Escherichia coli strains DH5α or JM109 were grown at 37°C either on agar-solidified LB medium or in LB liquid medium supplemented with 5 g L⁻¹ NaCl under continuous shaking at 200 rpm. To select for the presence of certain plasmids the medium was supplemented with 100 μg mL⁻¹ ampicillin, 35 μg mL⁻¹ chloramphenicol, or 50 μg mL⁻¹ spectinomycin. C. necator (obtained from the German Collection of Microorganisms and Cell Cultures, DSMZ) was grown at 37°C in LB liquid medium supplemented with 2.5 g L⁻¹ NaCl under continuous shaking at 200 rpm. Synechocystis was cultivated in yBG11 (Shcolnick et al., 2007) liquid medium under continuous shaking at 150 rpm, or BG11 (Stanier et al., 1979) solidified with 1.5% (w/v) Bacto agar (Becton Dickinson) and supplemented with 3 g L⁻¹ Na₂S₂O₃. The cyanobacterial growth media were buffered with 10–50 mM HEPES to pH 7.2. Photoautotrophic growth conditions were set to 30°C, ambient CO₂, constant light illumination with 50 μmol photons m⁻² s⁻¹, and 75% (v/v) humidity. For the selection of mutants, the media were supplemented with 10 μg mL⁻¹ chloramphenicol, 20 μg mL⁻¹ spectinomycin, or 50 μg mL⁻¹ kanamycin. A non-motile, glucose tolerant strain of Synechocystis, originally received from Martin Hagemann (Rostock University, Germany), was used as the wild type (WT). The mutant Synechocystis(Δhox) that is devoid of the endogenous [NiFe]-hydrogenase was obtained from Kirsten Gutekunst (Kassel University, Germany). In particular, hoxEFUYH (sll1220-sll1226) have been replaced by a kanamycin resistance cassette (Appel et al., 2020).

In silico work was performed using the software Geneious (Biomatters). Genetic constructs were generated through standard molecular cloning procedures and maintained on plasmids in E. coli DH5α. DNA processing and recombination were performed using FastDigest restriction endonucleases (Thermo Scientific), T4 DNA ligase (Thermo Scientific), and FastAP thermosensitive alkaline phosphatase (Thermo Scientific) following the manufacturer’s instructions. The obtained constructs were verified by Sanger sequencing. Information about used oligonucleotides and plasmids is given in Supplementary Tables S1, S2.

The sequences for the design of synthetic operons encoding the H₂-sensing complex and the corresponding maturases were obtained from the megaplasmid pHG1 of C. necator (Schwartz et al., 2003). The hoxJ sequence was modified to code for a variant exhibiting an amino acid substitution from serine to glycine at position 422, denoted as HoxJ* (Lenz and Friedrich, 1998). The hoxC sequence was altered to instead encode the variant HoxC(D15H) (Gebler et al., 2007). The gene sequences were codon-usage optimized for Synechocystis using the web-based tool JCat (Grote et al., 2005). The inducible promoters P_rhaBAD (Behle et al., 2020) and P_nrsB (Englund et al., 2016) were fused to the hox and hyp operons, respectively. The constructs were also equipped with unique restriction endonuclease sites flanking each operon, effective ribosome binding sites (RBS*, Heidorn et al., 2011) upstream of each ORF, as well as standardized transcription terminators (BioBrick BBa_B0015, Registry of Standard Biological Parts, 2003). Tag-encoding sequences were added to the 3′ ends of the open reading frames of hoxB (3xFLAG-tag), hoxJ* (3xFLAG-tag) and hypX (Strep-tag), respectively. The synthetic operons were chemically synthesized (Eurofins Genomics) and provided on plasmids, denoted as pHox2 and pHyp. For the maintenance in Synechocystis, the synthetic operons were transferred, either separately or combined, into the plasmid pSHDY_P_rhaBAD::mVenus_P_J23119::rhaS (Behle et al., 2020) via XbaI/BcuI and BcuI/PstI sites, respectively. Thereby, the P_rhaBAD::mVenus cassette was replaced. The resulting plasmids were named pHySe_Hox and pHySe_Hox_Hyp (Figure 2). The full sequences including annotations are provided in the Supplementary Data.

Plasmids of the pFO series and derivatives of pSB1A2_P_trc1O (Huang et al., 2010), all of which harbor various combinations of expression cassettes for the genes hoxA, hoxJ* and sfgfp, were generated via the Gibson assembly procedure (Gibson et al., 2009). For this, respective sequences were amplified via PCR using primers with 5′ extensions to create homologous overhangs for the desired assembly with other DNA fragments. The P_SH promoter, including the 5′ untranslated upstream region of hoxF (Zimmer et al., 1995; Schwartz et al., 1998), was amplified from genomic DNA of C. necator using primer pair P17/P18. Together with the sfgfp reporter gene and a downstream BioBrick BBa_B0015 transcription terminator, which were generated through PCR with primers P16/Sam_102 from pSEVA351-sfgfp (Opel et al., 2022), it was used for the assembly of pFO6 utilizing KpnI-treated pSEVA351 (Martínez-García et al., 2020) as vector. The hoxA gene was amplified from gDNA of C. necator using P25 that additionally contained the ribosome binding site BioBrick BBa_B0034 (Registry of Standard Biological Parts, 2003) as 5′ extension and P26 fusing a sequence encoding a hexahistidine-tag at the 3′ end of the gene. The linear fragment was used for the assembly with BcuI-cut pSB1A2, yielding pSB1A2_P_trc1O-hoxA. pSB1A2_P_trc1O-hoxA^D55A has a substitution at the codon coding for the amino acid at position 55 of HoxA from 5′-GAT-3′ (Asp) to 5’-GCC-3′ (Ala). It was generated with the DpnI-digested product from an inverted PCR, taking pSB1A2_P_trc1O-hoxA as template and the primer pair P52/P53, as well as the homologous overhangs-supplying HoxA(D55A) double-stranded DNA fragment. The P_trc1O-hoxA and P_trc1O-hoxA^D55A constructs were PCR-amplified using primers P49 and P50 from either pSB1A2_P_trc1O-hoxA or pSB1A2_P_trc1O-hoxA^D55A, thereby fused to a 3’sequence encoding a Strep-tag II, instead of the His-tag, to be each inserted into BcuI-linearized pSEVA351, yielding pFO25 and pFO26, respectively. Analogously, these two synthetic gene constructs were inserted into BcuI-cut pFO6, which resulted in pFO27 and pFO28, respectively. To obtain pFO45 and pFO46, pHox2 was first subjected to an inverted PCR using P83 and P84, thereby deleting hoxB^FLAG and hoxC^D15H as well as the particular upstream situated RBS*. This was followed by AQUA cloning, creating pHox5 that encodes the P_rhaBAD::hoxJ*^FLAG cassette. The latter was excised by restriction with XbaI and inserted into XbaI-linearized pFO27 and pFO28, yielding pFO45 and pFO46, respectively. Sequences of the pFO series are provided in the Supplementary Data.

Synechocystis WT as well as Synechocystis(Δhox) parental cells were made electro-competent and transformed via electroporation as described previously (Brandenburg et al., 2021). Plasmid-harboring strains were selected on BG11 agar plates containing appropriate antibiotics. Plasmid presence was verified by colony PCR using suitable primers and the GoTaq MasterMix (Promega) according to the manufacturer’s instructions. Recombinant E. coli JM109 strains were generated by electroporation of electro-competent cells via standard procedures.

For the isolation of RNA, Synechocystis cells were grown until reaching an OD₇₅₀ ~ 0.8. The cultures were subsequently supplemented with final concentrations of 0.1% (w/v) L-rhamnose and 5 μM NiSO₄. After 24 h, cells were harvested by rapid vacuum filtration applying sterilized polyether sulfone filters (pore size 0.8 μm, PALL). RNA isolation was performed as described previously (Bolay et al., 2022). The RNA samples were treated with RNase-free DNase I (Thermo Scientific) according to the manufacturer’s instructions. Afterwards, cDNA was generated by applying the high-capacity cDNA reverse transcription kit (Thermo Scientific) as given in the manufacturer’s instructions. A total of ~0.4 ng cDNA were used as template for quantitative PCR. Amplification of specific regions within either the rnpB gene or hypX were performed using the GoTaq MasterMix (Promega) according to the manufacturer’s instructions and primer pairs rnpB_114F/rnpB_226R and S30/P88, respectively (Supplementary Table S1).

Synechocystis cells were grown in presence of elevated CO₂ concentration of 2% (v/v) to an OD₇₅₀ of ~2. To induce expression of the hox and hyp genes the medium was supplemented with final concentrations (f.c.) of 0.2% (w/v) L-rhamnose and 5 μM NiSO₄. In addition, 17 μM (f.c.) ferric ammonium citrate was added to foster hydrogenase maturation similar to previous reports (Lupacchini et al., 2021). Samples were collected by centrifugation after 24 and 48 h. Cells were resuspended in 750 μL TBS lysis buffer (100 mM Tris, 150 mM NaCl, 1 mM PMSF, pH 7.5) and transferred to 2 mL Precellys tubes (Bertin), together with a mixture of glass beads (Sartorius) of 0.09–0.15, 0.17–0.18, and 0.5 mm diameter. Cell disruption was performed using a Precellys Evolution homogenizer (Bertin) equipped with a Cryolys cooling system (Bertin) for 4 × 30 s at 10.000 rpm with 30 s interim breaks for cooling. The samples were subsequently separated in supernatant (soluble extract) and sediment (crude extract) by centrifugation and subjected to protein concentration determination using a Bradford dye reagent ready-to-use solution (Thermo Scientific) according to the manufacturer’s instructions. Cell suspensions of recombinant E. coli JM109 strains were analogously treated to obtain soluble protein extracts. Those cells were beforehand cultivated as described for GFP fluorescence determination, but using 0.5% (w/v) D-glucose instead of glycerol. Protein separation was performed via SDS-PAGE using a total amount of 20 μg protein for each sample. For immunodetection via Western blots the separated proteins were transferred to nitrocellulose membranes of 0.45 μm pore size (GVS), followed by hybridization with either a Strep-Tactin horse radish peroxidase (HRP) (IBA Lifesciences GmbH) or a monoclonal ANTI-FLAG M2-Peroxidase conjugate (Sigma-Aldrich) according to the manufacturer’s instructions. Chemiluminescence was detected by using the substrate solutions WesternSure PREMIUM Chemiluminescent (LI-COR) or WesterBright ECL (advansta) and the Fusion FX7 EDGE V0.7 imaging system (VILBER), following the manufacturer’s instructions.

Pre-cultivation of Synechocystis was performed as described above and expression of heterologous genes was induced by 0.1% (w/v) L-rhamnose and/or 2.5 μM NiSO₄. After 48 h, cells were harvested and disrupted analogously but using an alternative lysis buffer (5% (v/v) glycerol, 50 mM KPO₄, 1 mM PMSF, pH 8). A Synechocystis strain harboring the hoxFUYHW genes encoding the SH from C. necator (Lupacchini et al., 2021) served as control. Approximately 400 μg of soluble proteins were separated via native PAGE and subjected to in-gel staining as previously described (Lupacchini et al., 2021) with few modifications. These concerned the supplementation of 90 μM phenazine methosulfate, additionally to 800 μM NAD⁺ and 60 μM nitro blue tetrazolium (NBT) in an H₂-saturated activity buffer (50 mM Tris, pH 8). Furthermore, the incubation time was increased from ~0.5 h to ~2.5 h. Hydrogenase activity, i.e., the release of electrons from H₂ oxidation, is indicated via a step-wise reduction of the electron transfer mediators NAD⁺ and/or phenazine methosulfate and the colorimetric dye nitro blue tetrazolium, which finally results in a visible precipitation of formazan (Ponti et al., 1978). The gels were subsequently decolorized from the remaining loading dye in activity buffer overnight. Afterwards, presence of HoxB and HoxJ* proteins was confirmed by blotting the same polyacrylamide gel and hybridizing the membrane with antibodies against the attached 3xFLAG-tag as described above.

Recombinant Synechocystis strains containing plasmids pFO25 (negative control), pFO6 (P_SH::sfgfp), pFO27 (P_SH::sfgfp + P_trc1O::hoxA), or pFO28 (P_SH::sfgfp + P_trc1O::hoxA^D55A) were analyzed in vivo regarding GFP fluorescence. The detection was performed as described previously (Opel et al., 2022). GFP fluorescence determination in E. coli JM109 was conducted for recombinant strains harboring the following plasmids: pSEVA351 (negative control), pFO27 (P_SH::sfgfp + P_trc1O::hoxA), pFO45 (P_SH::sfgfp + P_trc1O::hoxA + P_rhaBAD::hoxJ*), and pFO46 (P_SH::sfgfp + P_trc1O:: hoxA^D55A + P_rhaBAD::hoxJ*). Single colonies from selective LB agar plates were picked to inoculate liquid LB medium pre-cultures that were grown for ~18 h at 37°C. 1% (v/v) of these suspensions were taken to inoculate second pre-cultures using M9* medium, supplemented with 0.001% (w/v) thiamine, 2 mM MgSO₄, 0.4% (v/v) glycerol, US* trace elements solution, and buffered to pH 7.2. The M9* pre-cultures were incubated for ~24 h at 37°C. A volume of 100 μL of these cell suspensions were added to 10 mL M9* medium in baffled shake flasks, and the cells were further cultivated at 30°C instead, due to the temperature sensitivity of HoxA (Zimmer et al., 1995). These main cultures were supplemented with 10 μM IPTG and/or 0.2% L-rhamnose after 8 h, followed by another 16 h of cultivation. For GFP fluorescence determination, samples were diluted to an OD₆₀₀ of ~0.5 with TBS buffer (100 mM Tris, 150 mM NaCl, pH 7.5) in a final volume of 1,200 μL. Technical triplicates (each 200 μL) were transferred into an opaque black flat microtiter 96-well-plate (Nunc), followed by fluorescence measurements using an Infinite 200 PRO microplate reader (Tecan; gain: 123, integration time: 20 μs, excitation bandwidth: 9 nm, emission bandwidth: 20 nm, z-position: 2000 μm, 25 flashes) and excitation/emission wavelengths of 485 nm/520 nm, respectively. Furthermore, the same sample was used to measure the absorption at 600 nm in a transparent flat microtiter 96-well-plate (Nunc), also using the Infinite 200 PRO microplate reader (bandwidth: 9 nm, 25 flashes). The blank of the TBS buffer background was subtracted from the values detected for the cell suspensions. The fluorescence intensities were finally normalized by division of respective OD₆₀₀ values.

To functionally produce the RH of C. necator in Synechocystis, we rationally designed two operons in silico that were generated via chemical synthesis and inserted it into a vector that can be maintained in Synechocystis (Figures 2A,B). In case of the active site-containing HoxC subunit, we made use of the amino acid exchange variant HoxC(D15H), which, in contrast to the native protein, supports H₂-dependent growth of C. necator at an O₂ level of up to 10% (Gebler et al., 2007). The resulting gene cluster for the biosynthesis of the H₂-sensing complex comprised hoxB, hoxC^D15H, and hoxJ*. For proper maturation of the catalytic [NiFe] center we also utilized the corresponding accessory genes hypA1B1F1CDEFX (Buhrke et al., 2001; Bürstel et al., 2016). In C. necator, these genes are organized in an operon structure with partially overlapping open reading frames (Schwartz et al., 2003). To ensure their correct expression in the cyanobacterial target organism, we altered the spatial organization by linking each hox and hyp gene by an artificial spacer region, resulting in separate translational units (Figure 2B). Furthermore, the synthetic ribosome binding site RBS* which functions in Synechocystis (Heidorn et al., 2011), was introduced upstream of every single gene to enable efficient translation initiation. For facile detection of the proteins, sequences encoding either a 3xFLAG-tag or a Strep-tag were fused to hoxB, hoxJ*, and hypX (Figure 2B). The hoxJ* and hypX genes were chosen because they are the dorsal genes of each particular operon. Detection of both proteins is considered representative of upstream gene expression.

All protein-coding sequences were codon-usage optimized for translation in Synechocystis. To drive the polycistronic transcription of hox and hyp gene clusters, we used the L-rhamnose-inducible promoter P_rhaBAD from E. coli (Behle et al., 2020) and the nickel ion (Ni²⁺)-dependent promoter P_nrsB of Synechocystis (Englund et al., 2016), respectively. Thus, this setup permits a selective induction as well as a tight and tunable transcription of the synthetic gene constructs P_rhaBAD::hoxB^FLAGC^D15HJ*^FLAG (hox operon) and P_nrsB::hypA1B1F1CDEFX^Strep (hyp operon) in Synechocystis. Moreover, both operons were equipped with insulating transcription terminators at their 3′ end.

The constructs were inserted into the replicative vector pSHDY_P_rhaBAD::mVenus _P_J23119-rhaS (hereafter referred to as pSHDY), which encodes the heterologous transcriptional regulator RhaS that enables rhamnose-inducible gene expression in Synechocystis (Behle et al., 2020). The P_rhaBAD::mVenus cassette present in this pSHDY construct (Figure 2A) was replaced with the synthetic hox operon resulting in the plasmid pHySe_Hox. Subsequently, the hyp operon was inserted downstream to obtain the plasmid pHySe_Hox_Hyp (Figure 2B). The resulting plasmids pHySe_Hox and pHySe_Hox_Hyp, as well as the precursor construct pSHDY, were used individually for the transformation of Synechocystis. For transformation, a strain devoid of the endogenous [NiFe]-hydrogenase, designated Synechocystis(Δhox) (Appel et al., 2020), was used to prevent subsequent cross-reactions with the RH activity. Plasmid presence was verified in all obtained clones (Figure 2C).

C-terminal linkage with 3x-FLAG (HoxB & HoxJ*) or Strep-tags (HypX), enabled protein detection by commercially available antibodies targeting the corresponding tag. The Synechocystis strain harboring pHySe_Hox_Hyp was grown in the presence of L-rhamnose and Ni²⁺ to trigger hox and hyp gene expression, respectively (see section “Materials and methods” for specific inducer concentrations). To confirm heterologous gene expression, we performed immunoblotting to detect HoxB and HoxJ* using protein extracts of samples collected 24 and 48 h after induction. In addition to the crude extract, we also analyzed the soluble protein fraction obtained by centrifugation. In fact, distinct bands were detected, and their intensity increased according to the induction time. The bands represent HoxB-FLAG and HoxJ*-FLAG, as no signal was detected in the same size range in the Synechocystis wild-type strain (WT) (Figure 3). Consistent with the expected cytoplasmic localization of the RH, a strong signal for HoxB-FLAG was observed in the soluble extracts. However, the signal associated with HoxJ*-FLAG, which is also thought to be soluble, was predominantly present in the crude extract indicating partial protein misfolding or membrane association. Nevertheless, even though the signal was quite weak, a significant part was also found in the soluble fraction, in particular 48 h after induction. Thus, both fusion proteins were specifically detected, confirming the expression of the corresponding genes.

In case of HypX-Strep, however, no specific signal was observed in cells containing pHySe_Hox_Hyp (not shown), indicating either no translation or protein instability. To verify that the synthetic hyp operon is at least transcribed, we extracted total RNA from the same strain and performed classical reverse transcriptase (RT)-PCR targeting hypX. The reversely transcribed copy DNA (cDNA) for hypX was only detected in the Synechocystis strain containing pHySe_Hox_Hyp but not in the WT (Figure 4). That the band obtained is indeed a result of cDNA amplification of a hypX transcript was verified by a parallel RNA sample that was not treated with reverse transcriptase and consequently showed no bands for hypX and the housekeeping gene rnpB. As hypX is situated at the 3′ end of the synthetic gene cluster, we assume, that transcription of the upstream situated hyp genes also occurred. However, this analysis did not confirm the translation of the hyp gene transcripts. Nevertheless, sufficient synthesis of the maturation apparatus for the RH could be assumed, as indicated by the subsequent analysis (see below).

To investigate if the recombinant gene expression indeed results in the formation of active H₂-sensing RH, we analyzed the H₂ oxidation activity in soluble extracts of photoautotrophically grown cells of Synechocystis strains harboring the plasmids pSHDY, pHySe_Hox, and pHySe_Hox_Hyp. To this end, we used an in-gel activity assay under an H₂ atmosphere (Buhrke et al., 2004), which has also been used recently to confirm the activity of hydrogenases in Synechocystis (Lupacchini et al., 2021). The recombinant strains were cultivated in the presence of different inducer combinations to achieve independent expression of the hoxBCJ* genes (L-rhamnose) and the hypA1B1F1CDEX operon (Ni²⁺). Soluble protein extracts were prepared and subjected to native polyacrylamide gel electrophoresis. Strikingly, in-gel H₂ oxidation activity was detected only for Synechocystis(pHySe_Hox_Hyp) induced with both L-rhamnose and Ni²⁺ (Figure 5A, In-gel staining panel). The resulting activity bands were located at the same positions as the bands in the immunoblot analysis (also based on the native gel), showing HoxB-FLAG in complex with HoxC as the hydrogenase core module, and potentially HoxJ*-FLAG (Figure 5A, Western Blot panel). As the production of both HoxB-FLAG and HoxJ*-FLAG has been confirmed by a previous Western blot (Figure 3), a separate detection of both proteins has not been performed in this case. No H₂ oxidation activity was detected in the corresponding native gel, when the Hyp proteins required to produce catalytically active RH were absent, either due to the lack of the hyp genes (in case of strain pHySe_Hox) or the inducer Ni²⁺ (in case of strain pHySe_Hox_Hyp) (Figure 5A). Thus, expression of both the hox and hyp gene clusters is required to obtain detectable H₂ oxidation activity for the H₂-sensing RH in Synechocystis. While most [NiFe]-hydrogenases are inactivated by traces of O₂ (Shafaat et al., 2013), the RH activity has shown to be O₂-tolerant (Ash et al., 2015). Remarkably, the RH activity was observed in extracts of O₂-evolving photoautotrophically grown Synechocystis cells, indicating that both the maturation process and the catalytic activity of the RH occurred in the presence of O₂.

The RH activity in Synechocystis is comparatively low, as demonstrated by the long incubation time of ~2.5 h required to obtain detectable bands derived from H₂-dependent NBT reduction in the activity gel. A Synechocystis control strain containing the highly active SH from C. necator (Lupacchini et al., 2021) showed significantly stronger signal intensities after ~0.5 h already. The low signal strength of the RH in Synechocystis could not be increased by enhanced L-rhamnose levels, suggesting a saturation at 0.05% (w/v) rhamnose and consequently no limitation of the RH structural proteins (Figure 5B). Altogether, these results demonstrate for the first time the functional production of a recombinant regulatory hydrogenase with low H₂ turnover activity in a cyanobacterium.

The final goal is to couple the functional RH with the associated kinase HoxJ* and the cognate response regulator HoxA to establish an H₂-sensing signal transduction cascade in Synechocystis. In C. necator, HoxA positively controls the transcription of the genes encoding the SH and the MBH through binding to the promoters P_SH and P_MBH, respectively (Zimmer et al., 1995; Schwartz et al., 1998). The HoxA activity is modulated by HoxJ*-mediated phosphorylation, with transcriptional activation by HoxA in its non-phosphorylated state (Lenz and Friedrich, 1998). To drive HoxA-mediated gene expression in Synechocystis, the hoxA gene was set under control of the IPTG-inducible P_trc1O promoter, which has been shown to provide sufficient constitutive expression due to the lack of the lac repressor LacI in the cyanobacterial host (Huang et al., 2010). Furthermore, a sfgfp reporter gene encoding the superfolder green fluorescent protein (Pédelacq et al., 2006) (hereafter referred to as GFP) was fused to the HoxA-dependent promoter P_SH (Figure 6A). The synthetic gene constructs P_SH::sfgfp and P_trc1O::hoxA were introduced into Synechocystis WT, either separately or combined on a replicative plasmid. However, no significant GFP fluorescence beyond background activity was detected in any strain carrying the reporter gene construct alone or in combination with P_trc1O::hoxA (Figure 6B). An inactivation of HoxA by unspecific phosphorylation could be excluded because the GFP fluorescence was similar in a reporter strain containing the phosphorylation-insensitive variant HoxA(D55A) instead of HoxA. This variant cannot be phosphorylated at the crucial aspartate at position 55 and has been shown to be always active (Lenz and Friedrich, 1998). Although several attempts have been made to improve hoxA expression, e.g., by using different promoters as well as codon-usage optimized gene variants, HoxA-dependent sfgfp expression has not yet been achieved in Synechocystis (not shown).

However, to validate the general functionality of our genetic constructs and to establish HoxA and HoxJ*-dependent gene expression in a heterologous system, we introduced the same replicative plasmids harboring P_trc1O::hoxA and P_SH::sfgfp into E. coli JM109. As this host contains LacI, hoxA expression is IPTG-inducible. Indeed, a ~ 5-fold higher GFP fluorescence was detected in cells grow in presence of IPTG compared to cells grown in absence of IPTG (Figure 6C). Under non-induced conditions, the GFP fluorescence was similar to the background autofluorescence of control cells harboring the empty vectors. Thus, in contrast to Synechocystis, the expected activity of HoxA to promote reporter gene expression via the P_SH promoter was confirmed in E. coli.

Moreover, another synthetic gene construct encoding HoxJ* was included in the study to demonstrate the modulation of HoxA activity. HoxJ*-mediated phosphorylation of HoxA is expected to inactivate the response regulator (Lenz and Friedrich, 1998; Buhrke et al., 2004). A gene encoding HoxJ* carrying a C-terminal FLAG-tag (HoxJ*^FLAG) was set under control of the L-rhamnose-inducible promoter P_rhaBAD, and the resulting plasmid was introduced into the strain already harboring the Ptrc1O::hoxA construct. In general, this promoter could also be used in Synechocystis (Behle et al., 2020). Again, IPTG-induced hoxA gene expression resulted in GFP fluorescence. Remarkably, a significant decrease in GFP fluorescence was observed in cells that were grown in presence of both IPTG and L-rhamnose (Figure 6C). This correlates well with the exclusive detection of HoxJ* in protein extracts of cells grown in presence of L-rhamnose and carrying the respective gene construct (Figure 6D). The GFP level decreased to 50% compared to values obtained with strains either lacking the hoxJ* gene or that do not sufficiently express hoxJ* due to the absence of L-rhamnose. HoxA(D55A), which cannot be inactivated by HoxJ* through phosphorylation (Lenz and Friedrich, 1998), was again included as a control. As expected, a higher reporter signal was obtained with the corresponding strain in the presence of both IPTG an L-rhamnose than with the strain expressing wild-type hoxA as well as hoxJ* (Figure 6C). This result also suggests that the response regulator in E. coli JM109 is not inactivated by unspecific phosphorylation. Thus, for HoxA, an effective ~30% reduction in reporter signal strength was achieved by HoxJ* compared to HoxA(D55A), demonstrating the desired modulation. Moreover, the HoxJ*-mediated decrease of HoxA-dependent GFP fluorescence was further tunable by different amounts of L-rhamnose (Figure 6E).