If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Institute of Metabolism and Systems Research, College of Medical and Dental Sciences, University of Birmingham, Birmingham B15 2TT, United KingdomCentre for Endocrinology, Diabetes and Metabolism (CEDAM), Birmingham Health Partners, Birmingham B15 2TH, United Kingdom
Sulfur is present in the amino acids cysteine and methionine and in a large range of essential coenzymes and cofactors and is therefore essential for all organisms. It is also a constituent of sulfate esters in proteins, carbohydrates, and numerous cellular metabolites. The sulfation and desulfation reactions modifying a variety of different substrates are commonly known as sulfation pathways. Although relatively little is known about the function of most sulfated metabolites, the synthesis of activated sulfate used in sulfation pathways is essential in both animal and plant kingdoms. In humans, mutations in the genes encoding the sulfation pathway enzymes underlie a number of developmental aberrations, and in flies and worms, their loss-of-function is fatal. In plants, a lower capacity for synthesizing activated sulfate for sulfation reactions results in dwarfism, and a complete loss of activated sulfate synthesis is also lethal. Here, we review the similarities and differences in sulfation pathways and associated processes in animals and plants, and we point out how they diverge from bacteria and yeast. We highlight the open questions concerning localization, regulation, and importance of sulfation pathways in both kingdoms and the ways in which findings from these “red” and “green” experimental systems may help reciprocally address questions specific to each of the systems.
Sulfur (S) is an essential nutrient for all life forms. It is present in a plethora of metabolites of primary and secondary metabolism, most prominently in the amino acids cysteine and methionine, and cofactors such as iron–sulfur clusters, lipoic acid, and CoA. In the majority of these metabolites, sulfur is present in its reduced form of organic thiols; however, some compounds contain S in its oxidized form of sulfate (
is formed from sulfate by two ATP-dependent steps: adenylation, i.e. the transfer of the AMP moiety of ATP to sulfate to form adenosine 5′-phosphosulfate (APS) by ATP sulfurylase (ATPS), and the phosphorylation of APS at its 3′-OH group by APS kinase. The two enzymes are either fused into a single enzyme PAPS synthase (PAPSS) in the animal kingdom or occur as independent proteins in the green lineage (
). The by-product of PAPS-dependent sulfation reactions, 3′-phosphoadenosine 5-phosphate (PAP), is finally dephosphorylated to AMP by 3′-nucleotidases. This reaction to remove PAP is important beyond the sulfation pathways, as PAP accumulation has many additional physiological effects (
). Particularly fungi and some bacteria require PAPS for sulfate reduction and synthesis of cysteine. In these organisms, the activated sulfate in PAPS is reduced to sulfite by PAPS reductase, and after further reduction to sulfide, it is incorporated into cysteine (
). The green lineage as well as a large number of bacterial taxa, however, use APS for sulfate reduction by APS reductase, whereas Metazoa do not possess the ability to reduce sulfate and are dependent on sulfur-containing amino acids in their diet (
) wrote that sulfate “is of limited use to higher organisms except for sulfation and detoxification reactions,” without any further discussion of the topic. Since then, things have dramatically changed with growing evidence of the importance of sulfation pathways in both kingdoms. In addition, convergent findings in the green (plant) and red (animal) biochemistry of sulfur, e.g. recognition of hydrogen sulfide as a gaseous signal (
), revealed the value of comparative analysis of the same pathways in very different models. Here, we compare the mechanisms the two lineages, red and green, evolved to perform and control sulfation pathways. Given their importance for the metabolism of specific compounds and for the general sulfur metabolism, we extend the scope of our comparison to the enzymes providing the active sulfate and removing the by-product PAP. We aim to identify open questions common to both humans and plants as well as questions where knowledge from one lineage might be useful to inform research in the other.
Activation of sulfate to PAPS
The organification or activation of sulfate to PAPS by ATPS and APS kinase initiates sulfation pathways (
Differential subcellular localization and expression of ATP sulfurylase and 5′-adenylylsulfate reductase during ontogenesis of Arabidopsis leaves indicates that cytosolic and plastid forms of ATP sulfurylase may have specialized functions.
), whereas all four Arabidopsis isoforms possess N-terminal chloroplast-targeting peptides. Cytosolic activity is caused by alternative translation of the ATPS2 transcript, producing two different proteins: one with the target peptide transported into the chloroplast and one without the peptide located in the cytosol (
). However, given the major role of plastids for synthesis of PAPS and the presence of PAPS transporters in plastid envelopes, the role of cytosolic ATPS is not obvious. Interestingly, both human PAPS synthases are also regulated on the level of cellular localization of the enzyme between the nucleus and cytosol, even though a function of PAPS in the nucleus is completely unknown (
Because of its position at the beginning of the pathway, ATPS is a good candidate for controlling sulfate assimilation. Indeed, early findings in Brassica napus showed that ATPS activity and transcript levels were down-regulated by downstream products of sulfate assimilation, cysteine and GSH, and were up-regulated by sulfate starvation (
). However, the subsequent enzyme in the primary sulfate assimilation pathway in plants, the APS reductase, is regulated more strongly and was shown by metabolic flux control analysis to be the major control point of the pathway (
Sulphur flux through the sulphate assimilation pathway is differently controlled by adenosine 5′-phosphosulphate reductase under stress and in transgenic poplar plants overexpressing gamma-ECS, SO, or APR.
). Glucosinolates are part of the plant immune response to pathogens and herbivores as well as plant natural products responsible for smell, taste, and health effects of cruciferous vegetables, but also anti-nutrients for animal feed (
Although essential and sufficient for sulfate reduction, ATPS has to be coupled with the APS kinase for sulfation pathways. This enzyme, ubiquitous in nature and highly conserved in structure and sequence, shows the same localization in plants as ATPS. Arabidopsis possesses four APS kinase genes, which encode three plastidic and one (APK3) cytosolic isoform (
). APS kinase phosphorylates APS produced by ATPS and thus competes with APS reductase for this substrate. The two enzymes represent entries into the two branches of sulfate assimilation: a primary reductive assimilation pathway and a secondary oxidized sulfur metabolism involving sulfation pathways (
). Interestingly, it is the loss of two plastidic APS kinase isoforms APK1 and APK2 that results in strongly reduced accumulation of sulfated metabolites, such as glucosinolates, and not the disruption of the cytosolic enzyme APK3 (
). This, on the one hand, again challenges the significance of cytosolic APS and PAPS synthesis; on the other hand, it shows the necessity of intracellular PAPS transport. Indeed, a PAPS transporter has been identified in chloroplast envelope membranes, part of the glucosinolate co-expression network, whose mutation shows a phenotype similar to apk1 apk2 mutants (see below and Ref.
). The reduced synthesis of PAPS in apk1 apk2 results in a shift of sulfur flux from the secondary to the primary sulfur assimilation pathway, increased accumulation of reduced sulfur compounds, and highly-reduced glucosinolate levels (
APS kinase is regulated on both transcriptional and post-transcriptional levels. The genes are part of the glucosinolate transcriptional network, under control by a family of six MYB transcription factors in Arabidopsis and thus co-expressed with genes providing the main substrate for PAPS (
). In addition, according to the demand-driven concept, sulfate starvation represses APS kinase to channel the scarce sulfur to the primary sulfate assimilation. Excitingly, redox regulation of APS kinase enzyme activity through dimerization of the protein and formation of disulfide bridges has been revealed in a structural analysis (
In contrast to plants with separate proteins possessing ATPS and APS kinase activities, vertebrate and invertebrate genomes feature these activities in single polypeptides with a C-terminal ATPS domain and an N-terminal APS kinase domain (
). It is interesting to ask why this second sulfate-activating complex has evolved and has been strictly maintained in animals. Possibly, this has been a requirement for the expansion of sulfation pathways in animals. Two PAPS synthase genes would allow us to selectively support different sulfation pathways, either via transcriptional co-regulation (
Such a subfunctionalization is also indicated by the fact that PAPS synthases 1 and 2 cannot complement each other. Genetic defects for PAPSS1 have never been reported so far. However, mutations in the gene coding for PAPSS2 are associated with bone and cartilage malformations as well as a steroid sulfation defect (
The mechanistic question is what makes the two PAPS synthase enzymes so different. Certainly, the two genes are differentially regulated to a certain extent, and transcriptional co-regulation with certain sulfotransferases has been reported (
). Diverse catalytic activities were purported to explain the observed functional difference, based on only a 5-fold difference in kcat/Km values when treating bifunctional PAPS synthases as pseudo one-step Michaelis-Menten enzymes (
). In light of these findings, the PAPSS gene fusion could also be thought of as a solubility anchor of a more stable domain for another less stable domain, among other factors.
Benefit of being fused together
For bifunctional PAPS synthases, answering questions on whether and how the individual domains functionally interact with each other continue to drive our understanding in the field. Channeling of the APS intermediate between the two domains of human PAPS synthases was initially hypothesized but was subsequently ruled out based on kinetic (
). The crystal structure of full-length human PAPSS1 shows dimers of APS kinase and ATP sulfurylase each with large interacting surfaces but only a weak interaction between those sulfurylase and kinase domain dimers (
). Therefore, the equilibrium can be shifted by the removal of the products, e.g. linking with inorganic pyrophosphatase to remove pyrophosphate or with enzymes utilizing APS. The animal PAPS synthase is clearly a mechanism for the latter, but not the only one in nature (
), and the need for increased catalytic efficiency may have led to other gene fusions. In filamentous fungi, ATPS is also fused with APS kinase; however, the kinase domain of the fusion protein is at the C-terminal end and functions only as an activation domain to modulate activity of ATPS without having a kinase activity (
). ATPS in the diatom Thalassiosira pseudonana and other Protozoans is fused to both APS kinase and inorganic pyrophosphatase, whereas in the dinoflagellate Heterocapsa triquetra ATPS is fused with the other APS-utilizing enzyme, APS reductase (
). In addition to this diversity, at least three different ATP sulfurylase enzymes have evolved independently, the plant and animal enzyme, the bacterial GTPase-linked enzyme, and one mainly found in cyanobacteria and green algae (
In plants, the need for APS for primary assimilation is greater that for the sulfation pathways, and therefore, a mechanism shifting the equilibrium from APS to PAPS is not advantageous. Another interesting evolutionary aspect of plant ATPS is that although it is largely a plastidic enzyme, it is in no way related to the ATPS in cyanobacteria, the precursors of plant plastids (
). Because chlorophyte ATPS is of cyanobacterial origin, it seems that the common precursor must have contained both forms, plastidic and Eukaryotic, and the sister lineages, plants and green algae, each retained a different one.
Core sulfation pathways
Sulfotransferases are the first enzymes of the core sulfation pathways. They transfer sulfate from PAPS to the hydroxyl or amino group of a wide variety of acceptors: carbohydrates, lipids, peptides, hormone precursors, xenobiotics, and other molecules (
). These aryl sulfotransferases display a different fold, but retain a similar spatial arrangement of the active-site residues, indicative of convergent evolution. They are covered in more detail elsewhere (
). Different research communities abbreviate mammalian cytosolic sulfotransferases to SULT, plant enzymes to SOT (or SULT), and Golgi enzymes according to their main activity/substrate (e.g. HS6ST for heparan sulfate-6-O-sulfotransferase, and TPST for tyrosylprotein sulfotransferase). For consistency, we will keep the different abbreviations.
The main differences between animal and plant sulfation pathways are the number of genes with more than 50 genes for human SULTs and 21 SOT genes in Arabidopsis, whereas the 17 human sulfatases do not have counterparts in plant genomes.
Arabidopsis and the human sulfotransferase repertoire
Sulfotransferases are grouped into categories, such as soluble or membrane-bound, cytosolic or Golgi-located, substrate preference for low-molecular-weight substrates, or the larger carbohydrates, proteins, or proteoglycans (
) list 22 genes for sulfotransferases in Arabidopsis (but one of them is annotated as a pseudogene). Phylogenetic analysis of protein sequences from the final 21 and 52 genes representing the Arabidopsis and human sulfotransferase repertoires, respectively, reveals that Arabidopsis SOTs 1–18 share high-sequence similarity with human cytoplasmic SULTs. In fact, these two groups share a higher degree of similarity with each other than human Golgi and cytoplasmic SULTs (Fig. 3). This is illustrated by a structural overlay of AtSOT18 and human SULT1A1 (Fig. 4). Hence, any new insight on one class of these sulfotransferases may have direct applicability to the other (
). For example, the flexibility of the main substrate-binding loops—elucidated in part with new analytical and computational tools—is the molecular basis for the broad specificity of the sulfotransferase SULT1A1 (
), with possible implication also for AtSOT12 described below. This flexibility makes it difficult to search for pharmacologically useful and isoform-specific inhibitors of sulfotransferases using computational docking (
). Such inhibitors are useful because human sulfotransferases metabolize many drugs and may thus interfere with various pharmacological interventions. Another regulatory mechanism, allosteric regulation, has only recently been described in sulfation pathways (
). The structure demonstrated evolutionary conservation of the sulfotransferases between humans and plants (Fig. 4) and suggested a loop-gating mechanism as responsible for substrate specificity for the sulfotransferase in plants (
); however, whether these in vitro data are relevant in vivo needs to be confirmed.
Similar to the desulfo-glucosinolate SOTs, only in vitro substrate specificities are known for other SOTs. While AtSOT5, AtSOT8, and AtSOT13 transfer sulfate to flavonoids, AtSOT10 modifies the plant hormones, brassinosteroids (
Tyrosine sulfation is a major post-translational regulation of secreted proteins and peptides in both animals and plants. However, this modification seems to be confined to multicellular Eukaryotes, as TPSTs have not been found in either bacteria or yeast (
). They share 67% amino acid identity with each other. As Caenorhabditis elegans and Drosophila only contain one TPST gene, a gene duplication may have occurred at the invertebrate–vertebrate transition.
Elucidating the biological roles of individual TPST isoforms through biochemical and structural studies of recombinant TPST proteins was a challenge for a long time. In 2013, Teramoto et al. (
). These structures are remarkable for two reasons. First, the core protein fold and the 5′-phosphosulfate–binding (5′-PSB) and the 3′-phosphate–binding (3′-PB) motifs involved in PAPS co-factor binding are structurally conserved even in these sulfotransferases only distantly related to their cytoplasmic counterparts (
). Second, they explain the mechanism of substrate recognition. Protein substrates need to locally unfold to bind to TPSTs in a deep active-site cleft, a process similar to the one known for tyrosine kinases (
Indeed, despite structural and mechanistic similarities, the two TPST isoforms display notable functional differences. TPST1-knockout mice show reduced body weight and fewer litters due to increased fetal death in uterus, whereas male fertility is not affected (
). This suggests that TPST1 reaction products which would not be sulfated in the TPST1 knockout, have a role in females during development of embryos. TPST2-knockout mice, however, primarily show male infertility (
), the picture emerges that tyrosine sulfation acts as macromolecular glue to strengthen interactions of proteins with other proteins or other biopolymers. A recent biophysical study clearly illustrates this point. In the complex of an N-terminal sulfated part of the chemokine receptor CCR5 and its CCL5 ligand NMR revealed that high-affinity binding is attributed to sulfate-mediated twisting of the two N termini (
). Identifying more and more sulfated proteins is expected in the near future due to advances in MS that allow better recovery of sulfated peptides and unambiguous distinction from their isobaric phosphorylated counterparts (
). Despite the importance of the tyrosine sulfation, however, the corresponding sulfotransferase remained elusive in plants, as no homologous proteins to the animal enzyme could be found. AtTPST was identified in Arabidopsis after isolation of the enzyme from the microsomal fraction and proteomics analyses (
Sulfatase genes encode proteins with a broad range of substrate specificity. The extracellular endoglucosamine 6-sulfatases, SULF-1 and SULF-2, target highly sulfated extracellular heparan sulfate domains, which are involved in growth factor signaling, tumor progression, and protein aggregation diseases (
). This unit produces the placental estrogens, estradiol and estriol, from fetal adrenal androgens via fetal adrenal sulfation, fetal hepatic hydroxylation, and placental desulfation, further downstream conversion, and release into the maternal circulation (
). In adults, STS is also expressed in many other tissues allowing for the uptake of sulfated steroid precursors and their desulfation.
Plants do not seem to possess sulfatase activity. This poses an obvious question for the catabolism of sulfated secondary metabolites. Glucosinolates are an important pool of sulfur, which can be recycled during sulfur starvation. Glucosinolate degradation is part of their anti-herbivore activity, which is initiated by tissue damage, bringing the glucosinolates into contact with thioglucosidases (myrosinases). Removal of the sugar moiety leads to chemical rearrangement of the aglycones to form volatile isothiocyanates or nitriles and release of sulfate (
). Soil bacteria, however, can metabolize these compounds by sulfatases, releasing the sulfate, and thus improving plant sulfur nutrition. Hence, releasing sulfate via sulfatase activity is the mechanism of some plant growth-promoting bacteria (
). Attempts to engineer intracellular or excreted sulfatase in plants, to make the organic sulfate available to plants, failed so far, most likely because of the need for the post-translational activation by production of formylglycine.
The nucleotide PAP is produced during PAPS-dependent sulfation pathways. It is also formed during CoA-dependent fatty acid synthetase activation (
). To prevent the toxic effects of PAP, dedicated PAP phosphatases are found in all kingdoms of life. Most of the enzymes from higher Eukaryotes show multiple specificity toward PAP or PAPS, and they also impact inositol signaling by removing phosphate from inositol bis- and triphosphates (
), all representing small and negatively charged substrates.
Lithium is known to influence many different proteins, and PAP phosphatases belong to the most sensitive targets for lithium inhibition. Mechanistically, lithium inhibition is well-understood for the PAP phosphatase CysQ from Mycobacterium tuberculosis. Lithium replaces one of a cluster of magnesium ions bound in the active site of the enzyme (
), due to the diagonal relationship between magnesium and lithium; these elements, diagonally adjacent in 2nd and 3rd periods of the periodic table, display a number of similar properties. As the negative amino acids in the catalytic center are highly conserved, it is highly likely that the same mechanism is in place in other PAP phosphatases.
In many microorganisms PAP phosphatases are strongly associated with sulfate assimilation, because accumulation of PAP also inhibits PAPS reductase, an essential enzyme in sulfate reduction. In yeast, loss of PAP phosphatase Met-22 leads to methionine auxotrophy (
Plant PAP phosphatases and PAP-dependent stress signaling
Plant PAP phosphatase SAL1 belongs to the most pleiotropic plant genes. It was first identified in rice as a protein complementing an inability to grow on sulfate in cysQ mutants of E. coli and met22 yeast mutants (
). Since then, SAL1 has been found in numerous genetic screens for a number of unrelated phenotypes and is therefore described under many different names. A common denomination, FIERY1 or FRY1, comes from a screen for mutants in abscisic acid and stress signaling, where its loss-of-function resulted in hyperinduction of the luciferase reporter gene driven by stress-responsible promoter (
). Thus, oxidative stress leads to oxidation of a redox cysteine pair in SAL1 and strong inactivation of the enzyme. This in turn results in accumulation of PAP, its transport to the nucleus, and induction of expression of stress-response genes (
It has to be noted that in Arabidopsis SAL1 is a member of a small gene family with seven members. SAL1 is, however, the only gene that has been found in the numerous genetic screens and that, when disrupted, causes the various phenotypes. Two additional isoforms, AHL and SAL2, were confirmed to function as PAP phosphatase (
). Although this is clear evidence for PAP being the causal metabolite for many phenotypes, the reason why in WT Arabidopsis AHL does not suffice to metabolize PAP remains to be elucidated. Another unsolved question is the physiological relevance of PAPS dephosphorylation.
The alteration in glucosinolate synthesis is the first direct metabolic link of SAL1 with sulfation pathways (
). Interestingly, combining the fou8 mutant with apk1 apk2 resulted in alleviation of many of the phenotypic alterations connected with loss of SAL1 function, strongly suggesting that PAP was the responsible metabolite (
), whereas the “Golgi PAP phosphatase” (gPAPP) is obviously located in the Golgi apparatus. For its side activity toward inositols, however, gPAPP is also known as inositol monophosphatase domain containing 1 (IMPAD1). The catalytic domain of this type II transmembrane protein is in the lumen of the Golgi (
); however, a functional role in tumorigenesis remains to be established. Patients with truncation mutations in gPAPP/IMPAD1 are characterized by short stature, joint dislocations, brachydactyly, and cleft palate (
). Hence, there was speculation whether BPNT1 is the actual target for lithium as a treatment for bipolar disorder. At least in C. elegans, lithium causes BPNT1-mediated selective toxicity to specific neurons and leads to behavior changes (
) reported a mismatch between broad expression of BPNT1, but measurable PAP accumulation only in liver, duodenum, and kidneys.
Interesting questions about how BPNT1 is regulated in a tissue-specific manner, whether redox regulation plays a role, and what involvement this PAP phosphatase has in further regulatory pathways remain to be answered. In worms at least, a genetic interaction of BPNT1 and the exoribonuclease XRN2 in polycistronic gene regulation has recently been reported (
Subcellular localization and transporters in sulfation pathways
The products of sulfation pathways represent intracellular metabolites as well as external proteins/peptides and carbohydrates. Therefore, the sulfotransferase enzymes have to be located in at least two compartments of the cytosol and Golgi apparatus. However, animal sulfate activation occurs in the cytoplasm and the nucleus as PAPS synthases shuttle between these compartments (
). Conserved nuclear localization and export signals govern this subcellular distribution, including a nuclear localization signal at the very N terminus of the APK domain as well as an atypical nuclear export signal at the APK dimer interface (
A complementary mechanism to having enzymes in multiple compartments is that the substrates and/or sulfated products themselves can traffic around the cell. Many low-molecular-weight compounds such as steroids are believed to be membrane-permeable. Notably, however, a recent study challenges the dogma of freely membrane-permeable steroids (and maybe also other smaller compounds). Okamoto et al. (
Plants also require transporters for function of sulfation pathways. Although all the PAPS-dependent sulfotransferases are located outside of plastids, the majority of the APS kinase activity is located within the chloroplast (
). Hence, plant cytosolic sulfation pathways are dependent on export of PAPS from plastids and those in the Golgi additionally on import of cytosolic PAPS. The first plant PAPS transporter was identified through co-expression with genes for glucosinolate synthesis and transport assays in liposomes (
) and belongs to the ADP/ATP carriers of the mitochondrial carrier family. AtPAPST1 also transports PAP, which has to be imported to plastids for degradation by SAL1; therefore, the transporter most probably serves as a PAP/PAPS antiporter (
). The loss-of-function mutant papst1 accumulates desulfo-glucosinolate precursors and shows decreased glucosinolate levels similar to but to a lower extent than apk1 apk2, suggesting the existence of a second plastidic PAPS transporter. Indeed, AtPAPST2 was recently identified in Arabidopsis as a transporter located dually in membranes of chloroplasts and mitochondria (
). Localization links AtPAPST2 to SAL1, which is also present in plastids and mitochondria. Thus, it seems that AtPAPST1 has a major role in exporting PAPS from chloroplast to cytosol for sulfation reactions and AtPAPST2 in importing PAP into the organelles for degradation by SAL1 (
). It also seems that the two transporters, AtPAPST1 and AtPAPST2, are not sufficient to explain all phenotypes connected to movement of PAPS and PAP between cytosol and the organelles, particularly the accumulation of glucosinolates and their desulfo-precursors. This metabolic phenotype can be expected to be found in mutants of the additional transporter gene(s) and enable their identification.
Natural genetic variation
The enzymes connected to sulfation pathways show a large variation between the different lineages and taxa. Many of them are found in several isoforms, further expanding their variation. However, the individual gene/enzyme isoforms also show variability within a single species, in different accessions and populations or even in individuals. Rare genetic mutations have been extremely informative in the study of many components of the sulfation pathways (
). Because of vastly increased sequencing capacities, such genetic variation is now studied on a population scale, both in plants and in humans.
Human genetic variation and clinical outcomes
Genetic defects in the gene for human PAPSS1 have not been reported so far. Gene defects in human PAPSS2, however, have been known to cause various forms of bone and cartilage malformation, due to an under-sulfation of the extracellular matrix (
). A subsequent study with two brothers carrying the compound–heterozygous mutations G270D and the frameshift mutation W462Cfs*3, resulting in an early termination codon, in PAPSS2 confirmed disrupted sulfation of DHEA, the most abundant steroid in the human circulation (