32S/34S isotope fractionation in plant sulphur metabolism



  • Sulphur is a crucial microelement for plant metabolism, involved in response mechanisms to oxidative stress, C1 metabolism, electron transfer, secondary metabolism and post-translational peptide modifications. However, relatively little is known about sulphur transport and partitioning in plants. Although key steps of sulphur assimilation have been shown using labelling with radioactive 35S in past decades, the use of a natural tracer, allowing the examination of metabolic commitments and mass balance, is desirable. Sulphur stable isotopes (32S and 34S) have been proven to be useful for the investigation of the origin of sulphur atoms in geochemistry and for the detection of the origin of sulphur atoms. Nevertheless, their use for the study of primary sulphur metabolism has been impeded by our lack of knowledge of 32S/34S isotope fractionations and convenient methods for δ34S analyses. Here, we review documented 32S/34S isotope fractionations that may apply to sulphur metabolism, and explain how they should yield disparities amongst sulphur-containing plant compounds.


Each year, nearly 0.3 gigatons of elemental sulphur (S) are incorporated into biological organic matter of photosynthetic organisms (Ivanov, 1981). Plants absorb S from the environment, mostly in the form of sulphate (inline image) in the soil. The elemental S composition of plant matter is ≈ 0.2%, that is, 2 mmol S m−2 in most plant leaves. This represents an S assimilation flux (required to form plant matter) of c. 0.002–0.02 μmol m−2 s−1 during leaf metabolism. Such a small flux (up to 1/500th of photosynthesis only) is not negligible, however; it represents up to one-tenth of the respiratory metabolic flux (Krebs cycle activity) in the light (Tcherkez & Ribas-Carbo, 2012). Carbon skeletons required to sustain S assimilation into organic compounds are provided by glycolysis (acetyl-CoA), respiration (aspartic acid, Asp, which derives from oxaloacetate) and photorespiration (serine, Ser). Therefore, S assimilation is not inconsequential for carbon primary metabolism. Organic S is allocated to diverse compounds, such as free amino acids (cysteine, Cys; methionine, Met), proteins, cofactors (e.g. S-adenosyl-methionine, SAM; S-adenosyl-homocysteine, SAhC), antioxidants (glutathione), sulphate groups (attached to proteins or sugars), Fe–S centres and secondary metabolites (e.g. glucosinolates).

Disentangling S metabolism and biosynthetic pathways associated with S-containing compounds has involved labelling experiments with radioactive 35S since the 1970s (see Mudd & Datko, 1990; Sun et al., 1996; Fismes et al., 1999). However, there is now considerable interest in stable S isotopes (32S and 34S), as they provide a natural way to trace S atoms and metabolism without requiring costly or radioactive labelling. Taken as a whole, the natural 34S abundance (denoted as δ34S, Box 1) in total plant matter is believed to be close to that of source S, and thus δ34S values are often used to detect pollution by acidic, S-containing rainwater. Nevertheless, there is some disparity amongst δ34S values in plant organs and, furthermore, substantial 32S/34S isotope effects occur in most chemical reactions that form or break bonds involving S atoms. That is, chemical reactions involving S discriminate between 32S and 34S, so that the δ34S value of the product differs from that of the substrate. Indeed, most irreversible reactions favour the 32S isotope, thereby enriching in 34S the substrate molecules left behind. It is likely therefore that S-containing metabolites have contrasting δ34S values as a result of enzymatic isotope effects. Regrettably, isotopic measurements of δ34S values in metabolites purified from plant organs are presently scarce. However, the influence of S isotope fractionation during metabolism should be recognized: compound-specific δ34S values should be informative to the understanding of metabolic fluxes, and isotopic differences between metabolites could participate in explaining the isotopic divergence between plant organs.

Box 1. Sulphur isotope abundance and isotope effects

The 34S abundance is quantified with the delta value:

display math

where R is the 34S/32S isotope ratio. The δ34S value is thus expressed relative to the international reference material CDT (FeS (troilite) from the Canyon Diablo meteorite (in which Rstandard = 0.045005)). δ34S values are usually small and expressed in per mil (‰). Isotope effects in chemical reactions are defined as the quotient of the isotope ratio in the substrate to that in the product: α = Rsubstrate/Rproduct. Isotope effects are either kinetic (32k/34k) or thermodynamic (32Keq/34Keq). Most metabolic reactions involving S-containing compounds are irreversible and isotope effects are thus mostly kinetic. The isotope fractionation is defined as: Δ = α – 1. As Δ values are generally small (0–0.040), they are usually expressed in per mil. Rearranging the equation defining the isotope effect yields:

display math

Reactions that run to completion cannot fractionate between isotopes as they eventually consume all the substrate molecules. Therefore, irreversible reactions can fractionate only if they are not complete. The maximal value of the isotope effect is observed at the beginning of the reaction. Reversible reactions generally favour the heavy isotope (34S) when: the product involves more bonding around the S atom, or when the product is more oxidized than the substrate. At full equilibrium, hydrogen sulphide (H2S) is therefore 34S depleted compared with sulphur dioxide (SO2), which is, in turn, 34S depleted compared with sulphonate (inline image) or sulphate (inline image). For example, the equilibrium between H2S and SO2 (S exchange) favours 32S in H2S by ≈ 6‰ at 800 K (32Keq/34Keq ≈ 1.006) (Thode, 1991). This isotopic trend (the more reduced, the more 34S depleted) is hardly observed in (bio)chemical systems, however, because of irreversible reactions that fractionate between S isotopes and thus prevent isotopic equilibrium from being reached.

Here, we review 32S/34S isotope effects in key reactions and explain how they can be exploited to investigate plant S metabolism. The deliberate aim of this article is to stimulate research in S isotopes and 32S/34S fractionations, so as to improve our understanding of S fluxes and partitioning in plants.

Basics of S isotopes

There are five natural isotopes of S, 32S (≈ 95.02%), 33S (≈ 0.75%), 34S (≈ 4.21%), 35S (traces), 36S (≈ 0.02%), among which 35S is radioactive (the others being stable). As a result of the very small amount of 33S and 36S, 34S has been proven to be more useful for biochemistry (e.g. synthesis of elemental 34S0) and more measurable. Indeed, S is a microelement, and thus too low an isotope abundance would be impossible to measure with sufficient precision. However, the use of 33S for nuclear magnetic resonance (NMR) studies should be recognized (Hobo et al., 2010) despite the considerable cost of 33S standards and the intrinsic difficulty of 33S NMR analyses. Iron sulphide from the Canyon Diablo meteorite is used as a reference material to express δ34S values (Box 1). Meteoritic S appears to be the most natural reference as meteorites were formed from protosolar S from which all S-containing compounds and minerals on Earth eventually originate. In geological material, there is considerable variation in 34S abundance (for example, δ34S in coal varies between – 30% and + 30‰). That said, sedimentary sulphides are generally 34S depleted (δ34S roughly between – 50% and + 10‰). The strong 34S depletion in sulphide minerals and most organic S in marine sediments is attributed almost entirely to the process of microbial sulphate reduction (see ‘32S/34S fractionation in sulphate and sulphite assimilation'). Indeed, porewater H2S, which is strongly 34S depleted as the product of bacterial sulphate reduction, is the S source for sulphide minerals and most sedimentary organic matter, which, in turn, carry a 34S depletion. By contrast, evaporites (+ 5 to + 40‰) and seawater sulphate (≈ + 20‰) are 34S enriched (Thode, 1991), thereby counterbalancing the 34S depletion of sulphides. The S isotope composition (δ34S) of the volcanic input into the S cycle is close to 0‰, whereas sediment production buries 34S-depleted organic matter and pyrite; thus, by mass balance, the residue (sulphate, for which the main reservoir is seawater) becomes 34S enriched. In other words, δ34S values that diverge from zero in rocks indicate a sedimentary, biological or (bio)chemical origin of S, whereas volcanic, igneous rocks have a δ34S value close to 0‰. S that goes through the biogeochemical S cycle is eventually 34S depleted, as a result of isotope effects in biochemical reactions, and thus 34S depletion is considered to reflect a biological signature.

δ34S measurements

The S isotope composition is commonly measured using elemental analysis coupled to isotope ratio mass spectrometry (EA-IRMS; for technical details, see Grassineau et al., 1998), in which organic S is oxidized to SO2 and subsequently analysed by the mass spectrometer (masses 64 [32S16O2] and 66 [34S16O2 or 32S16O18O] amu). In biological material, the S content (%S) is rather low, and therefore S detection may be an issue. Recent EA-IRMS equipment alleviates this difficulty either with a high dynamic range (coping with large differences in elemental intensities) or a purge-and-trap technology (temperature-dependent trapping of SO2 by adsorption). Quite commonly, most EA-IRMS systems require between 30 ng and 3 μg of elemental S to provide a reliable δ34S value (meaning that, for ordinary plant organic matter, samples of 2–5 mg should be prepared). Recently, a very promising technology, based on gas chromatography coupled to ICP-MS (inductively coupled plasma mass spectrometry), has been applied to S-containing compounds and δ34S determination (Amrani et al., 2009). Other recent reports (Paris et al., 2013) have further shown that ICP-MS can be used to measure nanomolar quantities of dissolved sulphate in solution. When coupled with preparative high-performance liquid chromatography (HPLC), this should be a convenient way to measure δ34S values in individual S-containing compounds.

S absorption by plants

Sulphur absorption and distribution in plants involves sulphate transporters. In Arabidopsis, 14 sulphate transporters encoded by SULTR genes have been identified, among which some have a low (SULTR 2.1) or high (SULTR 1.1, SULTR 4.2) affinity for sulphate (for a review, see Hawkesford & De Kok, 2006). Some uncertainty remains as to whether sulphate absorption by roots is comparable with that of nitrate, that is, is associated with an influx larger than assimilation, thus leading to an efflux. Up to now, little evidence of a sulphate efflux from roots has been found (but see Cram, 1983), and sulphate contained by plants has been shown to be 34S enriched, suggesting that sulphate is absorbed and that (34S-enriched) excess sulphate molecules that are not utilized are retained rather than lost into the soil. The δ34S of whole-plant organic matter is rather close to that of source S (depletion of c. 1‰ only; Monaghan et al., 1999), showing a very small net fractionation against 34S during plant S acquisition, possibly explained by a fractionation of 3‰ against 34S during sulphate diffusion (Fig. 1). However, the 34S enrichment in plant sulphate suggests that S reduction by metabolism fractionates against 34S. Although the isotope effect (32k/34k, see Box 1) associated with the plant enzymes has not been measured to date, a significant isotope effect is extremely likely, as bacterial sulphate reduction is accompanied by an isotope fractionation of up to 40‰ (see ‘32S/34S fractionation in sulphate and sulphite assimilation’).

Figure 1.

Sulphur natural isotope composition (δ34S) in different organs of wheat (Triticum aestivum var. Soissons) grown in the field. The natural abundance in soil sulphate (precipitated as BaSO4 from soil samples) is also indicated (bottom). The apparent isotope enrichment (Δδ34S = δ34Sorgan – δ34Ssoil) is indicated on the right-hand side with bars. Redrawn from source data of Tea et al. (2003, 2007) and Tea (2004).

34S abundance in plant organs

There is currently little information about the organ-specific δ34S value. The most extensive study under natural conditions has been carried out on wheat, in which different plant parts have been sampled and compared with source sulphate (Fig. 1). Although there is little Δ34S fractionation at the whole-plant level (total plant organic matter isotopically close to source sulphate), roots and stems are depleted relative to soil sulphate by ≈ 2‰, and leaves and grains are enriched up to ≈ 2‰. It should be noted that the 34S enrichment in grains is also reflected in most proteins (δ34S in albumins, gliadins and globulins near + 3.5‰), with some larger values in low-molecular-weight glutenins (+ 4.6‰) (Tea et al., 2003). The relative 34S enrichment in leaf whole matter is not caused by the prevalence of (34S-enriched) sulphate amongst leaf compounds, but, rather, is the result of the natural 34S enrichment in S arriving at sink organs, simply because proteins themselves are 34S enriched. Furthermore, as high-molecular-weight glutenins are relatively less Cys-rich than low-molecular-weight glutenins (or gliadins), it is likely that Cys is naturally 34S enriched relative to Met (Cys and Met are the only two S-containing amino acids). The S content (%S) does not seem to be related to δ34S in proteins, as gliadins are much more S rich than glutenins (≈ 1.5% S relative to ≈ 0.5% on average), but are not particularly 34S enriched or depleted relative to the latter. It should be noted that the consistent pattern seen in wheat by Tea et al. (2003) differs from the results of Monaghan et al. (1999) (on the same species) which are much more variable, with no clear pattern. This discrepancy might have originated from the growth conditions (field vs hydropony, respectively), which may have influenced the root/shoot S reduction partitioning.

It is noteworthy that the plant 34S pattern of Fig. 1 is quite similar to that documented for δ15N values, with a 15N enrichment in leaves, which originates from nitrogen allocation (for a review, see Tcherkez & Hodges, 2008): nitrate reduction by roots fractionates against inline image and thus remaining, nonreduced nitrate molecules allocated to leaves are 15N enriched. It is tempting to suggest a similar mechanism for sulphate: excess 34S-enriched sulphate molecules left behind by reduction in roots would be allocated to leaves. In wheat, it has been shown with 35inline image labelling that between 80% and 95% of absorbed 35S is translocated to shoots at all growth stages and, furthermore, between 12% and 32% of 35S translocated to the shoots is then retranslocated to roots (Larsson et al., 2006). That is, between 35% and 85% of the 35S label eventually found in roots is inherited from shoots. Therefore, it is likely that, in wheat, the proportion of sulphate reduced by roots is modest, most of plant S assimilation taking place in leaves. However, in that study, organic and inorganic 35S were not analysed separately (35S abundance in whole organic matter), and thus the export of organic S-containing compounds by roots remains possible. On short 35inline image labelling (4 h) in Arabidopsis, reduced S in roots represented 30–40% of total recovered S (Kopriva et al., 1999). Furthermore, a clear activity of the key enzyme responsible for sulphate reduction (adenosine 5′-phosphosulphate (APS) reductase, APR) has been detected in roots, and has been found to represent 30–60% of shoot activity (when expressed in nmol mg protein−1 min−1). The cycling of S (i.e. the export of organic S from shoots back to roots) is probably not negligible, particularly under S-limited conditions (Cooper & Clarkson, 1989; Bell et al., 1995), and thus attenuates the isotopic difference caused by S reduction partitioning. That said, the isotopic difference of c. 4‰ between roots and leaves (Fig. 1) provides evidence for a kinetic isotope effect during S assimilation in roots, thereby enriching in 34S sulphate molecules exported to shoots. If we assume that root reduction represents 30–40% of plant S reduction, this 4‰ difference suggests that the effective isotope fractionation associated with S reduction is at least 7–11‰ (see the specific discussion in the next section).

32S/34S fractionation in sulphate and sulphite assimilation

Sulphate transfer from the soil solution to root cells is probably associated with an isotope fractionation. Such a fractionation is probably similar to that against 15N in nitrate, that is, of a few per mil (Kohl & Shearer, 1980; Mariotti et al., 1982). Calculations based on reduced mass ratios (estimating diffusion-based fractionation) suggest an isotope effect of 1.0016 (= [(1/18 + 1/96)/(1/18 + 1/98)]1/2, where 18, 96 and 98 are the masses of H2O, 32inline image and 34inline image, respectively). The modelling of sulphate absorption by bacterial cultures has predicted an inverse isotope effect of 0.997 (Brunner et al., 2005). An inverse isotope effect (against 32S) is not possible, however, as diffusion processes always mitigate against the heavy isotope.

Once inside plant cells, sulphate is first ‘activated’ to APS by ATP sulphurylase (Fig. 2). The associated isotope effect is believed to be unity (Brunner et al., 2005), and this would agree with the rather small isotope effect presumed for sulphate transfers (Cleland & Hengge, 2006). APS reduction to AMP + inline image by APR is likely to fractionate against 34S substantially. In bacteria, sulphate reduction shows a quite variable fractionation, between 2‰ and 45‰ (apparent isotope effect between 1.002 and 1.045) (Kaplan & Rittenberg, 1964; Detmers et al., 2001), and very large values (up to 60‰) have been found recently (Sim et al., 2011). In situ, bacterial sulphate reduction seems to be associated with an isotope effect of 1.020–1.030 depending on the reaction rate (Einsiedl, 2009). However, the effective isotope fractionation is probably lower, as APS does not accumulate and, furthermore, ATP sulphurylase and APR are associated physically (Cumming et al., 2007), so that sulphate is channelled to sulphite, with no possible discrimination of the substrate by APR. This channelling is believed to be required by the rather unfavourable equilibrium between ATP + inline image and APS + pyrophosphate (Keq = 10−8–10−6): the efficient withdrawal of APS by APR would thus be crucial to displace the ATP sulphurylase reaction to the right (APS production) (Sun & Leyh, 2006). That said, other enzymatic activities may use APS as a substrate, such as APS kinase which forms phospho-APS (PAPS; see also ‘Primary S assimilates (Cys, Met)’). The effective isotope fractionation of sulphite production thus depends on the proportion of APS allocated to sulphate reduction, and is thus presumably between 0 (if APS is completely used for sulphite production) and, say, 30‰ (if a very small fraction of APS is used for sulphite production). In planta, it is likely that APR does not consume APS quantitatively, as allelic variations in APR2 (encoding APR) have been shown to explain sulphate content in Arabidopsis (Loudet et al., 2007).

Figure 2.

Sulphur isotope effects (32k/34k) in the main steps of plant sulphur metabolism. Some of the values indicated here are assumed from similar reactions in vitro (see main text). Values above ‘1’ mean that the 32S isotope is favoured during the reaction. Question marks (?) indicate uncertainty about the numerical value. The possibility that excess inline image escapes the root (sulphate efflux) is shown. APS, adenosine 5′-phosphosulphate; Fd-SiR, ferredoxin-dependent sulphite reductase; homo-Cys, homocysteine; Met, methionine; Me-THF, methyl-tetrahydrofolate; PAPS, phospho-APS; SAhC, S-adenosyl-homocysteine; SAM, S-adenosyl-methionine; Ser, serine; THF, tetrahydrofolate.

Once inline image is formed, it is reduced by (ferredoxin-dependent) sulphite reductase (SiR). SiR, which evolves sulphide, is not likely to fractionate between S isotopes significantly as the cell content in sulphite is always very limited (sulphite is a chemically reactive species and Km of SiR is ≤ 10 μM; Nakayama et al., 2000), suggesting that sulphite reduction is fully committed. In addition, the catalytic cycle of SiR involves essentially irreversible steps (Crane et al., 1995, 1997), and thus no isotope fractionation is possible during the catalytic cycle. Sulphite can form H2S + inline image by (nonenzymatic) disproportionation, and this favours 34S in sulphide and 32S in sulphate (Habicht et al., 1998) (Fig. 2). This reaction tends to counterbalance the isotope fractionation against 34S during sulphate reduction to sulphide (ATP sulphurylase + APS reductase). Sulphite can further be oxidized by sulphite oxidase. Catalysis by the plant enzyme is extremely rapid, with turnover (kcat) rates at least 10 times faster than those of the animal enzyme (≈ 100 s−1 site−1; Hemann et al., 2005) with a low Km(sulphite) (10 μM; Brody & Hille, 1999), making the inline image oxidation step hardly limiting – and thus the isotope fractionation is certainly minimal.

Taken as a whole, there is a limited isotope fractionation (a few per mil only) associated with net S absorption and reduction because of (1) the lack of sulphate efflux from the pool of absorbed sulphate; (2) the large metabolic commitment of activated sulphate (APS) to its reduction; and (3) the metabolic partitioning between reactions involving S (reduction, PAPS formation, disproportionation). That said, the major S assimilates, Met and Cys, should tend to be 34S depleted compared with source sulphate. In vegetables, proteins (in which S is represented by Cys and Met) are consistently depleted by up to 3.8‰ and sulphate is enriched by up to 5‰ compared with total matter (Tanz & Schmidt, 2010). It should be kept in mind, however, that subtle changes in physiological conditions (S partitioning and the nature of prevalent S-containing compounds) may easily alter this pattern and modify the net S isotope fractionation.

Primary S assimilates (Cys, Met)

Cys is the primary S assimilate formed by sulphide fixation onto o-acetyl-Ser (OAS), by the combined action of Ser acetyl transferase and OAS thiol-lyase, which form a proteic complex. This step is believed to be critical for S assimilation as, under OAS excess, the complex dissociates and, under Cys excess, OAS thiol-lyase may act as a sulphydrase (for a review, see Hawkesford & De Kok, 2006). HS addition to OAS probably fractionates between S isotopes by 4–12‰, which is the range described in the nucleophilic addition of R–S (Kwart & Stanulonis, 1976). Cys is the precursor of Met: o-phosphohomo-Ser is combined to Cys by cystathionine γ-synthase to form cystathionine, during which an R–SH (thiol) group is transformed into R–S–R′ (thioether). This reaction is believed to be in equilibrium and thus a thermodynamic isotope fractionation is expected (perhaps near −5‰, the minus sign meaning a fractionation against 32S). The thioester group is then cleaved by the cystathionine β-lyase, giving homocysteine. Cleavage of S–R bonds fractionates by 8–18‰ (Saunders & Asperger, 1957; Friedberger & Thornton, 1976; Hargreaves et al., 1976). Homocysteine is then methylated by Met synthase to yield Met, using methyl tetrahydrofolate as a cofactor. Attaching a methyl group to the thiol S atom of homocysteine probably fractionates against 34S (like the methylation of SAhC to SAM), but the fractionation value is currently unknown and no similar reaction is documented for S isotopes. However, homocysteine levels are always very low (homocysteine is cytotoxic, including in human and animal cells) and the Met synthase reaction is certainly nearly fully committed.

Taken as a whole, Met is probably naturally 34S depleted relative to Cys (note that such a conclusion has already been reached earlier from a consideration of grain proteins; see the section on ‘34S abundance in plant organs’), but the specific δ34S difference between them is uncertain. The fractionation values quoted above (apart from Met synthase) suggest that Cys might be 34S depleted by 7–25‰ compared with Met.

Methylation cycles

Cys and Met are the main sources of S atoms found in downstream metabolites. The most important of these for cellular metabolism is SAM, which is a methyl group donor that exchanges CH3 with tetrahydrofolate. Adenosylation of Met does not fractionate between S isotopes (Markham et al., 1987), and thus SAM has the same isotope composition as Met. By contrast, SAM demethylation to SAhC fractionates against 34S by c. 13‰ (in the case of cathecol-O-methyltransferase; Rodgers et al., 1982), thereby 34S depleting SAhC. If SAM and SAhC levels were in the steady state, the net isotope effect would be negligible, that is, there would be no 34S difference between SAM and SAhC, because SAhC would be completely recycled back to SAM by methylation. However, it seems that the SAM/SAhC ratio is very large in plants (for a review, see Hanson & Roje, 2001) suggesting that SAhC instantly reforms SAM as soon as C1 plant metabolism generates SAhC during SAM-dependent methylations. In other words, SAhC might end up to be minimal and 34S depleted relative to SAM.

The metabolic relationships between SAM and SAhC are nevertheless complicated by the involvement of the S-methyl-methionine (SMM) cycle and SAhC hydrolase. SMM is formed by the SAM-dependent methylation of Met, which yields SAhC (SAM + Met → SAhC + SMM). Presumably, this reactions 34S depletes both SMM and SAhC. SMM is then recycled to Met (SMM + homocysteine → 2Met). SMM is believed to play different roles in plant metabolism, including homocysteine detoxification (Hanson & Roje, 2001), but the main function of SMM production is certainly the exchange of reduced S and methyl groups between plant organs via the phloem (Bourgis et al., 1999). SAhC hydrolase cleaves SAhC into adenosine + homocysteine. The latter may be recycled back to Met by Met synthase. SAhC hydrolysis is a thermodynamically disfavoured reaction (Keq = 5 × 10−7 M) which may only proceeds forward at extremely low levels of homocysteine. Both reactions (SMM cycle, SAhC hydrolase) probably lead to a 34S depletion in SMM and homocysteine, respectively, thereby 34S depleting further Met relative to Cys. However, the δ34S difference between SAhC and SAM remains somewhat intangible – unless clear flux patterns are known in SAM synthetase, methyltransferases, Met S-methyl transferase and SAhC hydrolase.

Cys- and Met-derived metabolites

Cys and Met are at the origin of many S-containing metabolites, including methyl sulphide (or methane thiol, CH3–SH), glutathione and glucosinolates (glucosinolates may contain both sulphate-derived and Cys-derived S). The δ34S value in glutathione is certainly identical to that in Cys because glutathione synthesis does not make or break bonds in which S atoms are involved.


Cys-derived S atoms in glucosinolates are 34S depleted by up to 6‰ compared with the molecular average and, in sinigrin, the Cys-derived S atom is 13.8‰ depleted relative to the sulphate group (Tanz & Schmidt, 2010). It is possible that a kinetic isotope fractionation occurs during the replacement of –OH groups by –SH (exchanged from Cys) during the synthesis of glucosinolates. This tends to 34S enrich Cys (increasing further the isotopic difference between Cys and Met).

Methyl sulphides

There should be a kinetic isotope fractionation of c. 10‰ during Met hydrolysis to homoserine + CH3–SH (Friedberger & Thornton, 1976) and c. 15–18‰ during Met spontaneous decomposition to 2-aminobutyrate + CH3–SH (Saunders & Asperger, 1957). In addition, there is a thermodynamic isotope fractionation against 34S of 3–5‰ during the dissociation of R–S–CH3 into RH + CH3–SH (where R is an organic backbone) (Amrani et al., 2008). That is, (di)methyl sulphide should be quite 34S depleted relative to source Met, should the dissociation be at equilibrium or not. Dimethyl sulphide emission by crops or wetland plants is estimated to be up to 0.2 g S m−2 yr−1 (Aneja et al., 1979; Aneja, 1990), that is, c. 90 pmol m−2 s−1. This flux is rather small (the average S assimilation rate is near 10 nmol m−2 s−1, that is, at least 100 times larger) and thus the potential isotope effect should be fully expressed (leading to 34S-depleted (di)methyl sulphide). To our knowledge, there is presently no δ34S determination in methyl sulphides liberated by plants, probably because of the very low emission rate which does not allow the substantial amount required for mass spectrometry 32S/34S analyses to be reached (the first δ34S measurement in dimethyl sulphide has been performed very recently on ocean water samples and reported by Amrani et al., 2012).

Hydrogen sulphide

Apart from excess H2S in S assimilation (excess sulphate reduction when S availability is large), H2S emission by plants originates from Cys cleavage by either Cys synthase (this happens when Cys is used instead of H2S as an S donor) or Cys desulphydrases (Schmidt, 2005). Therefore, emitted sulphide should inherit the isotope signature of Cys and thus be 34S depleted, for which evidence has been provided (reviewed in Trust & Fry, 1992).

Sulphated compounds and sulphonates

Plants contain many sulphated compounds, such as glucosinolates, proteins, APS and PAPS. Sulphate used by metabolism is certainly 34S enriched because of the kinetic isotope fractionation against 34S during S reduction and assimilation (see ‘32S/34S fractionation in sulphate and sulphite assimilation’). The sulphate donor of sulphotransferases is PAPS (produced by APS kinase) (Hoefgen & Hesse, 2008). There is no isotope effect in the synthesis of PAPS from APS, as the sulphate group is not involved in the reaction and only forms weak bonds with the amine groups of asparagine (Asn) and arginine (Arg) residues in APS binding (Sekulic et al., 2007). The kinetic isotope effect associated with sulphate transfer from PAPS to the various acceptors is not well documented. It seems that sulphotransferases proceed via a concerted mechanism with a single transition state that has a weakly charged SO3 moiety (Cleland & Hengge, 2006). The isotope fractionation is thus likely to be small because the reaction is essentially monodirectional (maximal commitment). Therefore, the δ34S value in sulphated metabolites is probably similar to that of substrate sulphate, and thus this should capture the natural 34S enrichment in cellular sulphate molecules. Indeed, sulphate groups in glucosinolates are systematically 34S enriched (Tanz & Schmidt, 2010).

By contrast, sulphate ester hydrolysis fractionates against 34S. Using phenyl phosphosulphate as a model for PAPS, the limiting step has been shown to be cleavage of the S–O bond, suggesting that the S isotope fractionation during sulphate group cleavage is large (Benkovic & Hevey, 1970). The acid hydrolysis of aryl sulphate monoesters has been shown to fractionate against 34S by 15–18‰ (Burlingham et al., 2003). This value is probably at the upper limit of that which would be expected for S–O cleavage, but sulphate hydrolysis in metabolism certainly tends to enrich in 34S the sulphated compounds left behind.

Most S-containing lipids do not have sulphate, but sulphonate, groups (inline image; no bridging oxygen with the carbon backbone). The major sulpholipid in plants is sulphoquinovodiacylgycerol (SQDG) (Dörmann & Hölzl, 2009). From a consideration of 35S labelling on isolated chloroplasts, SQDG has been long assumed to be synthesized directly from either APS or PAPS as a sulph(on)ate donor (Kepplinger-Sparace & Mudd, 1989). However, the biosynthetic pathway has recently been demonstrated to involve sulphite fixation to UDP-Glc, forming UDP-sulphoquinovose, which is then attached to diacylglycerol (for a recent review, see Frentzen, 2004). Sulphite originates from the reduction of APS, with a rather small isotope effect (see earlier, section on ‘32S/34S fractionation in sulphate and sulphite assimilation’), and thus is probably isotopically close to source sulphate, that is, 34S enriched. UDP-sulphoquinovose synthase probably fractionates against 34S. In the bacterial sulphonation of phosphoenolpyruvate by phosphosulpholactate synthase, a positively charged intermediate attacked by inline image (nucleophilic attack) has been hypothesized (Graham et al., 2002). This step may be accompanied by a large isotope effect as it may be rather limiting (low Km; reversibility of the formation of the charged species). Nonenzymatic desulphonation of aromatic compounds fractionates against 34S by up to 17‰ (Baliga & Bourns, 1966), showing that the CH2inline image bond requires energy to be formed and broken. It is probable therefore that sulpholipids are 34S depleted compared with source sulphite.

Polysulphide compounds and S oxidation

Polysulphide groups are relatively common in the plant metabolome, for example, in oxidized glutathione, dicysteine and lipoic acid. Although the S atoms come from their precursors (Cys) as thiol (–SH) groups, dimerization (oxidation) forms S–S bonds and fractionates between S isotopes. Disulphide bridges are formed by oxidoreduction reactions (involving, for example, NADPH as a reductant) which are mostly reversible. At full equilibrium, polysulphide groups are 34S enriched by c. 6‰ compared with H2S/HS (Amrani et al., 2006). Therefore, reduced compounds with disulphide bonds should be 34S enriched compared with their reduced counterparts, depending on their relative abundance (close to or far from redox equilibrium).

Fractionation in S incorporation from pollutants

As a result of the rather small net isotope fractionation in natural S plant nutrition, S isotopes (δ34S values) have been used to trace the origin of S atoms in lichens (Wadleigh, 2003) and plants (for a review, see Trust & Fry, 1992). Indeed, acidic rains and the deposition of particulate S influence the natural S isotope composition, because such S sources can be incorporated by metabolism. There is little isotope fractionation (≈ 1‰ at most) associated with SO2 dissolution from the gas to the liquid phase, but SO2 hydration into sulphite (inline image) fractionates by c. −11‰ (34S-enriched inline image) at full equilibrium (Eriksen, 1972). Of course, in plant cells, it is unlikely that the equilibrium is reached (Keq ≈ 105 at pH 7) because sulphite is removed efficiently by sulphite reductase and sulphite oxidase. Therefore, there is little effective fractionation during the incorporation of pollutant S. Indeed, in spruce leaves (Picea abies), the δ34S value of total organic matter approached that of deposited SO2 at large S content (%S) (Gebauer et al., 1994). It should be noted that the most abundant tropospheric S-containing gas is COS (carbonyl sulphide), which has been shown to decompose (with oxygen) to SO (sulphur monoxide) with a fractionation of 21‰ against 34S (Hattori et al., 2012), thereby depleting in 34S the pool of atmospheric S oxides and enriching in 34S the remaining COS. COS may be absorbed by plants, where it undergoes hydration to H2S; this reaction may also fractionate against 34S. However, the flux represented by this reaction is extremely small, c. 1/10 000th only of S assimilation (Stimler et al., 2010), and is thus unlikely to influence the δ34S of plant matter.

Metabolite 34S pattern

In this ongoing review, we have explained that many 32S/34S isotope fractionations occur in plant S metabolism and that this should lead to substantial δ34S differences between metabolites. A typical (assumed) δ34S pattern is presented in Fig. 3. It should be kept in mind that, apart from isotopic determinations in glucosinolates, sulphate, grain proteins and H2S (see all previous sections), there are presently few data on compound-specific δ34S values. Nevertheless, δ34S values may be of importance to better understand S metabolism. As a general rule, a fractionation arises in a metabolic pathway when the isotopically sensitive (fractionating) reaction becomes limiting, and when there are metabolic branching points (for further details, see Tcherkez et al., 2011). By contrast, fully committed pathways cannot fractionate between isotopes because of the quantitative consumption of the substrate. The small net S isotope fractionation associated with S incorporation in plant organic matter is a good example of a committed pathway in which absorbed sulphate molecules are mostly retained (reduced or redistributed), preventing any significant isotope discrimination at the plant scale (see ‘34S abundance in plant organs’). At the cellular level, however, the multiple fates of the key S assimilates, Cys and Met, and enzymatic isotope effects are such that downstream metabolites are certainly isotopically dissimilar (Fig. 3). From a consideration of the known fractionation described above, the δ34S range of metabolites (relative to absorbed sulphate) should be −35% to 0‰, with the largest presumed isotopic difference between methyl sulphides and residual sulphate. The isotope composition in key metabolites, such as Met, Cys, SAhC and SMM, is very likely to vary depending on the metabolic commitment to their biosynthesis: a large synthesis flux or a fully committed pathway would impede fractionation, so that Cys and Met would be hardly depleted relative to sulphate, and SAhC and SMM would be hardly 34S depleted relative to Met. Similarly, the isotopic offset between Met and Cys should provide clues on several metabolic reactions, such as the SMM cycle, which depletes Met in 34S. That is, the Δδ34S value (difference between Met and Cys) presumably correlates with the contribution of the SMM cycle to Met (re)formation.

Figure 3.

Presumed (theoretical) δ34S values in plant sulphur-containing metabolites. APS, adenosine 5′-phosphosulphate; Cys, cysteine; DMS, dimethylsulphide ((CH3)2S); homo-Cys, homocysteine; Met, methionine; MMS, monomethylsulphide (CH3SH); PAPS, phospho-APS; SAhC, S-adenosyl homocysteine; SAM, S-adenosyl methionine; SMM, S-methyl methionine. Values are relative to absorbed sulphate (at 0‰).


At present, despite the small number of studies, it is clear that the δ34S value in metabolites does not simply reflect the S isotope composition in source S, but is also influenced by fractionations and metabolic pathways. This situation seems rather different to that in animals, as controlled mammal feeding experiments have shown that there is minimal 32S/34S fractionation between diet and hair (although on a diet based on C4 plants, an isotope fractionation of +4‰ was observed, probably as a result of low digestibility) (Richards et al., 2003). Similarly, collagen δ34S values seem to reflect local environment δ34S values (Richards et al., 2001). Thus, the natural disparities in δ34S values amongst plant organs and metabolites may be providential to investigate S metabolism and partitioning in a noninvasive manner. Further studies on compound-specific δ34S values are thus desirable so as to describe plant S metabolic allocation patterns and to build an integrated model of S incorporation in plants. Such a model would be of practical importance to optimize S use efficiency and thus productivity or grain quality in S-limited crops, and to appreciate alterations in S metabolism in natural vegetation under a high level of pollution input.


G.T. thanks the Institut Universitaire de France for financial support.