Potential of biological sulphur recovery from thiosulphate by haloalkaliphilic Thioalkalivibrio denitrificans

ABSTRACT The aim of this study was to investigate the potential for elemental sulphur recovery from sulphurous solutions under aerobic and anoxic conditions by haloalkalophilic Thioalkalivibrio denitrificans at 0.8–19.6 g S2O3 2−-S L−1 and 0.2–0.58 g NO2 L−1, respectively. The experiments were conducted as batch assays with haloalkaline (pH 10 and ≥ 14 g Na+ L−1) thiosulphate solution. Aerobically, the highest biotransformation rate of thiosulphate obtained was 0.03 h−1 at 8.5 g L S2O3 2−-S. Based on Monod model, the maximum substrate utilisation rate (qm) was 0.024 h−1 with half saturation constant (Ks) 0.42 g S2O3 2−-S L−1 at initial [S2O3 2--S] of 14 g L−1. S0 accumulated at [S2O3 2−-S] ≥ 1.5 g L−1 (10% yield at initial 9.5 g S2O3 2−-S L−1) and the highest S0 yield estimated with the model was 61% with initial [S2O3 2--S] of 16.5 g L−1. Anoxically, the maximum nitrite removal rate based on Monod modelling was 0.011 h−1 with Ks = 0.84 g NO2 − L−1. Aerobically and anoxically the maximum specific growth rates (µm) were 0.046 and 0.022 h−1, respectively. In summary, high-rate aerobic biotransformation kinetics of thiosulphate were demonstrated, whereas the rates were slower and no S0 accumulated under anoxic conditions. Thus, future developments of biotechnical applications for the recovery of S0 from haloalkaline streams from the process industry should focus on aerobic treatment. Highlights Haloalkaline S2O3 2− biotransformations kinetics by Thioalkalivibrio denitrificans Aerobic thiosulphate-S bioconversion up to 0.024 h−1 with Ks = 0.42 g S2O3 2−-S L−1 10% S0 yield with initial 9.5 g S2O3 2--S L−1 in aerobic condition Anoxic NO2 removal up to 0.01 h−1 with Ks = 0.84 g NO2 − L−1 GRAPHICAL ABSTRACT

-S L −1 and 0.2-0.58 g NO 2 L −1 , respectively. The experiments were conducted as batch assays with haloalkaline (pH 10 and ≥ 14 g Na + L −1 ) thiosulphate solution. Aerobically, the highest biotransformation rate of thiosulphate obtained was 0.03 h −1 at 8. L −1 . Aerobically and anoxically the maximum specific growth rates (µ m ) were 0.046 and 0.022 h −1 , respectively. In summary, high-rate aerobic biotransformation kinetics of thiosulphate were demonstrated, whereas the rates were slower and no S 0 accumulated under anoxic conditions. Thus, future developments of biotechnical applications for the recovery of S 0 from haloalkaline streams from the process industry should focus on aerobic treatment.

Introduction
To support circular economy, the recovery and recycling of sulphurous compounds (i.e. HS − , S 2 O 3 2− ) from different industrial streams (i.e. pulp and paper, petrochemical, mining and fertiliser) are gaining increasing attention [1][2][3][4][5]. These compounds can also cause environmental and health concerns if released without treatment [6]. From the economic point of view, the use of sulphurous compounds contributes directly to operational costs, by raising the need for other chemicals and water as well as by causing corrosion [7]. As an example, maintaining the Na/S balance is crucial for efficient chemical pulping, and recovery of sulphur compounds from the process would be beneficial. It would decrease the requirement of additional Na supply because S accumulates in the process more than Na [8]. In pulping, the Na/S balance is generally controlled by purging the electrostatic precipitator-ash (ESP-ash) from the recovery boiler which results in reduction of Na and S in the recycled streams [9]. Complementing large-scale chemical processes with the biological recovery of excess sulphur is a promising approach [10]. Oftentimes the industrial streams, like the ones from pulp and paper industry, have alkaline pH and high salt content [11]. Moreover, the oxygen supply in aerobic treatment of these concentrated solutions might become process limiting. Therefore, biological processes for sulphur recovery would preferably be based on anoxic or microaerophilic processes. Haloalkaline process/wastewater streams may contain other constituents in addition to sulphurous and sodium-based compounds. These streams such as spent sulphidic caustic from the petrochemical industry contains phenols and benzene that may be toxic to chemolithotrophic SOB as well as organo-sulphur compounds (for example, methanethiol) that potentially interfere with biotransformations of sulphurous compounds [12][13][14]).
Haloalkaliphilic sulphur oxidising bacteria (SOB) are potent organisms for sulphur recovery from industrial steams. Haloalkaliphilic SOB use inorganic sulphur compounds including sulphide, polysulphide, thiosulphate, polythionates and elemental sulphur as electron donor [15]. Most SOB grow aerobically. Biooxidation of partially oxidised sulphurous compounds coupled to NO 3 − or NO 2 − reduction is possible under anoxic conditions by some haloalkaliphilic species of SOB [16,17] ) [16]. Moreover, Thioalkalivibrio spp. has tolerance to high salinity (e.g. up to 4.3. M Na + ) and alkaline pH (up to 10.6) [15,18]. T. denitrificans uses oxygen or nitrite/nitrous oxide as an electron acceptor during oxidation of sulphurous compounds in microaerophilic and anoxic environments, respectively [17] (for a review, see [16]). Therefore, it can be a better option with haloalkaline sulphurous solutions than for example aerophilic T. versutus. Besides oxidising sulphide to sulphate, T. denitrificans can also disproportionate partially oxidised sulphur oxyanions to elemental sulphur and sulphate [17] and was, for these reasons, selected as a model organism for this study. Elemental sulphur would be the desired product of sulphide conversion due to its separability from liquid phase and potential uses in various fields [19]. For example, the produced elemental sulphur could be used as a fertiliser or electron acceptor for denitrification [19][20][21]. The biological sulphur recovery process Thiopaq (Shell-Paques) in which HS-is converted to S 0 by SOB in the presence of oxygen. This process can be used internally or for fertiliser production has been applied to recover sulphur from natural gas, refinery gas and synthetic gas since 1993 [22]. Moreover, S 0 can be applied in mining and metallurgy for the recovery and removal of metals from wastewaters via biosulphidogenesis [23]. Due to the hydrophilic nature of biologically produced sulphur, it is more readily biologically available than chemically produced sulphur [19,20].
The aim of this study was to determine thiosulphate biotransformation potential by microaerophilic/denitrifying T. denitrificans under haloalkaline conditions and for the recovery of elemental sulphur from saline and alkaline sulphurous streams. In case of toxic concentration of HS − , chemical oxidation of to S 2 O 3 2− could be used as pre-treatment prior to thiosulphate bioconversion step [21]. Therefore, thiosulphate was selected as a model sulphurous compound of this study. The biotransformation kinetics of T. denitrificans have not been comprehensively studied and especially not at high (up to 19.6 g L −1 ) thiosulphate concentrations. The earlier studies on T. denitrificans have focused on the growth kinetics [17], pH limitation and N 2 O reducing activity [17]. The sensitivity of T. denitrificans to NO 2 − has been reported by Sorokin et al. [17], but anoxic kinetics or the potential of elemental sulphur production have not been investigated. Therefore, the specific objectives of this study were the following: (i) determination of the biotransformation rates of thiosulphate and nitrite by T. denitrificans under aerobic and anoxic conditions, respectively; (ii) determination of the kinetics of elemental sulphur and sulphate formation at different initial concentrations of thiosulphate in aerobic condition; (iii) determination of elemental sulphur production yield by T. denitrificans at a chosen concentration by in aerobic conditions; (iv) determination of qPCR-based growth kinetics and yields of T. denitrificans in presence of oxygen or nitrite, and (v) model fitting of aerobic biotransformations.

Inoculum and growth medium
Thioalkalivibrio denitrificans strain ALJD was obtained from the German Collection of Microorganisms and Cell Cultures GmbH (DSMZ). The stock culture was routinely grown with the 925 alkaliphilic sulphur respiring medium recommended by DSMZ on an orbital shaker at 30 ± 1°C and 150 rpm under aerobic condition [24]. The growth medium included: 20 g L −1 Na 2 CO 3 , 10 g L −1 NaHCO 3 , 5 g L −1 NaCl, 1 g L −1 K 2 HPO 4 , 0.5 g L −1 KNO 3 , 0.05 g L −1 MgCl 2 and 2% (v/v) trace element solution (see preparation from [24]). As energy source, 4.5 g L −1 sterile filtered (0.2 µm polyethersulfone membrane syringe filter, VWR International, North America) S 2 O 3 stock solution was added to the medium of the stock culture. The inoculum used during both aerobic and anoxic experiments was actively growing. The biotransformation activity of the stock culture under anoxic condition (nitrite as electron acceptor) was seen to decrease by repeatedly transferring the culture to a fresh anoxic medium (data not shown), thus the aerobically grown stock culture was used as inoculum during all experiments. The stock solution of nitrite (as NaNO 2 ) that was added to the cultures at the beginning of the anoxic experiments, was purged with N 2 gas and sterile filtered (0.2 µm polyethersulfone membrane syringe filter, VWR International, North America) in an anoxic chamber. In addition, the cultures (including the inoculum) were purged with N 2 gas prior to the addition of nitrite stock in the anoxic chamber.

Kinetic experiments
The experiments in the presence of air were implemented as liquid cultures with a working volume of 100 mL in 250 mL Erlenmeyer flasks [24]. To enable air transfer to the flasks, the caps were loosened slightly. The alkaline growth medium was the same as described in Section 2.1. The culture preparation was done as previously described by Hajdu-Rahkama et al. [24]. In brief, different concentrations (0.8, 1.5, 3, 6, 8.5, 14, 16.5 and 19.6 g L −1 ) of thiosulphate-S were added to duplicate cultures with 10% (v/v) inoculum taken from the stock culture. All flasks were incubated for 14 days in an orbital shaker at 150 rpm and 30°C. In order to ensure comparability of the kinetic results at the different thiosulphate-S concentrations, the inoculum used was in the same growth phase.

Determination of sulphur formation
Quantitative determination of elemental sulphur formation under aerobic conditions was implemented as a separate batch experiment to enable validation of kinetic modelling of the elemental sulphur formation. The contents of the medium and growth phase of the 10% (v/v) inoculum were the same as used during the aerobic the kinetic experiments (Section 2.2.1). Altogether 12 identical cultures were prepared with 9.5 g L −1 concentration of S 2 O 3 2--S which was a mid-range concentration used during the kinetic experiments. After the removal of sample (6 mL) for sulphur compounds determination by ion-chromatography, the full solid content was collected by vacuum filtration of the whole culture volume. This sacrificial sample collection was carried out from duplicate cultures every second day. During this incubation, the temperature and shaking were the same as of the inoculum and the duration of the experiment was 10 days.

Anoxic experiments in the presence of nitrite
Anoxic experiment was implemented to compare the thiosulphate transformation kinetics with nitrite as an electron acceptor to the rates obtained with oxygen as the terminal electron acceptor. Nitrite was used because T. denitrificans is missing nitrate reductase and can thus utilise nitrate only as nitrogen source for biomass formation. T. denitrificans can use nitrite as electron acceptor at concentration at least up to 4.14 g NO 2 − L −1 , but only after prolonged adaptation to increasing nitrite concentrations [17]. The anoxic experiment was conducted in 160 mL serum bottles (60 mL working volume) at 30°C and 150 rpm in an arbitrary shaker. The growth medium was the same as used for the aerobic experiments. To enable the investigation of the toxicity of nitrite only, the initial concentration of S 2 O 3 2− in the serum bottles was 3.9 ± 0.2 g L −1 , as this concentration was found to be non-inhibitory (similar as of stock culture) during the aerobic experiments and resulted in excess thiosulphate concentration based on the stoichiometry of nitrite biotransformation. The used initial concentrations of nitrite were 0.2, 0.3 and 0.58 g L −1 . The 10% (v/v) inoculum added to each culture was in the same growth phase. The length of the incubations depended on the timing of full nitrite consumption. This duration was 3, 6 and 8 days for the cultures with initial nitrite concentrations of 0.2, 0.3 and 0.58 g L −1 , respectively.

Analytical methods
The concentration of thiosulphate (S 2 O 3 2− ), sulphate (SO 4 2 − ) and nitrite (NO 2 − ) were measured with ion chromatograph from 0.45 µm sterile filtered (Chromafil Xtra polyester membrane filter, Macherey-Nagel, Germany) samples as reported by Di Capua et al. [19]. Dionex IonPac AS22 anion exchange column (Thermo Scientific) was used with the ion-chromatography. The quantity and quality of the elemental sulphur formation were measured from samples that were vacuum filtered (1.2 µm GF/C glass microfiber filter, Whatman) and dried overnight at 105°C. The sulphur content of these samples was determined by using elemental analyser (Flash Smart, Thermo Fischer Scientific) coupled to a Thermal Conductivity Detector (TCD and supplied with helium as carrier gas [24]. The initial and end pH of the batch cultivations was measured with a pH 3210 metre (WTW, Germany) and SenTix 81 pH-electrode (WTW, Germany). The change in biomass concentration was estimated from the starting and endpoint culture sample 16S rRNA copy numbers. For modelling purposes, the biomass concentration is typically measured as dry weight (volatile suspended solids, VSS), protein content and/or total nitrogen [12,18,25,26]. However, measuring the VSS content is not possible in the presence of elemental sulphur which has a boiling point of 440°C. Total-N content quantification can be challenging from low biomass concentration. Therefore, quantitative real-time polymerase chain reaction (qPCR) was used for the determination of the biomass concentration as it has been commonly used in microbial ecology studies [26,27]. The DNA samples were taken after the same durations (14 days) from all aerobic cultures, while in the case of anoxic cultures after all, nitrite was consumed (4, 6, 9 days with 0.2, 0,3 and 0.58 g L −1 NO 2 − concentrations, respectively). The copy numbers were measured with quantitative polymerase chain reaction (qPCR). Prior to the qPCR, the DNA was extracted from cell pellets (2 mL samples centrifuged at 2800 g and 4°C for 15 min) by using DNeasy PowerSoil Kit (Qiagen). The qPCR was conducted with Step One Plus Real-Time PCR (AB Applied Biosystems) as reported by Rinta-Kanto et al. [27]. The qPCR gene copy number of 5.8, which is the average of Gammaproteobacteria [28], was used for the conversion of copy numbers to cell numbers.
Monitoring the dissolved oxygen (DO) concentration in the shake flasks with pure culture was not possible aseptically and therefore was not done in this study.

Kinetic calculations
The kinetic calculations applied on the results of aerobic experiments were similar as described by Hajdu-Rahkama et al. [24]. The calculations used were as summarised in Table 1.

Thiosulphate and nitrite utilisation kinetics
The biological substrate utilisation rate (SUR) is directly proportional to the active microorganism concentration [29] which was also the case in our bio-assays ( Figure S1). Factors such as concentration of the substrate, possible inhibitory compounds, and the environmental conditions (temperature, pH, pressure, etc.) that influence the concentration of microorganisms also increases the volumetric reaction rate as, for example, demonstrated in our earlier research [30,31]. Moreover, aerobic substrate utilisation can also be limited by the mass transfer of oxygen and nutrient availability [32,33].
The kinetics of both S 2 O 3 2− -S and NO 2 − utilisation by T. denitrificans were described by Monod equation (Equation (1)) [29][30][31]: -S] to [S 0 ]. d μ is the specific growth rate, μ m is the maximum specific growth rate and Y m is the yield. The yield was only calculated with the aerobic experiments.
where q is specific SUR (g-S/ (g-VSS h) −1 or simply h −1 ), q m is the specific maximum substrate (S 2 O 3 2− -S) utilisation rate or specific nitrite (NO 2 − ) reduction rate (h −1 ), respectively, and K s is half saturation concentration (g L −1 ) of the analyte. As it was mentioned before, the SUR is proportional of the biomass concentration, thus the kinetic equation takes the following form: and where X is the biomass concentration as g-VSS L −1 .
More information about how the kinetic constants were obtained can be seen from S1. Besides using Monod modelling for the kinetic calculations of S 2 O 3 2− -S utilisation, Haldane model has been also applied to see if there is inhibition by the substrate (Equation (4)) with the aerobic batch cultures: The lag phases of biotransformations observed at different initial thiosulphate concentrations were omitted in the model fitting to the experimental data.

Product formation kinetics under aerobic conditions
The aerobic biotransformation reactions of thiosulphate together with their Gibbs-free energy changes have been given in Equations (5)- (9). Depending on the available oxygen concentration, thiosulphate is mainly converted to elemental sulphur (Equation (5)) and sulphate (Equation (6)) by T. denitrificans. In case oxygen is not limited, elemental sulphur is further converted to sulphate (Equation (7)). Although unlikely, some of the elemental sulphur may first become oxidised to sulphite (Equation (8)) which is then further oxidised to sulphate by T. denitrificans (Equation (9)). [25,34,35]. Moreover, Ang et al. [25] has also reported the formation of thiosulphate as metabolic intermediate of elemental sulphur oxidation to sulphate by Thioalkalivibrio versutus: According to Equations (3) and (4), conversion of thiosulphate produces two fractions that are SO 4 2− (f 1 ) and S 0 (f 2 ). A detailed description of this calculation of the two fractions was reported by Hajdu-Rahkama et al. [24].

Product formation kinetics under anoxic conditions
Under anoxic conditions with nitrite as electron acceptor, biotransformation of thiosulphate by T. denitrificans is shown in Equation (10) [10]: Based on this pathway, the formation of 1 mol SO 4 2− requires 1.33 mol of NO 2 − . Sorokin et al. [17] reported nitrous oxide (N 2 O) formation during reduction of NO 2 − to N 2 gas by T. denitrificans. The equations of denitrification of gaseous NO 2 − to N 2 gas (Equations (11)-(13)) are as follows:

Growth of T. denitrificans at different thiosulphate and nitrite concentrations
Similarly, as in Hajdu-Rahkama et al. [24] the cell growth was estimated based on the results of qPCR copy numbers. The copy number was converted to g L −1 by using 6.25 × 10 −10 g dry weight for cell formula of C 5 H 7 NO 2 [36]. Then, the specific growth rates (μ, h −1 ) were calculated by using Monod (Equation (14)). It was not possible to calculate the K s from the results of [S 2 O 3 2 − -S] of the anoxic experiments, thus also with the anoxic specific growth rate calculation, the K s from the aerobic experiments were used. The yield with aerobic condition was calculated as reported by Hajdu-Rahkama et al. [24]: where μ is the specific growth rate calculated from experimental data and μ m is the maximum specific growth rate (h −1 ). The growth of the biomass and the consumption of thiosulphate are connected as follows: where Y is the biomass growth yield (g L −1 biomass/ g L −1 S 2 O 3 2− -S or g biomass/ g S 2 O 3 2− -S).

Model validation with experimental data
At the end of this study, the SUR, sulphate production rate (SPR 1 ) and elemental sulphur production rate (SPR 2 ) kinetic models for the aerobic experiments were statistically verified with experimental data from the sulphur formation aerobic batch experiments (see Section 2.5.2). For this verification, regression analysis was applied.

Results and discussion
Biotransformations of thiosulphate by T. denitrificans in aerobic and anoxic conditions were studied and the batch experimental data was used to derive the SUR model. With aerobic biotransformation results, this SUR model was further used to create the SPR models (sulphate and elemental sulphur production). The aerobic biotransformation models were validated with the data of an independent batch experiment. Moreover, both aerobic and anoxic growth kinetics of T. denitrificans were determined.

Kinetics of thiosulphate biotransformation
As shown in Figure 1, thiosulphate was biotransformed at all studied initial substrate concentrations (0.8-19. -S] of 1.5 g L −1 and higher, elemental sulphur was produced and based on visual observations the quantity increased with increasing thiosulphate concentration. Once thiosulphate was removed at low (1.5-6 g S 2 O 3 2− -S L −1 ) initial concentrations, the elemental sulphur was further oxidised to sulphate, resulting in the removal of sulphur particles ( Figure S4). At initial thiosulphate-S concentrations from 0.8 to 3 g L −1 , thiosulphate was completely biotransformed within 14 days while this thiosulphate conversion efficiency was only 76%, 70%, 39%, 35% and 30% at initial substrate concentrations of 6, 9, 14, 16.5 and 19.8 Hajdu-Rahkama et al. [24] studied the thiosulphate biotransformations of T. versutus using similar experimental design conditions (pH 10 and 0.6-1.2 Na + ) and thiosulphate concentrations (0.8-17.6 g S 2 O 3 2− -S L −1 ) at 150 rpm and 30°C. In their study, the thiosulphate utilisation rate increased from 0.03 to 0.08 h −1 by increasing the thiosulphate concentration while in this study with T. denitrificans, the highest SUR was only 0.03 h −1 . Sorokin et al. [17] reported severe growth inhibition of T. denitrificans by forced aeration in batch culture. Thus, T. denitrificans as a microaerophile is likely more sensitive to oxygen than the aerobic T. versutus.
Monod fitting of the experimental data resulted in q m of 0.025 h −1 and K s of 0.42 g S 2 O 3 2− -S L −1 (Figure 2(a) and Equation (3) (Figure 1(b-d)). The Haldane model (Equation (4)  The sulphur formation started after 4 days and increased   Figure 2(c)). The measured sulphate formation, especially at initial 0.2 and 0.3 g NO 2 − L −1 , was different than predicted from the stoichiometry (Equation (10)). This indicates other fates for nitrite such as reduction to gaseous N 2 O followed by partial loss to gas phase due to stirring. No elemental sulphur was formed based on visual observations and therefore, SPR was not modelled. The substrate inhibition constant or K i value estimated by using Haldane model was 30 g NO 2 − L −1 .

Sulphate and elemental sulphur formation
During the anoxic experiments, the highest conversion of S 2 O 3 2− -S to SO 4 2− of 60% could be explained by nitrite reduction at initial 0.3 g NO 2 − L −1 (Figure 4, Equation (10)). Sulphate production reduced by 21% when increasing the initial NO 2 − concentrations to 0.58 g L −1 . The sulphate production with initial 0.2 g NO 2 − L −1 was 56%.

Estimation of thiosulphate biotransformation under aerobic conditions
The fractions of thiosulphate biotransformed into SO 4 2− -S and S 0 were calculated by using the results of kinetic experiments (Section 2.2.1) and the SUR models (Section 3.1.1). Finally, the models were validated with the results of the independent batch (Section 3.1.2).

Estimation of SPR
At initial S 2 O 3 2--S of 0.8 g L −1 and below no elemental sulphur was formed, as shown by the calculated fractions of f 1 and f 2 ( Figure 5(a)). The highest f 2 formation (61%) was obtained with initial 16.5 g S 2 O 3 2--S L −1 while above this concentration it decreased. The yields of S 0 formation as a function of time were as shown in Figure 5(b). In the study of Hajdu-Rahkama et al. [24] with T. versutus, this highest yield was 45% with initial 17.6 g S 2 O 3 -S L −1 when the lag phases were not omitted from the kinetic calculations under similar conditions. Calculating the sulphur formation yield similarly as with T. versutus, the highest S 0 yield by T. denitrificans would be close to the one by T. versutus.

SUR and SPR model validation
The validation of the SUR and SPR model parameters was done by using the experimental results with initial 9.5 g S 2 O 3 2− -S L −1 concentration (Section 3.1.2). The results of model validations are shown in Figure 6. The regression analysis (confidence bound to 95%) resulted in high correlation (R 2 > 0.95) between the kinetic models and the experimental data.

Growth of T. denitrificans
The kinetics of the growth of T. denitrificans was estimated by using qPCR copy numbers and K s from the SUR model.

Growth kinetics under aerobic condition
As shown in Figure 7 Figure 7(b) shows that the rate of thiosulphate biotransformation did not increase after 0.03 h −1 (initial 8.6 g S 2 O 3 -S L −1 ) with the increasing biomass concentration. This indicates that a third factor, in addition to thiosulphate and biomass concentrations, controlled the overall biotransformation rate. In shake flask bio-assays, aeration is intensive and we suggest that dissolved oxygen concentration was actually the controlling factor for the microaerophilic T. denitrificans. Experimentally using batch bio-assays demonstration of this phenomenon is very challenging and requires continuous-flow bioreactor experimentation. Neither sulphate nor elemental sulphur are toxic and therefore, product inhibition is out of question in this case. Different kinetic parameters from this and other studies with similar haloalkaline SOB as T. denitrificans are summarised in Table 2. In aerobic condition, T. versutus has much higher K s (1.74 g S 2 O 3 2− -S L −1 ) than T. denitrificans (0.42 g S 2 O 3 2− -S L −1 ), higher q m (+0.059 h −1 ) and higher yield (+0.09 g cell/g S 2 O 3 2− S) of sulphur formation. Table 3 compares the results obtained with initial 2.5 g S 2 O 3 2− -S L −1 in aerobic and anoxic conditions (containing 0.2-0.56 g NO 2 − L −1 ). With 2.5 g S 2 O 3 2− -S L −1 , the aerobic µ was higher (0.04 h −1 ) than that of the anoxic with nitrite (0.02 h −1 ). At slightly higher substrate concentration (2.56 g S 2 O 3 2− -S L −1 ), Sorokin et al. [17] reported µ m of 0.045 h −1 with N 2 O (3.96 g L −1 ) based on their batch bio-assays. Further, they also reported anoxic (with NO 2 − ) growth rate of 0.038 h −1 but not substrate utilisation kinetics by T. denitrificans in a continuous chemostat culture [17]. Therefore, the chemostat growth rates were lower than those obtained in our batch assays (0.046 h −1 ). Further studies are needed to optimise electron acceptor supply in anoxic bioreactors for thiosulphate biotransformation.

Limiting factors of aerobic biotransformations
This study showed high yield elemental sulphur accumulation by T. denitrificans in the presence of sufficient thiosulphate and oxygen (from air) concentrations. In industrial scale applications, bioprocesses are always open systems. Based on our earlier work  with T. versutus [24] and the results of this study, both T. denitrificans and T. versutus would be likely catalysts of thiosulphate biotransformations. Although the rate of aerobic biotransformation is higher by T. versutus, in oxygen-limited conditions, the application of microaerophilic T. denitrificans can be more suitable.
The experimental design (shake flasks) and pure culture (a requirement for aseptic conditions) did not allow monitoring and control of DO concentration although it is an important variable that influences the final product formation in thiosulphate biotransformation. Therefore, the DO concentration effects and optimisation for elemental sulphur formation by T. denitrificans should be delineated in bioreactor studies that allow the possibility for DO control and continuous monitoring. Adjusting the DO concentration to an adequate level is crucial when the aim is to produce elemental sulphur, thus preventing its further oxidation to sulphate [21,38]. At low DO concentrations, which are preferred for S 0 formation, reading the actual values is often challenging, therefore controlling the oxygen supply based on the oxidation redox potential (ORP) is a better approach [21,39]. In addition, bioreactor provides steady-state conditions and thus, gives additional information about the practical applicability of this bioprocess. The desired DO levels can be maintained in continuous-flow bioreactors such as fluidised bed bioreactor where completely mixed conditions are maintained via high-rate recirculation [40]. This has been demonstrated under haloalkaline conditions in a Thioalkalivibrio versutus amended fluidised bed bioreactor [41]. In a practical application, the bioreactor would serve as a kidney removing the excess/ accumulating sulphur from the process stream. Some of the haloalkaline streams may contain organic constituents that can be toxic towards SOB [13] and therefore, the potential inhibitory effects of these constituents should be determined in future studies.

Conclusions
Under haloalkaline conditions (∼pH 10 and 14-28 g Na + L −1 ) aerobic and anoxic thiosulphate biotransformation batch bio-assays with Thioalkalivibrio denitrificans used in this work resulted in the following conclusions: 4. The maximum aerobic and anoxic specific growth rates were 0.046 and 0.022 h −1 , respectively, which *As S 2 O 3 2− -S; N.D.: not determined; N.R.: not reported. **As g L −1 h −1 . Table 3. Comparison of the kinetic constants of T. denitrificans under aerobic and anoxic conditions obtained in this study.  indicate partial inhibition by nitrite. The highest aerobic growth yield was 0.22 g cells/ g S 2 O 3 2− -S. 5. In summary, aerobic/microaerobic biotransformations producing elemental sulphur under haloalkaline conditions have potential for development of sulphur recovery from saline and alkaline industrial sulphurous streams.