For the first time a functional protein was fused to a PHA synthase resulting in PHA granule formation and display of the respective function at the PHA granule surface. The GFP reporter protein was N-terminally fused to the class I PHA synthase of Cupriavidus necator (PhaC) and the class II PHA synthase of Pseudomonas aeruginosa PAO1 (PhaC1), respectively, while maintaining PHA synthase activity and PHA granule formation. Fluorescence microscopy studies of GFP–PHA synthase attached to emerging PHA granules indicated that emerging PHA granules locate to cell poles and to midcell representing the future cell poles. A rapid oscillating movement of GFP–PHA synthase foci from pole to pole was observed. In cell division impaired Escherichia coli, PHA granules were localized between nucleoids at regular spacing suggesting that nucleoid occlusion occurred. Accordingly, anucleate regions of the E. coli mukB mutant showed no regular spacing, but PHA granules with twofold increased diameter were formed. First evidence was provided that the cell division and the localization of GFP–PHA synthase foci are in vivo co-located.
Spherical PHA granules are formed inside bacterial cells due to the activity of PHA (/polyhydroxyalkanoate) synthases, which are the key enzymes of PHA biosynthesis. These unique enzymes catalyze the stereo-selective polymerization of (R)-3-hydroxyacyl-CoA to PHA with concomitant release of CoA . PHA, a polyester, serves as a reserve polymer.
As soon as the substrate, (R)-3-hydroxyacyl-CoA thioester, is intracellularly provided, the PHA synthase starts to catalyze the formation of a high molecular weight polyester (n > 1000). The growing polyester chain remains covalently attached to the enzyme  and converts the soluble enzyme into an amphipathic molecule. Presumably, a self-assembly process similar to micelle formation occurs and leads to the formation of spherical inclusions with an amorphus polyester core and PHA synthase covalently attached to the surface . These PHA granules increase in size while the attached synthases continuously incorporate precursor from the cytosol into the growing polyester chain and thereby into the PHA granule core. It remains to be determined whether larger granules occur due to fusion events or whether simple increase in size due to ongoing polymerization takes place. It is also unclear, whether PHA granules arise simultaneously or derive from one original particle. Usually 5–8 PHA granules are deposited intracellularly exploiting almost the entire cell volume, when maximum PHA accumulation is achieved. PHA granules are surrounded by a phospholipid membrane  with embedded or attached proteins  consisting of the PHA synthase [6,7], the intracellular PHA depolymerase [8,9], amphiphilic phasin proteins [10–12], PHA-specific regulator proteins [13–15] and additional proteins  with yet unknown function.
The PHA synthases can be divided into four classes with respect to subunit composition and substrate specificity. Recent fusion protein analysis indicated that fusion points located at the N-terminus of the conserved α/β-hydrolase fold region, and at a variable surface-exposed loop based on the current protein model, are tolerated by the synthases [1,17]. These data were suggesting that N-terminal fusions of reporter proteins would maintain enzyme activity and enable in vivo monitoring of PHA granule formation. Recently, it was demonstrated that the N-terminal domain of the PhaF phasin could be used as a polypeptide tag to attach fusion proteins to PHA granules . Further recent studies used Nile Red staining and fusion of GFP reporter protein to phasin proteins showed the localization of PHA granules . In this study, we N-terminally fused GFP to two different PHA synthases belonging to class I and II, respectively. Since only PHA synthases are required for PHA granule formation, the labeling of these enzymes should enable monitoring of early stages of PHA granule formation and positioning. Both GFP fusions remained active and led to formation of GFP-labeled PHA granules. This molecular tool leading to covalent labeling of the PHA granule surface was employed to monitor the in vivo formation of PHA granules and provided evidence for an important role of the cell division machinery.
2Material and methods
2.1Bacterial strains and growth conditions
Bacterial strains used in this study are listed in Table 1. Escherichia coli strains were grown in LB medium supplemented with 0.2% (w/v) decanoate sodium salt, if required. All E. coli strains were maintained at 37°C except of the mukB mutant which was maintained at 25°C. When required, antibiotics were used at the following concentrations: ampicillin, 75 μg ml−1; chloramphenicol, 25 μg ml−1; kanamycin, 50 μg ml−1; aztreonam, 1 μg ml−1 and tetracycline, 12.5 μg ml−1. Pseudomonas aeruginosa was grown at 37°C in 300 ml baffled flasks containing 50 ml of mineral salt medium containing either 0.05% (w/v) or 0.2% (w/v) ammonium chloride and 1.5% (w/v) sodium gluconate as carbon source, and if required antibiotics were added to appropriate concentrations. The antibiotic concentrations used for P. aeruginosa were as follows: gentamicin 300 μg ml−1 and aztreonam 5 μg ml−1. All chemicals were purchased from Sigma-Aldrich (St. Louis, Mo., USA).
Table 1. Bacterial strains, plasmids and oligonucleotides used in this study
Source or reference
aATCC, American Type Culture Collection, Manassas, VA, USA.
pBluesrcriptSK− containing Nde I/Bam HI-fragment comprising phaC from pHAS
pCWE containing SpeI adaptor in Nde I-site
pCWEA containing Spe I-inserted gfp gene derived from pPROBE-NT by PCR
pBHR80 containing SpeI adaptor in Nde I-site
pBHR80A containing Spe I-inserted gfp gene derived from pPROBE-NT by PCR
pBBR1JO5 containing an Xba I/Bam HI fragment from pBHR80Agfp comprising the gfp-phaC1 gene
5′-P-TATG GCTCTGCACTAGT CACTGC-CA-3
5′-P-TATG GCAGTGACTAGT GCAGAG-CA-3′
2.2Isolation, analysis and manipulation of DNA
DNA sequences of new plasmid constructs were confirmed by DNA sequencing according to the chain termination method using the model ABI310 automatic sequencer. All other genetic techniques were performed as described by Sambrook et al. .
Plasmids used in this study are listed in Table 1. Plasmids used to produce GFP–PHA synthase fusions were constructed as follows. A Spe I site harbouring adaptor was generated by hybridization of the oligonucleotides adaptor and adaptor reverse. The adaptor was inserted into Nde I site of pBHR80 and pCWE, respectively. The Spe I site was used to insert the gfp gene in frame with the respective PHA synthase gene. The gfp gene coding region was amplified by PCR using oligonucleotides 5′-gfp-Spe I and 3′-gfp-Spe I which provided Spe I sites. To investigate the GFP–PHA synthase in the natural host a broad host range construct was generated by subcloning the Xba I/Bam HI DNA fragment from pBHR80Agfp into the respective sites of pBBR1JO-5.
2.3Expression of GFP–PHA synthase gene
Except for P. aeruginosa cultivations, induction of production of GFP–PHA synthase fusions was achieved by addition of IPTG to the culture medium to 0.5 mM. For PHA synthase localization in cells filamented by inhibition of FtsI, aztreonam, a FtsI inhibitor, was added to cells at OD600 0.2, growth was continued to OD600 0.4–0.6. Then IPTG was added to the cells and growth was continued as long as required.
2.4In vivo GFP–PHA synthase activity
In vivo PHA synthase activity was obtained by analyzing PHA content of the respective bacterial cells. The amount of accumulated PHA corresponds to the relative in vivo PHA synthase activity. The PHA contents were qualitatively and quantitatively determined by gas chromatography/mass spectrometry (GC/MS) after conversion of the PHA into 3-hydroxymethylester by acid-catalyzed methanolysis.
2.5Localization of GFP–PHA synthase in the presence of substrate excess or limitation
Recombinant E. coli was used because metabolic control of substrate provision is available and to allow comparison between relevant mutants.
Class II enzymes accepting substrates ranging from C6–C14 atoms of 3-hydroxyacyl-CoA thioesters. To test the localization of GFP-PhaC1 from P. aeruginosa in the presence of excess substrate, either E. coli XL1-Blue cells were cultivated in the presence of acrylic acid and with decanoate as carbon source or E. coli fadB mutant LS1298 was cultivated with decanoate as carbon source [21,22]. To test the localization under substrate limitation the respective cells were grown in LB medium.
Class I enzymes accepting R-3-hydroxybutyryl-CoA thioester. To test the localization of GFP-PhaC from Cupriavidus necator in the presence of excess substrate, the fusion protein was co-produced with enzymes relevant for (R)-3-hydroxybutyryl-CoA provision. Plasmid pMCS69 encodes the enzymes β-ketothiolase and acetoacetyl-CoA reductase . To test the localization under substrate limitation the respective cells did not harbour plasmid pMCS69.
To investigate PHA granule formation mediated by GFP-PhaC1 in the natural host P. aeruginosa, we used plasmid pBBR1JO5-C1gfp and nitrogen-dependent regulation of substrate provision . Cells were grown in mineral salt medium with sodium gluconate as carbon source. For PHA accumulating conditions, cells were grown under nitrogen limitation with 0.05% (w/v) NH4Cl, while under non-PHA accumulating conditions 0.2% (w/v) NH4Cl was added to the media.
Bacteria were mounted onto a 1% agarose or LB agar pad on a 15-well slide (ICN). Unless otherwise indicated, staining for nucleoids was by incubation in the presence of 0.2 μg ml−1 DAPI for 5 min and/or staining for PHA granules was by incubation in the presence of 0.5 μg ml−1 Nile Red for 15 min immediately before mounting.
The spacing of GFP foci was measured as follows. The interval and regularity of spacing of GFP–PHA synthase foci was determined by measuring the distance from one end of the cell to each of >20 individual GFP foci per strain. For the ΔminCDE mutant, only filaments greater than or equal to two normal cell lengths were tabulated. For analysis of GFP–PHA synthase foci within nucleate or anucleate regions of chromosome partitioning mutants, the spacing between two foci was included only when both foci were located within the same nucleate or anucleate region.
3Results and discussion
3.1Co-localization of GFP–PHA synthase with PHA granules
GFP–PHA synthases either belonging to class I or class II enzymes showed in vivo activity as indicated by polyester accumulation. Co-localization of GFP–PHA synthase foci (Fig. 1A) and Nile Red foci (Fig. 1B) was demonstrated using stationary growth phase cells and providing excess of substrate. Under these conditions, cells producing either class I or class II GFP-PHA synthase were packed with GFP-labeled PHA granules. Thus, labeling of PHA granules could be achieved via N-terminal fusion of the reporter protein GFP to the PHA synthase. When analyzing exponential growth phase cells, dividing cells clearly showed GFP–PHA synthase foci at the poles and at the potential cell division site which provides the future poles (Fig. 1C–E). These data were suggesting that the cell division and the localization of GFP–PHA synthase foci are in vivo co-located. Moreover, the GFP–PHA synthase fusions provide an efficient tool to monitor the biogenesis and subcellular distribution of early stages of PHA granules.
3.2Emergence of PHA granules and subcellular distribution
The formation of PHA granules was investigated with temporal resolution in the absence or presence of excess substrate. In the absence of excess substrate PHA granules remained at early stage development not detectable by Nile Red staining. Only the GFP–PHA synthase foci were localized to the cell poles with dominating foci at one pole and to potential cell division sites determining future poles (Fig. 2). However, in the presence of substrate Nile Red stainable PHA granules were formed in stationary cells residing in the cytosol and descending from the poles (Fig. 2). In early growth stages (after 2 h incubation and induction), GFP–PHA synthase appeared already focused at both poles in the presence of excess substrate, while in the absence of substrate, the localization of GFP–PHA synthase was delayed. These data suggest that under PHA accumulating conditions the covalently attached and growing polyester chains enhance subcellular localization. This was found in recombinant E. coli as well as in the natural host P. aeruginosa (Fig. 2).
3.3Subcellular distribution and recognition of polar positional information
Analysis of the outer membrane protein IcsA from Shigella in E. coli indicated that IcsA localizes to sites of future poles independent of tubulin-like FtsZ, MinCDE, and nucleoid occlusion. To investigate how polar positional information is recognized by PHA synthase, we analyzed localization of the PHA synthase and PHA granule formation in E. coli cells inhibited by aztreonam and in E. coliΔminCDE and mukB mutants.
3.3.1Rapid real-time movement of early GFP-PhaC1 foci inside the bacterial cell
Analysis of cells of E. coli (pBHR80Agfp) from early growth stages and cultivated under non-PHA accumulating conditions showed a rapid oscillating movement of small GFP-PhaC1 foci between the poles and between the pole and, if present, a potential cell division site (Fig. 3 and supplementary video files as Fig. 3A–C). A maximum speed of >2 μm s−1 was detected, when analyzing movement from pole to pole (Fig. 3A). However, cells at different growth stages showed a different speed of oscillating movement (data not shown). The slowest movement was <0.5 μm s−1 (Fig. 3B). The GFP-PhaC1 foci presumably representing early stage PHA granules were not detectable by Nile Red staining. Since only the MinCDE proteins in E. coli were found to show an oscillatory rapid movement, expression of GFP-PhaC1 was analyzed in a ΔminCDE mutant to determine a potential role of these oscillating proteins. The MinCDE proteins enable precise determination of cell division site  by functioning as a negative regulator of FtsZ ring placement . In the absence of Min proteins, FtsZ forms rings at each of the three potential division sites, leading to a mix of short filaments, anucleate mini cells, and normal-sized cells. In contrast to wildtype E. coli, in the ΔminCDE mutant no GFP foci were detectable indicating a role of Min proteins in oscillating movement of PHA synthase and presumably early PHA granules (data not shown). The PHA synthase or small emerging PHA granules might interact with the Min proteins.
3.3.2Upon inhibition of septation, GFP–PHA synthase localized to or near cell division sites
We analyzed localization of GFP–PHA synthase in E. coli treated with aztreonam, which blocks FtsI, a cell division protein that is recruited towards the end of the cell division cascade [28,29]. Cells were analyzed in the absence and presence of excess substrate, respectively. Stationary growth phase cells cultivated for 16 h were analyzed. As expected, treatment with aztreonam induced the formation of long filaments (Fig. 4). GFP fluorescent foci from GFP–PHA synthase belonging to class I and II enzymes and the corresponding Nile Red fluorescent foci as well as DAPI-stained nucleoids localized at regularly spaced intervals along the length of the filaments and at the extreme ends of the filaments (Fig. 4 and data not shown). The GFP–PHA synthase foci were localized to gaps between the nucleoids (Fig. 4), which correspond to potential cell division sites, suggesting that GFP–PHA synthase was positioned at or near the cell division site. Treatment of the natural host P. aeruginosa (ΔphaC1-Z-C2) producing the GFP–PHA synthase with aztreonam led to filamented cells exerting subcellular localization of the PHA synthase and PHA granules consistent to the findings in E. coli (data not shown). Only PHA synthase was used as control. Nile Red fluorescent foci of granules formed by only PHA synthase were analyzed showing that GFP is not interfering with localization (data not shown). Since filamentous cells often show Z-ring formation at only every second or fourth potential cell division site  we found first evidence that the PHA synthase is presumably independent of FtsZ ring formation for its localization. GFP–PHA synthase formed fluorescent foci at regularly spaced intervals along the length of the filaments and at the extreme ends of the filaments (Fig. 5). The spacing of the GFP–PHA synthase foci of 3 ± 1 μm (mean ± SD) was similar to the spacing of cell division sites (Fig. 4). This spacing places approximately one focus per cell length. GFP–PHA synthase foci showed similar spacing intervals. Overall, these data indicated that PHA synthase localizes to or near potential division sites in filamentous cells independently of the FtsZ ring formation.
3.3.3Localization of GFP–PHA synthase foci is dependent of nucleoid occlusion
The localization of GFP–PHA synthase and corresponding PHA granules between nucleoids (Fig. 4) suggested that nucleoid occlusion causes positioning. Thus, the nucleoid could compete regarding space requirement inside the cell with PHA granule formation. If nucleoid occlusion plays a role in PHA synthase targeting, we assumed that PHA synthase localization would be random in anucleate cells or anucleate segments of filaments. Therefore, we investigated positioning of GFP–PHA synthase in anucleate segments of filamented cells. Null mutants in mukB exert defects in nucleoid structure and segregation, leading to about 5–13% of cells being anucleate . The localization of GFP–PHA synthase foci in mukB cells that were filamented with aztreonam was examined by quantifying only those cells that contained anucleate segments as indicated by DAPI staining. GFP–PHA synthase foci were found at irregular intervals along the filaments length of anucleate segments but were at regular intervals in nucleate cells (Fig. 5C and data not shown). The foci were found between nucleoids similar to those observed for foci in wildtype cells filamented with aztreonam (Fig. 4). Interestingly, PHA granules in nucleoid-free regions showed an about twofold increase of average diameter with about 1 μm and were clearly visible as GFP-labeled rings mainly organized in only one row along the cell length (Fig. 5A, B). Presumably, nucleoids and PHA granules compete with respect to intracellular space along the cell length. These data indicated that GFP–PHA synthase and PHA granule localization to or near potential cell division sites as well as PHA granule size is dependent of nucleoid occlusion.
We thank P. de Boer, M.B. Goldberg and W. Margolin for providing strains and J. Overhage for providing plasmid pBBR1JO5. B.H.A.R. was supported by a Massey University Research Funding grant and start-up funding from the Institute of Molecular Biosciences at Massey University.
Appendix A Supplementary data
Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.femsle.2005.05.027.