Biochemical Society Transactions

Frontiers in Biological Catalysis

Structural, mechanistic and regulatory studies of serine palmitoyltransferase

Jonathan Lowther , James H. Naismith , Teresa M. Dunn , Dominic J. Campopiano


SLs (sphingolipids) are composed of fatty acids and a polar head group derived from L-serine. SLs are essential components of all eukaryotic and many prokaryotic membranes but S1P (sphingosine 1-phosphate) is also a potent signalling molecule. Recent efforts have sought to inventory the large and chemically complex family of SLs (LIPID MAPS Consortium). Detailed understanding of SL metabolism may lead to therapeutic agents specifically directed at SL targets. We have studied the enzymes involved in SL biosynthesis; later stages are species-specific, but all core SLs are synthesized from the condensation of L-serine and a fatty acid thioester such as palmitoyl-CoA that is catalysed by SPT (serine palmitoyltransferase). SPT is a PLP (pyridoxal 5′-phosphate)-dependent enzyme that forms 3-KDS (3-ketodihydrosphingosine) through a decarboxylative Claisen-like condensation reaction. Eukaryotic SPTs are membrane-bound multi-subunit enzymes, whereas bacterial enzymes are cytoplasmic homodimers. We use bacterial SPTs (e.g. from Sphingomonas) to probe their structure and mechanism. Mutations in human SPT cause a neuropathy [HSAN1 (hereditary sensory and autonomic neuropathy type 1)], a rare SL metabolic disease. How these mutations perturb SPT activity is subtle and bacterial SPT mimics of HSAN1 mutants affect the enzyme activity and structure of the SPT dimer. We have also explored SPT inhibition using various inhibitors (e.g. cycloserine). A number of new subunits and regulatory proteins that have a direct impact on the activity of eukaryotic SPTs have recently been discovered. Knowledge gained from bacterial SPTs sheds some light on the more complex mammalian systems. In the present paper, we review historical aspects of the area and highlight recent key developments.

  • external aldimine
  • pyridoxal 5′-phosphate (PLP)
  • quinonoid intermediate
  • serine palmitoyltransferase (SPT)
  • sphingolipid
  • Sphingomonas sp


SLs (sphingolipids) are one of eight categories of lipid of the LIPID MAPS classification system [1]. Mammalian cells produce a vast array of SLs that function not only as structural elements of cellular membranes, but also as potent bioactive signalling molecules (Figure 1). They comprise a sphingoid base that can be amide-linked to a second fatty acid to give ceramides and derivatized further by the addition of a headgroup such as a sugar or phosphate. Structural diversity in the SL family arises from variations in the acyl chains and the large number of possible headgroups [2].

Figure 1 The series of enzymatic steps that makes up SL de novo biosynthesis in mammals

SPT (green) catalyses the first rate-limiting step; the enzyme names for other major steps in the pathway are in red.

SLs are also produced by plants, fungi and some bacteria, but the biosynthetic pathway in each can differ. For example, sphingosine is the major sphingoid base found in mammals, whereas phytosphingosine is found in fungi (Figure 2A). SL biosynthesis can also vary between species; the yeast Saccharomyces cerevisiae produces solely complex SLs containing a phosphoinositol headgroup, whereas Pichia pastoris produces both phosphoinositol-based and glucosylceramide-based GSLs (glycosphingolipids) [3]. SL biosynthesis also varies between bacterial species. GSLs are ceramides isolated from the Gram-negative bacterium Sphingomonas paucimobilis that contain either saturated and unsaturated sphingoid bases derived from palmitic acid that are attached to a second fatty acyl group derived from myristic acid [4]. By comparison, SLs in the dental pathogen Porphyromonas gingivalis contain unusual iso-branching on both the sphingoid base and the amide-linked fatty acid [5]. Therefore some organisms express anabolic enzymes for SL biosynthesis not found in others. Conversely, the first enzymatic step of de novo SL biosynthesis is common in all SL-producing organisms studied to date (Figure 2A). This step is carried out by the PLP (pyridoxal 5′-phosphate)-dependent SPT (serine palmitoyltransferase) [6]. The SPT gene is indispensable and thus can be used as a marker for identifying species of the SL-producing bacterium Sphingomonas in environmental samples [7]. SPT catalyses a Claisen-like condensation between L-serine and an acyl-CoA thioester (CoASH) substrate (typically C16 palmitoyl) or an acyl-ACP (acyl-carrier protein) thioester substrate, to form the first SL metabolite 3-KDS (3-ketodihydrosphingosine) (Figure 2B). Hence this condensation reaction forms the sphingoid base or LCB (long-chain base) found in all subsequent intermediate SLs and complex SLs in the organism. An abridged version of the SPT enzymatic reaction shown in Figure 3(A) indicates three important intermediates: (I) the holo-form or internal aldimine of the enzyme with PLP cofactor bound to the active-site Lys265 [SpSPT (S. paucimobilis SPT) numbering] via a Schiff base; (II) the external aldimine with L-serine following transimination; and (III) the quinonoid intermediate (carbanion equivalent) following deprotonation at Cα that attacks the incoming thioester substrate.

Figure 2 The key step in sphingolipid biosynthesis is conserved across all organisms studied to date

(A) The de novo pathway for SL biosynthesis can differ between species, but the first step, catalysed by SPT, is common across mammals, plants, fungi and bacteria. 3-KDS, 3-ketodihydrosphingosine; P-CoA, palmitoyl-CoA. (B) All SPTs condense L-serine with a fatty acyl derived from either an acyl-CoA substrate or an acyl-ACP substrate.

Figure 3 Mechanism and structure of bacterial SPT

(A) Abridged mechanism for the reaction catalysed by SPT. The reaction is dependent on its active-site PLP cofactor. (B) Structural comparison of the SPT internal aldimine with the L-serine-bound external aldimine form.

Three-dimensional structures of bacterial SPTs

SPT is a member of the AOS (α-oxoamine synthase) family of PLP-dependent enzymes and is classified in the fold-type I family of which aspartate aminotransferase is the prototype [8]. Three-dimensional crystal structures of several AOS enzymes have been solved including AONS (8-amino-7-oxononanoate synthase) [9], ALAS (5-aminolaevulinate synthase) [10], KBL (2-amino-3-oxobutyrate:CoA ligase) [11] and CqsA (cholera quorum-sensing autoinducer-1 synthase) [12,13]. Isolation of a homodimeric and water-soluble native SPT from the Gram-negative bacterium S. paucimobilis (SpSPT) [14] paved the way for structural studies on this enzyme (Table 1). The first crystal structure of the holoenzyme at 1.3 Å (1 Å=0.1 nm) resolution (PDB code 2JG2) revealed a symmetrical SPT homodimer with PLP cofactor bound to a conserved Lys265 in each subunit [15] (Figure 3B). The PLP cofactor is held in place via several interactions with active-site residues (at the dimeric interface from both subunits), the most important being π-stacking between the PLP pyridine ring and a conserved His159. Correct positioning of the PLP cofactor in the active site is vital, since addition of the L-serine substrate to SPT His159 mutants led to an abortive transamination reaction [16]. Subsequent studies captured the enzyme with the L-serine substrate bound as an external aldimine Schiff base following transimination at Lys265 [17]. In this structure, a non-conserved arginine residue (Arg378) had undergone a large swing into the active site to interact with the carboxy moiety of the PLP–L-serine external aldimine substrate (Figure 3B). Active-site arginine residues are crucial for recognition of the L-serine carboxy moiety during the catalytic cycle in this SPT homologue. In fact, mutation of a second active-site arginine residue in SwSPT (Sphingomonas wittichii SPT) (Arg370 equivalent to Arg390 in SpSPT), conserved in all SPTs and across the AOS family, resulted in an enzyme that could not form the subsequent quinonoid intermediate [species (III) in Figure 3A] in the presence of a palmitoyl-CoA analogue [18].

View this table:
Table 1 List of bacterial SPT homodimers and the current components of the human and yeast SPT complexes

The three-dimensional structures of two other SPT homologues, SmSPT (Sphingobacterium multivorum SPT) (38% homology) and SwSPT (70% sequence identity), have since been elucidated. The external aldimine form of SmSPT revealed an enzyme with an architecture very similar to that of SpSPT [19]. One major difference was that the carboxy group of the PLP–L-serine external aldimine formed was bound via two water molecules to two residues (Ser81 and Met271) on the opposite monomer, indicating that structural variations occur at the active site among homologous SPTs. Elucidation of the holoenzyme structure of SwSPT, from the dioxin-degrading bacterium S. wittichii, uncovered further variations [20]. For example, the entrance to the active site is wider than in its SpSPT homologue, suggesting that SwSPT may utilize a larger acylated-ACP thioester substrate rather than an acyl-CoA thioester. This prediction is strengthened by the fact that a gene encoding a putative bacterial Type II ACP resides immediately upstream of the SwSPT gene. Phylogenetic analysis of AOS enzymes has since found six bacterial SPT genes, including SwSPT, associated with ACP genes [21].

Other bacterial SPT isoforms

Along with the structural characterization of SPTs from the Sphingomonas strains described above, other SPTs have been characterized in terms of PLP-binding and substrate specificity from bacterial species including S. multivorum, Sphingobacterium spiritivorum and Bdellovibrio stolpii (Table 1) [22]. The PLP spectrum for each of these homologues was different from their S. paucimobilis counterpart. The SPTs from S. multivorum and B. stolpii were found to be peripheral proteins associated with the inner cell membrane, and therefore may more closely resemble the ER (endoplasmic reticulum)-bound eukaryotic SPTs. Furthermore, the SPT from B. stolpii displayed substrate inhibition by palmitoyl-CoA, a characteristic also found in the yeast [23] and mammalian SPTs [2426].

Recently, a study investigating the function of SLs in the gut-dwelling commensal bacterium Bacteroides fragilis suggested that SLs play a key role in survival strategies that allow the bacterium to persist under harsh conditions in the intestine [27]. Bacterial cultures grown in the presence of the SPT inhibitor myriocin were more susceptible to stressful conditions, suggesting a role for SLs in survival. Also in this study, a putative SPT gene was identified in B. fragilis by BLAST search [27]. Studies in our laboratory have confirmed that a purified recombinant version of this enzyme has SPT activity (E.K. Bower, J. Lowther, J. Wadsworth, S.A. McMahon, J.H. Naismith and D.J. Campopiano, unpublished work). Kinetic evaluation revealed an overall slower turnover compared with SPTs from Sphingomonas strains, as well as substrate inhibition by the acyl-CoA substrate similar to the SPT from B. stolpii.

Inhibitors of SPT

Several natural products (e.g. myriocin) have been identified as potent SPT inhibitors (Figure 4) and are routinely used as ‘blockers’ of SL biosynthesis in cells. Surprisingly, few complete investigations into the exact mechanism of inhibition by each have been undertaken. We previously carried out a detailed study on inhibition of SPT by the antibiotic L-cycloserine. This broad-spectrum inhibitor usually disables its target by forming an irreversible 3-hydroxyisoxazole–PLP adduct at the active site of several PLP-dependent enzymes such as alanine aminotransferase. However, prolonged incubation with SPT resulted in formation of PMP (pyridoxamine 5′-phosphate) and a small aldehyde product, β-amino-oxyacetaldehyde, both observed at the enzyme active site using X-ray crystallography and confirmed by MS [28]. Elucidation of the structure of another SpSPT crystal grown in the presence of cycloserine revealed an elongated Lys265 side chain, in keeping with in situ modification by the cycloserine-generated aldehyde (J. Lowther, S.A. McMahon, J.H. Naismith and D.J. Campopiano, unpublished work). Studies into the mechanism of inactivation by the SPT inhibitors L-penicillamine, β-chloro-L-alanine and myriocin are ongoing.

Figure 4 Natural product inhibitors of SPT

HSAN1 and deoxy-SLs

Many groups have contributed to the cloning and characterization of the SPT-encoding genes from higher organisms [2935]. It is thought that eukaryotic SPT contains a core heterodimer that spans the outer membrane of the ER [36] comprising a ‘non-PLP-binding’ LCB1 subunit (encoded by the SPTLC1 gene) together with a PLP-containing LCB2 subunit (encoded by either SPTLC2 or SPTLC3). In the present paper, we describe LCB1 as unable to bind PLP, but this has yet to be proven since the PLP-binding capacity of each subunit has not been analysed. We infer this from sequence homology/alignment studies; LCB2 contains all of the residues conserved among AOS family members involved in PLP binding and catalysis, i.e. a key lysine residue, two histidine residues and an aspartate residue, whereas LCB1 has none of these residues. Nevertheless, LCB2 is not expressed (and may not be stable) in the absence of LCB1, so the active catalytic SPT complex is thus a LCB1–LCB2 heterocomplex. Whether LCB1 contributes catalytic residues to the PLP active site remains to be determined. Specific mutations identified in either SPTLC1 or SPTLC2 cause a rare genetic disorder called HSAN1 (hereditary sensory and autonomic neuropathy type 1) [3739]. Three-dimensional structures of two bacterial SpSPT mimics, containing mutations similar to those identified in the LCB1 subunit of HSAN1 patients, revealed subtle structural changes at the enzyme active site [17]. Dunn and colleagues showed that yeast or CHO (Chinese-hamster ovary) cells expressing human heterotrimeric SPT [an LCB1–LCB2 heterodimer along with an ssSPT (SPT small subunit) that is required for optimal SPT activity, see below] carrying the HSAN1-causing mutation (C133W) in the LCB1 subunit gave rise to an enzyme with enhanced ability to condense L-alanine with the acyl-CoA substrate [40]. It therefore appears that HSAN1 mutations cause structural perturbations in the human enzyme, resulting in an altered specificity for the amino acid substrate. This relaxed specificity for amino acid substrate is consistent with the ‘gain-of-function’ phenotype of elevated levels of deoxysphingoid bases (deoxy-LCBs) arising from condensation of alanine and glycine with palmitoyl-CoA that is observed in HSAN1 patients and in HSAN1 transgenic mice [41].

The absence of the C1-OH in the deoxy-LCBs precludes phosphorylation by LCB kinase and degradation by the LCB-phosphate lyase in the only known pathway for LCB degradation. This may be in keeping with a late onset of disease in HSAN1 patients, i.e. these lipids gradually accumulate to toxic levels over time. It is also interesting to note that deoxysphingoid bases were also observed in ‘healthy’ cells, suggesting that wild-type SPT has some degree of amino acid substrate freedom to generate these lipids normally. An exciting promising pilot study results suggest that a treatment may be on the horizon for HSAN1 sufferers. A new report shows that oral administration of the natural substrate L-serine could compete with L-alanine and glycine, prevented accumulation of deoxy-SLs and improved HSAN1 symptoms [42].

It is important to stress that deoxy-SLs are natural metabolites that occur at low concentrations in all mammalian cells with a wild-type SPT. For example, deoxysphinganine was identified in kidney cells following addition of the CerS (ceramide synthase) inhibitor fumonisin B1 [43]. More recently, deoxyceramide has been found attached to an isolated human CD1b receptor [44]. Specific functions for deoxy-SLs are still unknown and, because they cannot be phosphorylated, the mechanism by which they are degraded or removed from the cell has yet to be revealed. This will be a particular interesting avenue of study, since it appears that their accumulation over a long period of time causes cytotoxicity in HSAN1. Further research is also required to understand the natural promiscuity of wild-type SPT and whether this phenomenon is unique to SPT among all AOS enzymes.

Identification of new subunits and regulators of the eukaryotic SPT complex

The SPT complex from higher organisms appears to comprise a core LCB1–LCB2 heterodimer (Table 1). The PLP-containing LCB2 subunit carries out the condensation reaction and, although the LCB1 subunit appears to lack the residues that bind PLP, it is still required for activity [38]. Hornemann et al. [31] identified a second isoform of LCB2 (LCB3, encoded by the SPTLC3 gene) that is expressed only in certain tissues. These two isoforms are functionally redundant in plants [45], but may have distinct functions in mammals since mice lacking the Sptlc2 gene were embryonic lethal [46]. It is interesting that some higher eukaryotes have two LCB2-like isoforms and others (e.g. Drosophila) have only one. Identification of novel components of SPT other than the core LCB1–LCB2 heterodimer has meant that the current model of the human SPT complex has advanced greatly in recent years. A third small subunit (Tsc3) associated with the heterodimer and required for maximal SPT activity was initially found by Dunn and colleagues in yeast [23], but bioinformatic approaches failed to identify candidate activators of the LCB1–LCB2 heterodimers from higher eukaryotes. However, poor correlation of SPT activity with increased SPTLC1/SPTLC2 expression in mammalian cells and the failure to reconstitute SPT activities comparable with those in mammalian microsomes upon expression of mammalian SPTLC1 and SPTLC2 in yeast led Dunn and her collaborators to search for an additional factor similar to the activity-enhancing Tsc3 subunit they had found previously in yeast [47]. Intriguingly, two novel small subunits (ssSPTa and ssSPTb) with no homology with Tsc3p were identified. Either of these small subunits can enhance activity >10-fold when bound to the LCB1–LCB2 (SPTLC1/SPTLC2 or SPTLC1/SPTLC3) heterodimer. Furthermore, these small activating subunits of SPT also confer distinct specificities for acyl-CoA substrates. The recent exciting discovery by two independent groups (led by Jonathan Weismann [48] and Amy Chang [49]) that the yeast ORM (orosomucoid) 1/ORM2 proteins also associate with and negatively regulate SPT activity has added an additional layer of complexity and prompted the authors to coin the term ‘SPOTS complex’ [SPT, ORM1/2, Tsc3, Sac1 (phosphatase)] to incorporate these novel components (Table 1). The ORMDL (orosomucoid-like) proteins are part of a family of ER proteins that were found to play a role in lipid homoeostasis and control of protein quality and trafficking [50]. The ORMs themselves are controlled in yeast through phosphorylation by yeast YPK1 [51], and it appears that the human ORMDL1/ORMDL2/ORMDL3 homologues interact with the LCB1–LCB2 heterodimer [48]. That the SPT complex may form a high-molecular-mass octameric complex comprising four LCB1–LCB2 heterodimers (and possibly associated ssSPT, ORM and Sac1 subunits) embedded in the membrane [52,53] raises further questions about the organization and regulation of SPT. Exactly how the cellular SL/ceramide levels are relayed back to this SPOTS complex to turn it on and off is not clear. Detailed studies of the topological organization and interactions between the components of the SPOTS complex will shed light on how these components interact to insure that this committed and rate-limiting enzyme of SL synthesis is properly regulated.


Work on SPT and SLs by D.J.C., J.L. and J.H.N. is supported by the Biotechnology and Biological Sciences Research Council [grant numbers BB/F009739/1 and BB/I013687/1]. The collaboration between D.J.C. and T.M.D. is supported by a Biotechnology and Biological Sciences Research Council United States Partnering Award (USPA) [grant number BB/G53045X/1].


  • Frontiers in Biological Catalysis: Biochemical Society Annual Symposium No. 79 held at Robinson College, Cambridge, U.K., 10–12 January 2012. Organized and Edited by David Leys (Manchester, U.K.), Andrew Munro (Manchester, U.K.), Emma Raven (Leicester, U.K.) and Martin Warren (University of Kent, U.K.).

Abbreviations: ACP, acyl-carrier protein; AOS, α-oxoamine synthase; ER, endoplasmic reticulum; GSL, glycosphingolipid; HSAN1, hereditary sensory and autonomic neuropathy type 1; LCB, long-chain base; ORM, orosomucoid; ORMDL, orosomucoid-like; PLP, pyridoxal 5′-phosphate; SL, sphingolipid; SPT, serine palmitoyltransferase; SmSPT, Sphingomonas multivorum SPT; SpSPT, Sphingomonas paucimobilis SPT; SwSPT, Sphingomonas wittichii SPT; ssSPT, SPT small subunit


View Abstract