Characterization of homologous sphingosine-1- phosphate lyase isoforms in the bacterial pathogen Burkholderia pseudomallei

Article abstract of DOI:10.1194/jlr.M071258

Sphingolipids (SLs) are ubiquitous elements in eukaryotic membranes and are also found in some bacterial and viral species. As well as playing an integral structural role, SLs also act as potent signaling molecules involved in numerous cellular pathways and have been linked to many human diseases. A central SL signaling molecule is sphingosine-1-phosphate (S1P), whose breakdown is catalyzed by S1P lyase (S1PL), a pyridoxal 5?-phosphate (PLP)-dependent enzyme that catalyzes the cleavage of S1P to (2E)-hexadecenal (2E-HEX) and phosphoethanolamine. Here, we show that the pathogenic bacterium, Burkholderia pseudomallei K96243, encodes two homologous proteins (S1PL2021 and S1PL2025) that display moderate sequence identity to known eukaryotic and prokaryotic S1PLs. Using an established MS-based methodology, we show that recombinant S1PL2021 is catalytically active. We also used recombinant human fatty aldehyde dehydrogenase to develop a spectrophotometric enzymecoupled assay to detect 2E-HEX formation and measure the kinetic constants of the two B. pseudomallei S1PL isoforms. Furthermore, we determined the X-ray crystal structure of the PLP-bound form of S1PL2021 at 2.1 ? resolution revealing that the enzyme displays a conserved structural fold and active site architecture comparable with known S1PLs. The combined data suggest that B. pseudomallei has the potential to degrade host SLs in a S1PL-dependent manner.-McLean, C., J. Marles-Wright, R. Custodio, J. Lowther, A. J. Kennedy, J. Pollock, D. J. Clarke, A. R. Brown, and D. J. Campopiano. Characterization of homologous sphingosine-1-phosphate lyase isoforms in the bacterial pathogen Burkholderia pseudomallei.

Full text of DOI:10.1194/jlr.M071258

Characterisation of homologous sphingosine 1-phosphate lyase (S1PL) isoforms in the
bacterial pathogen Burkholderia pseudomallei
1
Christopher McLean , Jon Marles-Wright , Rafael Custodio , Jonathan Lowther , Amanda J.
2#
3
1
1
Kennedy, Jacob Pollock , David J. Clarke , Alan R. Brown , Dominic J. Campopiano
1
3
1*
1
EastChem School of Chemistry, David Brewster Road, The University of Edinburgh, Edinburgh, EH9 3FJ,
UK.
2
Institute of Quantitative Biology, Biochemistry and Biotechnology, School of Biological Sciences, The
University of Edinburgh, Max Born Crescent, Edinburgh, EH9 3BF, UK.
3
*
Geoffrey Pope Building, University of Exeter, Stocker Road, Exeter, EX4 4QD, UK.
To whom correspondence should be addressed. Email Dominic.Campopiano@ed.ac.uk
Current address:
#
School of Biology, Newcastle University. Newcastle upon Tyne, NE1 7RU.
1

Abstract
Sphingolipids (SLs) are ubiquitous elements in eukaryotic membranes and are also found in some bacterial and
viral species. As well as playing an integral structural role, SLs also act as potent signalling molecules involved
in numerous cellular pathways and have been linked to many human diseases. A central SL signalling molecule
is sphingosine-1-phosphate (S1P) whose breakdown is catalysed by sphingosine-1-phosphate lyase (S1PL), a
pyridoxal 5’-phosphate (PLP) dependent enzyme that catalyses the cleavage of S1P to (2E)-hexadecenal (2E-
HEX) and phosphoethanolamine (PE). Here we show the pathogenic bacterium Burkholderia pseudomallei
K96243 encodes two homologous proteins (S1PL2021 and S1PL2025) that display moderate sequence identity
to known eukaryotic and prokaryotic S1PLs. Using an established mass spectrometry-based methodology we
show that recombinant S1PL2021 is catalytically active. Also, we used recombinant human fatty aldehyde
dehydrogenase (FALDH) to develop a spectrophotometric, enzyme-coupled assay to detect 2E-HEX formation
and measure the kinetic constants of the two B. pseudomallei S1PL isoforms. Furthermore, we determined the x-
ray crystal structure of the PLP-bound form of S1PL2021 at 2.1 Å resolution revealing the enzyme displays a
conserved structural fold and active site architecture comparable with known S1PLs. The combined data suggest
that B. pseudomallei has the potential to degrade host SLs in a S1PL-dependent manner.
Keywords
Sphingolipids, sphingosine-1-phosphate lyase, pyridoxal 5’-phosphate, assay, fatty aldehyde
dehydrogenase, Burkholderia
Abbreviations: SLs, sphingolipids; S1P, sphingosine-1-phosphate; S1PL, sphingosine-1-phosphate lyase; PLP,
pyridoxal 5’-phosphate; FALDH, fatty aldehyde dehydrogenase; 2E-HEX, (2E)-hexadecenal, PE,
phosphoethanolamine; PDB, Protein Data Bank; ESI-MS, electrospray ionisation mass spectrometry.
2

Introduction
Sphingolipids (SLs) are a diverse family of natural metabolites composed of a fatty acyl chain attached to a
polar headgroup derived from L-serine that together form the sphingoid base core (1). Amphipathic in nature,
SLs are important structural components of eukaryotic membranes that can associate with each other and with
other membrane components (such as cholesterol and proteins) in regions called lipid rafts (2). As well as this
structural role, SLs are now recognised as fundamental signalling molecules with a vast array of functions within
cells. A ‘sphingodynamics’ model has been proposed to encompass both intra and extracellular signalling by
all bioactive SL intermediates and their metabolites (3). Sphingosine-1-phosphate (S1P) is one such SL that
regulates a plethora of physiological processes in humans by activating five different G-protein coupled S1P
receptors (named S1PRs 1-5) (4). Perturbing the cellular levels of S1P can have serious consequences and has
been linked to diseases such as cancer, diabetes, atherosclerosis, osteoporosis and neurodegeneration (5, 6). The
balance between S1P and ceramide, another signalling molecule that often has opposing cellular effects to S1P,
is governed by a ‘sphingolipid rheostat’ (6, 7). Remarkably, the pyridoxal 5′-phosphate (PLP) dependent
enzyme sphingosine-1-phosphate lyase (S1PL) is the only known enzyme capable of irreversibly removing
sphingoid bases from the SL pool. S1PL catalyses the retro-aldol like breakdown of S1P to 2E-
hexadecenaldehyde (2E-HEX) and phosphoethanolamine (PE) (Fig. 1). In eukaryotes the 2E-HEX is oxidised
by fatty aldehyde dehydrogenase (FALDH, also known as ALDH3A2) to the hexadecenoic acid which links the
SL and glycerol phospholipid metabolic pathways (8). Saba and Hannun were the first to clone a S1PL gene
(
also known as dpl1 and BST1) from Saccharomyces cerevisiae in 1997 (9) and subsequently the human
homologue was characterised in 2000 (10). Since S1P plays a role in numerous diseases, enzymes in the S1P
biosynthetic and degradative pathways are now seen as attractive drug targets(3). To that end, Novartis recently
published results of a medicinal chemistry study that identified potent and specific inhibitors of human S1PL
(
11).
Nearly all Gram negative bacteria such as Escherichia coli, Legionella pneumophila and Burkholderia
pseudomallei contain lipopolysaccharides (LPS) as the major lipid component in their cell membranes. In
contrast, the number of bacteria that produce SLs, examples being Sphingomonas spp.(12) and Myxobacterium
spp. (13), is relatively small. However, it is interesting to note that many of the bacteria that are part of the
human microbiome such as Bacteroides fragilis and Porphyromonas gingivalis also produce SLs (14–16). This
3

raises the interesting question of what is the function of bacterial SLs? Moreover, if such species contain an
S1PL enzyme, what role does it play in bacterial metabolism in general and in the host-microbe interaction in
particular? Recent studies have begun to address these questions. A bacterial S1PL homologue (LegS2) was
identified in the intracellular human pathogen L. pneumophila (17, 18) and its role as a virulence factor in the
pathogenic mechanism of this organism was explored. B. pseudomallei is an intracellular Gram negative
pathogen that causes serious disease in humans and animals and as such, is classified by the Centre for Disease
Control (CDC) as a category B biothreat agent. B. pseudomallei infection results in melioidosis, a febrile and
potentially fatal disease which has a high mortality rate (19). To date, there have been no studies on SL
metabolism in Burkholderia, so we began by investigating if this species contains any of the known SL
metabolic enzymes and the recent work on the L. pneumophila focussed our attention on S1PL. We searched
the B. pseudomallei K96243 genome (20, 21) and surprisingly, found two ORFs BPSS2021 and BPSS2025,
whose gene products (S1PL2021 and S1PL2025 respectively) displayed high sequence identity to each other
and moderate sequence conservation to confirmed S1PLs from bacteria, yeast and humans. We felt this unusual
gene duplication merited further study, thus we set out to characterise the biochemical properties of these
putative B. pseudomallei S1PLs as a prelude to investigating their role in the host-pathogen interaction.
A number of assays that monitor S1PL activity have been reported in the literature. Many use either radioactive
3
14
(
e.g. H or C) or fluorescently-labelled (e.g. BODIPY) S1P analogues that generate labelled 2E-HEX aldehyde
that can be detected by scintillation counting or fluorescent methods respectively (22–25). Other assays rely on
the chemical derivatization of the HEX aldehyde product using chemical reagents, then applying mass
spectrometry (MS) to quantify the amount of material (LC-ESI-MS/MS, GC-MS, HPLC-ESI-QTOF etc) (26–
2
8). Each approach has its merits with respect to sensitivity, limit of detection and the ability to work with
purified enzyme, cell-free extracts or in whole cells. However, the drawbacks include the use of radioactive or
non-natural substrates, limited throughput and the requirement for the extraction of the S1PL reaction products.
First, using an established MS-based methodology we validated that the recombinant S1PL2021 displayed S1PL
activity (26). Then, to further characterise the two closely-related B. pseudomallei S1PLs we generated a
coupled spectrophotometric method for the convenient measurement of S1PL activity. This assay couples the
activity of these new S1PLs with a recombinant form of human fatty aldehyde dehydrogenase (FALDH) that is
part of the 2E-HEX catabolic pathway. In this way we were able to measure the production of 2E-HEX via the
4

+
reduction of the FALDH NAD cofactor to NADH. Furthermore, we determined the X-ray crystal structure of
the S1PL2021 isoform at 2.1 Å and revealed that it has features in common with bacterial, yeast and human
S1PL homologs. As well as reporting a convenient assay for S1PL activity, this study lays the foundation for
future studies of the role of these novel S1PLs in the growth and pathogenicity of Burkholderia.
5

MATERIALS AND METHODS
Materials
Plasmids and E. coli competent cells for cloning were purchased from Novagen and New England Biolabs. The
E. coli BL21 (DE3) competent cells used for protein expression were from Agilent. All chromatography
columns were from GE Healthcare. Hotstart Pfu polymerase was from Agilent. Restriction enzymes were from
New England BioLabs. Oligonucleotide primers were sourced from Sigma. Substrate S1P (CAS No. 26993-30-
6
) was from Cayman Chemical Company. (2E)-hexadecenal (2E-HEX) was synthesized in house using a
published method (29). All other buffers and reagents were sourced from Sigma.
Sequence alignment and prediction tools
Sequence data searches were performed using the BLAST suite of programs. Multiple sequence alignments
were carried out using the Clustal series of programs (30) with structural modelling performed using the Phyre2
server (31). To find S1PL homologs in the curated B. pseudomallei genome (www.burkholderia.com) the S1PL
sequence from Symbiobacterium thermophilim (Uniprot: Q67PY4, PDB: 3MAU) was used to carry out a
standard BLAST analysis. This recovered two hit genes, annotated as BPSS2021 (Uniprot: Q63IP8) and
BPSS2025 (Uniprot: Q63IP4) with moderate sequence identities to the template (43 and 44% respectively). The
encoded proteins, referred to as S1PL2021 and S1PL2025 share ~86% sequence identity.
Cloning and expression of BPSS2021 (S1PL2021)
Sequence analysis suggested that the BPSS2021 gene could be expressed from two start codons to give two
open reading frames (ORFs). Therefore, PCR forward and reverse primers were designed based on the longest
ORF (S1PL2021 N-EXT) using the following primers:
S1PL2021 forward 5' CCATGGTGCGTTCGATCGC 3'
S1PL2021 reverse 5' CTCGAGCGGACAGTCGGTAAACAG 3'
6

These primers incorporated NcoI and XhoI restriction sites giving a C-terminal six-histidine tagged fusion. A
positive pGEM/ S1PL2021 N-EXT clone was digested and the isolated S1PL gene ligated into pET-28a. This
gave a pET-28a/S1PL2021 N-EXT construct that expressed a protein of 500 aa. To clone the shorter ORF
(
S1PL2021), a forward primer was designed to initiate expression from an ATG start codon 60 base pairs
downstream of the annotated start codon;
S1PL2021 TRUNC forward: 5' CCATGGATCTCGAAGAAGG 3'.
This PCR product was also cloned into pET-28a via a pGEM intermediate to give a pET-28a/S1PL2021plasmid
encoding a protein of 480 aa. Plasmids pET-28a/S1PL2021 N-EXT and pET-28a/S1PL2021 were used to
transform E. coli BL21 (DE3) competent cells and selection was carried out on agar plates containing kanamycin
30
30
(
30 µg/ml, LB/Kan ). Single colonies were used to inoculate 2x250 ml of LB/Kan broth and overnight
cultures were grown at 37 °C with shaking at 250 rpm. The overnight cultures were used to inoculate 4l of fresh
3
0
LB/Kan broth and grown to an A600 of 0.6-0.8. Protein expression was induced by addition of IPTG to a final
concentration of 0.1 mM and growth was continued for 5 h at 30°C. Cells were harvested by centrifugation
(
Sorvall RC5-B centrifuge) at 5000 rpm for 15 minutes at 4°C.
Cloning and expression of the S1PL2021 K271A mutant
S1PL2021C mutants were prepared using the method of Naismith and Liu with primers were designed to
introduce the relevant mutations (32).
K271A forward 5’CGATACGCACGCGTTCGGCTACGGCCCGAAGGGC3’
K271A reverse 5’CGCCCTTCGGGCCGTAGCCGAACGCGTGCGTATCG3’
PCR reactions were set up using Hotstart Pfu polymerase with completed reactions digested with Dpn1 enzyme
to degrade template DNA. The post digestion mixture was transformed into high cloning efficiency cells
30
30
(
C2987- New England Biolabs) and spread on LB/Kan plates. Colonies were picked into LB/Kan and grown
overnight with plasmid DNA extracted from cells using a Qiaprep miniprep kit (Qiagen). Sequencing was
7

performed to ensure mutagenesis had occurred. Expression of the S1PL2021 mutants mirrored that of the
wildtype with the exception that induction temperature was reduced to 28°C.
Cloning and expression of BPSS2025 (S1PL2025)
The coding sequence for the BPSS2025 gene was amplified from B. pseudomallei K96243 chromosomal DNA
using primers:
S1PL 2025 forward: 5’ CCATGGATCTGGAGGAAGGCATCAG 3’
S1PL 2025 reverse 5’ CTCGAGAAGCGGGCAATCCGTGAAC 3’
to give an expressed protein sequence that aligned with the shorter S1PL2021 form described above. The
purified PCR product was cloned into the pGEM T-Easy vector by TA ligation and positive clones were isolated
and sequenced to confirm the identity of the gene. A positive pGEM/S1PL2025 clone was digested with NcoI
and XhoI and the isolated S1PL gene ligated into pET-28a to give a C-terminally six His-tag protein of 480 aa.
Expression of the gene and purification of the protein product was essentially identical to that for S1PL2021.
Cloning and expression of full length and truncated FALDH
The full length FALDH gene sequence (coding isoform 2) was purchased as a cDNA clone from Imagenes in a
pOTB7 vector (Uniprot: P51648). Full length FALDH (isoform 1 and 2) is predicted to contain a C-terminal,
membrane-bound helix thus using this as a template, full length and truncated versions for the FALDH isoform
1
(the major natural form) coding sequences were amplified by PCR using forward and reverse primers with
NdeI/NcoI and XhoI restriction enzyme sites.
FALDH forward: 5’ CATATGGAGCTCGAAGTCCGGCGGGTCCGACAGG.
FALDH reverse: 5’ CTCGAG(TCA)TAATATTCCGCCTTGACAAGCACAGCGGCTACAAT.
Truncated FALDH reverse: 5’ CTCGAGATCCACCTTTGACTGGCTGTTGGG.
The purified PCR products were cloned into pGEM, sequenced verified, and then cloned into pET-28a as
described for the S1PL constructs as an N-terminally hexa-his tagged protein. pET-28a plasmids containing
8

both versions of FALDH (full length and C-terminal truncation) were used to transform E. coli BL21 (DE3)
30
competent cells and selection was carried out on LB/Kan plates. Single colonies were used to inoculate 2x250
3
0
ml of LB/Kan broth and overnight cultures were grown at 37 °C with shaking at 250 rpm. The overnight
30
cultures were used to inoculate 4.5l of fresh LB/Kan broth and grown to an A600 of 0.6-0.8. Protein expression
was induced by addition of IPTG to a final concentration of 0.1 mM and growth was continued for 5 h at 22 °C.
Cells were harvested by centrifugation (Sorvall RC5-B centrifuge) at 5000 rpm for 15 minutes at 4°C.
General procedure for the purification of six his-tagged enzymes (S1PL2021, S1PL2025 and FALDH)
All purification steps were carried out at 4°C. All buffers contained 2 0mM HEPES (pH 7.5), 400 mM NaCl
and 10% glycerol. 100 µM PLP was included in the buffer for S1PL isolation. Binding buffers also contained
1
0 mM imidazole and one EDTA-free protease inhibitor tablet (Roche). Wash buffer contained 30 mM
imidazole with elution buffer containing 300 mM imidazole. Cell pellets (~4 g) were resuspended in Ni-NTA
binding buffer 10:1 (v/w). Cells were lysed by sonication (Soniprep 150) for 15 cycles (30s on, 30s off) on ice.
In the case of full length FALDH, 0.1% N-lauroylsarcosine was added to aid solubilisation of the recombinant
protein. The lysed cell suspension was centrifuged at 20,000 rpm for 30 minutes. The cell-free extract was
filtered (0.45µm syringe filter) before addition of 2 ml Ni-NTA resin (Qiagen) pre-equilibrated with binding
buffer. Binding was carried out by rotating the suspension at 4°C for 1 hour. The resin was collected by gravity
flow in a fritted plastic disposable column, washed with 30 ml wash buffer, and the bound protein eluted with
5
ml elution buffer. Fractions were analysed by SDS PAGE and the 300 mM imidazole eluate loaded onto a
pre-equilibrated HiPrep 16/600 Superdex S-200 size exclusion column. Recombinant protein was eluted at
a flow rate of 1 ml/min in buffer. The molecular weights of S1PL or FALDH were calculated from a calibration
curve using known molecular mass standards (GE Healthcare). The purity of the recombinant proteins were
analysed by SDS PAGE. The fractions were pooled and protein concentration was calculated by A280 using the
relevant extinction coefficients as determined using the Scripps Protein Calculator.
S1PL UV-visible spectroscopy
9

All UV-visible spectra were recorded on a Cary 50 spectrophotometer (Varian) and analysed using Cary WinUV
software (Varian). S1PL enzymes (15-20 µM) were maintained in holo-form by dialysing against buffer
containing 20 mM HEPES (pH 7.5), 400 mM NaCl, 10% glycerol and 100 µM PLP. Prior to spectroscopic
measurements excess PLP was removed using a PD-10 desalting column (GE Healthcare). A stock solution of
S1P (1 mg/ml) was prepared in 1% triton X-100 and sonicated. A stock solution of phosphoethanolamine (PE,
1
00 mM) in water. S1P and PE were added to the enzymes and scans recorded between 300-500 nm at
appropriate time intervals.
Assay of S1PL activity by LC/MS
The activity of S1PL was determined by detecting the formation of the (2E)-hexadecenal semicarbazone (2E-
HEX SC) product by LC/MS using the method of Berdyshev et al (26). Briefly an enzyme stock solution was
prepared by dialysing 10 µM S1PL in 20 mM HEPES (pH 7.5), 400 mM NaCl, and 100 µM PLP. A 2.6 mM
S1P stock was prepared with additional sonication. Using these stocks an enzyme assay was prepared to a final
volume of 500 µl containing 6 µM S1PL and 0.26 mM S1P substrate in a reaction buffer containing 20 mM
HEPES, 400 mM NaCl and 0.1 mM PLP. Control assays were prepared accordingly and all samples were
incubated for 20 minutes at 37°C with methanol then added to stop the reaction. The 2E-HEX was extracted
and the semicarbazone derivative produced by heating reaction aliquots with 0.2 ml of 5 mM semicarbazide
hydrochloride in methanol containing 5% formic acid at 40°C for 2 h. The semicarbazones were analysed by
mass spectrometry with samples injected via an Agilent 1200 HPLC with interfaced auto-sampler (Agilent
Technologies). Analysis was in ESI –positive mode on a micro-TOF2 (Bruker Daltonics) using a standard ESI
source. Ion source conditions were as follows; the ESI voltage was held at 4.5 kV at a temperature of 200°C
with a Hexapole RF of 70V peak to peak and skimmer 2 potential of 20V.
Assay of FALDH activity using UV-vis spectroscopy
The soluble version of FALDH, lacking its C-terminus, was assayed to determine its suitability as a coupling
enzyme with S1PL. This was done by monitoring the formation of NADH in the presence of a synthetic 2E-
+
HEX substrate. Reaction aliquots were prepared to a final volume of 500 µl containing 1 mM NAD , 500 nM
1
0

FALDH in a reaction buffer containing 20 mM HEPES (pH 7.5), 400 mM NaCl. Reactions were monitored by
UV-Vis at 340nm for 30 minutes at 37°C. Subsequent assays were performed in a Synergy HT plate reader and
the data plotted using GraphPad (non-linear regression analysis) to gain kinetic data for FALDH. The reaction
was linear over ~5 minutes and the lower limit of detection was ~4 µM 2E-HEX.
S1PL:FALDH coupled-enzyme assay
+
S1PL enzyme assays were prepared to a final volume of 500 µl containing 6 µM S1PL, 1 mM NAD , 500 nM
FALDH in a reaction buffer containing 20 mM HEPES (pH 7.5), 400 mM NaCl and 0.1 mM PLP. The reaction
was initiated by the addition of S1P from the stock solution to give a final concentration of 400 µM. Control
samples lacking enzyme or substrate were also prepared. Reactions were monitored by UV-Vis at 340nm for
3
0 minutes at 37°C on a spectrophotometer. For kinetic data the assays were then transferred to a Synergy HT
+
plate reader and these contained 6 µM S1PL, 1 mM NAD , 500 nM FALDH in a reaction buffer containing 20
mM HEPES (pH 7.5), 400 mM NaCl and 0.1 mM PLP (final well volume 150 µl). This time substrate S1P (0-
1
75 µM) was prepared in 1% triton X-100 and added to separate reaction aliquots in triplicate. Reactions were
monitored at 340nm for 30 minutes at 37°C. Kinetic data was fit using GraphPad as described previously.
X-ray crystallography
S1PL2021 at 6 mg/ml was crystallised by hanging-drop vapour diffusion at 18°C. Crystallisation experiments
were set up in 96 well MRC plates with commercially available screens from Molecular Dimensions Limited
with 100 nl drops of protein and 100 nl well solution and equilibrated against 70 µl of well solution. Crystals
grew in a well solution of 20% (v/v) polypropylene glycol 400, 10% (v/v) 1-propanol (Molecular Dimensions
MIDAS screen). Crystals were cryoprotected with mineral oil and then flash cooled by immersion in liquid
nitrogen. Datasets were collected on beamline I03 at the Diamond Light Source (Didcot, UK) at 100K using a
Pilatus 6M detector. Diffraction data were integrated and scaled using XDS (33) and symmetry related
reflections were merged with Aimless (34). The resolution cut off used for structure determination and
refinement was determined based on the CC1/2 criterion proposed by Karplus and Diederichs (35) The structure
1
1

of S1PL2021 was determined by molecular replacement using a model based on PBD code: 3MAD, modified
to match the target sequence using Chainsaw (36). A single solution comprising two molecules in the
asymmetric unit was found using Phaser (37). The initial model was rebuilt using Phenix.autobuild (38)
followed by cycles of refinement with Phenix.refine (39) in the Phenix programme suite (40) and manual
rebuilding in Coot (41) The final model was refined with isotropic B-factors and the model was validated using
MolProbity (42) Structural superimpositions were calculated using Coot. Crystallographic figures were
generated with PyMOL (43)
1
2

RESULTS
Identification of putative Burkholderia S1PLs by BLASTp search
BLASTp hits for novel bacterial S1PLs were identified in the genomes of various species of Burkholderia by
using the 507 amino acid sequence of the confirmed bacterial S1PL from S. thermophilum, StS1PL
(
Uniprot:Q67PY4) as the query sequence (20, 44). Several putative S1PLs with >40 % sequence identity were
identified in the known human pathogen B. pseudomallei and in B. thailandensis, a non-pathogenic model
organism (45). Strikingly, all representatives of these species each encode two putative S1PLs, with the two
genes located in close proximity within each species. This is exemplified by B. pseudomallei strain K96243
which contains two genes in close proximity with locus IDs of BPSS2021 (S1PL2021) and BPSS2025
(
S1PL2025) that display 86.3 % amino acid sequence identity between them and 43.9 % and 43.1 % identity
respectively to the confirmed StS1PL (supplementary Fig. S1 and S2). These two predicted S1PLs also
contained the conserved residues involved in binding the PLP cofactor (44).
To characterise the recombinant forms of these putative S1PLs we began by expressing S1PL2021 in E. coli,
but the ORF displayed two potential start codons. The sequence identity between S1PL2021 with StS1PL was
low at the N-termini (supplementary Fig. S2) with the pairwise alignment beginning at M47 of StS1PL. This
aligns well with a conserved MDLEEG motif found in S1PL2021 and the other S1PL homologs in Burkholderia
species and two clones were prepared; one that began at the conserved MDLEEG motif (named S1PL2021) and
another that expressed an extended construct (named S1PL2021 (N-EXT)).
Purification and characterisation of S1PL2021 and S1PL2021 N-EXT.
Recombinant S1PL2021 was expressed in E. coli from plasmid pET-28a as a hexa-histidine (6His) C-terminal
fusion. A standard purification procedure using immobilised nickel affinity (IMAC) and size-exclusion
chromatography (SEC) yielded milligram amounts of pure recombinant protein S1PL2021 that was
predominantly dimeric in solution (supplementary Fig. S3A and S3B). A similar strategy was used to prepare
recombinant S1PL2025 (supplementary Fig. S3C and S3D). Analysis of S1PL2021 suggested that PLP was
bound to the protein since the UV-visible spectrum displayed absorbances at ~330 nm and ~420 nm that are
characteristic of PLP-dependent enzymes (supplementary Fig. S3E). A similar PLP binding spectrum was
1
3

observed with S1PL2025 (supplementary Fig. S3F). In contrast, the alternative construct, S1PL2021 N-EXT
purified in an identical manner to the S1PL2021 but failed to display any PLP binding (supplementary Fig.
S3G). These results confirmed that these truncated versions of S1PL2021 and S1PL2025 bound PLP and these
forms of the protein were used throughout the study.
Substrate and product binding to S1PL2021.
When the substrate S1P (100 µM) was added to the PLP-bound form of S1PL2021 (15 µM) and the resulting
mixture analysed by UV-Vis spectroscopy, the peak at 420 nm decreased along with an increase in the 330 nm
region over a period of 30 mins suggesting that S1P is binding to S1PL2021 (Fig. 2A). Addition of the product
PE (10 mM) to the protein also showed changes in the UV-visible region (maxima at ~320 nm) with an
additional blue shift to ~395 nm, although this occurred at a much slower rate than for S1P (Fig. 2B). According
to the putative S1PL mechanism proposed by Bourquin et al (44) these spectroscopic changes are due to the
formation of the PLP-S1P and PLP-PE external aldimine intermediates that are bound to S1PL (supplementary
Fig. S4). We carried out similar UV-vis analysis of the binding of S1P and PE to the S1PL2025 enzyme (Fig.
2
C and 2D). The changes in the UV-vis spectra were similar when we compare the incubation of S1PL2021 and
S1PL2025 with S1P. In contrast, when we compared the spectra from the incubations of the product PE with
both enzymes, we did not observe the 420 to 395 nm shift in the S1PL2025 enzyme and the changes took longer
to occur compared with the S1PL2021 enzyme. These combined data suggest that, despite having highly similar
amino acid sequences (86 %), the two S1PL isoforms display small differences in their cofactor, substrate and
product binding that deserves further investigation. We also used UV-vis spectroscopy to show that S1PL2021
could also bind the non-phosphorylated substrate D-erythro sphingosine (Fig. 2E). Addition of this substrate
led to similar changes in the PLP region of the UV-vis spectrum that were observed for S1P, although this took
significantly longer (~4 hours).
Confirmation that S1PL2021 is an S1PL
To determine if the B. pseudomallei S1PL2021 and S1PL2025 displayed S1PL activity we used an established
mass spectrometry-based method developed by Berdyshev and colleagues that has been used to detect S1PL
1
4

activity in mammalian liver microsomes (26). This method relies on extracting the 2E-HEX aldehyde product
+
+
of the S1PL reaction and forming a stable semicarbazone derivative (2E-HEX SC, [M+H] C17
H
34
N O ,
3
predicted m/z = 296.3 Da, observed m/z = 296.3 Da) using semicarbazide hydrochloride (Fig. 3A). Recombinant
S1PL2021 was incubated with 260 µM S1P and aliquots taken for semicarbazone derivitisation. Additional
positive and negative controls were prepared and treated in an identical manner and the results analysed using
ESI-MS (Fig. 3B-E). The positive control used previously synthesized 2E-HEX and gave a single peak at 296.3
Da matching the expected semicarbazone derivative (Fig. 3B). This peak was absent in the no enzyme control
+
39NO
P+ m/z = 380.2 Da, observed m/z = 380.4 Da, Fig. 3C).
with the S1P substrate ([M+H] , predicted C18
H
5
The reaction catalysed by S1PL2021 resulted in a signal with m/z = 296.3 Da indicating the production of the
E-HEX SC species and confirming S1PL2021 displayed S1PL activity (Fig. 3D). In contrast, the no substrate
2
control lacked both the above peaks (Fig. 3E).
FALDH coupled assay suitability/optimisation
The ESI-MS method confirmed that S1PL2021 displays S1PL activity, therefore we went on to develop
a convenient continuous spectroscopic assay in order to perform steady-state kinetic analysis of S1PL. We took
advantage of the activity of fatty aldehyde dehydrogenase (FALDH), the next enzyme in the SL metabolic
pathway (Fig. 1). The human FALDH (also known as ALDH3A2) has been cloned and recombinant enzyme
has been isolated from E. coli by Lloyd et al. (46) so this presented an attractive candidate for the development
of a coupled assay. A commercially available cDNA clone (Source BioScience, product code
IRAUp969B0810D in plasmid pOTB7) was used to construct an expression clone of the full length FALDH
protein (isoform 1), however, in our hands, this initial construct was insoluble (data not shown). A sequence
alignment between human FALDH and rat ALDH3A1, which lacks a hydrophobic C-terminal transmembrane
region, guided the preparation of a second, truncated clone (supplementary Fig. S5). This shorter variant
expressed well in E. coli and we were able to isolate mg quantities of recombinant His-tagged human FALDH
(
supplementary Fig. S6). By monitoring aldehyde substrate-dependent, NADH formation at 340 nm we
established that the fully saturated substrates, hexadecanal and tetradecanal and, most importantly, the
unsaturated 2E-HEX, were substrates for this recombinant FALDH (Fig. 4A). Using varying 2E-HEX substrate
concentrations (0-260 µM) the reaction was linear over the first 3 minutes and the limit of detection was
1
5

determined to be ~4 µM 2E-HEX (supplementary Fig. S7A). We performed Michaelis Menten analysis and
-1
determined the kinetic values for 2E-HEX with a K
M
= 19.4 ± 3.2 µM and kcat = 0.22 ± 0.01 s recorded. (Fig.
4
B, Table 1). The active FALDH was then coupled to both S1PL2021 and S1PL2025 in the presence of S1P
+
and NAD at 30°C to determine the kinetic constants of both S1PL enzymes. The initial reaction rates were
linear over a S1P concentration range of 0-175 µM. Therefore S1PL activity could be quantified in a continuous
fashion (down to a S1P substrate concentration of ~10µM) (supplementary Fig. S7B and S7C) and allowed the
determination of the kinetic constants of both B. pseudomallei S1PLs. The values for each are as follows;
-1
S1PL2021 (K
M
= 48.2 ± 9.4 µM and kcat = 0.015 ± 0.003 s ) and S1PL2025 (K
M
= 50.2 ± 8.3 µM and kcat =
-1
0
.008 ± 0.0005 s ) (Fig. 4C and 4D, Table 1). Thus this new assay offered a convenient way of measuring the
activity of recombinant S1PL. Based on the S1PL sequence alignment residue K271 of S1PL2021 is predicted
to bind the essential PLP cofactor (supplementary Fig. S2). A S1PL2021 K271A mutant was prepared and when
used in the coupled assay this variant was essentially inactive in comparison to the wild type enzyme (Fig. 4E).
We also used the assay to determine if S1PL2021 used D-erythro sphingosine as a substrate and found that it
displayed ~8% activity compared with S1P (Fig. 4F).
Structural characterisation of S1PL2021
The further characterise the Burkholderia pseudomallei S1PL enzymes we subjected S1PL2021 to
crystallisation trials using commercially available screens. Crystals with a monoclinic morphology were grown
in 20 % (v/v) PEG400, 10 % (v/v) 1-propanol. Diffraction data on cryocooled crystals were collected on a single
crystal at Diamond Light Source (beamline I03) to a maximum resolution of 2.1 Å in space group P1 2 1. The
1
structure of S1PL2021 was determined by molecular replacement using a monomer of the S. thermophilum
S1PL structure as the template (PDBID: 3MAD). The final refined model contains a dimer of S1PL with the
PLP cofactor present as an internal aldimine with residue K271 in each chain (Fig. 5A). There was no clear
electron density present for residues before L10; between residues H430 and A455; and for the His6 tag after
residue P472 at the C-terminus; residues in these regions were omitted from the final model.
The S1PL2021 structure conformed to the typical type I PLP dependent fold seen in the group II decarboxylase
subfamily (11, 44). The overall architecture of the protein closely matches homologous structures, with excellent
1
6

overall alignment of S1PL2021 with published structures (supplementary Fig. S8A, supplementary Table S1).
The structure has an N-terminal domain (residues 10-55), which is primarily involved in dimerization; a large
cofactor-binding central domain (residues 56-320); a short C-terminal domain (residues 321 – 430); and a C-
terminal extension (431 – 471) (Fig. 5A). The latter domain is partially disordered, with the region between
residues 431 and 455 not visible in the experimental electron density; this region sits close to the cofactor and
substrate-binding region and may change conformation on ligand binding.
The cofactor-binding site at the dimer interface shows excellent electron density for the PLP cofactor, which
was found to bind to K271 via a Schiff base as an internal aldimine (Fig. 5B). Coupled with the high
conservation of active site residues with other members of the S1PL family across bacteria, yeast and human
(
supplementary Fig. S8B, supplementary table S2), our structural data confirm the assignment of S1PL2021 as
a member of the sphingosine-1-phosphate lyase family (EC number 4.1.2.27) and given the high sequence
identity of S1PL2025 (86 %) suggest that this is also a S1PL family enzyme. The cluster of residues involved
in PLP binding are highly conserved across the members of the S1PL family, with these residues adopting the
same orientation in published crystal structures (supplementary Fig. S8B). Analysis of the active site shows that
the pyridine ring is held in place between C235 and H159 with π-stacking interactions to the latter residue; the
nitrogen of the pyridine group forms a hydrogen bond to D233 (fig. 5B). The PLP phosphate group forms
hydrogen bonds with the sidechain of H270 and the backbone nitrogens of G126, T127, and S313 from the
partner chain. The K271 chain is held in place through a hydrogen bond between D268 and the sidechain
nitrogen that forms the Schiff base with PLP.
The dimer observed in the crystal structure of S1PL2021 is the native form of this protein (Fig. 5A), as the
cofactor-binding site and ligand-binding funnel is formed between the two monomers (supplementary Fig. S9A
and S9B). The dimerization interface buries nearly 50 % of the total surface area of each monomer with a total
2
buried area of 6430 Å ; the interface is stabilised by 52 hydrogen bonds and 32 salt bridges. The observation
that S1PL2021 is unable to bind PLP when expressed in the full-length form as annotated in the genome is
explained by the fact that any significant extension of the N-terminus of the structure from that seen in the
crystal structure would potentially occlude the substrate binding funnel and PLP binding site (supplementary
Fig. S9A). The substrate-binding funnel of S1PL2021 is formed between the two monomers, with one monomer
providing a primarily positively charged surface and the other a negatively charged surface (supplementary Fig.
1
7

S9B). The missing region of the protein that forms the C-terminal extension is likely to play a role in substrate
recognition given its position over the funnel and its inherent flexibility in this structure.
1
8

DISCUSSION
Sphingolipids and ceramides have been shown to be important players in various eukaryotic cell
functions. The amino alcohol core structure of all SLs is built from L-serine and a long chain fatty acid and
further elaborated by a series of enzymes that increase their complexity by phosphorylation, oxidation and
acylation (1). Cells maintain a balance of SLs and ceramides via highly regulated synthetic and metabolic
pathways whose control mechanisms are still not fully understood. Perturbation of this balance has now been
linked to a number of diseases such as cancer, atherosclerosis, autoimmunity and infection (5, 6). A key
regulatory mechanism involves the breakdown of the potent signalling molecule S1P by S1PL and S. cerevisiae
has been a useful model organism where the pathways for de novo biosynthesis, recycling and degradation of
SLs have been delineated (47). Indeed, the first S1PL gene (dpl1) was characterised in S. cerevisiae by Saba
and colleagues and they showed that a Dpl1 mutant strain was not only highly sensitive to exogenous
sphingosine but also accumulated intracellular S1P upon exposure to sphingosine (9). In contrast to higher order
organisms, relatively few studies have been carried out to investigate the role of SLs biosynthesis and
metabolism in bacterial species. Of these, it is interesting to note that microbes from the human microbiome –
P. gingivalis from the oral cavity and the gut commensal Bacteroides fragilis make SLs with unusual structural
features (14–16). Their roles in bacterial metabolism, as well as their interaction within the host environment,
are currently under investigation (48) .
Burkholderia are a group of interesting bacterial species that can cause life-threatening infections and
have a reputation as being highly resistant to antibiotics (49–51) They appear to lack the ability to synthesise de
novo SLs since analysis of their genomes suggest that the biosynthetic genes are absent. We were therefore
interested to discover that B. pseudomallei K96243 encodes two highly homologous ORFs that display moderate
(
~40%) sequence identity to the S. thermophilum S1PL query sequence (supplementary Fig. S1 and S2). The
two ORFs, BPSS2021 and BPSS2025 (annotated as decarboxylases in the genome) also display 86 % sequence
identity to each other. Since a more detailed sequence and structural analysis revealed that both ORFs contained
conserved residues important for binding the PLP cofactor and enzyme catalysis we hypothesised that these
genes encode S1PLs. Therefore, we cloned, expressed and purified recombinant forms of both S1PLs and UV-
vis analysis showed that they bound the PLP cofactor (supplementary Fig. S3). Monitoring the changes in the
UV-vis spectrum of PLP-dependent enzymes can be a useful tool to study how substrates, products and
1
9

inhibitors interact with these enzymes. The PLP binding region of holo-S1PL2021 showed changes upon
addition of 100 µM S1P with an initial sharp rise in the 330 nm region, followed by a slower increase at this
wavelength (Fig. 2A). At the same time there was a decrease in the absorbance at 420 nm. Addition of the
product PE to S1PL2021 led to the appearance of a peak at 320 nm and a decrease at 420 nm with a shift to 395
nm (Fig. 2B). We carried out the same substrate and product binding analyses with S1PL2025 (Fig. 2C and 2D)
and the general trend was similar, unsurprising as they are 86% identical. In contrast, in their studies with the
S. thermophilum S1PL, Bourquin and colleagues observed the transient appearance of two peaks (403 nm and
4
20 nm) which they tentatively assigned as the PLP-S1P and PLP-PE external aldimines in the putative S1PL
reaction mechanism (intermediates 3 and 7 respectively in supplementary Fig. 4). These peaks appeared
immediately upon S1P addition, and then reduced over 5 minutes until the spectrum resembled that of the
starting PLP-bound, holo- form of the enzyme. Moreover, they noted that the yeast S1PL UV-vis spectrum did
not show any changes upon S1P addition and they concluded that this enzyme was inactive. Our mass
spectrometry and FALDH coupled assay (described below) prove both Burkholderia S1PL2021 and S1PL2025
are active. Given that the S1PLs did not regenerate the holo- form spectrum, it suggests differences between the
Burkholderia S1PLs and the S. thermophilum S1PL with which they share ~40% sequence identity. Our results
could indicate the slow release of products (PE or 2E-HEX) or the build up of a hitherto uncharacterised
intermediate.
The ability to measure S1PL activity using a convenient method would help expand structural and
mechanistic studies of the enzyme in all organisms. Since the first discovery of S1PL activity in mammalian
14
cells by Stoffel and colleagues, (52) who used a C-labelled substrate, a number of S1PL assays have been
developed that use radioactive, fluorescent and fluorogenic substrates and substrate analogues. One of the
3
earliest methods used incorporation of H labels into the S1P substrate and thin layer chromatography (TLC)
and liquid scintillation counting (LSC) analysis of the products. This was applied to the study of the S1PL
enzyme in rat liver and values of K
M
= 9 µM and Vmax = 0.082 pmol/min/µg were determined (22). Bandhuvula
and colleagues synthesized a fluorescent, NBD-labelled S1P homolog (S1P-C18NBD) that was used to analyse
-4
the human S1PL in HEK293 cells and gave values of K
M
= 14.6 µM and Vmax = 3.4 x 10 pmol/min/µg (23).
These authors then prepared a more stable fluorescent S1P-C14 Bodipy analogue that was used to analyse the
-4
human enzyme again in HEK293 cells that gave values (K = 35 µM, Vmax = 5.1 x 10 pmol/min/µg) in good
M
2
0

agreement with the previous study (24). An alternative, clever strategy retained the phosphorylated amino
alcohol structure of the natural S1P substrate and combined it with an ether-linked, coumarin-based reporter to
generate a substrate that, upon S1PL-catalysed cleavage, breaks down by β-elimination to a highly-fluorescent
umbelliferone product. Although highly sensitive, this non-native substrate displays a high K (152 µM) for the
M
S1PL in mouse embryonic fibroblasts (MEFs) and the fluorescence was significantly reduced in the presence
of Triton-X100 that is used to solubilise the S1P (25). Mass spectrometry (MS)-based techniques have also been
developed that involve the derivitization of the 2E-HEX aldehyde product with reagents (usually hydrazides)
that improve ionization efficiency. The first MS-based method was developed by Berdyshev and colleagues to
characterise the S1PL activity in mouse liver microsomal extracts (K
M
= 5.7 µM, Vmax = 0.171 pmol/min/µg)
(
26). A similar gas chromatography/electron ionisation (GC/EI) MS method used pentafluorobenzoyloxime
derivitization of the 2E-HEX to analyse the S1PL in MEFs (K = 6 µM, Vmax = 0.374 pmol/min/µg) (27).
M
Novartis have recently identified the human S1PL as an attractive target for the development of therapeutics
and developed a high throughput, MS-based method to screen for novel inhibitors (11). By using
nicotinylhydrazide (Isoniazid) modification of the 2E-HEX they obtained the first data on purified, recombinant
human S1PL (K
using an internal D-labelled standard to also determine the values for recombinant human S1PL in good
agreement with the value determined by Novartis (K = 4.67 µM, Vmax = 2.6 pmol/min/µg) (53).
M
= 5.2 ± 0.9 µM). Most recently, Suh and colleagues developed a stable isotope dilution method
M
We decided to use an MS-based method to confirm that the recombinant B. pseudomallei S1PL2021
isoform catalysed the degradation of S1P to 2E-HEX (Fig. 3). Since this end-point method involves a number
of liquid handling steps (derivitization, organic extraction) and requires an internal standard, we sought to
develop a convenient, continuous spectroscopic assay that could be used for S1PL analysis. Our choice of
coupling enzyme was informed by our knowledge of the SL degradative pathway (Fig. 1). In mammals, the 2E-
HEX aldehyde product of the S1PL enzyme is oxidised to the acid by a fatty aldehyde dehydrogenase (FALDH,
+
also known as ALDH3A2, UniProt Code: P51648) in an NAD -dependent manner that can be monitored at 340
nm. Human FALDH has been the subject of various studies since it has been linked to the SL-metabolic disease
Sjögren–Larsson syndrome (8). Previously, recombinant, full-length human FALDH had been isolated by Lloyd
et al who showed that extraction of the enzyme from E. coli required the use of the anionic detergent N-
lauroylsarcosine (46). This enzyme displays broad substrate specificity and could use saturated fatty aldehydes
2
1

of varying chain lengths (from C5 to C18) with K
M
values in the low µM range (4-35 µM) and kcat values (~1 s-
1)
that would deem it suitable as a coupling enzyme. We prepared recombinant, N-terminally His-tagged human
FALDH by relatively straightforward, soluble expression in E. coli once we had removed the C-terminal 36
residues that contained a predicted potential transmembrane (TM) domain (supplementary Fig. S5). Of note,
Lloyd et al did not prepare 2E-HEX but did show that the FALDH used the commercially available, unsaturated
cis-11-hexadecenal as a substrate. We found that our truncated, recombinant FALDH was not only active with
the saturated C14 and C16 fatty aldehydes but, more importantly, also used the synthetic 2E-HEX as a substrate
-1
and we determined kinetic values (K = 19.4 ± 3.2 µM and kcat = 0.22 ± 0.01 s ) that suggested that it would
M
make a suitable coupling partner for the determination of S1PL activity (Fig. 4A and B).
During our work a detailed x-ray structure and mechanistic study of the human FALDH was published
with a view to understanding the substrate specificity of the enzyme and the impact of mutations that cause SLS
(
54). In this paper the FALDH construct contained an N-terminal streptavidin affinity tag and, in comparison to
our truncation, only removed the C-terminal 22 amino acids, the minimum based on the proposed TM region.
This construct was catalytically active with C8, C12 and C16 saturated aldehyde substrates in the presence of
0
.1% Triton X100 but they noted that removal of a 40 amino acid section resulted in a drastic reduction in
activity with the longer acyl-chain length substrates. The x-ray structure of the FALDH (PDB: 4QGK, 2.1 Å
resolution) revealed that, as well as being involved in the TM interaction, the C-terminal domain also contained
a gatekeeper helix that is proposed to be involved in substrate specificity (supplementary Fig. 5B). This structure
allowed us to rationalise how our truncation of the C-terminal hydrophobic element increased the solubility of
our FALDH construct whilst retaining enough of the helix to maintain good activity towards long acyl chain
substrates, crucially including 2E-HEX (supplementary Fig. S5A and S5B). We combined our FALDH with the
+
purified Burkholderia S1PL2021 and S1PL2025 enzymes in the presence of PLP, NAD and S1P solubilised in
Triton X100. We observed that the assay was linear over a practical substrate range (0-175 µM) and determined
-1
kinetic values for S1PL2021 (K
M
= 48.2 ± 9.4 µM and kcat = 0.015 ± 0.003 s ) and S1PL2025 (K
µM and kcat = 0.008 ± 0.0005 s ) (Fig. 4C and Fig. 4D, Table 1). Since both isoforms display such high sequence
identity it is not surprising that they display similar K values for S1P, however we noted that the S1PL2025
M
= 50.2 ± 8.3
-1
M
was 50% slower than the S1PL2021 isoform. Although we used this assay to only study the Burkholderia S1PLs,
we think that this convenient FALDH-coupled method will be of use to those working with S1PLs from other
2
2

organisms (e.g. yeast and human). It will be interesting to determine if it can also be applied not only to purified
recombinant enzymes and HTP S1PL inhibitor screens, but also whole cell lysates.
To explore the structures of these novel Burkholderia S1PLs we used our optimised expression system
for S1PL2021 to prepare crystals that diffracted to 2.1 Å. Using the S. thermophilum S1PL structure (PDB:
3
MAD) as a template for molecular replacement the structure refined well and the enzyme displayed the
canonical Type I fold of the PLP superfamily of enzymes (Fig. 5A, suppl. Fig. 8A) (55). Furthermore, it belongs
to the group II decarboxylase subfamily which is typical of S1PLs from other species. The monomer has a three
domain architecture, a short N-terminal domain, a large central domain that binds the well resolved PLP cofactor
bound as an internal aldimine/Schiff base to K271, and an extended C-terminal domain (Fig. 5A). As well as
K271, comparison with other S1PLs identifies conserved residues involved in PLP binding; C235 and H159
binds to the pyridine ring, residues G126, T127, H270, S313* (from the other subunit) coordinate the phosphate
group and D233 is involved in salt bridge formation of the pyridine nitrogen (Fig. 5B, supplementary table S2).
An overlay of all the well-refined structures of S1PLs in the PDB with PLP and/or Pi bound shows that the
S1PL2021 active site architecture is well conserved (Suppl. Fig. 8B, supplementary table S2). Given its role in
anchoring the PLP cofactor it is not surprising that mutation of the conserved K271 drastically reduces S1PL
activity to 2% when compared to wild type (Fig. 4E). Both S1PLs showed activity with S1P as a substrate but
we also wished to explore if the phosphate of S1P was required for activity. Since we had observed binding of
D-erythro sphingosine to S1PL2021 (Fig. 2E) by UV-vis spectroscopy, it was interesting to find that this
isoform did catalyse breakdown of this non-phosphorylated sphingolipid (~8% activity compared to S1P, Fig.
4
F). Thus it appears that the Burkholderia S1PLs display some substrate promiscuity and may be able to break
down a range of both phosphorylated and non-phosphorylated sphingolipids. In future it will be interesting to
carry out a comprehensive substrate analysis with these enzymes and explore their specificity with respect to
acyl chain length, C1 head group, the oxidation state at C4-C5 (sphingosine/sphinganine) and the C2-C3
stereochemistry (erythro/threo).
Although the crystal structures of S1PLs from S. thermophilum, S. cerevisiae, human and L.
pneumophila have been solved the exact nature of substrate binding and enzyme catalysis are still unclear since
the structure of a S1P-bound complex has not been determined (44, 56, 57). For example, in the putative
mechanism, the active site base that deprotonates the S1P hydroxyl leading to the retro-aldol like cleavage of
2
3

the substrate has not been identified (supplementary Fig. S4). Bourquin et al described a S. thermophilum
S1PL:PLP-PE product complex as well as a S1PL:PLP internal aldimine with a phosphate ion bound and
together these structures hinted at the residues involved in binding the phosphate of S1P. Despite screening
numerous conditions we were unable to trap the wild type S1PL2021:PLP-S1P external aldimine complex since
we assume the enzyme catalyses the rapid break down of the S1P. We recently used mutagenesis of an active
site lysine that allowed us to capture the PLP-dependent enzyme serine palmitoyl transferase (SPT) bound to
the inhibitor myriocin (58). We used a similar strategy to screen crystals of the S1PL2021 K271A mutant for
S1P binding but this has so far failed to yield a structure (data not shown). To date the exact residues involved
in binding the S1P acyl chain are still not known but an overlay of all available S1PLs suggests the presence of
hydrophobic funnel on the surface which leads down into the active site (supplementary Fig. S9). It is interesting
that one of the differences between the two enzymes is at position 401; this is a histidine in S1PL2021 and
glutamine in S1P2025. This change could contribute to the 50% difference in activity between the two isoforms.
In future, swapping the histidine at position 401 for glutamine (and vice versa) between the two forms, as well
as mutating other potential residues should identify those involved in substrate binding and catalysis. Our
continuous S1P assay will be useful for screening the activity of these and any other mutants.
Since human S1PL has recently been identified as an attractive target for drug discovery, both Novartis
and AbbVie have undertaken medicinal chemistry programmes to identify potent inhibitors (56, 59). The
determination of the structures of human S1PL and of the “humanised” S. thermophilum S1PL in complex with
various inhibitors (PDB codes : (PDB codes: 4Q6R, 5EUE and 5EUD respectively) have highlighted how these
molecules block the substrate-binding funnel. It will also be interesting to test these and other S1PL inhibitors
against various bacterial S1PLs with a view to screen them for antibacterial activity.
To our knowledge, this is the first time that two highly-related S1PLs have been identified in the same
organism. At present, this phenomenon appears to be exclusive to species within the Burkholderia genus. The
Burkholderia genus encompasses a remarkably diverse and versatile group of bacteria including
environmentally-beneficial species as well as pathogens of animals, humans and plants. It is notable that our
search of available genomes identified that Burkholderia-encoded S1PLs are restricted to B. pseudomallei and
B. thailandensis, both of which are facultative intracellular species. This is consistent with the bacterial S1PLs
contributing to pathogenesis during infection of host cells, as has recently been described for Legionella-
2
4

encoded S1PL (57). The ability to invade, survive and replicate within host cells is central to the pathogenesis
of B. pseudomallei (60) and whilst B. thailandensis is a highly attenuated relative to B. pseudomallei, it displays
a similar intracellular phenotype (61) and as such is a valuable model organism for studying the intracellular
lifestyle of B. pseudomallei. SLs have been shown to be key players in cell homeostasis during stress such as
that caused by bacterial infection (62). They mediate crosstalk between apoptosis and autophagy, the two major
mechanisms of controlled cell death (63). Recently, studies on L. pneumophila, the causative agent of
Legionnaires’ disease, have sought to understand the role that the L. pneumophila S1PL plays in its ability to
survive intracellularly (57). It was found that L. pneumophila S1PL activity alone prevented an increase in
sphingosine levels in infected host cells and this led to an inhibition of autophagy during macrophage infection.
This work revealed the disruption of host SL biosynthesis as a novel mechanism used by intracellular pathogens
to inhibit autophagy.
Since Burkholderia appear to not produce sphingolipids such as S1P, the identity of the “true”
intracellular substrate(s) for these enzyme is the subject of conjecture. We have shown that S1PL2021 can also
degrade sphingosine as well as S1P, and it will be interesting to study a comprehensive range of potential
substrates with each S1PL isoform to define their specificity. Burkholderia would most likely encounter high
concentrations of S1P only during infection of a mammalian host and this would require the S1PLs to be
secreted. In a recent complementary study we have discovered that the two S1PL homologs present in B.
thailandensis (BTH_II0309 and BTH_II0311, Fig. S1) are found in the extracellular supernatant in a pH-
dependent manner (Custodio et al, Mol Microbiol. 2016 Sep 15. doi: 10.1111/mmi.13531). We are currently
investigating whether both these S1PLs form part of a higher order complex as pairs of homo- or hetero- dimers
but initial native gel electrophoresis suggests the presence of larger complexes. We also found that when
expressed alone, BTH_II0309 and BTH_II0311 could each partially restore resistance to 10 µM sphingosine
treatment of a yeast Dpl1 mutant and full resistance was displayed by cells expressing both S1PLs. Our future
work will determine whether the B. pseudomallei S1PL2021 and S1PL2025 can also complement the yeast Dpl1
mutant and are secreted in a similar way to the B. thailandensis isoforms. We studied the functional roles of the
Burkholderia S1PLs and found they were required for virulence in a Galleria wax moth larvae model and
survival in murine macrophages (Custodio et al.). Of particular relevance to the current study, we discovered
that a B. pseudomallei S1PL2025 mutant displayed a much reduced virulence phenotype (80% survival)
2
5

compared to the wild type organism (100% mortality) using a murine infection model in BALB/c mice. We are
currently constructing the B. pseudomallei S1PL2021 knock out and a double mutant to study their virulence
phenotypes. In summary, our work has characterised two S1PL homologs in B. pseudomallei and, when
combined with recent studies on the emerging roles of bacterial S1PLs in pathogenesis, suggests that these
enzymes merit further study.
Acknowledgements
The authors thanks the following for funding: The Biotechnology and Biological Sciences Research Council
(
BBSRC) for an EastBio Doctoral Training Programme PhD studentship award to C McLean (BB/J01446X/1)
and a grant awarded to DJ Campopiano (BB/I013687/1) that supported J Lowther and DJ Clarke. R Custodio
was supported by the Defence Science and Technology Laboratory under contract DSTLX-1000060221 (WP1).
We thank the staff of the Diamond Light Source, UK for help with data collection. The authors thank Prof. John
RW Govan (University of Edinburgh) for his suggestions regarding Burkholderia strains and enthusiastic
support of this work. We also thanks Dr. Kevin Ralston for help in the synthesis of 2E-HEX. The data associated
with this paper is available to download (http://dx.doi.org/10.7488/ds/1412).
2
6

References
1
. Merrill, A. H.. 2011. Sphingolipid and glycosphingolipid metabolic pathways in the era of sphingolipidomics.
Chem. Rev. 111: 6387–6422.
2
3
. Lingwood, D., and K. Simons. 2010. Lipid rafts as a membrane-organizing principle. Science. 327: 46–50.
. Fyrst, H., and J. D. Saba. 2010. An update on sphingosine-1-phosphate and other sphingolipid mediators.
Nat. Chem. Biol. 6: 489–497.
4
5
6
7
8
. Pyne, S., and N. J. Pyne. 2000. Sphingosine 1-phosphate signalling in mammalian cells. Biochem. J. 349:
385–402.
. Yester, J. W., E. Tizazu, K. B. Harikumar, and T. Kordula. 2011. Extracellular and intracellular sphingosine-
1-phosphate in cancer. Cancer Metastasis Rev. 30: 577–597.
. Spiegel, S., and S. Milstien. 2003. Sphingosine-1-phosphate: an enigmatic signalling lipid. Nat. Rev. Mol.
Cell Biol. 4: 397–407.
. Cuvillier, O., G. Pirianov, B. Kleuser, P. G. Vanek, O. A. Coso, S. Gutkind, and S. Spiegel. 1996. Suppression
of Ceramide-Mediated Programmed Cell Death by Sphingosine-1-Phosphate. Nature. 381: 800–803.
. Nakahara, K., A. Ohkuni, T. Kitamura, K. Abe, T. Naganuma, Y. Ohno, R. A. Zoeller, and A. Kihara. 2012.
The Sjrogren-Larsson Syndrome Gene Encodes a Hexadecenal Dehydrogenase of the Sphingosine 1-
Phosphate Degradation Pathway. Mol. Cell. 46: 461–471.
9
1
. Saba, J. D., F. Nara, A. Bielawska, S. Garrett, and Y. A. Hannun. 1997. Communicationsꢀ: The BST1 Gene
of Saccharomyces cerevisiae Is the sphingosine 1-phosphate lyase. 26087–26090.
0. Van Veldhoven, P. P., S. Gijsbers, G. P. Mannaerts, J. R. Vermeesch, and V. Brys. 2000. Human
sphingosine-1-phosphate lyase: cDNA cloning, functional expression studies and mapping to
chromosome 10q22(1). Biochim. Biophys. Acta. 1487: 128–134.
1
1. Weiler, S., N. Braendlin, C. Beerli, C. Bergsdorf, A. Schubart, H. Srinivas, B. Oberhauser, and A. Billich.
2014. Orally Active 7-Substituted (4-Benzylphthalazin-1-yl)-2- methylpiperazin-1-yl]nicotinonitriles as
Active-Site Inhibitors of Sphingosine 1-Phosphate Lyase for the Treatment of Multiple Sclerosis. J. Med.
Chem. 57: 5074–5084.
1
1
2. Martinez, M., and A. Ovalle. 2013. Sphingomonas paucimobilis. Rev. Chil. Infectol. 30: 107–110.
3. Keck, M., N. Gisch, H. Moll, F. J. Vorhölter, K. Gerth, U. Kahmann, M. Lissel, B. Lindner, K. Niehaus,
and O. Holst. 2011. Unusual outer membrane lipid composition of the gram-negative,
lipopolysaccharide-lacking myxobacterium Sorangium cellulosum So ce56. J. Biol. Chem. 286: 12850–
12859.
1
1
1
1
4. Wieland Brown, L. C., C. Penaranda, P. C. Kashyap, B. B. Williams, J. Clardy, M. Kronenberg, J. L.
Sonnenburg, L. E. Comstock, J. A. Bluestone, and M. A. Fischbach. 2013. Production ofꢀα-
Galactosylceramide by a Prominent Member of the Human Gut Microbiota. PLoS Biol. 11. e1001610.
5. An, D., C. Na, J. Bielawski, Y. a Hannun, and D. L. Kasper. 2011. Membrane sphingolipids as essential
molecular signals for Bacteroides survival in the intestine. Proc. Natl. Acad. Sci. U. S. A. 108 Suppl :
4666–4671.
6. Mun, J., A. Onorato, F. C. Nichols, M. D. Morton, A. I. Saleh, M. Welzel, and M. B. Smith. 2007. Structural
confirmation of the dihydrosphinganine and fatty acid constituents of the dental pathogen
Porphyromonas gingivalis. Org. Biomol. Chem. 5: 3826–3833.
7. Degtyar, E., T. Zusman, M. Ehrlich, and G. Segal. 2009. A Legionella effector acquired from protozoa is
involved in sphingolipids metabolism and is targeted to the host cell mitochondria. Cell. Microbiol. 11:
1219–1235.
2
7

1
8. Rolando, M., P. Escoll, T. Nora, J. Botti, V. Boitez, C. Bedia, C. Daniels, G. Abraham, P. J. Stogios, T.
Skarina, C. Christophe, D. Dervins-Ravault, C. Cazalet, H. Hilbi, T. W. T. Rupasinghe, D. Tull, M. J.
McConville, S. Y. Ong, E. L. Hartland, P. Codogno, T. Levade, T. Naderer, A. Savchenko, and C.
Buchrieser. 2016. Legionella pneumophila S1P-lyase targets host sphingolipid metabolism and restrains
autophagy. Proc. Natl. Acad. Sci. 113: 1901-1906.
1
2
9. White, N. J. 2003. Melioidosis. Lancet. 361: 1715–1722.
0. Winsor, G. L., B. Khaira, T. Van Rossum, R. Lo, M. D. Whiteside, and F. S. L. Brinkman. 2008. The
Burkholderia Genome Database: Facilitating flexible queries and comparative analyses. Bioinformatics.
24: 2803–2804.
2
1. Holden, M. T. G., R. W. Titball, S. J. Peacock, A. M. Cerdeño-Tárraga, T. Atkins, L. C. Crossman, T. Pitt,
C. Churcher, K. Mungall, S. D. Bentley, M. Sebaihia, N. R. Thomson, N. Bason, I. R. Beacham, K.
Brooks, K. A. Brown, N. F. Brown, G. L. Challis, I. Cherevach, T. Chillingworth, A. Cronin, B. Crossett,
P. Davis, D. DeShazer, T. Feltwell, A. Fraser, Z. Hance, H. Hauser, S. Holroyd, K. Jagels, K. E. Keith,
M. Maddison, S. Moule, C. Price, M. A. Quail, E. Rabbinowitsch, K. Rutherford, M. Sanders, M.
Simmonds, S. Songsivilai, K. Stevens, S. Tumapa, M. Vesaratchavest, S. Whitehead, C. Yeats, B. G.
Barrell, P. C. F. Oyston, and J. Parkhill. 2004. Genomic plasticity of the causative agent of melioidosis,
Burkholderia pseudomallei. Proc. Natl. Acad. Sci. U. S. A. 101: 14240–14245.
2
2
2
2
2. Van Veldhoven, P. P., and G. P. Mannaerts. 1991. Subcellular localization and membrane topology of
sphingosine-1-phosphate lyase in rat liver. J. Biol. Chem. 266: 12502–12507.
3. Bandhuvula, P., H. Fyrst, and J. D. Saba. 2007. A rapid fluorescence assay for sphingosine-1-phosphate
lyase enzyme activity. J. Lipid Res. 48: 2769–2778.
4. Bandhuvula, P., Z. Li, R. Bittman, and J. D. Saba. 2009. Sphingosine 1-phosphate lyase enzyme assay using
a BODIPY-labeled substrate. Biochem. Biophys. Res. Commun. 380: 366–370.
5. Bedia, C., L. Camacho, J. Casas, J. L. Abad, A. Delgado, P. P. Van Veldhoven, and G. Fabriàs. 2009.
Synthesis of a fluorogenic analogue of sphingosine-1-phosphate and its use to determine sphingosine-1-
phosphate lyase activity. Chembiochem. 10: 820–822.
2
6. Berdyshev, E. V, J. Goya, I. Gorshkova, G. D. Prestwich, H.-S. Byun, R. Bittman, and V. Natarajan. 2011.
Characterization of sphingosine-1-phosphate lyase activity by electrospray ionization-liquid
chromatography/tandem mass spectrometry quantitation of (2E)-hexadecenal. Anal. Biochem. 408: 12–
18.
2
7. Reina, E., L. Camacho, J. Casas, P. P. Van Veldhoven, and G. Fabrias. 2012. Determination of sphingosine-
1-phosphate lyase activity by gas chromatography coupled to electron impact mass spectrometry. Chem.
Phys. Lipids. 165: 225–231.
2
8. Lüth, A., C. Neuber, and B. Kleuser. 2012. Novel methods for the quantification of (2E)-hexadecenal by
liquid chromatography with detection by either ESI QTOF tandem mass spectrometry or fluorescence
measurement. Anal. Chim. Acta. 722: 70–79.
2
3
3
3
9. Liu, Z., Y. Gong, H. Byun, and R. Bittman. 2010. An improved two-step synthetic route to primary allylic
alcohols from aldehydes. New J. Chem. 34: 470–475.
0. Chenna, R.. 2003. Multiple sequence alignment with the Clustal series of programs. Nucleic Acids Res. 31:
3497–3500.
1. Kelley, L. A., and M. J. E. Sternberg. 2009. Protein structure prediction on the Web: a case study using the
Phyre server. Nat. Protoc. 4: 363–371.
2. Liu, H., and J. H. Naismith. 2008. An efficient one-step site-directed deletion, insertion, single and multiple-
site plasmid mutagenesis protocol. BMC Biotechnol. 8: 91.
3
3
3. Kabsch, W.. 2010. Xds. Acta Crystallogr. Sect. D Biol. Crystallogr. 66: 125–132.
4. Evans, P. R.. 2011. An introduction to data reduction: Space-group determination, scaling and intensity
statistics. Acta Crystallogr. Sect. D Biol. Crystallogr. 67: 282–292.
2
8

3
3
3
3
5. Karplus, A. P. and K. Diederichs. 2012. Linking Crystallographic Model and Data Quality. Science 336:
1030–1033.
6. Stein, N.. 2008. CHAINSAW: A program for mutating pdb files used as templates in molecular replacement.
J. Appl. Crystallogr. 41: 641–643.
7. McCoy, A. J., R. W. Grosse-Kunstleve, P. D. Adams, M. D. Winn, L. C. Storoni, and R. J. Read. 2007.
Phaser crystallographic software. J. Appl. Crystallogr. 40: 658–674.
8. Terwilliger, T. C., R. W. Grosse-Kunstleve, P. V. Afonine, N. W. Moriarty, P. H. Zwart, L. W. Hung, R. J.
Read, and P. D. Adams. 2007. Iterative model building, structure refinement and density modification
with the PHENIX AutoBuild wizard. Acta Crystallogr. Sect. D Biol. Crystallogr. 64: 61–69.
3
4
9. Afonine, P. V., R. W. Grosse-Kunstleve, N. Echols, J. J. Headd, N. W. Moriarty, M. Mustyakimov, T. C.
Terwilliger, A. Urzhumtsev, P. H. Zwart, and P. D. Adams. 2012. Towards automated crystallographic
structure refinement with phenix.refine. Acta Crystallogr. Sect. D Biol. Crystallogr. 68: 352–367.
0. Adams, P. D., P. V. Afonine, G. Bunkóczi, V. B. Chen, I. W. Davis, N. Echols, J. J. Headd, L. W. Hung, G.
J. Kapral, R. W. Grosse-Kunstleve, A. J. McCoy, N. W. Moriarty, R. Oeffner, R. J. Read, D. C.
Richardson, J. S. Richardson, T. C. Terwilliger, and P. H. Zwart. 2010. PHENIX: A comprehensive
Python-based system for macromolecular structure solution. Acta Crystallogr. Sect. D Biol. Crystallogr.
66: 213–221.
4
1. Emsley, P., B. Lohkamp, W. G. Scott, and K. Cowtan. 2010. Features and development of Coot. Acta
Crystallogr. Sect. D Biol. Crystallogr. 66: 486–501.
4
2. Chen, V. B., W. B. Arendall, J. J. Headd, D. A. Keedy, R. M. Immormino, G. J. Kapral, L. W. Murray, J.
S. Richardson, and D. C. Richardson. 2010. MolProbity: All-atom structure validation for
macromolecular crystallography. Acta Crystallogr. Sect. D Biol. Crystallogr. 66: 12–21.
4
4
3. The PyMol Molecular Graphics System, Version 1.8, Schrodinger, LLC.
4. Bourquin, F., H. Riezman, G. Capitani, and M. G. Grütter. 2010. Structure and function of sphingosine-1-
phosphate lyase, a key enzyme of sphingolipid metabolism. Structure. 18: 1054–1065.
4
4
5. Haraga, A., T. E. West, M. J. Brittnacher, S. J. Skerrett, and S. I. Miller. 2008. Burkholderia thailandensis
as a model system for the study of the virulence-associated type III secretion system of burkholderia
pseudomallei. Infect. Immun. 76: 5402–5411.
6. Lloyd, M. D., K. D. E. Boardman, A. Smith, D. M. van Den Brink, R. J. A. Wanders, and M. D. Threadgill.
2007. Characterisation of recombinant human fatty aldehyde dehydrogenase: implications for Sjögren-
Larsson syndrome. J. Enzyme Inhib. Med. Chem. 22: 584–590.
4
4
4
5
5
5
5
5
7. Obeid, L. M., Y. Okamoto, and C. Mao. 2002. Yeast sphingolipids: Metabolism and biology. Biochim.
Biophys. Acta - Mol. Cell Biol. Lipids. 1585: 163–171.
8. Backhed, F., R. E. Ley, J. L. Sonnenburg, D. A. Peterson, and J. I. Gordon. 2005. Host-Bacterial Mutualism
in the Human Intestine. Science 307: 1915–1920.
9. Schweizer, H. P.. 2012. Mechanisms of antibiotic resistance in Burkholderia pseudomallei: implications for
treatment of melioidosis Herbert P Schweizer. Future Microbiol. 7: 1389–1399.
0. Mahenthiralingam, E., T. A. Urban, and J. B. Goldberg. 2005. The multifarious, multireplicon Burkholderia
cepacia complex. Nat. Rev. Microbiol. 3: 144–156.
1. Vandamme, P, J. Govan, and J. LiPuma. 2007. Diversity and Role of Burkholderia spp. In Burkholderia:
Molecular Microbiology and Genomics. T. Coenyne and P. Vandamme, Horizon Bioscience, 1–28.
2. Stoffel, W. Lekim, D. Sticht, G.. 1969. Distribution and properties of dihydrosphingosine-1-phosphate
aldolase. Hoppe-Seyler’s Physio. Chem. 350: 1233–1241.
3. Suh, J. H., A. Eltanawy, A. Rangan, and J. D. Saba. 2016. A facile stable-isotope dilution method for
determination of sphingosine phosphate lyase activity. Chem. Phys. Lipids. 194: 101–109.
4. Keller, M. A., U. Zander, J. E. Fuchs, C. Kreutz, K. Watschinger, T. Mueller, G. Golderer, K. R. Liedl, M.
2
9

Ralser, B. Kräutler, E. R. Werner, and J. A. Marquez. 2014. A gatekeeper helix determines the substrate
specificity of Sjögren-Larsson Syndrome enzyme fatty aldehyde dehydrogenase. Nat. Commun. 5: 4439.
5
5. Christen P. and P. K. Mehta. 2001. From cofactor to enzymes. The molecular evolution of pyridoxal-5’-
phosphate-dependent enzymes. Chem. Rec. 1: 436–447.
5
6. Weiler, S., N. Braendlin, C. Beerli, C. Bergsdorf, A. Schubart, H. Srinivas, B. Oberhauser, and A. Billich.
2014. Orally Active 7-Substituted (4-Benzyl-phthalazin-1-yl)-2-methyl-piperazin-1-yl]-nicotinonitriles
as Active-site Inhibitors of Sphingosine-1-Phosphate Lyase for the Treatment of Multiple Sclerosis. J.
Med. Chem. 57: 5074–5084.
5
7. Rolando, M., P. Escoll, T. Nora, J. Botti, V. Boitez, C. Bedia, C. Daniels, G. Abraham, P. J. Stogios, T.
Skarina, C. Christophe, D. Dervins-Ravault, C. Cazalet, H. Hilbi, T. W. T. Rupasinghe, D. Tull, M. J.
McConville, S. Y. Ong, E. L. Hartland, P. Codogno, T. Levade, T. Naderer, A. Savchenko, and C.
Buchrieser. 2016. Legionella pneumophila S1P-lyase targets host sphingolipid metabolism and restrains
autophagy. Proc. Natl. Acad. Sci. U. S. A. 113: 1901–1906.
5
5
8. Wadsworth, J. M., D. J. Clarke, S. A. Mcmahon, J. P. Lowther, A. E. Beattie, P. R. R. Langridge-Smith, H.
B. Broughton, T. M. Dunn, J. H. Naismith, and D. J. Campopiano. 2013. The chemical basis of serine
palmitoyltransferase inhibition by myriocin. J. Am. Chem. Soc. 135: 14276–14285.
9. Dinges, J., C. M. Harris, G. A. Wallace, M. A. Argiriadi, K. L. Queeney, D. C. Perron, E. Dominguez, T.
Kebede, K. E. Desino, H. Patel, and A. Vasudevan. 2016. Hit-to-lead evaluation of a novel class of
sphingosine 1-phosphate lyase inhibitors. Bioorganic Med. Chem. Lett. 26: 2297–2302.
6
0. Jones, A. L., T. J. Beveridge, and D. E. Woods. 1996. Intracellular survival of Burkholderia pseudomallei.
Infect.Immun. 64: 782–790.
6
1. Wand, M. E., C. M. Müller, R. W. Titball, and S. L. Michell. 2011. Macrophage and Galleria mellonella
infection models reflect the virulence of naturally occurring isolates of B. pseudomallei, B. thailandensis
and B. oklahomensis. BMC Microbiol. 11: 11.
6
6
2. Hannun, Y. A. 1996. Functions of ceramide in coordinating cellular responses to stress. Science. 274: 1855–
1859.
3. Young, M. M., M. Kester, and H.-G. Wang. 2013. Sphingolipids: regulators of crosstalk between apoptosis
and autophagy. J. Lipid Res. 54: 5–19.
3
0