l-Serine Biosensor-Controlled Fermentative Production of l-Tryptophan Derivatives by Corynebacterium glutamicum

Simple Summary l-tryptophan is an amino acid found in proteins. Its derivatives, such as hydroxylated or halogenated l-tryptophans, find applications in the chemical and pharmaceutical industries, for example, in therapeutic peptides. Biotechnology provides a sustainable way for the production of l-tryptophan and its derivatives. In the final reaction of l-tryptophan biosynthesis in bacteria, such as Corynebacterium glutamicum, another amino acid, l-serine, is incorporated. Here, we show that C. glutamicum TrpB is able to convert indole derivatives, which were added to cells synthesizing l-serine, to the corresponding l-tryptophan derivatives. The gene trpB was expressed under the control of the l-serine-responsive transcriptional activator SerR in the C. glutamicum cells engineered for this fermentation process. Abstract l-Tryptophan derivatives, such as hydroxylated or halogenated l-tryptophans, are used in therapeutic peptides and agrochemicals and as precursors of bioactive compounds, such as serotonin. l-Tryptophan biosynthesis depends on another proteinogenic amino acid, l-serine, which is condensed with indole-3-glycerophosphate by tryptophan synthase. This enzyme is composed of the α-subunit TrpA, which catalyzes the retro-aldol cleavage of indole-3-glycerol phosphate, yielding glyceraldehyde-3-phosphate and indole, and the β-subunit TrpB that catalyzes the β-substitution reaction between indole and l-serine to water and l-tryptophan. TrpA is reported as an allosteric actuator, and its absence severely attenuates TrpB activity. In this study, however, we showed that Corynebacterium glutamicum TrpB is catalytically active in the absence of TrpA. Overexpression of C. glutamicum trpB in a trpBA double deletion mutant supported growth in minimal medium only when exogenously added indole was taken up into the cell and condensed with intracellularly synthesized l-serine. The fluorescence reporter gene of an l-serine biosensor, which was based on the endogenous transcriptional activator SerR and its target promoter PserE, was replaced by trpB. This allowed for l-serine-dependent expression of trpB in an l-serine-producing strain lacking TrpA. Upon feeding of the respective indole derivatives, this strain produced the l-tryptophan derivatives 5-hydroxytryptophan, 7-bromotryptophan, and 5-fluorotryptophan.


Introduction
Amino acids and their derivatives can be produced by chemical synthesis, enzyme catalysis, or de novo by fermentation [1]. Fermentative production of the food and feed amino acids L-glutamate and L-lysine is a multi-billion USD market. The annual production of the amino acids L-tryptophan and L-serine amounts to 41,000 and 350 tons, respectively [2]. In nature, L-tryptophan is synthesized from L-serine, and structural analyses of aminoacyl-tRNA synthetases indicate that the fixation of L-tryptophan (and L-tyrosine) in gene. Binder et al., 2012 showed that pSenSer is a useful tool for visualizing intracellular L-serine concentrations [24]. Zhang et al., 2018 employed this biosensor's fluorescence output to isolate L-serine-overproducing C. glutamicum cells after atmospheric and room temperature plasma mutagenesis [11].
An L-tryptophan-responsive biosensor has not yet been developed for C. glutamicum. In C. glutamicum, the transcriptional repressor LtbR binds to a 12-bp motif between the −10 promoter region and the ribosome binding site of the trpL gene and prevents transcription of the trp operon [25]. Albeit lacking a TrpR homolog, C. glutamicum has a putative operator sequence with similarity to the operator DNA binding sequence of the TrpR repressor of E. coli and was detected upstream of the −10 promoter region of the trp operon [26,27]. Transcription of the trp operon is controlled by attenuation and deletion of the leader peptide gene trpL and improved overproduction of L-tryptophan [28] and aromatic aldehydes [29]. Several L-tryptophan-producing C. glutamicum strains have been developed [30][31][32]. Typically, genes of the common chorismate pathway and the dedicated L-tryptophan branch are overexpressed, and key enzymes are feedback-desensitized variants, combined with an improved supply of precursors from the central carbon metabolism and with enhanced redox cofactor regeneration.
In this study, the production of L-tryptophan and its derivatives was based only on the reaction catalyzed by tryptophan synthase β-subunit TrpB. The β-reaction has already been employed for the preparation of a wide range of L-tryptophan derivatives [33][34][35]. As a single-catalyst platform, L-tryptophan analogue conversion from the widely available indole analogues is accessible [36][37][38]. Here, it was found that the catalytic activity of C. glutamicum TrpB is not attenuated in the absence of its allosteric actuator TrpA and readily reacts with various indoles and L-serine. Since L-serine is a co-substrate of the TrpB reaction and its availability has to be on par with the demand for L-tryptophan biosynthesis, a C. glutamicum L-serine-overproducing strain [16,39] was endowed with the L-serine biosensor to drive the expression of trpB. This enabled the production of L-tryptophan derivatives from intracellularly synthesized L-serine and exogenously added indole derivatives.

Strains and Culture Conditions
The genotypes of the bacterial strains used in this study are listed in Table 1. Escherichia coli DH5α [40] was used as a cloning host and for plasmid propagation. C. glutamicum ATCC 13032 was used as a host organism for evaluation of L-serine-responsive biosensors and production of L-tryptophan derivatives. Luria-Bertani medium was used as the standard medium for E. coli, brain heart infusion (BHI) medium (Difco) as the complex medium, and CGXII [41] as minimal medium for C. glutamicum, with 40 g L −1 glucose as carbon source. Whenever appropriate, the following were supplemented in the medium: kanamycin (25 µg mL −1 ), tetracycline (5 µg mL −1 ), and 1 mM IPTG. E. coli and C. glutamicum were grown at 37 • C and 30 • C, respectively, in baffled shake flasks on a rotary shaker (160 rpm or 120 rpm). For C. glutamicum growth experiments, overnight pre-cultures were prepared as described above, harvested, and washed with TN buffer pH 6.3 (50 mM Tris-HCl, 50 mM NaCl), and the main medium is inoculated to an optical density at 600 nm (OD 600 ) of 1. Growth and production experiments with the C. glutamicum SER1 and SER2 were grown in CGXII minimal medium under folate-starvation condition as seed cultures and the main culture in CGXII minimal medium supplemented with 0.1 mM folate [16]. Growth medium for SER2 strain was supplemented with 0.25 g L −1 (1.22 mM) L-tryptophan for growth. Growth of C. glutamicum cultivated in 500 mL baffled shake flasks was monitored by measuring OD 600 using a V-1200 spectrophotometer (VWR, Radnor, PA, USA). HT cultivations were performed in 48-well Flowerplates with the BioLector cultivation system (m2p-labs GmbH, Baesweiler, Germany) [42,43]. Formation of biomass was recorded as a backscatter light intensity at 620 nm wavelength, while the fluorescence of the yellow

Molecular Genetic Techniques, Plasmids, and Strain Construction
Standard molecular genetic techniques were performed as described elsewhere [46]. Chromosomal deletions were performed using the non-replicable suicide vector pK19mobsacB [47] in a two-step homologous recombination [41]. The genomic regions flanking the respective gene for homologous recombination and, unless otherwise stated, the genes for cloning were amplified from C. glutamicum ATCC13032 gDNA using ALLin HiFi DNA Polymerase according to manufacturer (highQu GmbH, Kraichtal, Germany). Cloning of the amplified PCR products was performed using restriction enzyme digestion and Gibson assembly [48]. The oligonucleotides and plasmids used in this study are listed in Tables 2 and 3, respectively.   Table 2. Oligonucleotides used in this study.

Name
Oligonucleotide Sequence (5 to 3 ) L-serine-responsive biosensors were assembled in pGP21 plasmid; pGP21 was designed based on the vector pGP2 [49]. The plasmid pGP2 was modified to contain the T7te and rrnB T1 terminator for the divergent expression of the biosensor parts (i.e., transcription factor and reporter gene/biosynthetic trpB gene). To this end, pGP21-F and pGP21-R primers ( Table 2) were used to amplify the T7te and rrnB T1 terminator from the pRG-Duet2 [50] vector. The regulatory unit, serR (NCgl0581), and the intergenic region to serE were amplified using the primers serR-F and serE-R, while primer set mVen-F and mVen-R were used to amplify the reporter gene mVenus NB from pEB1-mVenus NB plasmid (Addgene plasmid # 103986). These DNA fragments were cloned into BamHI-digested pGP21. Km R , oriV Ec , sacB, lacZα; E. coli-C. glutamicum shuttle vector for construction of insertion and deletion mutants in C. glutamicum [47] pK19mobsacB ∆serA pK19mobsacB with a construct for in-frame deletion of serA This work pK19mobsacB ∆trpBA pK19mobsacB with a construct for in-frame deletion of trpBA This work pEKEx3 Spec R , P tac lacI q , pBL1 oriV Cg , sacB, lacZα; E. coli-C. glutamicum expression vector [52] pEKEx3-trpBA pEKEx3 overexpressing trpBA This work pEKEx3-trpB pEKEx3 overexpressing trpB This work To introduce single-point mutation into pSenSer1, site-directed mutagenesis (SDM) was performed and primers were designed according to Ref. [53]. A PCR reaction of 50 µL was prepared to contain 3 units (PfuTurbo DNA polymerase (Agilent Technologies, Santa Clara, CA, USA), 10 ng plasmid, 125 ng forward and reverse primers, 200 µM dNTPs, and Pfu DNA polymerase reaction buffer). To construct pSenSer1-serE'-26 aa and pSenSer2-serE'-26, primer sets SDM1-F/SDM1-R and SDM2-F/SDM2-R were used, respectively. To construct pSdS-trpB, pSenSer1 biosensor plasmid was double-digested with EcoRV and BamHI to remove mVenus. The trpB gene was then amplified with primers trpB-F and trpB-R and subsequently ligated to the linearized pSenSer1 by Gibson Assembly.

Quantification of L-Serine and Aromatic Compounds
Extracellular L-serine was quantified by high-performance liquid chromatography (HPLC) (1200 series, Agilent Technologies Deutschland GmbH, Böblingen, Germany). The culture supernatants were collected at specific time points and centrifuged (20,200× g, 10 min 4 • C) for analytical analysis. Samples for L-serine analysis were derivatized with ortho-phthaldialdehyde (OPA) (Schneider, Niermann, and Wendisch 2011). The amino acid separation was performed by a pre-column (LiChrospher 100 RP18 EC-5 µ (40 mm × 4 mm), CS-Chromatographie Service GmbH, Langerwehe, Germany) and a column (LiChrospher 100 RP18 EC-5 µ (125 mm × 4 mm), CS-Chromatographie Service GmbH, Langerwehe, Germany). Detection was carried out with a fluorescence detector (FLD G1321 A, 1200 series, Agilent Technologies) with the excitation and emission wavelengths of 230 nm and 450 nm, respectively. Aromatic compounds were analyzed from the culture supernatants according to Ref. [54]. A mobile phase gradient of buffer A (0.1% trifluoroacetic acid dissolved in water) and buffer B (acetonitrile) was used, and detection was performed with Diode Array Detector (DAD G1315B, 1200 series, Agilent Technologies) with scanning window of 210 nm to 330 nm.
For the quantification of intracellular L-serine, 1 mL of broth for each strain grown in minimal medium was collected at indicated time points. It was then centrifuged for 10 min at 20,200× g, 4 • C. Pellets were treated with 5% HClO 4 in an ice bath for 30 min. After centrifuging at 20,200× g for 10 min and 4 • C, the supernatant was neutralized with K 2 CO 3 solution and centrifuged again at 13,000 rpm for 10 min, 4 • C. The supernatant was then ready for amino acid quantification or to be stored at −20 • C [55]. The biomass concentration was determined according to the correlation CDW = 0.353 OD [56].

Design and Construction of an L-Serine-Responsive Biosensor
To design a biosensor that responds to elevated intracellular L-serine concentrations, the C. glutamicum native SerR-P serE biosensor first described and used by Binder et al., 2012 as a tool for the visualization of intracellular L-serine [24] was adapted. The EYFP reporter gene was replaced with the mVenus gene since mVenus has been reported to be a rapidly maturing monomer with moderate acid sensitivity as compared to EYFP [57]. The resulting biosensor plasmid called pSenSer1 (Figure 1) was transformed into the SER1 strain, a strain secreting L-serine under the folate-starvation condition [16], to evaluate the capability of the biosensor in monitoring the intracellular accumulation of its effector molecule. gene was replaced with the mVenus gene since mVenus has been reported to be a rapidly maturing monomer with moderate acid sensitivity as compared to EYFP [57]. The resulting biosensor plasmid called pSenSer1 ( Figure 1) was transformed into the SER1 strain, a strain secreting L-serine under the folate-starvation condition [16], to evaluate the capability of the biosensor in monitoring the intracellular accumulation of its effector molecule.  Strains C. glutamicum WT and SER1 were transformed with either the empty vector pGP21 or biosensor plasmid pSenSer1 and were grown in CGXII minimal medium supplemented with 0.1 mM folate in the BioLector system to monitor fluorescence, growth, and L-serine production. To determine the correlation between the fluorescence output signal and effector concentration, intracellular L-serine concentrations at certain time points were measured ( Figure 2A). As expected, no discernable fluorescence signal was observed using WT strains and SER1 (pGP21), while SER1 harboring pSenSer1 exhibited fluorescence ( Figure 2A). Measurable amounts of L-serine were detected in the cytosol of both SER1 (pGP21) and SER1 (pSenSer1), while L-serine was not detected in the WT strains (at concentration < 0.12 mM). Interestingly, until 5 h of cultivation, only autofluorescence was detected in the SER1 (pSenSer1) strain and only at 10 h and later, a significant fluorescence was recorded. This may indicate that it takes time for enough L-serine to induce mVenus gene expression and, subsequently, to accumulate mVenus fluorescence reporter protein.
The highest fluorescence was recorded at 72 h, which also corresponded to the highest intracellular L-serine measured. Similarly, Stolz et al. also observed a graded rise in L-serine production over time when the SER1 strain was cultivated under the folate-limiting condition in shake flasks [16]. There was no significant difference in the production concentration of the two SER1 strains (i.e., SER1 empty vector strain and the sensor strain produced 60.6 ± 5.3 and 61.9 ± 0.5 mM L-serine, respectively, in BioLector cultivation), suggesting that L-serine production is not affected by the biosensor plasmid ( Figure 2B). signal and effector concentration, intracellular L-serine concentrations at certain time points were measured ( Figure 2A). As expected, no discernable fluorescence signal was observed using WT strains and SER1 (pGP21), while SER1 harboring pSenSer1 exhibited fluorescence ( Figure 2A). Measurable amounts of L-serine were detected in the cytosol of both SER1 (pGP21) and SER1 (pSenSer1), while L-serine was not detected in the WT strains (at concentration ˂ 0.12 mM). Interestingly, until 5 h of cultivation, only autofluorescence was detected in the SER1 (pSenSer1) strain and only at 10 h and later, a significant fluorescence was recorded. This may indicate that it takes time for enough L-serine to induce mVenus gene expression and, subsequently, to accumulate mVenus fluorescence reporter protein. The highest fluorescence was recorded at 72 h, which also corresponded to the highest intracellular L-serine measured. Similarly, Stolz et al. also observed a graded rise in L-serine production over time when the SER1 strain was cultivated under the folatelimiting condition in shake flasks [16]. There was no significant difference in the production concentration of the two SER1 strains (i.e., SER1 empty vector strain and the sensor strain produced 60.6 ± 5.3 and 61.9 ± 0.5 mM L-serine, respectively, in BioLector cultivation), suggesting that L-serine production is not affected by the biosensor plasmid ( Figure  2B). Since the applicability of a biosensor requires that the relationship between the effector molecule and the reporter output is defined, the dose-response of pSenSer1 was analyzed in an L-serine auxotrophic mutant that was constructed by the deletion of the native serA gene in the wild-type C. glutamicum. Strain C. glutamicum ATCC 13032 ΔserA was, therefore, unable to synthesize L-serine; thus, the exogenous addition of L-serine to the medium could be used to modulate the intracellular L-serine concentration detectable by pSenSer1. As shown in Figure 3, a clear increase in pSenSer1 output was measured with Since the applicability of a biosensor requires that the relationship between the effector molecule and the reporter output is defined, the dose-response of pSenSer1 was analyzed in an L-serine auxotrophic mutant that was constructed by the deletion of the native serA gene in the wild-type C. glutamicum. Strain C. glutamicum ATCC 13032 ∆serA was, therefore, unable to synthesize L-serine; thus, the exogenous addition of L-serine to the medium could be used to modulate the intracellular L-serine concentration detectable by pSenSer1. As shown in Figure 3, a clear increase in pSenSer1 output was measured with increasing concentrations of L-serine supplemented. A plateau was not reached with the highest L-serine concentration tested, i.e., 200 mM.
Two variants of the pSenSer1 biosensor were constructed and tested. Plasmid pSenSer1 contains a transcriptional fusion of the promoter of serE with the promoterless mVenus gene. The construct allows for transcription of an mRNA comprised of 79 base pairs (bp) of serE, a linker sequence of 24 bp, and 717 bp of the mVenus coding sequence [24]. Translation of this mRNA may, thus, not only lead to a full-length mVenus fluorescence reporter protein but, in addition, also to a peptide initiating at the translational start codon of serE. To test if the putative synthesis of this peptide extends well into the mVenus gene, but in a different open reading frame, the last nucleotide of the 79 bp serE sequence was deleted in pSenSer1-serE'-26aa. By this manner, the resulting mRNA only codes for a defined short SerE peptide of 26 amino acids in addition to the full-length mVenus protein ( Figure 1B). In the second variant, pSenSer2-serE'-26aa ( Figure 1C), the translational start codon of the regulatory gene serR was changed from the less-preferred GTG to the canonical ATG. This change was expected to increase the SerR regulator protein concentration, which may positively affect the biosensor amplitude.  Two variants of the pSenSer1 biosensor were constructed and tested. Plasmid pSenSer1 contains a transcriptional fusion of the promoter of serE with the promoterless mVenus gene. The construct allows for transcription of an mRNA comprised of 79 base pairs (bp) of serE, a linker sequence of 24 bp, and 717 bp of the mVenus coding sequence [24]. Translation of this mRNA may, thus, not only lead to a full-length mVenus fluorescence reporter protein but, in addition, also to a peptide initiating at the translational start codon of serE. To test if the putative synthesis of this peptide extends well into the mVenus gene, but in a different open reading frame, the last nucleotide of the 79 bp serE sequence was deleted in pSenSer1-serE'-26aa. By this manner, the resulting mRNA only codes for a defined short SerE peptide of 26 amino acids in addition to the full-length mVenus protein ( Figure 1B). In the second variant, pSenSer2-serE'-26aa ( Figure 1C), the translational start codon of the regulatory gene serR was changed from the less-preferred GTG to the canonical ATG. This change was expected to increase the SerR regulator protein concentration, which may positively affect the biosensor amplitude.
The variants pSenSer1-serE'-26aa and pSenSer2-serE'-26aa were used to transform the L-serine auxotrophic strain C. glutamicum ATCC 13032 ΔserA and used to compare with ATCC 13032 ΔserA (pSenSer1). A sigmoidal dose-response curve was observed for all three strains (Figure 3). The finding of little fluorescence when 0.5 to 25 mM of L-serine was supplemented may be due to co-metabolization of L-serine with glucose at high utilization rates, as observed previously [58]. The biosensor output was highest with the highest tested concentration of L-serine (200 mM). Comparing the three biosensor variants revealed that the signal amplitude was highest with pSenSer1, followed by pSenSer1-serE'-26aa before pSenSer2-serE'-26aa. Thus, while all three biosensors were suitable, pSenSer1 was chosen for experiments aiming at dynamic expression regulation in response to intracellular L-serine. The variants pSenSer1-serE'-26aa and pSenSer2-serE'-26aa were used to transform the L-serine auxotrophic strain C. glutamicum ATCC 13032 ∆serA and used to compare with ATCC 13032 ∆serA (pSenSer1). A sigmoidal dose-response curve was observed for all three strains ( Figure 3). The finding of little fluorescence when 0.5 to 25 mM of L-serine was supplemented may be due to co-metabolization of L-serine with glucose at high utilization rates, as observed previously [58]. The biosensor output was highest with the highest tested concentration of L-serine (200 mM). Comparing the three biosensor variants revealed that the signal amplitude was highest with pSenSer1, followed by pSenSer1-serE'-26aa before pSenSer2-serE'-26aa. Thus, while all three biosensors were suitable, pSenSer1 was chosen for experiments aiming at dynamic expression regulation in response to intracellular L-serine.

Construction of an Indole-Essential C. glutamicum Strain
L-Serine is condensed with IGP to yield L-tryptophan in the ultimate reaction of L-tryptophan biosynthesis. More specifically, tryptophan synthase subunit A (TrpA) cleaves IGP to yield G3P and indole, and wherein the latter is condensed with L-serine in a β-substitution reaction, releasing water by subunit B (TrpB) (Scheme 1).
While extensive studies on the mechanism of the enzymatic reaction of tryptophan synthase have shown that the two subunits of the complex have low catalytic efficiencies outside the complex [59], it was tested if the absence of TrpA can be compensated for by extracellular addition of indole. To this end, the genes for both subunits trpBA were deleted in a base C. glutamicum C1* strain, yielding an L-tryptophan auxotrophic strain. C1* was used as a base strain to avoid possible effects of L-serine overproduction in strain SER2. This strain was transformed with the empty IPTG-inducible expression plasmid pEKEx3 (negative control), pEKEx3-trpBA (positive control of genetic complementation), and with pEKEx3-trpB. As expected, all strains grew when supplemented with 1 mM L-tryptophan, whereas only the genetic complementation strain C. glutamicum C1* ∆trpBA (pEKEx3-trpBA) could grow without L-tryptophan supplementation ( Figure 4). Notably, strain C1* ∆trpBA (pEKEx3-trpB), which lacks trpA, could grow when induced with 1 mM IPTG and supplemented with 1 mM indole. This indole-essential growth showed that catalysis by C. glutamicum TrpB-even outside of the native tryptophan synthase complex with TrpA-is sufficient to provide L-tryptophan for growth.

Construction of an Indole-Essential C. glutamicum Strain
L-Serine is condensed with IGP to yield L-tryptophan in the ultimate reaction of Ltryptophan biosynthesis. More specifically, tryptophan synthase subunit A (TrpA) cleaves IGP to yield G3P and indole, and wherein the latter is condensed with L-serine in a βsubstitution reaction, releasing water by subunit B (TrpB) (Scheme 1). Scheme 1. Tryptophan synthase enzymatic reaction. TrpA (top, in gray) catalyzes the retro-aldol cleavage on IGP (indole-3-glycerol phosphate) to G3P (glyceraldehyde-3-phosphate). Indole diffuses to the TrpB subunit (bottom, in black), which performs the β-substitution reaction between indole and L-serine, releasing water (not shown) and L-tryptophan. Decorations in the indole moiety are shown in red.
While extensive studies on the mechanism of the enzymatic reaction of tryptophan synthase have shown that the two subunits of the complex have low catalytic efficiencies outside the complex [59], it was tested if the absence of TrpA can be compensated for by extracellular addition of indole. To this end, the genes for both subunits trpBA were deleted in a base C. glutamicum C1* strain, yielding an L-tryptophan auxotrophic strain. C1* was used as a base strain to avoid possible effects of L-serine overproduction in strain SER2. This strain was transformed with the empty IPTG-inducible expression plasmid pEKEx3 (negative control), pEKEx3-trpBA (positive control of genetic complementation), and with pEKEx3-trpB. As expected, all strains grew when supplemented with 1 mM Ltryptophan, whereas only the genetic complementation strain C. glutamicum C1* trpBA (pEKEx3-trpBA) could grow without L-tryptophan supplementation ( Figure 4). Notably, strain C1* trpBA (pEKEx3-trpB), which lacks trpA, could grow when induced with 1 mM IPTG and supplemented with 1 mM indole. This indole-essential growth showed that catalysis by C. glutamicum TrpB-even outside of the native tryptophan synthase complex with TrpA-is sufficient to provide L-tryptophan for growth.

On-Demand Conversion of Indole and L-Serine to L-Tryptophan via L-Serine-Responsive Biosensor
After establishing that C. glutamicum TrpB can salvage the growth of an L-trypto-

On-Demand Conversion of Indole and L-Serine to L-Tryptophan via L-Serine-Responsive Biosensor
After establishing that C. glutamicum TrpB can salvage the growth of an L-tryptophanauxotrophic mutant strain upon trpB overexpression using an IPTG-inducible plasmid and indole supplementation, trpB was cloned into pSenSer1 plasmid for L-serine-controlled expression, replacing mVenus ( Figure 5A). The constructed pSdS-trpB was used in the following experiments to dynamically control the expression of the native trpB gene according to intracellular L-serine availability. In comparison, an IPTG-inducible trpB expression vector (pEKEx3-trpB) was utilized. When the switch-plasmid pSdS-trpB was used to transform L-serine-producing strain SER2 (Table 1), supplementation of 1 mM indole was sufficient to allow for growth, while the empty vector carrying negative control strain SER2 (pGP21) could not grow ( Figure 5B). Furthermore, there was no observable difference in growth behavior between SER2 carrying the switch-plasmid pSdS-trpB for dynamic regulation or the fully induced IPTG-regulated plasmid pEKEx3-trpB. This indicated that strain SER2 provided enough L-serine and that the switch-plasmid pSdS-trpB was sufficient to promote L-tryptophan biosynthesis from exogenous indole and intracellular L-serine. Since both strains grew comparably ( Figure 5B), we continued to use SER2 (pSdS-trpB) for further experiments.

Fermentative Production of L-Tryptophan Derivatives by C. glutamicum SER2 (pSdS-trpB) from Indole Derivatives
In principle, the tryptophan synthase β-subunit TrpB is able to react with many indole analogues [60] in the absence of TrpA (Scheme 1). After having shown that C. glutamicum TrpB efficiently converts indole to L-tryptophan in the absence of TrpA, it was tested if C. glutamicum TrpB is promiscuous with respect to the substrate indole, i.e., whether it also accepts indole analogues.

Fermentative Production of L-Tryptophan Derivatives by C. glutamicum SER2 (pSdS-trpB) from Indole Derivatives
In principle, the tryptophan synthase β-subunit TrpB is able to react with many indole analogues [60] in the absence of TrpA (Scheme 1). After having shown that C. glutamicum TrpB efficiently converts indole to L-tryptophan in the absence of TrpA, it was tested if C. glutamicum TrpB is promiscuous with respect to the substrate indole, i.e., whether it also accepts indole analogues.
To this end, either 5-fluoroindole (5-FI), 7-bromoindole (7-BrI), or 5-hydroxyindole (5-HI) were added exogenously as substrates to C. glutamicum SER2 (pSdS-trpB) cells growing in minimal medium supplemented with 0.25 g L −1 (1.22 mM) L-tryptophan. Growth was affected by 1 mM 5-FI and 1 mM 7-BrI, but not by 3 mM 5-HI ( Figures 6A  and 7A). Furthermore, growth was also inhibited by the product 5-HTP (Supplementary Figure S1), as was shown previously for 7-bromo-L-tryptophan (7-BrTP) [61]. In all the cultivations, about 50 to 60 mM of L-serine was produced, and the glucose was completely consumed after 48 h (data not shown). The conversion of the indole derivatives to the respective L-tryptophan derivatives 5-fluoro-L-tryptophan (5-FTP), 7-BrTP, or 5-hydroxy-Ltryptophan (5-HTP) was followed by HPLC. Quantification of 5-HI, 5-HTP, 7-BrI, 7-BrTP, and 5-FI was performed using commercially obtained standards. The formation of 5-FTP was shown by LC-MS analysis ( Figure 7B).    Taken together, the native TrpB of C. glutamicum was shown to be active not without TrpA as a stand-alone enzyme, but also to accept indole analogues as subst Taken together, the native TrpB of C. glutamicum was shown to be active not only without TrpA as a stand-alone enzyme, but also to accept indole analogues as substrates in addition to its native substrate indole. On-demand expression of trpB gene was achieved by placing it under the control of an L-serine-responsive biosensor, and this strategy has been demonstrated to be useful in producing L-tryptophan derivatives of importance in industry, such as 7-BrTP, 5-HTP, and 5-FTP.

Discussion
The production of L-tryptophan derivatives 5-FTP, 7-BrTP, and 5-HTP was achieved by applying the L-serine-dependent expression of the native trpB from C. glutamicum in an L-serine-overproducing C. glutamicum strain. L-tryptophan scaffolds, L-tryptophan derivatives with decorations in their indole moiety, are widespread in bioactive natural products. The synthesis of L-tryptophan analogues such as the ones produced in this study is of interest because such functionalized derivatives lead to modified properties of biological effects that may subsequently lead to enhanced and/or novel pharmaceutically active compounds. Due to the inherent reactivity and oxidative nature of the indole ring of L-tryptophan, L-tryptophan stimulates oxidative reactions, electrophilic aromatic substitutions, and alkylation reactions [62]. However, in order to achieve L-tryptophan diversification, several different enzymes would be needed for each modification reaction, complicating the process.
Employing TrpB as a versatile catalyst to facilitate the production of L-tryptophan analogues offers a simplified route as compared to utilizing different enzymes to introduce the functional group in L-tryptophan. The unprecedented discovery that C. glutamicum TrpB efficiently converts indole to L-tryptophan even in the absence of its allosteric activator, TrpA, is of significance because it revealed the possibility of TrpB as being a promiscuous enzyme that accepts indole analogues. In recent years, the significance of tryptophan synthase as a biocatalytic platform has been largely investigated by the group of 2018 Nobel Prize Laureate Frances Arnold. Only the β-subunit of the tryptophan synthase catalyzes the formation of L-tryptophan derivatives. In order to release the allosteric modulation of TrpB from the α-subunit, directed evolution on TrpB was applied to identify mutations that emulate the effect of TrpA binding and to allow the high catalytic activity of TrpB in isolation. As a result, TrpB progenies were developed that have efficient stand-alone activity not only with indole as a substrate but also allowed systematic expansion to indole analogues [60]. TrpBs from thermophilic and hyperthermophilic organisms have been extensively subjected to directed evolution to recapitulate allosteric activation in the absence of TrpA. TrpB from thermophilic organisms [EC 4.2.1.122] has been described to possess indole-salvaging activity and has a lower Km for indoles compared to the TrpB found in mesophilic organisms [EC 4.2.1.20] (https://www.brenda-enzymes.org/enzyme.php?ecno=4.2.1.122, accessed on 4 April 2022). Besides the inherent property of thermophilic TrpB, choosing TrpB from organisms that thrive at relatively high temperatures is strategically advantageous because its thermostability simplifies enzyme purification from a mesophilic E. coli host and, subsequently, allows screening at elevated temperatures [63]. For instance, a TrpB mutant from Pyrococcus furiosus Pf TrpB 0B2 with 83-fold improved catalytic efficiency compared to the wild-type was obtained after the introduction of mutations that were found to accelerate catalysis through the same mechanism as TrpA effector binding [63,64].
Our approach also yielded 5-FTP. Fluorination at different positions of the indole ring of L-tryptophan results in altered intrinsic properties due to the high strength and polarization of the C-F bond. Fluorinated tryptophan analogues are usually used as "intrinsic" probes of protein local structures; for instance, the absorbance spectra of 5-FTP extend to longer wavelengths than L-tryptophan itself and, hence, enable preferential excitation of the fluorinated analogue in fluorescence experiments [65]. To the best of our knowledge, no report has been made on the occurrence of a nucleophilic halogenase that specifically fluorinates L-tryptophan. There is, however, a report on a fluorinating enzyme 5 -fluoro-5 deoxyadenosine synthase from the soil bacterium Streptomyces cattleya [66], which participates in the biosynthesis of fluoroacetate and fluorothreonine [67]. On the other hand, brominated compounds, such as 7-BrTP, are precursors of TMC-95A, and it may be a promising candidate for tumor and inflammation therapies, because it is biologically active as a protease inhibitor against the chymotrypsin-like, trypsin-like, and peptidylglutamyl-peptide-hydrolyzing activities of the 20S proteasome of eukaryotic cells [68,69]. The fermentation process for 7-BrTP production based on halogenase enzymes RebH and RebF from Lechevalieria aerocolonigenes in a recombinant C. glutamicum strain has been achieved before [61].
A more simplistic and convenient approach in incorporating a halogen into L-tryptophan has been previously described by expressing the tryptophan synthase from Salmonella enterica on a high copy number plasmid in E. coli. Treating the cell lysates as synthetic reagents and mixing them with haloindole and serine in a buffer resulted in successful syntheses of 4-, 5-, 6-, or 7-halo L-tryptophan [34]. Notably, the highest yield was afforded by 5-FTP (63% after one cycle with enzyme and improved to a 100% yield after two further reaction cycles) compared to all the other analogues tested. In our process, no 5-FI was detected in the samples after 48 h, unlike in the case of 5-HI and 7-BrI (Figures 6 and 7). It was, therefore, inferred that the conversion yield to 5-FTP, particularly in our biotransformation process, is high. Goss and Newill 2006 rationalized that the shorter the carbon-halogen bond, the greater the ease of conversion, as was observed in the trend 5-FI > 5-ClI > 5-BrI, and that comparatively lower yields could be postulated for 4-or 7-substituted indoles due to steric hindrance [34]. This may also be the explanation why, in this study, less than 50% biotransformation for both 5-HI and 7-BrI was observed and, although quantification was not possible for 5-FTP due to a lack of standard, the 5-FI peak was gone after 48 h. The process described in this study does not require prior cell lysis, reducing manual labor, and it decreases the cost because L-serine is already produced de novo from cheap sugar.
5-HTP is a metabolic intermediate in the biosynthesis of the neurotransmitter serotonin. It is prepared either by extraction from Ghana seeds or through the enzymatic reaction catalyzed by L-tryptophan-5-hydroxylase with L-tryptophan as the direct precursor. However, the extraction method generates pollutants to the environment because of the organic solvents used, and the T5H for biosynthesis is unstable when expressed in microorganisms and requires a special cofactor 5,6,7,8-tetrahydrobiopterin that must be generated through additional enzymatic reactions [70]. Recently, directed evolution was performed on E. coli tryptophan synthase to generate a mutant with increased catalytic activity towards 5-HTP conversion from L-serine and 5-HI, reaching a yield of 86.7% [71]. In this study, 42 ± 0.91% conversion was observed for 5-HTP. In the future, directed evolution may also be applied to C. glutamicum TrpB to improve its catalytic efficiency towards 5-HI.
As far as we are aware, this is the first study using an L-serine-responsive biosensor for an on-demand production platform for L-tryptophan derivatives through dynamic expression control of the biosynthetic gene trpB. Moreover, we provide a first proof of concept that C. glutamicum TrpB accepts not only indole but also indole analogues as substrates. Since the C-C-bond-forming reaction for the formation of L-tryptophan analogues is only catalyzed by the β-subunit of the tryptophan synthase, TrpA was bypassed, and it was shown that C. glutamicum TrpB is catalytically efficient as a stand-alone enzyme for the conversion of L-serine and indole (derivatives) to L-tryptophan (derivatives). Until now, it has been widely described that native TrpBs lose most of their activity in the absence of allosteric activation from their corresponding TrpAs [72,73], and both TrpA and TrpB had to be present for the activity of native, unevolved tryptophan synthase towards the synthesis of L-tryptophan derivatives. However, the likelihood that C. glutamicum TrpB activity can be further increased upon complexing with TrpA cannot be disregarded, and further detailed enzymatic analysis may have to be performed in the future.
In this study, the ability of the SerR-based L-serine biosensor for dynamic pathway control was harnessed in order to produce L-tryptophan analogues on-demand in an L-serine C. glutamicum overproducer. The specificity of the L-serine biosensor to cytosolic L-serine was shown before by Binder et al., 2012 in a peptide-dose assay wherein the addition of Ser3-OH resulted in fluorescent cultures, while this was not the case for either Ala-Ala or Thr-Ala [24]. The sensor transfer curve or the characterization of the output response of pSenSer1 and its derivatives with its cognate effector L-serine was achieved by quantifying the fluorescent output response in relation to the supplied L-serine in an L-serine-auxotrophic mutant. L-serine is known to be channeled into the central metabolism by several organisms [74]; hence, the fluorescence output could only be related to free L-serine inside the cells. Netzer et al., 2004 described that the co-utilization of glucose and L-serine in C. glutamicum has three distinct phases: first, cellular growth occurs during the co-metabolism of both glucose and L-serine; second, the cells enter a non-growth phase while L-serine is still being utilized; and, last, the cells stop utilizing L-serine [58]. The intracellular fate of L-serine during the second phase was determined via isotopomer labelling, and it was found that L-serine is either utilized for the synthesis of other cellular metabolites via pyruvate or oxidized to produce energy for maintenance. Similar to the described second phase that was observed in this study, the maximum OD 600 was reached at 25 mM L-serine, but a significant signal-to-noise fluorescence ratio was only observed at 50 mM L-serine. From 50 mM to 200 mM L-serine, the biosensor showed a gradedfluorescence output with increased L-serine supplementation. In other words, mVenus is expressed when the cells stopped growing and L-serine is still consumed but no longer utilized for growth. This makes the system suitable for the dynamic pathway control as it allows sensing the significantly elevated intracellular L-serine level and, subsequently, causes the expression of the output gene.
The SerR transcriptional activator employed in pSenSer1 belongs to the family of LysR-type transcriptional regulators (LTTRs). LTTRs are highly conserved and ubiquitous among bacteria and described as global transcriptional regulators, acting as either activators or repressors of single genes or genes in an operon. They were also observed to be often divergently transcribed to their target gene, but could also be located elsewhere on the bacterial chromosome [75]. While the exact DNA-binding sequence of SerR upstream of serE is unknown, LTTR broadly binds at −40 to −20 bp as the activation site of its cognate regulated gene [75]. Nevertheless, an electrophoretic mobility shift assay (EMSA) confirmed that SerR binds to the upstream region of serE [12]. Knowledge on the accurate transcription factor binding sequence is useful in tuning the biosensor response [76] and may be helpful in future fine-tuning of the L-serine biosensor constructed here.
All three constructed L-serine-responsive biosensors, namely pSenSer1, pSenSer1-serE'-26aa, and pSenSer2-serE'-26aa, exhibited varied dynamic range, i.e., the difference between the minimum response value and the maximum output response in the dose-response curve, although the maximum response saturation was not reached during the L-serine feeding experiment. Therefore, each of the three could be utilized depending on the goal of the application. Sensor pSenSer2-serE'-26aa, for instance, could be used when a tighter response is needed. Furthermore, pSenSer1 was chosen to dynamically express trpB since the largest signal amplitude was observed using this biosensor. In this way, the production of L-tryptophan analogues was coupled to the elevated intracellular L-serine as sensed by the biosensor.

Patents
Patent applications regarding the results described here were not filed.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.