiring bacterial sensing modules into surrogate
hosts using a synthetic receptor platform. b Architecture and functional components of EMeRALD platform. The EMeRALD platform is composed of
swappable ligand-binding domains (LBD) that can be plugged into a synthetic receptor scaffold consisting of the DNA-binding domain (DBD) of the CadC
protein, a juxtamembrane (JM) linker, and a transmembrane region. The resulting synthetic transmembrane receptor is activated via ligand-induced
dimerization and triggers reporter gene expression. c Rewiring bile salt sensing into E. coli using the EMeRALD platform. Enteropathogenic bacteria detect
intestinal bile salts as host environmental cues for activating their virulence pathway. We plugged the V. cholerae bile salt receptor TcpP, and its cofactor
protein TcpH, into the EMeRALD platform operating in the surrogate host E. coli to build a synthetic bile salt receptor controlling expression of a
reporter gene.

function, using the primary bile salt taurocholic acid (TCA) as a We then assessed the versatility of the EMeRALD platform by
ligand (Supplementary Figs. 1–3). These data suggest that the connecting the VtrA/VtrC sensor system from V. parahaemolyticus36
RseP homolog present in E. coli (UniProt:P0AEH1) can degrade (Fig. 2c). We built a dual-expression system consisting of P9-CadC-
TcpP when not bound by TcpH. We also found that the relative VtrA and P5-VtrC, and tested its response to its canonical ligand
expression level of CadC-TcpP and TcpH were critical para- taurodeoxycholic acid (TDCA). We found that the VtrA/VtrC
meters affecting system performance (Supplementary Fig. 3). EMeRALD system was functional with a slightly higher LOD, similar
We then placed both proteins under the control of constitutive dynamic range and signal strength than the TcpP/TcpH EMeRALD
promoters42. We used the strong constitutive promoter P5 for system (Fig. 2d and Supplementary Fig. 6). These results highlight
TcpH and three different constitutive promoters of increasing the modularity and scalability of the EMeRALD platform, which
strengths, P9, P10, and P14, for CadC-TcpP (Fig. 2a) and tested supports the connection of different sensing modules to the receptor
their response to TCA41 (Fig. 2b and Supplementary Figs. 4–5) scaffold.
(Promoter strength: P5 > P14 > P10 > P9 (ref. 42)). We found that
the P9-CadC-TcpP variant had the lowest LOD, highest dynamic
range, and highest signal strength. These data confirm that the Synthetic bile salt receptors exhibit different specificity pro-
stoichiometry between CadC-TcpP and TcpH is a key parameter files. We then assessed the specificity profile of the synthetic bile
influencing receptor performance. salt receptors. Bile salts are classified in two categories: primary

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

a b
TcpP TcpH Bile salts

150
P14
Periplasm P10
< X P9
100
Cytosol RseP

RPU
DBD
50

0
0 50 100 150 200
TCA [ μM]
TcpH CadC-TcpP GFP
P5 P14 PCadBA
P15A
P10
P9

c VtrA VtrC d
Bile salts

150

Periplasm
Cytosol 100

RPU
DBD

50

0
0 50 100 150 200
TDCA [ PM]
VtrC CadC-VtrA GFP
P5 P9 PCadBA
P15A

e Primary bile salts Secondary bile salts

80
TcpPH
60
RPU

40

20

0
80 CA GCA TCA CDCA GCDCA TCDCA UDCA GUDCA DCA GDCA TDCA LCA

VtrAC
60
RPU

40

20

0
CA GCA TCA CDCA GCDCA TCDCA UDCA GUDCA DCA GDCA TDCA LCA

Fig. 2 Design, implementation, and characterization of synthetic bile salt sensors. a Overview of the TcpPH-EMeRALD system. The CadC DNA-binding
domain (DBD) is fused to the transmembrane and periplasmic domains of TcpP. Three constitutive promoters (P14, P10, and P9) were tested to tune the
transcription level of CadC-TcpP. Transcription of the TcpH cofactor is under the control of the constitutive promoter P5. In the absence of bile salts, CadC-
TcpP is probably degraded by an endogenous E. coli homolog of the V. cholerae protease RseP. In the presence of bile salts, CadC-TcpP dimerizes and forms
a stable complex with TcpH that protects it from proteolysis. The CadC-TcpP dimer then activates downstream expression of the GFP reporter. b Transfer
function of TcpPH-EMeRALD receptors controlled by different promoters in response to increasing concentrations of the bile salt taurocholic acid (TCA).
c Overview of the VtrAC-EMeRALD system. The CadC DBD is fused to the transmembrane and periplasmic domains of VtrA. CadC-VtrA and VtrC are
under the control of the P9 and P5 promoters, respectively. Bile salts binding to VtrA/VtrC heterodimeric complexes promote oligomerization of CadC-
VtrA and activate downstream expression of the GFP reporter. d Transfer function of VtrAC-EMeRALD receptor to increasing concentrations of the bile
salt taurodeoxycholic acid (TDCA). e Bile salt specificity profiles for TcpPH-EMeRALD and VtrAC-EMeRALD systems. The full names and molecular
structure of the different bile salts are listed in Supplementary Fig. 1. The curve graphs (b, d) correspond to the mean value of three replicates performed in
triplicate on three different days (n = 3 biologically independent samples). The bar graph (e) corresponds to the mean value of three replicates performed
in triplicate on three different days (n = 3 biologically independent samples). Green dots correspond to the values for all replicates. Error bars: ±SD. RPU
reference promoter units. Cells growing in exponential phase were incubated with bile salts for 4 h before flow cytometry measurement.

4 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y ARTICLE

bile salts (including TCA) are produced by the liver while sec- region of the TcpP-sensing module. Further kinetic analysis
ondary bile salts arise from modification of primary bile salts by revealed that the sequence variation of this loop region changes
gut microbiome metabolism. Primary bile salts are upregulated in reaction speed and system interaction of bile salts with the
serum and urine of patients with liver disease31–33. Previously synthetic receptor, and variant V18 has 13-fold increase in ligand
identified virulence activating factors for V. cholerae include affinity and faster response at low ligand concentration compared
TCA, glycocholate, and cholic acid41. We measured the response to wild type TcpP (Supplementary Fig. 11).
of the bactosensor to a panel of 12 different bile salts, including To better understand the sequence features influencing the
both primary and secondary types (Fig. 2e and Supplementary response of TcpP to bile salts, we sequenced the whole pool of
Figs. 7 and 8). Interestingly, while no sensing module was specific enriched variants by NGS. Surprisingly, the sequence features of
for a single bile salt species, the CadC-TcpP system was highly functional variants were different from those expected from
specific for primary conjugated bile salts (especially TCA and natural TcpP homologs (Fig. 3f and Supplementary Fig. 12). First,
GCDCA), and did not respond to secondary bile salts. On the and in contrast with wild type TcpP homologs, we observed a
other hand, the CadC-VtrA system had a larger spectrum of bile strong depletion of long-chain, negatively charged amino acids
salt specificity, mainly responding to secondary conjugated bile (Asp and Glu) along with long-chain polar amino acids (Asn and
salts GDCA and TDCA. Due to the link between primary bile Gln) at position 211. Lysine at position 211 also appeared to be
salts and liver diseases, we selected the TcpP/TcpH system to depleted in functional variants (despite being commonly found at
develop a bile salt bactosensor for medical diagnosis. this position in other TcpP homologous proteins). Second, amino
acids with bulky aromatic side chain such as Phe and Tyr, and
hydrophobic side chain such as Leu were highly conserved in
Directed evolution of TcpP-sensing module for improving selected functional variants, strongly indicating the important
LOD and higher sensitivity. Sensor sensitivity and LOD are key role of hydrophobic residues at position 211 in the C-terminal
parameters for biosensors applications. We aimed to identify key loop region for the function of V. cholerae TcpP. We chose the
residues determining the sensitivity of the TcpP-sensing module, best engineered variant, termed TcpP18, for further development
and targeted those to improve synthetic receptor sensitivity and of a clinical bactosensor.
LOD. To do so, we coupled comprehensive mutagenesis with
functional screening and next-generation sequencing (NGS), an
approach that supports the identification of functional variants Development of a colorimetric version of the bactosensor.
together with the sequence determinants within local structural Colorimetric assay provides a simple and intuitive method for
motifs43–45. This strategy has also been used to engineer ortho- simple and direct estimation of test results by the naked eye. In
gonal two-component systems46. addition, colorimetric assays support straightforward develop-
Transition from intramolecular to intermolecular disulfide ment of quantitative assays using smartphone-based platforms for
bonds between TcpP monomers is a key determinant of TcpP POC or home-based diagnosis47. We used TcpP18 coupled with
response to bile salts and is mediated by two cysteine residues, the reporter beta-galactosidase LacZ (termed TcpP18–LacZ) and
Cys207 and Cys218 (ref. 41). By performing multiple sequence its substrate chlorophenol red-β-D-galactopyranoside (CPRG) to
alignments of different TcpP bacterial homologs (Supplementary provide a colorimetric output48 (Fig. 4a, see “Methods” for
Fig. 9), we found a significant conservation of the amino acids details). Similarly to the biosensor equipped with a GFP output,
flanked by these two cysteines (Fig. 3a). Secondary structure the bile salt specificity profile of the TcpP18–LacZ system was
prediction and ab initio 3D prediction using the Rosetta modeling slightly shifted from TCA to GCDCA (Fig. 4b and Supplementary
suite (Fig. 3b and Supplementary Fig. 10) suggested that each Fig. 13). We thus evaluated the LOD and signal output threshold
cysteine was located in rigid beta sheets separated by a flexible of TcpP18–LacZ in response to increasing concentrations of
loop region between Asn210 and Gln213. This loop propensity to GCDCA. We also explored the influence of varying cell density
form a turn would allow the two beta sheets and the cysteines to and incubation time (Fig. 4c and Supplementary Figs. 14 and 15).
come in close proximity and form an intramolecular disulfide By adjusting cell density or incubation time we improved the
bond. We hypothesized that the flexibility of the turn region dynamic and operating ranges of TcpP18–LacZ (Fig. 4c and
between Cys207 and Cys218 was a key parameter controlling the Supplementary Figs. 14 and 15). After optimization, the
transition rate between the two states, and that altering its amino Tcp18–LacZ demonstrated linear response to GCDCA within the
acid composition could change the system’s sensitivity to bile concentration from 0 to 40 μM in 1 h (Supplementary Fig. 15).
salts. This allowed us to tune the threshold activation level to match
We thus built a comprehensive mutational library (NNK x 4, various clinical levels associated with specific liver-related medical
theoretical library complexity ≌1.05 × 106 variants, see “Methods” conditions.
for details) targeting the NYEQ residues inside the turn, and
cloned it into the plasmid constitutively expressing CadC-TcpP
and producing GFP in response to bile salts (Fig. 3c). The Bactosensor-mediated detection of elevated bile salts levels in
resulting library was induced with TCA, and GFP-positive serum from patients with liver transplant. We then prototyped
variants were isolated by fluorescence-activated cell sorting. We our bactosensor for the detection of bile salts in clinical condi-
performed three rounds of enrichment (200 μM of TCA as ligand tions. To do so, we tested the sensor on samples from patients
in first and second rounds of selection, and 20 µM for the third having undergone liver transplantation. After liver transplant, the
round) and observed an increasing fraction of the cell population main complications are bile ducts stenosis and acute cellular
responding to different ligand concentrations (20–80 µM) rejection (ACR). In order to detect these complications at an early
(Fig. 3d). We collected, cultured, and sequenced single variants stage, liver tests are performed regularly. Serum bile salt con-
and tested their response to TCA (Fig. 3e). We found that centration has been shown to be a good indicator for the
comprehensive mutagenesis of residues Asn210 to Gln213 could assessment of liver dysfunction after liver transplantation49. A
alter the limit-of-detection, the sensitivity, and the fold activation field-deployable method for bile salt assessment would greatly
of our biosensor. The 3.3-fold difference in LOD between V18 improve the monitoring of these patients, ultimately allowing fine
and V22 (EC50 from 28.3 to 92.5 µM, Table 1) indicated the broad grained monitoring performed at-home by the patients
range of sensitivity engineering obtained by mutating the loop themselves.

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

a 207 * 218 b
A.wondanis
A.logei
Target loop: NYEQ
V.cholerae
V.vulnificus
Photobacterium
V.panuliri
V.owensii
Cys 207
V.campbellii
V.hyugaensis
Cys 218
V.harveyi
V.Jasicida
PSIPRED

c
4 NNK library Single colony
TCA analysis

induction sorting Next Generation

Sequencing

GFP (-) GFP (+)

3x

d
Wild type Evo1 Evo2 Evo3
OFF ON 200 µM 200 µM I:80 µM II: 40 µM III: 20 µM

- ligand 1.4% 8.8% 39.0% 9.9% 3.8%

Cell #

+ ligand

GFP fluorescence intensity (A.U.)

e f
140
2.1

105
1.5
Information content
RPU

70 wt_NYEQ
1
V3_DFGV

35 V16_DFLT
0.5
V18_VFSD

0 0
210 211 212 213
0 50 100 150 200
Taurocholic acid [PM]

Fig. 3 Directed evolution of TcpP for improving LOD and sensitivity. a Multiple sequence alignment of the C-terminal periplasmic domains from TcpP
homologous proteins. Colors indicate sequence identities. Secondary structures prediction by PSIPRED is shown below. The red line box indicates the
region of interest for mutational scanning. b Ab initio modeling of the C-terminal region of the TcpP-sensing module. The two cysteine residues are shown
as spheres. The amino acid residues NYEQ in the loop region of interest are labeled as sticks. c Schematic diagram of the screening procedure to obtain
functional TcpP loop variants. The cell library was submitted to several rounds of sorting based on fluorescent signal output produced in response to bile
salts. Individual clones were then isolated and characterized while the sequence of the mutated loop for the whole pool of isolated variants was analyzed by
NGS. d Screening conditions and results of TcpP directed evolution. The gate used for cell sorting is colored in green. The x-axis indicates the fluorescence
intensity in arbitrary units (AU). The gate for fluorescence-activated cell sorting is determined using the fluorescence distribution of TcpPH-EMeRALD
sensor with or without the 200 μM TCA (left panel). e Transfer function of TcpP functional variants with improved sensitivity in response to increasing
concentrations of TCA. Amino acid sequences of the loop regions are indicated. The curve graphs correspond to the mean value of three replicates
performed in triplicate on three different days (n = 3 biologically independent samples). Error bars: ±SD. RPU reference promoter units. f Sequence logos of
the flexible loop region from selected TcpP functional variants derived from NGS data.

We tested our bactosensor in 21 serum samples from liver years (see Supplementary Tables S2 and S3). These patients had
transplantation patients (Fig. 5a and Supplementary Fig. 16). The received a liver transplant for end-stage liver disease as a result of
patients were followed at the Montpellier hospital after their liver alcoholic related liver disease or non-alcoholic fatty liver disease,
transplant, most of them having been performed in the last 2 chronic cholangitis, or liver cancer. A complete hepatic check-up

6 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y ARTICLE

Table 1 Functional analysis of selected TcpP functional C-terminal loop region of V. cholerae TcpP. The exact role of
variants. Tyr211 is still unclear, but this residue may affect TcpP–TcpP
dimer formation or TcpP–TcpH interaction. While the unique
presence of Tyr at the 211 position in V. cholerae was visible on
# Variant Sequence EC50 (μM) Response Max
multiple sequence alignments, the link between this residue and
EC50 (RPU) fold change
TcpP function could not be inferred by this approach.
V22 YYVL 92.531 56.19 69.06 This discovery exemplifies how our synthetic receptor platform
TcpPwt NYEQ 89.405 56.5 164.74
offers a powerful strategy to interrogate the sequence–function
V7 WYVH 86.029 53.37 51.31
V78 FYES 84.668 59.62 69.06
relationship of bacterial sensing modules in a massively parallel
V11 YYIV 82.167 53.83 31.76 fashion. Such studies in the natural pathogenic host would have
V14 TFLA 81.636 62.97 222.67 been tedious because of complex pathway regulations and safety
V3 DFGV 61.48 61.18 142.21 issues would have limited the final bactosensors to a few expert
V19 FFKA 59.186 61.29 127.75 groups. In contrast, our platform provides a straightforward and
V16 DFLT 42.058 59.34 75.73 scalable strategy to study pathogen signaling in surrogate hosts,
V18 VFSD 28.344 58.99 84.92 usable by the larger scientific community. The EmeRALD system
coupled with the deep-mutational scanning, methods to navigate
protein sequence space52–55, designer libraries, and Flow-Seq
was performed, and serum bile salts were also measured using an analysis45 can serve as a general platform for studying bacterial
enzymatic assay (Supplementary Table S4 for clinical data). We sensing modules, engineering their ligand specificity and their
noticed that our bactosensor exhibited a negative response to response properties. In particular, our system is ideal for under-
some clinical serum samples. Therefore, we tried to minimize the standing transmembrane one-component signaling, which
potential interferences by sample dilution and compared the mechanism and regulation are currently underexplored16. On the
readouts with serum bile salt concentration measured by same line, EMeRALD receptors could be applied to the discovery
enzymatic assay to determine the proper dilution factor of inhibitors of virulence pathways through screening of chemical
(Supplementary Fig. 16). We found that patients who had a high or natural substances libraries. Finally, the bile salt-sensing sys-
potential of ACR after liver transplantation (serum bile acid tem could be applied to control the activity of engineered pro-
>37 μM)50 had significant and visible colorimetric signal changes biotics upon arrival in the gut, as recently proposed in Bacteroides
in bactosensor assays (Supplementary Fig. 17). Three patients in thetaiotomicron56.
particular raised our attention: patients #5, 10, and 13. These Here we show a pilot application of the EMeRALD technology
three patients had elevated serum bile salts concentration. Two of to the field of medical diagnostics by detecting abnormal bile salt
them (5 and 10) presented abnormalities in their hepatic levels in patient samples. As bile salts dysregulation and gut
enzymatic values (Aspartate Aminotransferase (ASAT), Alanine dysbiosis are critical in the pathogenesis of liver diseases or gas-
Aminotransferase (ALAT), gamma-glutamyl transferase (GGT), trointestinal cancer57–62, there is an urgent need for POC assays
Alkaline Phosphatase (ALP), and bilirubin). For these patients, for bile salt monitoring63. Our bactosensor results correlate well
the bile salt bactosensor produced the strongest colorimetric with hospital tests and produce a very strong signal detectable
change easily detectable with the naked eye (Fig. 5b). These with the naked eye for the three most critical patients, using very
results indicate that our bactosensor is able to provide a simple, small sample volumes, demonstrating the potential of our tech-
reliable, and cost-effective method for monitoring patient nology for future POC estimation of liver dysfunction.
condition after liver transplantation. Yet, several improvements are needed to translate our platform
into a POC or home diagnostics system. First, given the size of
Discussion our patient cohort and its diversity, further studies are required to
Microbes detect and process myriad environmental signals, pro- fully validate our approach in a clinical context. Second, while our
viding a vast sensing repertoire for engineering biosensors usable work demonstrates the clinical applicability of our platform, our
for several applications. However, the intricacy of signaling net- assay by itself is still not compatible with POC operation. In
works and our limited understanding of their biochemical particular, it still includes several liquid handling and incubation
properties restrain their direct use. Here we presented a general steps, such as cell growth, spinning, and lysis. Lyophilizing the
strategy to rewire sensing modules of interest into a well- bactosensors once the first preparations steps have been done
characterized synthetic receptor platform, EMeRALD, enabling would provide a simple test format. Several examples have shown
their fine tuning and repurposing for biomedical applications. In the possibility to lyophilize bacteria and maintain their function,
the future, these receptors could be coupled with genetic circuits even on paper64. The coupling of whole-cell biosensors with
performing multiplexing logic, memory, and signal amplification microfluidics devices into a lab-on-chip apparatus has also been
to engineer even more sophisticated whole-cell biosensors10,51. extensively documented51,65 and could be applied to our system.
We were able to connect two different bile salt-sensing mod- Finally, the use of pigment reporters or the implementation of
ules having different specificity profiles, the TcpP/H and VtrA/C autolysis systems could suppress the additional step to lyse
systems, which respond mostly to primary and secondary bile cells66–68.
salts, respectively. Importantly, our capacity to engineer synthetic Regarding the biological sample itself, serum preparation from
bile salt signaling using only these protein domains demonstrates blood requires additional manipulations. While those can be
that these modules are the only essential components required for performed in POC using low-tech devices such as the paperfuge,
bile salt sensing in their natural host. By performing directed detecting bile salts in more readily accessible physiological fluid
evolution of TcpP, we improved the sensitivity and decreased the might be advantageous. Urine offers such an opportunity, with
LOD of the sensor. In addition, we discovered previously the added benefits of a non-invasive collection process. We
unknown amino acid sequence features influencing TcpP func- indeed validated the function of our bactosensor to detect exo-
tion and potentially relevant for V. cholerae virulence. For genously bile salts in urine samples (Supplementary Fig. 19).
instance, we found a stringent functional requirement for amino However, detecting endogenous bile salts in patients’ urines, even
acids with a nonpolar aromatic side chain at position 211. This with high bile salt levels, proved unreliable. We attribute this issue
requirement indicates rigorous steric interactions located in the to the fact that most bile salts excreted in urines are sulfonated

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

a TcpP18 - LacZ Preparation Incubation Readout

15 mins - 1 hour 6 – 8 mins

LacZ

LacZ

CPRG CPR
Lysis / CPRG

[Bile salts]
Aliquot:
LacZ OD 0.1 ~ 1.0
PCadBA / 50 µL rxn
P15A
Harvest 37 C RT
w/o shaking w/o shaking

b 3.5
3.0
2.5
2.0
ΔA580

1.5
1.0
0.5
0.0
-0.5
CA GCA TCA CDCA GCDCA TCDCA UDCA GUDCA DCA GDCA TDCA LCA

c
USER SPECIFICATIONS TUNING OPTIMIZED ASSAY

Disease-specific
bile salts concentration range
LOW: e.g. early cirrhosis

e.g. Non-alcoholic Cell Incubation Input Output

HIGH density
fatty liver disease time

60 mins : 0 – 40 μM 30 mins : 0 – 140 μM

4 4
OD 0.1 OD 0.1
OD 0.2 OD 0.2
3 3
OD 0.5 OD 0.5
A580

OD 1.0 OD 1.0
A580

2 2

1 1

0 0
0 10 20 30 40 50 0 25 50 75 100 125 150
GCDCA concentration [μM] GCDCA concentration [μM]

60 min 30 min
OD:1 OD: 1

GCDCA [μM] 0 1 2.5 5 10 20 30 40 GCDCA [μM] 0 20 40 60 80 100 120 140

Fig. 4 Colorimetric assay for bactosensor-mediated bile salts detection. a Schematic diagram for the design and operation procedure of the TcpP18–LacZ
system for bile salt detection. The TcpP18–LacZ uses the TcpP loop variant V18 as sensing module for lower LOD and higher sensitivity and LacZ as a
colorimetric reporter, with CPRG as a substrate which is converted to chlorophenol red turning the reaction from yellow to purple. b Bile salt specificity
profile of the TcpP variant V18 characterized using TcpP18–LacZ sensor. Response of TcpP18–LacZ was quantified as ΔA580 (the difference in absorbance
at 580 nm (A580) with or without ligand bile salts). The bar graph corresponds to the mean value of three replicates performed in triplicate on three
different days (n = 3 biologically independent samples). Error bars: ±SD. Cells growing in the exponential phase were incubated with bile salts for 4 h
before flow cytometry analysis. c Optimization and quantification of TcpP18–LacZ response to the bile salt glycodeoxycholic acid (GCDCA). The LOD and
response dynamic range of TcpP18–LacZ were fine-tuned by varying the cell concentration and incubation time. Data points correspond to the mean value
of three replicates performed in triplicate on three different days (n = 3 biologically independent samples). Error bars: ±SD. See “Methods”, main text, and
SI for details.

8 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y ARTICLE

Bile
salts Bacterial biosensor assay
Colorimetric

LacZ
+
5 μL
Serum
X 21
Liver transplantation patients

Serum bile salt concentration [PM]

2.0 200
Bacterial sensor
Sigma
Bactosensor ('A580)

1.5 150

(Sigma assay)
1.0 100

0.5 50

0.0 0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
-0.5 -50

Fig. 5 Bactosensor-based pathological bile salt detection in clinical samples. a Serum samples from 21 liver transplantation patients being followed at a
Montpellier hospital were analyzed using a bactosensor equipped with the TcpP18–LacZ sensor as described in Fig. 4a, with a 10-fold dilution and 2 h
incubation time (Supplementary Fig. 14). b Comparing the analysis result of 21 clinical serum samples between TcpP18–LacZ and a bile salt assay kit. The
response of TcpP18–LacZ is shown in purple bars, left axis. The serum total bile salt concentration measured by a bile salt enzymatic assay kit is labeled in
green asterisks, right axis. The bar graph corresponds to the mean value of three replicates performed in triplicate on three different days (n = 3 biologically
independent samples). Error bars: ±SD. See Supplementary Fig. 18 for plots with all replicates represented.

after they pass through the kidney69, and that our biosensor with p15a origin of replication and chloramphenicol resistance gene by isothermal
seems unable to detect sulfonated versions of bile salts (Supple- Gibson assembly. All experiments were performed using E. coli strain NEB10β
(New England Biolabs). Plasmids and materials will be made available through
mentary Fig. 20). This issue could be solved by treating samples Addgene.
or expressing in bacteria a bile acid sulfate sulfatase70, or engi-
neering the TcpP or VtrA modules to detect sulfonated bile salts. Functional characterization of synthetic bile salt receptors with sfGFP fluor-
The specificity shift observed with TcpP-18 compared to the wild escence outputs. For the experiments of constructs with constitutive promoters
type TcpP suggests that such engineering is feasible. (Figs. 2b, d, e, and 3e), plasmids encoding different constructs were transformed
We observed that our bactosensor had relatively low or nega- into chemically competent E. coli NEB10β (New England Biolabs), plated on LB
tive responses to several serum samples (Fig. 5b and Supple- agar plates supplemented with 25 μg/mL chloramphenicol and incubated at 37 °C
overnight. For each measurement, three fresh colonies were picked and inoculated
mentary Fig. 16). This behavior may be due to differential matrix into 5 mL of LB/chloramphenicol and grown at 37 °C with vigorous shaking for
effects arising in samples from different patients. Large-scale 16–18 h. In the next day, the cultures were diluted 1:100 into 1 mL of LB/chlor-
analysis of clinical samples chemical composition coupled with amphenicol medium with different concentration of bile salts in 96 deep-well plates
close monitoring of the diet and medical treatments for each (Greiner bio-one), incubated at 37 °C with vigorous shaking for further 4 h and
analyzed by flow cytometry. All experiments were performed at least three times in
patient could help identify the causes of this effect. triplicate on three different days. For bile salt specificity profiles of TcpPH-
Finally, our sensing platform based on genetically engineered EMeRALD and VtrAC-EMeRALD systems (Fig. 2e), experiments were performed
organisms will require containment systems before being with the same protocol, and using each bile salt at a 80 µM concentration. For the
deployed71,72, together with open ethical and societal debate73. experiments of constructs with inducible promoters (Supplementary Fig. 2), the
overnight cultures were diluted 1:100 into 1 mL of LB/chloramphenicol medium
Addressing most of these challenges entails repurposing existing with different concentrations of IPTG, 1.5 mM of benzoic acid, and different
systems and frameworks, together with solving engineering concentrations of bile salts in 96 deep-well plates. And then follow the same
puzzles that are within current trends of synthetic biology and procedures as constructs with constitutive promoters. All chemicals used in this
biosensor research. research were purchased from Sigma-Aldrich.
Here we have shown that the EMeRALD platform, which can
accommodate natural or synthetic sensing modules, provides a Calculation of relative promoter units (RPUs). Fluorescence intensity mea-
surements among different experiments were converted into RPUs by normalizing
versatile and scalable solution to develop new sensing modalities in them according to the fluorescence intensity of an E. coli strain NEB10β containing
bacteria, with the potential to help address practical challenges. We a reference construct and grown in parallel for each experiment. We used the
anticipate that EMeRALD receptors will be repurposed for other constitutive promoter J23101 and RBS_B0032 as our in vivo reference standard
applications in medical diagnostics, bacterial therapeutics, and and placed superfolder GFP as a reporter gene in plasmid pSB4K5. We quantified
the geometric mean of fluorescence intensity (MFI) of the flow cytometry data and
environmental monitoring, supporting the transition of bacterial calculated RPUs according to the following equation:
biosensors towards a wide range of real-world applications.
RPU ¼ ðMFIsample Þ=ðMFIreference promoter Þ ð1Þ

Methods
Plasmids and strains. Genetic parts of constructs used in this study are provided Flow cytometry analysis. Flow cytometry was performed using an Attune NxT
in Supporting Information. All constructs were cloned into plasmid J64100_p15A cytometer coupled with high-throughput autosampler (Thermo Fisher Scientific)

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

and Attune NxT™ Version 2.7 Software. In all, 30,000 cells were collected for each cell pellets were resuspended with LB medium (without antibiotics) at 1.1× of the
data point. Experiment on Attune NxT were performed in 96-well plates with final OD. Five microliters of different bile salts (prepared in LB medium) was
setting; FSC: 200 V, SSC: 380 V, green intensity BL1: 440 V (488 nm laser and a added into 45 μL of resuspended cells in 96-well plates. The plate was incubated at
510/10 nm filter). Flow cytometry data were analyzed using FlowJo 10.0.8r1 37 °C for 10–60 min without shaking. After incubation, 50 μL of B-PER solution
(Treestar Inc., Ashland, USA). All raw data values are listed in Supplementary (Thermo Fisher) with 800 μM CPRG was added into the cell culture directly. Due
Information. The gating strategy is shown in Supplementary Fig. 21. to the low permeability of substrate CPRG, it would take more than 2 h to have
slight visual colorimetric change, and the difference between with or without ligand
is much less than the lysis method. The mixture was further incubated for 6 min at
4 x NNK library construction. Using P9-CadC-TcpP plasmid as a template, the
room temperature and then 50 μL of 1 M sodium carbonate was added to stop the
insert with 4 x NNK library was amplified by Phusion Flash High-Fidelity PCR
reaction. The absorbance at 580 nm was measured with a plate reader (Biotek;
Master Mix (Thermo Fisher Scientific) with the primers I5 and I3; the vector was
Cytation3). For the clinical sample assay, serum samples were heat-inactivated by
amplified with the primers V5 and V3 (primer sequence details are listed in
incubation in a 56 °C water bath for 30 min. Five microliters of serum samples were
Supplementary Table 9). The purified fragments were first digested with BsaI-HFv2
added into 45 μL of resuspended cells in a 96-well plate and incubated at 37 °C
(New England Biolabs), and then ligated by T4 ligase at 4 °C overnight. Five
without shaking for 2 h before lysis.
micrograms of ligation product was transformed into E.coli strain NEB10β with
electroporation.
Statistical analysis. The signal outputs with significant differences (compared
with the signal without ligand) verified by unpaired two-tailed Student’s t-test are
Cell sorting. Cell sorting was performed using a S3 cell sorter (Bio-rad) with Bio- marked by asterisk (*p < 0.05, **p < 0.01, ***p < 0.001, and ****p < 0.0001,
rad S3 software Prosort version 1.6. In total, 100,000 cells under different induction respectively). Dose–response curves were fitted using a nonlinear regression model
conditions (as shown in Fig. 3) were gated and collected in SOC medium at each with Hill Slope (four-parameter dose–response curve). All data analysis was per-
round of sorting. Experiments on Bio-rad S3 cell sorter were performed with formed on GraphPad Prism (Version 8.0.2).
setting; FSC: 400 V, SSC: 284 V, green intensity FL1: 680 V (488 nm laser and a
510/10 nm filter). The sorted cells were further inoculated in 10 mL of LB/chlor-
amphenicol medium at 37 °C with vigorous shaking for a further 16–18 h. The cell Patient samples. Serum samples were consecutively collected from hospitalized or
cultures positively selected cells were then applied to the next round of selection. ambulatory liver transplant patients at the Hepatology and Liver Transplantation
For the first two rounds, the loop variant library was induced with 200 μM of TCA. Unit, Hôpital Saint-Eloi in Montpellier (France) in the month of July 2020. All
In the third round of evolution, cells were further induced at different con- patients underwent liver transplantation between November 2006 and April 2020.
centrations of TCA to isolate more sensitive variants. For each round of screening, Clinical and biological data were also recorded for every patient included in the
cells from an overnight culture were diluted 1:100 in LB with or without TCA and study from their medical files. Abnormal liver enzymes were defined by alanine
grown for 16 h at 37 °C before being sorted. The gating strategy is shown in Fig. 3d transaminase > 41 U/L, aspartate transaminase > 40 U/L, alkaline phosphatase >
and Supplementary Fig. 22. 130 U/L, GGT > 60 U/L or total bilirubin (TBil) >21 µmol/L. The ethics committee
of the University Hospital of Montpellier granted ethical approval (Number:
198711) and all patients signed an informed consent.
Rosetta modeling. Ab initio structural modeling. Structural models of the TcpP
C-term segment (residues 182 to 211) were generated using the ROSETTA ab initio
3D prediction protocol74 with 1D sequence and 2D predicted secondary structure Reporting summary. Further information on research design is available in the Nature
as input data. Research Reporting Summary linked to this article.

Next-generation sequencing. Cells sorted from the third round of evolution were Data availability
further induced with 320 μM of TCA to collect the fully activated TcpP variants. The authors declare that all data supporting the findings of this study are available within
Plasmids extracted from sorted cells were further amplified (primer sequence the paper, its supplementary information files, or public repositories. NGS data have been
details are listed in Supplementary Table 9) and added UMI barcodes with first deposited in the NCBI SRA database with the BioProject ID: PRJNA714981. The flow
round primers (see Supplementary Materials for sequence details). After PCR clean cytometry data have been deposited in FlowRepository (https://flowrepository.org), with
up, the first round PCR products were further amplified by second round PCR IDs: FR-FCM-Z3K6 (Fig. 2b), FR-FCM-Z3KL (Fig. 2d), FR-FCM-Z3KM (Fig. 2e), FR-
primers with NGS index. The PCR products were further purified with AMPure XP FCM-Z3KT (Fig. 3e_V3_7_11), FR-FCM-Z3KU (Fig. 3e_V14_16_18), FR-FCM-Z3KV
paramagnetic beads to remove the contaminants. Samples were sequenced on (Fig. 3e_V19_22_78), FR-FCM-Z3KR (Fig. S3A), and FR-FCM-Z3LZ (Fig. S3B). Source
Illumina PE250 platform at Novogen (Hong Kong, China) with paired-end reads of data are provided with this paper.
250 bases.
Code availability
NGS data processing. The datasets generated and analyzed during the current Python scripts used to process NGS data are available from github at: https://github.com/
study are available in the NCBI SRA repository (BioProject ID: PRJNA714981; hungjuchang/NGS-Sequence-Counts_PWM_PSSM-calculator.git
https://www.ncbi.nlm.nih.gov/bioproject/714981). Sequence counts were converted
into amino acid sequences and listed in Supplementary Data 1. The python scripts
for NGS data processing are available from Github Page: https://github.com/ Received: 7 June 2021; Accepted: 12 August 2021;
hungjuchang/NGS-Sequence-Counts_PWM_PSSM-calculator.git. The resulting
data were used for sequence logo preparation by R package Logolas version 1.3.1
(ref. 75).

Time-course measurements. For Supplementary Fig. 11, bacteria cells containing

different CadC-TcpP promoter variants (with P5-TcpH) were streaked on LB plus References
1.5% agar plates and grown overnight at 37 °C. Threes different colonies were 1. Lee, S., Huang, H. & Zelen, M. Early detection of disease and scheduling of
inoculated into 1 mL of LB plus antibiotics in a 2 mL 96 deep-well plates (Thermo screening examinations. Stat. Methods Med. Res. 13, 443–456 (2004).
Fisher Scientific, 278606) sealed with AeraSeal film (Sigma-Aldrich, A9224-50EA) 2. Glasziou, P., Irwig, L. & Mant, D. Monitoring in chronic disease: a rational
and incubated at 37 °C for 16 h with shaking (300 r.p.m.) and 80% of humidity in a approach. BMJ 330, 644–648 (2005).
Kuhner LT-X (Lab-Therm) incubator. After overnight growth the cells were 3. Rohr, U.-P. et al. The value of in vitro diagnostic testing in medical practice: a
diluted 1:250 into 200 µL 96-well plates with M9 minimal medium with 0.4% status report. PLoS ONE 11, e0149856 (2016).
glycerol plus different TCA concentration. Measurements were done on a Cyta- 4. Dunn, J., Runge, R. & Snyder, M. Wearables and the medical revolution. Per.
tion3 microplate reader (Biotek Instruments, Inc.). Absorbance at 600 nm and GFP Med. 15, 429–448 (2018).
(excitation 485 nm, emission 528 nm, gain 80) were measured every 10 min for 4 h 5. Yang, Y. & Gao, W. Wearable and flexible electronics for continuous
(linear range of growth). Data were collected using a Gen5 Microplate Reader and molecular monitoring. Chem. Soc. Rev. 48, 1465–1491 (2019).
Imager Software version 3.03. Raw data were processed by subtracting auto- 6. Nayak, S., Blumenfeld, N. R., Laksanasopin, T. & Sia, S. K. Point-of-care
fluorescence and normalizing by OD values. Michaelis–Menten enzyme kinetics diagnostics: recent developments in a connected age. Anal. Chem. 89, 102–123
was evaluated by using GraphPad Prism (version 8.0.2) nonlinear (2017).
regression model. 7. Gui, Q., Lawson, T., Shan, S., Yan, L. & Liu, Y. The application of whole cell-
based biosensors for use in environmental analysis and in medical diagnostics.
Colorimetric assay of bactosensors for bile salt detection. Overnight cultures of Sensors 17, 1623 (2017).
TcpP–LacZ were diluted in 1/250-fold into 100 mL of LB with 25 μg/mL chlor- 8. Quillardet, P., Huisman, O., D’Ari, R. & Hofnung, M. SOS chromotest, a
amphenicol, incubated at 37 °C for 4 h to reach an OD600 of around 0.4. The cells direct assay of induction of an SOS function in Escherichia coli K-12 to
were put on ice for 30 min and then spun down at 4 °C and 2200g for 10 min. The measure genotoxicity. Proc. Natl Acad. Sci. USA 79, 5971–5975 (1982).

10 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y ARTICLE

9. Hicks, M., Bachmann, T. T. & Wang, B. Synthetic biology enables 42. Mutalik, V. K. et al. Precise and reliable gene expression via standard
programmable cell‐based biosensors. Chemphyschem 21, 132–144 (2020). transcription and translation initiation elements. Nat. Methods 10, 354–360
10. Courbet, A., Endy, D., Renard, E., Molina, F. & Bonnet, J. Detection of (2013).
pathological biomarkers in human clinical samples via amplifying genetic 43. Fowler, D. M. et al. High-resolution mapping of protein sequence-function
switches and logic gates. Sci. Transl. Med. 7, 289ra83 (2015). relationships. Nat. Methods 7, 741–746 (2010).
11. Chang, H.-J., Voyvodic, P. L., Zúñiga, A. & Bonnet, J. Microbially derived 44. Chang, H.-J. et al. Loop-sequence features and stability determinants in
biosensors for diagnosis, monitoring and epidemiology. Microb. Biotechnol. antibody variable domains by high-throughput experiments. Structure 22,
10, 1031–1035 (2017). 9–21 (2014).
12. Menezes, A. A., Montague, M. G., Cumbers, J., Hogan, J. A. & Arkin, A. P. 45. Cambray, G., Guimaraes, J. C. & Arkin, A. P. Evaluation of 244,000 synthetic
Grand challenges in space synthetic biology. J. R. Soc. Interface 12, 20150803 sequences reveals design principles to optimize translation in Escherichia coli.
(2015). Nat. Biotechnol. 36, 1005–1015 (2018).
13. Chang, H.-J. & Bonnet, J. Synthetic receptors to understand and control 46. McClune, C. J., Alvarez-Buylla, A., Voigt, C. A. & Laub, M. T. Engineering
cellular functions. Methods Enzymol. 633, 143–167 (2020). orthogonal signalling pathways reveals the sparse occupancy of sequence
14. Chang, H.-J. et al. A modular receptor platform to expand the sensing space. Nature https://doi.org/10.1038/s41586-019-1639-8 (2019).
repertoire of bacteria. ACS Synth. Biol. 7, 166–175 (2018). 47. Roda, A. et al. Smartphone-based biosensors: a critical review and
15. Ulrich, L. E., Koonin, E. V. & Zhulin, I. B. One-component systems dominate perspectives. Trends Anal. Chem. 79, 317–325 (2016).
signal transduction in prokaryotes. Trends Microbiol. 13, 52–56 (2005). 48. Sicard, C. et al. A rapid and sensitive fluorimetric β-galactosidase assay for
16. Jung, K., Fabiani, F., Hoyer, E. & Lassak, J. Bacterial transmembrane signalling coliform detection using chlorophenol red-β-D-galactopyranoside. Anal.
systems and their engineering for biosensing. Open Biol. 8, 180023 (2018). Bioanal. Chem. 406, 5395–5403 (2014).
17. Lindner, E. & White, S. H. Topology, dimerization, and stability of the single- 49. Kumar, S. et al. Non invasive diagnosis of acute cellular rejection after liver
span membrane protein CadC. J. Mol. Biol. 426, 2942–2957 (2014). transplantation—current opinion. Transpl. Immunol. 47, 1–9 (2018).
18. Schlundt, A. et al. Structure-function analysis of the DNA-binding domain of 50. Janßen, H. et al. Serum bile acids in liver transplantation—early indicator for
a transmembrane transcriptional activator. Sci. Rep. 7, 1051 (2017). acute rejection and monitor for antirejection therapy. Transpl. Int. 14,
19. Bruix, J., Han, K.-H., Gores, G., Llovet, J. M. & Mazzaferro, V. Liver cancer: 429–437 (2001).
approaching a personalized care. J. Hepatol. 62, S144–S156 (2015). 51. Wan, X. et al. Cascaded amplifying circuits enable ultrasensitive cellular
20. Kullak-Ublick, G. A. et al. Drug-induced liver injury: recent advances in sensors for toxic metals. Nat. Chem. Biol. 15, 540–548 (2019).
diagnosis and risk assessment. Gut 66, 1154–1164 (2017). 52. Rollins, N. J. et al. Inferring protein 3D structure from deep mutation scans.
21. Marcellin, P. & Kutala, B. K. Liver diseases: a major, neglected global public Nat. Genet. 51, 1170–1176 (2019).
health problem requiring urgent actions and large-scale screening. Liver Int. 53. Stiffler, M. A., Hekstra, D. R. & Ranganathan, R. Evolvability as a function of
38(Suppl 1), 2–6 (2018). purifying selection in TEM-1 β-lactamase. Cell 160, 882–892 (2015).
22. Ahmed, M. Acute cholangitis—an update. World J. Gastrointest. Pathophysiol. 54. Starr, T. N., Picton, L. K. & Thornton, J. W. Alternative evolutionary histories
9, 1–7 (2018). in the sequence space of an ancient protein. Nature 549, 409–413 (2017).
23. Schuppan, D. & Afdhal, N. H. Liver cirrhosis. Lancet 371, 838–851 (2008). 55. Russ, W. P. et al. An evolution-based model for designing chorismate mutase
24. Asrani, S. K., Devarbhavi, H., Eaton, J. & Kamath, P. S. Burden of liver enzymes. Science 369, 440–445 (2020).
diseases in the world. J. Hepatol. 70, 151–171 (2019). 56. Taketani, M. et al. Genetic circuit design automation for the gut resident
25. Giannini, E. G., Testa, R. & Savarino, V. Liver enzyme alteration: a guide for species Bacteroides thetaiotaomicron. Nat. Biotechnol. https://doi.org/10.1038/
clinicians. CMAJ 172, 367–379 (2005). s41587-020-0468-5 (2020).
26. McGill, M. R. The past and present of serum aminotransferases and the future 57. Tsuei, J., Chau, T., Mills, D. & Wan, Y.-J. Y. Bile acid dysregulation, gut
of liver injury biomarkers. EXCLI J. 15, 817–828 (2016). dysbiosis, and gastrointestinal cancer. Exp. Biol. Med. 239, 1489–1504 (2014).
27. Maldonado-Valderrama, J., Wilde, P., Macierzanka, A. & Mackie, A. The role 58. Ridlon, J. M., Kang, D. J., Hylemon, P. B. & Bajaj, J. S. Bile acids and the gut
of bile salts in digestion. Adv. Colloid Interface Sci. 165, 36–46 (2011). microbiome. Curr. Opin. Gastroenterol. 30, 332–338 (2014).
28. Pataia, V., Dixon, P. H. & Williamson, C. Pregnancy and bile acid disorders. 59. Liu, H.-X., Keane, R., Sheng, L. & Wan, Y.-J. Y. Implications of microbiota and
Am. J. Physiol. Gastrointest. Liver Physiol. 313, G1–G6 (2017). bile acid in liver injury and regeneration. J. Hepatol. 63, 1502–1510 (2015).
29. Gulamhusein, A. F. & Hirschfield, G. M. Primary biliary cholangitis: 60. Cao, H. et al. Secondary bile acid-induced dysbiosis promotes intestinal
pathogenesis and therapeutic opportunities. Nat. Rev. Gastroenterol. Hepatol. carcinogenesis. Int. J. Cancer 140, 2545–2556 (2017).
17, 93–110 (2020). 61. Nguyen, T. T., Ung, T. T., Kim, N. H. & Jung, Y. D. Role of bile acids in colon
30. Trottier, J. et al. Profiling circulating and urinary bile acids in patients with carcinogenesis. World J. Clin. Cases 6, 577–588 (2018).
biliary obstruction before and after biliary stenting. PLoS ONE 6, e22094 62. Chen, J., Thomsen, M. & Vitetta, L. Interaction of gut microbiota with
(2011). dysregulation of bile acids in the pathogenesis of nonalcoholic fatty liver
31. Sugita, T. et al. Analysis of the serum bile acid composition for differential disease and potential therapeutic implications of probiotics. J. Cell. Biochem.
diagnosis in patients with liver disease. Gastroenterol. Res. Pract. 2015, 717431 120, 2713–2720 (2019).
(2015). 63. Liu, Y., Rong, Z., Xiang, D., Zhang, C. & Liu, D. Detection technologies and
32. Luo, L. et al. Assessment of serum bile acid profiles as biomarkers of liver metabolic profiling of bile acids: a comprehensive review. Lipids Health Dis.
injury and liver disease in humans. PLoS ONE 13, e0193824 (2018). 17, 121 (2018).
33. Elsheashaey, A., Obada, M., Abdelsameea, E., Bayomy, M. F. F. & El-Said, H. 64. Stocker, J. et al. Development of a set of simple bacterial biosensors for
The role of serum bile acid profile in differentiation between nonalcoholic quantitative and rapid measurements of arsenite and arsenate in potable
fatty liver disease and chronic viral hepatitis. Egypt Liver J. 10, 50 (2020). water. Environ. Sci. Technol. 37, 4743–4750 (2003).
34. Kato, T., Yoneda, M., Nakamura, K. & Makino, I. Enzymatic determination of 65. Volpetti, F., Petrova, E. & Maerkl, S. J. A microfluidic biodisplay. ACS Synth.
serum 3 alpha-sulfated bile acids concentration with bile acid 3 alpha-sulfate Biol. 6, 1979–1987 (2017).
sulfohydrolase. Dig. Dis. Sci. 41, 1564–1570 (1996). 66. Din, M. O. et al. Synchronized cycles of bacterial lysis for in vivo delivery.
35. Xue, Y. et al. Redox pathway sensing bile salts activates virulence gene Nature 536, 81–85 (2016).
expression in Vibrio cholerae. Mol. Microbiol. 102, 909–924 (2016). 67. McNerney, M. P., Michel, C. L., Kishore, K., Standeven, J. & Styczynski, M. P.
36. Li, P. et al. Bile salt receptor complex activates a pathogenic type III secretion Dynamic and tunable metabolite control for robust minimal-equipment
system. Elife 5, e15718 (2016). assessment of serum zinc. Nat. Commun. 10, 5514 (2019).
37. Nelson, E. J., Harris, J. B., Morris, J. G. Jr, Calderwood, S. B. & Camilli, A. 68. Hui, C.-Y. et al. Genetic control of violacein biosynthesis to enable a pigment-
Cholera transmission: the host, pathogen and bacteriophage dynamic. Nat. based whole-cell lead biosensor. RSC Adv. 10, 28106–28113 (2020).
Rev. Microbiol. 7, 693–702 (2009). 69. Alnouti, Y. Bile Acid sulfation: a pathway of bile acid elimination and
38. Rivera-Cancel, G. & Orth, K. Biochemical basis for activation of virulence detoxification. Toxicol. Sci. 108, 225–246 (2009).
genes by bile salts in Vibrio parahaemolyticus. Gut Microbes 8, 366–373 70. Tazuke, Y. et al. A new enzymatic assay method of sulfated bile acids in urine.
(2017). 臨床化学 21, 249–258 (1992).
39. Childers, B. M. & Klose, K. E. Regulation of virulence in Vibrio cholerae: the 71. Gallagher, R. R., Patel, J. R., Interiano, A. L., Rovner, A. J. & Isaacs, F. J.
ToxR regulon. Future Microbiol. 2, 335–344 (2007). Multilayered genetic safeguards limit growth of microorganisms to defined
40. Pédelacq, J.-D., Cabantous, S., Tran, T., Terwilliger, T. C. & Waldo, G. S. environments. Nucleic Acids Res. 43, 1945–1954 (2015).
Engineering and characterization of a superfolder green fluorescent protein. 72. Lee, J. W., Chan, C. T. Y., Slomovic, S. & Collins, J. J. Next-generation
Nat. Biotechnol. 24, 79–88 (2005). biocontainment systems for engineered organisms. Nat. Chem. Biol. 14,
41. Yang, M. et al. Bile salt–induced intermolecular disulfide bond formation 530–537 (2018).
activates Vibrio cholerae virulence. Proc. Natl Acad. Sci. USA 110, 2348–2353 73. Bhatia, P. & Chugh, A. Synthetic biology based biosensors and the emerging
(2013). governance issues. Curr. Synth. Syst. Biol. 1, 2332–0737 (2013).

NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications 11

ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-021-25538-y

74. Simons, K. T., Bonneau, R., Ruczinski, I. & Baker, D. Ab initio protein structure Additional information
prediction of CASP III targets using ROSETTA. Proteins 3, 171–176 (1999). Supplementary information The online version contains supplementary material
75. Dey, K. K., Xie, D. & Stephens, M. A new sequence logo plot to highlight available at https://doi.org/10.1038/s41467-021-25538-y.
enrichment and depletion. BMC Bioinformatics 19, 473 (2018).
Correspondence and requests for materials should be addressed to J.B.

Acknowledgements Peer review information Nature Communications thanks the anonymous reviewer(s) for
We thank members of the synthetic biology group and the CBS for fruitful discussions. their contribution to the peer review of this work. Peer reviewer reports are available.
We are grateful to the patients for participating in this study and providing their samples,
and to the personnel of the Montpellier CHU hospital for collecting and preparing the Reprints and permission information is available at http://www.nature.com/reprints
samples. This work was supported by an ERC starting grant “COMPUCELL” to J.B. J.B.
also acknowledges support from the INSERM Atip-Avenir program and the Bettencourt- Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in
Schueller Foundation. The CBS acknowledges support from the French Infrastructure for published maps and institutional affiliations.
Integrated Structural Biology (FRISBI) ANR-10-INSB-05-01.

Author contributions Open Access This article is licensed under a Creative Commons
H.-J.C., A.Z., G.C. and J.B. designed experiments. H.-J.C., A.Z., P.L.V., E.F.-R. performed Attribution 4.0 International License, which permits use, sharing,
experiments. H.-J.C., A.Z., P.L.V., E.F.-R., G.C. and J.B. analyzed experiments. I.C., L.M., adaptation, distribution and reproduction in any medium or format, as long as you give
M.M. and G.-P.P. collected clinical samples and analyzed clinical data. J.G. performed appropriate credit to the original author(s) and the source, provide a link to the Creative
bioinformatic analysis and structural modeling of TcpP. H.-J.C., J.G. and M.C.-G. ana- Commons license, and indicate if changes were made. The images or other third party
lyzed the structural data. H.-J.C. and J.B. wrote the paper. All authors participated in material in this article are included in the article’s Creative Commons license, unless
elaborating the final version of the manuscript and approved it. indicated otherwise in a credit line to the material. If material is not included in the
article’s Creative Commons license and your intended use is not permitted by statutory
Competing interests regulation or exceeds the permitted use, you will need to obtain permission directly from
The authors declare the following competing interests: two patent applications related to the copyright holder. To view a copy of this license, visit http://creativecommons.org/
this work have been deposited. H.-J.C., J.G., M.C.G. and J.B. are named inventors on licenses/by/4.0/.
Patent application number PCT/EP2021/066224 (filled 16 June 2021), which is related to
the TcpP/TcpH receptor system described in this article. H.-J.C. and J.B. are named
inventors on the patent application EP21305165.9 (filled 8 Feb 2021), which is related to © The Author(s) 2021
the VtrA/VtrC receptor system described in this article. The remaining authors declare
no competing interests.

12 NATURE COMMUNICATIONS | (2021)12:5216 | https://doi.org/10.1038/s41467-021-25538-y | www.nature.com/naturecommunications