Glycolytic Breakdown of Sulfoquinovose in Bacteria: a Missing Link in the Sulfur Cycle

Abstract

Sulfoquinovose (6-deoxy-6-sulfo-d-glucopyranose), formed by the hydrolysis of the plant sulfolipid, is a major component of the biological sulfur cycle. However, pathways for its catabolism are poorly delineated. We examined the hypothesis that mineralization of sulfoquinovose to inorganic sulfate is initiated by reactions of the glycolytic and/or Entner-Doudoroff pathways in bacteria. Metabolites of [U-¹³C]sulfoquinovose were identified by ¹³C-nuclear magnetic resonance (NMR) in strains of Klebsiella and Agrobacterium previously isolated for their ability to utilize sulfoquinovose as a sole source of carbon and energy for growth, and cell extracts were analyzed for enzymes diagnostic for the respective pathways. Klebsiella sp. strain ABR11 grew rapidly on sulfoquinovose, with major accumulations of sulfopropandiol (2,3-dihydroxypropanesulfonate) but no detectable release of sulfate. Later, when sulfoquinovose was exhausted and growth was very slow, sulfopropandiol disappeared and inorganic sulfate and small amounts of sulfolactate (2-hydroxy-3-sulfopropionate) were formed. In Agrobacterium sp. strain ABR2, growth and sulfoquinovose disappearance were again coincident, though slower than that in Klebsiella sp. Release of sulfate was still late but was faster than that in Klebsiella sp., and no metabolites were detected by ¹³C-NMR. Extracts of both strains grown on sulfoquinovose contained phosphofructokinase activities that remained unchanged when fructose 6-phosphate was replaced in the assay mixture with either glucose 6-phosphate or sulfoquinovose. The results were consistent with the operation of the Embden-Meyerhoff-Parnas (glycolysis) pathway for catabolism of sulfoquinovose. Extracts of Klebsiella but not Agrobacterium also contained an NAD⁺-dependent sulfoquinovose dehydrogenase activity, indicating that the Entner-Doudoroff pathway might also contribute to catabolism of sulfoquinovose.

Since its discovery nearly half a century ago (4), the plant sulfolipid (sulfoquinovosyldiacylglycerol [SQDG]) has become recognized as a major component of the biological sulfur cycle (20). For plants, it accounts for at least 10 mg/g of dry weight of leaf tissue (21), yet our understanding of its biosynthesis and function in photosynthetic tissues is still emerging (2). A pathway for SQDG biosynthesis was proposed with data from biochemical experiments (30), and, recently, genes have been isolated that encode proteins that catalyze the conversion of UDP-glucose to UDP-sulfoquinovose (49) and the use of the latter for SQDG formation (35). Ultimately, the source of sulfur for the synthesis of SQDG is inorganic sulfate from the soil (2, 21, 28). However, many soils are becoming sulfur deficient as a result of intensive agriculture and the diminution of sulfur deposition in rainfall following measures to reduce sulfur emissions to the atmosphere (7, 22, 36, 50). In these circumstances, mineralization and recycling of organic sulfur become critically important in sustaining the sulfur cycle, and it is, therefore, surprising that little is known about the pathways by which sulfur is released from sulfolipid in plant tissues undergoing degradation in soils.

The usual view of SQDG catabolism (19) is that the lipid is rapidly hydrolyzed by plant acyl hydrolases to form the deacylated product, sulfoquinovosylglycerol (3, 39). Further attack by β-galactosidase yields sulfoquinovose (37). This reaction is significant in some plants (10) but is usually much less active than deacylation (38, 39). Studies on sulfoquinovosylglycerol metabolism in plant leaf tissue also showed formation of C-2 and C-3 sulfonated intermediates, indicating a possible role for the so-called sulfoglycolytic pathway (24) for the initial stages of catabolism prior to sulfate release. However, the formation of sulfate from sulfoquinovose is thought to be dominated by soil microorganisms (26, 29, 42), and Martelli and Benson (26) showed that a putative Flavobacterium from soil could grow on methyl α-sulfoquinovoside and accumulate intracellular cysteate, sulfoacetate, and sulfate. However, nothing is known of the reactions involved in the mineralization of the sulfur of sulfoquinovose.

If sulfoquinovose is considered an analog of glucose 6-phosphate, then the Embden-Meyerhoff-Parnas (glycolysis) pathway (16) is a prime candidate for sulfoquinovose degradation in bacteria. In addition, the Entner-Doudoroff (14) pathway operating via 6-phosphogluconate may serve as a bypass for reactions of the upper glycolytic pathway (13). In a preliminary communication (33) we reported the isolation of five bacterial strains (four from forest leaf mould and one from activated sewage sludge) that grew on sulfoquinovose as a sole source of carbon. Two of these strains achieved mineralization of sulfoquinovose to inorganic sulfate. In the present paper, using a combination of metabolite identification by ¹³C-nuclear magnetic resonance (NMR), analysis for enzymes diagnostic for the operation of the respective pathways, and direct tests on the action of these enzymes on sulfoquinovose, we examined the hypothesis that these bacterial strains used the Embden-Meyerhoff-Parnas and Entner-Doudoroff bypass pathways to initiate catabolism of sulfoquinovose.

MATERIALS AND METHODS

Chemicals.

The potassium salt of sulfoquinovose (potassium 6-deoxy-6-sulfonato-d-glucopyranose) was prepared and characterized as previously described by Roy and Hewlins (34). [U-¹³C]sulfoquinovose was prepared in a yield of 38% by a scaled-down version of the same method, starting from 250 mg of [U-¹³C]glucose (99 atom% ¹³C; Aldrich, Gillingham, Dorset, United Kingdom). All substrates, cofactors, and enzymes for use in enzyme assays were from Sigma-Aldrich.

Sodium sulfolactate (2-hydroxy-3-sulfopropionate) was obtained by the deamination of l-cysteic acid by diazotization in acid conditions as described elsewhere (48), except that 2 M HCl was replaced by 2 M acetic acid. The barium salt of dl-sulfopropandiol (dl-2,3-dihydroxypropanesulfonate) was obtained from 1-tosylglycerol (46) by displacement of the tosyl group with sulfite as described for the preparation of sulfoquinovose (34).

Isolation and culturing of bacteria.

Bacteria were isolated from forest leaf mould and activated sewage sludge by enrichment culture (five rounds), as reported previously (33), in a minimal medium containing 0.1% (wt/vol) sulfoquinovose (3.5 mM, K⁺ salt) in basal salts lacking any other added sulfur or carbon and containing (in grams per liter) K₂HPO₄ (3.5); KH₂PO₄ (1.5); NH₄Cl (0.5); NaCl (0.5); MgCl₂ · H₂O (0.15); and 2 ml of trace elements solution. Trace element solution contained (in milligrams per liter) FeCl₃ · 6H₂O (120); CoCl₂ · 6H₂O (20); Cu(NO₃)₂ · 6H₂O (25); MnCl₂ · 4H₂O (15); ZnCl₂ (75); and Na₂MoO₄ · 6H₂O (15). Axenic strains were isolated and maintained on nutrient agar. Isolates were tested for the ability to grow axenically on the original enrichment medium. Of the five successful isolates (ABR1, ABR2, ABR6, ABR9, and ABR11) obtained by Roy et al. (33), strains ABR2 and ABR11 were selected for further study because of their ability to liberate inorganic sulfate from sulfoquinovose. All cultures were grown at 30°C with shaking (150 rpm).

Inocula for starting cultures for metabolic experiments were obtained by inoculating 20 ml of a phosphate-basal salts medium containing 3.5 mM sulfoquinovose (sole carbon source) from a colony on a nutrient agar plate (no older than 1 week from subculturing) and incubating overnight. The phosphate-basal salts medium contained no added source of sulfur, but no extra precautions were taken to ensure that it was sulfur free.

16S rRNA analysis.

Preliminary identification of strains was achieved by sequence analysis of genes encoding 16S rRNA with the PCR primers 63f (5′-CAG GCC TAA CAC ATG CAA GTC-3′) and 1387r (5′-GGG CGG WGT GTA CAA GGC), designed and evaluated by Marchesi et al. (25) for the amplification of 16S rRNA genes. Sequences were compared to known DNA sequences encoding 16S rRNA in the composite nonredundant database (including GenBank, EMBL, DDBJ, and PDB) by using the BLAST search program (1). Strains were also identified independently by 16S rRNA gene sequence analysis using the MicroSeq 500 Microbial Identification System (Applied Biosystems) at the National Collection for Industrial and Marine Bacteria (Aberdeen, Scotland).

Metabolic experiments.

Independent cultures were established by using either ¹³C natural abundance sulfoquinovose (for measurements of growth, sulfoquinovose disappearance, and sulfate production) or ¹³C-enriched sulfoquinovose (for detection of metabolites by NMR). In order to give acceptable NMR spectra, it was necessary to elevate the concentration of sulfoquinovose (¹³C-enriched) in the basal salts medium to 10.6 mM (0.3% [wt/vol] of the K⁺ salt). Samples of the cultures were taken at intervals (see Results) for ¹³C-NMR spectroscopy and generally were used without removal of the cells. In some cases, the cells were removed by centrifugation for 5 min at 9,000 × g and the supernatant was taken to dryness in a rotary evaporator at 30°C. The residue was exchanged three times with D₂O (deuterium oxide, ²H₂O) to allow the ¹H-NMR spectra to be obtained. For the growth experiments, the composition of the medium was the same as that for the NMR experiments, except that ¹³C natural abundance sulfoquinovose was used. Growth was followed by measuring attenuance of light at 600 nm (D₆₀₀). Samples of the culture medium for the determination of sulfate and sulfoquinovose (as reducing sugar) were obtained by centrifuging for 5 min at 9,000 × g to remove the cells.

Analytical methods.

Sulfoquinovose was determined quantitatively, as reducing sugar, by the dinitrosalicylic acid method of Sturgeon (43). Although this method is not specific for sulfoquinovose and is unreliable when the sugar concentration is less than about 0.5 mM (because the calibration curve becomes nonlinear), it was nevertheless useful for showing disappearance of reducing sugars during growth of bacteria. Specific detection of sulfoquinovose in growth media was achieved by thin-layer chromatography (TLC) on phosphate-impregnated silica gel plates (18) with the solvent system acetone:propan-2-ol:0.1 M formic acid (2:2:1 by volume) (15). Reducing sugars and oxidizable compounds were detected on TLC plates with aniline-diphenylamine and alkaline KMnO₄, respectively.

Sulfate was determined by the turbidimetric method of Sörbo (41). The reproducibility of the method was improved by using plastic, rather than glass, test tubes.

NMR.

NMR spectra were recorded at 20°C in 5-mm-diameter tubes on a Bruker AMX360 spectrometer operating at 360 MHz for ¹H and 90 MHz for ¹³C. The ¹³C spectra were recorded with a pulse angle of 30° and a delay time of 2 s with 32k data points collected over 21,700 Hz. For the metabolic experiments, 1,300 accumulations were made over 1 h, with 5% of D₂O added to the culture medium for field-frequency lock. For determination of ¹³C spectra of standard samples, the solvent was the same phosphate-basal salts medium as that used for growth, supplemented with 5% D₂O. All ¹H spectra were obtained by dissolving samples (standards or dried metabolites) in D₂O. Chemical shifts are given relative to ¹H (δ = 0) and ¹³C (δ = −2.6 ppm) signals in sodium 4,4-dimethyl-4-silapentanesulfonate used as the external standard. The number of protons attached to each ¹³C nucleus was determined by the DEPT-135 method, and the ¹H-¹³C correlation was done by HETCOR. Standard Bruker software (XWINNMR) was used throughout.

Preparation of cell extracts.

Flasks containing 100 ml of basal salts plus 10.6 mM sulfoquinovose were inoculated with 5-ml volumes of starter cultures grown for 48 h on basal salts plus 3.5 mM sulfoquinovose. Culture flasks were incubated at 30°C, and when the D₆₀₀ reached about 0.4 (44 and 16 h for strains ABR2 and ABR11, respectively), cells were harvested by centrifugation (12,100 × g for 10 min; Beckman JA20 rotor). The cell pellets from 100-ml batches of culture medium were resuspended separately in 1-ml aliquots of 10 mM-Tris/HCl buffer (pH 7.6), and the cells were ruptured by sonication for 40 s at 0°C (four times, 10 s on and 20 s off, 50% duty cycle, power setting 7; Lucas Dawes). The lysates were centrifuged as before, and the supernatants containing 1.8 to 4 mg ml⁻¹ of protein were stored on ice and used for subsequent enzyme assays on the same day.

Enzyme and protein assays.

Glucose 6-phosphate dehydrogenase activity in cell extracts was determined in microtiter plate assays by the glucose 6-phosphate-dependent reduction of NAD⁺ or NADP⁺ to NAD(P)H (6). Each well contained 83 mM Tris/HCl buffer (pH 7.5), 6.6 mM MgCl₂, 1.1 mM glucose-6-phosphate, 0.4 mM NAD⁺ or NADP⁺, and up to 50 μl of cell extract (containing 20 to 100 μg of protein) in a total volume of 240 μl. Plates containing all reagents except cofactors and enzyme were preincubated for 10 min at 30°C in a Thermomax microtiter plate reader (Molecular Dynamics, Wokingham, United Kingdom). Enzyme sample was then added, the plate was reequilibrated for 2 min, and the reactions were started by addition of cofactor. Formation of NAD(P)H was monitored by its absorbance at 340 nm over a period of 5 min, with measurements recorded automatically every 10 s. For some assays, glucose-6-phosphate was replaced with sulfoquinovose to determine sulfoquinovose dehydrogenase activity.

The apparent K_m for sulfoquinovose was estimated by measuring rates at sulfoquinovose concentrations of 1.18 to 35.4 mM with the NAD⁺ concentration kept constant at 0.5 mM. The apparent K_m for NAD⁺ was estimated by measuring rates at NAD⁺ concentrations in the range 0.5 to 2.75 mM with 1.18 mM sulfoquinovose. Data were fitted to the Michaelis-Menten equation by using Sigmaplot graphics software (Jandel Scientific, Erkrath, Germany). Glucose 6-phosphate dehydrogenase from Leuconostoc mesenteroides (nominal specific activity of 833 U/mg, diluted to 1 U/ml, 48 μl per well; Sigma-Aldrich) was used as a positive control to validate the assay method. Incubation mixtures lacking glucose 6-phosphate and sulfoquinovose were used to check for the absence of nonspecific reduction of NAD(P)⁺ (negative controls). Incubations lacking commercial enzyme or cell extract showed no nonenzymatic reduction of either cofactor in the presence of either glucose 6-phosphate or sulfoquinovose. The limit of detection of the assay was determined as twice the standard deviation from the mean of all of the no-enzyme and no-substrate controls (n = 21).

Phosphofructokinase activity was determined in coupled assays in which ADP that was formed in the reaction of ATP with fructose 6-phosphate was reacted with phosphoenolpyruvate to liberate pyruvate, which in turn was reduced to lactate by NADH (6). Assay mixtures contained (per well) 70 mM-Tris/HCl buffer (pH 8.5), 1.4 mM MgSO₄, 4.6 mM KCl, 2.3 mM fructose 6-phosphate, 1.3 mM ATP, 0.86 mM fructose 1,6-bisphosphate, 0.75 mM phosphoenolpyruvate, 0.5 mM NADH, 2 U of rabbit muscle pyruvate kinase, 4 U of rabbit muscle lactate dehydrogenase, and up to 80 μl of cell extract in a final volume of 224 μl. All reagents except cell extract were preincubated for 10 min at 30°C in the plate reader, and reactions were started by adding enzyme sample. The rate of disappearance of NADH was measured by monitoring its absorbance at 340 nm over a period of 5 min, with measurements recorded automatically every 10 s. Phosphofructokinase from Bacillus stearothermophilus (nominal specific activity of 0.61 U/mg, diluted to 5 U/ml, 10 μl per well; Sigma-Aldrich) was used as a positive control to validate the assay method. In some experiments, fructose 6-phosphate was replaced with either glucose 6-phosphate or sulfoquinovose as potential substrates for kinase activity.

For all enzyme assays, each rate of reaction was calculated by using the first 10 data points (over 100 s) in the initial linear part of the curve. One unit of enzyme activity was defined as the amount of enzyme that liberates 1 μmol of product per min under the assay conditions. Protein concentration in extracts was determined by the method of Bradford (9) with bovine serum albumin as a standard.

Nucleotide sequence accession numbers.

DNA sequences of 16S rRNA gene fragments were deposited in the EMBL nucleotide sequence database, with accession numbers AJ564106 and AJ564107 for strains ABR2 and ABR11, respectively.

RESULTS

Characterization of the organisms.

Analysis of ca. 200-bp fragments of genes carrying 16S rRNA retrieved by using primers recommended by Marchesi et al. (25) indicated that the nearest phylogenetic neighbors of ABR2 and ABR11 were Agrobacterium tumefaciens (89% identity) and Klebsiella (98% identity for both K. planticola and K. ornithinolytica), respectively. Further analysis (National Collection for Industrial and Marine Bacteria, Aberdeen, Scotland), using MicroSeq 500 (Applied Biosystems) and ca. 450-bp sequences, confirmed the genus match (97.87%) for ABR2 as Agrobacterium, with a provisional species identification as A. tumefaciens, and produced a match (99.79% identity) for ABR11 as either K. ornithinolytica or K. planticola. Normally this level of similarity would provide species identification, but given the similarities of 16S rRNA sequences in the family Enterobacteriaceae, species resolution was not possible. The strains were thus designated Agrobacterium sp. strain ABR2 and Klebsiella sp. strain ABR11.

Metabolic experiments with Klebsiella sp. strain ABR11.

Figure 1a shows the growth of Klebsiella sp. strain ABR11 on 10.6 mM sulfoquinovose as the sole added source of carbon and sulfur, together with the changes in sulfoquinovose and sulfate concentrations in the growth medium. Rapid exponential growth in the first 8 h of incubation was coincident with extensive utilization of sulfoquinovose (determined as reducing sugar), the concentration of which fell to <1 mM by the end of the exponential phase. The apparent residual sulfoquinovose (see Fig. 1a) was probably an artifact arising from the low specificity of the analytical method, because no sulfoquinovose (or other reducing sugar) was detected by TLC (limit of detection, <0.1 mM). No inorganic sulfate was detected in the growth medium for at least 9 h, but small amounts of sulfate, equivalent to about 10% of the sulfoquinovose utilized, were found in the medium after 24 h of growth. The discrepancy between sulfoquinovose degraded and sulfate liberated indicated the occurrence of a major accumulation of sulfur-containing intermediate(s). The concentration of sulfate rose to about 5 mM (equivalent to 50% of initial sulfoquinovose concentration) from 9 to 55 h, corresponding to a period of little growth.

FIG. 1.

Growth of Klebsiella sp. strain ABR11 (a) and Agrobacterium sp. strain ABR2 (b) in phosphate-basal salts medium containing 10.6 mM sulfoquinovose. Symbols: filled circles, culture attenuance at D₆₀₀; hollow circles, sulfoquinovose as reducing sugar; triangles, inorganic sulfate.

The ¹³C-NMR spectrum of [U-¹³C]sulfoquinovose was identical to the natural abundance spectrum described by Roy and Hewlins (34), who gave full details of the unequivocal assignments for the chemical shifts, based on the ¹³C-¹H correlation in the two-dimensional NMR spectrum (34). After growth for 3 h in 10.6 mM [¹³C]sulfoquinovose, the ¹³C-NMR spectrum of the culture showed a small doublet at 53.7 ppm superimposed on the C-6 signal of sulfoquinovose (53.3 ppm) together with a singlet at 160 ppm, showing the presence of H¹³CO₃⁻, which subsequently persisted in all spectra. During the next 9 h, the resonances from sulfoquinovose steadily decreased while signals at 53.7, 64.7, and 68.1 ppm steadily increased. In these spectra, the resonance at 68.1 ppm was superimposed on that of C-5α of sulfoquinovose (68.7 ppm).

By 24 h the signals from sulfoquinovose had disappeared, leaving the above three resonances as the dominant ones, together with the bicarbonate signal (160 ppm) and some minor peaks. Table 1 gives the characteristics of both major and minor resonances, and Table 2 gives the chemical shifts for chemically synthesized sulfopropandiol and sulfolactate. Comparison of the major metabolite resonances with those of the standard compounds shows very close agreement with the chemical shifts for sulfopropandiol. After 24 h of growth the bacteria were removed from a sample of the culture by centrifugation. The supernatant was rotary evaporated to dryness, and the residue was exchanged with D₂O. The ¹³C-NMR spectrum was unchanged from that of the bacterial suspension, except for the disappearance of a minor singlet at 124.6 ppm. This showed that the major metabolite accumulated extracellularly while the compound responsible for the lost singlet could be either intracellular or, less probably, volatile. Moreover, the ¹H-NMR spectrum showed that the major metabolite contained five protons, and the ¹H-¹³C correlation in the DEPT-135 spectrum showed that the 53.7, 64.7, and 68.1 ¹³C resonances were associated with 2, 2, and 1 proton, respectively. These data indicate that the groups present were -CH₂-SO₃⁻, -CH₂-OH, and >CH-OH, and the only possible structure that is consistent with these groups and with the coupling constants shown in Table 1 is sulfopropandiol. The intensities of the resonances of the -CH₂SO₃⁻ groups in the 24-h spectrum of the sulfopropandiol product were comparable with those initially present for sulfoquinovose, showing a high degree of conversion.

TABLE 1.

NMR data for metabolites of [U-¹³C]sulfoquinovose produced by Klebsiella sp. strain ABR11

Major resonances			Minor resonances
13C chemical shiftsa	1H chemical shifts	Structural unit identified	13C chemical shiftsa	Structural unit identified
53.7b (d, J = 38)	2.9, 3.0	—CH2SO3− of sulfopropandiol	23.3c (d, J = 52)	CH3− of acetate
64.7b (d, J = 41)	3.5, 3.6	—CH2OH of sulfopropandiol	55.3d (d, J = 37)	—CH2SO3− of sulfolactate
68.1b (dd, J = 38, 41)	4.0	>CHOH of sulfopropandiol	69.1d (dd, J = 37, 55)	>CHOH of sulfolactate
			124.6c (s)	Unknown
160.0 (s)		HCO3−	178.7d (d, J = 55)	—COO− of sulfolactate
			181.4c (d, J = 52)	—COO− of acetate

^aChemical shifts are given in parts per million and the coupling constants, J, in hertz; d, doublet; dd, double doublet; s, singlet. Resonances are listed in increasing order of chemical shifts in the ¹³C spectrum, and signals considered to arise from a given compound are aligned vertically.

^bAccumulating up to 24 h and disappearing thereafter.

^cTraces in the 48-h spectrum.

^dSignificant between 24 and 72 h of culture.

TABLE 2.

Chemical shifts (in parts per million) from NMR spectra for natural-abundance sulfopropandiol and sulfolactate prepared by chemical synthesis

Compound	Structural unit	13C resonance	1H resonance(s)
Sulfopropandiol, HOCH2·CH(OH)·CH2SO3−	—CH2SO3−	53.6	2.9, 3.0
	—CH2OH	64.6	3.5, 3.6
	>CHOH	68.0	4.0
Sulfolactate, −OOC-CH(OH)·CH2SO3−	—CH2SO3−	55.3	2.9, 3.2
	>CHOH	69.1	4.2
	—COO−	178.7

Minor signals (55.3, 69.1, and 178.7 ppm) that were just visible in the 24-h ¹³C-spectrum of culture medium showed small increases in intensity during the following 24 h, and these changes were accompanied by a marked decrease in signal intensity from sulfopropandiol. The chemical shifts of the weak signals (Table 1) corresponded closely to those of synthetic sulfolactate (Table 2), and the coupling constants of the respective signals are entirely consistent with the assignments for the structural units of sulfolactate shown in Table 1. Although sulfolactate did not accumulate to high concentrations, the signals in the 48-h spectrum persisted to the end of the experiment (72 h). A metabolite with an identical NMR spectrum was also observed (unpublished results) as the major accumulating product during degradation of [U-¹³C]sulfoquinovose by Pseudomonas sp. strain ABR1 isolated from forest leaf mould (33). When this metabolite was recovered from the growth medium and subsequently incorporated (approximately 10 mM) in fresh basal salts medium inoculated with Klebsiella sp. strain ABR11 previously grown overnight on sulfoquinovose, there was no growth and no change in the ¹³C-NMR spectrum over a period of 48 h. This observation and the low but persistent accumulation of sulfolactate formed from sulfoquinovose in Klebsiella sp. strain ABR11 cultures indicated that sulfolactate was not degraded by this organism.

Of the remaining minor resonances in the Klebsiella sp. 24-h spectrum, the coupled (J = 52 Hz) doublets at 23.3 and 181.4 ppm (Table 2) were probably from ¹³CH₃¹³COO^−. The remaining minor singlet at 124 ppm in the 12- and 24-h spectra that was also seen with Agrobacterium sp. strain ABR2 (see below) must arise from a ¹³C-1 compound (i.e., a ¹³C atom not attached to any other ¹³C atom). This material was either intracellular or volatile because it was absent from freeze-dried culture supernatant, but its identity remains unknown.

Metabolic experiments with Agrobacterium sp. strain ABR2.

Growth of Agrobacterium sp. strain ABR2 on 10.6 mM sulfoquinovose (Fig. 1b) was significantly slower than that of Klebsiella sp. strain ABR11. However, the growth curves for Agrobacterium sp. strain ABR2 in 3 mM (data not shown) and 10.6 mM sulfoquinovose were identical for the first 7 h (after which growth ceased in the former because of substrate depletion), thus indicating that sulfoquinovose concentration was not limiting the growth rate.

Figure 1b also shows the changes in concentration of sulfoquinovose and sulfate during growth of Agrobacterium sp. strain ABR2. As with Klebsiella sp. strain ABR11, there was close correspondence between reduction in sulfoquinovose concentration (measured as reducing sugar) and growth, but sulfoquinovose was not exhausted under these conditions. That the reducing sugar remaining in the medium was indeed sulfoquinovose was shown by TLC: after growth for 24 h the only compound detected, with aniline-diphenylamine (for reducing sugar) or KMnO₄ (for all sugars), was sulfoquinovose. During growth, sulfate slowly appeared in the medium in molar concentrations less than those of the sulfoquinovose which had been metabolized; at 10 and 24 h the sulfate accounted for about 0.35 and 0.7 mol/mol, respectively, of the sulfoquinovose utilized.

No accumulations of degradation intermediates were detected for this organism by ¹³C-NMR. The utilization of [¹³C]sulfoquinovose was obvious because of the rapid appearance in the culture medium of the H¹³CO₃⁻ resonance at 160 ppm and the steady decrease in the intensities of the sulfoquinovose resonances (data not shown). There was a minor singlet at 124 ppm which was also seen with Klebsiella sp. strain ABR11.

Enzyme activities.

Crude extracts of sulfoquinovose-grown cells of Agrobacterium sp. strain ABR2 and Klebsiella sp. strain ABR11 both contained phosphofructokinase activity, with the latter having the higher specific activity (by about sixfold) when assayed with the standard method (Table 3). When fructose 6-phosphate was replaced with glucose 6-phosphate, the respective activities were slightly increased (1.42- and 1.24-fold, respectively), indicating the presence of active phosphoglucose isomerase in the crude extracts that rapidly converted glucose 6-phosphate to fructose 6-phosphate prior to the action of phosphofructokinase. When glucose 6-phosphate was replaced by sulfoquinovose, apparent sulfoquinovose kinase activity was present in each extract, with activities 1.61- and 1.71-fold higher than the activities measured with fructose 6-phosphate directly.

TABLE 3.

Phosphofructokinase activity in extracts of Agrobacterium sp. strain ABR2 and Klebsiella sp. strain ABR11 grown on sulfoquinovose

Source of enzyme	Kinase activitya (U/mg) towards
	Fructose 6-phosphate	Glucose 6-phosphate	Sulfoquinovose
Extracts of ABR2	0.36 ± 0.02	0.51 ± 0.04	0.58 ± 0.05
Extracts of ABR11	2.16 ± 0.82	2.68 ± 0.18	3.71 ± 0.06
B. stearothermophilusb	1.1 × 103	NTc	NT

^aSpecific enzyme activities were measured by coupling ADP production to NADH oxidation in a coupled enzyme assay involving pyruvate kinase and lactate dehydrogenase (see text) and are expressed as units per milligram of protein ± standard deviation based on triplicate assays.

^bPurified phosphofructokinase from B. stearothermophilus (nominal 1.6 × 10³ U/mg; Sigma-Aldrich) was used to validate the assay method.

^cNT, not tested; glucose 6-phosphate and sulfoquinovose are not substrates for the pure phosphofructokinase in the absence of phosphoglucose isomerase.

Crude extracts of Agrobacterium sp. strain ABR2 showed good glucose 6-phosphate dehydrogenase activity with either NAD⁺ or NADP⁺ but no detectable activity when glucose 6-phosphate was replaced by sulfoquinovose with either cofactor (Table 4). In contrast, Klebsiella sp. strain ABR11 showed an NADP⁺-dependent dehydrogenase active on glucose 6-phosphate but not on sulfoquinovose and an NAD⁺-dependent dehydrogenase active on sulfoquinovose but not on glucose 6-phosphate. The glucose 6-phosphate dehydrogenase from L. mesenteroides used to validate the assay (Table 4) gave high activity for glucose 6-phosphate (comparable with the manufacturer's nominal activity) but no detectable activity towards sulfoquinovose with either cofactor. The apparent K_m values for the NAD⁺-specific sulfoquinovose-dehydrogenase in crude extracts of Klebsiella sp. strain ABR11 were 0.21 ± 0.06 and 0.54 ± 0.10 mM for NAD⁺ and sulfoquinovose, respectively.

TABLE 4.

Glucose 6-phosphate dehydrogenase and sulfoquinovose dehydrogenase activities in extracts of Agrobacterium sp. strain ABR2 and Klebsiella sp. strain ABR11 grown on sulfoquinovose

Source of enzyme	Cofactor	Glucose 6-phosphate dehydrogenasea		Sulfoquinovose dehydrogenasea
		Concn in extract (U/ml)	Sp act (U/mg)	Concn in extract (U/ml)	Sp act (U/mg)
Extracts of ABR2	NAD+	4.66 ± 0.54	2.6 ± 0.3	NDb
	NADP+	18.14 ± 1.25	10.1 ± 0.7	ND
Extracts of ABR11	NAD+	ND		1.04 ± 0.02	0.26 ± 0.02
	NADP+	9.75 ± 0.33	2.44 ± 0.08	ND
L. mesenteroidesc	NAD+	1.99 ± 0.30	(1.81 ± 0.27) × 103	ND
	NADP+	1.38 ± 0.13	(1.25 ± 0.12) × 103	ND

^aEnzyme activities were measured by monitoring rates of cofactor reduction (see text) and are expressed as units per milliliter of extract and units per milligram of protein. Values are means ± standard deviations based on triplicate assays.

^bND, not detectable; i.e., measured activity was below the limit of detection of the assay (0.41 U/ml of extract, calculated as two times the standard deviation of 21 control assays lacking either enzyme or the non-cofactor substrate).

^cStock solution of enzyme from Sigma (nominally approximately 3 × 10³ U/ml and 833 U/mg), used to validate the assay method, was diluted approximately 3,000-fold before use.

DISCUSSION

The initial rapid exponential growth of Klebsiella sp. strain ABR11 was accompanied by equally rapid and complete disappearance of sulfoquinovose (Fig. 1a) but with no release of inorganic sulfate until the much slower second phase of growth (after 10 h). Therefore, cells rapidly assimilate some of the carbons of sulfoquinovose and accumulated the remainder as sulfonated intermediate(s) that were then degraded more slowly.

A hypothesis for formation of sulfonated intermediates via glycolytic reactions is depicted in Fig. 2, which shows the reactions of the upper glycolytic pathway starting from glucose 6-phosphate (reactions 1 to 7 and including the branch via dihydroxyacetone phosphate to glycerol, reactions 8 and 9) and also the corresponding reactions for sulfoquinovose. The action of aldolase on 6-deoxy-6-sulfofructose 1-phosphate (reaction 3′) would yield dihydroxyacetone phosphate and sulfolactaldehyde (sulfo-analog of glyceraldehyde 3-phosphate). Triosephosphate isomerase (reaction 4) would convert dihydroxyacetone phosphate to glyceraldehyde 3-phosphate and the onward reactions (5-7) towards pyruvate, thus enabling growth. Sulfolactaldehyde might reasonably be expected to follow a parallel route (reactions 5′ and 6′). Triosephosphate isomerase would also allow conversion of sulfolactaldehyde (reaction 4′) to 2-keto-3-hydroxypropanesulfonate (sulfo-analog of dihydroxyacetone phosphate) and then onwards (reaction 8′) towards glycerol.

FIG. 2.

Proposed metabolism of sulfoquinovose via reactions of the Embden-Meyerhoff-Parnas (glycolysis) pathway. Pathways to the left of the dotted line are for glucose 6-phosphate. Enzymes catalyzing each reaction are the following: reaction 1, phosphoglucose isomerase; reaction 2, phosphofructokinase; reaction 3, aldolase; reaction 4, triosephosphate isomerase; reaction 5, glyceraldehyde 3-phosphate dehydrogenase; reaction 6, phosphoglycerate kinase; reaction 7, phosphoglycerate isomerase; reaction 8, glycerophosphate dehydrogenase; reaction 9, glycerophosphatase. Reactions to the right of the dotted line (with primed numbers for the reaction steps) show the proposed corresponding reactions for sulfoquinovose. Dihydroxyacetone phosphate produced from C-1-C-3 of sulfoquinovose at the aldolase step (3′) provides assimilable carbon for growth. The C-4-C-6 carbon fragment bearing the sulfonate group is converted to the observed intermediates shown in bold. Pi, inorganic phosphate.

Although it is a reasonable hypothesis that the CH₂-SO₃⁻ and CH₂-O-PO₃⁻ structures are sufficiently similar to allow sulfoquinovose and its metabolites to be accommodated in the active sites of the enzymes of the pathway, it is much less likely that reactions in which the C-6 phosphate ester bond is broken will also occur for the much more stable C-SO₃⁻ bond. The C-6-phosphate bond is broken in two reactions: hydrolysis of 1-phosphoglycerol to glycerol (reaction 9) and isomerization of 3-phosphoglycerate to 2-phosphoglycerate (reaction 7). Thus, the corresponding analogs derived from sulfoquinovose, namely sulfopropandiol and sulfolactate, would be expected to accumulate (Fig. 2).

NMR data (Tables 1 and 2 and associated text) provided unequivocal evidence that sulfoquinovose was converted by Klebsiella sp. strain ABR11 predominantly to sulfopropandiol as the major degradation intermediate accumulating during the first 24 h of growth, and to sulfolactate in much smaller amounts, during the slower second phase of growth. Sulfopropandiol does not seem to have been detected hitherto in bacteria, but it occurs in diatoms (11). Sulfolactate formation accompanies sporulation of Bacillus subtilis (8, 48), but this occurs in the complete absence of sulfoquinovose, when cells sporulate in response to nutrient deprivation. B. subtilis may produce sulfolactate from cysteate, as occurs in mammals (47). Sulfolactate and sulfopropandiol were both detected in the alga Chlorella during [³⁵S]sulfate incorporation and metabolism experiments (39).

The key reaction shown in Fig. 2 is the initial conversion of sulfoquinovose to 6-deoxy-6-sulfofructose (reaction 1′), presumably catalyzed by an enzyme related or identical to 6-phosphoglucose isomerase, although nonenzymatic base-catalyzed conversions of aldoses to ketoses are also well known. Indeed, glucose 6-sulfate was converted spontaneously to fructose 6-sulfate in the presence of Tris buffer (15). For the second step (reaction 2′), heart muscle phosphofructokinase is known (27) to convert fructose 6-sulfate to 1-phosphofructose 6-sulfate and that the latter is converted to dihydroxyacetone phosphate and glyceraldehyde 3-sulfate by aldolase. If these reactions can occur with sugar 6-sulfate esters, then the analogous reactions with 6-deoxy-6-sulfosugars seem quite plausible. Table 3 shows high activity of phosphofructokinase (measured as fructose 6-phosphate-dependent formation of ADP from ATP) in cell extracts of Klebsiella sp. strain ABR11 grown on sulfoquinovose. Moreover, replacing fructose 6-phosphate with glucose 6-phosphate or sulfoquinovose in the assays produced the same activity in crude extracts (Table 3), indicating tandem operation of (i) glucose phosphate isomerase to convert aldo- to keto-isomers and (ii) the phosphofructokinase itself.

The wide specificity of aldolase (45) makes it very likely that 6-deoxy-6-sulfofructose 1-phosphate would be a substrate of this enzyme (reaction 3′), and indeed the reverse condensation reaction has been demonstrated in vitro (5). The immediate products of aldolase action on sulfoquinovose would be sulfolactaldehyde (previously identified in algal extracts [39]) and dihydroxyacetone phosphate (Fig. 2, reaction 4). Triosephosphate isomerase, needed to process the latter through the lower glycolytic pathway, could also operate in the opposite direction (reaction 4′) to convert sulfolactaldehyde to 2-keto-3-hydroxypropanesulfonate, which may then be reduced to the major metabolite sulfopropandiol via a reaction catalyzed by glycerophosphate dehydrogenase (Fig. 2, reaction 8′). Dihydroxyacetone sulfate (the sulfate ester analog of dihydroxyacetone phosphate) is reduced in vitro by glycerophosphate dehydrogenase from rabbit skeletal muscle (17). It is therefore reasonable to propose a similar reaction with the sulfonate analog, 2-keto-3-hydroxypropanesulfonate. The accompanying oxidation of NADH to NAD⁺ may serve to recycle NADH produced by glyceraldehyde 3-phosphate dehydrogenase (reaction 5) and thus allow continued oxidation of this compound in the same way that production of lactic acid or ethanol from pyruvate facilitates anaerobic fermentations in other organisms. When sulfoquinovose was exhausted after about 10 h, the accumulated sulfopropandiol was slowly depleted and sulfolactate appeared in small concentrations, presumably via a reversal of reactions 8′ and 4′ and then reactions 5′ and 6′. Sulfolactate was not utilized by this organism, so these reactions fail to release assimilable carbon. Nevertheless, they may be useful to the cell insofar as they provide energy in the form of ATP (reaction 6′) and reducing equivalents as NADH (reactions 8′ and 5′).

Glyceraldehyde 3-phosphate is also formed in the Entner-Doudoroff pathway, which is ancient (32) and widely distributed in bacteria (12), including Klebsiella (40), and bypasses the upper glycolytic pathway (13) by routing glucose 6-phosphate via gluconate 6-phosphate and 2-keto-3-deoxy-6-phosphogluconate to glyceraldehyde 3-phosphate and pyruvate. The analogous pathway with sulfoquinovose would yield sulfolactaldehyde and pyruvate. The key initiating enzyme activities (glucose 6-phosphate dehydrogenase and sulfoquinovose dehydrogenase) were both detected in extracts of Klebsiella sp. strain ABR11 (Table 4), thus opening a second possible route from sulfoquinovose to sulfolactaldehyde and, thence, to the observed metabolites. The two dehydrogenases appeared to be distinct because they use different cofactors (Table 4). Commercial glucose 6-phosphate dehydrogenase from L. mesenteroides, although highly active on glucose 6-phosphate with either NAD⁺ or NADP⁺, was inactive on sulfoquinovose with either cofactor. The molecular and catalytic characterization of the apparently unique sulfoquinovose dehydrogenase would allow determination of its genetic origins and the basis for its specificity.

Although growth and disappearance of sulfoquinovose were slower in Agrobacterium sp. strain ABR2 than in Klebsiella sp. strain ABR11, sulfate liberation was faster. The discrepancy between sulfate released and sulfoquinovose degraded was much smaller than that in Klebsiella sp. strain ABR11, indicating a much smaller accumulation of sulfonated intermediate(s). Indeed, ¹³C-labeled intermediates were not detectable by NMR during conversion of sulfoquinovose to bicarbonate in strain ABR2. Although ABR2 possessed glucose 6-phosphate dehydrogenase (Table 4), there was no significant activity towards sulfoquinovose with either NAD⁺ or NADP⁺. On the other hand, there was detectable phosphofructokinase activity in crude extracts, using either fructose 6-phosphate or glucose 6-phosphate or sulfoquinovose as substrates. Thus, the glycolytic pathway is the probable route for sulfoquinovose catabolism in this organism. The kinase activity towards sulfoquinovose was more than six times higher in ABR11 than in ABR2, which is consistent with the faster growth and substrate depletion in the former strain.

Sulfolactate, detected as a minor metabolite in this study, was a major product in Pseudomonas spp. (unpublished results), yet we found no evidence for catabolism of sulfolactate in any of our isolates. In contrast, the slow disappearance of sulfopropandiol after 24 h of culture was accompanied in Klebsiella sp. strain ABR11 by the liberation of sulfate into the medium, implying that sulfopropandiol undergoes desulfonation. Agrobacterium sp. strain ABR2 also liberated sulfate from sulfoquinovose but did so via unknown intermediates. Degradation of other primary (23, 44) and secondary (31) aliphatic sulfonates occurs by oxygen insertion at the sulfonated carbon to yield structures equivalent to sulfite adducts of carbonyl groups from which sulfite dissociates spontaneously. Such reactions would convert sulfopropandiol and sulfolactate to glyceraldehyde and glycerate, respectively, both of which are readily assimilable and unlikely to be detected. These desulfonations are potentially critical steps in the mineralization of sulfolipid sulfur, for which further studies are now needed.