Boosting Dibenzothiophene Biodesulfurization Through Implantation of a Refactored DBT Pathway in a Tailored Pseudomonas putida Chassis

ABSTRACT

This study reports the efficacy of a rationally designed Pseudomonas putida strain to bring about the specific removal of S atoms from dibenzothiophene (DBT), the model heterocyclic sulfur‐containing component of raw petroleum. The emphasis on DBT as a model compound stems from its prevalence in fossil fuels and its resistance to hydrodesulfurization, which positions it as a critical target for improving biodesulfurization technologies. To this end, we explored the combinatorial space of the dsz operon of the naturally occurring strain Rhodococcus qingshengii IGTS8—known to achieve dibenzothiophene degradation—by re‐engineering the native regulation of the operon, generating permutations of the order of the cognate genes and their ribosomal‐binding sites, testing the effects of multicopy versus monocopy doses and introducing the resulting constructs in the tailor‐made host. The combination that emerged as best in terms of catalytic efficacy, moderate physiological burden, and durability was one in which the original dsz operon was refactored by [i] reordering its native gene order to dszBCA, [ii] decompressing their naturally occurring translational coupling with optimised ribosomal‐binding sites, [iii] engineering its constitutive expression with a heterologous promoter and [iv] inserting the thereby refactored pathway in the Tn7 site of the genome‐edited strain P. putida EM384, which is optimised for greater stability and hosting harsh redox reactions. The resulting P. putida DS006 exhibited exceptional DBT desulfurization activity as well as efficiency in model biphasic biodesulfurization systems.

This study engineered a Pseudomonas putida strain to efficiently remove sulfur from dibenzothiophene (DBT) by reordering and optimising the dsz operon from Rhodococcus qingshengii IGTS8, enhancing catalytic activity and stability. The modified strain, P. putida DS006, showed high desulfurization efficiency in model biodesulfurization systems.

1. Introduction

In addition to CO₂ release, the combustion of fossil fuels poses an equally significant environmental challenge due to the concurrent emission of sulfur oxides (SOx), which contribute to acid rain and pose public health risks (Sadare and Daramola 2023). Current stringent regulations mandate the use of low‐sulfur fuels, while the depletion of low‐sulfur crude oil reserves necessitates the development of effective desulfurization technologies. Traditional hydrodesulfurization (HDS) processes, while effective, are costly and environmentally harmful, requiring high temperatures, pressures, and the use of hydrogen gas, which leads to the production of toxic byproducts like hydrogen sulfide (H₂S)—conditions that become more intensive when the deep removal of heterocyclic S‐compounds is sought (Martínez et al. 2017).

Biodesulfurization (BDS) has emerged as a promising alternative to HDS, particularly for the removal of recalcitrant sulfur compounds like dibenzothiophene (DBT) and its alkylated derivatives that are resistant to deep hydrodesulfurization. BDS employs biphasic systems in which microbial suspensions acting as biocatalysts (aqueous phase) come in contact with oil products (organic phase) and selectively remove sulfur from these compounds, without breaking carbon–carbon bonds, thus preserving the fuel's energy content (Bachmann et al. 2014). Among various microorganisms explored, Pseudomonas putida stands out due to its robustness and adaptability to diverse environmental conditions, making it an ideal candidate for industrial applications (Martínez et al. 2022).

The core of the BDS process involves the 4S metabolic pathway, which converts DBT into 2‐hydroxybiphenyl (2‐HBP) and inorganic sulfate, through a series of enzymatic reactions encoded by the dszABC genes (Figure 1). This operon was initially identified in the environmental Gram‐positive bacterium Rhodococcus qingshengii IGTS8, formerly known as Rhodococcus erythropolis (Thompson et al. 2020). However, homologues of the dsz genes are present across various genera (Corynebacterium, Brevibacterium, Gordonia and Nocardia) (Gallagher et al. 1993; Matos Neto et al. 2023; Thompson et al. 2020). Despite its widespread abundance, the structure of the operon represents a rather non‐optimal configuration concerning gene order, gene overlaps, and a sulfur‐repressed promoter. Although in natural environments, this dsz operon configuration may reflect evolutionary trade‐offs that prioritise adaptability and survival over efficiency, it places under question its suitability for industrial applications (Monticello 2000; M. Z. Li et al. 1996).

FIGURE 1.

Genetics and metabolism of biodesulfurization in Rhodococcus qingshengii IGTS8. (A) Genetic organisation of the natural dsz operon. The genes dszA, dszB, and dszC, which are located on the plasmid pSOX (GenBank: CP029300.1, partially annotated) encode the enzymes involved in the 4S metabolic pathway. The promoter of the operon, Pdsz, is represented by a yellow arrow (M. Z. Li et al. 1996). The green semicircle, with its DNA sequence in bold below, depicts the predicted ribosome binding site (RBS) of the dszA gene, while the predicted RBS sequences of dszB and dszC are shown in bold. The green arrow represents the 3´end of dszA, while the light blue arrow indicates the start of the dszB gene. (B) The dszD gene, which is located in the chromosome (GenBank: NZ_CP029297.1, partially annotated) encodes for an oxidoreductase that provides for the necessary reducing power. It should be noted that while the accession numbers correspond to the complete plasmid and chromosomal sequences, the specific coordinates of the dsz genes are not provided due to the partial annotation of these sequences. (C) The enzymatic stepwise conversion of dibenzothiophene (DBT) into 2‐hydroxybiphenyl (2‐HBP). This metabolic conversion is carried out via the sequential actions of DszC, DszA, and DszB, while the reductive power required is supplied through the action of the NADH‐FMN oxidoreductase, DszD. The pathway includes the intermediate formation of dibenzothiophene sulfoxide (DBTO) and dibenzothiophene sulfone (DBTO₂), demonstrating the sulfur‐specific cleavage process while preserving the carbon skeleton of the substrate.

Despite advancements in identifying microorganisms with BDS activity and optimising their growth conditions, the industrial application of BDS is hindered by the low activity and stability of these organisms in petroleum‐based environments. To address these challenges, enhancing bacterial traits such as solvent tolerance and desulfurization efficiency through genetic engineering is crucial (Kilbane 2006).

This study aims to enhance the BDS capabilities of P. putida by first exploring different strains, natural and optimised ones, and then rearranging the dsz genes and evaluating their performance in comparison to the model strain R. qingshengii IGTS8 under bioreactor and biphasic system conditions. By leveraging deep genome engineering, we aim to create robust P. putida strains with improved desulfurization efficiency and stability, paving the way for more efficient and economically viable biodesulfurization processes. This research contributes to the field by demonstrating the potential of genome engineering to optimise microbial biocatalysts for industrial BDS applications, thus offering a sustainable alternative to traditional desulfurization methods.

2. Results and Discussion

2.1. Assessing 2‐HBP Resistance in Different Bacterial Hosts

To identify the best bacterial host for the BDS process, we evaluated the ability of various bacteria (Table 1) to withstand high concentrations of 2‐HBP, a key metabolite in the BDS reaction (Figure 1). We selected E. coli MG1655 (Blattner et al. 1997), P. putida S12 (Hartmans et al. 1990), KT2440 (Bagdasarian et al. 1981), and two genome‐reduced variants of KT2440: EM42 and EM384 (Martínez‐García et al. 2014). The main difference between EM42 and EM384 is that the latter retains the flagella, which has been linked to solvent tolerance (Segura et al. 2001, 2004).

TABLE 1.

Bacterial strains used in this study.

Strain	Description/relevant characteristics	References
Escherichia coli
DH5α	Cloning host; F− λ− endA1 glnX44(AS) thiE1 recA1 relA1 spoT1 gyrA96 (NalR) rfbC1 deoR nupG Φ80(lacZΔM15) Δ(argF‐lac)U169 hsdR17(rK—mK+)	(Grant et al. 1990)
DH5α λpir	Cloning host; λ pir lysogen of DH5α	(Platt et al. 2000)
CC118 λpir	Cloning host; Δ(ara‐leu) araD Δ lacX74 galE galK phoA20 thiE1 rpsE (SpR) rpoB (Rif R) argE (Am) recA1	(Herrero et al. 1990)
HB101	Mating helper strain; F− λ− hsdS20(rB—mB−) recA13 leuB6 (Am) araC14 Δ(gpt‐proA)62 lacY1 galK2 (Oc) xyl‐5 mtl‐1 thiE1 rpsL20 (SmR) glnX44(AS)	(Boyer and Roulland‐dussoix 1969)
MG1655	F− λ− ilvG rfb‐50 rph‐1	(Blattner et al. 1997)
Pseudomonas putida
KT2440	Wild‐type strain	(Bagdasarian et al. 1981)
EM42	KT2440 derivative; Δprophages ΔTn7 ΔendA1 ΔendA2 ΔhsdRMS Δflagellum ΔTn4652	(Martínez‐García et al. 2014)
S12	Wild‐type strain	(Hartmans et al. 1990)
EM384	KT2440 derivative; Δprophages‐ΔTn7 ΔendA1 ΔendA2 ΔhsdRMS	(Martínez‐García et al. 2014)
DS001	EM384 with the pTn7‐M‐DS001 integrated into the attTn7 site	This work
DS002	EM384 with the pTn7‐M‐DS002 integrated into the attTn7 site	This work
DS003	EM384 with the pTn7‐M‐DS003 integrated into the attTn7 site	This work
DS004	EM384 with the pTn7‐M‐DS004 integrated into the attTn7 site	This work
DS005	EM384 with the pTn7‐M‐DS005 integrated into the attTn7 site	This work
DS006	EM384 with the pTn7‐M‐DS006 integrated into the attTn7 site	This work
DS007	EM384 with the pTn7‐M‐DS007 integrated into the attTn7 site	This work
DS008	EM384 with the pTn7‐M‐DS008 integrated into the attTn7 site	This work
Rhodococcus qingshengii
IGTS8	Wild‐type strain	ATCC 53968

Bacterial growth was quantified under varying concentrations of the stressor 2‐HBP (600, 800, and 1100 μM) and in control conditions (without the stressor). Figure 2 illustrates the growth responses of all strains at the different 2‐HBP concentrations.

FIGURE 2.

Growth curves of the different bacterial strains under varying concentrations of 2‐hydroxybiphenyl (2‐HBP). The growth of each strain was monitored over 24 h and measured as optical density at 600 nm (OD₆₀₀). The experiment was performed twice with four technical replicates, and the average and standard deviations (STDEV) plotted.

P. putida EM384 consistently demonstrated superior growth across all 2‐HBP concentrations. At 600 μM, EM384 had the fastest growth, followed by KT2440 and S12, while E. coli MG1655 and EM42 showed much slower growth. This pattern continued at 800 μM, with EM384 maintaining robust growth, outperforming KT2440 and S12, while MG1655 and EM42 showed minimal growth. At 1100 μM, EM384 again showed the highest growth capacity, while KT2440 and S12 experienced moderate inhibition. MG1655 and EM42 were severely inhibited in this condition. It is interesting to note that the poor behaviour observed for EM42 is probably caused by the absence of flagella.

These results highlight the superior resilience of P. putida EM384 under high 2‐HBP concentrations, suggesting that its genetic modifications offer advantages in toxic environments. As a result, P. putida EM384 was selected as the optimal chassis for evaluating biodesulfurization capacity with various dsz operon configurations.

2.2. Optimising Dsz Operon Arrangement to Enhance Biodesulfurization

The biodesulfurization operon genes dszA (ID: AY294402.1), dszB (ID: AY294403.1) and dszC (ID: AY294404.1) from R. qingshengii IGTS8 were cloned into the pSB1C3 repository vector (Table S1) using the SevaBrick Assembly approach (Damalas et al. 2020). This allowed the exploration of various gene arrangements and ribosome binding site (RBS) combinations to optimise operon functionality. Most constructs (DS003 to DS008) contained an optimised and strong RBS (BBa_B0034) (Elowitz and Leibler 2000). However, DS001 contained the native RBS sequence along with an additional RBS designed by Gallardo et al. (1997), while DS002 retained only the native RBS (Figure 3A). These two provided controls to assess the effects of different RBS designs and gene arrangements.

FIGURE 3.

The different dsz operon rearrangements in multicopy plasmids and their impact on growth and biodesulfurization. (A) Schematic representation of the eight dsz operon cassettes (DS001p to DS008p) used in the study. Gene expression in all constructs is driven by the IPTG inducible Ptrc promoter (green arrow). The light orange semicircle depicts the native RBS, the dark orange, the one used in (Gallardo et al. 1997), while the purple represents the optimised BBa_B0034. For clarity, the native RBSs of dszB and dszC are not drawn. All genetic constructs were assembled using the SevaBrick approach. The constructs are not drawn to scale. (B and C) The growth curves of each strain carrying the corresponding plasmid in LB and minimal medium, respectively. (D and E) Desulfurization activity of the cells carrying the different plasmid operons when grown in LB and minimal medium, respectively. All strains were induced with IPTG at the beginning of the experiment (t = 0). Desulfurization activity was measured in cells collected at the mid‐exponential phase (8 h of growth) and quantified in units per milligram of dry cell weight (DCW) based on the levels of produced 2‐HBP and DBTO₂ under the standard biodesulfurization assay conditions (described in the Experimental Procedures section). Statistical analysis using one‐way ANOVA showed significant differences in DBTO₂ and 2‐HBP production in both LB and minimal medium (p < 0.0001). Each experiment was performed twice and the average and standard deviation (STDEV) plotted.

Constructs DS003 to DS008 maintained the optimised RBS while varying gene order to explore operon efficiency. All constructs were then cloned into a broad‐host‐range plasmid with IPTG‐controlled expression (lacI‐Ptrc; Table 2) and transferred to P. putida EM384.

TABLE 2.

Plasmids used in this study.

Name	Relevant features	References
pRK600	oriColE1, tra + mob + of RK2; CmR	(Keen et al. 1988)
pSEVA234	pBBR1 (lacIqPtrc); KmR	(Martínez‐García et al. 2023)
pTn7‐M	oriR6K; MCS flanked by Tn7L and Tn7R; KmR GmR	(Zobel et al. 2015)
pTNS2	oriR6K; Tn7 transposose (tnsA‐tnsB‐tnsC‐tnsD); ApR	(Choi et al. 2005)
pSEVAb234 (DS001p)	ori(pBBR1); lacI q ‐P trc  → dszA_dszB_dsz; KmR	This study
pSEVAb234 (DS002p)	ori(pBBR1); lacI q ‐P trc  → dszA_dszB_dszC; KmR	This study
pSEVAb234 (DS003p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszA_BBa_0034‐dszB_BBa_0034‐dszC; KmR	This study
pSEVAb234 (DS004p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszA_BBa_0034‐dszC_BBa_0034‐dszB; KmR	This study
pSEVAb234 (DS005p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszB_BBa_0034‐dszA_BBa_0034‐dszC; KmR	This study
pSEVAb234 (DS006p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszB_BBa_0034‐dszC_BBa_0034‐dszA; KmR	This study
pSEVAb234 (DS007p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszC_BBa_0034‐dszA_BBa_0034‐dszB; KmR	This study
pSEVAb234 (DS008p)	ori(pBBR1); lacI q ‐P trc  → BBa_0034‐dszC_BBa_0034‐dszB_BBa_0034‐dszA; KmR	This study
pTn7‐M‐DS001	oriR6K; lacI q ‐P trc  → dszA_dszB_dszC; KmR GmR	This study
pTn7‐M‐DS002	oriR6K; lacI q ‐P trc  → dszA_dszB_dszC; KmR GmR	This study
pTn7‐M‐DS003	oriR6K; lacI q ‐P trc  → BBa_0034‐dszA_BBa_0034‐dszB_BBa_0034‐dszC; KmR GmR	This study
pTn7‐M‐DS004	oriR6K; lacI q ‐P trc  → BBa_0034‐dszA_BBa_0034‐dszC_BBa_0034‐dszB; KmR GmR	This study
pTn7‐M‐DS005	oriR6K; lacI q ‐P trc  → BBa_0034‐dszB_BBa_0034‐dszA_BBa_0034‐dszC; KmR GmR	This study
pTn7‐M‐DS006	oriR6K; lacI q ‐P trc  → BBa_0034‐dszB_BBa_0034‐dszC_BBa_0034‐dszA; KmR GmR	This study
pTn7‐M‐DS007	oriR6K; lacI q ‐P trc  → BBa_0034‐dszC_BBa_0034‐dszA_BBa_0034‐dszB; KmR GmR	This study
pTn7‐M‐DS008	oriR6K; lacI q ‐P trc  → BBa_0034‐dszC_BBa_0034‐dszB_BBa_0034‐dszA; KmR GmR	This study

The capacity of the plasmid‐borne P. putida strains (Figure 3A) was tested in both rich (LB) and minimal (M9cit) media. Among these, DS005p and DS006p yielded the highest biomass in LB (~1.2 mg/mL; Figure 3B) and in M9cit (~0.6 mg/mL; Figure 3C), indicating a lower metabolic burden of these genetic arrangements. DS008p, when compared to strains DS005p and DS006p, showed lower growth in LB (~0.9 mg/mL) but comparable growth in M9cit (~0.6 mg/mL), possibly due to a more balanced enzyme expression under nutrient‐limited conditions (Bentley et al. 1990; Glick 1995). DS001p and DS002p consistently showed the lowest biomass (LB: ~0.8 mg/mL and M9cit ~0.3 mg/mL), suggesting their gene arrangements hindered metabolic efficiency by imposing a higher metabolic burden.

Besides, the BDS activity of each strain was measured by monitoring the simultaneous production of DBTO₂ (blue bars) and 2‐HBP (orange bars) with exponential phase cells grown in both LB (Figure 3D) and M9cit (Figure 3E) media, using the standard BDS resting cell assay described in the Experimental Procedures section.

In LB, strains DS004p, DS005p, and DS006p demonstrated the highest desulfurization activity, at levels ~25 Units/mg_DCW, indicating optimal operon function since only a minimal amount of DBTO₂ was detected in the assay. DS001p cells, however, accumulated mostly DBTO₂ with minimal HBP production, likely due to suboptimal dszB expression (Monticello 2000) or imbalanced operon components (G. Q. Li et al. 2008; M. Z. Li et al. 1996).

In M9cit, strains DS005p and DS006p also maintained relatively high BDS activity (~15 Units/mg_DCW), while DS001p accumulated mostly DBTO₂. The high DBTO₂ levels in DS001p suggest sensitivity to the plasmid‐encoded nature of the constructs or differences in regulatory mechanisms, leading to incomplete conversion of DBTO₂ to HBP (Radde et al. 2024). The poor performance of DS001p across both media types suggests a need for better gene expression or enzyme balance to avoid metabolic bottlenecks. The variation in growth and desulfurization activity is linked to the different gene configurations in the plasmid‐encoded dsz operons (Kilbane 2006). Strains DS004p, DS005p and DS006p, which showed the highest biomass and BDS activity, likely benefit from gene arrangements that enhance the expression of the dsz genes or improve protein–protein interactions (Hirasawa et al. 2001; Oldfield et al. 1997) leading to efficient desulfurization and full conversion of DBT to 2‐HBP by the corresponding resting‐cell biomass. In contrast, DS001p and DS002p, which performed poorly, likely have suboptimal gene arrangements that reduce pathway efficiency, causing DBTO₂ accumulation and incomplete desulfurization, especially under the plasmid system's metabolic burden.

Building on these findings, our next goal was to integrate those constructs into the genome of P. putida EM384 to explore whether a more stable configuration, but monocopy, translates into an enhanced BDS activity. For this, all dsz constructs were cloned into the pTn7‐M vector, making use of the mini‐Tn7 transposase system (Zobel et al. 2015), and the resulting plasmids were named pTn7‐M‐DS001 to pTn7‐M‐DS008 (Figure 4A and Table 1). The growth curves of the recombinant EM384 strains, containing the different dsz constructs, were measured over time in LB (Figure 4B) and M9cit (Figure 4C). Strains DS005 and DS006, with dszB as the first gene in the operon, showed significantly higher final biomass in both media, particularly in LB (1.80 ± 0.08 mg/mL and 1.86 ± 0.10 mg/mL at 16 h, respectively). In contrast, the other strains yielded lower biomass production in LB (DS001: 1.07 ± 0.10 mg/mL; DS002 1.07 ± 0.11 mg/mL, and DS007 1.01 ± 0.10 mg/mL). These final biomass differences were more subtle in minimal medium (Figure 4C). The higher biomass production observed in strains DS005 and DS006 suggests that this specific operon configuration promotes a more efficient desulfurization pathway, possibly due to increased or more balanced enzyme production. However, we cannot rule out the possibility that these results may be influenced by an unexpected mutation elsewhere in the genome. Interestingly, the final biomass production of the genome‐integrated variants (DS006 and DS005; Figure 4B,C) also outperformed the plasmid‐borne strains (DS006p and DS005p; Figure 3B,C).

FIGURE 4.

The different dsz operon rearrangements integrated into the genome and their impact on growth and biodesulfurization. (A) Schematic representation of the eight dsz operon cassettes (DS001 to DS008) used in this study. Gene expression in all constructs is driven by the IPTG inducible Ptrc promoter (green arrow). The light orange semicircle depicts the native RBS, the dark orange the one used in (Gallardo et al. 1997), while the purple represents the optimised BBa_B0034. For clarity, the native RBSs of dszB and dszC are not drawn. All genetic constructs were assembled using the SevaBrick approach. The constructs are not drawn to scale. (B and C) The growth curves of each strain carrying each operon in LB and minimal medium, respectively. (D and E) Desulfurization activity of the cells carrying the different operons when grown in LB and minimal medium, respectively. All strains were induced with IPTG at the beginning of the experiment (t = 0). Desulfurization activity was measured in cells collected at the mid‐exponential phase (8 h of growth) and quantified in units per milligram of dry cell weight (DCW) based on the levels of produced 2‐HBP and DBTO₂ under the standard biodesulfurization assay conditions (described in the Experimental Procedures section). Statistical analysis using one‐way ANOVA showed significant differences in DBTO₂ and 2‐HBP production in both LB and minimal medium (p < 0.0001). Each experiment was performed twice and the average and standard deviation (STDEV) are presented.

We further investigated whether increased biomass led to improved desulfurization activity. Again, the resting cell assays showed that all genome‐integrated variants (Figure 4D,E) resulted in higher specific BDS activity than their plasmid‐borne counterparts (Figure 3D,E). DS006 cells had significantly higher specific desulfurization activity in both media, peaking at 36 Units/mg_DCW in LB. This highlights the importance of placing the dszB gene at the start of the operon, consistent with findings by (G. Q. Li et al. 2007), who showed that the upstream positioning of dszB enhances DBT desulfurization rates in R. qingshengii IGTS8. Previous studies have shown that other dsz gene orders can disrupt enzyme activities, leading to inefficient metabolic pathways. Our data for DS007 (dszC → A → B) and DS008 (dszC →B → A) align with the findings of Chen et al. 2008, which showed that placing dszC earlier in the operon results in suboptimal DBT conversion. This misalignment leads to intermediate accumulation that results in decreased growth (Oldfield et al. 1997; Yang et al. 2019) as seen in the relatively increased DBTO₂ levels in DS007 and DS008 (Figure 4D,E). In contrast, DS006 activity assays showed very low DBTO₂ levels (Figure 4D,E), indicating minimal intermediate accumulation, suggesting an optimal metabolite flux through the BDS pathway, meaning that the Dsz enzymes are produced in appropriate ratios, allowing efficient reactions without toxic intermediates build‐up. Balanced enzyme expression is crucial because intermediates like DBTO₂ can inhibit downstream steps of the route, as previous studies have demonstrated that an excess of DBTO₂ hampers DszA and DszC activities, reducing pathway efficiency and harming cellular health (Oldfield et al. 1997; Yang et al. 2019). In the case of DS001 and DS007, the high DBTO₂ accumulation indicates bottlenecks within the metabolic route, likely due to suboptimal dszB expression, slowing pathway progression, and reducing biomass production, emphasising the importance of gene order for a balanced metabolic flux. The high desulfurization activity of DS006 cells grown in both rich and minimal media is particularly notable, as a minimal medium is more cost‐effective for industrial applications. Therefore, its ability to degrade DBT under nutrient‐limited conditions makes it a strong candidate for industrial biodesulfurization processes.

We next focused on identifying the optimal dsz‐operon expression time for maximum biodesulfurization activity. For this, strain DS006 was grown in the minimal medium (M9cit) and induced by adding IPTG at lag, early, mid, late exponential, and stationary phase (Figure S2). Then, we determined the specific BDS activity of cells removed from the culture at 12 and 14 h. The highest BDS efficiency was observed when cells were induced at mid‐exponential (6 h of growth; Figure S2). This suggests that dsz enzyme expression during the early stages of growth enhances enzyme activity, maximising desulfurization efficiency in DS006. This induction timing will be used in the following experiments.

2.3. Comparative Analysis of the Biodesulfurization Capacity of P. putida DS006 and R. qingshengii IGTS8

Our work continued with a comparative analysis between P. putida DS006 and R. qingshengii IGTS8 to evaluate the performance of the optimised dsz operon configuration in the recombinant EM384 in comparison with the wild‐type strain in bioreactor conditions (Figure 5). The comparison was performed in a 7‐L bioreactor with controlled parameters: pH 7, temperature 30°C, stirring speed 400 rpm, and oxygen supply 1 vvm. DS006 was grown in M9cit while IGTS8 was cultured in the optimised CDM‐ethanol (Glekas et al. 2022; Martzoukou et al. 2022). As always in this work, BDS activity was measured in Units/mg_DCW through resting cell assays, using cell biomass collected from various growth phases.

FIGURE 5.

Comparative analysis of the bacterial growth and BDS activity in a 7‐L bioreactor. In all graphs, data for DS006 are depicted in orange colour while for IGTS8 in purple. (A) Bacterial growth of DS006 and IGTS8 over time; (B) Specific biodesulfurization activity of DS006 and IGTS8 cells expressed as (Units/mg_DCW); (C) Volumetric BDS activity (Units/L) = (specific BDS activity in Units/mg_DCW) × (biomass concentration in mg_DCW/mL) × (1000 mL/L); and (D) Volumetric BDS productivity (Units/L/h) = (Volumetric BDS activity in Units/L)/(Corresponding growth time in h). Each experiment was performed twice and the average and standard deviation (STDEV) are presented.

As shown in Figure 5B, IGTS8 exhibits significantly higher maximum specific BDS activity (expressed as Units/mg_DCW of 2‐HBP) than DS006. However, this difference is mitigated when biomass levels between the strains are considered. Figure 5C indicates that the disparity in volumetric BDS activity (Units/L) between these species is less pronounced than the difference in their specific BDS activity, primarily attributed to the higher biomass yield of DS006.

From an industrial perspective, an additional critical factor is the process duration that is incorporated in productivity. An effective biocatalyst ideally combines high desulfurization capacity with shorter overall process times. Figure 5D depicts that DS006 achieves over 100% higher overall volumetric BDS productivity (Units/L/h) compared to IGTS8, due to the substantially faster growth rate of DS006, completing its growth cycle 2.7 times faster than IGTS8 (Figure 5A). This emphasises the potential of the recombinant DS006 as a biocatalyst for fossil fuel biodesulfurization.

In addition, examination of the BDS activity of DS006 (Figure 5B) reveals that it remains relatively stable during the exponential growth phase. In contrast, IGTS8 cells show a decline in their specific activity during this phase. Calzada et al. 2011, utilising P. putida CECT5279 and plasmid‐born Dsz enzymes, reported maximal activity of the DszA and DszC enzymes during late exponential growth, while DszB exhibited optimal activity during early exponential growth. This variation in activity contributed to the gradual decline in desulfurization capacity during the exponential growth. Comparable reductions in BDS activity were also observed in R. qingshengii IGTS8 by Prasoulas et al. 2021, in a 20‐l bioreactor. In contrast, DS006, harbouring the dszBCA operon, stabilises desulfurization activity during the exponential growth. We hypothesise that increased DszB protein levels, potentially due to this specific genetic operon configuration, may enhance and sustain the desulfurization activity of this recombinant biocatalyst. Further research is needed to explore the molecular mechanisms behind this observation.

2.4. Biodesulfurization Performance of P. putida DS006 and R. qingshengii IGTS8 in a Biphasic BDS System

While the initial bioreactor experiments demonstrated the strong potential of the recombinant P. putida DS006, it was necessary to validate further its performance in a more complex and industrially relevant setting. The performance of both P. putida DS006 and wild‐type R. qingshengii IGTS8 was next compared in a lab‐scale biphasic BDS system using cells grown under optimal conditions (M9cit for DS006 and CDM‐ethanol for IGTS8; see Experimental Procedures section for details; Figure 6A). Both strains were cultivated in 1 L flasks (200 mL working volume) at an initial biomass concentration of 0.01 mg/mL, with samples withdrawn from the cultures at three growth phases: (i) mid exponential (M in Figure 6B,C), (ii) late exponential (L in Figure 6B,C), and (iii) stationary (S in Figure 6B,C). At each phase, the culture broths were mixed with n‐dodecane containing 3 mM DBT in a 50/50 volumetric ratio. Then, biomass concentration and 2‐HBP production were measured in the aqueous and organic phases, respectively.

FIGURE 6.

Comparison of organic phase 2‐HBP volumetric productivity between P. putida DS006 and R. qingshengii IGTS8 under two different process conditions. (A) Schematic representation of the two species ( P. putida EM384 DS006 and R. qingshengii IGTS8) used for the biphasic biodesulfurization experiment. (B) Table summary of the fitted parameters (Equations 2 and 3; see Section 4) derived from the experimental data of growth and subsequent biphasic experiments with the recombinant DS006 and the wild type IGTS8 cells. (C) 2‐HBP productivity comparison. For the biphasic system only, a process time of 24 h was considered for both strains, while in the growth & biphasic condition, the corresponding growth time for each phase and strain was added. M: Indicates mid exponential; L: Late exponential and S: Stationary phase. Statistical significance was evaluated with a two‐tailed t‐test (p < 0.05) and when relevant indicated with an asterisk within the figure.

The mathematical analysis of the data (Figure 6B and Figure S3) reveals a notable difference in the growth parameters of the two species, particularly the maximum specific growth rate, μ _max. Similar to bioreactor experiments, this rate is significantly higher in DS006 (0.47 h⁻¹) than in IGTS8 (0.08 h⁻¹). Another interesting finding is that both species exhibited relatively stable growth patterns after being removed from the fermentation broth and transferred to the biphasic system (Figure S3). This stability is significant from a biotechnological perspective, as it suggests that continuing biomass growth can enhance biodesulfurizing activity during contact with the organic phase. However, it is important to note that this may not hold when actual petroleum products are used, as they may contain compounds that could inhibit microbial growth.

Equally interesting results were obtained regarding the parameters of activity loss (Figure 6B). The initial BDS specific activity of the cells (μmole 2‐HBP in the organic phase per gram cells in the aqueous phase per hour) as expressed by the sum of the activity constants (y ₀ + y ₁ − Equation (3) at the initiation of the biphasic system when t = 0), is generally higher for IGTS8 than for the recombinant DS006. This initial activity gradually decreases for both species with growth phase time, but at a much faster rate in IGTS8 compared to DS006. In fact, at the late logarithmic and stationary phases, the initial specific biodesulfurization activities are almost equal in both microorganisms.

The values of the desulfurization activity of the deactivation constant, k _d, are significantly higher for DS006 compared to IGTS8. At first glance, this means that in general, Pseudomonas cells lose their desulfurizing activity faster over time. However, a very important difference between the two microbial strains was revealed from the non‐linear regression fitting results: In DS006, the non‐linear regression fitting of the experimental data in Equation (3) always converged at a certain value of the parameter y ₀ (Figure 6B). This means, at least from a mathematical point of view, that a very significant proportion of the initial cell‐specific desulfurizing activity (expressed by the value of the parameter y ₀) appears to be unaffected by time and remains constant throughout the biphasic growth—for example, it is not dependent on the k _d. For IGTS8, this quantity converged to zero (or to a much smaller value) for the cells from all growth phases. While this may be a mathematical specificity of the non‐linear regression algorithm used, it suggests that the actual desulfurization activity loss is similar for both strains or at least not as different as the corresponding k _d values imply.

All the above is summarised in Figure 6C, which illustrates the overall productivity of the biphasic system for the cells from the different growth phases. The calculations consider both the duration of the biphasic system and the additional cultivation time required for the corresponding biomass production: only biphasic and growth & biphasic. In the first case, the total volumetric productivity is maximal with cells obtained at the late‐logarithmic growth phase, with the biomass of the genetically modified DS006 showing only a 10%–20% higher 2‐HBP productivity than the corresponding IGTS8 biomass. The relative advantage of DS006 increases substantially—surpassing 300%—when the growth duration required to reach the end of the exponential growth is considered (Growth & Biphasic), as this period is markedly longer for IGTS8. In summary, although R. qingshengii IGTS8 initially displayed higher specific BDS activity, P. putida DS006 emerged as a more promising candidate for biodesulfurization applications due to its faster growth, stable desulfurization activity, and superior total productivity over extended biphasic system operation times. The sustained enzyme activity of DS006 makes it a more efficient biocatalyst for industrial‐scale applications. Further studies are needed to investigate the molecular mechanisms underlying this enhanced biodesulfurization activity and to fully maximise the potential of recombinant DS006 for industrial processes, which will be the focus of future work.

3. Conclusion

Taken together, the results above show that the integration of a reshaped biodesulfurization operon (dsz) in a tailor‐made Pseudomonas putida strain leads to a whole‐cell catalyst that by large outperforms the efficacy of the reference, naturally occurring isolate R. qingshengii IGTS8. As shown, this was because the restructured operon improved its expression, minimised its metabolic burden to the host by reordering genes with optimised ribosomal‐binding sites, and was implanted in a chassis more adapted to host harsh reactions. This work thus highlights that the engineered P. putida strain and potentially others generated under the methodological and conceptual frame of Synthetic Biology may open sustainable alternatives to conventional chemical desulfurization processes. By engineering greater stability and enhanced kinetics in biphasic systems, the duration and cost of industrial biodesulfurization mediated by such bacterial catalysts could be significantly reduced, thereby speeding up the industrial process.

4. Experimental Procedures

4.1. Bacterial Strains, Plasmids, Media, Chemicals, and Growth Conditions

The bacterial strains used are depicted in Table 1. A summary of the plasmids used is shown in Table 2 while the repository plasmids are indicated in Table S1.

LB was utilised as the nutrient‐rich medium (10 g/L of tryptone, 5 g/L of yeast extract, and 5 g/L of NaCl). When required, P. putida cells were grown on M9cit minimal medium (Sambrook and Russell 2001) while R. qingshengii IGTS8 was cultured on sulfur‐free chemically defined medium (CDM) as described in (Glekas et al. 2022; Martzoukou et al. 2022). Both minimal media were supplemented as needed with different carbon sources (citrate for P. putida or ethanol for R. qingshengii ) at a final concentration of 0.33 mol of C per litre. Solid media was prepared by adding 1.5% (w/v) agar. When required, antibiotics were used at the following final concentrations: 50 μg mL⁻¹ kanamycin; 10 μg mL⁻¹ gentamycin; 30 μg mL⁻¹ chloramphenicol; and ampicillin, 150 μg mL⁻¹ for E. coli and 500 μg mL⁻¹ for P. putida . To induce the expression of heterologous DNA, isopropyl‐β‐D‐1‐thiogalactopyranoside (IPTG) was added at a final concentration of 1 mM.

The majority of growth studies on strains and bacterial strains were performed in 96‐well plates incubated at 30°C with a stirring velocity of 600 rpm (150 μL of culture per well—four wells per condition and time point). Growth kinetics was monitored by following the OD₆₀₀ (Multiskan GO Microplate Spectrophotometer, Thermo Fisher Scientific, Waltham, MA USA) at specific time intervals, using a calibration curve to correlate OD₆₀₀ with dry cell weight concentration (Figure S1).

To evaluate the stress caused by various 2‐HBP concentrations (600, 800, and 1100 μM) to the different bacterial strains, cells were grown with and without the stressor in a 96‐well microtiter plate and their growth (OD₆₀₀) was monitored every 30 min using a Spectramax iD3 (Molecular Devices LLC; San Jose, CA, USA) for 24 h.

Bioreactor growth studies of selected P. putida transformants and R. qingshengii IGTS8 were performed in a 7 L Labfors 3 Bioreactor (Infors HT, Switzerland) at a working volume of 6 L. All batches were subjected to identical culture conditions of temperature (30°C), agitation (400 rpm), and aeration (1 vvm). The pH was maintained at 7.0 by adding either 5 M NaOH or 2 M HCl. P. putida was grown in M9cit minimal medium with citrate (11 mM) as the carbon source, while R. qingshengii IGTS8 was cultured in CDM medium with ethanol (165 mM) as the sole carbon source and dimethyl sulfoxide (DMSO) (1.3 mM) as the sole sulfur source (Olmo et al. 2005). Bioreactors were inoculated from overnight‐grown cultures to an initial OD₆₀₀ of 0.05 in the bioreactor. Biodesulfurization (BDS) activity was determined following the methodology described below.

4.2. Plasmids and Strain Construction

For DNA manipulation, standard laboratory protocols were followed (Sambrook and Russell 2001). When required, genomic DNA from R. qingshengii IGTS8 was isolated following the phenol‐chloroform method (Wilson 2001). Plasmid DNA was isolated via the QIAprep Spin Miniprep kit (Qiagen, USA). Oligonucleotides were synthesised by IDT (USA) and the complete set is listed in Table S2. The genes of interest were amplified using the Q5 high‐fidelity DNA polymerase (New England BioLabs, MA, USA) following the manufacturer's guidelines. In some cases, the standard PCR protocol was adapted to a biphasic one to amplify complex constructs (see Table S2). This consists of using a low annealing temperature (50°C) in the initial phase (10 cycles) and then increasing the hybridization temperature to 76°C for the final set (25 cycles). For colony PCR, the 2× Master Mix (Thermo Fisher Scientific, MA, USA) was used. PCR products were purified using the NucleoSpin Gel and PCR Clean‐up kit (Macherey‐Nagel, Germany). DNA concentration and purity were determined using a μDrop plate with a Multiskan GO microplate spectrophotometer (Thermo Fisher Scientific, MA, USA). Restriction enzymes were purchased from New England BioLabs (MA, USA). The assembly of the different dsz genetic constructs was performed using the primer sets listed in Table S2 (Damalas et al. 2020). The PCR products and digested plasmids were separated by DNA electrophoresis with 0.8% (w/v) agarose gels and visualised using the Molecular Imager VersaDoc (Bio‐Rad, USA). The correctness of plasmid constructs was determined by DNA sequencing. Recombinant DNA was integrated into the genome of P. putida by the use of the Tn7 transposon (pTn7‐M (Zobel et al. 2015) and Tn7 transposase (pTNS2; (Choi et al. 2005)). These plasmids were introduced into P. putida EM384 recipient cells by a four‐partite mating as described (Martínez‐García et al. 2011; Martínez‐García and De Lorenzo 2012). Briefly, E. coli CC118λpir harbouring the different pTn7‐DS00n (n denotes one of the eight constructed cassettes) plasmids was used as the donor strain, along with the mating‐helper E. coli HB101 (pRK600) and E. coli DH5alpha λpir carrying the Tn7 transposase (pTNS2). Mating mixtures were incubated for 24 h at 30°C and resuspended in 10 mM MgSO₄ and dilutions plated on a selective medium (M9cit supplemented with Gm).

4.3. Biodesulfurization Assay

The biodesulfurization activity of the various recombinant P. putida strains and the wild‐type R. qingshengii IGTS8 was determined using cells from different growth stages as described in (Glekas et al. 2022; Martzoukou et al. 2022; Prasoulas et al. 2021). In brief, a calculated number of cells were washed with 100 mM Hepes, pH 8, and incubated for 30 min in a 2 mL tube and shaken at 1200 rpm in the presence of 1 mM DBT in the same buffer (total volume 300 μL). Then, an equal volume of acetonitrile was added to stop the BDS reaction. After that, the tubes were centrifuged and the supernatant was used to determine the concentrations of the final product, 2‐hydroxybiphenyl (2‐HBP), and the intermediate, dibenzothiophene‐2,2‐dioxide (DBTO₂), by HPLC with fluorescence detection (Martzoukou et al. 2022). Desulfurization activity was expressed in units (U) per milligram of dry cell weight (DCW), where 1 U corresponds to the release of one nanomole of product (2‐HBP and/or DBTO₂) per hour under the specified assay conditions. Each experiment was performed twice with three samples per condition.

4.4. Biodesulfurization Performance in a Biphasic Setup Model

The performance of the selected recombinant DS006 and IGTS8 was simultaneously evaluated in model biphasic BSD systems. The study involved the growth of each strain in their corresponding optimal media (200 mL cultures in conical flasks) and the use of samples from the growing fermentation broth in lab‐scale biphasic systems. The latter was conducted in 2 mL tubes with 1.2 mL working volume at a 50/50 v/v organic to aqueous phase ratio. The organic phase was comprised of 3 mM DBT in n‐dodecane. The fermentation broth from different growth phases (different biomass concentrations) was used as the aqueous phase. Multiple tubes were prepared for each condition and placed in an orbital shaker at 1000 rpm and 30°C. Three tubes were removed at each time point, and the phases were separated by mild centrifugation (3 min, 1000 rpm). Biomass concentration was determined in the aqueous phase by OD₆₀₀, while 2‐HBP concentration was determined in the organic phase by HPLC.

The growth kinetics data were fitted using a logistic‐modified equation to determine key growth parameters (biomass concentration at any given time; Figure S3).

$$\frac{\mathrm{d}{\rm X}}{\mathrm{d}t}={\mu }_{\max }·X·\left(1-\frac{X}{X_F}\right)$$ (1)

where X is the biomass concentration (mg/mL); μ _max is the apparent maximum specific growth rate (h⁻¹); and X _F represents the final biomass concentration (mg/mL). Equation (2) (the integrated version of Equation 1) was used to determine the values of μ _max and X _F for each microbial strain by fitting the experimental data using the non‐linear regression routines of SigmaPlot software (Ver. 12.0), where X ₀ is the initial biomass (mg/mL), and t is the growth time of the process (hours).

$$X=\frac{X_F}{1+\left(\frac{X_F-X_0}{X_0}\right)·e^{-{\mu }_{\max }·t}}$$ (2)

For the kinetic analysis of the biphasic system, where an organic phase (containing DBT) and an aqueous one (biomass) coexist, we used a simplified model structured according to the following: (i) the rate of 2‐HBP produced in the organic phase is proportional to the concentration of the biocatalyst in the aqueous phase; (ii) the incorporation of a first‐order deactivation term, assuming that the decline in biocatalyst activity follows a simple exponential decay pattern; (iii) the environmental conditions (pH, T, and substrate availability) remained constant throughout the experiments. Based on these assumptions, the rate of 2‐HBP production was modelled through the following equation:

$$\frac{\mathrm{d}\left[\mathrm{HBP}\right]}{\mathrm{d}t}=X·\left(y_0+y_1·{\mathrm{e}}^{-k_d·t}\right)$$ (3)

where [HBP] is the concentration of 2‐HBP in the organic phase (μΜ), X is the concentration of biomass in the aqueous phase (mg/mL), y ₀ and y ₁ are activity constants (μmole 2‐ΗΒP/mg cells/h), k _d is the deactivation constant (h⁻¹), and t is the time after the initiation of the biphasic system (h). The constant y ₀ corresponds to a portion of biocatalyst activity that remains constant over time, whereas y ₁ represents a fraction of activity that decreases exponentially with time (such as enzyme inactivation or loss of cellular viability). The deactivation constant k _d describes the rate at which the biocatalyst activity decreases over time due to various factors (enzyme denaturation, substrate inhibition, or cellular stress). The value of constants y₀, y₁, and k_d was determined by fitting the experimentally obtained 2‐HBP concentration combined with the biomass growth data (from Equation 2), using the non‐linear regression routines of SigmaPlot (Ver. 12.0).

Author Contributions

Panayiotis D. Glekas: writing – original draft, conceptualization, methodology, investigation, formal analysis, validation, visualization. Ioannis Papageorgopoulos: investigation, formal analysis, visualization, writing – original draft. Stamatios G. Damalas: formal analysis, methodology. Víctor de Lorenzo: writing – review and editing, supervision. Diomi Mamma: writing – review and editing, conceptualization, supervision, validation. Esteban Martínez‐García: writing – review and editing, supervision, methodology, validation, conceptualization. Dimitris G. Hatzinikolaou: conceptualization, funding acquisition, writing – review and editing, methodology, validation, supervision.

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1. Calibration curves of OD₆₀₀ and Dry cell weight. The chart shows the linear relationship between optical density (OD₆₀₀) and dry cell weight (DCW, g/L) for P. putida EM384 grown in two different media: LB (orange) and M9cit with cit(green). Error bars represent the standard deviation of triplicate measurements. The linear regression equations for each medium are as follows: LB (y = 0.3693x, slope range 0.3607 to 0.3779) and M9cit (y = 0.3229x, slope range 0.3142 to 0.3304). These curves demonstrate a positive correlation between OD₆₀₀ and DCW, with the cells grown in LB media showing a slightly higher dry cell weight compared to those grown in M9cit media at equivalent optical densities.

Figure S2. Effect of induction time on the specific biodesulfurisation activity of the genetically modified strain P. putida EM384 DS006. Different colour points indicate the growth phase/time points when the inducer was added to the culture.

Figure S3. BDS in a model biphasic system, using cells (whole broth) collected during the mid/late exponential and stationary phase. The blue circles represent biomass growth and the blue line is the non‐linear regression fitting of Equation (2). The red triangles correspond to 2‐HBP concentration and the red line corresponds to Equation (3) fitting. Left diagrams: P. putida DS006, Right diagrams: R. qingshengii IGTS8. The green line indicates the time of biomass withdrawal from the culture for use in the biphasic system.

Table S1. Plasmids that were used in this study.

Table S2. Oligonucleotides used in this study.

Table S3. Oligonucleotides with its specific usage.

Glekas, P. D. , Papageorgopoulos I., Damalas S. G., et al. 2025. “Boosting Dibenzothiophene Biodesulfurization Through Implantation of a Refactored DBT Pathway in a Tailored Pseudomonas putida Chassis.” Microbial Biotechnology 18, no. 9: e70148. 10.1111/1751-7915.70148.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.