Conversion of D-hamamelose into 2-carboxy-D-arabinitol and 2-carboxy-D-arabinitol 1-phosphate in leaves of Phaseolus vulgaris L.

[1-14C]Hamamelose (2-hydroxymethyl-D-ribose) was synthesized by reaction of ribulose 5-phosphate with potassium [14C]cyanide, catalytic hydrogenation of the resulting cyanohydrin, and dephosphorylation of the product. Its identity was established by a chromatographic comparison with hamamelose isolated from the bark of witch hazel (Hamamelis virginiana L.). Following vacuum infiltration of the [1-14C]hamamelose into leaf discs from Phaseolus vulgaris L., 14C-labeled 2carboxy-D-arabinitol (CA) and 2-carboxy-D-arabinitol 1-phosphate (CA1P) were formed, in the dark. Conversion of hamamelose to both CA and CA1P in the leaf discs was inhibited by dithiothreitol and sodium fluoride, although at high concentrations of these inhibitors conversion into CA was still evident when conversion into CA1P was totally inhibited. Wheat (Triticum aestivum L.) leaves converted hamamelose into CA without formation of CA1P. Leaves from P. vulgaris contained 68 nmol.g-1 fresh weight of hamamelose in the light and 35 nmol.g-1 fresh weight in the dark. A pathway for the biosynthesis of CA1P from Calvin cycle intermediates is proposed which includes the sequence: hamamelose --> CA --> CA1P.

activase (6) and rendered noninhibitory by a specific CA1P phosphatase, which converts it to 2-carboxy-D-arabinitol (CA) and inorganic phosphate (7)(8)(9). CA occurs in many plants and is not confined to the chloroplast (10). CA, administered to leaves through the petiole, is converted to CA1P in a subsequent period of darkness (11). However, since CA is present in species that make little CA1P, such as wheat (Triticum aestivum L.) (10,12), it may serve other purposes, apart from its role as a precursor of CA1P.
The ready conversion of CA into CA1P in the leaf tissue of some plants, taken together with the identification of CA1P phosphatase, led to the conclusion that CA1P and CA participate in a metabolic substrate cycle in vivo, remote from mainstream metabolism (11). However, Andralojc et al. (13) demonstrated that up to 8% of recently assimilated carbon could be incorporated into CA1P and CA. This indicates a considerable flux of new carbon from Calvin cycle intermediates into CA and CA1P. Of the Calvin cycle intermediates, the most obvious potential precursor is fructose 1,6-bisphosphate (FBP) which undergoes intramolecular rearrangement in vivo and in vitro to form hamamelose 2 1 ,5-bisphosphate (HBP) (14). Consistent with this suggestion is the increased incorporation of newly assimilated radiolabel into CA1P when assimilation takes place at low irradiance (13); conditions under which FBP concentrations would be relatively high (15). In leaves of an alpine primrose (Primula clusiana Tausch), HBP is metabolized to hamamelose 2 1 -phosphate, hamamelose 5-phosphate and hamamelose (14,16). Hamamelose (2-hydroxymethyl-D-ribose) is very widely distributed among plant species (17).
In this study, radiolabeled hamamelose was prepared in order to explore its metabolism in leaves. It was specifically metabolized to CA in the light and to CA and CA1P in the dark. These results therefore provide a link, through hamamelose, between CA1P (or CA) and reports of the synthesis of HBP from FBP.
Plant Material-French bean (Phaseolus vulgaris cv. Tendergreen) and wheat (Triticum aestivum cv. Alexandria) were grown in a glasshouse with supplementary lighting to ensure a photoperiod of 16 h, with a minimum photon flux density (PFD) of approximately 200 mol⅐m Ϫ2 ⅐s Ϫ1 . Leaves were harvested between 14 and 21 days after sowing.
Vacuum Infiltration of Leaf Discs/Leaf Strips-10 discs were cut from seedling leaves of P. vulgaris using a cork borer of 13-mm diameter. Stacks of 10 discs accumulated in the cork borer were transferred to the bottom of a 10-ml plastic syringe barrel. Just prior to infiltration, the plunger of the syringe was pushed into place to exert gentle pressure on the stack of discs. Syringes containing the disc samples were supported vertically in a large vacuum desiccator with their nozzles submerged in 0.5 ml of [ 14 C]hamamelose (4 ϫ 10 6 dpm; 58 mCi⅐mmol Ϫ1 ) or [ 14 C]CA (2.2 ϫ 10 6 dpm; 55.3 mCi⅐mmol Ϫ1 ) held in a small glass cup. (In some experiments, inhibitors were infiltrated along with hamamelose; see Table II.) A vacuum was applied until the substrate solution began to boil (2.3 kPa, 20°C). The desiccator was held at this vacuum for 20 s, and then air was readmitted. The solution filled the space below the plunger and entered all the leaf discs, giving complete infiltration. The discs were quickly removed and spread out in plastic Petri dishes, with their upper surfaces downward, without removing adhering substrate solution (which served to attach the discs to base of the dish). To prevent desiccation, the lids of the Petri dishes were lined with a wet filter paper. The dishes were inverted under metal halide lamps, providing a PFD of 200 mol⅐m Ϫ2 ⅐s Ϫ1 , followed by a period of darkness (as indicated) prior to freezing in liquid nitrogen, to await analysis. The samples of 10 discs weighed approximately 0.25 g, fresh weight (fw). In experiments using wheat seedlings, 1-cm leaf strips (totalling 0.25 g, fw, per sample) were treated in the same way.
Metabolite Extraction-Infiltrated, frozen leaf material was ground to a fine powder in liquid nitrogen and mixed rapidly with 1 ml of ice-cold 3.5% (v/v) trifluoroacetic acid, 0.15% (w/v) 8-hydroxyquinoline. Extract clarification, and removal of hydrophobic and cationic compounds using C 18 and Dowex 50-H ϩ columns, respectively, was as described previously (13). The treated extracts were evaporated to dryness in vacuo over NaOH pellets and anhydrous CaCl 2 , then immediately rehydrated and stored at Ϫ25°C, awaiting HPLC. Extracts to be analyzed for hamamelose were neutralized with Dowex 1 carbonate directly after the Dowex 50 treatment, before freezing.
Leaf material infiltrated with [ 14 C]CA was subsequently analyzed for [ 14 C]CA1P. In this case, CA1P was extracted as a complex with Rubisco, after which the protein was separated from low molecular weight components by gel filtration (1). Protein-bound metabolites were then released by thermal denaturation and resolved by anion exchange HPLC (13).
In spite of differences in sample size and detector, HPLC elution profiles obtained using the same column chemistry were displayed with common axes, by expressing the detector signal as a percentage of the highest signal in the same chromatogram.
Radioisotope Dilution Assay for Hamamelose-A known weight (between 0.90 and 1.50 g, fw) of leaf material from P. vulgaris was extracted as described above in 2 ml of 3.5% (v/v) trifluoroacetic acid, containing 10.5 nmol of radiolabeled hamamelose (2.8 mCi mmol Ϫ1 ). Hamamelose was purified from these extracts and the endogenous hamamelose deduced from the dilution of radiolabel, as evidenced by the altered PAD/radioactivity ratio, using a similar approach to that described for the quantitation of CA1P (12).
Lactonization of Resolved Extracts-The peak of radioactivity with an identical retention time to CA, derived from leaf material infiltrated with radiolabeled hamamelose (Fig. 3, track B), was collected and treated with an excess of Dowex 50-H ϩ . Hydrochloric acid was then added, to a final concentration of 10 mM (pH Ϸ 3), and the sample was evaporated in vacuo over NaOH pellets and anhydrous CaCl 2 . Just before subsequent HPLC analyses, the samples were rehydrated in water (Fig. 4) or 5 mM H 2 SO 4 (Fig. 6).
Phosphatase Treatment of Resolved Components-The peak of radioactivity with an identical retention time to CA1P, derived from leaf material infiltrated with radiolabeled hamamelose (Fig. 3, track B), was treated with Dowex 50-H ϩ , followed by evaporation in vacuo, over NaOH pellets and anhydrous CaCl 2 . The sample was dissolved in 0.3 ml of 10 mM TEA. Two units of alkaline phosphatase (bovine intestine, Sigma, UK) were added, and the mixture was kept at 25°C for 2 h, after which it was mixed with 0.02 ml of Dowex 50-H ϩ . The resulting solution was analyzed by HPLC (Fig. 5).
Rubisco Inhibition Assay-Acid-stable neutral plus anionic metabolites, derived from hamamelose-infiltrated leaf material, were resolved by anion exchange HPLC (Fig. 3) and collected in a series of 0.5-ml fractions as they emerged from the column. Aliquots (35 l) were incubated for 5 min in 0.5 ml of 100 mM Bicine, pH 8.2 (NaOH), 20 mM MgCl 2 , 10 mM NaHCO 3 , and 10 g of Rubisco (previously carbamylated by exposure to the same concentrations of Bicine, MgCl 2 , and NaHCO 3 , for 40 min at 37°C). Rubisco activity was then determined, following addition of 0.5 ml of 100 mM Bicine, pH 8.2 (NaOH), 20 mM MgCl 2 , 10 mM NaH 14 CO 3 (0.5 mCi mmol Ϫ1 ), and 0.66 mM ribulose 1,5-bisphosphate. The assay was stopped with 0.1 ml of 10 M formic acid after 5 min. The samples were oven-dried, and the acid-stable 14 C was determined by liquid scintillation counting.

RESULTS
Purified and Chemically Synthesized ( 14 C-Labeled) Hamamelose-Authentic hamamelose, from hamamelitannin, was resolved from common sugars by two distinct HPLC procedures. Thus, using anion exchange HPLC with 40 mM NaOH as eluent, hamamelose (retention time 16.0 min; Fig. 1, track C, peak 5) was fully resolved from galactose, glucose, fructose, and sucrose. It was also resolved from glucose, sucrose, and ribose by means of IMP HPLC, using 5 mM H 2 SO 4 as eluent (retention time 15.5 min; Fig. 2, track A). These two HPLC procedures exploit different properties of hamamelose, and so their combined use provides a powerful tool in the separation of this sugar from complex mixtures (see below).
Reduction of the products of the reaction of K 14 CN and D-ribulose 5-phosphate (see "Materials and Methods") yielded two components that were resolved by anion exchange HPLC, with retention times similar to sugar monophosphates and each accounting for 20 -25% of the detected radiolabel. Radiolabel was also incorporated into an unresolved mixture of the 5-phosphates of 2-carboxyarabinitol and 2-carboxyribitol (30 -40% of radiolabel) and an unidentified neutral component (10 -15% of label). The sugar constituent of only one of the sugar phosphate peaks had identical chromatographic properties to hamamelose, as evidenced by HPLC (Figs. 1, track B, and 2, track D) and also by TLC (not shown) and so was identified as 2-C-(hydroxymethyl)-D-ribose 5-phosphate (hamamelose 5-phosphate). The sugar liberated by phosphatase treatment of this compound was therefore [1-14 C]hamamelose (using nomenclature adopted by Beck et al. (5)) and was used in the following experiments.
Metabolism of [ 14 C]Hamamelose in Leaf Discs-Leaf discs were vacuum-infiltrated with solutions of the [ 14 C]hamamelose, illuminated (PFD 200 mol⅐m Ϫ2 ⅐s Ϫ1 ) for 2 h, then placed in total darkness for 6 h. Analysis of the acid extracts by anion-exchange HPLC at neutral pH revealed three radiolabeled peaks (Fig. 3, track B). The first radiolabeled component to emerge from the column was neutral, having the same retention time as hamamelose (approximately 2.8 min; peak 1). This was followed by a weakly anionic component (peak 2), then a strongly anionic component (peak 3), and these had retention times identical to CA (6.2 min) and CA1P (15.7 min), respectively. Each of these peaks was subjected to further analysis to confirm their identities.
The Neutral Component Is Hamamelose-Material from the neutral peak (Fig. 3, track B) was shown exclusively to be hamamelose, by means of IMP (Fig. 2, track C). This demonstrates that the hamamelose administered had not been metabolized to any other neutral component in the course of the experiment.
The Weakly Anionic Component Is CA-When the component with the same retention time as CA (Fig. 3, track B, peak 2) was dehydrated in acid and then rehydrated immediately before anion exchange HPLC (at pH 6.0) it had a shorter retention time, consistent with a neutral compound. This is illustrated in Fig. 4, which shows the elution profiles of [ 14 C]CA standards before (track A) and after (track B) such treatment, as well as of the treated product of [ 14 C]hamamelose metabolism (track C). This behavior is consistent with the formation of a lactone (which is uncharged) and would be expected if the compound were CA.
Anion exchange HPLC at pH Ͼ 12 (i.e. at or above the pK a of sugar hydroxl groups) (23), using sodium hydroxide as eluent, gives baseline resolution of CA from other weakly anionic components (Fig. 5, track A), including gluconic acid (GA), which can also form a lactone. Under these conditions, the weakly anionic compound derived from [ 14 C]hamamelose (Fig. 5, track C) runs as a single peak with the same retention time as the [ 14 C]CA standard (Fig. 5, track B).
Finally, both authentic CA and the weakly anionic product of hamamelose metabolism were dehydrated/rehydrated in dilute mineral acid (to ensure complete lactonization) prior to IMP.  Fig. 1, track A), following removal of NaOH. Amount loaded was derived from 0.11 g, fw, of leaf material. Track C (RID), further resolution of peak (from Fig. 3, track B) with the same retention time as hamamelose. Material containing 4,800 dpm was applied without further treatment. Again, the compound in question yielded a single peak (Fig. 6, track C) with the same elution profile as the CA (lactone) control (Fig. 6, track B). Furthermore, 96% of the applied radiolabel was recovered in each of the peaks (of tracks B and C). We concluded that this compound was exclusively CA.
The Strongly Anionic Component Is CA1P-The component with the same retention time as CA1P (Fig. 3, track B, peak 3) was collected as it emerged from the column, in a series of fractions, collected at 0.5-min intervals. Analysis of these fractions showed that radioactivity and ability to inhibit Rubisco coincided in three separate experiments (Fig. 7, A, B, and C), suggesting that radiolabeled CA1P had been formed. The amounts of CA1P present (60 -80 nmol⅐g Ϫ1 , fw) were estimated by reference to a standard CA1P-inhibition curve, constructed contemporaneously, and are similar to published values (12,13).
Phosphatase treatment of this compound, followed by anion exchange HPLC, yielded a single peak of radioactivity (Fig. 5, track E) with a retention time identical to that of an authentic sample of CA1P treated in the same way (track D), and of authentic CA (track B). Additionally, 97% of the radiolabel associated with the strongly anionic component was accounted for beneath the peak of track E (Fig. 5), following phosphatase treatment. This confirms the identity of the strongly anionic peak as exclusively CA1P.
Hamamelose Content of P. vulgaris Leaves-The conversion of infiltrated hamamelose exclusively into CA and CA1P by leaves of P. vulgaris strongly suggests that hamamelose is a precursor of these compounds in vivo, for which it must be present in the leaves of P. vulgaris. A combination of anionexchange and IMP HPLC of the acid-stable, neutral compounds extracted from leaf material demonstrated the presence of hamamelose and allowed an estimate of its amount. An acid extract of leaves was treated with Dowex 50-H ϩ and then neutralized using excess Dowex 1 carbonate, prior to HPLC. The supernatant was first analyzed by anion exchange HPLC at high pH, yielding several components with short retention times, followed by three large peaks corresponding to glucose, fructose and sucrose. Finally, a single, small peak emerged with a retention time identical to that of hamamelose (Fig. 1,  track A). This peak was collected and treated immediately with excess Dowex 50-H ϩ to remove Na ϩ ions and acidify. This was further resolved by IMP HPLC, revealing 2 major peaks (Fig. 2,  track B). The first to emerge had a retention time similar to sucrose (track A), while the second, larger, peak had a retention time identical to that of hamamelose and was symmetrical, indicating base line resolution from other components. We concluded that this (latter) peak was hamamelose.
A radioisotope dilution assay was conducted to measure the hamamelose content of leaves. This involved addition of 14 Clabeled hamamelose of known specific radioactivity to the leaf material at the (initial) acid extraction, followed by isolation of FIG. 4. Effect of lactonization on the retention of CA. Anion exchange HPLC with isocratic elution using 0.05 M sodium acetate, pH 6.0. Column eluate was fractionated (0.5-min intervals), and radioactivity was determined. Track A, chemically synthesized [ 14 C]CA, in open chain form (10 nmol, 11,100 dpm). Track B, chemically synthesized [ 14 C]CA, following dehydration at low pH (promoting lactonization). 9 nmol (9,400 dpm) applied. Track C, radioactive peak with the same retention time as CA (from Fig. 3, track B, peak 2) following dehydration at low pH. 8,000 dpm applied.  Fig. 3, track B, peak 2): 3,400 dpm applied. Track D (RID), 9.5 nmol (5,100 dpm) of chemically synthesized [ 14 C]CA1P, pretreated with alkaline phosphatase. Track E (RID), radioactive peak with same retention time as CA1P (from Fig. 3, track B, peak 3) following alkaline phosphatase treatment. 2,600 dpm applied.  Fig. 3. Fractions, collected at 0.5-min intervals, were analyzed for radioactivity and Rubisco inhibitory activity. Results for three separate leaf extracts are shown. Amounts applied were equivalent to 0.08 g, fw, of leaf material. the hamamelose in the extract by the two sequential HPLC fractionations. The specific radioactivity of the purified hamamelose was measured and used to determine the endogenous hamamelose. Table I shows that the hamamelose content of P. vulgaris was between 35 and 72 nmol⅐g Ϫ1 , fw. Consistent with its proposed role as a precursor of CA1P (which accumulates in the dark) the hamamelose content of dark adapted leaves was 50% lower than that of light adapted leaves (treatment 3). Transfer from normal to low irradiance (treatment 2) did not significantly alter the hamamelose content (Table I).
Reaction Sequence for Hamamelose Conversion into CA and CA1P-We have shown that hamamelose is specifically converted into both CA and CA1P in leaf discs of P. vulgaris. Similarly, [ 14 C]CA administered to leaf discs by vacuum infiltration is also converted into [ 14 C]CA1P, in a subsequent period of darkness (Fig. 8). Since no other radiolabeled compounds were detected following infiltration of [ 14 C]CA (not shown), we can conclude that CA is converted directly into CA1P in infiltrated leaf discs.
One approach to determining whether the conversion of hamamelose into CA and CA1P is sequential, and to establish a likely order of synthesis, is to find out if either CA or CA1P can be derived from hamamelose without the accompanying synthesis of either CA1P or CA, respectively. Moore and Seemann (10) reported that wheat contains substantial levels of CA, but has very little CA1P (12). We infiltrated wheat leaves with [ 14 C]hamamelose and found that it was converted exclusively into CA (Fig. 3, track C and inset). The identity of [ 14 C]CA, formed from hamamelose in wheat, was established as described above. This shows that hamamelose can be converted directly into CA in the absence of CA1P synthesis. In addition, DTT and fluoride have been shown to inhibit the conversion of CA into CA1P (11). We assessed the effect of these reagents on the conversion of hamamelose into CA and CA1P in leaf discs from P. vulgaris (Table II). Both DTT and fluoride reduce the conversion of hamamelose into CA and CA1P at both concentrations. However, at the higher concentrations of each inhibitor (50 mM DTT and 100 mM NaF), no radiolabel could be detected in CA1P, although significant amounts (1.5-3.0 ϫ 10 4 dpm g Ϫ1 , fw) still appeared in CA. In other words, in P. vulgaris as in wheat, conversion of hamamelose into CA does not require the synthesis of CA1P.
Rate and Light Dependence of Hamamelose Metabolism-Leaf discs were infiltrated with [ 14 C]hamamelose, then incubated at room temperature in the light (PFD 200 mol⅐ m Ϫ2 ⅐s Ϫ1 ) for up to 4 h, after which they were frozen in liquid nitrogen immediately and subsequently analyzed for radiolabeled products. Radiolabeled CA was apparent in illuminated leaf discs after only 30 min and was produced faster as time increased, but no [ 14 C]CA1P was detected (Fig. 9, open symbols).
After 2 h of illumination, a proportion of the leaf discs were transferred to complete darkness for the indicated times, after which the amounts of 14 C-labeled products were determined (Fig. 9, solid symbols). Synthesis of [ 14 C]CA continued at the rate which had been established in the preceding period of light, for at least 2 h, after which the rate declined. A small but significant amount of [ 14 C]CA1P was present after the first hour of darkness (Fig. 9). Although the appearance of [ 14 C]CA1P was considerably slower than the initial rate of [ 14 C]CA synthesis in the dark, it was sustained for at least 6 h. C]hamamelose and the indicated concentrations of DTT or sodium fluoride (NaF). They were then illuminated for 2 h (PFD ϭ 200 mol⅐m Ϫ2 ⅐s Ϫ1 ), followed by 6 h of darkness. Proportion of radiolabel in CA and CA1P (the only radiolabeled products detected) was then determined by HPLC analyses of the leaf extracts. Total (100%) radioactivity recovered was 1.2-1.5 ϫ 10 6 dpm g Ϫ1 , fw. ]CA was vacuum infiltrated into a series of (0.25 g, fw) leaf disc samples, which were then illuminated for 2 h, transferred to darkness for the indicated times, and then extracted, and the [ 14 C]CA1P content determined.
FIG. 9. Light and time dependence of the conversion of hamamelose into CA or CA1P. Leaf discs of P. vulgaris were infiltrated with [ 14 C]hamamelose, then incubated at room temperature in the light (PFD ϭ 200 mol⅐m Ϫ2 ⅐s Ϫ1 ) for the indicated periods, after which the radiolabeled CA (E) or CA1P (Ⅺ) was determined. After 2 h in the light, a proportion of the discs were transferred to darkness for the indicated periods, after which the amounts of 14 C-labeled CA (q) and CA1P (f) were determined. the amounts of FBP in the chloroplast may affect the flux of photosynthate into CA and/or CA1P. Sassenrath-Cole and Pearcy (15) have shown that FBPase responds rapidly to changes in light intensity, and that this is reflected in the amounts of fructose and sedoheptulose bisphosphates. In particular, increased amounts of these two metabolites were shown to accompany a fall in light intensity (15). These observations are consistent with the greater incorporation of recently assimilated carbon into CA and CA1P at low PFD than at high PFD, reported previously (13). Hence, changes in the ambient concentration of FBP may well affect the amount of recently assimilated carbon incorporated into CA and CA1P. If HBP is synthesized in the chloroplast, then its structural similarity to the transition state intermediate of the carboxylation reaction of Rubisco would suggest that it, too, would inhibit Rubisco. Indeed, it may prove to be the daytime inhibitor reported previously (25). This may be another reason why Rubisco activase is necessary, ensuring that Rubisco does not become irreversibly inhibited during the day.
It is puzzling that, in the light, there is a net flow of carbon away from CA1P (dephosphorylation to CA) while de novo precursors of CA1P are simultaneously being synthesized. CA1P, dephosphorylated to CA in the light, should provide sufficient carbon skeletons (along with the relatively large pool of vacuolar CA) (10) to meet the demand for rephosphorylation to CA1P, in an ensuing period of darkness. Perhaps these branch chain sugars serve some other purpose as well, for which a regular supply from recently assimilated carbon is needed.