Control of sulfur partitioning between primary and secondary metabolism


For correspondence (fax +44 1603 450014; e-mail


Sulfur is an essential nutrient for all organisms. Plants take up most sulfur as inorganic sulfate, reduce it and incorporate it into cysteine during primary sulfate assimilation. However, some of the sulfate is partitioned into the secondary metabolism to synthesize a variety of sulfated compounds. The two pathways of sulfate utilization branch after activation of sulfate to adenosine 5′-phosphosulfate (APS). Recently we showed that the enzyme APS kinase limits the availability of activated sulfate for the synthesis of sulfated secondary compounds in Arabidopsis. To further dissect the control of sulfur partitioning between the primary and secondary metabolism, we analysed plants in which activities of enzymes that use APS as a substrate were increased or reduced. Reduction in APS kinase activity led to reduced levels of glucosinolates as a major class of sulfated secondary metabolites and an increased concentration of thiols, products of primary reduction. However, over-expression of this gene does not affect the levels of glucosinolates. Over-expression of APS reductase had no effect on glucosinolate levels but did increase thiol levels, but neither glucosinolate nor thiol levels were affected in mutants lacking the APR2 isoform of this enzyme. Measuring the flux through sulfate assimilation using [35S]sulfate confirmed the larger flow of sulfur to primary assimilation when APS kinase activity was reduced. Thus, at least in Arabidopsis, the interplay between APS reductase and APS kinase is important for sulfur partitioning between the primary and secondary metabolism.


Sulfur is an essential nutrient that is present in all organisms in the amino acids cysteine and methionine, many co-enzymes and prosthetic groups, and various other metabolites. Plants fulfil most of their demand for sulfur by taking up inorganic sulfate from the soil. Before assimilation into bioorganic compounds, the sulfate must be activated by adenylation with ATP sulfurylase to adenosine 5′-phosphosulfate (APS). APS is a branching point of sulfate assimilation (Figure 1). For synthesis of cysteine and other sulfur-containing compounds of primary metabolism, APS is reduced by APS reductase to sulfite, which is subsequently reduced to sulfide by sulfite reductase. Sulfide is incorporated into the amino acid skeleton of O-acetylserine (OAS) to form cysteine (reviewed by Leustek et al., 2000; Kopriva, 2006; Davidian and Kopriva, 2010). The form of active sulfate used for sulfation in secondary metabolism, e.g. as the last step of synthesis of glucosinolates (Mugford et al., 2009), is 3′-phosphoadenosine 5′-phosphosulfate (PAPS), which is formed from APS by APS kinase. The enzymes responsible for the reductive steps, APS reductase and sulfite reductase, are exclusively present in the plastids (Koprivova et al., 2001; Kopriva et al., 2009), but ATP sulfurylase and APS kinase are localized in both plastids and cytosol (Rotte and Leustek, 2000; Mugford et al., 2009). OAS and cysteine synthesis occurs in plastids, the cytosol and mitochondria.

Figure 1.

 Scheme of plant sulfate assimilation.

Both primary and secondary sulfur-containing products are important for plant fitness and their ability to cope with stress (Rausch and Wachter, 2005). The tripeptide glutathione (GSH), the major thiol in plants, is involved in defence against oxidative stress, detoxification of heavy metals or xenobiotics, and also in biotic interactions (Gullner et al., 2001; Parisy et al., 2007; Rouhier et al., 2008; Foyer and Noctor, 2009). Of the secondary sulfated metabolites, the best known class are the glucosinolates that occur in the Capparales, which are important for plant interactions with herbivores and pathogens (Halkier and Gershenzon, 2006; Bednarek et al., 2009; Clay et al., 2009). Thus, coordination and regulation of sulfur fluxes to these various classes of compounds is important for understanding of the complex plant defences against pathogens. Indeed, the levels of glucosinolates and GSH are strongly altered in Arabidopsis plants lacking two isoforms of APS kinase (Mugford et al., 2009). The apk1 apk2 mutants possessed only approximately 15% of the level of foliar glucosinolates compared to wild-type Col-0 plants, while the level of GSH was increased approximately twofold in these mutants (Mugford et al., 2009, 2010). Clearly, reduction in APS kinase activity results in lower availability of PAPS, which limits synthesis of glucosinolates and other sulfated compounds (Mugford et al., 2009). On the other hand, over-expression of APS reductase leads to a very high accumulation of reduced sulfur-containing compounds, including sulfite and thiosulfate (Tsakraklides et al., 2002; Martin et al., 2005). This resulted in adverse effects in transgenic plants, including chlorosis and growth inhibition (Tsakraklides et al., 2002; Martin et al., 2005). This is not surprising as APS reductase exerts very high control over sulfate assimilation (Vauclare et al., 2002).

To further dissect the control of sulfur partitioning between primary and secondary metabolism, we over-expressed APS kinase in Arabidopsis as well as an APR-B isoform of APS reductase from the moss Physcomitrella patens (Kopriva et al., 2007). In addition, we analysed mutants with reduced levels of APS kinase (apk1 apk2) (Mugford et al., 2009) or APS reductase (apr2) (Loudet et al., 2007). Here we report the effects of modulation of these two enzymes on levels of sulfur-containing metabolites and on the flux through sulfate assimilation.


Over-expression of APS kinase in Arabidopsis

In Arabidopsis, APS kinase is present in the cytosol and the plastids, and each compartment alone is capable of producing enough PAPS for plant survival (Mugford et al., 2009, 2010). We expressed the APS kinase gene from Escherichia coli under the control of the constitutive 35S promoter and targeted the enzyme to the cytosol (cAPK) or to the plastids (tAPK). We chose the bacterial gene to avoid possible problems with gene silencing in the transgenic plants. The segregation ratio in the T3 generation for kanamycin-resistant T2 progeny was determined, five lines of each construct with a 3:1 ratio of resistant to sensitive seedlings, indicating a single insertion, were selected, and homozygous plants were recovered. The lines showed various expression levels of the transgene (Figure 2a,b). In contrast to apk1 apk2 plants, which are significantly smaller than wild-type, neither type of transgenic plants displayed marked growth or developmental phenotypes. The transgenic plants set flower at the same time as Col-0, and the seed yield and germination were not affected.

Figure 2.

 Levels of transgene mRNA in the transgenic lines. Total RNA was isolated from leaves of transgenic plants expressing (a) bacterial APS kinase in the cytosol, (b) bacterial APS kinase in plastids, and (c) the APR-B isoform of APS reductase from Physcomitrella patens targeted to plastids. Steady-state mRNA levels for bacterial APS kinase (a, b) and APR-B (c) were determined by semi-quantitative RT-PCR and compared to those of actin. The transcripts were not detectable in wild-type plants. The results are presented in arbitrary units as means ± SD from three independent RNA preparations.

Neither cAPK nor tAPK plants showed any alteration in total glucosinolate content or in the levels of individual indolic and aliphatic glucosinolates in leaves, with the exception of the cAPK5 line, in which the levels of both classes were reduced (Figure S1). Thus, although reduction in APS kinase leads to low levels of these secondary compounds, it is not possible to increase their content by over-expression of this enzyme. Similarly, over-expression of APS kinase did not consistently affect cysteine and GSH levels (Figure S2). Although the GSH content in the cAPK3 line was lower than that in Col-0 and other lines, it was slightly higher in cAPK4. Similarly, only one tAPK line, tAPK3, showed a reduction of both thiols, and GSH levels were slightly reduced in tAPK1. In Arabidopsis, APS kinase is part of the glucosinolate biosynthesis network, which also comprises APS reductase (Yatusevich et al., 2010). Disruption of APS kinase in apk1 apk2 mutants resulted in increased transcript levels for genes involved in glucosinolate synthesis, such as methylthioalkylmalate synthase 3 (MAM3), desulfoglucosinolate sulfotransferase (SOT17) or C-S lyase SUPERROOT1 (SUR1) (Figure 3a) (Mugford et al., 2009). Interestingly, the steady-state levels of these three transcripts were significantly elevated in cAPK lines, but only MAM3 and SOT17 mRNA levels were increased in tAPK plants (Figure 3).

Figure 3.

 Expression analysis of genes involved in glucosinolate synthesis. Total RNA was isolated from leaves of transgenic plants expressing (a) bacterial APS kinase in the cytosol, (b) bacterial APS kinase in plastids, and (c) the APR-B isoform of APS reductase from Physcomitrella patens targeted to plastids. Steady-state mRNA levels for MAM3, SOT17 and SUR1 were determined by semi-quantitative RT-PCR in the five transgenic lines (dark grey) and expressed relative to wild-type levels. The values for wild-type Col-0 (light grey) were set to 1. In addition, transcript levels in mutants apk1 apk2 (a) and apr2 (c) are shown in white. The results are means ± SD from three independent RNA preparations. Values marked with asterisks are significantly different from those for wild-type plants (Student’s t-test; ≤ 0.05).

The activity of APS reductase, which controls the reductive pathway of sulfate assimilation, was increased in all five cAPK lines (Figure 4). The lines expressing APS kinase in plastids exhibited increased APR activity, except the tAPK5 line that had the lowest transgene expression level. The APS reductase activity did not differ in apk1 apk2 plants compared to Col-0, while the disruption of APR2 in the apr2 mutant (Loudet et al., 2007) resulted in approximately 75% reduction in total APS reductase activity (Figure 4d). The activity of ATP sulfurylase, the enzyme supplying APS for both APS kinase and APS reductase, was not affected in any of the transgenic lines over-expressing APS kinase (Figure S3). As APS kinase activity could not be determined, because the current assay is not sensitive enough for measurements in plant extracts, the mRNA levels of the four APS kinase isoforms were compared in the transgenic lines and found to be unaffected (data not shown).

Figure 4.

 APR activity in transgenic lines. (a–d) APR activity was measured in protein extracts from leaves of transgenic plants expressing (a) bacterial APS kinase in the cytosol, (b) bacterial APS kinase in plastids, (c) the APR-B isoform of APS reductase from Physcomitrella patens targeted to plastids, and (d) mutants in the APR2 isoform of APS reductase and the APK1 and APK2 isoforms of APS kinase. (e) The activity of the APR-B isoform was determined in plants over-expressing the corresponding gene. Results are means ± SD from three independent plants. Values marked with asterisks are significantly different from those for wild-type plants (Student’s t-test; ≤ 0.05).

Over-expression of APR-B from Physcomitrella patens in Arabidopsis

Next we aimed to study the effects on sulfate assimilation of moderately increased APS reductase activity. We over-expressed the APR-B isoform of APS reductase from P. patens (Kopriva et al., 2007), and targeted it to plastids using the transit peptide from the Rubisco small subunit. We chose to express the APR-B form in order to avoid the negative effects that result from high activity (Tsakraklides et al., 2002). In addition, we wished to determine whether this enzyme is also functional in the context of flowering plants. The transgene expression was verified by RT-PCR in five homozygous tAPRb lines with single T-DNA insertions (based on segregation rates) (Figure 2). Similar to the APS kinase over-expressing lines, no growth or developmental defects were observed, and foliar glucosinolate levels were not affected in tAPRb plants (Figure S1). However, all five transgenic tAPRb lines accumulated cysteine, and the GSH content was increased in three lines (Figure 5). For comparison, the levels of cysteine and GSH were significantly higher in apk1 apk2 mutants as described previously (Mugford et al., 2009), but were not affected in the apr2 mutant. The increase in thiol content was much higher than in previous experiments (Mugford et al., 2009, 2010), probably due to use of a richer soil. Over-expression of APR-B did not affect the transcript levels of glucosinolate biosynthesis genes. The same was true for disruption of the APR2 gene, as glucosinolate levels in the apr2 mutant were unaffected (Figure 3).

Figure 5.

 Thiol levels in transgenic lines. Cysteine (a, b) and GSH (c, d) levels were determined by HPLC in leaves of transgenic plants expressing the APR-B isoform of APS reductase from Physcomitrella patens targeted to plastids (a, c), and in mutants in APR2 isoform of APS reductase and APK1 and APK2 isoforms of APS kinase (b, d). Results are means ± SD from three independent plants. Values marked with asterisks are significantly different from those for wild-type plants (Student’s t-test; ≤ 0.05).

Using specific reaction conditions, it is possible to assay the two forms of APS reductase, endogenous APR and APR-B, separately. APR-B activity was confirmed in all five lines. The assay is not completely prohibitive towards the higher plant APR, so some sulfite production was also detected in Col-0; however, the APR-B activity in all transgenic lines was significantly greater than that in Col-0 (Figure 4e). The endogenous APR activity was not affected in the transgenic plants (Figure 4c). APR-B over-expression thus increased the total APS reductase capacity only by approximately 10%. ATP sulfurylase activity was unaffected in the tAPRb transgenic plants; however, the activity was higher in apk1 apk2 and apr2 mutants (Figure S3). The mRNA levels for the four APS kinase isoforms did not differ between Col-0 and the APR-B over-expressing lines (data not shown).

Effects of manipulation of APS kinase and APS reductase on flux through sulfate assimilation

To better assess how manipulation of APS kinase and APS reductase affects sulfate assimilation, we determined the flux through the pathway using [35S]sulfate. In addition to the well characterized mutants apr2 and apk1 apk2, we chose two transgenic lines for each construct (cAPK2, cAPK5, tAPK1, tAPK3, tAPRb1 and tAPRb5) for analysis. The two APRb lines possessed the highest APR-B activity, and the APS kinase over-expressing lines showed similarly high transcript levels of the transgene. To assess the role of cytosolic APS kinase separately, we also analysed a mutant lacking the cytosolic isoform (apk3-1) (Mugford et al., 2009). To determine the effect of a reduction in the glucosinolate pool on primary sulfate assimilation, the hag1 mutant deficient in the synthesis of aliphatic glucosinolates was also included (Gigolashvili et al., 2007). Sulfate uptake was slightly but significantly reduced in tAPRb1 and cAPK5 lines but was not affected in any other transgenic line (Figure 6a). However, sulfate uptake increased twofold in the apk1 apk2 mutant. Disruption of the major APS kinase isoforms APK1 and APK2 led to a significant increase of 35S incorporation into proteins and thiols, and an approximately sixfold higher labelling of methionine (Figure S4), and therefore an approximately twofold increase in flux through the reductive part of sulfate assimilation, calculated as incorporation of 35S into cysteine, GSH, proteins and methionine (Figure 6b). Disruption of cytosolic APK3 did not affect primary sulfate assimilation. Surprisingly, over-expression of APS kinase in both cytosol and plastids increased the incorporation of 35S into thiols (Figure S4), resulting in increased flux through primary assimilation pathway (Figure 6b). Disruption of APR2 did not affect flux into the reduced sulfur pools. Even the small increase in APS reductase activity in tAPRb1 and tAPRb5 plants led to increased flux through the pathway (Figure 6b). The hag1 mutants also showed a small but significant increase in flux through primary sulfate assimilation (Figure 6b).

Figure 6.

 Flux through sulfate assimilation. Two-week-old seedlings of Col-0 and ten transgenic lines or mutants as indicated were incubated for 4 h with 0.2 mm [35S]sulfate, and incorporation of 35S into thiols (glutathione and cysteine), methionine, proteins and glucosinolates was quantified. (a) Sulfate uptake, (b) flux through primary assimilation determined as the sum of incorporation into thiols, methionine and proteins, and (c) incorporation into glucosinolates. Data are means ± SD from three biological replicates. Values marked with asterisks are significantly different (< 0.05) from those for wild-type plants.

Approximately 2% of the sulfate taken up was incorporated into glucosinolates in Col-0, compared to approximately 9% into reduced sulfur compounds. Labelling of the glucosinolate pool was surprisingly increased in the apk1 apk2 mutant despite significantly lower glucosinolate levels. On the other hand, as expected, lower 35S incorporation rate into glucosinolates was found in the hag1 plants (Figure 6c). The lines over-expressing APS kinase or APR-B did not show any consistent alterations: glucosinolate labelling was increased in tAPRb5 and slightly reduced in cAPK1 (Figure 6c).

These experiments confirmed the important role of APS reductase in the control of sulfate assimilation, and showed the importance of interplay of this enzyme with APS kinase in regulation of sulfur partitioning between primary and secondary metabolism. However, it appears that, while disruption of secondary sulfur metabolism redirects the flow of sulfur to primary assimilation, this simple regulation does not take place in the opposite direction.


Over the last few decades, sulfur nutrition has become important for crop improvement, because of a higher incidence of sulfur deficiency caused by lower atmospheric deposits of sulfur. Sulfur deficiency is connected to lower yields and quality and higher susceptibility to diseases (Rausch and Wachter, 2005; Dubuis et al., 2005). In addition, the increased acrylamide content in some food products is caused by sulfur-deficient crops (Muttucumaru et al., 2006). Thus, a better understanding of the control of sulfur metabolism is essential to improve crop sulfur nutrition, especially under conditions of low sulfur availability. An important target for improvement is sulfur use efficiency, as some crop plants, especially oilseed rape and other brassicas, have a very high demand for sulfur. Oilseed rape produces large amounts of glucosinolates, which are important for protection against herbivores and insects, but also represent an important sink for sulfur that cannot be utilized for protein synthesis and growth (Halkier and Gershenzon, 2006). Understanding of the control of sulfur partitioning between primary and secondary metabolism is thus a prerequisite to optimize sulfur fluxes in these plants and so improve sulfur use efficiency, particularly in glucosinolate-producing brassica crops.

Our previous work suggested important roles of the two enzymes utilizing APS in the control of sulfur partitioning. Several lines of evidence suggest that APS reductase is the key enzyme in the control of primary sulfate assimilation. Control flux analysis using Arabidopsis roots and inhibition of the enzyme by thiols revealed that the contribution of APS reductase to the control of sulfate reduction reaches 90% (Vauclare et al., 2002). In a recent report, we showed that APS reductase also has very high control over the flux in poplar, under many but not all conditions (Scheerer et al., 2010). Reduction in APS reductase activity in Arabidopsis, resulting either from natural variation or disruption of the APR2 isoform, leads to accumulation of sulfate (Loudet et al., 2007). On the other hand, over-expression of APS reductase results in accumulation of reduced sulfur compounds (Tsakraklides et al., 2002). Accordingly, although over-expression of APR-B resulted in a very small increase in the total capacity to reduce sulfate, it was sufficient to cause increased thiol levels (Figures 4 and 5), without any apparent negative effects on growth and development. However, whilst disruption of APR2 results in 75% reduction in total APS reductase activity and the apr2 mutants accumulate sulfate (Loudet et al., 2007), cysteine and GSH levels are not affected in the mutant. Thus, although a lower capacity for APS reduction leads to accumulation of the first metabolite in the pathway, sulfate, the levels of the end-products are under more complex control and are not decreased.

APS kinase was shown to be important for the synthesis of sulfated secondary metabolites, as disruption of two isoforms, APK1 and APK2, resulted in strongly reduced glucosinolate levels (Mugford et al., 2009, 2010). Clearly, the lower PAPS availability in apk1 apk2 mutants limited synthesis of mature glucosinolates (Mugford et al., 2009). However, the glucosinolate levels were not affected by over-expression of APS kinase. Although the availability of PAPS is limiting for glucosinolate synthesis, glucosinolate levels cannot be boosted by a higher supply of this sulfate donor. Thus, under conditions of high PAPS availability, glucosinolate synthesis appears to be limited by the rate of synthesis of the core structure. Nevertheless, over-expression of APS kinase resulted in increased transcript levels of several genes involved in glucosinolate biosynthesis. It therefore appears that PAPS or its derivatives may be part of the signalling circuit regulating synthesis of these metabolites. However, increased mRNA levels of MAM3 and SOT17 had no effect on the glucosinolate levels, indicating further levels of regulation of glucosinolate synthesis.

The analysis of apk1 apk2 mutants suggested that changes in enzyme activities in one branch of sulfate assimilation may also affect the other branch. The disruption of APS kinase led to a simultaneous reduction in glucosinolate levels and accumulation of thiols (Mugford et al., 2009). However, no effects on glucosinolates were detected in plants with higher or lower APS reductase activity. Similarly, over-expression of APS kinase had no impact on thiol levels, even though it caused an increase in APS reductase activity (Figures 4 and 5). Thus, although several reports showed a high flux control coefficient for APS reductase, the thiol levels are not always correlated with the enzyme activity, and the control of GSH levels in particular is far more complex.

The measurements of flux through sulfate assimilation revealed that the impact of the manipulation of APS reductase and APS kinase on the pathway is far greater than implied from the metabolite data. The slightly higher APS reduction capacity due to over-expression of APR-B increased the flux through the pathway as already indicated by the higher thiol levels (Figures 4–6), highlighting the role of this enzyme in flux control, in agreement with previous reports (Vauclare et al., 2002; Scheerer et al., 2010). It is important to note that APR-B is a different type of APS reductase than the one present in flowering plants, as it does not require the iron–sulfur cluster as a co-factor and lacks a glutaredoxin-like domain (Kopriva et al., 2007). Thus, it may be expected that not all levels of regulation act on the APR-B form, particularly the post-transcriptional regulation observed in Arabidopsis (Bick et al., 2001; Koprivova et al., 2008), The clear effect that introduction of this gene had on thiol levels and flux, despite only a minimal increase in total activity, indicates the importance of such fine regulation. The increased thiol levels in apk1 apk2 mutants may also be explained by increased flux through the reductive part of sulfate assimilation. Interestingly, the APS reductase activity in the apk1 apk2 mutant was not different from that of Col-0 (Figure 5; Mugford et al., 2009), so that the increase in flux must be caused by another mechanism, possibly due to increased sulfate uptake capacity (Figure 6a). This is quite unusual, given the previously determined high flux control coefficient of APS reductase, but not unique, as flux was also not correlated with the enzyme activity in several transgenic lines of poplar affected in glutathione synthesis (Scheerer et al., 2010). However, as ATP sulfurylase activity was higher in the apk1 apk2 mutant than in Col-0 (Figure S3), it is possible that, in this particular case, this enzyme has higher control over the flux than APS reductase. Indeed, mutations in ATP sulfurylase result in an accumulation of sulfate, a similar phenotype to that of apr2 mutants (Liang et al., 2010). In addition, sulfite reductase has been shown to exert high control over sulfate assimilation, as reduction of its expression results in highly compromised plant growth (Khan et al., 2010). Alternatively, it is possible that the increased flux is caused simply by greater availability of APS for reduction, enabling higher flux through the reductive branch of the pathway without a need for increased reduction capacity.

Surprisingly, the flux through primary sulfate assimilation was increased in plants over-expressing APS kinase both in plastids and the cytosol. Although no changes in metabolite levels were seen, the flux was increased, mostly due to higher labelling of the thiols. This can be explained by the stimulation of APS reductase activity in cAPK and tAPK plants. Thus, there seems to be a strong correlation between the increases in APS reductase activity and flux in tAPRb, cAPK and tAPK plants, further supporting the important role of this enzyme in control of flux through sulfate assimilation. However, there are also other mechanisms by which the flux can be increased, as is the case in apk1 apk2 or hag1 plants and some of the experiments described by Scheerer et al. (2010). In agreement with the metabolic flux control theory (Fell, 1997), the control of flux and product concentration appear to be uncoupled. For example, although increased APS reductase activity in cAPK plants leads to increased flux, it does not affect the levels of cysteine and GSH. This indicates a possible weakness in using metabolite data for analysis of regulation of metabolic pathways, and shows clearly the necessity of flux analysis for better understanding of control of plant metabolism. It should be noted, however, that the usefulness of flux analysis of the mutants and stable transgenic plants is also limited, as it represents analysis of a system that has adapted to the disturbance. To determine dynamic responses of metabolic fluxes to changes in expression of individual genes, inducible promoters would have to be used.

Another example of uncoupling of metabolite and flux data is the unexpected increase in labelling of glucosinolates in apk1 apk2 plants (Figure 6). Two factors appear to contribute to this result. Firstly, both donors of reduced sulfur for glucosinolate synthesis, GSH and methionine, are labelled to a much higher degree in apk1 apk2 than in other genotypes. Therefore, the specific activity of the sulfur used for glucosinolate synthesis is much higher, and the synthesis rate appears to be higher that it actually is. However, there is probably a much higher glucosinolate synthesis rate in apk1 apk2 plants. Because of their low levels, the turnover of glucosinolates is faster and the synthesis rate must be faster as well. The same effect was not observed in the hag1 mutant, but even here the reduction in glucosinolate labelling (25%) is lower than the reduction in total glucosinolate content (50%) (Gigolashvili et al., 2007; Sønderby et al., 2007). In addition, aliphatic glucosinolates, which accumulate to a lesser degree in hag1, contain more sulfur atoms than indolic glucosinolates and so contribute more to the labelling. Thus, even here it appears that the synthesis rate of the aliphatic glucosinolates is greater in the mutant, and does not correspond to the lower metabolite levels. Interestingly, the increased flux in the hag1 mutant corresponds to the increased incorporation of 35S into proteins and thiols in the myb51 mutant (Yatusevich et al., 2010). It therefore appears that reduced synthesis of glucosinolates, either by disruption of APS kinase or the MYB factors controlling synthesis of the aliphatic and indolic branches, triggers an increase of flux through primary assimilation. Such increase in flux is important to increase availability of the glucosinolate precursors (methionine and GSH) or to synthesize alternative sulfur-containing metabolites for protection against herbivores.

Altogether the data show clearly that both APS reductase and APS kinase are important for control of sulfate assimilation. Both enzymes are capable of affecting flux through the reductive branch of the pathway. The sulfate reduction rate is responsive to changes in APS reductase, and possibly also to the availability of the first intermediate, APS. It appears that avoiding the competition for APS can drive the synthesis of reduced sulfur compounds when levels of APS kinase are diminished, but cannot increase the production of secondary metabolites when the activity of APS reductase is low. However, this model may be limited to glucosinolate-producing Brassicacae. In any case, it appears that, although both enzymes participate in the control of partitioning between primary and secondary metabolism, the control is not a straightforward ‘push and pull’ mechanism, and requires further components, presumably involved in provision of sulfur acceptors.

Experimental Procedures

Plant material and growth conditions

The selection of apr2, apk1 apk2 and apk3 mutants has been described previously (Loudet et al., 2007; Mugford et al., 2009, 2010). Plants were grown in a controlled environment room under a short-day 10 h light/14 h dark cycle at constant temperature of 22°C, 60% relative humidity, and light intensity of 160 μmol m−2 sec−1. A minimum of four whole rosettes were harvested when the plants were 5 weeks old, and immediately frozen in liquid nitrogen. Tissue was then either freeze-dried (for glucosinolate analysis) or ground under liquid nitrogen in a pestle and mortar (for RNA extraction, enzyme activity and thiol measurements).

Over-expression of APS kinase and APS reductase

The binary vector pBINAR-TKTP (Wirtz and Hell, 2003), which allows 35S-driven expression and plastid targeting of the expressed proteins due to the presence of a transketolase plastid transit peptide, was used for the transformation. For the plastidic and cytosolic targeting of APS kinase, the gene was amplified from Escherichia coli DNA using primers ECAPKBAMF (5′-GCGGGATCCCATGGCGCTGCATGACG-3′) or ECAPKASPF (5′-GCGGGTACCATGGCGCTGCATGACG-3′), respectively, together with ECAPKHINDR (5′-GCGAAGCTTTCAGGATCTGATATATCG-3′). The PCR products were cloned into pGEM-T (Promega, and fully sequenced. The fragments were then cut with BamHI and HindIII or Asp718 and HindIII, respectively, and cloned into pBINAR-TKTP cut with the same enzymes. Use of the BamHI site allows translational fusion with the transit peptide. To express APR-B from Physcomitrella patens, the PpAPR-B fragment was amplified from an APR-B-expressing pET14 plasmid (Kopriva et al., 2007) using primers APRBBAMF (5′-GCGGGATCCCGCATCTGCAGTGCCTG-3′) and APRBHINDR (5′-CGCAAGCTTTTATGTTCTACCACCAACG-3′), cloned into pGEM-T, sequenced, and introduced into pBINAR-TKTP. The resulting binary plasmids were transformed into Agrobacterium tumefaciens GV3101 (pMP90) using the freeze–thaw method (Höfgen and Willmitzer, 1988). Arabidopsis ecotype Col-0 plants were transformed by the floral-dip method (Clough and Bent, 1998). Transgenic plants were selected on GM agar medium (Valvekens et al., 1988) containing 50 mg L−1 kanamycin sulfate. The segregation ratio in the T3 generation for kanamycin-resistant T2 progeny was determined, and five lines of each construct with a 3:1 ratio of resistant to sensitive seedlings, indicating a single insertion, were selected for further analysis.

RNA extraction and expression analysis

The expression analysis was performed as described by Koprivova et al. (2008). Total RNA was isolated from the leaf material by phenol/chloroform/isoamylalcohol (25/24/1) extraction and LiCl precipitation. Aliquots of 1 μg were reverse-transcribed using SuperScript reverse transcriptase (Invitrogen, For semi-quantitative PCR, equivalents of 40 ng of total RNA were amplified using GoTaq Flexi DNA polymerase (Promega) in 20 μl reactions with primers specific for the transgenes EcAPK and PpAPR-B. Primer sequences are given in Table S1. For standardization, the cDNA was amplified using primers derived from ACTIN 7 (At5g09810). To compare the expression levels of glucosinolate biosynthetic genes, PCR was performed using primers for MAM3, SOT17 and SUR1 (Table S1). The reactions were stopped after 26, 28 and 21 cycles for EcAPK, PpAPR-B and actin, respectively, and after 30, 28 and 25 cycles for MAM3, SOT17 and SUR1, respectively, at which time the reactions were still in exponential phase as determined from preliminary experiments. To compare the effects of the transformations on endogenous APS kinase, the mRNA levels of the four isoforms were determined in the same RNA preparations. It was not possible to measure APS kinase activity, as the foliar activity is below the detection limit of our method. Ten microlitres of the PCR products were subjected to electrophoresis on ethidium bromide-containing 1% agarose gels. The resulting band intensity on an UV transilluminator (Bio-Rad, was calculated using the Quantity One® software package (Bio-Rad).

Enzyme assays

APS reductase activity was determined as the production of [35S]sulfite, assayed as acid-volatile radioactivity formed in the presence of [35S]APS and dithioerythritol as the reductant (Koprivova et al., 2008). To measure the APR-B activity specifically, MgSO4 was replaced by 1 m KNO3, the concentration of APS was increased to 100 μm, and 10 μg E. coli thioredoxin was added (Wiedemann et al., 2007). ATP sulfurylase activity was measured as the APS- and pyrophosphate-dependent formation of ATP (Cumming et al., 2007). The protein concentrations were determined according to the Bradford method (Bradford, 1976) with bovine serum albumin as the standard.

Analysis of sulfur-containing metabolites

Analysis of cysteine and GSH levels was performed as described previously (Koprivova et al., 2008) using 20–30 mg of leaf material. Glucosinolates were extracted from 20 mg crushed freeze-dried leaf material. Quantification of intact glucosinolates was performed as described by Brown et al. (2003) and Mugford et al. (2010). The levels of the following glucosinolates were determined: 3MSOP (3-methylsulfinylpropyl glucosinolate), 4MSOB (4-methylsulfinylbutyl glucosinolate), 5MSOP (5-methylsulfinylpentyl glucosinolate), 7MSOH (7-methylsulfinylheptyl glucosinolate), 4MTB (4-methylthiobutyl glucosinolate), 8MSOO (8-methylsulfinyloctyl glucosinolate), 3OHP (3-hydroxypropyl glucosinolate), 4OHB (4-hydroxybutyl glucosinolate), 7MTH (7-methylthioheptyl glucosinolate) and 8MTO (8-methylthiooctyl glucosinolate) (all aliphatic glucosinolates), and I3M (indol-3-ylmethyl glucosinolate), 4OHI3M (4-hydroxy-indolyl-3-methyl glucosinolate), 4MOI3M (4-methoxyindol-3-methyl glucosinolate) and 1MOI3M (1-methoxyindol-3-ylmethyl glucosinolate) (all indolic glucosinolates).

Determination of flux through sulfate assimilation

The flux through sulfate assimilation was measured as incorporation of 35S from [35S]sulfate into thiols and proteins essentially as described by Kopriva et al. (1999) and Vauclare et al. (2002). The plants were grown for 14 days on vertical MS/phytagel (Sigma, plates. The plants were transferred into 24-well plates containing 2 ml of MS nutrient solution adjusted to a sulfate concentration of 0.2 mm and supplemented with 5.6 μCi [35S]sulfate (Hartmann Analytic, to a specific activity of 1900 kBq per nmol sulfate, and incubated in the light for 4 h. After incubation, the seedlings were washed three times for 1 min with 2 ml of cold non-radioactive nutrient solution, carefully blotted with paper tissue, weighed, transferred into 1.5 ml tubes, and frozen in liquid nitrogen. The plant tissue was extracted at a 1:10 w/v ratio in 0.1 m HCl. Ten microlitres of the extract were added to 1 ml Optiphase HiSafe3 scintillation cocktail (Perkin Elmer,, and the radioactivity was measured in a scintillation counter (Beckman, to determine sulfate uptake. To measure 35S incorporation into proteins, these were precipitated from 100 μl of the extract using 25 μl 100% trichloroacetic acid (TCA) as described by Kopriva et al. (1999). After 15 min on ice, the precipitate was collected by centrifugation (at 10 000 g for 10 min at 4°C), washed once in 100 μl 1% TCA, and once in 200 μl EtOH, and dissolved in 100 μl 0.1 m NaOH. The radioactivity was determined after addition of 1 ml scintillation cocktail in the scintillation counter.

To determine the radioactivity in thiols, 100 μl of the extract was mixed with 100 μl 0.1 m NaOH and 2 μl 0.1 m DTT and incubated in dark at 37°C for 15 min. Then 23 μl of 1 m Tris pH 8.0 and 10 μl of 100 mm monobromobimane was added, mixed and incubated in the dark at 37°C for 15 min. Finally, 22.5 μl of 50% acetic acid was added, mixed and centrifuged for 15 min at 10 000 g. Aliquots (200 μl) of the solution were transferred into HPLC vials. Standard thiol analysis was performed as described above with an injection volume of 100 μl. The HPLC was connected to a fraction collector, and fractions of 0.8 ml were collected in 6 ml scintillation vials. The radioactivity in these fractions was determined in the scintillation counter after addition of 2 ml scintillation solution. The same extract was used to determine the radioactivity in methionine. When resolved on an ion-exchange column (IC-PAK Anion, 4.6 × 75 mm, Waters,, methionine is eluted in the first two fractions, while thiols and sulfate are detected in fractions 14–18.

The radioactivity in glucosinolates was determined by a modification of their quantitative analysis, namely collection of the eluted fractions after sulfatase treatment into scintillation vials. Two millilitres of scintillation solution were added and radioactivity was determined in a scintillation counter.


This research was supported by the UK Biotechnology and Biological Sciences Research Council (grant number BB/D009596/1).