Biochemical Characterization of GacI, a Bifunctional Glycosyltransferase–Phosphatase Enzyme Involved in Acarbose Biosynthesis in Streptomyces glaucescens GLA.O

Abstract

Acarbose, a pseudotetrasaccharide produced by several strains of Actinoplanes and Streptomyces, is an α-glucosidase inhibitor clinically used to control type II diabetes. Bioinformatic analysis of the biosynthetic gene clusters of acarbose in Actinoplanes sp. SE50/110 (the acb cluster) and S. glaucescens GLA.O (the gac cluster) revealed their distinct genetic organizations and presumably biosynthetic pathways. However, to date, only the acarbose pathway in the SE50/110 strain has been extensively studied. Here, we report that GacI, one of the proteins that appear to be different between the two pathways, is a bifunctional glycosyltransferase family 5 (GT5)–phosphatase (PP) enzyme that functions at two different steps in acarbose biosynthesis in S. glaucescens GLA.O. In the acb pathway, the GT and the PP reactions are performed by two different enzymes. Truncated GacI proteins having only the GT or the PP domain showed comparable catalytic activity with the full-length GacI, indicating that domain separation does not significantly affect their respective catalytic activity. GacI, which is widely distributed in many Streptomyces, represents the first example of naturally occurring GT5-PP bifunctional enzyme biochemically characterized.

Graphical Abstract

Introduction

Pseudooligosaccharides are a group of natural products whose structures highly resemble oligosaccharides but possess one or more pseudosugar (cyclitol) moieties.^{1, 2} Members of this family of compounds have outstanding inhibitory activities against a variety of sugar hydrolytic enzymes,³ underlying the commercial use of some of them to treat human and plant diseases. Validamycin A, a pseudotrisaccharide from several variants of Streptomyces hygroscopicus, is a potent trehalase inhibitor used to control fungal infections in rice plants.⁴ Acarbose (1) (Figure 1A), a pseudotetrasaccharide produced by several strains of Actinoplanes and Streptomyces,² is an α-glucosidase inhibitor that is clinically used to control type II diabetes.⁵ Due to their clinical and commercial importance, their biosynthesis and potential methods to improve their production have been subjects of investigations.

Figure 1.

Biosynthesis of acarbose. (A) Proposed biosynthetic pathway to acarbose in Actinoplanes sp. SE50/110; (B) Genetic organizations of the acarbose clusters in Actinoplanes sp. SE50/110 (the acb cluster) and Streptomyces glaucescens GLA.O (the gac cluster). The acarbose/acarviosin cluster in S. coelicoflavus ZG0656 is identical to the gac cluster. The acbI (encodes a GT) and acbJ (encodes a PP) genes in the acb cluster are fused as gacI in the gac cluster.

The biosynthetic gene clusters (BGCs) of acarbose (1) and related pseudotetrasaccharides have been identified in several actinobacteria. However, close inspections of those BGCs revealed significant variations between the BGCs in strains of Streptomyces and non-Streptomyces. For example, the genetic organizations of the clusters in Actinoplanes (e.g., Actinoplanes sp. SE50/110) are considerably different from those in Streptomyces (e.g., S. glaucescens GLA.O and S. coelicoflavus ZG0656) (Figure 1B), leading to the notion that the biosynthetic pathways of acarbose in these organisms are different.^6–8 However, so far, only the acarbose pathway in Actinoplanes sp. SE50/110 (the Acb pathway) has been fully studied (Figure 1A).

In the SE50/110 strain, the pseudosugar moiety of acarbose is derived from 2-epi-5-epi-valiolone (EEV, 3), a cyclization product of sedoheptulose 7-phosphate (2) catalyzed by the cyclase enzyme AcbC (Figure 1A).⁹ Subsequently, EEV (3) undergoes phosphorylation by the kinase AcbM,¹⁰ epimerization by AcbO,¹¹ dehydration by AcbL,⁷ and reduction by AcbN to give valienol 7-phosphate (V7P, 7).¹² V7P (7) is then phosphorylated to valienol 1,7-diphosphate (V1,7PP, 8) by the kinase AcbU, dephosphorylated to valienol 1-phosphate (V1P, 9) by the phosphatase AcbJ, and converted to GDP-valienol (GDP-V, 10) by the nucleotidyltransferase AcbR.¹³ The 4-aminodeoxysugar moiety of acarbose is derived from dTDP-4-amino-4,6-dideoxyglucose (dTDP4a6dGlc, 11), which is synthesized from glucose 1-phosphate catalyzed successively by the glucose 1-phosphate thymidylyltransferase AcbA, the dTDP-glucose 4,6-dehydratase AcbB, and the dTDP-4-amino-4,6-dideoxyglucose transaminase AcbV. Subsequently, dTDP4a6dGlc (11) is coupled with maltose by AcbI to give 4-aminoDGG (12).¹³ The final assembly of acarbose (1) is accomplished by coupling between GDP-valienol and 4-aminoDGG by AcbS,¹³ which is a new member of a growing family of pseudoglycosyltransferases (PsGTs).¹⁴

In contrast to the Acb pathway in Actinoplanes sp. SE50/110, the Gac pathway in S. glaucescens GLA.O (Figure 1B, Tables S1 and S2) appears to adopt a slightly different strategy to produce acarbose.⁸ A clear distinction was observed in the early steps of the pathway, where the Gac pathway is more similar to that of validamycin than to the Acb pathway.¹⁵ In addition, variations can also be seen in genes involved in the latter part of the pathway. For example, the gac cluster lacks a gene corresponding to the phosphatase acbJ gene from the acb cluster, whereas gacI, which corresponds to acbI, encodes a glycosyltransferase family 5 (GT5) as well as a phosphatase (PP), suggesting that the gene product possesses both catalytic functions.

Here, we report the biochemical characterization of GacI as a bifunctional GT5–PP enzyme consisting of an N-terminal GT domain and a C-terminal PP domain connected by a short linker region. GacI is the first example of bifunctional GT–PP enzymes biochemically characterized and is believed to be a product of gene fusion between an acbI homologue and an acbJ homologue in S. glaucescens GLA.O.

Experimental Section

Strains, cultivation, and genetic manipulation

Genomic DNA (gDNA) of S. glaucescens GLA.O was prepared using the GeneElute Bacterial Genomic DNA Kit (Sigma). E. coli strains DH10b and BL21(DE3)/pLysS were used as hosts for DNA manipulation and recombinant protein production, respectively (Table S2). PCR was conducted in 35 cycles using PrimeSTAR GXL DNA polymerase (Takara Bio). Oligodeoxyribonucleotide primers were purchased from Sigma. DNA fragments were separated with agarose gel electrophoresis (0.8%) and recovered using the E.Z.N.A. Gel-Extraction Kit (Omega Bio-Tek). Restriction endonucleases were purchased from New England Biolabs. Preparation of plasmid DNA was performed using EconoSpin all-in-one silica membrane mini spin columns (Epoch Life Science). All other DNA manipulations were done according to standard protocols. DNA sequencing was done at the Center for Quantitative Life Sciences (CQLS) Core Laboratories, Oregon State University. BD Difco LB medium was used for the E. coli culture.

Construction of the expression vectors

The genes gacI, gacI-PP, gacI-GT1, gacI-GT2, gacI-GT3, and gacI-GT4 were PCR amplified from gDNA of S. glaucescens GLA.O (Table S3) using their corresponding primers (Table S4). The amplicons were digested with appropriate restriction enzymes and ligated with the pRSET B vector (Thermo Fisher). All plasmids are listed in Table S5. Each plasmid was transferred into E. coli BL21(BE3)/pLysS (Promega).

Preparation of recombinant GacI, GacI-PP, GacI-GT1, GacI-GT2, GacI-GT3 and GacI-GT4

Transformants harboring each gene were grown on LB agar plates containing ampicillin (100 μg/mL) and chloramphenicol (35 μg/mL) until colonies were formed. The colonies were picked and cultivated in LB medium (1–3 mL) containing the same antibiotics. After 12–14 h of cultivation at 37°C in a shaker (200 rpm), the cultures (1 mL) were transferred into a fresh LB medium (100 mL) containing the same antibiotics and shaken (37°C, 200 rpm) until OD₆₀₀ reached 0.4. After cooling to 4°C on ice, IPTG (100 μM) was added, and the bacteria were cultivated for another 15–18 h at 16°C. The cells were harvested by centrifugation (4200 rpm, 4°C, 10 min), and the cell pellets were washed with buffer A (30 mM HEPES pH 8.0, 300 mM NaCl, 10% glycerol). Subsequently, the pellets were frozen with dry ice-acetone or liquid nitrogen and stored at –80°C until use.

To prepare the recombinant proteins, the pellets were resuspended with buffer A, sonicated (2 W, 10 sec, 10 times), and centrifuged (13000 rpm, 30 min, 4°C) to remove the cell debris. Ni-NTA (UBPBio) was used to purify the recombinant proteins. The buffer was exchanged with Tris-HCl buffer (20 mM, pH 7.5) and the protein solution was concentrated using an Amicon Ultra centrifugal filter (MWCO 10K or 30K) and confirmed with the Bradford assay (Bio-Rad). The molar concentrations of the proteins were calculated based on the assumption that each of the protein solutions contain only the desired protein.

Enzymatic reactions of GacI and GacI-PP for the phosphatase activity

For LC-Q-TOF-MS analysis, a reaction mixture (10 μL) containing Tris-HCl buffer (20 mM, pH 7.5), GacI or GacI-PP (2 μM) and substrate (1 mM) was incubated at 30 °C for 3 h. All substrates, V1,7PP, 1-epi-V1,7PP, V7P, and Aca7P, have been prepared and reported in our previous publications.^{13, 16} The reaction mixtures were analyzed by LC-Q-TOF-MS using the below-described method.

LC-Q-TOF-MS analysis

High-resolution mass spectrometry (HR-MS) analysis was performed on an Agilent 1260 HPLC connected to an Agilent 6545 Q-TOF, using an InfinityLab Poroshell HILIC column (150 × 4.6 mm, 2.7 μm) at a flow rate of 0.4 mL/min and the following gradient method. Line A was an aqueous ammonium formate solution (10 mM), and line B was acetonitrile. The column was pre-equilibrated with 10% A/90% B. After injection, the mobile phase composition was maintained for 2 min and then changed to a linear gradient from 10% A/90% B to 90% A/10% B over a period of 20 min.

Measurements of the initial velocity of the PP activity of GacI toward several substrates

To get the initial velocity of GacI or GacI-PP, a reaction mixture (10 μL) containing Tris-HCl buffer (20 mM, pH 7.5), GacI or GacI-PP (1 μM), and each substrate (1 mM) were incubated at 30°C for 5 min. The reactions were then quenched with 10% SDS and Tris-HCl buffer (20 mM, pH 7.5), PNP (2 U/mL), and MESG (400 μM) were added to make a total volume of 50 μL. Subsequently, the reaction mixtures were analyzed by a SpectraMax iD3 microplate reader (Molecular Devices) equipped with a SoftMax Pro (version 7) processor. The changes in A₃₆₀ were monitored at 30 °C for 10 min. Boiled enzymes (100 °C for 10 min) were used as negative controls. The data were collected in triplicate.

Kinetic studies of the PP activity of GacI

To obtain the kinetic values of the PP activity of GacI, reaction mixtures (40 μL) containing Tris-HCl buffer (20 mM, pH 7.5), PNP (2 U/mL), MESG (400 μM), and V1,7PP (1500 μM, 1000 μM, 750 μM, 500 μM, 300 μM, 200 μM, 150 μM, 100 μM or 0 μM) were pre-incubated at 30 °C and the A₃₆₀ was monitored until no changes were observed. Subsequently, GacI (1.0 μM) and Tris-HCl buffer were added to make 50 μL reaction mixtures and the changes in A₃₆₀ were monitored at 30 °C for 10 min. The data were collected in triplicate. A Lineweaver-Burk plot was used to calculate the K_m and k_cat values.

Enzymatic reactions of GacI or GacI-GTs for the GT activity

To confirm the activity of GacI or GacI-GTs, reaction mixtures (10 μL each) containing Tris-HCl buffer (20 mM, pH 7.5), GacI or GacI-GTs (1 μM), dTDP4a6dGlc (1 mM),¹⁷ and maltose (1 mM) were incubated at 30 °C for 3 h. The reaction mixtures were analyzed for the expected product 4-aminoDGG by LC-Q-TOF-MS using the method described above for LC-Q-TOF-MS analysis.

Enzymatic reactions of GacI for the GT activity using several sugar acceptors

To determine whether GacI can catalyze a coupling between dTDP4a6dGlc and oligosaccharides to yield 4-aminoDGG, reaction mixtures (10 μL each) containing Tris-HCl buffer (20 mM, pH 7.5), GacI (1 μM), dTDP4a6dGlc (1 mM), and either maltose, maltotriose, maltotetraose or maltopentaose (1 mM) were incubated at 30 °C for 3 h. The reaction mixtures were analyzed for the expected product 4-aminoDGG by LC-Q-TOF-MS using the method described above. Maltotriose and maltopentaose were purified over Sephadex LH-20 to remove maltose contaminant before use.

The initial velocity of the GT activity of GacI and GacI-GT3 and GacI-GT4

Reaction mixtures (10 μL each) containing Tris-HCl buffer (20 mM, pH 7.5), GacI or GacI-GTs (1 μM), dTDP4a6dGlc (1 mM), and maltose (1 mM) were incubated at 30 °C for 10 min. After addition of 50% aqueous acetonitrile (90 μL), the samples were filtered, and the reaction mixtures were analyzed by LC-Q-TOF-MS using the method described above. The data were collected in triplicate. The peak areas of the reaction products were obtained with Agilent MassHunter software and the values were used to obtain initial velocities based on the standard curve obtained from a series of concentrations of authentic 4-aminoDGG.

Kinetic studies using the GT domain of GacI

To obtain the kinetic values of the GT activity of GacI, reaction mixtures (10 μL) containing Tris-HCl buffer (20 mM, pH 7.5), maltose (1 mM), MgCl₂ (1 mM), GacI (1 μM), dTDP4a6dGlc (2000 μM, 1000 μM, 750 μM, 500 μM, 250 μM, 100 μM, 50 μM, 25 μM or 0 μM) were incubated at 30 °C for 5 min. The reaction mixtures were processed using the same procedure described above. A Lineweaver-Burk plot was used to calculate the K_m and k_cat values.

Construction of truncated domains

The alignments were obtained and edited by Geneious Prime using the ClustalW algorithm (version 2.1) with standard parameters.¹⁸ The modeled structure was obtained by using AlphaFold v2.1.0 (Google Colaboratory version).¹⁹ The obtained modeled structure was analyzed and drawn with Pymol (version 2.3.2).²⁰

Pull-down assay

The sequence of Factor Xa recognition site was obtained from the pColdI vector. The primers having this sequence were used for the amplification of gacI-PP (gacI-PP-FaXa_F and R) and pRSET B (pRSET-FaXa_F and R), respectively. Each PCR-amplified fragments were ligated using the NEBuilder HiFi DNA assembly mix (NEB). The recombinant GacI-PP having a Factor Xa recognition site located after the N-terminal His₆-tag was purified with Ni-NTA and treated with Factor Xa (NEB) in the presence of CaCl₂ (2 mM) at 16°C for overnight. The reaction mixture was then passed through a Ni-NTA column to remove the cleaved His₆-tag and the unreacted GacI-PP. The flow-through fraction was concentrated and incubated with His₆-tagged GacI-GT3 at 30°C for 2 h. Subsequently, Ni-NTA was added to the mixture and incubated at 4°C for 1 h. The mixture was transferred into a column and washed with buffer A (20 mL) containing imidazole (20 mM) and eluted with buffer A containing imidazole (200 mM). Fractions were analyzed with SDS-PAGE.

Results and Discussion

Biochemical characterization of the phosphatase activity of GacI

To characterize the catalytic function of GacI, the gene was PCR amplified from the chromosomal DNA of S. glaucescens GLA.O, ligated into the pRSET B vector, and expressed in Escherichia coli BL21(BE3)/pLysS. The 110 kDa His₆-tagged recombinant protein was purified with Ni-NTA column (Figure S1) and characterized for its phosphatase activity using a number of possible substrates, i.e., valienol 7-phosphate (V7P, 7), valienol 1,7-diphosphate (V1,7PP, 8), 1-epi-valienol 1,7-diphosphate (1-epi-V1,7PP), and acarbose 7-phosphate (Aca7P) (Figure 2A). These compounds have previously been proposed to be involved in acarbose biosynthesis.^{10, 12, 16, 21} Interestingly, LC-MS analysis of the reaction mixtures showed that GacI was able to dephosphorylate all of the tested substrates (Figures 2B, 2C, and S2).

Figure 2.

Biochemical characterization of the GacI phosphatase activity. (A) dephosphorylation of V1,7PP, 1-epi-V1,7PP, V7P, and Aca7P; (B) ESI (–) EIC (m/z 255.0275) corresponding to V1P in GacI reactions with V1,7PP; (C) ESI (+) EIC (m/z 199.0577) corresponding to valienol in GacI reactions with V7P; (D) Initial velocities of GacI with V1,7PP, 1-epi-V1,7PP, V7P, or Aca7P as substrates; E) Initial velocities of GacI and GacI-PP with V1,7PP as a substrate. (n = 3, error bar is S.D.)

Next, we determined the initial velocities of GacI reactions with V7P, V1,7PP, 1-epi-V1,7PP, and Aca7P by quantifying the inorganic phosphate as the reaction side product using the MESG assay (Figure S3).²² Reactions with V7P and V1,7PP showed high initial velocities of 5.69 ± 0.94 μM/min and 5.33 ± 1.47 μM/min, respectively, whereas those with 1-epi-V1,7PP and Aca7P showed lower initial velocities of 2.13 ± 0.29 μM/min and 0.33 ± 0.09 μM/min, respectively (Figure 2D, Table S6). The results suggest that GacI can effectively dephosphorylate both V7P and V1,7PP but not 1-epi-V1,7PP or Aca7P, which are consistent with the substrate preference of AcbJ from the acarbose pathway in Actinoplanes sp. SE50/110.¹³ Although both V7P and V1,7PP can be effectively dephosphorylated by GacI, the natural substrate for GacI is believed to be V1,7PP, as its dephosphorylation product valienol 1-phosphate (V1P, 9) has recently been identified as an intermediate in acarbose biosynthesis (Figure 1A).¹³ Therefore, the kinetic parameters for GacI were determined with V1,7PP as a substrate using the MESG assay, and the results fitted well into the Michaelis-Menten and Lineweaver-Burk plots to give the apparent K_m and k_cat values for V1,7PP of 707.4 ± 101.2 μM and 8.87 ± 1.72 min⁻¹, respectively (Figure S4, Table S7).

Biochemical characterization of the glycosyltransferase activity of GacI

To evaluate the glycosyltransferase activity of GacI, the protein was incubated with dTDP-4-amino-4,6-dideoxy-D-glucose (dTDP4a6dGlc, 11)¹⁷ and maltose, which are the substrates of AcbI in the acb pathway,¹³ and the reaction mixture was analyzed by LCMS. As expected, the results showed that the GacI reaction with dTDP4a6dGlc and maltose produced 4-aminoDGG (12) (Figure 3A). In addition, substrate specificity of GacI for sugar acceptor was evaluated using maltose, maltotriose, maltotetraose and maltopentaose, and the results showed that only maltose was recognized as a substrate (Figure S5). Kinetic parameters for GacI were determined based on the area under curves of the generated 4-aminoDGG peaks in LC-MS (Figures S6 and S7). The apparent K_m and k_cat values for dTDP4a6dGlc were 305.8 ± 89.1 μM and 1.07 ± 0.28 min⁻¹, respectively (Figure S7 and Table S7).

Figure 3.

Biochemical characterization of the GacI glycosyltransferase activity. A. Coupling reaction of dTDP4a6dGlc and maltose to give 4-aminoDGG catalyzed by GacI; B. ESI (+) EIC (m/z 510.1793) of GacI reactions with dTDP4a6dGlc and maltose; C. Comparison of the initial velocities of GacI, GacI-GT3 and GacI-GT4 based on the peak areas in mass chromatogram. (n = 3, error bar is S.D.)

The above results clearly showed that GacI is a bifunctional enzyme possessing both glycosyltranferase (GT) and phosphatase (PP) activities. The GT domain is a member of the GT family 5 (GT-5) and the phosphatase domain can recognize phosphorylated cyclitols as substrates. Therefore, GacI is not only unique in its catalytic functions, but also is the first example of naturally occurring GT-PP bifunctional enzyme that has been biochemically characterized.

Individual PP and GT domains retain their respective activity

To investigate if the fusion of the two domains is critical for their activity, truncated proteins, each contains only one of the domains, were prepared and their activities were evaluated. To determine positions where a start codon or a stop codon may be inserted, the amino acid sequences of GacI, AcbI and AcbJ were aligned and analyzed (Figure 4A). To generate an independent PP domain (GacI-PP), a start codon was introduced before W753, because the start codon of AcbJ aligned with W753 of GacI. The PCR amplified DNA fragment was introduced into the pRSET B vector and expressed in E. coli BL21(BE3)/pLysS. To our delight, the truncated recombinant protein of GacI-PP was expressed in a soluble form. The protein was then incubated with V1,7PP and the formation of V1P was confirmed by LC-MS (Figure 2B). Comparisons of the reaction efficiencies between the full-length GacI and the monodomain GacI-PP showed that both enzymes have similar phosphorylase catalytic efficiencies with comparable initial velocities of 5.33 ± 1.47 μM/min and 4.66 ± 1.07 μM/min, respectively (Figure 2E and Table S8). The results indicate that the separation of the two domains does not affect the PP domain activity.

Figure 4.

Construction of truncated GacI proteins. (A) Partial amino acid alignments of GacI, AcbI and AcbJ. The numbers are corresponding to the residue number of GacI. Positive (+) mark shows the putative catalytic residues of the phosphatase domain as assigned based on the alignment of the proteins with the well-characterized phosphatases; (B) A modeled GacI structure generated using Alphafold2 (Google colab version); (C) Expanded modeled structure around the linker region. The numbers are the same as those shown in Figure 4A.

Next, the effect of domain separation on the activity of the GT domain was investigated. Based on the multiple amino acid sequence alignments, we identified two locations where a stop codon may be introduced. The first one was right after P733, which is the end of the aligned region with AcbI, and the second one was right after V751 or before the region aligned with AcbJ starts (Figure 4A). However, unfortunately, the expected gene products, GacI-GT1 and GacI-GT2, were not produced in soluble forms. A modeled structure of GacI created using the Alphafold2 program (Figures 4B and 4C)¹⁹ showed that V751 is located within a β-sheet structure. On the other hand, P733 is located further upstream of V751, in the loop region before the two β-sheet structures (Figure 4C), suggesting that the two β-sheet structures at the C-terminus of the GT domain are important for solubility. Therefore, two other truncated proteins (GacI-GT3 and GacI-GT4) were designed based on the modeled structure. First, we introduced a stop codon after V755, which is the end of the second β-sheet structure, to produce GacI-GT3 (Figures 4A and 4C). We also constructed GacI-GT4, in which a stop codon was introduced at the end of the linker loop and right before the beginning of the next β-sheet structure, which starts from L764 (Figures 4A and 4C). The putative catalytic residues (the aspartate residues) of the PP domain, predicted from the amino acid alignment of the PP domain with those of well-characterized PP enzymes, are located right downstream of this position (Figures 4A and S8),²³ suggesting that both constructs or at least GacI-GT4 should contain all sequence necessary for the GT activity. As expected, we were able to obtain the recombinant proteins GacI-GT3 and GacI-GT4 in soluble forms (Figure S1). However, based on the SDS-PAGE, both proteins appeared to be less stable than the PP domain. The semi-purified proteins were then incubated with dTDP4a6dGlc (11) and maltose. LC-MS analysis of the reaction mixtures showed that both GacI-GT3 and GacI-GT4 were able to convert dTDP4a6dGlc (11) and maltose to 4-aminoDGG (12) (Figure 3B). Given that the only difference between GacI-GT2 and GacI-GT3 is that GacI-GT3 contains four additional amino acid residues between V-751 and V-755, these residues may be critical for the stability or solubility of the recombinant proteins.

To investigate if the truncated GacI-GT3 and GacI-GT4 proteins retain the same catalytic efficiency as the full-length GacI, their initial velocities were determined and compared. The results showed that both GacI-GT3 and GacI-GT4 have almost identical initial velocities as that of the full-length GacI (Figure 3C and Table S9). Taken together, the data suggest that there are no significant catalytic changes in the GT and the PP activities of the proteins when the two domains were separated.

Physical interaction between the discrete GT and PP domains

To investigate whether the GT and PP domains interact with or have affinity to each other, a pull-down assay was conducted. First, a His₆-tag-free GacI-PP was prepared by introducing a protease recognition site into the region between the N-terminal His₆-tag and GacI-PP, and the His₆-tag was removed by incubating the recombinant protein with the restriction protease Factor Xa. The His₆-tag-free GacI-PP was subsequently incubated with His₆-tagged GacI-GT3 and the protein mixture was passed through a Ni-NTA column. The bound proteins were eluted with 200 mM imidazole solution and analyzed by SDS-PAGE (Figure S9). The results showed that the His₆-tag-free GacI-PP was co-eluted with His₆-tagged GacI-GT3, indicating that the two discrete domains were physically interacting with each other. However, since the catalytic efficiency of each domain is comparable with that of the full-length GacI, the significance of this physical interaction is unclear.

GacI homologues are ubiquitous in actinomycetes

Although our previous studies showed that acarbose or acarbose-related BGCs are widely distributed in many actinomycetes,¹³ the degree to which the gacI-like bifunctional GT-PP gene is distributed among those BGCs was unclear. Therefore, a follow-up study of those BGCs was performed and the results showed that from 139 strains that harbor acarbose or acarbose-related BGCs, 129 strains have a gacI-type gene, instead of a pair of acbI and acbJ genes, and all of them are strains of Streptomyces. The remaining 10 strains that harbor the acbI and acbJ genes are mostly uncommon actinomycetes such as Actinoplanes, Polymorphospora, Hamadaea, and Catenulispora. Although the above numbers may not accurately represent the actual ratio of gacI versus acbI-acbJ genes in nature, because there are more genome sequences of Streptomyces than other genera of actinomycetes in the NCBI database, the result nonetheless indicates that GacI homologues are ubiquitous in actinomycetes, particularly in Streptomyces. However, whether the fusion between the GT and the PP domains has any beneficial function for the organisms is still an open question.

In summary, GacI represents the first biochemically characterized GT-PP bifunctional enzyme from the acarbose biosynthetic pathway in S. glaucescens GLA.O and many other strains of Streptomyces. In these strains, GacI phosphorylates V1,7PP (8) to V1P (9) and catalyzes a coupling reaction between dTDP4a6dGlc (11) and maltose (Figure 5). In Actinoplanes sp. SE50/110, these functions are performed by two different proteins, the phosphatase AcbJ and the glycosyltransferase AcbI, respectively. Therefore, GacI is believed to be a product of gene fusion between homologues of acbI and acbJ. While the discrete GT and PP domains of GacI appear to physically interact with each other, individually they retain their respective activity at the level similar to the full-length GacI, leaving the benefit of the gene fusion and the physical interaction of the two domains remains unclear.

Figure 5.

GacI is a bifunctional glycosyltransferase and phosphatase enzyme that is involved in acarbose biosynthesis at two different stages.