Pseudomonas aeruginosa strains causing chronic pulmonary infections in cystic fibrosis patients produce high levels of alginate, an exopolysaccharide that confers a mucoid phenotype. Alginate is a linear polymer of d -mannuronate (M) and variable amounts of its C-5-epimer, l -guluronate (G). AlgG is a periplasmic C-5-epimerase that converts poly d -mannuronate to the mixed M+G sequence of alginate. To understand better the role and mechanism of AlgG activity, a mutant was constructed in the mucoid strain FRD1 with a defined non-polar deletion of algG . Instead of producing poly mannuronate, the algG deletion mutant secreted dialysable uronic acids, as does a mutant lacking the periplasmic protein AlgK. High levels of unsaturated ends and the nuclear magnetic resonance spectroscopy pattern revealed that the small, secreted uronic acids were the products of extensive polymer digestion by AlgL, a periplasmic alginate lyase co-expressed with AlgG and AlgK. Thus, AlgG is bifunctional with (i) epimerase activity and (ii) a role in protecting alginate from degradation by AlgL during transport through the periplasm. AlgK appears to share the second role. AlgG and AlgK may be part of a periplasmic protein complex, or scaffold, that guides alginate polymers to the outer membrane secretin (AlgE). To characterize the epimerase activity of AlgG further, the algG4 allele of poly mannuronate-producing FRD462 was shown to encode a protein lacking only the epimerase function. The sequence of algG4 has a Ser-272 to Asn substitution in a serine–threonine-rich and conserved region of AlgG, which revealed a critical residue for C-5-epimerase activity.
Alginate is a high-molecular-weight, linear polysaccharide composed of β-d-mannuronic acid (M) and its C5 epimer, α-l-guluronic acid (G), linked by β-1,4 glycosidic bonds (Evans and Linker, 1973). These two residues do not form a repeating unit. Rather, the structure of alginate is variable and can play an important role in its function. Alginate is initially synthesized as polymannuronic acid (polyM), and then a variable fraction of the mannuronate residues in the polymers is modified by C5-epimerase (AlgG) to l-guluronate. This results in alginate with random blocks of mannuronate residues (M-blocks) and mannuronate blocks interspersed by l-guluronate residues (MG-blocks) (Chitnis and Ohman, 1990; Franklin et al., 1994). However, G-blocks are not observed in alginate from P. aeruginosa (Sherbrock-Cox et al., 1984). Some M residues are also modified by esterification with O acetyl groups at either the O-2 and/or O-3 positions, producing an O-acetylated polymer (Davidson et al., 1977). O acetylation of alginate is required for P. aeruginosa to form microcolonies in a biofilm (Nivens et al., 2001) and also maximizes the resistance of mucoid P. aeruginosa to antibody-independent opsonic killing by conferring resistance to alginate-specific antibodies found in the sera of infected CF patients (Pier et al., 2001).
Most of the genes required for alginate biosynthesis are in the algD operon of 12 genes on the P. aeruginosa chromosome (Chitnis and Ohman, 1993). The activated precursor of alginate, GDP-mannuronic acid, is synthesized from fructose-6-phosphate via the action of AlgA (phosphomannose isomerase-GDP-d-mannose phosphorylase), AlgC (phosphomannomutase) and AlgD (GDP-mannose dehydrogenase) (May et al., 1991). AlgC is the sole known biosynthesis gene that is encoded by an unlinked gene, and is involved in both alginate and lipopolysaccharide (LPS) biosynthesis (Zielinski et al., 1991; Goldberg et al., 1993). The precursor, GDP-mannuronic acid, is polymerized and transported across the inner membrane by an unknown mechanism that may involve the gene alg8, which encodes a protein resembling β-glycosyl transferases (Saxena et al., 1995). Some of the M residues in the initial polymer, polyM, are epimerized to G residues by a periplasmic C-5-epimerase encoded by algG (Chitnis and Ohman, 1990; Franklin et al., 1994). The gene products of algI, algJ and algF are responsible for acetylation of some M residues (Franklin and Ohman, 1993; 1996). The algL gene in the algD operon encodes an alginate lyase that can depolymerize alginate (Schiller et al., 1993), but the function of which in alginate biosynthesis is undetermined. The algK gene encodes a periplasmic protein, and deletion of the gene results in the secretion of small uronic acids, instead of alginate, suggesting that AlgK is required for polymer production (Jain and Ohman, 1998). The roles of alg44 and algX are unclear. Alginate appears to be secreted across the outer membrane through a protein resembling a secretin encoded by algE (Rehm et al., 1994).
The algG gene was originally identified by screening survivors of random chemical mutagenesis for defects in epimerase activity. A derivative of the mucoid CF strain FRD1 was isolated, called FRD462, that retained the mucoid phenotype but secreted a polymer comprising only M residues (polyM). The mutation was mapped to the algG gene (Chitnis and Ohman, 1990). The AlgG protein was subsequently shown to be a periplasmic C-5-epimerase that acts directly upon polyM to yield the mixed M and G polymer (polyMG) of alginate (Franklin et al., 1994) (Fig. 1). In this study, we deleted algG from the FRD1 chromosome, which revealed a second function for AlgG, which is to protect new polymers from AlgL lyase during transport out of the cell. We also sequenced the algG4 allele from FRD462, which identified a critical residue for C-5 epimerase activity.
Deletion of algG blocks alginate polymer synthesis
Strain FRD1 is a typical CF isolate of P. aeruginosa, in that it has a highly mucoid colony morphology resulting from the overproduction of alginate (Ohman and Chakrabarty, 1981). Its derivative, FRD462, is defective in C-5- epimerase activity due to an algG4 mutant allele (Chitnis and Ohman, 1990; Franklin et al., 1994). In this study, we examined the effect of a defined, non-polar algG deletion in the FRD1 chromosome by constructing FRD1200, in which algG sequences were replaced with a non-polar gentamicin (Gm) resistance cartridge (Fig. 2). Interestingly, instead of producing polyM like FRD462 and hence being mucoid, the ΔalgG mutant had a non-mucoid phenotype. This indicated a defect in the production or release of alginate polymer. The mutation in FRD1200 was verified by polymerase chain reaction (PCR) analysis (see Experimental procedures). A complementation analysis was also performed to ensure that the ΔalgG::Gm mutation in the FRD1200 chromosome was not just polar on the expression of downstream genes. Introducing a previously described plasmid containing algG under a Ptrc promoter (pMF55) into FRD1200 restored the mucoid phenotype. Thus, AlgG appears to be a bifunctional protein with roles in epimerization and polymer production.
Loss of AlgG results in secretion of uronic acids
We showed recently that the loss of AlgK, another periplasmic protein encoded by the alginate biosynthetic operon (see Fig. 2), results in a non-mucoid phenotype and also in the secretion of dialysable uronic acids (Jain and Ohman, 1998). However, a defined role for AlgK that could explain why its loss results in the release of the subunits of the alginate was not apparent. Here, we tested whether the ΔalgG mutant FRD1200 might display a similar phenotype. Samples of culture supernatants of FRD1, FRD462 and FRD1200 were each subjected to equilibrium dialysis against an equal volume of buffer and tested for dialysable uronic acids. As expected, the supernatants of mucoid strains FRD1 and FRD462 had all uronic acids retained by the dialysis membrane because they contained high-molecular-weight polymers (Table 1). However, the supernatant of the ΔalgG mutant FRD1200 contained high levels of uronic acids that were diffusible through a dialysis membrane with 10 kDa cut-off pores (Table 1). Upon dialysis against 4 l of buffer, FRD1 and FRD462 polymeric uronic acids were still retained by the dialysis membrane, but all uronic acids from the ΔalgG mutant dialysed away into the large volume of buffer. This result demonstrated that ΔalgG mutants have a similar phenotype to ΔalgK mutants, suggesting that both AlgG and AlgK proteins are required for the secretion or polymerization of uronic acids (i.e. alginate).
Table 1. . Equilibrium dialysis of uronic acids secreted by FRD1 (wild type, mucoid), FRD462 ( algG4 ) and FRD1200 (Δ algG ). a
a. Samples (7 ml) of MAP culture supernatants of FRD1, FRD462, FRD1200 and FRD1100 were placed in dialysis bags (10 000 or 1000 Da cut-off) and dialysed against an equal volume (7 ml) of Tris-HCl buffer. Uronic acids (UAs) in/out of the dialysis bag were determined by the carbazole method and are shown as mg of uronic acids ml −1 . The data represent one experiment that was repeated several times with comparable results.
. After equilibrium dialysis, samples in the dialysis bags were then placed in 4 l of Tris-HCl buffer overnight, and the concentrations of UAs in the bags were determined. ND, not done.
To characterize the size of the secreted uronic acids better, we tested diffusion through membranes with an average pore size of 1 kDa, which would be predicted to retain small oligomers. The results of this equilibrium dialysis of culture supernatants showed that more uronic acids were retained than diffused with both the ΔalgG mutant and the ΔalgK mutant (Table 1). Upon nearly exhaustive dialysis, ≈ 10% of the carbazole-reactive uronic acids secreted by both mutants was still retained by the 1 kDa cut-off membrane (Table 1). This suggested that both mutants secreted a heterogeneous population of uronic acids as short oligomers.
Loss of AlgG or AlgK results in release of degraded alginate oligomers
One explanation for the oligouronide-producing phenotype was that loss of AlgK or AlgG caused a chain length defect such that only short alginate chains were formed. Alternatively, the loss of AlgK or AlgG in the periplasm could result in the formation of a polymer that was now accessible to periplasmic AlgL, which can degrade alginate. AlgL is encoded by the alginate biosynthetic operon (see Fig. 2) and is a periplasmic alginate lyase (Schiller et al., 1993). The latter hypothesis could be readily evaluated by testing the secreted uronic acids for unsaturated bonds that would be formed at the non-reducing ends of oligomers if glycosidic linkages are cleaved by a lyase (see Experimental procedures; Preiss and Ashwell, 1962). Also, AlgL protein was partially purified from FRD1 for these tests. When AlgL was incubated with culture supernatants of FRD1 and FRD462 containing polymeric uronic acids, a threefold increase in unsaturated ends was observed with each (i.e. 3.0 OD units; see Table 2), indicating that AlgL from FRD1 exerted lyase activity on both alginate and polyM as expected. In contrast, the equivalent amount of uronic acids from the ΔalgG or ΔalgK mutants showed the same high levels of unsaturated ends (i.e. 3.0 OD units) even without the addition of purified lyase. This suggested that dialysable uronic acids secreted by ΔalgG and ΔalgK mutants were the result of rapid degradation of the newly formed polymer by AlgL in the periplasm.
Table 2. . Relative measurement of unsaturated ends resulting from the activity of AlgL, an alginate lyase.
a. Samples of MAP culture supernatants containing 0.2 mg of uronic acids (UAs) in a thiobarbituric acid assay for unsaturated bonds formed at the non-reducing end. Samples containing polymers were pretreated to remove acetyl groups, and some were incubated with partially purified AlgL, an alginate lyase from P. aeruginosa . The A 535 values (without background subtraction) represent relative amounts of unsaturated ends. The data represent one experiment that was repeated several times with comparable results.
The nature of these dialysable uronic acids secreted into culture supernatants was examined further using nuclear magnetic resonance (NMR) spectroscopy. An NMR analysis of polyM produced by FRD462 yields only one peak designated M-1 (Chitnis and Ohman, 1990; Franklin and Ohman, 1996). The oligouronides released by ΔalgG mutant FRD1200 were concentrated, desalted and then analysed by 1H-NMR spectroscopy at 400 MHz. The spectrum between 6.0 and 4.5 p.p.m. (Fig. 3) produced a pattern that was strikingly similar to the results of other studies that analysed the products of alginates degraded by AlgL from Azotobacter vinelandii, an alginate lyase homologous to AlgL of P. aeruginosa (Rehm et al., 1996). Cleavage of an M-M bond results in an M on the reducing end (labelled Mred) and ΔM on the non-reducing end. Δ denotes an unsaturated 4-deoxy-l-erythro-hex-4-enepyranosyluronate residue, which is formed at the non-reducing ends of alginate when degraded by alginate lyase (Ertesvåg et al., 1998). A small M-1 peak was observed representing 1H on M residues internal to the ends. The area of this peak is so small that most of the oligouronides cannot contain any internal residues, suggesting that unsaturated dimers are the main product produced by FRD1200. Any G-1 peak would be expected at 5.1 p.p.m. (Ertesvåg et al., 1998), but none was observed because no G residues would be formed in the absence of AlgG. The nature of the signal at 4.9 p.p.m. is currently unknown. As a control, the components of MAP medium, in which the cells were grown, were examined by NMR, and no effect on this spectrum was observed (data not shown). Overall, these data demonstrate that the dialysable uronic acids being secreted by ΔalgG mutants are small oligouronides resulting from alginate degradation by AlgL. Thus, AlgG and AlgK probably play a role in protecting new polymers from degradation by AlgL during transport through the periplasm to the outer membrane.
Complementation analysis shows that AlgG4 plays a role in polymer formation without C-5-epimerization
The algG4 allele in FRD462 appears to encode an altered protein that still plays its role in protecting polymers from AlgL during transport despite its C-5-epimerization enzymatic defect. To test this hypothesis, the algG4 allele was amplified from the FRD462 chromosome by PCR and cloned into a broad-host-range Ptrc expression vector to form pSJ209. When the plasmid-borne algG4 on pSJ209 was conjugated into FRD1200 (ΔalgG::Gm), the mucoid phenotype was restored, indicating complementation for polymer formation. FRD1200(pMF55), expressing wild-type algG in trans, produced the same phenotype. Next, we determined whether FRD1200(pSJ209) expressing algG4 was secreting polyM. To test this, an alginate G-lyase from Klebsiella aerogenes (see Experimental procedures) was used that preferentially cleaves glycosidic bonds between M and G residues (Boyd and Turvey, 1977). This G-lyase reacts with alginate and forms unsaturated bonds at the non-reducing ends of the cleaved polymer, which can be measured in the thiobarbituric acid assay. This G-lyase has been shown to depolymerize alginate from FRD1, but it produces only a weak reaction with the polyM made by FRD462 (Chitnis and Ohman, 1990). This was verified with our preparation of the G-lyase (Table 3). We also showed that FRD1200(pMF55) containing wild-type algG in trans produced a polymer that reacted with the G-lyase. In contrast, the polysaccharide secreted by FRD1200(pSJ209) expressing algG4 showed only a weak reaction with G-lyase (Table 3). These genetic complementation results verified that AlgG4 retains its functional role in polymer formation, but is defective in C-5-epimerization activity. In addition, purified AlgL was shown to degrade both polyM produced by FRD1200(pSJ209) as well as polyMG produced by FRD1, producing> 2 units of activity in the thiobarbituric acid assay (data not shown). This reaction was consistent with the hypothesis that the small oligouronides secreted by ΔalgG and ΔalgK mutants were the result of AlgL degradation of newly formed polymers.
Table 3. . Reaction of a Klebsiellal -guluronate lyase on polymers produced by P. aeruginosa expressing plasmid-borne algG or algG4 .
Reaction with G-specific lyase (%)
The algG or algG4 allele was expressed from the chromosome (FRD1, FRD462) or in trans from plasmids (pMF55, pSJ209). A 0.2 mg sample of deacetylated polymer from each was treated with 20 µl of a G-specific lyase from Klebsiella. The TBA assay was to quantify the unsaturated bonds formed at the non-reducing ends. Units represent the A535 after subtraction of background and also the percentage activity relative to the FRD1 alginate reaction. The reaction of polyM with G-lyase indicates that enzyme was in excess. The data represent one experiment that was repeated several times with comparable results.
Sequence analysis of AlgG4
The C-5-epimerase catalytic site of AlgG is unknown. However, the specific alteration in AlgG4, defective in C-5-epimerase activity, could reveal information about its structure–function relationships. To this end, the sequence of the algG4 allele from FRD462 was determined and compared with that of the published sequence of wild-type algG (Franklin et al., 1994). This revealed that the algG4 allele has only one mutation, a transition of G to A at position 815. Upon translation, this mutation results in the replacement of serine (AGC) by an asparagine (AAC) at residue 272 in the 544-amino-acid protein. This result suggests that Ser-272 is a critical residue required for C-5-epimerase activity in AlgG.
In contrast to most other polysaccharides, alginate does not contain a repeating sequence along the polymer chain, and so there can be substantial heterogeneity among alginates from different organisms (Smidsrød and Draget, 1996). This compositional variability, as well as the sequential distribution of M and G along the chain, can have profound effects on chain flexibility, viscosity in aqueous solution and in the ability to form gels in the presence of divalent cations (Smidsrød and Draget, 1996). Alginate is produced by organisms for a variety of purposes, and its composition often appears to play a role. For instance, in brown seaweeds, alginate has a structural role in the tissues, where stiff structures are observed to contain more G-rich alginate than flexible structures (Smidsrød and Draget, 1996). In A. vinelandii, alginate composition plays a role in the formation of cysts (Sadoff, 1975). A. vinelandii has an AlgG-like C-5-epimerase that is required to form MG-blocks, but it also secretes extracellular C-5- epimerases (called AlgE) that can introduce GG-blocks into the alginate of the cyst wall (Valla et al., 2001). In P. aeruginosa, alginate production is associated with chronic pulmonary disease in CF patients and in microcolony biofilm formation (Nivens et al., 2001), but a role for M-G composition is as yet uncharacterized.
When mutants of mucoid P. aeruginosa were discovered that produced polyM as a result of algG mutation, this showed that C-epimerization is not a required step for polymer formation (Chitnis and Ohman, 1990). Also, in vitro biochemical studies show that AlgG is a C-5-epimerase whose action occurs remarkably at the polymer level (Franklin et al., 1994). To characterize AlgG further, we constructed a derivative of the CF isolate FRD1 with a defined algG deletion by replacing it with a non-polar gentamicin resistance cartridge. Interestingly, the non-polar ΔalgG::Gm mutation conferred a non-mucoid phenotype, instead of polyM production like the previously described algG4 point mutant (Chitnis and Ohman, 1990). This finding was consistent with a similar ΔalgG mutant in Pseudomonas fluorescens that also failed to produce detectable amounts of alginate (Morea et al., 2001). These data suggested that AlgG has a second function in alginate polymer synthesis.
Another important phenotype of the ΔalgG mutant was the secretion of dialysable uronic acids instead of polymerized alginate. Loss of AlgK, another periplasmic protein encoded by the algD operon, also results in a non-mucoid phenotype and the secretion of dialysable uronic acids (Jain and Ohman, 1998), suggesting that AlgG and AlgK share a similar function. None of the uronic acids secreted by ΔalgG and ΔalgK mutants was retained by dialysis membranes with a cut-off of 10 000 Da, but a small amount of the uronic acids was retained by a pore size that would restrict>1000 Da proteins. This suggested that the uronic acids secreted included some oligomers and, thus, at least some polymerization of alginate was taking place in ΔalgG and ΔalgK mutants.
We tested the hypothesis that newly synthesized alginates in the ΔalgG and ΔalgK mutants were susceptible to depolymerization by AlgL, a periplasmic alginate lyase in P. aeruginosa encoded by a gene that is also located in the algD operon (Schiller et al., 1993). It remains a mystery why the operon for alginate biosynthesis should encode an enzyme for alginate degradation. If glycosidic linkages were hydrolysed by AlgL lyase, then the secreted uronic acids should have unsaturated moieties at the non-reducing ends. Indeed, compared with the polymers of FRD1 and FRD462, threefold more unsaturated ends were observed in the dialysable uronic acids, which indicated hydrolysis of the polymeric form of alginate. The oligomers secreted by an ΔalgG mutant were also examined by 1H-NMR. The NMR spectrum produced a pattern that was characteristic of AlgL lyase degradation of polyM in vitro, including the formation of 4-deoxy-l-erythro-hex-4-enepyranosyluronate residues, as shown recently (Ertesvåg et al., 1998). The spectrum also suggested that the predominant species is a dimer. In a parallel investigation, an ΔalgG mutant of P. fluorescens was shown by mass spectroscopy to produce predominantly a dimeric oligomannuronic acid that is unsaturated at the non-reducing end (M. Gimmestad, H. Sletta, H. Ertesvåg, K. Bakkevig et al., submitted). These data together are consistent with the hypothesis that ΔalgG and ΔalgK mutants were polymerizing alginate, but it was highly susceptible to the lyase activity of AlgL, resulting in secretion of short oligouronides. A likely possibility is that AlgG and AlgK are both part of a periplasmic scaffold that guides alginate polymers to the outer membrane secretin (AlgE) and protects it from the lyase activity of AlgL. Future studies will examine the nature of this proposed scaffold complex by tests for protein–protein interactions.
AlgG is thus bifunctional, required for both release of high-molecular-weight polymer from the cell and for C-5-epimerization of M residues in alginate. AlgG's C-5- epimerization activity on β-d-mannuronate (M) is an inversion of the configuration at carbon atom 5 to form the corresponding 5-epimer, α-l-guluronate (G) (see Fig. 1). The AlgG4 variant was shown to be defective in its epimerase activity, but it still protected polymer from AlgL lyase activity, perhaps by assembling in a periplasmic scaffold for polymer transport. A sequence analysis revealed a single mutation in algG4, resulting in a replacement of Ser-272 for Asn in the middle of the 544-amino-acid protein. This suggests that Ser-272 is critical for C-5-epimerization activity.
The amino acid sequence of AlgG from P. aeruginosa was recently compared with the AlgG sequences in P. fluorescens and A. vinelandii, which show 67% and 60% overall identity respectively (Morea et al., 2001). Interestingly, a serine residue is not found at relative position 272 in the other two AlgG proteins but, instead, there is a threonine residue that is apparently equivalent for function. Asparagine is only slightly larger than threonine, suggesting that R-group size is not the cause of epimerase dysfunction in AlgG4. Serine and threonine, which can be found at position 272 in wild-type AlgG sequences, are highly related for function in that they are the only two amino acids with aliphatic hydroxyl side-chains. For example, in serine proteases (e.g. trypsin), this aliphatic hydroxyl group of the reactive serine is critical for the initial reaction with substrate that leads to hydrolysis. Although Ser-272 is located in a region that shows extensive homology among AlgG proteins (Fig. 4), there is extensive homology throughout the AlgG proteins (Morea et al., 2001). It was also observed that this region tends to be rich in conserved serine/threonine (S/T) residues, some of which may play a role in enzymatic activity or conformation. Substitutions by site-directed mutagenesis in this region may reveal more about its importance to C-5-epimerization activity. A computer analysis of P. aeruginosa AlgG structure (Chou–Fasman) predicts that α-sheets dominate the amino-terminus, that the centre region is particularly abundant in β-strands with Ser-272 located at a turn and that the C-terminus has a mix of both (Fig. 4).
A blast analysis search for AlgG-like proteins also produced the secreted alginate C-5-epimerase of A. vinelandii, AlgE7. Although AlgE7 has weak overall homology to AlgG, residues 321–479 show 46% similarity, which interestingly is outside the Ser-272 region (Fig. 4). The residue analogous to AlgG's Asp-324 in AlgE7 was predicted to be important for the catalytic activities of AlgE7, which include C-5-epimerase and alginate lyase activities (Svanem et al., 2001). Both Ser-272 and Asp-324 are relatively close in the central β-rich region. In a parallel investigation of algG mutants in P. fluorescens that produce polyM, the four substitutions identified were in the region with similarity to AlgE7 (M. Gimmestad, H. Sletta, H. Ertesvåg, K. Bakkevig et al., submitted). Studies are in progress to identify other residues important for C-5-epimerzation activity to understand AlgG structure and function better.
Bacterial strains, plasmids and media
Pseudomonas aeruginosa strains used in this study are shown in Table 4 . E. coli and P. aeruginosa were routinely grown in L broth (10 g of tryptone, 5 g of yeast extract, 5 g of NaCl per litre). Plasmids were transferred from E. coli to P. aeruginosa by triparental matings, with the conjugative helper plasmid pRK2013 ( Figurski and Helinski, 1979 ). After the conjugation, P. aeruginosa was selected using a 1:1 mixture of Pseudomonas isolation agar (Difco) and L agar (Sigma). MAP medium was used for growing cultures before the uronic acid assay. MAP is a defined medium that promotes alginate production and has been described previously ( Franklin and Ohman, 1996 ). Antibiotics were used at the following concentrations (per ml): ampicillin (Ap) at 100 µg, carbenicillin (Cb) at 300 µg, gentamicin (Gm) at 15 µg for E . coli and 300 µg for P. aeruginosa , and kanamycin (Km) at 40 µg. Sucrose was added to L agar at a concentration of 7.5%.
Table 4. .P. aeruginosa strains and plasmids used.
Contains algL encoding a periplasmic alginate lyase of P. aeruginosa FRD1, Apr
Restriction enzymes used to construct plasmids were purchased from Boehringer Mannheim/Roche Molecular Biochemicals or New England Biolabs. Klenow polymerase was used to blunt end restriction fragments. Plasmid DNA was extracted using the Qiaprep Spin Miniprep kit (Qiagen). Production of large-scale plasmid DNA was done using the Qiagen Plasmid Midiprep kit (Qiagen). DNA fragments for cloning were purified from agarose gels using the QIAEX II Gel Extraction System (Qiagen). Oligonucleotides were custom synthesized by Operon Technologies. DNA sequences were determined using the Big Dye Terminator cycle sequencing ready reaction kit and an automated sequencer from Applied Biosystems. Primers for sequencing were designed using the wild-type algG sequence, and numbered a total of 14, which read sequence off both strands of DNA. Sequence data were aligned and analysed with dnastar software by Lasergene on an Apple Macintosh computer.
Construction and confirmation of algGΔ::Gm mutation by PCR
The strategy for construction of an algGΔ mutant was to delete the gene and replace it with a non-polar selectable marker to preserve the transcription of downstream genes (see Fig. 2). pEX100T is a gene replacement vector with oriT for transfer by conjugation, bla for β-lactamase activity and sacB for sucrose sensitivity. A 3.7 kb DNA fragment containing algG was cloned into pEX100T (Schweizer and Hoang, 1995) and called pSJ155. A non-polar gentamicin resistance cassette (Jain and Ohman, 1998), obtained from pSJ12 on a SmaI fragment, was cloned into pSJ155 such that it replaced a HindIII fragment located in the internal one-third of algG. This plasmid, pSJ166, was conjugated to FRD1 with helper plasmid pRK2013 (Figurski and Helinski, 1979). Merodiploids were selected as carbenicillin-resistant colonies. Gene replacement mutants were obtained by growing merodiploids in the absence of selection, followed by selection for gentamicin and sucrose resistance, with screening for carbenicillin sensitivity. PCR was used to confirm insertion of the gentamicin cassette in the chromosome. The primers used were specific to the 5′ end of the Gmr gene and to the 3′ end of the algG gene. When these primers were used to amplify DNA of the wild-type FRD1, no PCR product was obtained, as expected. However, when these primers were used to amplify DNA of the mutant strain under the same conditions, a 1.3 kb fragment was obtained, which corresponded to the length of the gentamicin cassette (770 bp) plus the C-terminal fragment of algG downstream of the HindIII site (530 bp).
Cloning of algG4 by PCR
Primers used for PCR amplification were made using the sequence of the wild-type algG gene and were specific to the 5′ and 3′ ends of the gene. The 1.6 kb PCR product was first cloned into the vector pET-28(b). The algG4 allele was subsequently subcloned into pMF54, the parent vector of pMF55, to produce pSJ209. The algG4 allele of FRD462 was cloned by PCR amplifying the gene directly from DNA in an overnight culture of FRD462 diluted 1:10 in saline, and 5 µl was used as the template in a 100 µl PCR. The primers used for amplifying the gene were designed using the published algG sequence. The 5′ primer included an XbaI site, the ribosomal binding site of algG and the start codon of the gene (5′-CGA TCT AGA AGG AAA CCG GAC ATG CCC GAC ATT TCC CTT-3′). The 3′ primer contained a XhoI site downstream of the 3′ end of the algG gene (5′-GCT CTC GAG GTC CTG GAG TTC GGC CTG GCT TTC CAC GGG-3′). Amplification was performed using Pfu polymerase to give a product of 1.6 kb. This was cloned into the XbaI–XhoI sites of pET-28(b) (Novagen) to give the plasmid pSJ206. The fragment was later subcloned into the XbaI–XhoI sites of the Pseudomonas expression vector pMF54 (Franklin et al., 1994).
Uronic acid assay
Pseudomonas aeruginosa cultures to be tested for uronic acid secretion were grown for 20 h at 37°C in 10 ml of MAP medium. The cells were removed by centrifugation at 9000 r.p.m. for 1 h. For equilibrium dialysis, supernatants were placed in dialysis tubing (Spectra/Por membrane, MWCO 10 000 or 1000, 1.8 ml cm −1 ) and dialysed against an equal volume of 10 mM Tris-HCl, pH 7.6, at 4°C for 16 h. The two fractions corresponding to the dialysed (inside the tubing) and dialysate (outside the tubing) were then tested for uronic acid content by the method of Knutson and Jeanes (1968 ). Briefly, 30 µl of the sample was treated with 1.0 ml of borate–sulphuric acid reagent (0.1 M borate in concentrated sulphuric acid) to hydrolyse the polymers, if present. Next, the sample was mixed with 30 µl of carbazole reagent (0.1% in ethanol) and heated to 55°C for 30 min. The absorbance was determined at 530 nm using a spectrophotometer. Uronic acid concentrations were determined from a standard curve using Macrocystis pyrifera alginate (Sigma).
Preparation of lyases and assay of lyase activity
The gene aly of K. aerogenes encodes the secreted enzyme l-guluronate lyase, which specifically cleaves glycosidic linkages adjacent to G residues present in MG or GG blocks (Boyd and Turvey, 1977). Plasmid pRC5 has the aly gene cloned in it and can be overexpressed in E. coli to yield extracellular lyase (Caswell et al., 1989). E. coli cells containing pRC5 were grown overnight at 37°C and harvested the next day by centrifugation (10 000 g for 45 min). The l-guluronate lyase in the supernatant was ammonium sulphate precipitated (80% saturation), resuspended in lyase buffer (50 mM phosphate buffer, pH 7.0, 50 mM NaCl), dialysed extensively against lyase buffer and stored at 4°C.
The gene algL, located in the alginate biosynthetic operon, encodes a lyase that cleaves glycosidic bonds between residues of alginate. Plasmid pMF106 has the FRD1 algL gene cloned in the Pseudomonas expression vector pMF54, and the enzyme was obtained from E. coli cells containing this plasmid. Bacterial cells were sonicated, followed by centrifugation to remove unlysed cells and the insoluble membrane-associated fraction. The supernatant, containing AlgL activity, was stored at 4°C.
Alginate or polyM polymer was obtained by centrifuging 20 h cultures of the strains secreting them (8000 r.p.m. for 60 min) and using the supernatant as the sample. The samples were first deacetylated with 1 N NaOH (65°C for 15 min) and then acidified with 1 N HCl to neutralize the pH. Deacetylated polysaccharides were treated with either Klebsiella Aly or P. aeruginosa AlgL lyase for 1 h at room temperature. A colorimetric assay (Preiss and Ashwell, 1962) described by Chitnis and Ohman (1990) was used to measure the unsaturated residues formed at the non-reducing ends of the lyase-cleaved polymers or in secreted uronic acids. Briefly, the samples were treated sequentially with periodic acid (0.025 N in 0.125 N sulphuric acid) at room temperature for 40 min, sodium arsenite (2% in 0.5 N HCl) at room temperature for 1 min and 2-thiobarbituric acid (0.6%, pH 2.0) at 65°C for 30 min. The reactions were cooled, centrifuged, and absorbance was read at A535.
Isolation of oligouronides for 1H-NMR studies
FRD1200 was grown for 2 days at 30°C in the defined MAP medium. The bacteria were removed by centrifugation, and the supernatant was filtered to remove any unpelleted bacterial cells. Filtrate (180 ml) was dialysed (MWCO 12–14 000 Da) twice against an equal amount of deionized water to yield 400 ml of sample outside the dialysis tubing, which was concentrated by freeze-drying. The freeze-dried material was then subjected to desalting. First, the sample was dissolved in 50 ml of deionized water. Activated charcoal (2.5 g) and 2.5 g of celite were added, and the suspension was stirred for 2 h and centrifuged for 30 min at 11 000 g. The charcoal–celite mixture was washed twice with 50 ml of water, and the oligouronides were extracted three times by the addition of 50 ml of 80% ethanol. The oligouronides were then concentrated by evaporation (rotavapour) and freeze-drying. These samples were analysed by proton-NMR as described previously (Ertesvåg et al., 1998).
We gratefully acknowledge the assistance of Wenche Strand in the NMR analysis. This work was supported by Veterans Administration Medical Research Funds (D.E.O.), Public Health Service grants AI-19146 (D.E.O.) and AI-46588 (M.J.F.) from the National Institute of Allergy and Infectious Diseases, the Center for Biofilm Engineering at Montana State University (M.J.F.), NSF-supported Engineering Research Center Co-operative Agreement EEC-8907039 (M.J.F.) and by grants from The Norwegian Research Council and FMC Biopolymers AS (S.V.).