Background

Implementation of a bio-based economy relies on the development of efficient microbial bioprocesses to produce commodity chemicals and fuels. Advances in metabolic pathway and microbial host engineering have led to an ever-expanding toolbox that is enabling the development and optimisation of biological production routes to a diverse array of compounds. These include oleochemicals [1], gaseous fuels [2], petrochemical derivatives [3], biomaterials [4] and bioplastics [31]. Default settings were used during the database initialisation, replicon specification and build to identify metabolic pathways from the genome data. These pathways were then used to inform manual homology searches and improve genome annotation through NCBI BLASTp (https://blast.ncbi.nlm.nih.gov/Blast.cgi) with varied parameters decided based on the similarity of the sequence in question with those in the database.

Structure prediction and structure homology searches

Validation of the annotation of putative carbon fixation and thiosulfate utilisation pathway genes in the genome sequence of H. rowanensis (Additional file 1: Fig. S2) was performed by generating three dimensional models using the Alphafold 2.0 server [32]. This was performed using the recommended default parameters for prokaryotic peptide sequences. These structures were viewed and interrogated in Pymol [33]. The DALI webserver [34] was used to match the predicted structures to the entire PDB database by homology with the recommended default settings. The resulting match with the lowest RMSD alignment was viewed and manual alignments performed in Pymol. The resulting alignment was interrogated for comparison of key residues from the predictions in the active site or those interacting with bound ligands in the published PDB structures.

Quantitation of ectoine and PHA production

Ectoine was harvested from Halomonas cultures using a modified ‘bacterial milking’ method described previously [35]. Cultures (50 mL) were harvested by centrifugation at 4000g for 5 min at room temperature. The pellet was resuspended in 5 mL of ultrapure water (18.2 MΩ·cm) and vortexed for 1 min until homogeneous. The suspension was incubated for 10 min at room temperature and centrifuged as before. The milky white supernatant was retained and the ectoine content was quantified by HPLC using an Agilent 1260 Infinity HPLC with a 1260 ALS autosampler, TCC SL column heater, a 1260 refractive index detector (RID). Samples were run on an Agilent Hi-Plex-H column (300 × 7.7 mm) using 2.5 mM sulfuric acid as the mobile phase (0.6 mL/min) at 60 °C for 30 min. Ectoine quantitation (retention time of 6.3 min) was performed by comparing peak areas to a standard curve generated from authentic standards. A salinity gradient in the thiosulfate medium was used to find the optimum for ectoine production which is salinity dependent. Different water sources with a variety of minerals were tested as well as salinity from added NaCl.

PHA production of H. rowanensis and H. bluephagenesis TD1.0 were performed using a modification of the method described previously [23]. Cultures (50 mL) in growth medium (LB60 ± 4% (w/v) glucose and/or thiosulfate medium) were incubated at 37 °C with 180 rpm agitation for 48 h. Aliquots (50 mL) were centrifuged at 4000g for 10 min, followed by two cell pellet washes in 5 mL water with centrifuging as before. The pellets were flash frozen in liquid nitrogen and freeze dried overnight. Dried cell pellet (12–86 mg without glucose, 85–190 mg with glucose feed) in 4 mL airtight glass vials were dissolved in 1 mL chloroform, followed by the addition of the derivatization reagent (1 mL; 3% H2SO4 in anhydrous methanol containing the internal standard 0.1% benzoic acid). Mixtures were incubated at 100 °C in a sand bath for 4–6 h then cooled to room temperature. Sodium bicarbonate (~ 1 g) was added to quench the reaction and the samples were refrigerated overnight to allow any cell debris to precipitate. The debris was centrifuged at 4000 g and the supernatant was clarified by passage through a filter tip. The methyl 3-hydroxybutyrate content (derivatized PHA monomer) was determined by GC using an Agilent Technologies 7890A GC equipped with an FID detector using an HP-5 column [36]. Compound identification was determined by comparison of the retention time with authenticated standards run under identical conditions. PHA content is expressed as the percentage of PHA mass as a fraction of the total dry cell weight (DCW) of the lyophilized cell pellet (% g/g DCW).

Radiometric bicarbonate uptake and fixation

Halomonas species were grown in LB60 or the thiosulfate medium [21], while ATCC290-medium for thiobacilli was used for growth of H. neapolitanus. Both E. coli and S. enterica were grown in LB medium. Cell cultures were grown to exponential phase, harvested by centrifugation, and were resuspended in Rubisco assay buffer (100 mM EPPS, pH 8.0, and 20 mM MgCl2). The 14C (H14CO3−) fixation assay was performed as described previously [37].

Cell cultures were grown to exponential phase using the medium above, harvested by centrifugation and resuspended in filtered nitrogen-bubbled culture medium to OD 600 nm of 4. The whole cell 14C uptake assay was performed according to previous studies [37] with several modifications. Radiolabelled NaH14CO3 (10 mM final concentration) was added, and the cultures were incubated at 37 °C for 10, 15 and 60 min. Cell permeabilization was performed by the addition of alkyltrimethylammonium bromide (MTA) to a final concentration of 0.03% (w/v). Assay termination was performed by adding formic acid [10% (v/v)] and boiling the mixtures on heating blocks (98 °C) until dry, removing any unconverted NaHCO3 as CO2 gas. The powders were resuspended in water (100 µl), mixed with scintillation cocktails (1 mL; Ultima Gold XR, PerkinElmer) and counted for radioactivity (Tri-Carb, PerkinElmer). Three biological repeats were performed for each sample and the data were normalised to the cell density (OD600 nm).

Propane production by H. rowanensis and H. bluephagenesis TD1.0

Introduction of plasmids pHAL7–CvFAPG462V and pHAL102–CvFAPG462V [17] into H. rowanensis was performed via conjugation, using the S17-1 E. coli donor strain, as described previously [2]. Propane assays were performed as described previously for Halomonas cultures, except the medium (20 mL) was either LB60 or thiosulfate medium at pH 7.0 and the butyric acid concentration was 10 mM [2]. Propane quantitation was performed by headspace sampling using an Agilent 490 Micro GC [2]. Halomonas TD1.0 propane assays were as previously described in LB60 or thiosulfate medium at pH 9 [2].

Results and discussion

Isolates identification and genome sequencing

A variety of native halophilic or halotolerant bacteria were isolated from a natural brine spring (pH 7.1) by enriching with heterotrophic culture medium specific for H. bluephagenesis [23], but at neutral pH. Partial 16S rDNA, sequence analysis of individual isolates revealed the presence of some Gram-positive Bacillus species (orange-coloured colonies) and a Kocuria sp. strain (small, white colonies). In addition, Gram-negative bacteria were present, including two potential Idomarina species (I9–I10) and several Halomonas species (isolates I3–I7) (Additional file 1: Table S1). The Halomonas isolates were small aerobic non motile rods (Fig. 1 inset), that grew heterotrophically on amino acid-based carbon sources.

Fig. 1
figure 1

Phylogenetic tree of ‘Old Biot’ Halomonas isolates. Analysis using apartial and bfull 16S rDNA sequences from PCR products and genome sequencing, respectively. Insets: light microscope images of H. rowanensis at 100X magnification. Strains: H. saliphila st. LCB169; H. flava st. YIM 94343; H. socia st. NY-011 16S; H. subglaciescola st. DSM 4683; H. halmophila st. ATCC 19717; H. elongata st. 1H9; H. alkaliphilia st. 18bAG and H. taeanensis st. BH539

Full size image

Partial genome sequences were obtained for Halomonas I4 and Idiomarina I9 isolates, with a more complete genome sequence generated for the fastest growing Halomonas strain I5 (66.5% annotated). This compares to the 89% annotation of the H. bluephagenesis TD01 strain (Genbank: GCA_923868895.1), although in the latter case a secondary correlation of the low sequence identity genes needs to be performed. The 16S rRNA genes for Halomonas strains I4 and I5 were identical, with a 99% identity to the known organism Halomonas taeanensis strain BH539 [38] (Fig. 1). This strain was more distinct from the robust industrial strain H. bluephagenesis TD1.0, which prefers a highly alkaline environment. We designated the I5 (and by homology I4) strain as Halomonas rowanensis.

For the two Idiomarina isolates, the 16S rDNA sequences differed in 2 base positions, suggesting two different species or strains were present. We decided to take forward the Halomonas isolates as chassis for recombinant propane production as the synthetic biology toolbox is more advanced than for Idiomarina, and our existing synthetic biology constructs were already tailored for Halomonas [2, 17, 20].

High salt tolerance of the new isolates

Each Halomonas isolate was screened for its tolerance towards salinity (NaCl) and optimal pH. The isolates were found to tolerate higher NaCl concentrations than the control H. bluephagenesis TD1.0 strain (Additional file 1: Fig. S2) but preferred neutral pH at higher salinities (9–12% NaCl). This is consistent with the naturally high salinity (~ 20%) and neutral pH observed for the native spring they were isolated from. Given our ultimate aim of biological propane production, we additionally tested the Halomonas isolates for tolerance towards the precursor butyric acid [2]. The highest butyric acid tolerance was found with H. rowanensis (80 mM), similar to engineered H. bluephagenesis TQ10 [20] and TD1.0 [36] (Additional file 1: Fig. S2).

The mechanism(s) of salt tolerance of the new isolates was investigated by genome mining for pathways known to be involved in osmoregulation. Genes associated with salt tolerance in Halomonas biemenensis include the sodium-translocating NADH:quinone oxidoreductase (nqrA) and an NAD-specific glutamate dehydrogenase (gdhB). These genes are associated with sodium efflux and the production of the secondary compatible solute glutamate [15]. Both genes were found to be present in the genomes of both the Halomonas and Idiomarina isolates I4, I5, I9 and I10.

Production of compatible solutes and PHA

Halomonas species are known to counteract osmotic pressure by the production and intracellular accumulation of significant levels of the compatible solute ectoine [39]. This compound is a valuable product within biotechnology and cosmetics industries and has many medicinal uses. For example, ectoine is used as biofunctional stabilizers, skin protectors and potential drugs for diseases, such as Alzheimer’s and rhinoconjunctivitis [40]. Annotation of the H. rowanensis genome revealed the presence of a likely ectoine biosynthesis pathway (Fig. 2a). This included the ectABC operon and genes for aspartate kinase (lysC) and aspartate semialdehyde dehydrogenase 1 (asd1) [40]. In addition, the gene for ectoine dioxygenase (ectD) was present, which produces hydroxyectoine from ectoine. The osmoregulated solute TRAP transporter (teaABC) was also identified, which mediates the uptake of ectoine and hydroxyectoine in Halomonas elongata [41].

Fig. 2
figure 2

Heterotrophic ectoine and PHB production of H. rowanensis. A Halomonas metabolic pathway to ectoine and PHB production. Gene annotations from H. rowanensis are shown in blue. B Ectoine production of H. rowanensis (green) and H. bluephagenesis TD1.0 (blue) in variable levels of salinity. Cultures were harvested and milked for ectoine with ultrapure water as described previously [35]. C PHB production of H. rowanensis and H. bluephagenesis TD1.0 in LB60 with and without 4% glucose. Cultures were grown for 24 h and the PHA concentration was determined by polymer extraction, degradation, derivatization and GC quantitation. Error bars represent one standard deviation of triplicate analyses. Enzymes: asd = aspartate semialdehyde dehydrogenase 1; EctA = L-2,4-diaminobutyric acid acetyltransferase; ectB = L-2,4-diaminobutyric acid transaminase; ectC = L-ectoine synthase; ectD = ectoine dioxygenase; lysC = aspartate kinase; phaA = 3-ketothiolase; phaB = NADPH-dependent acetoacetyl-CoA reductase; phaC = PHA synthase. The PHB titre and CDW data are available in Additional file 1: Table S2

Full size image

Ectoine can be obtained naturally from H. elongata by a ‘bacterial milking’ process [35]. We performed bacterial milking on H. rowanensis and a control H. bluephagenesis TD1.0 strain, the latter of which is known to generate large quantities of ectoine [12]. As expected, high quantities of ectoine were produced by both Halomonas strains under high salt conditions (10–15% NaCl; Fig. 2b). Both strains generated similar titres of ectoine at 15% salinity (0.97 g/L), comparable to titres achieved by H. elongata [35]. However, H. bluephagenesis TD1.0 generated nearly double the ectoine levels than H. rowanensis at only 10% NaCl (2.19 vs 1.20 g/L, respectively).

Halomonas species are also well-known as prolific PHA producers, generating up to 92% of dry cell weight (DCW) in strains engineered to upregulate the expression of the three native PHA genes (phaA, phaB and phaC; Fig. 2a [18]). The native PHA production titres of H. rowanensis in standard growth medium were marginally lower than H. bluephagenesis TD1.0 (15.38 ± 8.01% vs 21.90 ± 10.56%, respectively; Fig. 2c), with the low titres representative of harvesting long after the carbon source was depleted. Significantly higher titres were obtained when cultivated in the presence of high glucose levels, with H. rowanensis accumulating PHA at 48.05% ± 14.37% of the dry cell weight (Fig. 2c). This is equivalent to around 3.0 g/L PHA. Under the same growth conditions, H. bluephagenesis TD1.0 generated PHA at around 3.3 g/L. Therefore, H. rowanensis has a similar capacity as a PHA producer as the known industrial chassis H. bluephagenesis TD1.0, which is known to generate up to 82% DCW of PHA without any pathway engineering of the PHA biosynthesis operon [16, 23] and with multiple coproducts [36].

Chemoautotrophic growth of H. rowanensis from CO2 in wastewater

A key advantage of using Halomonas as an industrial microbial chassis is its ability to grow under non-sterile conditions in sea water and wastewater using simple inexpensive organic carbon sources [2]. We extended this approach further by testing H. rowanensis for chemoautotrophic growth on salinity adjusted seawater (Irish sea) and polluted river or canal water collected from waterways around the Greater Manchester region [42]. Significant growth was detected with H. rowanensis after repeated re-streaking into sterile domestic waterway samples, but not seawater, in the absence of exogenous carbon sources. This suggested either significant organic carbon sources were naturally present in the Manchester waterways or H. rowanensis is capable of chemoautotrophic growth on inorganic carbon (e.g., CO2). Control cultivations of H. bluephagenesis strains TD01 and TQ10 did not display significant growth under identical growth conditions as H. rowanensis, suggesting there was insufficient Halomonas-specific carbon sources available to support heterotrophic growth. Comparative heterotrophic growth of H. rowanensis was performed in wastewater supplemented with glycerol, which showed around 2.5-fold higher biomass production than (potential) chemoautotrophic growth (Additional file 1: Fig. S3).

Thermotolerant Halomonas stevensii is known to fix CO2 using thiosulfate as a sole energy source [21]. We further investigated both H. rowanensis and the H. bluephagenesis strains TD1.0 and TQ10 to determine if they are capable of chemoautotrophic or mixotrophic growth phenotypes using thiosulfate as the sole energy source [21]. After repeated subculturing into organic carbon-free thiosulfate medium, only H. rowanensis was capable of significant growth (Fig. 3a). The addition of NaHCO3 to the medium resulted in an enhancement in overall biomass production (Fig. 3b), suggesting H. rowanensis can fix CO2 into organic carbon [37, 43], similar to the growth of H. stevensii under chemoautotrophic conditions [21]. Given that H. rowanensis also grows efficiently under heterotrophic conditions, this suggests this organism may be a facultative chemoautotroph.

Fig. 3
figure 3

Chemoautotrophic growth of H. rowanensis. A Chemoautotrophic growth of three Halomonas species. Cultures were grown in thiosulfate growth medium containing 150 mM NaHCO3 at 30 °C for 48 h. B Growth of H. rowanensis with or without supplemental inorganic carbon. Cultures were grown in thiosulfate medium in the presence and absence of 150 mM NaHCO3 for 100 h at 30 °C. C Radiolabelled carbon fixation of non-permeabilised cultures. Cultures were grown in species-specific growth medium and the 14C (H14CO3−) fixation assay was performed as described previously [37] with the culture at OD 600 nm = 4.0. D Time course of 14C labelled bicarbonate fixation in six microorganisms. Cells (OD 600 nm = 4.0) were permeabilised with alkyltrimethylammonium bromide with NaH14CO3 addition to allow the measurement of total enzymatic carbon fixation activity independently of bicarbonate uptake systems. The H. bluephagenesis strain for parts C, D was TD1.0

Full size image

To assess the potential of CO2 fixation by H. rowanensis, we performed radiolabelling experiments with 1H14CO3− and looked for label incorporation into cellular biomass. The NaH14CO3 was fed externally during growth, so label incorporation is a measure of both the efficiency of bicarbonate uptake systems as well as carbon fixation and retention intracellularly. Radiolabelled bicarbonate incorporation was detected for H. rowanensis (2.21 ± 0.75 nmol/h/mL at OD 600 nm = 4.0; Fig. 3c), which is equivalent to 2.47 ± 0.84 nmol/h/g cells. This is approximately threefold lower than label incorporation under the same growth conditions by a known chemoautotroph H. neapolitanus (6.40 ± 1.36 nmol/h/mL). Interestingly, H. bluephagenesis TD1.0 also displayed 14C incorporation (1.42 ± 0.20 nmol/h/mL), despite a lack of growth in thiosulfate medium. This suggests it contains the machinery for carbon fixation, but not via thiosulfate as an energy source.

Further radiolabelled bicarbonate incorporation studies were performed with H. rowanensis and H. bluephagenesis TD1.0 under heterotrophic (LB60) and chemoautotrophic (thiosulfate) growth conditions at pH 7.0. In this case, cells were permeabilised to remove the dependence on bicarbonate uptake into the cells. The highest levels of 14C incorporation were seen with H. rowanensis cultivated under chemoautotrophic conditions (1.32 ± 0.03 nmol/mL), with a near fourfold reduction when grown initially in heterotrophic growth medium (0.34 ± 0.01 nmol/mL; Fig. 3d). This suggests a full carbon fixation pathway may be upregulated when the culture is transitioned into autotrophic growth medium, rather than the organism simply containing a few genes enabling it to survive brief periods of fixed carbon starvation.

Radiolabelled bicarbonate incorporation was apparent with H. bluephagenesis TD1.0 during the 1 h assay despite its lack of growth in thiosulfate-base chemoautotrophic medium. The incorporation was less efficient that H. rowanensis (0.57 ± 0.01 nmol/mL; Fig. 3d), but also showed the upregulation when cells were transferred into chemoautotrophic medium. This suggests that carbon fixation may be a common trait within the Halomonas genus.

Putative carbon fixation pathway

We examined the annotated genome sequences to detect potential pathways supporting both heterotrophic and carbon fixation metabolism. This was performed using an updated annotation of the genome with accuracies improved using BLAST sequence homology paired with RAST annotation. In addition, the overall enzyme class was confirmation in addition to the likely presence of the appropriate active site residues/arrangement via AlphaFold structural modelling [32]. As expected, the annotated genome contained all the expected genes required to support a heterotrophic lifestyle. This included complete pathways for glycolysis, TCA cycle with glyoxylate shunt, pentose phosphate and Entner–Duodoroff pathways [44]. Genes for a fully functional aerobic electron transport chain were also present, such as cytochrome c oxidase for utilising oxygen as a terminal electron acceptor and a proton translocating ATP synthase for chemiosmotic energy generation. The potential use of nitrate as an alternative terminal electron acceptor was also inferred by genes for respiratory nitrate reductase as well as genes permitting electron transfer from donor compounds formate and glycerol-3-phosphate.

Classic chemoautotrophic bacteria, such as H. neapolitanus, fix CO2 via the Calvin–Bensen–Bassham cycle, with energy supplied by sulfur oxidation pathways [45]. A carbon-concentration mechanism is present in H. neapolitanus via the presence of bicarbonate transporters, carbonic anhydrase and α-carboxysomes, which enclose the key carbon fixation enzyme ribulose 1,5-bisphosphate carboxylase/oxygenase (RubisCO) [46]. However, annotation of the H. rowanensis genome did not reveal the presence of any gene involved in the Calvin–Bensen–Bassham cycle or evidence of α-carboxysome formation. The quinoprotein dehydrogenase-associated SoxYZ-like carrier gene was found in the genome of H. bluephagenesis TD1.0, but no further genes involved in α-carboxysome formation were annotated.

The new annotated genome identified a putative complete reductive tricarboxylic acid (rTCA) cycle in H. rowanensis (Fig. 4) [47], which could account for the ability to fix CO2. This pathway has been found in other chemoautotrophs, such as Chlorobium, Desulfobacter hydrogenophilus and some members of the thermophilic Aquificales order and archaeal Thermoproteaceae family [48]. The rTCA cycle generates one molecule of oxaloacetate from four molecules of CO2 and requires 4–5 mol of adenosine 5′-triphosphate (ATP) [48]. Essential genes for this pathway are required to overcome key energetically unfavourable reverse reaction steps. This includes ATP citrate lyase, which catalyzes the cleavage of citrate into acetyl-CoA and oxaloacetate in a CoA- and ATP-dependent manner (Step 11 of Fig. 4). Other key enzymes are two of the four carbon dioxide-fixing enzymes 2-oxoglutarate:ferredoxin oxidoreductase (Step 7 of Fig. 4), and pyruvate:ferredoxin oxidoreductase (Additional file 1: Table S3) [48].

Fig. 4
figure 4

Putative carbon fixation pathway in H. rowanensis. Enzymes: 1 = phosphoenolpyruvate synthase; 2 = phosphoenolpyruvate carboxylase; 3 = malate dehydrogenase; 4 = fumarate hydratase; 5 = succinate dehydrogenase; 6 = succinyl-CoA ligase; 6a = γ-butyrobetaine,2-oxoglutarate dioxygenase; 7 = 2-oxoglutarate synthase; 8 = Isocitrate dehydrogenase [NADP]; 9–10 = aconitate hydratase; 11 = citrate lyase and 12 = NADP-dependent malic enzyme. Metabolites: PEP = phosphoenolpyruvate; αKG = α-ketoglutarate; 3-OHTMAB = 3-hydroxy-4-trimethylammoniobutanoate and 4-TMAB = 4-trimethylammoniobutanoate. Gene annotation was performed using a combined approach with Rapid Annotations using Subsystems Technology toolkit (RASTtk) along with BLASTp validation, using the KEGG carbon fixation pathways as guide for enzyme discovery. EC numbers for each enzyme can be found in Additional file 1: Table S3

Full size image

We identified at least one annotated gene for each step of the rTCA cycle (Additional file 1: Table S3), whose likely function were predicted by AlphaFold structural simulation and the DALI structural homology webserver (Additional file 1: Fig. S4). Other genes identified included enzymes catalysing the interconversion of malate to glyoxylate and acetyl CoA or pyruvate and those involved in the hydroxybutyrate [49] and hydroxypropionate [50] cycles. We also identified a gene for γ-butyrobetaine hydroxylase, which could potentially act as an alternative enzyme to catalyse steps 6–7 (succinate to α-ketoglutarate) in the presence of ascorbate [51]. Given the presence of key enzymes citrate lyase, NADPH dependent malic enzyme and malate dehydrogenase, the interconversion of citrate and pyruvate is likely possible.

Putative thiosulphate utilisation pathways in H. rowanensis

Chemolithoautrophic bacterium obtain energy to fix atmospheric CO2 typically by extracting electrons from reduced ferrous (Fe[II]), thiosulfate (S2O3−2) or ammonium (NH4+) ions [52]. For example, energy generation in H. neapolitanus and Paracoccus pantrophus proceeds via the oxidation of thiosulfate to sulfate using the Sox system [53, 54], delivering eight electrons to the quinone pool for ATP generation (Additional file 1: Fig. S5a). Alternatively, some chemo-organoheterotrophic bacteria, including Halomonas species, are known to derive energy from the oxidation of thiosulfate to tetrathionate (Additional file 1: Fig. S5b) via the action of thiosulfate dehydrogenase (tsaD) [55].

Annotation of the genomes of H. rowanensis and Halomonas strain I4 revealed the presence of a variety of genes involved in the transport of thiosulfate to sulfite, sulfide and elemental sulfur (Additional file 1: Table S4). However, there were no Sox genes system or tsaD annotated in either genome that could explain how H. rowanensis could derive energy to fix CO2 in chemoautotrophic thiosulfate medium. Given the relative incompleteness of the H. rowanensis genome compared to H. bluephagenesis TD01, it is likely that the genes involved in chemoautotrophic energy generation are buried within the putative proteins. Therefore, further work is required before the mechanism(s) of energy generation is determined for H. rowanensis and H. bluephagenesis TD01.

The presence of multiple thiosulfate transferases and sulfate adenyltransferases suggest H. rowanensis contains the assimilatory and dissimilatory sulfate reduction and oxidation pathways. The sulfur metabolism genes annotated in H. rowanensis suggests they play a role as a sulphur source for l-cysteine and related compounds (Fig. 5). For example, several putative thiosulfate sulfur transferases (e.g., rhdA and glpE) could reduce thiosulfate to sulfite [56]. Further NADPH-dependent reductions could lead to the production of sulfides, which can act as a sulfur donor to O-acetyl-l-serine to generate l-cysteine [57]. Genes for the conversion of sulfate to sulfite via adenosine-5′-phosphosulfate (APS) are also present (Fig. 5). As these pathways overall are net energy requiring, they are unlikely to be coupled directly to CO2 fixation.

Fig. 5
figure 5

Putative thiosulfate and other oxidised sulfur utilisation pathways in H. rowanensis. APS = adenosine-5′-phosphosulfate; PAPS = 3'-phosphoadenylyl sulfate; Enzymes: CGS1 = cystathionine γ-synthase; CysS = cysteine synthase; fccB = flavocytochrome c; rhdA/glpE/cysA1 = thiosulfate sulfur transferases; SAT = sulfate adenylyltransferase; sir/cysJ = sulfite reductase

Full size image

Chemoautotrophic propane, ectoine and PHA production

We wished to test the potential of chemoautotrophic H. rowanensis to act as a robust microbial chassis to produce secondary products by incorporating an established route towards bio-propane (bio-LPG) production. This was chosen as propane production in Halomonas is already established in the literature [2, 20, 36], so could be used as a basis of comparison for the new strain. This was performed by incorporating the recombinant fatty acid photodecarboxylase variant G462V from Chlorella variabilis NC64A (CvFAPG462V). This enzyme catalyses the decarboxylation of butyric acid to propane [58], and was previously shown to be functionally expressed in H. bluephagenesis TQ10 cultivated on biodiesel waste glycerol and amino acids [2, 17, 20]. A second strain H. bluephagenesis TD1.0 also generated bio-propane with a higher degree of fitness [36]. However, both strains required extensive pH control measures to prevent the culture from its natural tendency to shift the pH up to 9.0, which would rapidly abolish propane production [2]. By incorporating CvFAPG462V into H. rowanensis, we wished to investigate whether this neutral pH chassis would be more suitable for bio-propane production. We expressed CvFAPG462V in H. rowanensis under control of a weak (P7) or strong (P102) constitutive promoter and cultivated the organism under blue light to activate CvFAPG462V for propane production.

Propane production from CO2 was more significant when using the weaker P7 promoter (61.8 µg/L culture; Fig. 6a). Supplementation with butyric acid (substrate and carbon source) led to a 19-fold increase in propane titres (1.2 mg/L culture). These relatively low titres compared to prior studies with heterotrophic H. bluephagenesis TQ10 reflects differences in the growth rates between heterotrophic and autotrophic conditions, likely differences in the intracellular butyrate concentration and relatively high apparent kinetic constant of CvFAP [2].

Fig. 6
figure 6

Propane production of H. bluephagenesis TQ10 and/or H. rowanensis expressing plasmid borne CvFAPG462V with a (A) weak p7 and (B) strong p102 constitutive promoter. In part A, cultures (50 mL) were cultivated in LB60 medium (H. bluephagenesis TD1.0) or thiosulfate medium (H. rowanensis) for 6 h from a 1% inoculum at 30 °C and 200 rpm. Triplicate aliquots (1 mL) were sealed in 4 mL glass vials and incubated overnight at 30 °C and 180 rpm under a blue LED panel. For part B, H. rowanensis was cultivated as above in the two medium types with or without supplemental butyric acid (10 mM). Propane was quantified by Micro GC analysis of manual headspace samples. C Ectoine production of H. rowanensis in different growth media all with 10% (w/v) NaCl except the brine water which is ~ 20% NaCl. Cells were harvested by centrifugation and milked to release ectoine after 48 h in shake flasks at 30 °C. Ectoine and hydroxyectoine were quantified by HPLC. D Box plot of PHA production. PHB was accumulated over 48 h in LB60 and thiosulfate media in shake flasks at 30 °C. Cells from 50 mL were harvested by centrifugation and freeze dried. CDW was recorded and PHB concentration was determined by GC after derivatisation. Data are shown as the g PHB per g of freeze dried CDW starting material. HR = H. rowanensis; TD01 = H. bluephagenesis TD01 (wild-type isolate). The PHB titre and CDW data are available in Additional file 1: Table S2

Full size image

Surprisingly, switching to a higher strength promoter system yielded lower overall yields of propane under heterotrophic and chemoautotrophic growth conditions (Fig. 6b). This suggests the high expression of CvFAPG462V may impart some toxicity on the system. Overall, propane titres of H. rowanensis from CO2 (thiosulphate medium without butyric acid) were around tenfold lower than photoautotrophic propane production by Synechcocystis PCC 6803 (~ 11.1 ± 2.4 mg propane/L/day) [2]. However, the latter strain had been bioengineered to accumulate small chain fatty acids by knocking out the native fatty acyl ACP synthase gene (∆aas) and to co-express CvFAPG462V with a butyryl-ACP thioesterase from Bacteroides fragilis (Tes4) [59].

After establishing chemoautrophic and heterotrophic bio-propane production in H. rowanensis, we determined the levels of naturally occurring ectoine and PHA to see the potential of multiproduct generation. The highest titres of ectoine when H. rowanensis was cultivated on mineral enriched brine water with 1% (v/v) glycerol as the carbon source (11.31 ± 1.03 ectoine/DCW (%); Fig. 6c). Other enriched medium based on natural water systems gave good titres of ectoine (Fig. 6c). Cells grown chemoautotrophically on CO2 produced 20-fold lower titres of ectoine, with the degree of salinity determined to be important for ectoine production (Fig. 2B and Additional file 1: Fig. S6). Ensuring sufficiently high salinity under chemoautotrophic conditions will likely improve ectoine titres in H. rowanensis.

PHA production by H. rowanensis was essentially the same when cultivated under heterotrophic (LB60) or chemoautotrophic (thiosulfate) conditions (15.38 ± 8.01 vs 14.54 ± 9.72% PHA/DCW; Fig. 6d). These titres are slightly lower than comparable heterotrophic growth of H. bluephagenesis TD1.0 (21.90 ± 10.56% PHA/DCW), in the absence of glucose feeding. Therefore, optimal titres of PHA are generated under heterotrophic condition, where excess organic carbon can be supplied to stimulate PHA accumulation.

Further studies are required to optimise the CvFAP construct for H. rowanensis to increase recombinant propane titres in addition to naturally producing ectoine and PHA. Supplementing cultures with butyric acid is likely to switch the metabolism towards heterotrophic growth as Halomonas species are known to grow on butyric acid as a carbon source [2]. Therefore, while chemoautotrophic propane, ectoine and PHA have been demonstrated, it is likely that switching to conventional glucose fed and butyric acid supplemented cultivation will increase the titres of all three target compounds.

Conclusions and future perspectives

The recruitment of environmental isolates as de novo microbial chassis is not a new concept. The practice of screening for microorganisms of a specific genera or growth condition can sometimes lead to surprising growth phenotypes, which can open alternative approaches to achieve the desired bioproduction goal. The discovery of facultative chemoautotrophic H. rowanensis with relatively rapid growth on inorganic carbon sources and wastewater raises the possibility of dramatically reducing the carbon footprint of chemicals and fuels bioproduction by utilising industrial waste gases or bicarbonate from carbon capture technologies as the carbon (CO2) and energy (sulphur compounds) sources. This gives it an advantage as a potential robust industrial chassis for secondary product formation once it has been optimised to increase titres. The neutral pH optimum allows it to more easily express enzyme pathways that are not active at the high pH that H. bluephagenesis thrives at. The broader salt tolerance of H. rowanensis compared to H. bluephagenesis strains allows for non-sterile cultivation to be possible even at pH 7.0 as the culture thrives at the higher salinity of 10%.

Thiosulfate and inorganic sulphur contaminated wastewater can be found in the mining industry [60], steelworks [61], offshore gas, shale oil [62] and water dichlorination processes [63]. A potential scenario could be the capture of both industrial CO2 from the flue gas and thiosulfate-contaminated water found at steel mills as the basis for a bioproduction medium. This would effectively be a means to bioremediate thiosulfate while capturing CO2 emissions as both biomass and propane.

There are considerable cost savings to be achieved by exploiting the ability of halophiles to be cultivated under non-sterile growth conditions without the need for fresh water. The relative ease at which the Halomonas genetic toolbox was transmittable to an unmodified natural isolate highlights the potential of environmental screening to discover more unique potential microbial chassis. Overall, the utilisation of a chemoautotrophic halophilic industrial microbial chassis could improve the sustainability of the process in addition to an overall reduction in the carbon footprint. The generation of valuable biochemicals and fuels from CO2 could increase the energy security and could contribute towards global carbon management targets and help achieve clean air directives.