Riboflavin 3‘- and 5’-Sulfate, Two Novel Flavins Accumulating in the Roots of Iron-deficient Sugar Beet (Beta vulgaris)*

Roots from iron-deficient sugar beet grown in the presence of calcium carbonate exhibit a yellow color and autofluorescence typical of flavin-like compounds, whereas roots of control, iron-sufficient plants exhib- ited no yellow color and extremely low autofluorescence. The two major flavins whose accumulation is induced by iron deficiency have been shown to be different from riboflavin, FMN, and FAD by reversed-phase high performance liquid chromatography. These flavins, accounting for 82 and 15% of the total flavin concentration in deficient roots, have been shown un- equivocally to be riboflavin 3”sulfate and riboflavin 5’-sulfate, respectively, by electrospray-mass spec- trometry, inductively coupled plasma emission spectroscopy, infrared spectrometry, and ‘H nuclear mag- netic resonance. These flavin sulfates have not been standards. Normal 'H spectra were collected by using a HOD presaturation sequence. The sample (0.4 mg) was dissolved in CD,OD (0.7 ml), and the spectra was recorded at 30 "C after 1024 scans. The phase-sensitive two-dimensional proton double quantum-filtered COSY (DQCOSY) spectra were obtained with the usual sequence and phase cycle with full quadrature detection in the two dimensions by using the same sample as above with a 1.5- s repetition rate and a spectral window of 1000 Hz covering the region between 1.85 and 5.20 ppm. The spectral data were collected as 256 tl increments using 1024 points in the t2 dimension with 32 scans/ increment. The data set was filtered using a 90" sine bell function. Elemental Analysis-The isolated flavins were analyzed for total P and S content by ICP using a Perkin-Elmer P-40 apparatus. Spectral lines used were 180.731 and 213.618 nm for S and P, respectively.

Plants grown under limited iron supply may develop several different iron-acquisition mechanisms that are not expressed or underexpressed when iron supply is high. Two major strategies have been proposed in the recent literature (1,2). One of these mechanisms, Strategy 11, has been found in plant species within the Poaceae and is based mainly in the excretion of small molecules called siderophores. Once excreted to the medium surrounding the root, these siderophores may bind Fe(II1) that is subsequently made available to the plant. A second major type of mechanism, Strategy I, has been found in dicots and non-Graminaceae monocots. This latter strategy involves physiological changes at several levels, such as the * This work was supported by Grants CA 9/91 from the Consejo Asesor de Investigaci6n-Diputaci6n General de Arag6n (CONAI-DGA) and PB91-0057 from the Direccih General de Investigacibn Cientifica y Tecnica. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$$To whom correspondence should be addressed. Fax: 34-76-575620. development of increased capacities for proton excretion and for iron turbo-reducing activity, and morphological changes including the development of root hairs and transfer cells (1,2). Other physiological changes brought about by iron deficiency, such as the excretion of reducing compounds like riboflavin (Rbfl)' and phenolics, have been described in the literature and proposed to have a role in iron acquisition (3)(4)(5)(6)(7). However, the physiological significance of these released compounds has been frequently dismissed on the basis of the little reducing capacity of the total amount of the excreted substances (3,4,7). A detailed study on the nature of the flavin compounds released from iron-deficient sugar beet roots will be presented elsewhere. ' Flavin accumulation has been studied much less extensively than flavin excretion although several reports in the literature have mentioned the existence of yellow parts in the roots of plants grown under limited iron supply. The first report of this kind, to our knowledge, was by Weinstein et al. (8), which reported in 1954 that yellow and swollen root tips were characteristic symptoms of iron deficiency in sunflower. Similar observations were made later for sugar beet (9), sunflower (10, l l ) , pumpkin (12), and pepper (13). It has been generally assumed in the literature that the flavin compound causing the yellow color of the roots of iron-deficient plants was Rbfl since this was the flavin compound thought to be actively excreted to the nutrient solution (7). However, only in a few cases have flavins been extracted from the roots and analyzed, and in most cases the analysis was carried out by fluorescence techniques (6,14). In all instances the flavin being accumulated was reported to be Rbfl (6,14).
In this paper we confirm from fluorescence emission and excitation spectroscopy that the yellow color of the subapical root portions in iron-deficient plants is due to the presence of flavins. Two major flavin compounds, accounting for more than 97% of the total flavin content, are shown by HPLC to be different from Rbfl, FAD, and FMN. Furthermore, we show unequivocally by several analytical techniques (electrospray-MS, ICP, IR spectrometry, and 'H NMR) that the flavins induced by iron deficiency in sugar beet are Rbfl 3'sulfate and Rbfl 5'-sulfate, two flavins not previously found in biological systems. We hypothesize, based on the similar localization of flavins and that of iron reduction, that the accumulation of Rbfl sulfate(s) induced by iron deficiency may be an integral part of the iron-reducing system in sugar beet roots.

MATERIALS AND METHODS
Plant Culture-Sugar beet (Beta vulgaris L. cvs. Monohil and F58-554 H1) and spinach (Spinacia oleraeea L. cv. Matador) were grown in a growth chamber. Seeds were germinated and grown in vermiculite for 2 weeks. Seedlings were grown for 2 more weeks in nutrient solution (3/8-Hoagland nutrient solution with 22.4 pM iron) and then transplanted (4 plants per bucket) to 20-liter buckets containing half-Hoagland solution with 0 or 44.8 p~ iron. Spinach was grown with 0.5 p~ iron because at 0 p~ iron the roots exhibited necrosis and died quickly. Iron was added in the chelated commercial form Sequestrene 138 from Ciba-Geigy. The pH of the nutrient solution was raised with 1 mM NaOH and 1 g/liter CaCOa to simulate conditions usually found in the field that lead to iron deficiency; this treatment led to a constant pH of 7.5 throughout the 2-week growth period. Plants were grown with a photosynthetic photon flux density of 400 pE. m+. s-' at 25 "C, 80% relative humidity, and a photoperiod of 16 h light/8 h dark.
Extraction of Yellow Pigments from Roots-Root extracts were obtained from plants grown without iron for 10-12 days. Root tips exhibiting the yellow color were excised and ground, in dim light (less than 1 pE. m-*. s-') and at 4 "C, in a Teflon pestle glass homogenizer with ice-cold 0.1 M ammonium acetate, pH 6.1, 10% trichloroacetic acid, or 50 mM Tris-HC1, pH 6.8. All treatments extracted essentially all of the yellow pigment from the roots. The extract was centrifuged for 10 min at 12,000 X g. The trichloroacetic acid extracts were neutralized with 2 M phosphate buffer to pH 6.1 (15,16). A ratio of approximately 7 ml of extractant/g of fresh root was used.
Separation of Root Yellow Pigments by HPLC-Flavins were separated by HPLC on a 100 X 8-mm Waters Novapak Cla radial compression column (4-pm particle size). Samples were injected with a 20-pl loop, and mobile phases were pumped by a Waters M45 high pressure pump at a flow of 1 ml/min. Detection was made at 375 or 445 nm (absorption maxima of flavins) with a Shimadzu SPD-6AV detector. The column was equilibrated with 0.1 M ammonium acetate:methanol (pH 6.0) (95:5, by volume) for 10 min. The sample was injected into the column, and mobile phase A was pumped for another 2 min. A mixture of water:methanol (7030, by volume) was pumped for 23 min. A wash of 10 min with pure methanol was carried out before re-equilibrating the column as described above. Rbfl, FMN, and FAD standards were obtained from Sigma (Madrid, Spain).
Isolation of Yellow Pigments-For the isolation of flavins, neutralized trichloroacetic acid extracts were separated by HPLC as indicated above, and the two major flavin peaks were collected at the detector outlet. To concentrate the flavins, pooled fractions from several chromatograms were diluted with 10 volumes of water and passed through Sep-Pak C18 cartridges (Waters) pre-equilibrated with methanol. Flavins retained in the cartridges were eluted with pure methanol and dried under N2. After isolation both flavins were rechromatographed by HPLC, and their retention times were found to be identical to those of the two major peaks present in the sugar beet crude root extracts.
Molecular Absorption and Infrared and Fluorescence Spectra-Electronic absorption spectra were measured in water at room temperature from 200 to 600 nm with a Shimadzu 2101 PC computercontrolled spectrophotometer, using a 1-nm slit. Flavin concentrations were estimated from absorbances at 445 nm by assuming an t = 12.2 mM-' cm".
Infrared spectra (from 4000 to 400 cm") were obtained with a Bomem MB-120 spectrometer after 256 scans, from a pellet obtained from a mixture of 50 ng of sample and 5 pg of anhydrous KBr (infrared grade) pressed under vacuum at 15 tons. Emission and excitation fluorescence was measured from intact roots with a Perkin-Elmer (Beaconsfield, England) LS-50 computer-controlled spectrofluorometer equipped with a cuvette suitable for solid samples. Emission and excitation fluorescence from root extracts and isolated compounds dissolved in water were measured at room temperature with a Hitachi F-4500 computer-controlled spectrofluorometer. In all cases emission spectra were measured between 400 and 700 nm with an excitation beam of 450 nm (2.5-nm slit), and excitation spectra were measured between 250 and 600 nm for emission at 526 nm (5-nm slit). All spectra were acquired with approximately the same absorbance or fluorescence by adjusting the concentration of the different compounds. Spectra presented in the figures were obtained by multiplying each spectra by a different factor to facilitate direct comparison.
Electrospray-MS-Electrospray-MS spectra were obtained by using a Finnigan (San JosB, CA) TSQ 700 triple quadrupole mass spectrometer equipped with an Analytica of Branford (Branford, CT) electrospray interface. The HPLC fractions to be submitted to electrospray-MS analysis were evaporated, redissolved in MeOH, and introduced into the electrospray source at a rate of 1 pl/min using a Harvard Apparatus (Southnatick, MA) perfusion pump. Operating conditions were: electrospray voltage, 3500 V (positive ions) and 2800 V (negative ions); drying nitrogen flow, 5 ml/min; scan range, 200-1000 units; scan rate, 2 s. Spectra presented here are the average signal for 5-min acquisition. Rbfl, FMN, and FAD standards from Sigma were used for electrospray tuning and calibration.
'H NMR-One-and two-dimensional spectra were recorded on a Varian Unity 300 apparatus. Proton chemical shifts were referenced relative to tetramethylsilane standards. Normal 'H spectra were collected by using a HOD presaturation sequence. The sample (0.4 mg) was dissolved in CD,OD (0.7 ml), and the spectra was recorded at 30 "C after 1024 scans. The phase-sensitive two-dimensional proton double quantum-filtered COSY (DQCOSY) spectra were obtained with the usual sequence and phase cycle with full quadrature detection in the two dimensions by using the same sample as above with a 1.5s repetition rate and a spectral window of 1000 Hz covering the region between 1.85 and 5.20 ppm. The spectral data were collected as 256 tl increments using 1024 points in the t2 dimension with 32 scans/ increment. The data set was filtered using a 90" sine bell function.
Elemental Analysis-The isolated flavins were analyzed for total P and S content by ICP using a Perkin-Elmer P-40 apparatus. Spectral lines used were 180.731 and 213.618 nm for S and P, respectively.

Yellow Areas of Roots from Iron-deficient Sugar Beet Grown in the Presence of Calcium Carbonate Exhibit Fluorescence
Characteristics Typical of Flavins-When grown in nutrient solution with no iron added and buffered at high pH by calcium carbonate, the subapical portions of sugar beet roots developed a distinct yellow color. A complete anatomical and cytochemical study of these roots is being completed and will be presented in a separate study. Under these conditions the nutrient solution did not develop a yellow color. Conversely, when sugar beet was grown in unbuffered nutrient solutions, the medium developed a yellow color within a few days, and the only roots that exhibited yellowing were those situated above the solution level.' The yellow areas of iron-deficient sugar beet roots growing in the presence of calcium carbonate exhibited strong green fluorescence when illuminated with blue or UV light. The characteristics of this fluorescence from intact roots, which exhibited an emission peak at 523 nm (

Flavins Present in Roots of Iron-deficient Plants Are Free
Flavins-The possible presence of flavins linked noncovalently to proteins in iron-deficient sugar beet roots was investigated. Roots were extracted with 50 mM Tris-HC1, pH 8.0, and the extract was passed through a DEAE-cellulose column equilibrated with extraction buffer. A yellow band was weakly retained and then eluted with extraction buffer plus 0.1 M NaC1. A single peak exhibiting an absorbance maximum at 445 nm was recovered and concentrated by lyophilization. This peak was applied to a calibrated Sephadex G-75 superfine column pre-equilibrated with extraction buffer plus 20 mM NaCl and eluted using the same buffer. Again a single peak absorbing at 445 nm was found, with an elution volume indicating a molecular mass below 3000 Da.
Both the ex- tremely high flavin to protein ratio (approximately 30 on a w/w basis; protein measured as in Ref. 17) and the low A m / AM ratios of column eluates indicated that no significant amounts of any flavoprotein were present (not shown). Furthermore, boiling did not cause any shift in the absorption maxima. All these data indicated that most, if not all, of the flavins in the root extracts were not bound to proteins.

The Major Flavins Present in Roots of Iron-deficient Plants
Are Different from Rbfl, FMN, and FAD-Rbfl, FMN, and FAD standards can be easily separated by reverse-phase HPLC (Fig. 2 A ) . Retention times for these compounds were 19.5, 14.6, and 13.5 min, respectively. When a root extract was separated by the same method, three compounds with absorption spectra characteristic of flavins, with absorption maxima around 270, 375, and 445 nm, were found (Fig. 2B). The first and second peak (thereafter referred to as compounds X1 and X,, respectively) from root extracts exhibited retention times of 14.9 and 15.2 min. Therefore, their chromatographic mobility was intermediate between those of FMN and Rbfl. The mobility of flavin standards was found not to change when loaded along with root extracts. Compounds X1 and X, from root extracts accounted for approximately 82 f 3% and 15 & 3% of the total absorbance at 445 nm (mean & S.D. of four different extracts, each from a different batch of plants). A small peak with a mobility identical to that of Rbfl (retention time of approximately 19.5 min) was also found in the extracts, accounting for 3 $ 1% of the total absorbance at 445 nm. Similar ratios between compounds X, and Xz were found when the extracts were made with 10% trichloroacetic acid (Fig. 2B), 0.1 M ammonium acetate, pH 6.1, or 50 mM Tris-HC1, pH 6.8 (not shown). Extracta from control plants exhibited very low amounts (less than 1% of the total flavin present in iron-deficient plants) of Rbfl and FMN and traces of other unidentified compounds absorbing at 445 nm (not shown). Compounds XI and X2 never appeared in extracts from control plants.
The two major flavins (compounds X1 and X,) present in root extracts from iron-deficient plants were isolated by re- covering fractions from several chromatographic runs. The pooled fractions corresponding to these flavins were concentrated with Sep-Pak cartridges as described under "Materials and Methods." When rechromatographed, compounds XI and X2 gave distinct peaks of chromatographic mobilities identical to those present in crude root extracts (Fig. 2C). Each of the purified compounds was slightly contaminated with the other one, but this contamination was always below 0.5 and 3% for compounds XI and X*, respectively, as judged from the HPLC chromatograms.
Spectral Characteristics of Iron Deficiency-induced Flauins-The electronic absorption spectra of compounds X1 and X2 were very similar to those of Rbfl and FMN (Fig. 3A). All flavins exhibited absorption peaks at 222, 267, 373, and 445 nm. Some differences can be appreciated between flavins in the relative intensity of the two UV absorption peaks (Fig.  3A). The fluorescence emission spectra of all flavins were also remarkably similar, exhibiting a single peak at 526 nm (Fig.  3B). Different flavins also exhibited similar fluorescence excitation spectra, although some differences were apparent in the UV region of the spectra (Fig. 3C). These results illustrate how difficult it is to identify flavins based only on the electronic absorption and/or fluorescence spectral characteristics.
Identification of Flavins Present in Roots of Iron-deficient Sugar Beet-The two flavins separated from root extracts by HPLC were studied by electron impact-, thermospray-, and electrospray-MS (only electrospray-MS results are presented here). No molecular or structural information could be obtained from electron impact-MS using a solid probe introduction method, probably because of the thermolability of the compounds being analyzed. In agreement with the spectrophotometric data, initial results obtained by thermospray-MS indicated that these fractions contained compounds with a flavin-like structure, but clear molecular information could not be obtained. However, electrospray-MS, a soft ionization technique that uses no high temperature sources for ionization, provided relevant information on the molecular weight of compounds X1 and Xz.
The electrospray-MS spectra of Rbfl and FMN standards are presented in Fig. 4 -' for FMN) could be detected, indicating the presence of only one acidic proton in these compounds (Fig. 4, C and D ) .
A possible molecular weight of 456, identical to that FMN, prompted us to investigate the possible presence of P in the flavins isolated from iron-deficient sugar beet roots. However, both 31P NMR and ICP indicated the absence of any significant amount of P in both flavins. Another possibility could have been the presence of a -SO,H group, instead of a -P03H2 group, that could explain both the presence of a single acidic site and a molecular weight of 456. Differences between P and S include slightly different monoisotopic masses (0.01 units) and isotopic patterns, but these differences cannot be detected by using a low resolution quadrupole mass spectrometer. Indeed, the presence of S in the isolated compounds was confirmed by ICP emission spectroscopy, the results giving molar stoichiometries of 0.93 and 0.89 mol of S per mol of flavins X1 and X p , respectively. Infrared spectra showed a complex pattern of absorptions comparable with the Rbfl or FMN spectra with the exception of bands at 1251,1226,1066, and 1012 cm" that are fully compatible with the presence of a sulfate ester (results not shown).
The combined use of 'H NMR and DQCOSY (see Figs. 5 and 6) led to the assignment of the chemical shifts and coupling constants of the hydrogen network in the problem compounds X1 and X p (Tables I and 11). The comparison of the chemical shifts and coupling constants of these compounds with those of FMN (19), and N(3)-carboxymethyl-Rbfl (NCMR) (20) permitted elucidation of the structure of these flavins, including the position and nature of the substituent (Tables I and 11). The chemical shift of the pair of signals near 8 and 2.5 ppm and their relative integral (1:3) (Fig. 5, A and B ) were very similar to those found in Rbfl and Rbfl analogues. These signals correspond to two hydrogens at positions 6 and 9 (signals around 8 ppm) and the two methyl groups at positions 7 and 8 (signals close to 2.5 ppm) (21). Since the position of these signals is known to be very sensitive to ring substitution and redox state (21), it could be concluded that compounds XI and X2 have a ring system very similar to that of Rbfl in its oxidized state. The signals in the region 3.5-5.5 ppm come from a network of seven coupled hydrogens and correspond to the ribityl side chain. The chemical shift characteristics of compound Xz were closer to FMN than to NCMR however, X2 exhibited increased shifts for 5'(Ha) and 5'(Hb) and decreased shifts for 4'(H) and 3'(H) when compared with FMN (Table I). The coupling constants were also similar to those of FMN (Table II and Fig. 6). These data indicated that compound X p had a substituent more electronegative than a phosphate group in position 5', thus being in good agreement with the presence of a sulfate suggested by other techniques. In the case of compound X1, the chemical shifts for hydrogens in positions l', 4', and 5' were very similar to those found in NCMR a quite large increase in chemical shift appeared in the case of 3'(H) along with a smaller increase for 2'(H). This was indicative of a substituent in 3'. Furthermore, the coupling constants relating 3'(H) with the neighboring hydrogens were lower than in NCMR indicating the presence of a more electronegative group than hydroxyl (Table I1 and Fig. 6). From these data it was concluded that compounds X1 and X p were Rbfl 3"sulfate and Rbfl 5'-sulfate, respectively.
Rbfl3'-Sulfate Does Not Result from an Artifactual Transesterification During Extraction-No detectable interconversion was found between Rbfl 3"sulfate and Rbfl 5"sulfate when the lyophilized flavins were dissolved and stored for up to 24  at room temperature. Neutralization of the trichloroacetic acid flavin extract produced no detectable interconversion. Furthermore, when a known amount of Rbfl 5"sulfate was incubated with a Tris-HC1 (pH 6.8) root extract for 30 min at room temperature, no detectable decrease in Rbfl5"sulfate was found. All these data indicated that no artifactual interconversions occurred during extraction of the roots.

DISCUSSION
The data presented in this paper give new information on the characteristics of the flavin accumulation in the roots of iron-deficient sugar beet plants grown under low iron supply. We found that when sugar beet plants were grown in conditions similar to that found in the field in areas prone to result in iron deficiency, i.e. in the presence of CaC03, their subap-. ical root zones exhibited a characteristic yellow color. These yellow root zones emitted green fluorescence when illuminated with blue or UV light. The in vivo fluorescence emission and fluorescence excitation spectra of root tips and root extracts were indeed typical of a flavin compound. Furthermore, we analyzed the root extracts from these sugar beet plants by HPLC and found that the major flavins present were different from other free flavins commonly found in plants such as Rbfl, FAD, and FMN. The major flavin accumulating in sugar beet roots under iron deficiency was Rbfl 3'-sulfate, a novel flavin compound not previously reported in biological systems, which accounted for approximately 82% of the total flavin. Smaller amounts of Rbfl 5"sulfate (approximately 15% of the total flavin content) and minor amounts of Rbfl (approximately 3% of the total flavin content) were also found in root extracts. Whereas these data correspond to the sugar beet hybrid Monohil, Rbfl sulfates have also been found to accumulate in response to iron deficiency in the roots of a different sugar beet hybrid (F58-554 H1) and also in the roots of spinach, another member of the Chenopodiaceae (results not shown). Both plant species are considered as very efficient in iron acquisi-tion (9). Whether or not the accumulation of Rbfl sulfateb) is a general response of iron-efficient species to iron deficiency is currently being investigated.
Until now, the flavin enrichment in iron-deficient roots has been generally ascribed in the scientific literature to Rbfl. This was accepted by analogy to the idea, accepted so far, that the flavin compound excreted to the growing medium of several plant species grown in hydroponics under iron deficiency was Rbfl. Flavin accumulation in the roots has been mentioned for iron-efficient cultivars of a few plant species grown under low iron (9-13). In all cases the identification of the flavin as Rbfl was based on the absorption and/or the fluorescence spectra of the flavin excreted to the nutrient solution. However, both the absorption and the fluorescence spectra are practically identical for Rbfl and Rbfl sulfates (as well as for FMN and many other flavins, all of them containing an isoalloxazine ring) (see Fig. 3). Therefore, to ascertain whether or not the accumulation of Rbfl sulfate(s) is a ubiquitous response of plants to iron deficiency, a new re-evaluation of each case is necessary. HPLC has not been used previously to study the nature of the flavins accumulated in the roots of iron-deficient plants but was used in one instance to study the nature of flavins released to the medium by irondeficient plants (22). In that study one of the main components absorbing in the bIue region of the spectra was suggested to be Rbfl on the basis of the similarity of its retention time with that of pure Rbfl (22). In this context, it should be noted that when nutrient solutions containing any of the two flavins isolated in this work were aged for several days in the buckets used to grow the plants significant amounts of Rbfl were formed at the expense of the original compounds (not shown). This is the first time, to our knowledge, for a sulfated flavin to be reported in plants. Rbfl sulfates have been used as Rbfl analogs in flavin chemistry (23,24) but have never been found in biological systems. A large number of sulfated flavonoids, however, have been reported to date (25)(26)(27). Speculation suggests better solubility of the sulfated products, but very little information concerning the specific roles of sulfate esters is available (28). Other proposed functions for sulfated compounds include an alternative excretory mechanism or a storage role (27).
Our finding that Rbfl sulfates are located preferentially at the subapical root zone provides an explanation for the observation that some areas of roots from iron-deficient plants exhibit strong fluorescence. The fluorescence of iron-deficient roots has been described before in pepper (29) and cotton (30). In both cases the fluorescence was attributed to endogenous iron chelators of the phenolic acid type. Only recently Welkie et aE. (13) have indicated that Rbfl accumulated in adventitious pepper roots not submerged in the nutrient so-Iution and that this accumulation occurred in root zones similar to those reported previously to be fluorescent in the same species (29). In sugar beet roots flavin fluorescence was almost nonexisting in the root tips and very strong in the subapical root portion that is thicker and has root hairs (not shown).
To assess the possible physiological significance of the flavin accumulation in the roots of iron-deficient plants, it would be necessary to estimate the actual flavin concentrations in the root. The only reports of Rbfl concentrations in root tissues of iron-deficient plants, to our knowledge, refer to roots of sunflower (lo), tobacco (6, 14), and tomato (6).
The flavin concentration in sunflower roots was reported to be [8][9] in iron-sufficient plants and 23-79 p~ in irondeficient plants, the maximum values being obtained by keeping the pH high (10). In iron-deficient tobacco the total Rbfl  concentrations were reported to be in the range of 10-17 hmol g" fresh mass, roughly equivalent to 11-20 mM on a root water basis (6,14), whereas in iron-sufficient tobacco and in iron-deficient and iron-sufficient tomato the concentration was approximately 1-3 mM (6). However, since the maximum solubility of Rbfl in water is approximately 130-850 p~, depending on pH and other factors (18), claims of free Rbfl  ND, not data reported.
concentrations above 10 mM in root (and also leaf) tissue should be taken with caution. These values may have been caused by interferences in the fluorometric assay. Alternatively, the flavin in question may have been different from (i.e. more soluble than) Rbfl. In other plant materials, such as legume nodules, the concentration of Rbfl has been recently shown to be in the range of 60-80 nmol g" fresh weight (approximately 70-95 p~ on a nodule water basis) (31). Our first estimates indicate that the concentration of Rbfl sulfates may be at least 450 pM to 1 mM on a root water basis in the root subapical zone. This concentration could be even higher in case there was any flavin compartmentalization in the root. This relatively high concentration led us to examine the solubilities of Rbfl and Rbfl3"sulfate in water by measuring the absorbance of saturated flavin solutions. The maximum solubility found for Rbfl was 140 p~. However, with the limited amount of Rbfl3"sulfate that could be obtained from roots it was not possible to obtain saturated solutions; solutions as concentrated as 20 mM were still not saturated, thus indicating that Rbfl 3"sulfate is at least 150 times more soluble than Rbfl. This is in good agreement with the high solubility, close to 200 mM, of other Rbfl esters such as FMN (18).
Since the rates of Fe(II1) reduction by roots are 10-20 times higher in iron-deficient (high flavin content) than in ironsufficient (negligible flavin content) sugar beet (results not shown), it is tempting to postulate a possible relationship between the presence of flavin sulfates at high concentrations and reducing capacity. This possibility is supported by the fact that the concentration of flavin is very low in the root tips that have very little reducing activity and high in the subapical root portions that have root hairs and exhibit high reductase activity. However, some areas with high levels of fluorescence but without root hairs lack significant reducing capacity (not shown). Therefore, our data indicate that the presence of Rbfl sulfates is not the sole requirement for Fe(II1) reduction. The possible involvement of flavin redox intermediates in iron reduction was suggested by Cakmak et al. (32), and in some reviews speculation has been devoted to the possible roles of flavins in redox processes (1,7).
It remains to be elucidated whether flavin accumulation in the roots or flavin excretion to the medium is the true response of sugar beet and other iron-efficient plant species to iron deficiency. It is obvious that any possible plant mechanism directed toward acquiring iron, such as Strategy I, must have evolved under conditions leading to iron deficiency in nature. Under these natural conditions, for instance when calcium carbonate is present and external pH is high, flavin is accumulated in the roots and not excreted to the media. Significant flavin excretion takes place only in conditions unrelated to naturally occurring iron deficiencies, for instance when no calcium carbonate is present and the pH of the nutrient solutions decreases due to the functioning of the ATPase (7). These facts suggest that flavin accumulation may be an integral part of the response of dicots to iron deficiency.