INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Chronic pulmonary disease in patients with cystic fibrosis is commonly caused by mucoid strains of Pseudomonas aeruginosa that overproduce an exopolysaccharide called alginate. Conversion to the mucoid phenotype occurs as the result of mutations in alginate regulatory loci that may encode a stress response system (14, 16, 24, 25, 29). Alginate has been implicated in facilitating the persistence of P. aeruginosa in the lungs by conferring an antiphagocytic barrier (3, 37), frustrating phagocytic cells (42), obstructing the bronchioles by its high viscosity (17), and providing a mechanism for adherence (23, 32) leading to the formation of biofilms (21). Additionally, the high resistance of P. aeruginosa to antibiotics makes these strains intractable, and they can overwhelm the cystic fibrosis patient by their sheer numbers (16, 41).

Alginate is a linear polymer of high molecular weight composed of the uronic acids -D-mannuronate (M) and its C-5 epimer, -L-guluronate (G), which are linked by -1,4 glycosidic bonds (9). Analysis of P. aeruginosa alginate by nuclear magnetic resonance shows that the two moieties are present in blocks of poly-M and poly-MG but not poly-G (6). Most of the genes required for alginate biosynthesis are in the algD biosynthetic operon at 34 min on the 75-min P. aeruginosa chromosome (6, 8). The only known exception is algC, at 10 min on the chromosome, which encodes phosphomannose mutase, an enzyme that is also involved in lipopolysaccharide biosynthesis (15, 43).

The early steps in the biosynthesis of alginate to form GDP-mannuronic acid (Fig. 1) require the products of algA and algC to convert fructose-6-phosphate to GDP-mannose and then require the product of algD to convert this to the uronic acid form, GDP-mannuronate (for a review, see reference 26). However, the mechanisms by which this activated monomer is utilized to form the next intermediate, and how it is transported across the inner membrane and polymerized, are still largely unknown. The genes called alg8 and alg44, immediately downstream of algD (22), are required for alginate production (8). Sequence comparison studies suggest that the alg8 product may be a glycosyl transferase (34). The algG gene encodes a periplasmic C-5-epimerase that introduces Gs into alginate by epimerization of Ms at the polymer level in the periplasm (5, 11). Alginate is also modified by acetylation of M moieties in the periplasm at positions O-2 and/or O-3, and this requires the products of algI, algJ, and algF (12, 13). The algE gene encodes a protein that may form a porin in the outer membrane through which the polymer is excreted out of the cell (7, 33). The algL gene encodes an alginate lyase whose role in alginate biosynthesis is unclear, and algX is a gene of unknown function (36).

View larger version (12K): [in this window] [in a new window]   FIG. 1.   Pathway leading to the synthesis of alginate in P. aeruginosa. Known genes required at each step and the products they encode are indicated. The transport and polymerization enzymes are unknown.

A region of approximately 1.2 kb between alg44 and algE remained uncharacterized, although it may encode a potentially important component of the alginate biosynthetic machinery. A recent report showed that in strain 8873 this region contains an open reading frame, which was termed algK (1). Here we extend this initial observation and describe a genetic characterization of algK, which includes the effects of nonpolar mutations in FRD1, a cystic fibrosis isolate and mucoid strain. Our results suggest that AlgK plays a critical role in the formation of poly-M from activated monomers in the pathway leading to alginate secretion.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Bacterial strains, plasmids, and media. The bacterial strains and plasmids used in this study are shown in Table 1. P. aeruginosa and Escherichia coli were routinely cultured in L broth (10 g of tryptone, 5 g of yeast extract, 5 g of NaCl [per liter]). Triparental matings were used to mobilize plasmids from E. coli to P. aeruginosa with the conjugative helper plasmid pRK2013 (10). A 1:1 mixture of Pseudomonas Isolation Agar (Difco) and L agar (Difco) was used to select for P. aeruginosa following matings with E. coli. MAP, a defined medium that promotes alginate production by P. aeruginosa, was previously described (13). When used, antibiotics were at the following concentrations (micrograms per milliliter): ampicillin, 100; carbenicillin, 300; gentamicin, 15 for E. coli and 250 for P. aeruginosa; and kanamycin, 40. 

DNA manipulations. Restriction endonucleases were purchased from Boehringer Mannheim or New England Biolabs. Klenow polymerase was used to blunt the ends of restriction fragments. Large-scale preparations of plasmid DNA were made by using the QIAprep Plasmid Midi system (Qiagen). DNA fragments for cloning were purified from agarose gels by using the QIAEX II Gel Extraction system (Qiagen). Oligonucleotide primers were synthesized on an Applied Biosystems 380B synthesizer. DNA for sequence analysis was obtained from pALG2 (5), which contains the entire alginate biosynthetic operon from strain FRD1; a 1.2-kb KpnI fragment from pALG2 containing DNA between alg44 and algE was cloned into the single-stranded phage vector M13mp18 in both orientations, to generate pSJ5-1 and pSJ5-2. DNA sequences were determined on pSJ5-1 and pSJ5-2 with the Taq DyeDeoxy Terminator Cycle Sequencing system and an automated sequencer (model 373A) from Applied Biosystems. Primers used for sequencing were the M13mp18 universal forward and reverse primers as well as primers designed from the sequences obtained (a total of 32) to determine the sequence on both DNA strands. Adjacent sequences were determined with pMF100 as a template. Sequence data were analyzed with Lasergene software (DNA-Star) on a Macintosh (Apple) computer. Sequences were aligned with published sequences upstream and downstream of this region. Homology searches and alignments were performed with the Basic Local Alignment Search Tool Network Service at the National Center for Biotechnology Information, National Institutes of Health (2).

Gene expression under control of the T7 promoter. E. coli JM109(DE3), which carries the gene encoding T7 RNA polymerase under control of the inducible lacUV5 promoter, was used for expression of plasmid-encoded genes transcribed under control of the T7 promoter. Following the addition of isopropylthiogalactopyranoside (IPTG) for induction of rifampin (300 µg/ml) to inactivate the host RNA polymerase, the cells were labeled with 10 µCi of [³⁵S]methionine, and whole-cell proteins were subjected to sodium dodecyl sulfate-12.5% polyacrylamide gel electrophoresis as previously described (11).

Gene fusions with phoA. To construct algK-phoA translational fusions, algK DNA was amplified by PCR with pMF100 as the template. One primer was specific to a region 240 bp upstream of the algK start codon that contained an NcoI site at the 5' end (5'-TGGCCCATGGCACCCGGGTGAACTTCCAGGT-3'). The other primer used corresponded to oligonucleotides encoding either AlgK residues 120 or 340 (5'-AGGCTCTAGAGCCTCGCGGTGCTCGGCGTCG-3' and 5'-CTGGTCTAGAGCCTTGAGCAGGTGCCGCTCG-3', respectively), and each had an XbaI site downstream from the coding region. The PCR products were cut with NcoI and XbaI and cloned into the NcoI-XbaI sites of the expression vector pMF54 (11) to give rise to pSJ42 and pSJ43, respectively. An XbaI-XhoI fragment from pPHO7 (18), encoding alkaline phosphatase without a signal sequence, was ligated into the XbaI-XhoI sites of pSJ42 and pSJ43 to give rise to pSJ46 and pSJ47, respectively, encoding AlgK-PhoA with fusion joints at residue 120 or 340. Protein fusions with PhoA were verified by Western blot analysis, which was performed as previously described (27). The primary antibody used was rabbit anti-bacterial alkaline phosphatase (5 Prime3 Prime, Inc), and the secondary antibody was a horseradish peroxidase-conjugated goat anti-rabbit immunoglobulin G. The detection system consisted of a hydrogen peroxide substrate-based colorimetric assay. Positive activity of the phoA fusions (PhoA⁺) in E. coli and P. aeruginosa was tested by screening colonies for blue color on L agar containing 5-bromo-4-chloro-3-indolylphosphate (X-P) at a concentration of 40 µg/ml.

Replacement of chromosomal algK with a nonpolar gentamicin cassette. Plasmid pUCGm (39) contains a cassette encoding a polar gentamicin resistance (Gm^r) gene between transcriptional stop sequences. This was used as a template to amplify by PCR a DNA fragment encoding a nonpolar Gm^r gene, along with its promoter sequence but without the transcriptional stop sequences. The upstream primer used for this matched sequences 115 bp upstream of the start codon: 5'-CGCGCCCGGGTTGACATAAGCCTGTTCGGTTCGTAA-3'. The other oligonucleotide primer was complementary to sequences 12 bp downstream of the stop codon: 5'-CGCGCCCGGGAAGCCGATCTCGGCT-3'. Both primers were designed to produce SmaI sites at both ends of the PCR product. The 0.7-kb SmaI-digested PCR product was then ligated into the SmaI site of pBluescript II KS() to obtain pSJ12, which conferred Gm^r. A 3.8-kb EcoRI fragment from pMF100 containing algK was cloned into the EcoRI site of the gene replacement vector pEX100T (40) to obtain pSJ10; this was digested with KpnI (to delete most of algK and a few nucleotides upstream), blunt ended, and ligated to the Gm^r SmaI fragment from pSJ12 to obtain pSJ15. Triparental matings were used to mobilize pSJ15 from E. coli HB101 to P. aeruginosa strains with selection for Gm^r. The plasmid had a narrow host range, and Gm^r colonies obtained were due to homologous recombination with the chromosome. Colonies that had undergone double crossovers, leading to gene replacement, were identified by plating on L agar containing 7.5% sucrose, indicating loss of the sacB (sucrose sensitivity) gene on the plasmid, and also by screening for loss of the plasmid-encoded bla (carbenicillin resistance) marker.

Uronic acid assay. P. aeruginosa cultures were grown in 10 ml of MAP defined medium for 24 h, and cells were removed by centrifugation (10,000 × g for 1.5 h). The supernatants were placed in dialysis bags (Spectra/Por membrane) (molecular weight cutoff, 10,000; 1.8 ml/cm) and dialyzed against an equal volume of 10 mM Tris-HCl (pH 7.6) at 4°C overnight. The uronic acid concentrations in the dialyzed and dialysate fractions were determined by the carbazole method of Knutson and Jeanes (20). Briefly, 30 µl of the fraction was mixed with 1.0 ml of borate-sulfuric acid reagent (100 mM H₃BO₃ in concentrated H₂SO₄), and 30 µl of carbazole reagent (0.1% in ethanol) was added. The mixture was heated to 55°C for 30 min, and the absorbance at 530 nm was determined. Uronic acid concentrations were determined from a plot with Macrocystis pyrifera alginate (Sigma) as a standard.

Nucleotide sequence accession number. The nucleotide sequence data for the fragment containing algK from strain FRD1, a sputum isolate from a cystic fibrosis patient (28), have been deposited in the GenBank database under accession no. AF039535.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Nucleotide sequence of algK from P. aeruginosa FRD1 and its expression in E. coli. We determined the sequence of the DNA in a region of the alginate biosynthetic operon between alg44 and algE by using DNA derived from the alginate-overproducing (Alg⁺) cystic fibrosis isolate FRD1. Upon aligning the sequences at each end with those already known, an open reading frame (algK) of 1,428 bp (accession no. AF039535) was observed. This open reading frame predicted the synthesis of a 52-kDa polypeptide with an isoelectric point of 5.8 and a charge of 4.3. The stop codon of algK overlapped with the start codon of the downstream gene, algE, suggesting that these two genes may be translationally coupled. The AlgK sequence from P. aeruginosa 8873 was recently reported (1), and it differed at Arg-78 (AGG) in FRD1, which was a Ser (AGT) in strain 8873, and Phe-177 (TTC) in FRD1, which was a Leu (CTC) in 8873. Homology searches did not reveal any obvious similarity to a protein of known function that might suggest a function for AlgK.

View larger version (28K): [in this window] [in a new window]   FIG. 2.   Expression of algK in E. coli. (A) Partial restriction map of P. aeruginosa FRD1 DNA in pMF100, showing a 3.8-kb EcoRI fragment containing algK and alg44 that was used in this study for expression in E. coli under control of the T7 promoter (PT7). (B) Autoradiogram of [35S]methionine-labeled proteins expressed in E. coli JM109(DE3) from the vector pBluescript II KS() (lane 1) or the clone pMF100 (lane 2) in the presence of IPTG and rifampin. Sizes of molecular weight markers (lane 3), in thousands, are shown. The positions of the AlgK and Alg44 proteins expressed from pMF100 are indicated with arrows.

Periplasmic localization of AlgK. The hydrophilicity plot of the AlgK sequence predicted a generally hydrophilic protein, except for its amino terminus (approximately 25 residues), which displayed considerable hydrophobic character (Fig. 3A). This was followed by two potential signal peptidase cleavage (Ala-X-Ala) sites (35), suggesting that AlgK may possess a signal sequence for transport across the inner membrane. To determine if AlgK localized to the periplasm, algK DNA was manipulated to encode AlgK-PhoA fusion proteins containing the first 120 or 340 residues of AlgK (Fig. 3B). The 5' algK sequences were ligated upstream and in frame to a phoA gene lacking its own signal sequence. Expression of the resulting algK-phoA fusions in the broad-host-range expression vector pMF54 was examined in E. coli HB101. AlgK-PhoA fusion proteins of the appropriate sizes were observed in a Western blot analysis utilizing alkaline phosphatase antibodies (data not shown). When grown on X-P indicator plates, the E. coli strains producing both AlgK-PhoA fusions were blue (PhoA⁺). This indicated that the AlgK-PhoA proteins had localized to the periplasm, utilizing the AlgK signal sequence, followed by cleavage of the phosphate group on the chromogenic X-P substrate. When these plasmids were mobilized to P. aeruginosa PAO1, the colonies also turned blue on the X-P agar medium. Colonies of HB101 and PAO1 containing the vector were not blue. These studies indicated that AlgK normally resides in the periplasm. In addition, a cysteine residue was observed immediately after the presumptive signal sequence in AlgK (Cys-28), which suggests that the protein may undergo lipid modification (31); the other two cysteines in AlgK (Cys-195 and Cys-205) probably form a disulfide bond.

View larger version (24K): [in this window] [in a new window]   FIG. 3.   Localization of AlgK to the periplasm. (A) A hydrophilicity plot of the predicted AlgK polypeptide shows a generally hydrophilic protein except for the amino terminus, which represents a putative signal sequence. (B) Aligned to the hydrophilicity plot are shown DNA fragments encoding 5' portions of algK (solid lines) that were used to construct algK-phoA translational fusions. These algK-phoA fusions are under control of the Ptrc promoter (open arrow) in a broad-host-range expression vector (pMF54), and both showed alkaline phosphatase activity (PhoA+) in E. coli and P. aeruginosa as shown by blue colony formation on an X-P-containing medium.

Construction of a nonpolar algK deletion mutant. A defined algK deletion (algK) mutant was constructed in the mucoid FRD1 strain background to evaluate the potential role of this gene in alginate production. Our strategy was to replace algK in the chromosome with a marker that would not affect expression of any downstream genes in the operon that are essential for alginate precursor formation (e.g., algA). This was accomplished by first cloning the 3.8-kb EcoRI fragment (Fig. 2A) containing algK and adjacent sequences into pEX100T, a gene replacement vector that contains sacB, which confers sucrose sensitivity in P. aeruginosa (38). A nonpolar Gm^r cartridge was synthesized by PCR and used to replace a KpnI fragment in the clone to delete most of algK but none of the adjacent alg44 or algE sequences, forming pSJ15. When this EcoRI fragment containing alg44 algK::Gm^r was placed under control of the T7 promoter (pSJ137), AlgK expression was lost but Alg44 expression was not affected (data not shown). The algK::Gm^r construct in pEX100T (pSJ15) was introduced into mucoid FRD1 by triparental mating with selection for the vector-encoded bla (-lactam resistance) gene by plating on L agar with carbenicillin (Fig. 4A). Due to the narrow host range of the plasmid, the numerous colonies that formed were merodiploids resulting from homologous recombination with the chromosome by a single crossover. Approximately three-fourths of the colonies obtained were nonmucoid, and one-fourth were mucoid. This was as predicted and depended on whether the crossover had occurred in the 2.0 kb upstream of algK::Gm^r or in the downstream 0.5-kb region (Fig. 4A). The former would result in the integrated vector having a polar effect on the downstream biosynthetic genes, whereas the latter would allow the promoter of the Gm^r gene to drive transcription of downstream genes in the algD operon. The presence of colonies with the mucoid phenotype also verified that the Gm^r cassette developed here was nonpolar on downstream gene transcription. To remove the vector sequences and obtain the algK::Gm^r mutants (i.e., gene replacement), mucoid merodiploid colonies were grown in L broth (without antibiotic selection) to permit spontaneous recombination, followed by selection for gentamicin and sucrose resistances. The colonies obtained also demonstrated carbenicillin sensitivity, as expected, due to loss of the vector. All the algK::Gm^r mutants constructed in the mucoid FRD1 background were nonmucoid, indicating that AlgK was essential for alginate production.

View larger version (31K): [in this window] [in a new window]   FIG. 4.   Construction of algK mutants of P. aeruginosa. (A) Plasmid pSJ15 contains a 3.8-kb EcoRI (R) fragment of the algD operon in FRD1 cloned into a gene replacement vector (pEX100T), with a nonpolar Gmr cartridge (hatched arrow) replacing algK. Shown are the predicted genotypes of merodiploids that formed when pSJ15 integrated into the chromosome of mucoid FRD1 to form mucoid (Alg+) and nonmucoid (Alg) colonies. When a crossover occurs at site 1, the integrated vector has a polar effect on the transcription of downstream genes (open arrows), and the strain should be Alg. When a crossover occurs at site 2, the promoter of the Gmr gene should drive transcription of downstream genes in the algD operon so that all genes in the operon are expressed (black arrows), and this strain should be Alg+. (B) Diagrams of pSJ44, which expresses algK in trans, and the algD operons in the mutants used in this study along with their colony morphologies are shown. Note that in a strain overproducing alginate (FRD1), a nonpolar algK::Gmr mutation resulted in a small-colony (i.e., sick) phenotype (strain FRD1100) and that this phenotype could be suppressed (strain FRD1105) with an algJ::Tn501-3 mutation (shown as a pin) which was polar on algA. Open arrows indicate genes that are not expressed. The genes of the algD operon are not drawn to scale.

algK mutants display a small-colony phenotype that is suppressed by a mutation polar on algA. In addition to the loss of alginate production, another interesting phenotype of the algK mutants was their small-colony phenotype. One of the algK mutants, called FRD1100, which displayed the typical small-colony phenotype, was chosen for further study; its chromosomal replacement of algK by the Gm^r cartridge was verified by PCR analysis (see Materials and Methods). We also tested the effect of algK in trans in FRD1100. For this, the P. aeruginosa expression vector pMF54 (11) was used to clone the 3.8-kb EcoRI fragment that contained algK in the proper orientation relative to the plasmid's trc promoter, which formed pSJ44 (Fig. 4B). A similar plasmid (pSJ151) that contained the alg44 and algK::Gm^r alleles was constructed as a control. When pSJ44 and pSJ151 were conjugally transferred to FRD1100 (as well as to the other algK mutants) with selection for carbenicillin resistance, only the transconjugants with pSJ44 (i.e., algK⁺) exhibited both the wild-type mucoid (Alg⁺) and normal growth phenotypes. This also indicated that the chromosomal algK::Gm^r mutation was nonpolar.

algK algJ

View larger version (63K): [in this window] [in a new window]   FIG. 5.   Photographs of L-agar plates containing colonies of P. aeruginosa strains that formed after 24 h at 37°C. Note that in an Alg+ strain (FRD1) background, a nonpolar algK::Gmr mutation (strain FRD1100) resulted in Alg and small-colony (i.e., sick) phenotypes. These phenotypes could be suppressed with a polar algJ::Tn501-3 mutation (strain FRD1105). (A) algK mutant FRD1100. (B) algK algJ::Tn501-3 mutant FRD1105.

The algK mutant secretes unpolymerized uronic acids. The studies described above suggested that the small-colony (i.e., slow-growth) phenotype of algK mutants was due to the production of toxic alginate precursors. Given that AlgK localized to the periplasm, we hypothesized that a mannuronate precursor might still be transported to the periplasm and, if unpolymerized, may permeate through the outer membrane and into the medium. Initial tests showed that culture supernatants of strains grown in L broth (22 h, 37°C, with aeration) were lower in pH with FRD1100 than with FRD1105, and this was consistent with our hypothesis (data not shown). To test culture supernatants for unpolymerized uronic acids and to avoid the background interference that arises with L-broth cultures, the P. aeruginosa strains were grown in the defined medium MAP (11), which promotes alginate production. Uronic acids in the supernatants of Alg⁺ FRD1 cultures (grown for 22 h at 37°C with aeration) were compared to those of the algK mutant (FRD1100); the algK algJ::Tn501 double mutant (FRD1105) served as the background negative control. At first, each sample was dialyzed against just an equal volume (7 ml) of buffer, and then the uronic acid contents of both the dialyzed fraction and the dialysate were assayed (Table 2). As expected, the Alg⁺ FRD1 sample had uronic acids almost exclusively in the dialyzed fraction, because the uronic acids (i.e., mannuronate and guluronate) were in the polymeric (i.e., alginate) form, which is too large to pass through the dialysis membrane. This was also true of the Alg⁺ strain FRD1100(pSJ44) (data not shown). In contrast, the algK mutant FRD1100 showed approximately equal amounts of uronic acids in both fractions. The amount of uronic acid produced by the algK mutant, determined by adding the values observed in the dialyzed and dialysate fractions, was similar to that of the polymerized product of parent strain FRD1. When each fraction was then exhaustively dialyzed, all of the FRD1 uronic acids remained in the dialysis sac (Table 2). However, none of the FRD1100 uronic acids was detectable following this, because they had all dialyzed away. This suggests that algK mutants produced large amounts of a low-molecular-weight precursor of alginate that may have been properly localized but not polymerized. This high concentration of unpolymerized uronic acids in the environment of FRD1100 was apparently what was detrimental to its growth and led to the small-colony morphology observed.

View this table: [in this window] [in a new window]   TABLE 2.   Amounts and relative sizes of uronic acid-containing material secreted by Alg+ P. aeruginosa FRD1 and its Alg algK derivative FRD1100

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Alginate biosynthesis is comprehensively regulated by a number of factors that ultimately converge on the promoter of the algD biosynthetic operon to limit the production of enzymes for alginate biosynthesis (for a review, see reference 29). However, mucoid strains from cystic fibrosis patients appear to undergo deregulation, probably due to adaptive mutations occurring in vivo, which result in the copious overproduction of the viscous polymer. This conversion to the Alg⁺ phenotype is regarded as an important pathogenic mechanism. Alginate biosynthesis in P. aeruginosa also provides an attractive model system to better understand the secretion of exopolysaccharides in gram-negative bacteria. The steps in the pathway leading to the formation of an activated monomeric unit (GDP-mannuronate) have been well characterized (26). This involves the participation of two genes, algD and algA, which are the first and last genes of the large algD operon. Recently, the late steps involved in postpolymerization modifications, such as epimerization and acetylation, are beginning to be understood. These processes require the gene products of algG, algI, algJ, and algF in the operon (5, 11-13). However, the middle process in the pathway of alginate biosynthesis, in which the inner membrane is crossed by the next intermediate (monomer or oligomer) and the polymerization of these moieties to poly-mannuronic acid occurs, is still unknown.

A complete understanding of the alginate biosynthetic pathway was not possible until all of the components involved had been identified. There was one segment (~1.2 kb) of the algD operon, located between alg44 and algE, that was still uncharacterized. Our sequence analysis of this region from FRD1 revealed a potential open reading frame (1,428 kb), and we showed here that it encoded the predicted 52-kDa protein. During the final stages of this work, Aarons et al. (1) also sequenced this region from strain 8873 and called it algK; we have adopted their nomenclature. The amino terminus of the predicted protein resembled a signal sequence recognized by type II signal peptidase, suggesting that the algK gene may encode a lipoprotein (31). The hydrophilicity plot predicted a hydrophilic protein, suggesting its localization to the periplasm. We confirmed this by constructing algK-phoA protein fusions, which were positive for alkaline phosphatase activity and strongly suggested that the protein is indeed periplasmic. These studies complement those of Aarons et al. (1), who also predicted such localization by using -lactamase (bla) fusions.

A better understanding of AlgK function was obtained through the construction and analysis of an algK deletion mutant in the Alg⁺ FRD1 background. The mucoid or nonmucoid phenotype of such a mutant would show whether AlgK played a role either in polymer production or in modification of the polymer. The genetic manipulations involved in this construction were simplest if algK in the chromosome was replaced with a selectable marker. This laboratory has previously used Tn501 insertions to study the algD operon, but their polar nature requires that the downstream genes, which are essential for polymer formation, be provided in trans (12). Thus, it was more desirable here to have a selectable marker that would be nonpolar on downstream gene expression so that even the last gene (algA) would still be expressed, as this is required for the formation of all alginate precursors. Using PCR amplification, we constructed a Gm^r cassette that contained no transcriptional stop sequences, and thus it was predicted to be nonpolar. By replacing algK with this nonpolar Gm^r cassette (which still retained its promoter sequence), we obtained algK mutants. These demonstrated a nonmucoid phenotype, indicating that AlgK was required for alginate production rather than its modification. By providing algK in trans under control of a trc promoter, alginate production was restored, which indicated that the algK::Gm^r mutation was nonpolar on all downstream genes.

An interesting and useful phenotype of the algK mutants was their small-colony morphology, suggesting that they were sick as a result of the mutation. We hypothesized that such algK mutants produced an alginate precursor that was potentially toxic to the cells. We constructed an algK mutation in an AlgA (i.e., algJ::Tn501) strain background, and these double mutants grew as well as the parent strain, which was in contrast to the small colony size of the algK mutant. This implied that AlgK acted at a step subsequent to precursor formation. If there was no precursor available in the double mutant, then loss of algK did not have an effect on cell growth. Bringing algA in trans should then give the double mutant a growth defect. However, we were unsuccessful in attempts to transfer algA on a high-copy-number plasmid to such strains, suggesting that an algK mutation was deleterious when alginate precursors were produced. If algK was involved in the assembly and/or transport of alginate, then a deletion mutation of the gene could potentially result in accumulation of toxic levels of either GDP-mannose or a precursor polymer within the cells. In the case of K1 capsule formation in E. coli, mutations that block transport across the inner membrane lead to the accumulation of polymer within the cell, and this can be observed by electron microscopy (4, 30). However, our electron micrographs of FRD1 and the algK mutants constructed here showed no evidence of polymer accumulation in the cells (data not shown). Efforts to determine whether the algK mutation causes the accumulation of GDP-mannuronate within the cytoplasm are in progress.

However, it seemed reasonable that if AlgK was a periplasmic component involved in alginate production, then it may be a part of a polymer assembly apparatus in the periplasm. If AlgK did not affect precursor formation or transport across the membrane, then this might lead to accumulation of a monomeric or oligomeric precursor in the periplasm, which could diffuse through the outer membrane. The first suggestion that the algK mutation affects polymerization was the observation that the AlgK mutant supernatants were more acidic than those of an AlgA mutant, suggesting that uronic acids were being released by the cells. Indeed, the AlgK mutant was shown to produce large quantities of unpolymerized uronic acids. This material was readily dialyzable, indicating that it was of low molecular weight. This uronic acid material is currently being purified for further analysis. However, we would predict that it is some form of D-mannuronic acid, either monomer, short oligomer, or sugar nucleotide. This analysis should provide further insight toward our understanding of the immediate precursor of poly-M in the pathway of alginate biosynthesis and the process involved in polymer formation.