Export of Extracellular Polysaccharides Modulates Adherence of the Cyanobacterium Synechocystis

The field of cyanobacterial biofuel production is advancing rapidly, yet we know little of the basic biology of these organisms outside of their photosynthetic pathways. We aimed to gain a greater understanding of how the cyanobacterium Synechocystis PCC 6803 (Synechocystis, hereafter) modulates its cell surface. Such understanding will allow for the creation of mutants that autoflocculate in a regulated way, thus avoiding energy intensive centrifugation in the creation of biofuels. We constructed mutant strains lacking genes predicted to function in carbohydrate transport or synthesis. Strains with gene deletions of slr0977 (predicted to encode a permease component of an ABC transporter), slr0982 (predicted to encode an ATP binding component of an ABC transporter) and slr1610 (predicted to encode a methyltransferase) demonstrated flocculent phenotypes and increased adherence to glass. Upon bioinformatic inspection, the gene products of slr0977, slr0982, and slr1610 appear to function in O-antigen (OAg) transport and synthesis. However, the analysis provided here demonstrated no differences between OAg purified from wild-type and mutants. However, exopolysaccharides (EPS) purified from mutants were altered in composition when compared to wild-type. Our data suggest that there are multiple means to modulate the cell surface of Synechocystis by disrupting different combinations of ABC transporters and/or glycosyl transferases. Further understanding of these mechanisms may allow for the development of industrially and ecologically useful strains of cyanobacteria. Additionally, these data imply that many cyanobacterial gene products may possess as-yet undiscovered functions, and are meritorious of further study.


Introduction
The need for alternative and sustainable fuels has become widely accepted, and biofuel production in microalgae and cyanobacteria has been developed to that end. In our laboratory we have engineered the cyanobacteria, Synechocystis to produce high levels of fatty acid and to automatically process their own biomass in an inducible manner [1][2][3][4]. Unlike traditional sources of biofuels, the use of photosynthetic microorganisms does not require the use of cultivatable land or potable water. However, the large culture volumes required to produce biofuels at commercial scale are considerable. Traditional sources of biomass dewatering rely on potentially toxic flocculants, filtration or centrifugation, making biomass dewatering an energy intensive process [5]. In an attempt to diminish the energy requirements for biomass dewatering, cyanobacteria have been constructed with modulated adherence properties with potential for use in biomass harvest and bioremediation [6][7][8].
Both extracellular polysaccharides (EPS) and O-antigen (OAg) are carbohydrate moieties on the cell surface that mediate bacterial protection and provide a mechanism for bacterial cells to interact with their environments [9][10][11]. Between species, OAg is a highly variable, surface-exposed component of lipopolysac-charide in the outer membrane of Gram-negative bacteria. This key structure modulates susceptibility to phage [12,13], surface charge [14] and permeability of the outer membrane [14]. Similarly, EPS is a large, polymeric carbohydrate structure that serves to protect cells from environmental stress such as UV irradiation [15], heavy-metal toxicity [8,16,17], osmotic stress [8,18] and desiccation [19].
In this study, we completed a whole genome comparison of in silico translated genes from Synechocystis, E. coli K-12 and S. Typhimurium LT2. We searched for genes in Synechocystis that showed homology to genes involved in modulating the cell surface moieties in other species. Two of the genes we identified were slr0977 and slr0982. These genes encode products that are homologous to the Wzm and Wzt proteins in E. coli. Together, these gene products comprise the permease and oligosaccharide binding proteins that function together as a ABC-transporter [20,21].
We noted that slr0977 and slr0982 were located in a cluster of contiguous genes that were annotated as OAg transport genes in the Cyanobase genome database [22,23]. Subsequent mutation of these genes resulted in flocculating strains with modulated adherence properties. Surprisingly, neither the OAg structure nor its composition was altered in these mutants. Rather, we provide evidence that the phenotypes exhibited by these mutants are due to altered extracellular polysaccharide. Together with previously published studies, these data indicate multiple mechanisms for autoflocculation through disruption of EPS biosynthesis and/or export. This understanding is important for the construction of autoflocculating mutant strains.

Bioinformatics Searches and Analyses
The complete genomes for E. coli K-12 (NC_000913.2), S. Typhimurium LT2 (NP_459707.1), and Synechocystis 6803 (BA000022.2) were downloaded from NCBI GenBank and gene information was extracted using Perl scripts. The COG database [24] was also used in categorizing genes. All of the features were stored in a MySQL database. For gene identification, we used BLASTP implemented in NCBI blast-2.2.18 to identify possible homologous genes of E. coli or Salmonella in the Synechocystis genome with a threshold e-value less than 1.0 -4 . Three methods were used to define genes of synthesis and assembly of cell wall components: (1) Based on a set of genes in E. coli K-12 and S. Typhimurium LT2, the homologous genes in Synechocystis using were determined.
(2) a set of genes based on the functional descriptions of the COG database were determined, and then the homologs in Synechocystis were identified using BLASTP (3) We searched for genes directly based on the functional annotation of Synechocystis. Thus, 519 gene products potentially involved in the synthesis and assembly of the Synechocystis cell wall and cell-surface macromolecular components were identified.

Bacterial Growth Conditions
Cultures were inoculated to a starting OD 730 of 0.1 in BG-11 medium [25] from starter cultures. Cultures were incubated at 30uC with 40 mmol photons m 21 s 21 . Kanamycin was added at a concentration of 50 mg/mL when appropriate. To assess the effect of NaCl on EPS mutants, Synechocystis cultures were inoculated in triplicate to an OD 730 2 mL of BG-11 or BG-11+0.8M NaCl in test tubes. Every 24 h cultures were gently agitated with a vortex until adherent cells were fully in suspension. The OD 730 was then measured and recorded.

Vector Construction
DNA manipulation was carried out using standard procedures [26]. Suicide vectors were constructed in a two-step process. For each gene deletion, PCR primers (Table 1) were used to amplify ,400bp of genomic DNA from the flanking region of of each gene. Flanking regions were stitched together by a modified overlap extension PCR as previously described [27]. Briefly, BamHI and NdeI restriction sites were generated between the two flanking sequences. Thus, we created the plasmids py508 (slr0977 flanking regions), py512 (slr1610 flanking regions), y514 (sll0574-5 flanking regions) and py560 (slr0982 flanking regions). The Kan R -SacB cassette from pPsbA2ks [28] was purified following digestion with BamHI and NdeI and ligated into py508, py512, y514 and py560 to generate py509 (Dslr0977) py513 (Dslr1610), py515 (D(slr0977;Dsll0574-5)) and py561 (Dslr0982) respectively. In the case of py560, an NdeI site native to the 39 flanking region was removed by introducing a silent mutation using QuikChange II Site-Directed Mutagenesis Kit (Agilent Technologies) according to the manufacture's protocol, using primers MLF-68 and MLF-69.
To rule out second-site mutations that might have accumulated during the construction of mutant strains, we constructed vectors for genomic complementation. Complementation was carried out by replacing the Kan R -SacB in each mutant with the wild-type allele. Complementation vectors py618, py619 and py620 were constructed to complement the mutations Dslr0977, Dslr0982, Dslr1610 in SD506, SD553 and SD507, respectively. To construct these vectors, DNA from SD100 was amplified with primer pairs MLF-1/MLF-4, MLF-5/MLF-8 and MLF-9/MLF-12. The PCR products were ligated into pJET1.2 to generate py618, py619 and py620, respectively. SD553 was transformed with py619 to generate SD563. SD507 was transformed with py620 to generate SD564. Several attempts were made to complement the Dslr0977 strain by transformation of SD506 with py618. However, no transformants were recovered.

Transformation and Complementation
Mutants of Synechocystis were generated as previously described [1]. Briefly, 4 mg of suicide vector was added to 100 mL of a late log phase culture of and incubated overnight in BG-11 without antibiotic. The entire transformation mixture was then spread onto a Nuclepore Track-etch membrane (Whatman #111707) on a BG-11 plate. These were then incubated for three days at 30uC with 40 mmol photons m 21 s 21 . Following this incubation, membranes were transferred to a BG-11 plate containing 100 mg/mL kanamycin and incubated until single colonies appeared. To ensure complete segregation of the mutants, colonies were patched onto BG-11+6% sucrose and BG-11+100 mg/mL kanamycin. Kanamycin resistant colonies that were sucrose sensitive were further verified by PCR to ensure complete segregation.
To complement mutants in SD553 and SD507, 100 mL late log cultures of each strain were transformed with 4 mg of either pY619 or pY620, respectively to generate SD563 and SD564. Following an overnight incubation at 30uC with 40 mmol photons m 21 s 21 , transformation mixtures were resuspended in 1mL of BG-11 and allowed to recover for three days. Following the recover step, 100 mL of culture was then plated on BG-11+6% sucrose and incubated until single colonies arose. Isolated colonies were then patched onto BG-11+6% sucrose and BG-11+100 mg/mL kanamycin. Colonies that were sucrose resistant and kanamycin sensitive were verified by PCR and phenotypic assays (see crystal violet adherence assay below) for complete segregation.

Crystal Violet Adherence Assay
Synechocystis cultures were inoculated at an OD 730 of 0.1 in 2 mL of BG-11 medium in test tubes. Kanamycin was added at a concentration of 50 mg/mL when appropriate. In each experimental replicate, static cultures were grown in triplicate for 96 hours at 30uC with 40 mmol photons m 21 s 21 . The adherence of three cultures was then measured as follows: The culture medium containing non-adherent cells was decanted from each of the tubes. Each tube was gently washed 26 with 2 mL of BG-11. Tubes were then stained with 1% crystal violet in dd H 2 O for 15 min. The tubes were then gently rinsed 36 with dd H 2 O. Adherent cells were resuspended in 1 mL of DMSO and vortexed vigorously. The OD 630 of each tube was then measured. Each strain was analyzed in triplicate. Representative data of three separate experiments are presented.

LPS Extraction and Analysis
LPS extraction was carried out as described [29]. Briefly, 25 mL cultures of Synechocystis were inoculated to an initial cell density of 0.1 OD 730 in 50 mL of BG-11 in 125 mL Erlenmeyer flasks for 4 days at 30uC with 40 mmol photons m 21 s 21 , shaking at 115 rpm. Cultures were normalized for cell density and centrifuged at 6000 rpm in a Sorvall SS-34 rotor for 20 min. Bacterial pellets were resuspended in 5 mL TRI reagent (Sigma) in 15 mL conical tubes, vortexed and incubated at room temperature overnight with gentle shaking. Then, 1 mL of chloroform was added to each tube which was then vortexed well and incubated at room temperature for 15 min. Samples were then centrifuged at 11,000 rpm in a Sorvall SS-34 rotor for 10 min. The aqueous phase was transferred to a fresh tube and the organic phase was backextracted with 2 mL of dd H 2 O. Samples were frozen and lyophilized overnight. Samples were then resuspended in cold 0.375 M NaCl in 95% ethanol, transferred to 1.5 mL centrifuge tubes and washed 26 in 0.375 M NaCl in 95% ethanol, resuspended in 1mL of cold 100% ethanol and lyophilized overnight. Samples were then washed three times in cold Folch reagent (2:1 CHCl 3 :MeOH), lyophilized and resuspended in 100 mL dd H 2 O. LPS was analyzed by SDS-PAGE on a 4-15% gradient gel (Biorad) and then visualized by silver staining [30].

EPS Extraction and Analysis
EPS was extracted by mechanical disruption as described [31]. Cells were grown to early stationary phase (OD 730 ,1.0) and normalized for equal OD 730 . Fifty milliliters of cells were centrifuged for 20 minat 10,0006g (Sorvall SS-34). Samples were resuspended in 10 mL BG-11 and fixed with 60 mL of formaldehyde (Sigma) overnight at 4uC. 4 mL of 1 M NaOH was added to each sample, which was then vigorously agitated with a vortex for 15 sec. Samples were centrifuged for 20 min at 20,0006g (Sorvall SS-34). The supernatants were frozen, lyophilized and resuspended in 2 mL of water, each. At this point, samples were dialyzed three times against 4 L of water, lyophilized overnight and resuspended in 100 mL of water. 10 mL of each sample was loaded onto a 12% Mini-PROTEANH TGX TM Precast Gel (BioRad #456-1045) and stained with Alcian Blue. Briefly, gels were fixed with 12.5% trichloroacetic acid for 30 min and rinsed with water. Following incubation in a solution containing 1% periodate and 3% acetic acid, gels were washed five times in water. Gels were then incubated in 0.5% potassium disulfite for thirty min and washed three times in water. At this point, gels were stained with a solution of 0.5% Alcian blue in 3% acetic acid overnight. Prior to visualization, gels were destained with 7% acetic for 5 min or until EPS bands were visible.

Outer Membrane Protein Analysis
Outer membrane proteins were extracted as described [32]. Cultures were grown under static conditions as described above. Fifty milliliters of culture at OD 730 0.4 were centrifuged and the outer membrane faction was purified. Membrane fractions were resuspended in 20 mL of Laemmli buffer with 5.5% b-mercaptoethanol (Sigma). Each sample was loaded in its entirety onto a 12% SDS-PAGE gel and stained with Coomassie brilliant blue.

Glycosyl Composition Analysis of OAg and EPS
Glycosyl composition analysis was performed by combined gas chromatography/mass spectrometry (GC/MS) of the per-Otrimethylsilyl (TMS) derivatives of the monosaccharide methyl glycosides produced from the sample by acidic methanolysis. Between 200 and 300 mg of each sample was used for the analysis. The samples were placed into test tubes and 20 mg of inositol was added. Methyl glycosides were then prepared from the dry samples by methanolysis in 1 M HCl in methanol at 80uC (17 hours), followed by re-N-acetylation with pyridine and acetic anhydride in methanol (for detection of amino sugars). The samples were then per-O-trimethylsilylated by treatment with Tri-Sil (Pierce) at 80uC (0.5 hours). These procedures were carried out as previously described [33,34]. GC/MS analysis of the TMS methyl glycosides was performed on an Agilent 7890A GC interfaced to a 5975C MSD, using a Agilent DB-1 fused silica capillary column (30 mm 6 0.25 mm ID).

Results and Discussion
We wished to ascertain what genes in Synechocystis were likely involved in the biosynthesis of structural components of cell surface structures. To this end, whole genome comparisons of Synechocystis were carried out against E. coli K-12 and S. enterica serovar Typhimurium LT2. We identified over 500 putative genes potentially involved in the synthesis and assembly of the Synechocystis 6803 cell wall and cell-surface macromolecular components. Additionally, we identified gene products involved in LPS biosynthesis, polysaccharide biosynthesis, extracellular appendages, cell wall biosynthesis, metabolite biosynthesis and peptidoglycan biosynthesis (Tables S1).
The nomenclature for OAg biosynthesis has changed over the years [35]. Whereas the OAg ABC transporter components were once designated rfbA (permease component) and rfbB (ATP binding component), some of these are now given designations wzm (permease component) and wzt (ATP binding component) [35]. Our analysis revealed that Synechocystis contains two systems similar to the wzm/wzt genes that encode OAg transporters in E. coli. Genes encoding products annotated as involved in OAg synthesis and assembly, are present in the same gene cluster. Specifically, slr0983, slr0984 and slr0985 were also assigned the ''rfb'' designations rfbF, rfbG and rfbC, respectively. Interestingly, a similar gene arrangement (rfbFGC) was bioinformatically identified in Azotobacter vineladnii [36]. Following slr0985 in the Synechocystis gene cluster described here is slr1610, which is annotated as encoding a methyltransferase ( Table 2).
The genes slr0977 and sll0574 encode homologs of the integral membrane protein Wxm. The slr0977 and sll0574 gene products possessed 24% and 29% identity to Wzm in E. coli, respectively (Table 1). Similarly, slr0982 and sll0575 are annotated as rfbB [22]. These gene products are 34% and 41% identical to the ABC protein, Wzt in E. coli. Wzt possess a C-terminal nucleotide binding cassette and an N-terminal carbohydrate binding domain [37]. While the C-terminal, nucleotide binding regions of the slr0982 and sll0575 gene products are highly homologous, there is little homology at the N-terminal region of these predicted gene products. This lack of N-terminal homology is likely indicative of differential substrate specificity for the type of polysaccharides transported by these systems.

Disruption ABC Transporters Leads to Flocculent Cultures with Modulated Adherence Properties
In Synechocystis, genes encoding products that function in the same biosynthetic pathway are not consistently organized in operons [22]. It is interesting therefore, that the genes slr0977 and slr0982 are part of a nine-gene cluster in Synechocystis (Fig. 1A). These two ABC transporter components are separated by slr0978, slr0980, and slr0981, which are hypothetical genes with no known function. These genes are followed by slr0983 (glucose-1-phosphate cytidylyltransferase), slr0984 (CDP-glucose 4,6-dehydratase), slr0985 (dTDP-4-dehydrorhamnose 3,5-epimerase) and slr1610 (putative C-3 methyltransferase). We also observed that the genes sll0574 and sll0575 located distantly on the complementary strand of the genome (Fig. 1B) encoded for homologs of Wzm and Wzt respectively.
To investigate the role of slr0977 and slr0982 in cell surface modification, we generated mutants by deleting these genes in the SD100 background [1]. Mutant strains lacking different combinations of these genes were created through allelic replacement with a kanamycin-resistance cassette and the counter-selectable sacB gene (Table 3) [38]. The growth rates of Dslr0977, Dslr0982, Dslr1610, or the triple mutant D(slr0977;Dsll0574-5) were identical to wild type. However, mutants demonstrated readily apparent flocculent phenotypes and adherence to glass culture vessels (Fig. 2).
We quantified the adherence of each mutant to the growth vessel by the crystal violet binding assay (Fig. 2). All mutant strains tested were significantly better able to adhere to glass culture vessels. Following four days of static growth in a two milliliter culture, the mutant strains Dslr0977, Dslr0982, and Dslr1610 all demonstrated a 4-6 fold increase in binding (Fig. 2). The adherence of these strains was specific to glass vessels. Mutant strains grown in polystyrene flat-bottom 96 well plates (Costar or Nunc MaxiSorpH) did not demonstrate increased adherence relative to wild-type (data not shown). The ability of these mutants to flocculate and not adhere to plastic may have industrial applications. This is particularly true of production in closed photobioreactors (PBR), which are often constructed from plastics. Such flocculent strains would decrease the energy requirements for centrifugation or filtration during biofuel production. Concomitantly, such phenotypes may eliminate the risk of biofouling by cyanobacterial growth on the PBR surface. It is intuitive that Dslr0977 and Dslr0982 would demonstrate similar phenotypes given that they are both lacking genes encoding products in the same transport system. The failure to transport the same substrate likely accounts for this phenotype. It is interesting that the Dslr1610 mutant demonstrated a nearly identical pheno-type to both the Dslr0977 and Dslr0982 strains. The slr1610 gene encodes a putative methyltransferase. In OAg production, a methyltransferase is responsible for terminating carbohydrate polymerization [22]. Additionally, methylation of the carbohydrate chain is important for recognition by the ABC transporter for subsequent transport [39]. Given this, one parsimonious model is that in the absence of the slr1610 gene product, larger carbohydrate moieties are synthesized by the mutant than in wild-type. It is likely that such long carbohydrate chains would be unable to properly interact with the slr0977/slr0982 gene products for extracellular transport. It is unlikely that the phenotypes of these mutants are due to polar effects created while constructing the slr0977 or slr0982 deletions. This is evidenced by the fact that a similar mutation D(slr0978-81) (located between the slr0977 and slr0982 ORFs) was indistinguishable from wild-type in adherence, flocculence and growth (data not shown).

Mutant Adherence Occurs as a Monolayer
To further examine the adherence of cell to glass, wild-type and mutants were grown on sterile coverslips for four days in static culture and examined by fluorescence microscopy (Fig. 3). Consistent with crystal violet staining, wild-type cells adhered poorly to glass (Fig. 3A), whereas Dslr0977, Dslr0982, Dslr1610 and D(slr0977;Dsll0574-5) readily adhered to coverslips (Fig. 3B-E,  respectively). Furthermore, the adherence phenotypes of the Dslr0982 or Dslr1610 mutants is unlikely due to compensatory mutations as strains complemented with a wild-type allele were indistinguishable from wild-type ( Fig. 2 and Fig. 3E-F).
We did not observe complex, three-dimensional architecture consistent with biofilm formation [40]. Rather, we saw monolayers of cells strongly adhered to glass (Fig. 3A). These data are consistent with the idea that disruption of these gene products leads to a change in the charge of the cell surface. In fact, EPS biosynthesis mutants lacking sll1581 (gumB) and sll5052 (gumC) aggregated out of solution and had altered cell surface charges [8].
OAg Structure and Composition is Unaffected by Mutations in Dslr0977, Dslr0982, Dslr1610 and D(slr0977;Dsll0574-5) To determine if OAg was responsible for the modulated adherence and flocculation of the mutants generated, total LPS was extracted from mid-log cultures of wild-type or mutant strains and analyzed by SDS-PAGE (Fig. 4). A previously characterized fucose synthase mutant (Dsll1213) [41] was also processed as a positive control for disrupted OAg structures. As expected, the fucose synthase mutant demonstrated marked differences in size  and intensities of purified LPS species from the wild-type strain. However, LPS from the mutant strains demonstrate very little difference from wild-type (Fig. 4). Occasionally, some variation in band intensities was observed between the mutant strains and wild-type (e.g. Fig. 4 low molecular weight bands). However, these differences were sporadic and likely due to minor variations in growth stage from one experiment to another, rather than the introduced mutations.
Although there were no obvious differences in size of the OAg derived from Dslr0977, Dslr0982, Dslr1610 or D(slr0977;Dsll0574-5), it seemed possible that compositional changes in OAg accounted for the phenotypes of these strains. Therefore, purified OAg was analyzed by gas chromatography to determine its composition ( Table 4). The results of the overall composition of Synechocystis OAg are consistent with similar analyses [42]. Glucose was the major component in all wild-type and mutant strains (62.1-76.5%). OAg from mutant and wild-type strains contained similar amounts of glucose as well as of xylose, fucose, galactose, and mannose (Table 4).
Although no significant differences were observed in the OAg profile of the Dslr0982 mutant, increased concentrations of sugars such as rhamnose were detected. These differences were sporadic between experiments, but it is interesting to speculate that the slr0977 gene product may be interacting non-specifically with unidentified homologs of the slr0982 gene product (i.e. other ABC transporter components). Since such slr0982 homologs would have affinity for different carbohydrates, it may explain the sporadic display of unusual sugars in the OAg of this mutant. We have identified 157 putative glycosyl transferases in Synechocystis with homology to gene products of E. coli and Salmonella (Table S1). It is unclear how many of these function in EPS biosynthesis. By engineering strains that have altered glycosyl transferase profiles, it may be possible to engineer altered EPS for designer purposes such as re-capturing waste products from waste water [8,16,17].
EPS is Altered in Strains Dslr0977, Dslr0982, Dslr1610 and D(slr0977; sll0574-5) Surprisingly, OAg structure was not disrupted in the mutant strains Dslr0977, Dslr0982, Dslr1610 and D(slr0977;Dsll0574-5). It seemed likely, therefore, that these gene products play a role in the export of other carbohydrate substrates. To determine if these gene products functioned in EPS biosynthesis, EPS was extracted from stationary phase cultures of wild-type, Dslr0977, Dslr0982, Dslr1610 and D(slr0977;Dsll0574-5) strains. Samples from each culture were analyzed by SDS-PAGE and Alcian blue staining. The Dsll1213 mutant had been previously characterized as an EPS mutant [41] and was used as a positive control for EPS disruption (Fig. 5). The major band from EPS extraction (See Fig. 5, black arrow) is a large molecular weight polymer that runs above 250 kDa. The mutant strains Dslr0977, Dslr0982, Dslr1610 and D(slr0977;Dsll0574-5) all lacked the major EPS band. However, EPS from all mutants tested displayed a smaller EPS band similar to that from the Dsll1213 strain, previously demonstrated to produce altered EPS [41].
Given that the mutants we analyzed were deficient in EPS export compared to wild-type, we determined the composition of mutant EPS. Consistent with altered EPS sizes detected in SDS-PAGE (Fig. 5), we observed differences in both the total carbohydrate in the EPS samples and alterations in the sugar content ( Table 5). As previously described [31], Synechocystis EPS is primarily composed of glucose (,72%), fucose (,10%), xylose (,9%), mannose (,5%), and galactose (,3%) with trace percentages of ribose, rhamnose and N-acetyl glucosamine ( Table 5). EPS samples from Dslr0977 showed a marked decrease in glucose composition (35%) compared to wild-type samples and a concomitant increase in the percentage of ribose (29%). One possibility is that in the absence of the native pore, increased activity of a slr0977 gene product homolog leads to the improper transport of extracellular carbohydrates. In fact, this is likely due to the ABC transporter encoded by sll0574-5, as D(slr0977; Dsll0574-5) demonstrated carbohydrate composition nearly identical to wild-type (Table 5). EPS derived from Dslr0982 and Dslr1610 showed a modest increase in the percentage of ribose from wildtype. While the EPS from the Dslr1610 showed a marked decrease in glucose composition (50%), the EPS of the Dslr0982 mutant was composed of 80% glucose. In addition, the EPS extracted from Dslr0982 contained less fucose than the wild-type. This may indicate that in the absence of slr0982, the slr0977 gene product is capable of utilizing alternative ABC transporter components with a similar affinity for glucose as the slr0982 gene product. It is possible that changes in EPS or OAg can lead to altered outer membrane protein (OMP) profiles, which may play a role in cellular attachment. To determine if this was the case with our mutants, we isolated OMPs from mutant strains and analyzed them by SDS-PAGE. Proteins were visualized by staining with Coomassie brilliant blue (Fig. 5B). This analysis demonstrated that the Dslr0982 strain lacked a protein with A molecular wieght  of ,250 kDa, which likely represents the monomer that comprises the proteinaceous S-layer. This result indicates an important role in fucose production and transport in the attachment of the S-layer to the cell surface, as the fucose minus strain, Dsll1213, is also deficient in S-layer attachment [41].
As expected, no detectable fucose was present in the Dsll1213 strain, which is aberrant in cell wall structure, thylakoid membrane biogenesis, EPS as well as LPS production ( [41] and Fig. 4). Aggregation of the Dsll1213 strain is detectable microscopically [41]. However, a defect in fucose production does not illicit autoflocculation from the culture medium. This may be due to the pleotropic effects of the Dsll1213 mutation, which compensate cell surface charge or structures by an unknown mechanism. However, the differences between these strains may also indicate that perturbations of the transport machinery, rather than the EPS or LPS content per se are responsible for the cellular aggregation.
Here we have bioinformatically identified over five hundred gene products associated cell surface structures in Synechocystis with homology to genes in E. coli and Salmonella (Table S1). These data included slr0977/slr0982 and slr0574-5 which were annotated as gene products in involved in OAg transport. We provide evidence that they are, in fact, involved in the transport of a similar   carbohydrate structure, EPS. By modulating such cell surface moieties, strains can be developed for with autoflocculant phenotypes useful in biomass recovery. Additionally, genetically engineering cyanobacteria with altered cell surface carbohydrate structures in cyanobacteria will prove to be useful tools for bioremediation.

Supporting Information
Table S1 Gene products potentially involved in the synthesis and assembly of the Synechocystis cell wall and cell-surface macromolecular components. (XLS)