Accumulation of UDP-sulfoquinovose in a sulfolipid-deficient mutant of Rhodobacter sphaeroides.

The sulfolipid 6-sulfo-α-D-quinovosyl diacylglycerol is found in the photosynthetic membranes of all plants and most photosynthetic bacteria. Progress toward the elucidation of the pathway for sulfolipid biosynthesis has been slow in the past. However, the recent isolation of three genes of the photosynthetic bacterium Rhodobacter sphaeroides known to be involved in sulfolipid biosynthesis provides promising new opportunities. Two of the genes flank an open reading frame predicted to encode a protein with amino acid sequence similarity to sugar nucleotide-dependent glycosyltransferases. The UDP-sulfoquinovose:diacylglycerol sulfoquinovosyltransferase thought to catalyze the last step of sulfolipid biosynthesis belongs to this group of glycosyltransferases. To test whether this open reading frame encodes the sulfoquinovosyltransferase of R. sphaeroides, it was inactivated by gene replacement avoiding polar mutagenesis. The resulting sulfolipid-deficient mutant defines a new gene, designated sqdD. Mutant cells grown in the presence of [S]sulfate accumulate a water-soluble S-labeled compound. The purified compound was tentatively identified by co-chromatography with standards and enzymatic conversion as UDP-sulfoquinovose, the final precursor of sulfolipid biosynthesis. This result strongly suggests that the inactivation of sqdD causes a metabolic block in the last step of sulfolipid biosynthesis.

The sulfolipid 6-sulfo-␣-D-quinovosyl diacylglycerol is found in the photosynthetic membranes of all plants and most photosynthetic bacteria. Progress toward the elucidation of the pathway for sulfolipid biosynthesis has been slow in the past. However, the recent isolation of three genes of the photosynthetic bacterium Rhodobacter sphaeroides known to be involved in sulfolipid biosynthesis provides promising new opportunities. Two of the genes flank an open reading frame predicted to encode a protein with amino acid sequence similarity to sugar nucleotide-dependent glycosyltransferases. The UDP-sulfoquinovose:diacylglycerol sulfoquinovosyltransferase thought to catalyze the last step of sulfolipid biosynthesis belongs to this group of glycosyltransferases. To test whether this open reading frame encodes the sulfoquinovosyltransferase of R. sphaeroides, it was inactivated by gene replacement avoiding polar mutagenesis. The resulting sulfolipid-deficient mutant defines a new gene, designated sqdD. Mutant cells grown in the presence of [ 35 S]sulfate accumulate a water-soluble 35 S-labeled compound. The purified compound was tentatively identified by co-chromatography with standards and enzymatic conversion as UDPsulfoquinovose, the final precursor of sulfolipid biosynthesis. This result strongly suggests that the inactivation of sqdD causes a metabolic block in the last step of sulfolipid biosynthesis.
The sulfolipid sulfoquinovosyl diacylglycerol is found in all plants and most photosynthetic bacteria (1,2). In higher plants it is confined to the chloroplast (3). Since the discovery (4) and structural elucidation (5) of sulfoquinovosyl diacylglycerol by Benson and co-workers, different pathways for the biosynthesis of the 6-sulfoquinovose-containing lipid have been suggested (see summary in Ref. 1). Experimental verification of the proposed steps of the pathway is generally lacking, but evidence is accumulating that the immediate precursors from which sulfolipid is assembled are UDP-sulfoquinovose and diacylglycerol. The tentative identification of nucleotide sulfoquinovose in extracts of [ 35 S]sulfate-labeled cells of Chlorella (6) led Benson (5) to suggest that the transfer of the sulfoquinovose moiety from a nucleotide sulfoquinovose onto diacylglycerol may represent the last step of sulfolipid biosynthesis. More recently, it was demonstrated that chemically synthesized UDP-sulfoquinovose could stimulate sulfolipid biosynthesis by spinach chloroplast membranes which were biochemically preloaded with radioactive diacylglycerol (7). Since the effect was specific for UDP-sulfoquinovose as compared to other sulfosugar nucleotides, it was concluded that in spinach chloroplasts a UDPsulfoquinovose:diacylglycerol sulfoquinovosyltransferase catalyzes the last step of sulfolipid biosynthesis. Using synthetic UDP-sulfoquinovose as substrate, the biochemical properties of the enzyme and its association with the inner envelope membrane of the chloroplast could be determined (8).
Currently, the only genes known to encode proteins involved in the biosynthesis of sulfoquinovosyl diacylglycerol are sqdA, sqdB, and sqdC isolated from the photosynthetic bacterium Rhodobacter sphaeroides (9,10). Two of the genes, sqdB and sqdC, were found to flank an open reading frame predicted to encode a protein with an amino acid sequence similar to that of glycogenin (9), a sugar nucleotide-dependent glycosyltransferase, but no mutant affecting this open reading frame was available. Given this sequence similarity and the location of the open reading frame within one of the sulfolipid operons of R. sphaeroides, it is tempting to predict that the open reading frame may encode a protein catalyzing the transfer of sulfoquinovose during final assembly of the sulfolipid in analogy to the reaction in spinach chloroplasts. As a first step to test this hypothesis, we had to specifically inactivate this open reading frame, avoiding polar mutagenesis. This experiment would answer the general question whether the open reading frame represents a novel sqd gene of R. sphaeroides essential for sulfolipid biosynthesis. In addition, we set out to identify UDPsulfoquinovose in the mutant cells which we expected to accumulate due to a metabolic block in the final step of sulfolipid biosynthesis, if the protein encoded by the open reading frame would be involved in catalyzing this final reaction.
R. sphaeroides cell cultures were grown photoheterotrophically in Sistrom's medium (17,18) as described previously (9). To obtain low sulfate medium, sulfate salts in Sistrom's medium were substituted with chloride salts. E. coli strains were grown in Luria broth. Antibiotics were added at concentrations of 100 g/ml ampicillin, 50 g/ml kanamycin, or 10 g/ml tetracycline for E. coli cultures or 25 g/ml kanamycin for R. sphaeroides cultures.
Construction of the Inactivation Cassette and Plasmid pMR4D-The plasmid pMR1 containing the new inactivation cassette (see Fig. 1) was constructed by ligating a 1.3-kb 1 EcoRI/HaeII neomycin phosphotransferase gene fragment of pUC4K treated to remove sticky ends and redigested with PstI, a 300-bp EcoRI/HindIII fragment of pCHB500 treated to fill in the 3Ј-ends and redigested with PstI, which contained the cytochrome c 2 promoter (Pcyc) from Rhodobacter capsulatus (19), and the PstI vector fragment of pUC4K. This procedure ensured the cloning of the promoter fragment in proper orientation at the 3Ј-end of the neomycin phosphotransferase gene which was confirmed by restriction analysis.
Construction of the plasmid pMR4D used to inactivate the open reading frame flanked by the sqdB and sqdC genes was accomplished in three steps. A 3.6-kb partial PstI/HindIII fragment from pCHB160018 containing the sulfolipid operon from R. sphaeroides was cloned into pBluescript-KSII ϩ giving rise to plasmid pMR2. This plasmid was cut with EcoRI and ligated with a 1.6-kb EcoRI fragment of pMR1 containing the new inactivation cassette. In the resulting plasmid pMR2D1, a 53-bp fragment of the targeted open reading frame was replaced with the inactivation cassette. The proper orientation of the cassette was confirmed by restriction analysis with XhoI. Finally, a partial 3.6-kb PstI fragment of pMR2D1 containing the interrupted open reading frame followed by the sqdC gene was ligated with pSUP202 cut with PstI. The resulting plasmid pMR4D1 was used to inactivate the respective gene in R. sphaeroides wild type cells.
Genetic Procedures and Recombinant DNA Techniques-Plasmids were transferred from E. coli strain HB101 in a triparental, or from E. coli strain S17-1 in a biparental mating reaction as previously described (9). Routine recombinant DNA techniques were performed as described elsewhere (9,10).
Preparation and Thin Layer Chromatography of 35 S-Labeled Lipids and Water-soluble Compounds-Precultures of R. sphaeroides mutant MRD or wild type were grown photoheterotrophically in low sulfate Sistrom's medium supplemented with 100 M ammonium sulfate to midlog phase. Following 100-fold dilution in the same medium and addition of 15 Ci/ml sodium [ 35 S]sulfate (specific activity, 100 mCi/ mmol) the cells were incubated for 2 days and harvested by centrifugation. Lipid extracts were prepared as described elsewhere (9). Extracts containing water-soluble compounds, in particular sugar nucleotides, were prepared according to Bochner and Ames (20).
Lipids were separated by one-dimensional thin layer chromatography on ammonium sulfate-impregnated silica plates as described previously (9) with one modification to the solvent, the substitution of benzene with toluene.
Radiolabeled water-soluble compounds were separated on Silica Gel 60 F 254 plates (0.2 mm, Merck) developed with one of the following solvents as indicated in the text: solvent A, ethanol/water/acetic acid (20:10:1, by volume); solvent B, isopropanol/dioxane/H 2 O/acetic acid (20:10:10:1, by volume); and solvent C, isopropanol/H 2 O/acetic acid (35:10:1, by volume). The R F values for UDP-sulfoquinovose in the different solvents were: A, 0.7; B, 0.2; and C, 0.6. The 35 S-labeled compounds were visualized by autoradiography. Their relative abundance was determined by densitometry scanning of the respective lanes on the autoradiography employing a laser densitometer (LKB, Ultroscan XL). For preparation of labeled UDP-sulfoquinovose, the compound accumulating in the MRD mutant was extracted with water from the silica powder prepared by scraping the respective spot.
High Performance Liquid Chromatography of Synthetic Sulfosugar and Nucleotide Standards-Different sulfoquinovose nucleotide standards as well as sulfoquinovose 1-phosphate were available from a previous chemical synthesis (7,21). Sulfoquinovose was prepared from synthetic sulfoquinovose 1-phosphate by acid hydrolysis with 1 M aqueous hydrochloric acid for 1 h at 90°C followed by vacuum drying.
Analysis of sulfosugars and nucleotides was accomplished by reversed phase high performance liquid chromatography using a RP-18 column (Bondapak C18 from Waters) with a particle size of 10 m and the dimensions of 3.9 ϫ 300 mm. As mobile phase a mixture of 30 mM KH 2 PO 4 , 2 mM tetrabutylammonium hydroxide, adjusted to pH 6.0 with KOH (solvent A) and acetonitrile (solvent B) was used. For gradient elution at a flow rate of 1 ml/min a linear gradient from 0 to 30% of solvent B in A was run for 30 min, and the solvent mixture was kept for an additional 15 min at this ratio. The column was returned to start conditions with a linear gradient over 10 min ending with 100% solvent A followed by equilibration for 40 min. Since an on-line mass detector for sugar analysis was not available, different modes of detection were used: recording the UV absorption at 254 nm followed in line by radiodetection employing a solid-phase flow-through scintillation detector (Berthold LB 507) or manual collection of fractions for colorimetric determination of sugar content and parallel scintillation counting. The second combination was used to detect sulfoquinovose and sulfoquinovose 1-phosphate. Approximately 0.4-ml fractions were collected and dried in vacuum for 3 h. The residues were dissolved in 300 l of water. An aliquot of 100 l was subjected to scintillation counting. The remaining solution was rapidly mixed with 400 l of anthrone reagent (3 mg/ml in concentrated H 2 SO 4 ). Following incubation for 20 min at room temperature, the extinction was measured at 600 nm.
Enzymatic Hydrolysis of UDP-sulfoquinovose-The 35 S-labeled compound co-chromatographing with UDP-sulfoquinovose (5000 dpm) was dissolved in 17 l of reaction buffer (10 mM Hepes/KOH, 10 mM MgCl 2 , pH 7.4) and mixed with 10 l of a 10 mM solution of unlabeled synthetic UDP-sulfoquinovose. From this solution an aliquot of 7 l was used directly for analysis, and a second aliquot of 10 l was mixed with 0.4 unit of nucleotide pyrophosphatase (Sigma) dissolved in 10 l of reaction buffer. To the third 10-l aliquot of the UDP-sulfoquinovose mixture, an excess of alkaline phosphatase (0.6 unit, Sigma) in 10 l of reaction buffer was added in addition to nucleotide pyrophosphatase. After incubation for 3 h at room temperature, the reactions were stopped by the addition of 4 l of 11 M formic acid and kept on ice for 10 min. Precipitates were removed by centrifugation, and the solvent was removed under vacuum. The residues were dissolved in 20 l of water and subjected to further analysis.
Preparation of Chloroplast Membranes and Glycosyltransferase Assay-Approximately 20 g of leaf material of 6-week-old spinach plants were harvested and homogenized in 4°C cold isolation medium (40 mM Tricine/KOH, 300 mM sorbitol, pH 8.0) by four short bursts in a Waring blender. The extracts were kept at 4°C during the whole procedure. Following filtration through two layers of Miracloth, the suspension was centrifuged at 3,000 ϫ g for 5 min and the pellet was resuspended in 0.8 ml of isolation medium. The chloroplasts were purified by passage through a Percoll cushion (40% in isolation medium; modified according to Mills and Joy (22)) and washed in a small volume of isolation medium. The washed chloroplasts were broken in swelling buffer (10 mM Tricine/KOH, 4 mM MgCl 2 , pH 8.0), centrifuged at 4,500 ϫ g for 2 min, and the pellet was resuspended in 50 mM Tricine/KOH (pH 7.5). This pellet consists mostly of thylakoid membranes. In addition, about one-third of the envelope membranes found in intact chloroplasts is retained in the pellet.
Glycosyltransferase reaction mixtures of 30 l total volume contained approximately 5,000 dpm of the 35 S-labeled compound accumulating in the MRD mutant or 100,000 dpm UDP-[U-14 C]galactose (2.5 Ci/mol, Amersham Corp.) and broken chloroplast membranes equivalent to 35 g of chlorophyll in reaction buffer (50 mM Tricine/KOH, 30 mM MgCl 2 , pH 7.5). Following incubation for 1 h at 25°C, the reaction was stopped by the addition of 250 l of methanol/chloroform (1:1, v/v) and 100 l of 1 M KCl, 0.2 M phosphoric acid. Chloroform-soluble material was analyzed by thin layer chromatography as described above.
Quantitation of total chlorophyll in 80% acetone was done according to Lichtenthaler (23).

Inactivation of the Open Reading Frame Flanked by sqdB and sqdC-To
test whether the open reading frame located between sqdB and sqdC in the sulfolipid operon of R. sphaeroides represents a new gene involved in sulfolipid biosynthesis, we had to inactivate this putative gene, but allow for the expression of sqdC. From previous experiments it was known that sqdB and sqdC, which are essential for sulfolipid biosynthesis in R. sphaeroides, form a transcriptional unit with sqdC at the 3Ј-end. To ensure the expression of sqdC, we modified the inactivation cassette from pUC4K by cloning the cytochrome c 2 promoter of R. capsulatus in outward reading orientation behind the 3Ј-end of the neomycin phosphotransferase gene (Fig. 1). This promoter was successfully used before to express sqdC (10). After deleting a 53-bp EcoRI fragment, this new cassette was inserted into the center of the target open reading frame. The insertionally disrupted central portion of the operon carried by a plasmid not replicating in R. spha-eroides (pMR4D1) was transferred by conjugation into the wild type. The kanamycin-resistant exconjugants were tested by Southern hybridization for the replacement of the wild type DNA following a double crossover and the complete loss of the wild type genome copies. Most of the lines showed a hybridization pattern consistent with the predicted genomic restriction map (Fig. 2). In addition, no wild type copies could be detected. All lines showing this restriction pattern behaved identical during further experiments and were generally designated MRD.
No sulfolipid band could be detected by iodine staining on thin layer chromatograms of lipids extracted from three independent MRD lines. To use the most sensitive detection method available, we grew cells of one of the MRD lines and wild type in the presence of [ 35 S]sulfate. Comparison of a serial dilution of wild type lipid extract and undiluted MRD lipid extract by thin layer chromatography and autoradiography revealed that in MRD the sulfolipid content is reduced at least by a factor of 1000 (Fig. 3). Therefore, inactivation of the open reading frame flanked by sqdB and sqdC in the MRD lines results in a sulfolipid deficient mutant of R. sphaeroides.
Complementation Analysis Reveals a New Sulfolipid Gene of R. sphaeroides-To conclude that the open reading frame represents a gene crucial for sulfolipid biosynthesis in R. sphaeroides, we needed to demonstrate that the insertion in MRD did not negatively affect the expression of sqdB or sqdC. For this purpose we designed an experiment based on genetic complementation (Fig. 4). We introduced different DNA subfragments of the sulfolipid operon of R. sphaeroides in trans into three mutant lines, the line MRD described above or the lines CHB16 and CHB18 carrying a mutation in sqdB or sqdC, respectively (10). Gene expression from these DNA fragments was either under the control of the endogenous promoter or the cytochrome c 2 promoter of R. capsulatus. Mutant exconjugants were tested for their competence for sulfolipid biosynthesis indicating complementation by the respective fragment. All fragments containing the complete sqdB reading frame complemented line CHB16, all fragments containing the complete sqdC open reading frame complemented line CHB18, and all fragments containing the open reading frame in the center of the operon complemented line MRD (Fig. 4). The result obtained for clone 184 allowed us to answer in a decisive way the question whether the central open reading frame represents a new sulfolipid gene. This clone contains only the central open reading frame of the operon in its full length, but it nevertheless can complement the defect in MRD which carries an insertion in the genomic copy of this open reading frame. This result strongly suggests that sqdB and sqdC are expressed in MRD and that the sulfolipid deficiency in this line is due to the inactivation of the central open reading frame only. Therefore, we conclude that this open reading frame represents a new gene, tentatively designated sqdD, which is crucial for sulfolipid biosynthesis in R. sphaeroides.
Inactivation of sqdD Results in the Accumulation of a 35 S-Labeled Water-soluble Compound-The sequence similarity between the protein predicted to be encoded by sqdD and the glycosyltransferase glycogenin suggested that sqdD may encode the sulfoquinovosyl transferase catalyzing the last step of sulfolipid biosynthesis. A total loss of this enzymatic activity as predicted for MRD would lead to a block in the last step of sulfolipid biosynthesis and to the accumulation of precursors, particularly UDP-sulfoquinovose. To test this hypothesis, we incubated cells of wild type and line MRD in the presence of [ 35 S]sulfate which was readily taken up by sulfur-starved cells and incorporated into all compounds containing sulfur. We optimized the extraction procedure toward the isolation of sugar nucleotides (20) and separated the water-soluble compounds contained in the extracts by thin layer chromatography using conditions known to be optimal for UDP-sulfoquinovose (7). Using this approach we could detect a 35 S-labeled compound of much greater abundance in extracts of MRD cells (Fig.  5). Densitometric scanning of TLC lanes as shown in Fig. 5 indicated that the relative amount of the labeled compound was increased 8 -10-fold in the mutant. With the goal of performing structural analysis we attempted to purify a larger amount of this compound. However, even after considerable scale up we were not able to detect the compound by means other than autoradiography. In addition, after optimization of the labeling experiment still only 0.001% of the labeled sulfur in the culture medium (1 mCi/500 ml) was found in this compound. Based on the specific radioactivity of sulfate used in the experiment (100 mCi/mmol), we could estimate that approxi- mately 10 nmol of the compound were produced in a 500-ml culture of the MRD mutant containing approximately 0.5 g of wet cells. Purification of sufficient unlabeled material for detailed structural elucidation of this compound employing different methods of spectroscopy would require a considerable increase in the scale of the experiment. As a more feasible alternative, we isolated the radiolabeled compound in sufficient amount to permit an identification based on chromato-graphic behavior and substrate properties for defined enzymes.
The Accumulating Compound Co-chromatographs with Synthetic UDP-sulfoquinovose-Using three different thin layer chromatography systems as described under "Materials and Methods," we were able to demonstrate that the R F values for the labeled compound accumulating in MRD and synthetic UDP-sulfoquinovose were identical (data not shown). Furthermore, we employed high performance liquid chromatography to resolve sulfosugar nucleotides according to their constituent base. Four synthetic nucleoside sulfoquinovoses were mixed with a small sample of the labeled compound and analyzed. Given that the radio detector signal is slightly lagging behind due to the experimental setup, a clear match of retention times between the radiodetector signal indicative for the unknown compound and the UV detector signal for synthetic UDP-sulfoquinovose was apparent (Fig. 6).
The nucleotide structure of the labeled compound was confirmed by cleavage with specific enzymes and subsequent identification of the cleavage products by co-chromatography with standards. The labeled compound was mixed with an excess of synthetic unlabeled UDP-sulfoquinovose and treated either with nucleotide pyrophosphatase or a combination of nucleotide pyrophosphatase and alkaline phosphatase. The nucleotide pyrophosphatase cleaves sugar nucleotides releasing a sugar phosphate and a nucleoside monophosphate. In the presence of alkaline phosphatase, these products are converted to free sugar and the corresponding nucleoside. Following the enzymatic hydrolysis, the products were analyzed by high performance liquid chromatography. The elution of the base containing part of synthetic UDP-sulfoquinovose was monitored on-line by UV absorption, and the radioactive portion of the unknown compound by radio detection. Treatment with nucleotide pyrophosphatase resulted in complete disappearance of the UV signal for UDP-sulfoquinovose and the overlapping radio signal characteristic for the unknown compound. Concomitantly, a UV signal for uridine monophosphate and a radio signal for a radioactive compound escaping UV detection and eluting after uridine monophosphate appeared (data not shown). Combined treatment with nucleotide pyrophosphatase and alkaline phosphatase resulted in a UV signal for uridine and a radio signal for a new compound escaping UV detection and eluting ahead of uridine (data not shown). Under the conditions employed, a complete conversion of substrates was observed for the two different reactions.
To demonstrate that the radioactive hydrolysis products cochromatographed with the expected compounds, synthetic sulfoquinovosyl 1-phosphate and sulfoquinovose (40 g each) were added after termination of the reactions to the hydrolysis mixtures and subjected to high performance liquid chromatography. For this purpose the effluent was manually collected, and the fractions were monitored for the presence of sulfoquinovose (colorimetry with anthrone) and radioactive compounds. Nucleotide pyrophosphatase released a labeled product co-chromatographing with sulfoquinovosyl 1-phosphate (Fig. 7, right  traces). The additional treatment with alkaline phosphatase produced a labeled product co-chromatographing with sulfoquinovose (Fig. 7, left traces). Taken together, these results present strong biochemical evidence for the identity of the unknown compound with UDP-sulfoquinovose.
Chloroplast Membranes from Spinach Incorporate the 35 S-Labeled Part of the Compound into Sulfolipid-To demonstrate that the compound accumulating in MRD is a precursor of sulfolipid biosynthesis, we initially incubated French press extracts from R. sphaeroides wild type cells in the presence of a small aliquot of the compound. However, using this crude system we could not detect any incorporation of label into sulfolipid. We therefore used isolated spinach membranes which contain a well characterized UDP-sulfoquinovose:diacylglycerol sulfoquinovosyltransferase activity (8). Incubation of these membranes with a small aliquot of the labeled compound accumulating in MRD resulted in the formation of labeled sulfolipid (Fig. 8).
As a control, we incubated the membranes in parallel with UDP-[U-14 C]galactose to monitor the formation of monogalactosyl diacylglycerol. Given the well characterized substrate specificity of the spinach sulfoquinovosyltransferase, this result provides further biochemical evidence that the compound accumulating in MRD is the sulfolipid precursor UDP-sulfoquinovose. DISCUSSION Our objective was to test whether the open reading frame flanked by sqdB and sqdC in the sulfolipid operon of R. sphaeroides represents a new gene crucial for sulfolipid biosynthesis. Our strategy was to disrupt the open reading frame without affecting the expression of the neighboring sulfolipid genes. As a precautionary measure we designed an inactivation cassette containing the outward reading cytochrome c 2 promoter from R. capsulatus at the 3Ј-end of a neomycin phosphotransferase  7. Analysis of the radioactive cleavage products by high performance liquid chromatography following treatment of the 35 S-labeled unknown compound with hydrolytic enzymes. A mixture of the labeled unknown compound and synthetic UDP-sulfoquinovose (100 nmol) was treated either with nucleotide pyrophosphatase (triangles) or with a combination of nucleotide pyrophosphatase and alkaline phosphatase (circles). Following the termination of the reactions synthetic sulfoquinovose or sulfoquinovose 1-phosphate standards were added (40 g each). The two reaction mixtures were analyzed separately by high performance liquid chromatography. Sulfosugars (open symbols) and radioactivity (closed symbols) were monitored in manually collected fractions. The elution profiles from both experiments were combined in one figure since neither radioactivity nor sugars were detected at different retention times than indicated for each experiment. SQVP, sulfoquinovosyl phosphate; SQV, sulfoquinovose.
gene. Insertion of this cassette in proper orientation resulted in specific inactivation of the targeted open reading frame which was confirmed by complementation analysis. However, further experiments will be required to decide whether this cassette can be used in a general way for nonpolar mutagenesis in R. sphaeroides. The specific disruption of the targeted open reading frame led to sulfolipid deficiency thereby defining a new sulfolipid gene, which we tentatively designated sqdD. It is the fourth gene of R. sphaeroides which was shown to be crucial for sulfolipid biosynthesis using mutational analysis combined with genetic complementation (9,10). Three of the genes, sqdB, sqdD, and sqdC are located in one transcriptional unit (Fig. 4).
Of all four genes, only for sqdD a clearer picture emerges about the possible function of the gene product. First, the predicted amino acid sequence encoded by this gene shows sequence similarity to a UDP-glucose-dependent glycosyltransferase (10), an enzyme with functional analogy to the UDPsulfoquinovose-dependent sulfoquinovosyltransferase known to catalyze the last step of sulfolipid biosynthesis in spinach chloroplasts (8). Second, the MRD mutant described here which is unable to express sqdD accumulates a 35 S-labeled compound tentatively identified as UDP-sulfoquinovose. Since we were unable to purify a sufficient amount of unlabeled compound for structural analysis by spectroscopy, we alternatively applied two established experimental approaches to determine the possible chemical nature of the 35 S-labeled compound. One line of evidence was based on co-chromatography with synthetic standards, the second on the conversion of the UDP-sulfoquinovose by enzymes with well defined substrate specificities. The compound clearly co-chromatographed in the three different thin layer systems tested, as well as in the high performance liquid chromatography system with synthetic UDP-sulfoquinovose. It was completely cleaved by nucleotide pyrophosphatase, confirming a nucleotide structure. The resulting 35 S-labeled cleavage product co-chromatographed with sulfoquinovose 1-phosphate. Further dephosphorylation using alkaline phosphatase resulted in co-elution of the labeled cleavage product with sulfoquinovose. These independent observations were all in agreement with a UDP-sulfoquinovose structure for the compound accumulating in MRD. In addition, the labeled portion of the molecule was incorporated into a lipophilic compound co-chromatographing with sulfolipid following the incubation with isolated spinach chloroplast mem-branes. These membranes contain a sulfoquinovosyltransferase highly specific for UDP-sulfoquinovose as compared to other nucleotide sulfoquinovoses (7). Taken together, these results provide convincing evidence for a UDP-sulfoquinovose structure of the compound accumulating in the sqdD insertion mutant MRD. More than 30 years ago, a sulfurcontaining sugar nucleotide, presumably UDP-sulfoquinovose, was observed by paper chromatography of Chlorella extracts (6), a result which has not been reproduced since.
To address the question whether sqdD encodes the sulfoquinovosyltransferase of R. sphaeroides we tried to express this gene in E. coli in a functional form using different strategies. Currently, all our attempts have failed. Although the accumulation of presumably UDP-sulfoquinovose in a sqdD-inactivated line supports the idea that the sqdD protein at least may be involved in some aspect of this transfer reaction, it remains to be seen whether the holoenzyme requires additional proteins or factors missing in E. coli.
The MRD mutant line provides a new tool for sulfolipid research. Since it apparently accumulates UDP-sulfoquinovose it can be utilized as a source for radiolabeled UDP-sulfoquinovose, particularly if the yield can be increased. Currently, the multistep chemical synthesis of UDP-sulfoquinovose (7) represents no feasible alternative to the biochemical preparation of labeled UDP-sulfoquinovose with high specific activity as required for biochemical studies.
The accumulation of UDP-sulfoquinovose in a sulfolipid-deficient mutant of R. sphaeroides and the incorporation of the labeled part of this molecule into sulfolipid by isolated spinach chloroplast membranes provide further experimental evidence for the hypothesis that the head group of sulfolipid is derived from UDP-sulfoquinovose in photosynthetic bacteria and higher plants. Whether the transfer reaction itself proceeds in the same way in photosynthetic bacteria and plants must await further experiments. The question for the pathway of sulfolipid biosynthesis can be reduced to the question for the biosynthesis of UDP-sulfoquinovose. Further analysis of the proteins encoded by the four known sulfolipid genes of R. sphaeroides, particularly in vitro reconstitution experiments, may provide the answer to this question in the future.