Abstract
Phthalic acid esters (phthalates) are used as additives in various plastics and industrial applications. They are produced worldwide in huge amounts causing major pollution in the environment. Biodegradation of phthalates from the environment is an important route for their removal. In our previous work, we showed that Azoarcus sp. strain PA01 catabolizes o-phthalate via the anaerobic benzoyl-CoA pathway that involved two putative enzymes: the succinyl-CoA:o-phthalate CoA-transferase activates o-phthalate to o-phthalyl-CoA which is subsequently decarboxylated to benzoyl-CoA by o-phthalyl-CoA decarboxylase. In this work, we provide the information on the enzymes involved in the promising step of anoxic decarboxylation of o-phthalate to benzoyl-CoA. We have identified that there are two proteins are involved in decarboxylation step, of which only one does the actual decarboxylation but other one is essential. o-Phthalyl-CoA decarboxylase (PhtDa and PhtDb) encoded by the two genes PA01_00217 and PA01_00218 which catalyses the decarboxylation of activated o-phthalate to benzoyl-CoA. Both genes are originally annotated as an UbiD-like/UbiX-like protein. The gene with locus tag PA01_00217 is 1584 bp long coding for protein PhtDa (60 kDa), whereas PA01_00218 is 600 bp long codes for protein PhtDb (22 kDa). Here, we demonstrate that PhtDb is a flavin mononucleotide (FMN)-binding protein which does not function as a decarboxylase alone. Rather, PhtDb is assumed to generate a modified FMN-containing cofactor that is required by the PhtDa for decarboxylase activity. Alone, PhtDa does not function as a decarboxylase either. Recombinantly expressed PhtDa and PhtDb together showed activity for decarboxylation of o-phthalyl-CoA to benzoyl-CoA, only if PhtDb was previously incubated with FMN and dimethylallyl monophosphate. Phylogenetically, the proteins PhtDa and PhtDb are closely related to UbiD-like/UbiX-like enzymes that catalyses the decarboxylation of 4-hydroxy-3-octaprenylbenzoic acid to 2-octaprenylphenol, an intermediate step in ubiquinone biosynthesis. Furthermore, multiple sequence alignment and structural modelling of both proteins suggested that only PthDb possesses the binding site for FMN. These results strongly indicate that the flavin-containing cofactor is essential for decarboxylation of o-phthalyl-CoA to benzoyl-CoA during anaerobic o-phthalate degradation by Azoarcus sp. strain PA01.
Introduction
o-Phthalic acid (1,2-dicarboxybenzene) is a synthetic organic compound most used in the manufacturing of phthalic acid esters (PAEs). They are globally produced in huge quantities each year for a wide range of applications being the main plasticizers used in the polymer industry since the 1930s (Liang et al., 2008). They are commonly added (10% - 60%) to plastic materials, such as polyvinyl chloride (PVC), polyethylene terephthalate (PET), polyvinyl acetate (PVA), and polyethylene (PE), to improve extensibility, elasticity, and workability of the polymers. Phthalates are considered as priority industrial pollutants due to their chemical toxicity, and adverse effect on human health and animals (Giuliani et al., 2020). In environment, degradation of phthalate esters by bacteria involves the initial de-esterification step which releases readily degradable side chain alcohols and often accumulates phthalate. Decarboxylation of phthalate is a challenging and rate-limiting reaction in microbial degradation, especially for anaerobic bacteria (Kleerebezem et al., 1999). Because, aerobic phthalate decarboxylation is facilitated by oxygenase-dependent oxygenation reactions, forming a 3,4-dihydrodiol (in Gram-positive bacteria) or 4,5-dihydrodiol (in Gram-negative bacteria) which is later converted to dihydroxyphthalate which is then decarboxylated by decarboxylases to the common intermediate 3,4-dihydroxybenzoate (Batie et al., 1987; Chang & Zylstra, 1998; Eaton & Ribbons, 1982). Similarly, m-phthalate (isophthalate) and p-phthalate (terephthalate) are analogously decarboxylated leading to 3,4-dihydroxybenzoate (Fukuhara et al., 2010; FUKUHARA et al., 2008; Karegoudar & Pujar, 1985; Schläfli et al., 1994). In essence, aerobic phthalate-degrading bacteria introduce molecular oxygen into the phthalate ring that partially polarizes the ring, facilitating the difficult step of phthalate decarboxylation.
Due to absence of molecular oxygen, decarboxylation in anaerobic phthalate-degrading bacteria becomes challenging and regarded as the rate-limiting step (Kleerebezem et al., 1999). In the past, different hypotheses were proposed for anaerobic decarboxylation of phthalate to benzoate. For instance, Taylor and Ribbon (1983) suggested that phthalic acid is reduced by two electrons leading to 3,5-cyclohexadiene-1,2-dicarboxylic acid before its decarboxylation to benzoic acid (Taylor & Ribbons, 1983). Later, it was assumed that decarboxylation of phthalate involves formation of phthalate-coenzyme A ester, which is subsequently decarboxylated to benzoyl-CoA (Nozawa & Maruyama, 1988a; Nozawa & Maruyama, 1988b). In recent years, we provided the first experimental evidence that phthalate degrading anaerobic bacteria accomplish phthalate decarboxylation via a two-step enzyme reaction involves activation of o-phthalate to o-phthalyl-CoA, which is subsequently decarboxylated to benzoyl-CoA (Boll et al., 2002; Geiger et al., 2019; Junghare et al., 2019; Junghare et al., 2016). We demonstrated the formation of first intermediate o-phthalyl-CoA from o-phthalate and succinyl-CoA followed its decarboxylation to benzoyl-CoA in assay performed in cell-free extract of Azoarcus sp. strain PA01 (Junghare et al., 2016). Benzoyl-CoA is further metabolized by the enzymes of the anaerobic benzoyl-CoA degradation pathway (Breese et al., 1998; Fuchs, 2008; Gall et al., 2013). The genome of Azoarcus sp. strain PA01 possess the required genes for anaerobic benzoate-degradation pathway (Junghare et al., 2015).
Differential protein profiling using cell-extracts of benzoate-grown cells versus o-phthalate of Azoarcus sp. strain PA01 identified a set of proteins induced specifically with o-phthalate. Interestingly, phthalate-induced proteins were placed together in a single gene cluster that includes genes coding for the protein homologous to a solute transporter (locus tag PA01_00214), two CoA-transferases (PA01_00215 and PA01_00216), and the UbiD-like/UbiX-like decarboxylases (PA01_00217 and PA01_00218), respectively (Junghare et al., 2016). In phthalate degrading denitrifying bacteria, the CoA-transferases are involved in o-phthalate activation to o-phthalyl-CoA, and UbiD-like/UbiX-like decarboxylases in decarboxylating o-phthalyl-CoA to benzoyl-CoA (Ebenau-Jehle et al., 2017; Junghare et al., 2016; Mergelsberg et al., 2018). Among these, the genes involved in activation of phthalate to phthalyl-CoA were belonged to CoA transferase or CoA ligase family that were homologous to proteins involved in activation of aromatic acids to its CoA esters.
Interestingly, in all phthalate degrading bacteria including Azoarcus sp. PA01 the genes involved in decarboxylation were annotated as proteins homologous to UbiD-like/UbiX-like decarboxylases. UbiD-like and UbiX-like are proteins that widely distributed among prokaryotes, e.g., Escherichia coli with the carboxy-lyase activity converting 4-hydroxy-3-octaprenyl benzoic acid to 2-octaprenylphenol, an early step in ubiquinone biosynthesis (Gulmezian et al., 2007; Lin et al., 2015; Meganathan, 2001; White et al., 2015). Even though 4-hydroxy-3-octaprenyl benzoic acid and o-phthalyl-CoA are structurally quite different from each other, the genes PA01_00217 and PA01_00218 that are homologous to UbiD-like/UbiX-like proteins are suggested to be involved in the anaerobic decarboxylation of phthalyl-CoA to benzoyl-CoA by Azoarcus sp. strain PA01 (Junghare et al., 2016). To address the specific role of UbiD-like and UbiX-like decarboxylase, especially the question how these two proteins function together to catalyse the decarboxylation of o-phthalyl-CoA to benzoyl-CoA. We performed the cloning, expression and characterization of both the proteins UbiD-like/PhtDa (60 kDa) and UbiX-like/PhtDb (23 kDa). In our preliminary results we could provide the first biochemical evidence that both the proteins are involved in the decarboxylation of o-phthalyl-CoA to benzoyl-CoA, of which UbiX-like protein prepares modified prFMN co-cofactor that is required by UbiD-like for actual decarboxylation during the anaerobic degradation of o-phthalate by Azoarcus sp. strain PA01.
Results
Gene amplifications, cloning of PhtDa and PhtDb
Genomic DNA used for amplification of full-length o-phthalyl-CoA decarboxylase genes resulted in a 1.58 kb amplicon for PA01_00217 and 0.6 kb amplicon for PA01_00218 (Figure 1A). PCR products were cloned into the ampicillin resistance pET100/D-TOPO expression vector which adds 6His-tag (3kDa), Xpress Epitope to the N-terminal end of the overexpressed protein. Recombinant plasmid was used to transform E. coli TOP10 cells and grown on LB media containing ampicillin (100 µg/ml). Several positive clones were found with a gene of interest inserted into the pET100/D-TOPO expression vector. Recombinant clones were picked and screened based on the PCR product size of amplified PCR products using T7 primers (as discussed in method section) and confirmed by comparing the size of the respective genes plus additional 276 bp from cloning site of pET100/D-TOPO vector (Figure 1AB). The gene coding for PhtDa is 1584 bp long plus 276 and PhtDb is 600 bp plus 276 and expected size PCR products (using T7 primers) from positive clones are shown in Figure 1B.
Selected positive clones of E. coli TOP10 cells containing recombinant plasmids were propagated in LB media (ampicillin, 100 μg/ml) for plasmid DNA isolation. Purified recombinant plasmid DNAs were sequenced across the cloning site using T7 primers and inserted gene sequences were determined (data not shown). The sequence of recombinant genes displayed correct insertion of genes into the vector and showed 100% sequence identity with the original gene sequences available from the published draft genome sequence of Azoarcus sp. strain PA01 (Junghare et al., 2015b).
Heterologous expression, purification and protein identification
The genes PA01_00217 and PA01_00218 (Figure 1A), putatively coding for the enzymes PhtDa and PhtDb responsible decarboxylation of o-phthalyl-CoA to benzoyl-CoA, i.e., o-phthalyl-CoA decarboxylase, were individually expressed in E. coli BL21 star (DE3) and E. coli Rosetta 2 (DE3)pLysS host cells with a N-terminal His-tag introduced from pET expression vector. The purified recombinant pET100/D-TOPO plasmids (Figure 1C) containing genes coding for PhtDa (60 kDa) and PhtDb (60 kDa) were used to transform the chemically competent E. coli BL21 star (DE3) and E. coli Rosetta 2 (DE3)pLysS as the expression host cells and grown on LB media with ampicillin selection (100 µg/ml). Complete open reading frames of the cloned genes PA01_00217 and PA01_00218 are expected to encode the proteins of 527 (PhtDa) and 199 (PhtDb) amino acids in length with the computed molecular weight of about 59.54 kDa (̴60 kDa) and 21.99 kDa (̴22 kDa), respectively, plus 3 kDa 6His-tag.
In order to find the optimal expression conditions for obtaining soluble proteins, both the transformed E. coli expression host strains were grown under different combinations of growth conditions, including temperatures (37, 25 and 15 °C), culture media (LB and TB medium) varying IPTG concentrations (0.1 to 1 mM) and with or without glucose (1 %) plus ethanol (3 %), respectively, (data not shown). Protein expression optimization experiments revealed, E. coli Rosetta 2 (DE3)pLysS host cells were found to be suitable for optimal expression of soluble protein after induction with 0.5 mM IPTG (at OD between 0.5 - 0.6) and grown at 15 °C for about 18 - 20 h in the TB medium supplemented with 3 % ethanol and 1 % glucose (Figure 2A and 2B). Although, under these growth conditions protein over-expression was also observed at 37 °C and 25 °C, but expressed proteins formed inclusion bodies (insoluble protein). Finally, E. coli Rosetta 2 (DE3)pLysS cells cultured in TB media was selected for the large scale protein expression and purifications. Protein expression was up-scaled to 500 - 1000 ml culture volume.
His-tag on the recombinant proteins (PhtDa and PhtDb) allowed quick and easy purification from the supernatant. About 6 - 8 ml of supernatant prepared from 3 - 4 g of wet cell material lysate was used to load the Ni2+ ion chelated column (40 mg His-tagged protein binding capacity). By washing the Ni2+ ion chelated column with about 10 ml buffer A (as discussed above), his-tagged proteins were eluted with buffer B containing higher concentrations of imidazole (0.5 M). Eluted fractions were subjected to SDS-PAGE analysis and simultaneously, purified protein contents were determined using the Bradford assay. SDS-PAGE analysis of Ni2+ ion affinity chromatography purified proteins resulted the dense single protein bands at the expected size of about 62 kDa for PhtDa and 23 kDa for PhtDb that include plus 3 kDa 6His-tag (Figure 2C). Protein bands excised from the SDS gel and identified peptides by MS analysis, showed the expected molecular mass and high identity with the deduced amino acid sequence of genes PA01_00217 and PA01_00218 from the draft genome of the Azoarcus sp. strain PA01 (data not shown).
Enzyme activity, substrate specificity and cofactor requirement
Decarboxylase activity for conversion of o-phthalyl-CoA to benzoyl-CoA was determined with recombinantly expressed PhtDa and PhtDb in E. coli. Enzyme assay were performed directly with the E. coli transformant’s cell lysate or cell-free extracts containing respective expressed proteins PhtDA and PhtDb. Decarboxylation of o-phthalyl-CoA was demonstrated with E. coli cell-free extracts expressing PhtDa and PthDb proteins and LC-MS analysis of samples collected at the different time intervals revealed an increase of benzoyl-CoA (Figure 3). Decarboxylase activity was unaffected by oxygen exposure and was independent of oxygen. Neither one of these proteins could decarboxylate o-phthalyl-CoA to benzoyl-CoA alone. As expected, no decarboxylase activity was observed with isophthalyl-CoA or terephthalyl-CoA (Junghare et al., 2016). Furthermore, decarboxylase activity was observed only when, PhtDb was previously incubated with flavin mononucleotide (FMN) and dimethylallyl monophosphate (DMAP). No activity was observed when FMN was replaced with the flavin dinucleotide (FAD). This result strongly indicates that PhtDb is a FMN-binding/dependent protein that is essential for the decarboxylation of o-phthalyl-CoA to benzoyl-CoA. In addition to this, UV-visible scanned spectrum of FMN-DMAP incubated and His-tag purified PhtDb revealed the possible binding and modification of the FMN molecule to a PhtDb protein (Figure 4). UV-visible scanning of PhtDa and PhtDb showed the same absorption maxima at the wavelength of about 340 nm but at different intensity.
Phylogenetic position of PhtDa and PhtDb
A phylogenetic tree was constructed using MEGA 7.0 software (Kumar et al., 2016) based on the neighbor-joining method to determine the evolutionary relationship of PhtDa and PhtDb of Azoarcus sp. strain PA01 with the other known bacterial and fungal species UbiX-like/UbiD-like decarboxylases whose crystal structure are available. Amino acid sequence of total 13 different proteins were aligned with clustal W method and a single phylogenetic tree was generated for the evolutionary analysis (Figure 5). Among all the UbiD-like and UbiX-like enzymes from the bacterial and fungal species, the PhtDa enzyme (PA01_00217) with expected decarboxylase activity formed a discrete group with UbiD-like and a putatively flavin (FMN)-binding PhtDb (PA01_00218) grouped with UbiD-like enzymes, respectively (Figure 5). The proteins PhtDa and PhtDb were clearly separated from each other in the distinct clade and showed only about 30 % sequence identity to one another. Among the UbiD-like enzymes clade in the phylogenetic tree, PhtDa of Azoarcus sp. strain PA01 was found to be more closely related with the putative aromatic acid decarboxylase (Pad1, 4IP2_C) of P. aeruginosa and the ferulic acid decarboxylase (Fdc1, AHY75481) of S. cerevisiae showing about 23 % of sequence similarity. Whereas the PhtDb FMN-binding protein of Azoarcus sp. strain PA01 shared highest 64 % of amino acid sequence similarity with UbiX-like enzyme (P69772) of E. coli strain O157:H7 (Figure 5). The other closely related UbiX-like enzymes includes the phenolic acid decarboxylase (β-subunit, P94404) of B. subtilis, fungal phenylacrylic acid decarboxylases (Pad1/PadA1) of S. cerevisiae (P33751) and A. niger (A3F715) were positioned relatively distant with around 50 % amino acid sequence similarity, but all grouped together UbiX-like enzymes (Figure 5). However, the UbiX (Q9HX08) flavin prenyltransferase enzyme of P. aeruginosa strain PAO1 showed only about 44 % sequence identity with PhtDb and was distantly placed compared other UbiX (Figure 5).
Multiple sequence alignment and homology-based structure modelling
A BLAST search for proteins PhtDa and PhtDb found several sequences sharing significant similarity (data not shown). However, the BLAST results contain the enzymes with known functions that shared less similarity with PhtDa and PhtDb. For example, the phylogenetic analysis of PhtDa suggested, it shared only about 23 % sequence identity with the UbiD-like decarboxylase (Pad1) of P. aerugonisa (PDB code 4IP2_C, UniProt Q9I6N5) whose crystal structure is currently known (Jacewicz et al., 2013). Therefore, due to very less sequence similarity (23 %) of PhtDa to enzyme with known function, multiple sequence alignment and secondary structure prediction for PhtDa was not performed. On the other hand, BLAST search and phylogenetic analysis of PhtDb revealed about 64% sequence identity with FMN-dependent UbiX-like decarboxylase of E. coli O157:H7 (UniProt P69772) whose crystal structure (PDB code 1sbz) is currently resolved (Rangarajan et al., 2004). Other known UbiX-like enzymes of bacterial and fungal species also shared significant identity with PhtDb (Figure 5). Therefore, multiple amino acid sequence alignment and secondary structure predictions (using PDB code 1sbz as template) of PhtDb revealed highly conserved regions not only with the FMN-dependent UbiX-like decarboxylase of E. coli (P69772), but also with other UbiX-like enzymes from bacterial and fungal species (Figure 5). We have analyzed more closely the top 5 sequences, which all had lengths of ∼200 residues and had over 50 % sequence identity to PhtDb of Azoarcus sp. strain PA01. Among the ∼200 amino acid residues in the sequences, there are ∼50 residues conserved in all sequences, as highlighted (red) as shown in Figure 6. The majority of these enzymes sequences are the known FMN-binding prenyltransferase belonging to the superfamily of UbiX-like enzymes.
The three-dimensional (3-D) structures of many UbiD-like and UbiX-like enzymes are presently known. The 3-D structure prediction and modelling of PthDa and PhtDb were computed in SWISSMODEL workspace (https://swissmodel.expasy.org/interactive) using the deduced amino acid sequences of the PhtDa (PA01_00217) and PhtDb (PA01_00218) to find the best templates. Although, amino acid sequence of ferulic acid decarboxylase (Fdc1) from S. cerevisiae and putative aromatic acid decarboxylase (Pad1) of P. aeruginosa shared ∼23% amino acid sequence similarity with PhtDa (Figure 5). The homodimer crystal structure of Pad1 (PDB code 4iws) of P. aeruginosa (Jacewicz et al., 2013) was the best template and chosen for homology or template based structural modelling of PhtDa (Figure 7A). Whereas the homo 12-mer (dodecamer) crystal structure (PDB code 1sbz) of FMN-binding UbiX flavoprotein (Pad1) in E. coli O157: H7 (Rangarajan et al., 2004) was the best template for PhtDb and chosen for predicting the 3-D structure of PhtDb (Figure 7B). The predicted 3-D structures of PhtDa (homodimer) and PhtDb (homo 12-mer) in Azoarcus sp. strain PA01 were visualised (Figure 7AB) and compared and superimposed with the respective templates (PDB code 4zac and 1sbz) using the UCSF-Chimera package (Figure 7CD). PhtDb showed a high degree of secondary structure and motif conservation with the template (Figure 7D), compared to overlapping of PhtDa with its template (Figure 7C).
Native PAGE and protein identification by MS analysis
In order to investigate the protein complexes in phthalate-induced proteins by Azoarcus sp. strain PA01, especially to determine the PhtDa and PthDb protein-protein interactions and oligomer state. Native-protein gel electrophoresis was performed with cell-free extracts of o-phthalate versus benzoate-grown cells. Interestingly, only few native protein bands were visible on o-phthalate-grown cells compared to benzoate-grown cells (Figure 8A). Specifically phthalate-induced protein bands were estimated between 400 to 140 kDa that include protein bands labeled as 2, 3, 4 and 5 (Figure 8A). These spots were excised from the native gel and identified by MS. Protein identification of protein bands 2 and 3 (approximate estimated molecular size ranging between 390 to 350 kDa) resulted the identification majorly two proteins, namely PhtDa (60 kDa) and PhthDb (22 kDa) as shown in Table 1. Among the protein identification of the lower molecular weight protein bands (4, 5, 6 and 7) PhtDa was also identified, including the CoA-transferases (PhtSa and PhtSb protein) that are involved in the activation of o-phthalate to o-phthalyl-CoA reported in our recent findings (Junghare et al., 2016).
Furthermore, native-protein gel analysis of heterologously expressed and partially purified PhtDa and PhtDb, were performed individually and incubating together for about 1 h at room temperature to determine the protein-protein interactions. Surprisingly, when PhtDa (60 kDa) and PhtDb (22 kDa) were incubated together with FMN and DMAP, only a single band at ca. 400 kDa was visible on the native protein gel (Figure 8B). Furthermore, individual native protein gel analysis of purified PhtDa and PhtDb revealed the formation of two distinct bands visible at the estimated molecular size between ca. 280 to 420 kDa. Nevertheless, SDS analysis of purified PhtDa (60 kDa) and PhtDb (22 kDa) showed only a single bands at the expected size for the respective proteins (Figure 2C). These results strongly indicate that these high molecular size protein bands in native gel are most likely the multimeric complexes of PhtDa and PhtDb. However, at the moment it is still unclear whether the single band observed in the lane of the PhtDab incubated mixture (with FMN plus DMAP) is due to the PhtDa and PhtDb multimeric complex, because MS confirmation of the proteins is elusive.
Discussion
Present work demonstrates that PhtDa is the enzyme responsible for catalysing decarboxylation of o-phthalyl-CoA to benzoyl-CoA, and the decarboxylase activity of PhtDa depends on the FMN-dependent PhtDb protein constituting a two-enzyme component system. PhtDa (60 kDa) and PhtDb (22 kDa) were cloned into pET100/D-TOPO and overexpressed as N-terminal His-tagged proteins in E. coli Rosetta (DE3)pLysS host. Both proteins were purified to homogeneity by utilizing Ni2+ ion affinity chromatography (Figure 2C). The molecular weight of the recombinant proteins was determined by SDS PAGE analysis. Alternatively, proteins were identified by MS analysis and revealved that both protein bands showed expected molecular masses and high identity with the deduced amino acid sequenes of both gene from the genome sequence of Azoarcus sp. strain PA01 (Junghare et al., 2015b).
Enzyme assays with cell lysate or cell-free extracts of E. coli strains expressing PhtDa and PhtDb could readily convert o-phthalyl-CoA to benzoyl-CoA (Figure 3). Interestingly, none of these proteins possess decarboxylase activity alone. Although, PhtDa and PhtDb do not share sequence similarity with each other, but they belong to a new class of enzyme family and are homologous to UbiD-like and UbiX-like decarboxylases found in a wide range of bacterial and fungal species (Figure 5, 6 and 7). UbiD and UbiX are known to act in the decarboxylation of aromatic substrates (Zhan and Javor, 2003) but individually UbiX or UbiD were unable to perform decarboxylation (Jacewicz et al., 2013; Rangarajan et al., 2004; Kopec et al., 2011). Our results showing the involvement of PhtDa and PhtDb for decarboxylation reaction are in agreement with the previous studies. Because decarboxylation of phenylacrylic acid in S. cerevisae required Fdc1 (UbiD-like) and Pad1 (FMN dependent UbiX-like). Similarly, in E. coli O157:H7 or P. aeruginosa decarboxylation of 3-octaprenyl-4-hydroxybenzoate to 2-octaprenylphenol is also required two enzymes as UbiX-like (FMN-dependent) and an UbiD-like aromatic acid decarboxylase (Rangarajan et al., 2004; Jacewicz et al., 2013).
Furthermore, heterologously overexpressed PhtDa and PhtDb in E. coli only led to benzoyl-CoA formation, when PhtDb was pre-incubated with FMN and dimethylallylphosphate (DMAP). No decarboxylase activity was observed when FMN was replaced against flavin dinucleotide (FAD). These results strongly suggest that decarboxylase activity of PhtDa, is dependent on FMN-dependent PhtDb. Phylogenetically, PhtDb is closely related to the FMN-dependent UbiX-like enzyme which covalently modifies FMN molecule that is needed for the decarboxylase activity by UbiD-like decarboxylase in E. coli (White et al., 2015; Lin et al., 2015; Rangrajan et al., 2015; Payne et al., 2015). Multiple sequence alignment of PhtDb with the known UbiX flavin-binding proteins from bacterial and fungal species (Figure 5 and 6) revealed highly conserved amino acid residues. PhtDb showed highest sequence similarity (64 %) with the FMN-dependent UbiX-like enzyme of E. coli O157:H7 for which the FMN-binding site at the amino acid positions Ser36 and Arg122, and a nucleotide binding site at Thr9, Gly10, Ala11, Ser87 and Met88, respectively, has been established recently (White et al., 2015). Interestingly, PhtDb also possesses the conserved amino acid residues at these positions, suggesting a probable binding site for FMN and DMAP along with the similar FMN dependent enzymes (Figure 6). The addition of DMAP and FMN to PhtDb caused distinct changes in the FMN absorbance profile (Figure 4), which indicated possible binding and modification of FMN due to the formation of PhtDa:FMN:DMAP complex. UV-VIS measurement of purified PhtDb after incubation with FMN and DMAP showed absorption peak maxima at 384 nm and 458 nm (Figure 4B), which are the characteristic peaks for the FMN molecule (Lin et al., 2015). However, slight changes were observed in PhtD:FMN spectrum that consisted of a decrease in the absorbance intensity at 380 nm and ̴ 450 nm and the developed flat shoulders at 380 and 450 nm (Figure 4B). This strongly indicates an interaction of FMN with the PhtDb.
In addition to this, structural modelling of both PhtDa and PhtDb suggested that only PhtDb possesses the FMN-binding domain and forms a dodecameric complex (264 kDa; 12 FMN per 12 PhtDb), whereas PhtDa is a homodimer based on the known template structure (Rangarajan et al., 2004; (Figure 7B and 5C)). Superimposition of the model structure of both PhtDa and PhtDb with the respective templates of known crystal structures such as homodimer Pad1 in P. aeruginosa ((PDB code 4iws); Jacewicz et al., 2013) showed high overlapping and similarity in the overall fold of the secondary structure elements with the predicted model of PhtDa (Figure 7C). Whereas superimposition of the homo 12-mer crystal structure of FMN-dependent UbiX-like enzyme (PDB code 1sbz) in E. coli O157:H7 (Rangarajan et al., 2004) exhibited very high overlapping of predicted homo-12mer 3-D model of PhtDb (Figure 7D). Therefore, this accurately reflects that subunits in both PhtDa and PhtDb are arranged in a similar pattern with the oligomeric structures of their respective templates and thus it is most likely to have functional homology (Figure 7).
Native protein analysis of cell-free extract of Azoarcus sp strain PA01 as well as purified PhtDa and PhtDb revealed the formation of a protein complex of the expected molecular weight in analogy to the oligomeric structure of the related proteins as described in Figure 7. Based on the 3-D structure modelling, PhtDa (60 kDa) forms a 120 kDa homodimer while the PhtDb (22 kDa) is homo dodecamer forming a 264 kDa oligomer complex. Thus, both together (PhtDa and PhtDb) are expected to form a 384 kDa (approx. 390 kDa) protein complex. Native protein analysis of cell-free extract of Azoarcus sp. strain PA01 cells grown in o-phthalate showed the protein bands, particularly at approximate 380 and 350 kDa size (Figure 8A). MS analysis results (Table 1) of these two bands (2 and 3) surprisingly showed the identification of mainly PhtDa and PhtDb, with significant sequence coverage and score (Table 1) among the other proteins detected (data not shown). However, MS identification of the following lower molecular weight size protein bands (between 140 to 80 kDa), resulted in the identification of mainly PhtSa (43 kDa) and PhtSb (41 kDa), the expected subunits of the succinyl-CoA dependent CoA-transferase that is involved in the activation of o-phthalate to o-phthalyl-CoA (Junghare et al., 2016). The CoA-transferase are expected to form a tetramer (approximately 170 kDa) homologous to succinyl-CoA:(R)-benzylsuccinate CoA-transferase involved in toluene activation by T. aromatica (Leutwein and Heider, 2001). In addition to this, native protein analysis of partially purified PhDb incubated with FMN plus DMAP showed a single protein band at 390 kDa size, although the native protein analysis of PhtDa and PhtDb individually formed two distinct protein bands of 400 to 350 kDa (Figure 8B). Yet, the SDS-PAGE analysis of the purified PhtDa and PhtDb exhibited single bands at the expected sizes of 60 and 22 kDa (Figure 2C).
In conclusion, our experiments demonstrate that PhtDa and PhtDb constitute a two-component decarboxylase system for o-phthalyl-CoA decarboxylation similar to the UbiX and UbiD two-component decarboxylase system in which UbiX with its prenylated FMN cofactor is essential for the decarboxylase UbiD (Lin et al., 2015; White et al., 2015). Therefore, we strongly suggest that the prenylated FMN consequently plays a crucial role in the decarboxylation of o-phthalyl-CoA to benzoyl-CoA. However, how exactly prenylated FMN facilitates the decarboxylation of o-phthalyl-CoA to benzoyl-CoA is yet to be established.
Materials and methods
Materials and reagents
All chemicals used in this study were of analytical grade. Coenzyme A (CoA) trilithium salt, ampicillin, flavin mononucleotide (FMN), flavin dinucleotide (FAD), γ,γ-dimethylallyl monophosphate (DMAP), phthalate isomers and phthalic anhydride were from Sigma-Aldrich (Germany. o-Phthalyl-CoA’s used in this study were produced as described previously (Junghare et al., 2016). Materials and equipment for protein purification were obtained from GE Healthcare (Munich, Germany).
Strains and cultivation conditions
Azoarcus sp. strain PA01 was cultivated anoxically in non-reduced, freshwater mineral medium containing o-phthalate (2 mM) or benzoate (2 mM) as the growth substrate supplemented with 10 - 12 mM nitrate as electron acceptor as described previously (Junghare et al., 2015b; Junghare et al., 2016). Cells were harvested by centrifugation (10,000 x g for 10 min at 10°C) at an optical density (OD600nm) 0.2 - 0.3 and stored at −20 °C until further use. E. coli OneShot TOP10, E. coli BL21 star (DE3) and E. coli Rosetta 2 (DE3)pLysS (Invitrogen) cells were grown either in Luria Bertani (LB) (Sambrook et al., 1989) or in terrific broth (TB) (Tartoff et al., 1987) media (unless otherwise mentioned) supplemented with 100 µg/ml of ampicillin.
Gene amplifications, plasmid construction and sequencing
The genes encoding o-phthalyl-CoA decarboxylase, i.e., PA01_00217 (1584 bp) and PA01_00218 (600 bp) sequence can be found at the Integrated Microbial Genomes (IMG) under gene ID 2597201524 and 2597201525, (https://img.jgi.doe.gov/cgi-bin/m/main.cgi) respectively, in the draft genome sequence of Azoarcus sp. strain PA01 (Junghare et al., 2015b). For PCR amplification of the respective protein (PhtDa and PhtDb) coding genes, primers were synthesized (Microsynth, Switzerland). To facilitate directional cloning of the genes into champion pET100/D-TOPO expression vector which introduces a C-terminal 6x-His-tag (3 kDa), overhang of four nucleotides (CACC as underlined) was added to 5’ end of each forward primer of the respective genes. The gene PA01_00217 coding for PhtDa (60 kDa) was amplified using the primers: 217f, 5’-CACCATGAACGATCTGGCAACGA-3’ and 217r, 5’-CTTTGTGGTGGCAGCAGC-3’, and PA01_00217 coding for PhtDb (22 kDa) was amplified using primers: 218f, 5’-CACCATGAAAAGACTGATAGTGGGG-3’ and 218r, 5’-TGCGATTACTTCTGCGGC-3’, respectively. Gene amplification was performed using proof reading Phusion High Fidelity (HF) DNA polymerase (Finnzymes) in the PCR reaction (50 µl) containing: 2 µl of gDNA template (approx. 50 −100 ng); 2 µl of each primer (100 µM); 1 µl of 50 mM MgCl2 (unless otherwise mentioned); 10 µl of 5x Phusion HF buffer; 20 µl of dNTPs (500 µM); 0.5 µl of Phusion HF DNA polymerase (0.25 Units) and 12.5 µl of molecular grade water (for amplification of gene PA01_00218, MgCl2 was not added). A personal DNA thermal cycler (Eppendorf) was used for PCR reactions: initial denaturation at 98 °C for 5 min followed by 35 cycles of 98 °C for 30 s, 65 °C (70 °C for PA01_00217) for 40 s, 72 °C for 90 s and final extension at 72 °C for 10 min. PCR reactions were analysed by agarose (1 % w/v) gel electrophoresis and stained with ethidium bromide. Analysed PCR products were directly purified using DNA Clean and Concentrator kit (Zymo Research). DNA concentrations were determined using a nano UV spectrophotometer (Thermo Scientific).
The purified PCR products were cloned by mixing with the champion pET100/D-TOPO expression vector and ligated according to the manufacturer’s protocol (Invitrogen). The ligated vector containing the gene of interest (recombinant plasmid) was transformed into chemically competent E. coli OneShot TOP10 cells and grown on LB agar plates containing ampicillin (100 µg/ml). Positive clones containing the gene of interest were screened using the T7 primer pair as: T7, 5’-TAATACGACTCACTATAGGG-3’ and T7 reverse, 5’-TAGTTATTGCTCAGCGGTGG-3’, provided in the champion directional TOPO expression kit. Positive clones of E. coli TOP10 cells were propagated in LB medium (ampicillin, 100 µg/ml) and used for recombinant plasmid DNA isolation (Zypy Plasmid Miniprep, Zymo Research). Prior to plasmid transformation to expression host E. coli cells, the correct integration and sequence identity of the inserted gene was individually verified by sequencing the recombinant plasmids across the cloning site using the T7 primers (as mentioned above). DNA sequencing was performed at GATC-Biotech (Constance, Germany) and sequences were compared with the original gene sequences from the genome of Azoarcus sp. strain PA01 (Junghare et al., 2015b) for any mutation or base change using the BLAST (http://www.ncbi.nlm.nih.gov).
Optimization of protein expression and purifications
Chemically competent OneShot E. coli BL21 Star (DE3) and E. coli Rosetta 2 (DE3)pLysS cells were transformed with the purified recombinant vector. Transformant’s were grown either in 10 ml of LB liquid or onto the LB solid agar medium (for maintenance) containing ampicillin (100 µg/ml). Over-expression of recombinant proteins PhtDa and PhtDb were performed in LB medium or in TB medium containing 100 µg/ml of ampicillin supplemented with 1 % glucose and 3 % ethanol (unless otherwise mentioned). Pre-cultures were grown initially at 37 °C under shaking conditions (200 rpm). Cells were induced for protein expression at an optical density (OD600nm) between 0.5 to 0.6 by adding 0.3 - 0.5 mM of isopropyl-β-D-thiogalactopyranoside (IPTG) and grown at different temperatures (at 37 °C for about 4 h; 25 °C for about 7 h and at 15 °C for about 17 h). Cells were harvested by centrifugation (10,000 x g for 10 min at 8 °C) and stored frozen at −20 °C until further use or analysed for soluble protein expression. Protein expression optimization experiments were performed in small scale cultures using 10 to 100 ml of growth media.
For large scale protein expression, transformed cultures were grown in 500 or 1000 ml media. About 2 to 3 g wet cell material was suspended in 5 ml lysis buffer containing 50 mM Tris-HCl (pH 7.6), 200 mM NaCl, 100 mM KCl, 60 mM imidazole, 10 % glycerol and 5 mM dithiotritol designated as buffer A. Prior to cell lysis, the cell suspension was supplemented with 1 mg/ml lysozyme (Sigma), 0.01 mg/ml DNase I (Sigma), 50 mM L-glutamate, 50 mM L-arginine-HCl, 1 mg of protease inhibitor cocktail (Roche) plus dimethyl allyl monophosphate (DMAP) together with 1 mM flavin mononucleotide (FMN) or 1 mM flavin dinucleotide (FAD). The cell suspension was incubated on ice for about 30 min for enzymatic (lysozyme) cell lysis. Finally, cells were passed through a French pressure cell operated at 139 kPa at least for three times. Cell lysate was centrifuged at 27,000 x g for 30 min at 4 °C to remove the cell debris. The supernatant (cell-free extract) was either used directly for enzyme activity measurements or subjected to further enzyme purification.
His-tagged recombinant enzyme purification was performed using the Ni2+ ion affinity chromatography. E. coli supernatant was directly applied onto a 1-ml Ni2+ ion chelating poly-histidine affinity column (HisTrap HP, GE Healthcare) equilibrated with buffer A (50 mM Tris-HCl, pH 7.6; 200 mM NaCl; 100 mM KCl; 60 mM imidazole; 10 % glycerol and 5 mM dithiotritol). After washing the column with 10 ml of buffer A, the His-tagged proteins were finally eluted with 2 - 3 ml of elution buffer B (buffer A supplemented with 0.5 M imidazole final concentration). Imidazole was removed by passing the eluted protein fraction over a desalting NAPTM-25 column (GE Healthcare) equilibrated with buffer A and eluted with 2.5 ml of the same buffer. Fifty percent glycerol was added to the purified protein to a final concentration of 25 % (v/v) and stored at −20 °C. Protein contents were determined using the Bradford assay (using BAS as the standard).
Molecular weight determination by SDS-PAGE
Sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) analysis was used to determine the purity and molecular weight of the over-expressed proteins. SDS-PAGE was also used to determine the expression level of soluble protein in E. coli cell-free extracts or cell pellet, according to Laemeli method using the established protocol (Junghare et al., 2016). Protein samples were mixed with sample loading buffer (0.05 % bromophenol blue, 5 % β-mercaptoethanol, 10 % glycerol, 1 % SDS in 0.25M Tris-HCl buffer pH 6.8) and denatured by heating at 99 °C for 10 min. Samples were loaded onto the one-dimensional SDS gel (12 % resolving and 4 % stacking gel) and electrophoresis was performed using a BioRad Mini-Protean cell (8 cm x 6 cm x 1.0 mm) at constant voltage of 120 V for about 2 h. For molecular weight (MW) estimation of protein bands, 10 µl of the prestained protein ladder of MW range 10 - 180 kDa was used (Page Ruler, Thermo Fisher). Protein bands were visualised with 0.25 % Coomassie Brilliant Blue R-250 dye dissolved in 10 % acetic acid and 50 % methanol, followed by de-staining in 10 % acetic acid and 50 % methanol in distilled water. The detected protein band size was compared with standard protein MW markers for size estimation (Pre-stained protein molecular size marker 15 – 100 kDa, Themo Scientific). Expected size protein bands were excised from the SDS gel and given for peptide identification by electrospray mass spectrometry (MS) analysis at the Proteomics Facility, University of Konstanz. The MS spectra of peptides were analysed using the Mascot search engine (version 2.2.03, Matrix Science) to confirm the identity of the over-expressed proteins by comparison with the deduced amino acid sequence of the respective genes from the genome of the Azoarcus sp. strain PA01 (Junghare et al., 2015b).
Decarboxylase activity assays and cofactor requirement
o-Phthalyl-CoA decarboxylase activity of PhtDa and PhtDb enzymes overexpressed in E. coli cells was assayed according method described before (Junghare et al., 2016). The decarboxylase activity was assayed directly with the E. coli cell lysate containing overexpressed protein PhtDa and PhtDb for conversion of o-phthalyl-CoA to benzoyl-CoA. Standard anoxic enzyme assays (150 µl) contained: 100 µl tri-ethanolamine buffer (0.1M, pH 7.6), 10 - 20 µl cell lysates (PhtDa and PhtDb) and 1 mM dithionite. Enzyme reaction was started by the addition of 10 µl o-phthalyl-CoA and incubated at 30 °C for about 30 min. Samples (30 µl) were removed at different time intervals and proteins were precipitated with methanol (60 µl). Samples were centrifuged and the supernatants were analysed for benzoyl-CoA formation by liquid chromatography coupled with electrospray ionization-MS (LC-ESI- MS) as described previously (Junghare et al., 2016). The substrate specificity of the decarboxylase activity was assayed for decarboxylation of isophthalyl-CoA, terephthalyl-CoA or fluoro-o-phthalyl-CoA to benzoyl-CoA (fluoro-benzoyl-CoA in case fluoro-o-phthalyl-CoA). Effect of oxygen exposure on the stability and activity of PhtDa and PhtDb was assayed by performing enzyme assay under oxic condition.
The flavin dependent activity of recombinant PhtDb (PA01_00218) was determined using reconstitution experiments. Heterologously expressed PhtDb in E. coli cells were suspended in cell lysis buffer A (as mentioned above) and supplemented with 1 mM flavin dinucleotide (FAD) or flavin mononucleotide (FMN) plus about 10 µl of di-methyl allyl mono phosphate (DMAP). Cell were incubated on ice for about 30 min (enzymatic cell lysis) and finally opened by French Press treatment as described above. Cell lysate was either used directly for enzyme activity determinations or processed further by centrifugation (27, 000 x g for 30 min at 4 °C) for His-tag purification of recombinant protein. UV-visible spectrum (300nm to 500nm wavelength) of the purified PhtDb enzyme was recorded in the UV-vis spectrophotometer (Jasco V-630, spectrophotometer) for determination of FMN content. Spectrum of 1 µM FMN in elution buffer was used as reference.
Native protein PAGE analysis
Estimation of protein complex size and protein-protein interaction, native protein polyacrylamide gel electrophoresis (PAGE) was performed with the cell-free extract of Azoarcus sp. strain PA01 cells grown with o-phthalate and benzoate as the control. Additionally, native protein gel was also performed with the partially His-tag purified recombinant PhtDa (60 kDa) and PhtDb (22 kDa) proteins. Approximately, each of 30 µg of purified PhtDa and PhtDb proteins or soluble proteins from the cell-free extract of Azoarcus sp. strain PA01 cells were mixed with the native protein sample loading buffer (60 mM Tris-HCl, pH 6.8; 30 % glycerol; 0.01 % (w/v) bromophenol blue). PhtDa and PhtDb protein-protein interaction were investigated by incubating these two proteins together plus FMN and DMAP addition. Electrophoresis was performed using BioRad Protean Mini cell and the PreCast gel slab (BioRad) at 130 V for about 3 h. Protein bands were visualised by straining the gels with Coomassie Brilliant Blue R-250 as described above. Protein-protein interaction and protein complex sizes were estimated by comparing the stained protein bands with the native protein molecular weight markers range between 60 to 690 kDa (Amersham, GE Healthcare Life Sciences). Protein bands excised from the stained native protein gels were identified by MS analysis.
Sequence alignment and structure homology modelling
Protein similarity searches were performed using the BLASTp search tool available at the NCBI web page (http://www.ncbi.nlm.nih.gov/BLAST/). A phylogenetic tree was constructed to determine sequence divergence and the evolutionary positions using the MEGA 7 software package (Kumar et al., 2016). Amino acid sequences of the PhtDb and PhtDa of Azoarcus sp. strain PA01 and the characterized the UbiX-like and UbiD-like enzymes from bacterial and fungal species were considered for tree construction. Phylogenetic tree was calculated using neighbor-joining method and the bootstrap (1000 replicates) method was followed for corroborating the consistency of each node in the tree. For determination of conserved amino acid regions, the T-Coffee Expresso multiple sequence alignment package with the CLUSTAL W (1.83) alignment tool (http://tcoffee.crg.cat/apps/tcoffee/do:expresso) was used (Notredame et al., 2000). Aligned sequences were edited in the ESPript 3.0 program ((http://espript.ibcp.fr/ESPript/ESPript/index.php); Robert and Gouet, 2014). Homology-based 3-dimensional (3-D) models were constructed for PhtDa and PhtDb proteins to understand structural aspects and functional homology to the related UbiD-like and UbiX-like enzymes. The predicted structural models for PhtDa and PhtDb were generated through the SWISS-MODEL workspace in an automated mode (Arnold et al., 2006; Biasini et al., 2014). PhtDa structure was predicted using the template putative aromatic acid decarboxylase (Pad1) crystal structure (PDB accession code 4iws.1.A) from Pseudomonas aeruginosa (Jacewicz et al., 2013), whereas PhtDb structure was predicted using the FMN-dependent Ubix-like decarboxylase (PDB code 1sbz.1.A) from Escherichia coli O157:H7 (Rangarajan et al., 2004). Predicted structural model visualization and superimpositions analyses with the respective template structures were performed with the UCSF Chimera package (Pettersen et al., 2004).
Acknowledgments
MJ acknowledges the University of Konstanz for financial support. MJ greatly acknowledges Jasmin Frey for helpful suggestions and discussions during the cloning and heterologous protein expression experiments. MJ greatly thanks to Prof. Dieter Spiteller and Prof. Bernhard Schink for reading and improvement of the manuscript.