A multi-omics strategy resolves the elusive nature of alkaloids in Podophyllum species

Article abstract of DOI:10.1039/c4mb00403e

Podophyllum hexandrum and, to a much lesser extent P. peltatum, are sources of podophyllotoxin, extensively used as a chemical scaffold for various anti-cancer drugs. In this study, integrated omics technologies (including advanced mass spectrometry/metabolomics, transcriptome sequencing/gene assemblies, and bioinformatics) gave unequivocal evidence that both plant species possess a hitherto unknown aporphine alkaloid metabolic pathway. Specifically, RNA-seq transcriptome sequencing and bioinformatics guided gene assemblies/analyses in silico suggested presence of transcripts homologous to genes encoding all known steps in aporphine alkaloid biosynthesis. A comprehensive metabolomics analysis, including UPLC-TOF-MS and MALDI-MS imaging in situ, then enabled detection, identification, localization and quantification of the aporphine alkaloids, magnoflorine, corytuberine and muricinine, in the underground and aerial tissues. Interestingly, the purported presence of alkaloids in Podophyllum species has been enigmatic since the 19th century, remaining unresolved until now. The evolutionary and phylogenetic ramifications of this discovery are discussed. This journal is

Full text of DOI:10.1039/c4mb00403e

Volume 10 Number 11 November 2014 Pages 2737–3032
Molecular
BioSystems
Interfacing chemical biology with the -omic sciences and systems biology
www.molecularbiosystems.org
ISSN 1742-206X
PAPER
Norman G. Lewis et al.
A multi-omics strategy resolves the elusive nature of alkaloids in
Podophyllum species

Molecular
BioSystems
View Article Online
View Journal | View Issue
PAPER
A multi-omics strategy resolves the elusive nature
of alkaloids in Podophyllum species†
Cite this: Mol. BioSyst., 2014,
10, 2838
Joaquim V. Marques, Doralyn S. Dalisay, Hong Yang, Choonseok Lee,
Laurence B. Davin and Norman G. Lewis*
Podophyllum hexandrum and, to a much lesser extent P. peltatum, are sources of podophyllotoxin, extensively
used as a chemical scaffold for various anti-cancer drugs. In this study, integrated omics technologies (including
advanced mass spectrometry/metabolomics, transcriptome sequencing/gene assemblies, and bioinformatics)
gave unequivocal evidence that both plant species possess a hitherto unknown aporphine alkaloid metabolic
pathway. Specifically, RNA-seq transcriptome sequencing and bioinformatics guided gene assemblies/analyses
in silico suggested presence of transcripts homologous to genes encoding all known steps in aporphine
alkaloid biosynthesis. A comprehensive metabolomics analysis, including UPLC-TOF-MS and MALDI-MS
imaging in situ, then enabled detection, identification, localization and quantification of the aporphine alkaloids,
magnoflorine, corytuberine and muricinine, in the underground and aerial tissues. Interestingly, the purported
presence of alkaloids in Podophyllum species has been enigmatic since the 19th century, remaining unresolved
until now. The evolutionary and phylogenetic ramifications of this discovery are discussed.
Received 11th July 2014,
Accepted 15th August 2014
DOI: 10.1039/c4mb00403e
www.rsc.org/molecularbiosystems
Introduction
Podophyllum hexandrum and P. peltatum from the Berberidaceae
family are best known as sources of the medicinal aryltetralin
lignan, (ꢀ)-podophyllotoxin (PTOX 1, Fig. 1).1 It is, however,
generally commercially extracted from P. hexandrum rhizomes,
and is extensively used worldwide for semi-synthesis of the
topoisomerase II inhibitor anticancer drugs, teniposide (2),
etopophoss (3), and etoposide (4).2
Recently two cytochrome P450s (CyP450s) were reported from
P. hexandrum (CYP719A23) and P. peltatum (CYP719A24), capable
of facile methylenedioxy bridge formation, in the conversion of
(ꢀ)-matairesinol (18) into (ꢀ)-pluviatolide (19) (Fig. 2b), i.e. with the
latter being a proposed pathway intermediate to PTOX (1).3 However,
phylogenetic comparison of their corresponding proteins indicated
they were more closely aligned to the CYP719A family known to
catalyse an analogous reaction in isoquinoline alkaloid-producing
plant species,4 i.e. CYP719A23 and CYP719A24 share B59% identity
and B78% similarity to Coptis japonica CYP719A1. Both of these
CyP450s also have lower sequence identity/similarity (B26/B46%) to
CYP81Q1, which is reportedly involved in sesame lignan methylene-
dioxy bridge formation.5 Furthermore, in P. hexandrum, there are
Institute of Biological Chemistry, Washington State University, Pullman,
WA 99164-6340, USA. E-mail: lewisn@wsu.edu; Fax: +1 509 335 8206;
Tel: +1 509 335 2682
Fig. 1 Podophyllotoxin (1) scaffold and its anticancer drugs/derivatives.
† Electronic supplementary information (ESI) available: Purification of Podophyllum
hexandrum norcoclaurine synthase (Fig. S1) and expression levels of norcoclaurine
6-O-methyltransferase, corytuberine synthase and coclaurine N-methyltransferase in
P. hexandrum and P. peltatum (Fig. S2). See DOI: 10.1039/c4mb00403e
another two CyP450s, CYP81B56 and CYP81B57, with higher identity/
similarity (46/63% and 49/66%, respectively) to the CYP81Q1 involved
2838 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
Fig. 2 Alkaloid and lignan pathways. (a) Magnoflorine (13), (b) podophyllotoxin (1), and (c) sesamin (23) biosynthetic pathways. (d) Structures of
muricinine (24), berberine (25) and syringaresinol (26). Bold arrows represent known biosynthetic steps and dashed arrows represent currently unknown
steps. CNMT, coclaurine N-methyltransferase; CS, corytuberine synthase; DP, dirigent protein, HNMC 4OMT, 30-hydroxy-N-methyl-(S)-coclaurine
40-O-methyltransferase; NC 6OMT, norcoclaurine 6-O-methyltransferase; NCS, norcoclaurine synthase; NM C3H, N-methylcoclaurine 30-hydroxylase;
PLR, pinoresinol–lariciresinol reductase; PLS, pluviatolide synthase; PSS, piperitol/sesamin synthase; SDH, secoisolariciresinol dehydrogenase.
in the piperitol/sesamin (21/22) biosynthetic pathway that we had (Athens, OH), respectively, and maintained in Washington State
earlier discovered.6 However, the corresponding recombinant pro- University greenhouse facilities.
teins were catalytically inactive in our assays with matairesinol (18)
Chemicals
and other putative intermediates tested.
Using a combination of transcriptomics, metabolomics and
bioinformatics approaches, we discovered a hitherto undetected
metabolic pathway in both Podophyllum species, which afford
various aporphine alkaloids (Fig. 2a) ꢀ albeit those lacking a
methylenedioxy bridge functionality. In addition, we report the
successful use of MALDI metabolite imaging mass spectrometry,
together with laser capture microdissection (for total RNA
extraction), to determine spatial and temporal distribution of
aporphine alkaloids and transcripts responsible for their bio-
synthesis. Our findings suggest evolutionary linkages between
both lignan and alkaloid biosynthetic pathways.
Dopamine (5) was purchased from Sigma-Aldrich. Magnoflorine
(13), corytuberine (12) and norcoclaurine (7) authentic standards
were kindly provided by Dr Toni Kutchan (Donald Danforth
Plant Science Center).
4-Hydroxyphenylacetaldehyde (6) was produced from 4-hydroxy-
phenylethanol in a one-step reaction in 95% crude yield as follows:
4-hydroxyphenylethanol (15 mmol) in dichloromethane (10 ml)
was added to dichloromethane (200 ml) containing pyridinium
chlorochromate (20 mmol), over ice. Following a 2 h reaction,
during which the temperature reached room temperature, the
reaction mixture was filtered, with the supernatant evaporated
in vacuo, to afford the crude product, which was used without
further purification.
Experimental details
Plant material
Transcriptome sequencing and library assembly
Podophyllum hexandrum and P. peltatum plants were obtained Transcriptome sequencing, assemblies, and analyses were per-
from Digging Dog Nursery (Albion, CA) and Companion Plants formed as described previously.3
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2839

View Article Online
Molecular BioSystems
Paper
Metabolite extraction and analyses
100 to 1000 m/z with data obtained in the positive ion mode at
1000 Hz, the laser energy set at 350, and using leucine–
enkephalin as lock mass. The data was obtained at either
30 or 22 mm spatial resolution. The raw data was then processed
and ion maps were visualized in HDImagingt (v1.1) software.
Three hundred of the most intense peaks were obtained after
processing, with magnoflorine (13) readily identifiable. Collision
induced dissociation MS/MS analysis using 30 V collision energy
gave fragmentation patterns identical to authentic magnoflorine
(13). Ion mobility spectrometry was also used to obtain the drift
time of the ions identified in MALDI-TOF IMS.
Metabolite extraction was carried out as before.3 Briefly, different
tissues (i.e. rhizomes, stems and leaves) were individually harvested,
immediately frozen in liq. N2, ground and subsequently lyophilized.
After passage through a 150 mm sieve, each sample was individually
extracted with 10 ml mgꢀ1 methanol–water (7 : 3, v/v) for 5 min
in an ice bath and sonication; the corresponding extracts were
then maintained at ꢀ80 1C until analysis. For alkaloid metabolite
detection and isolation, samples were subjected to an Acquity
ultra performance liquid chromatography (UPLC, Waters) system,
utilizing a BEH C18 column (Waters, 1.7 mm particles, 2.1 ꢁ
50 mm) with a binary mobile phase of formic acid–water (0.1%)
(A) and formic acid–acetonitrile (0.1%) (B); detection was carried
out at both 280 nm and by use of electrospray ionization mass
spectrometry in the positive ion mode. A linear gradient (A : B) was
used at a flow rate of 0.3 ml minꢀ1: from 95:5 to 75:25 in 11 min,
to 60:40 in 5 min, to 0 :100 in 4 min, followed by 1.5 min at
0 : 100. The column temperature was held at 25 1C and the sample
injection volume was 5 ml. Masses were determined using a
Waters Xevo G2 Q-TOF mass spectrometer using leucine–
enkephalin, C28H37N5O7 with m/z 556.2771 [M + H]+ as a lock-
mass standard.
Toluidine blue O (TBO) stain for imaged sections
After MALDI imaging analysis, the MALDI plate was rinsed with
ethanol to remove the matrix. The tissue section was then
flooded with an aqueous 0.1% TBO solution for 3 min, and next
successively washed with water and ethanol. The TBO stained
tissue section was scanned using a standard flatbed scanner
(Epson Perfection V500 photo, Japan).
Gene cloning and sequencing
Putative alkaloid biosynthetic genes, with the exception of
30-hydroxy-N-methyl-(S)-coclaurine 40-O-methyltransferase, were
amplified from P. hexandrum cDNA using the primers described
in Table 1. Amplification was performed in a thermocycler
using PfuTurboDNA polymerase (Agilent Technologies) with
35 cycles of 94 1C for 30 s, 55 1C for 30 s, and 70 1C for 3 min,
and a final extension at 70 1C for 10 min. PCR products were
resolved in 1% agarose gels. Bands were extracted using QIAquick
Gel Extraction Kit (QIAGEN) and cloned into pENTRt/D-TOPOs,
following the manufacturer’s instructions. Sequencing was
performed by Eurofins. All sequences were deposited in the
GenBankt Data Bank under accession numbers KJ786954–
KJ794165 (Table 2 and Fig. 3 for NCS).
MALDI-imaging mass spectrometry (IMS)
Fresh plant sample tissue of a P. hexandrum rhizome and
P. peltatum rhizome and root tissues were obtained by cryo-
sectioning. Briefly, fresh tissues were embedded in ultra-pure
agarose (3%) and cryo-sectioned at 15 mm thickness. The cryo-
sections were placed on a MALDI target plate and 2,5-dihydroxy-
benzoic acid was applied at 40 mg mlꢀ1 (in methanol–water,
1 : 1, v/v) by robotic sprayer (TM-sprayert, HTX Technologies,
Carrboro, NC). The acquisition mass range was then set from
Table 1 Primers used for sequence confirmation
P. hexandrum norcoclaurine synthase (NCS) heterologous
expression and activity determination
Gene
Forward/reverse primers
Norcoclaurine synthase
ATGAGGATGGAAATTGTTTCCC
TTATATTTCAATCAAGCTCTTATTCTTGT After confirmation of its DNA sequence in the pENTR/D-TOPO
Norcoclaurine 6-O-
methyltransferase
Coclaurine
N-methyltransferase
N-Methylcoclaurine
30-hydroxylase
ATGGAAGCTCAAAAGGAGAACATATC
cloning vector, a D9 truncated PhNCS with addition of the
ATAAGGGAAAGCCTCAATTACAGA
ATGGCCGTTACTGAAGATTCC
MTGS7 signal peptide was generated by PCR using forward
TCATTTCCTGGTGAGTAGAAGATGA
ATGGATTCCCTGCACTGCC
(50-CACCATGACCGGTTCGCTATTTTTCTTAGCATGCGGAA-30) and
reverse (50-TATTTCAATCAAGCTCTTATTCTTGTT-30) primers.
This was subsequently cloned into the pET101/D expression
vector (Invitrogen) by the TOPO reaction, according to the
CTAGACAAAACATTTTGGAGAGATGC
ATGGAGCTAACAACAGCACTGC
CTACACCCTGGATTTCGGAATG
Corytuberine synthase
Table 2 GenBankt accession numbers
P. hexandrum
P. peltatum
GenBankt
accession
number
GenBankt
accession
number
Gene
Name
Name
Norcoclaurine synthase
PhNCS
KJ794165
KJ786954
KJ786956
KJ786958
KJ786960
KJ786962
PpNCS
KJ786964
KJ786955
KJ786957
KJ786959
KJ786961
KJ786963
Norcoclaurine 6-O-methyltransferase
Coclaurine N-methyltransferase
N-Methylcoclaurine 30-hydroxylase
30-Hydroxy-N-methyl-(S)-coclaurine 40-O-methyltransferase
Corytuberine synthase
PhNC6OMT
PhCNMT
PhNMC3H
PhHNMC4OMT
PhCS
PpNC6OMT
PpCNMT
PpNMC3H
PpHNMC4OMT
PpCS
2840 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
Fig. 3 Amino acid alignment of norcoclaurine synthase (NCS) genes. Argemone mexicana (gb|ACJ76785.1|), Papaver somniferum (gb|ACI45396.1|),
Thalictrum flavum (sp|Q67A25.1|) and the homologs found in P. hexandrum (PhNCS, accession KJ794165) and P. peltatum (PpNCS, accession KJ786964),
respectively.
manufacturer’s instructions. D9 PhNCS was next heterologously Laser microscope dissection of xylem, pith and cortex regions
expressed in BL21 (DE3) (Novagen) cells after induction with 1 mM from rhizomes
isopropyl b-D-1-thiogalactopyranoside at 30 1C.
Laser capture microdissection was used to dissect xylem, pith
and cortex cells from rhizomes of both P. hexandrum and
After 19 h induction, cells were harvested and recovered
after centrifugation (3900 ꢁ g) with pellets re-suspended in
P. peltatum using a PALM MicroBeam System (Zeiss Microscopy)
as described previously.9 Rhizomes were embedded with Tissue-
BugBuster Protein Extraction Reagent (Novagen) according to
the manufacturer’s instructions. Cell debris was removed by
Tek OCT (Miles, Inc., Elkhart, IN) compound. The frozen cryocut
centrifugation (10 000 ꢁ g for 20 min). Supernatants were
sections (10 mm) were obtained from a Leica CM 1950 Cryostat,
individually subjected to POROS MC 20 column (1 ꢁ 9 cm)
and then mounted on activated UV PEN slides (Applied Biosystems).
Slides were placed in cold water for 1 min to remove OCT, dried
chromatography with Ni2+ (NiSO4) as metal ion.8 After washing
with buffer A (5 mM Tris-HCl, pH 7.9, 500 mM NaCl, 50 ml)
containing 5 mM imidazole, recombinant NCS was eluted
in vacuo and stored at ꢀ20 1C. Each slide was subjected to
ethanol dehydration steps using 70%, 95% ethanol in water
with a stepwise imidazole gradient (5–500 mM, 100 ml) in
and 100% ethanol for 1 min and air dried before laser micro-
buffer A. Fractions containing recombinant NCS (i.e. eluting
dissection. For each region (i.e. xylem, pith and cortex) approxi-
between 100 and 400 mM imidazole) were buffer-exchanged
mately 8000 dissected cells were collected, combined and kept
to 100 mM Tris-HCl buffer, pH 7.0, and concentrated
in ꢀ80 1C until further use.
(B0.5 ml) using a Centricon Plus-20 (Amicon, Billerica, MA).
Fractions eluted with 300 mM, and containing NCS purified to
apparent homogeneity (Fig. S1, ESI†), were utilized in the
RNA extraction, cDNA synthesis and gene expression analysis
assays. All enzyme manipulations and purifications were carried Rhizomes, roots and shoots were freshly harvested from
out at 4 1C.
P. hexandrum and P. peltatum and immediately frozen in liq.
Enzymatic assays were performed at 37 1C in 220 ml final N2 until use. Total RNA was isolated from each organ indivi-
volume of 100 mM Tris-HCl buffer, pH 7.0, containing 5 mg of dually using Plant RNA Purification Reagent (Invitrogen) with
purified recombinant NCS, dopamine (5) and 4-hydroxyphenyl- contaminating genomic DNA digested with RNase-free DNase
acetaldehyde (6) at a final concentration of 5 mM. Enzymatic (Invitrogen). Total RNA was also extracted from dissected cells
reactions were carried out for 30 min and stopped with addition with Qiagen RNeasy Micro Kit (QIAGEN) according to the
of glacial acetic acid (10 ml).
manufacturer’s instructions.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2841

View Article Online
Molecular BioSystems
Paper
First strand cDNA was next synthesized using SuperScript III to genes possibly encoding an aporphine alkaloid biosynthetic
First-Strand Synthesis System (Invitrogen) with DNase treated pathway (Fig. 2a). That is, there were putative P. hexandrum and
RNA (1 mg for rhizome, root and shoot or 500 pg for dissected P. peltatum norcoclaurine synthases (NCS; Fig. 3 and Table 3)
cells) as template according to the manufacturer’s instructions. with circa 74/56% similarity and 59/38% identity to Thalictrum
PCR mixtures (20 ml) for gene expression analysis consisted of flavum NCS (UniProtKB accession Q67A25.1) affording entry
cDNA (1 ml), Platinums SYBRs Green qPCR SuperMix-UDG into the magnoflorine (13) pathway (Fig. 2a). In an analogous
(Invitrogen, 10 ml), Rox dye (Invitrogen, 0.4 ml), nuclease-free manner, homologs for all of the remaining steps were also
water (7.8 ml) and gene specific primers (0.4 ml of each). Primers identified (see Table 3). These genes, with the exception of
were designed using Primer3 software with melting temperature PhHNMC4OMT, were cloned from P. hexandrum and their
of 57–58 1C: for P. hexandrum, PhNCS-For (50-GTGCTTCACC sequences again demonstrated the validity of our assemblies.
ATGAGTCCAC-30) and PhNCS-Rev (50-TTCAACCTTCTCGAAT These analyses thus suggested the presence of a previously
GCAC-30), for P. peltatum PpNCS-For (50-TTTCCACAAGGAA undetected aporphine alkaloid biosynthetic pathway operative
CACCTC TC-30) and PpNCS-Rev (50-CAAACTCTAGGAATCCTC in both Podophyllum species.
CTTCA-30), as well as Actin-For (50-TCAGGCTGTTCTTTC
CCTTT-30) and (Actin-Rev 50-GGAAGGGCATAACCTT CGTA-30).
Triplicate reactions were performed using cDNAs synthesized
from independent total RNA isolations as follows: 95 1C for
15 min followed by 40 cycles of 94 1C for 30 s, 58 1C for 1 min,
and 72 1C for 30 s. Dissociation curves were obtained with
products confirmed as single products by running agarose gels.
Relative expression levels were normalized against P. hexandrum
and P. peltatum actin genes.
Norcoclaurine synthase activity
In order to further confirm the authenticity of the alkaloid bio-
synthetic pathway provisionally detected, the putative P. hexandrum
NCS (Fig. 3) was heterologously expressed in E. coli as a D9
truncated [i.e. lacking a putative signal peptide13], C-terminus
His-tag, recombinant protein. After purification by Ni affinity
chromatography (Fig. S1, ESI†), it was found capable of con-
verting dopamine (5) and 4-hydroxy-phenylacetaldehyde (6) into
norcoclaurine (7), with the product identified by comparison
with an authentic standard (Fig. 4). That is, norcoclaurine (7)
had an elution time of ca. 3.5 min (Fig. 4a) and a mass spectrum
with m/z 272.1272, corresponding to [M + H]+ (calc. 272.1287,
C16H18NO3) and characteristic fragments of m/z 255.1007
(calc. 255.1016, [M + H–NH3]+), 237.0909 (calc. 237.0915,
[M + H–NH3–H2O]+), 161.0591 (calc. 161.0602, [C10H9O2]+) and
107.0490 (calc. 107.0497, [C7H7O]+) (Fig. 4d). The enzymatically
produced norcoclaurine (7) had the same elution time (Fig. 4b)
and a near identical mass spectrum (Fig. 4e), whereas with a boiled
enzyme assay (Fig. 4c), norclaurine (7) was not detectable.
Results
Our earlier Podophyllum sp. transcriptome profiling of aerial and
underground tissues, followed by gene assembly and bio-
informatics analysis,3 established the validity of our assembly
processes for previously identified genes encoding known
biochemical steps to PTOX (1). This not only confirmed that
the assemblies for known biochemical steps were identical to
earlier known gene sequences,10–12 but it also resulted in our
identifying several putative gene candidates for remaining
(unknown) enzymatic steps.
Metabolomics-guided analysis of Podophyllum species
However, the finding that Podophyllum species encoded genes
more closely related on sequence similarity/identity to methylene- Metabolomic profiles from different tissues were next more
dioxy bridge forming CyP450s in alkaloid formation, rather than in-depth analysed, with an emphasis ꢀ based on the provisionally
lignan biosynthesis, raised the question of whether our gene annotated gene sequences above ꢀ on benzylisoquinoline
assemblies contained additional potential alkaloid pathway genes. alkaloids, using UPLC-TOF-MS high sensitivity and accuracy.
Accordingly, we re-examined the assemblies and transcripts These analyses resulted in detection of 3 peaks whose mass
from P. hexandrum and P. peltatum rhizome, root, stem and leaf spectral fragmentation patterns were consistent for aporphine
tissues (www.uic.edu/pharmacy/MedPlTranscriptome/database. alkaloids14,15 (Fig. 5). The most abundant peak in both
html). This resulted in several contigs being detected homologous P. hexandrum (Fig. 5a) and P. peltatum (Fig. 5e) was identified
Table 3 Alkaloid biosynthetic pathway homolog
Pathway genes
Species
Similarity/identity (%)
Data bank accession
number
P. hexandrum/P. peltatum homologs
P. hexandrum
P. peltatum
Norcoclaurine synthase (NCS)
Norcoclaurine 6-O-methyltransferase (NC 6OMT)
Coclaurine N-methyltransferase (CNMT)
Thalictrum flavum
T. flavum
T. flavum
Coptis japonica
T. flavum
C. japonica
Q67A25.1
AAU20765.1
Q5C9L6.1
BAB71802.1
AAU20767.1
Q9LEL5.1
74/59
73/41
69/48
70/48
73/41
71/40
56/38
85/59
82/65
83/66
85/59
82/63
N-Methylcoclaurine 30-hydroxylase (NM C3H)
30-Hydroxy-N-methyl-(S)-coclaurine 40-O-methyltransferase
(HNMC 4OMT)
Corytuberine synthase (CS)
C. japonica
BAF80448.1
74/37
85/56
2842 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
Fig. 4 Norcoclaurine (7) production by recombinant norcoclaurine
synthase. (a) Chromatogram of norcoclaurine standard (7). (b and c) Enzymatic
reaction products from heterologous (recombinant) PhNCS (b) and control
assay with boiled enzyme (c). Data shows that norcoclaurine (7) (elution time:
3.45 min) formation occurs only in presence of PhNCS. Mass spectra of
norcoclaurine (7) standard (d) and enzymatically produced 7 (e).
as magnoflorine (13) when compared with an authentic standard
(Fig. 5g and h). It had mass spectra essentially identical to the
standard, with a base peak of m/z 342.1712 (Fig. 5b), corre-
sponding to [M]+ (calc. 342.1700, C20H24NO4), accompanied by
fragments of m/z 297.1135 (calc. 297.1127, C18H17O4), resulting
from the loss of NH(CH3)2, and of m/z 265.0873 (calc. 265.0865,
C17H13O3), generated by further loss of MeOH, with these
fragments being characteristic of aporphine alkaloids.15 The
magnoflorine (13) from P. peltatum also gave identical data
(Fig. 5f). Interestingly, the main tissue for accumulation of
Fig. 5 Chromatograms and mass spectra of alkaloids from Podophyllum
species and standards. (a) Extracted ion chromatogram at m/z 297.1 ꢂ 0.5 Da
of P. hexandrum rhizome extract corresponding to alkaloid magnoflorine (13),
corytuberine (12) and muricinine (24). (b) to (d) Mass spectral fragmentation
patterns of magnoflorine (13, b), corytuberine (12, c), and putative muricinine
(24, d) in P. hexandrum rhizome extract. (e) Extracted ion chromatogram at
m/z 297.1 ꢂ 0.5 Da of P. peltatum rhizome corresponding to magnoflorine
(13). (f) Mass spectral fragmentation pattern of magnoflorine (13) in P. peltatum
rhizome extract. (g) and (i) Extracted ion chromatogram at m/z 297.1 ꢂ 0.5 Da
of authentic standards magnoflorine (13, g) and corytuberine (12, i). (h) and
magnoflorine (13) in P. hexandrum was in the rhizomes (Fig. 6a), (j) Mass spectral fragmentation patterns of authentic standards magnoflorine
(13, h) and corytuberine (12, j).
while in P. peltatum it was more abundant in the leaves (Fig. 6b),
respectively.
The second most abundant alkaloid detected was identified The different molecular ions, after their characteristic losses of
as corytuberine (12) by comparison with the authentic standard [NH(CH3)2 in magnoflorine (13), NH2CH3 in corytuberine (12)
(Fig. 5i and j). It had a mass spectrum (Fig. 5c) identical to and NH3 in muricinine (24)], produce the characteristic frag-
that of the authentic standard (Fig. 5j), consisting of an ion of ments of m/z 297 and 265.
m/z 328.1552, corresponding to [M + H]+ (calc. 328.1549,
Thus, our hypothesis that Podophyllum species had a complete
C
19H22NO4), accompanied by the same fragments of m/z 297 aporphine alkaloid biosynthetic pathway was fully validated.
and 265. The third alkaloid detected (Fig. 5a) was provisionally
identified as muricinine (24), with a peak of m/z 314.1388,
corresponding to [M + H]+ (calc. 314.1392, C18H20NO4) and the
MALDI-imaging mass spectrometry and qPCR of laser
microdissected tissues
same fragments of m/z 297 and 265 (Fig. 5d). The differences in the It was instructive to next compare and contrast the localization
mass spectra of the three detected alkaloids (Fig. 5b–d) thus reflect of magnoflorine (13) in situ with expression patterns of various
successive methylation of the secondary amine functionality. genes involved in its biosynthetic pathway in rhizome/root
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2843

View Article Online
Molecular BioSystems
Paper
there was no magnoflorine (13) molecular ion detected in the
vascular bundles of either the rhizome or the roots. In the
longitudinal tissue sample of a P. hexandrum root, magnoflorine
(13) was also mainly highly detectable in the emerging roots
(Fig. 7d, yellow arrows) with a lower ion intensity in parts of the
epidermis. For the P. peltatum rhizome cross-section, only a
quarter of the sampled cross-section was subjected to imaging
analysis (Fig. 8b), which included a newly developed root (green
arrowhead). Although less readily detectable in this tissue,
magnoflorine (13) was spatially localized in pith and epidermal
cells of the rhizome, and also where the root was emerging
(Fig. 8b). In the longitudinal section of P. peltatum rhizome
with newly developed shoots (Fig. 8d), magnoflorine (13) dis-
played a higher ion abundance in the region just before the
budding shoot. It was also detected in the vasculature of the
rhizome that it connects to (Fig. 8d, orange arrows).
Lastly, it was considered instructive to establish the location
of expression of the genes in the magnoflorine (13) pathway.
This was investigated, using qPCR from RNA extracted either
from different organs and tissues or cell types excised via laser
microdissection. First, in P. hexandrum, the emerging roots
were separated from the older roots (Fig. 7c, green dotted line),
whereas in P. peltatum, the rhizome was separated from the
budding shoot (Fig. 8c, green dotted line), with each RNA
sample then subjected to qPCR analysis. The expression levels
Fig. 6 Alkaloid content in Podophyllum species. Alkaloid accumulation
(% dry weight) in extracts from different tissues of P. hexandrum (a) and
P. peltatum (b).
tissues and in individual cell types (Fig. 7 and 8). From of NCS (Fig. 7e and 8e) correlated very well with the pattern
an anatomical perspective, underground tissues from both of alkaloid distribution as determined by MALDI-MS imaging
P. hexandrum and P. peltatum, respectively, were stained with (Fig. 7d and 8d). That is, in the case of the P. hexandrum, higher
TBO after MALDI IMS analysis for visualization. This includes: transcript and magnoflorine (13) levels were co-localized in the
P. hexandrum rhizome and its roots [Fig. 7a, with arrow heads emerging roots, while alternatively in P. peltatum, highest gene
indicating older (green) and younger (yellow) parts of the roots], expression and alkaloid (13) content were more evident in the
as well as a P. peltatum rhizome with a newly developed root rhizome, rather than the young, emerging shoots. All other
(green arrow head, Fig. 8a); longitudinal sections of P. hexandrum genes in the pathway gave similar results, namely norcoclaurine
root tissues with emerging lateral roots (indicated with yellow 6-O-methyltransferase (NC 6OMT), coclaurine N-methyltransferase
arrow heads, Fig. 7c), as well as a P. peltatum rhizome with newly (CNMT) and corytuberine synthase (CS) (Fig. S2, ESI†). Thus these
developed shoots (Fig. 8c, right of green broken line).
data strongly correlated magnoflorine (13) accumulation with site
Prior to magnoflorine (13) localization in situ, we first of gene expression. It was next of interest to establish patterns
examined whether 2,5-dihydroxybenzoic acid was a good ioni- of gene expression in individual cell types. In this regard, laser
zation matrix for detection of this alkaloid. With this matrix, microdissection was carried out to isolate xylem, cortex and
the magnoflorine (13) molecular ion of m/z 342.1705 [M]+ was pith-enriched cell preparations from P. hexandrum (Fig. 7f) and
readily detected in the positive ion mode (data not shown). P. peltatum (Fig. 8f), where expression of NCS was found to be
Its spatial distribution was thus next investigated in the various mainly associated with the xylem cells, thus spatially distinct
Podophyllum tissues. High resolution MALDI-TOF IMS showed from site of magnoflorine (13) accumulation. Similar results were
molecular ions at m/z 342.1707 (P. hexandrum) and m/z obtained with NC 6OMT, CS and CNMT genes (Fig. S2, ESI†).
342.1700 (P. peltatum) corresponding to [M]+, with the MS/MS
fragmentation patterns being in good agreement to an authentic
13 standard. The P. hexandrum tissues showed a higher magno-
florine (13) molecular ion intensity, which was B8 times higher
Discussion
compared to P. peltatum tissues, a finding in good agreement The presence of alkaloids in P. hexandrum has been a subject of
with our UPLC-MS analysis (Fig. 6).
debate since the 19th century, with the first report indicating
In the P. hexandrum cross-section, magnoflorine (13) molecular that it contained berberine (25, Fig. 2d). However, other researchers
ion intensity varied, being highest in the rhizome pith and later concluded that no alkaloids were present.16–18 More recently,
epidermal cells (Fig. 7b) as illustrated in the heat map (yellow Wani and co-workers reported very preliminary colorimetric
being the most intense). One younger root cross-section (Fig. 7b, findings in terms of faintly detecting alkaloids in P. hexandrum
yellow arrowhead) also had a higher molecular ion abundance aqueous extracts using Mayer’s reagent.19 No alkaloids were,
in the epidermal and cortical parenchyma cells. By contrast, however, identified. Additionally, these same investigators
2844 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
Fig. 7 Spatial localization of magnoflorine (13) and norcoclaurine synthase gene expression in P. hexandrum. MALDI imaging mass spectrometry images
of magnoflorine (13) (b and d) and expression levels of norcoclaurine synthase (e and f). For illustration, the MALDI plate is shown stained with toluidine
blue O, TBO, after MALDI-imaging (a and c). Green arrow heads indicate roots, whereas yellow arrow heads show emerging roots. Regions with
high magnoflorine (13) content in emerging roots of P. hexandrum (d) have higher NCS expression (e). Orange arrow heads indicate vasculature. The
green broken lines in C shows areas excised for qPCR analyses: magnoflorine (13) accumulates in tissues (d) with higher NCS expression level (e).
Abbreviations: p, pith; cp, cortical parenchyma. Relative intensity of metabolites depicted in heat map, where yellow is the highest intensity, and blue the
lowest. Bars = 5 mm.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2845

View Article Online
Molecular BioSystems
Paper
Fig. 8 Spatial localization of magnoflorine (13) and norcoclaurine synthase gene expression in P. peltatum. MALDI imaging mass spectrometry images
of magnoflorine (13) (b and d) and expression levels of norcoclaurine synthase (e and f). For illustration, the MALDI plate is shown stained with toluidine
blue O, TBO, after MALDI-imaging (a and c). Green arrow heads indicate roots, whereas yellow arrow heads show emerging roots. Orange arrow heads
indicate vasculature. The green broken lines in (c) shows areas excised for qPCR analyses: magnoflorine (13) accumulates in tissues (d) with higher NCS
expression level (e). Abbreviations: p, pith; cp, cortical parenchyma. Relative intensity of metabolites depicted in heat map, where yellow is the highest
intensity, and blue the lowest. Bars = 5 mm.
2846 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
failed to detect any positive colorimetric testing of the corre- but was perhaps minimized in favour of lignan biosynthesis
sponding methanolic extracts using three reagents (Mayer’s, that characterize these species. We can thus hypothesize
Wagner’s and Dragendorff) specific for alkaloids.20 By contrast, though that one or more alkaloid biosynthetic genes have been
we unambiguously established presence of a hitherto unknown recruited into the PTOX (1) and related lignan biosynthetic
aporphine alkaloid pathway being operative in both Podophyllum pathways, as exemplified by the CYP719A enzymes capable of
species, and the identity of the alkaloids so formed. This discovery converting matairesinol (18) into pluviatolide (19).3
relied initially on in depth Illumina transcriptome sequencing
Another interesting report was that when Arabidopsis
and bioinformatics ‘‘omics’’ analyses, but was then definitively was treated with the soil-borne vascular pathogen Verticillium
established by probing UPLC-TOF-MS metabolomic data, with longisporum, this apparently resulted in induced accumulation
magnoflorine (13), corytuberine (12) and putatively muricinine of lignans, piperitol (21), sesamolinol (23) and syringaresinol
(24) being identified.
(26) in the root tissue.26 Until then, methylenedioxy-bridge
As further proof, the deduced entry point pathway enzymatic containing lignans 21 and 23 had never been reported in
step, catalysed by NCS, was also shown to be functionally active, Arabidopsis. However, they are also well-known abundant food
by cloning the encoding D9 truncated gene from P. hexandrum, protective lignans in sesame oil, implying perhaps antioxidant
and demonstrating that recombinant NCS was capable of or defence roles. Accordingly, while the genes in Arabidopsis
forming norcoclaurine (7) from dopamine (5) and 4-hydroxy- leading to these lignans are currently unknown, formation of
phenylacetaldehyde (6). The remaining downstream pathway 21 and 23 raises the possibility that a ‘‘silent’’ metabolic path-
steps were also all cloned from P. hexandrum (with the excep- way is induced on pathogen attack.
tion of PhHNMC4OMT), with each cloned gene having the full
Aporphine alkaloid localization and metabolic pathway
considerations
sequences deduced earlier from the RNA-seq assemblies.
Yet, in contrast, prior work by Facchini and co-workers21 on
the phylogeny of benzylisoquinoline alkaloid biosynthesis in Metabolite localization at the cell, tissue and organ type level
various plant species, was unable to detect NCS activity in in situ, together with corresponding targeted gene expression
P. peltatum. This earlier inability to detect NCS activity may levels, are powerful tools in natural product biosynthetic
be due to the experimental conditions (and limited sensitivity) studies. In this work, we thus mapped the spatial distribution
of the assay conditions employed in that investigation.
of magnoflorine (13) in Podophyllum sp. mature/older rhizome
and root tissues using MALDI imaging mass spectrometry and
complemented these results with tissue and cell type targeted
Potential evolutionary ramifications
The biosynthetic pathways to the Podophyllum aryltetralin gene expression analysis. Our results demonstrated that in
lignans and aporphine alkaloids have many similarities, in mature/older rhizome/root tissues, the localization of magno-
terms of being derived from aromatic amino acids, and in florine (13) and gene transcripts in its metabolic pathway
undergoing somewhat analogous skeletal re-arrangements were distinct in terms of cell types. That is, the pathway genes
(i.e. inter- and/or intra-molecular coupling/cyclisations), hydroxyl- from NCS to CS are mainly detected in xylem cells, whereas
ations and methylations. Indeed, the hydroxylations, methyl- magnoflorine (13) is mainly found in pith, epidermal cells, and
ations and the C–C intramolecular cyclisation to form PTOX (1) newly emerging root sections (Fig. 7 and 8). This finding
and related lignans are essentially mirrored in the pathway apparently strongly suggests transport processes are in place
leading to the aporphine alkaloids (Fig. 2). In contrast to PTOX from one cell type for its formation to the other type(s)
(1), however, alkaloids displaying methylenedioxy bridge func- for storage. In the case of Podophyllum species, biosynthesis
tionalities were not detected in the present investigation.
apparently occurs in younger tissues and vasculature, with
Presence of alkaloids in Podophyllum species might be consid- accumulation in the rhizome pith, where transcript levels are
ered consistent with its evolutionary proximity to several benzyl- considerably lower.
isoquinoline alkaloid producing species in the Berberidaceae
It should be noted, though, that many factors also influence
family.21 On the other hand, current molecular phylogeny of this metabolite imaging, such as: different compounds having
family22,23 clearly places Podophyllum species in a monophyletic different ionization behaviours and so MALDI-imaging condi-
group (often designated Podophyllum group) which also includes tions must be optimized on a case-by-case basis. Besides that,
Dysosma and Diphylleia. All of these plant species are known to results must be interpreted keeping in mind that biosynthetic
accumulate PTOX (1) and/or related lignans,24,25 but not alkaloids. intermediates might not accumulate in sufficient amounts for
The results herein, however, now clearly demonstrate how our detection; different steps in a biosynthetic pathway can occur in
rapidly evolving multi-‘omics’ approaches can radically change different tissues/cell types; and metabolites can be transported
the way we consider natural product chemistry and chemo- to different tissues/cell types for storage, making the actual site
taxonomy, i.e. with increasingly more sensitive analytical technol- of biosynthesis more complex to decipher.
ogies, deeper and more accurate coverage of transcriptomes,
genomes, proteomes, and metabolomes, as well as more powerful understanding metabolism at a new level, with the potential to
bioinformatics. greatly shorten the time required for complete elucidation of
Nevertheless, these technological advancements allow for
The detection of alkaloid metabolism is also an indication natural product biosynthetic pathways and their temporal and
that evolution did not completely eliminate alkaloid metabolism, spatial organization.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2847

View Article Online
Molecular BioSystems
Paper
Conclusions
the NSF MRI support. We thank Dr Toni Kutchan (Donald
Danforth Plant Science Center) for magnoflorine, corytuberine
and norcoclaurine standards.
The application of multi-omics analyses (transcriptomics, bio-
informatics and metabolomics) enabled discovery of the aporphine
alkaloid pathway in Podophyllum species, and the evolutionary/
phylogenetic ramifications thereof. This discovery has potentially
additional benefits, both medicinally and economically. This is
because P. hexandrum rhizome extracts are the commercial source
of PTOX (1) for the semi-synthesis of anticancer drugs, and the
alkaloids corytuberine (12), magnoflorine (13) and muricinine (24)
can be easily recovered from methanolic and aqueous extracts.
They thus may be of additional economic value, since aporphine
alkaloids display a wide array of biological activities,27–29 ranging
from antioxidant28,29 to anti-inflammatory30 and psychoactive31
properties. Indeed, their presence should also be considered in
the context of the use of crude extracts of such plant species in
folk medicine. The discovery of this alkaloid pathway thus also
raises the possibility of increasing their levels using biotechnol-
ogical approaches; this could be achieved via effective transforma-
tion and/or transient silencing/upregulation (e.g. virus-induced
gene silencing).
Notes and references
´
1 M. Gordaliza, P. A. Garcıa, J. M. del Corral, M. A. Castro and
´
M. A. Gomez-Zurita, Toxicon, 2004, 44, 441–459.
2 C. Canel, R. M. Moraes, F. E. Dayan and D. Ferreira,
Phytochemistry, 2000, 54, 115–120.
3 J. V. Marques, K.-W. Kim, C. Lee, M. A. Costa, G. D. May,
J. A. Crow, L. B. Davin and N. G. Lewis, J. Biol. Chem., 2013,
288, 466–479.
4 N. Ikezawa, K. Iwasa and F. Sato, Plant Cell Rep., 2009, 28,
123–133.
5 E. Ono, M. Nakai, Y. Fukui, N. Tomimori, M. Fukuchi-
Mizutani, M. Saito, H. Satake, T. Tanaka, M. Katsuta,
T. Umezawa and Y. Tanaka, Proc. Natl. Acad. Sci. U. S. A.,
2006, 103, 10116–10121.
6 Y. Jiao, L. B. Davin and N. G. Lewis, Phytochemistry, 1998, 49,
387–394.
7 A. Pasquo, A. Bonamore, S. Franceschini, A. Macone,
A. Boffi and A. Ilari, Acta Crystallogr., Sect. F: Struct. Biol.
Cryst. Commun., 2008, 64, 281–283.
8 E. K. M. Ueda, P. W. Gout and L. Morganti, J. Chromatogr. A,
2003, 988, 1–23.
9 A. M. Patten, M. Jourdes, C. L. Cardenas, D. D. Laskar,
Y. Nakazawa, B.-Y. Chung, V. R. Franceschi, L. B. Davin and
N. G. Lewis, Mol. BioSyst., 2010, 6, 499–515.
10 A. T. Dinkova-Kostova, D. R. Gang, L. B. Davin, D. L.
Bedgar, A. Chu and N. G. Lewis, J. Biol. Chem., 1996, 271,
29473–29482.
11 D. R. Gang, M. A. Costa, M. Fujita, A. T. Dinkova-Kostova,
H.-B. Wang, V. Burlat, W. Martin, S. Sarkanen, L. B. Davin
and N. G. Lewis, Chem. Biol., 1999, 6, 143–151.
Abbreviations
CNMT
CS
CyP450
DP
Coclaurine N-methyltransferase
Corytuberine synthase
Cytochrome P450
Dirigent protein
HNMC 4OMT 30-Hydroxy-N-methyl-(S)-coclaurine 40-O-methyl-
transferase
NC 6OMT
NCS
NM C3H
PLR
Norcoclaurine 6-O-methyltransferase
Norcoclaurine synthase
N-Methylcoclaurine 30-hydroxylase
Pinoresinol–lariciresinol reductase
Pluviatolide synthase
PLS
PSS
PTOX
SDH
Piperitol/sesamin synthase
Podophyllotoxin
Secoisolariciresinol dehydrogenase
Toluidine blue O
´
12 Z.-Q. Xia, M. A. Costa, H. C. Pelissier, L. B. Davin and
N. G. Lewis, J. Biol. Chem., 2001, 276, 12614–12623.
13 L. Y. P. Luk, S. Bunn, D. K. Liscombe, P. J. Facchini and
M. E. Tanner, Biochemistry, 2007, 46, 10153–10161.
TBO
´
14 C. Stevigny, J.-L. H. Jiwan, R. Rozenberg, E. de Hoffmann
Acknowledgements
and J. Quetin-Leclercq, Rapid Commun. Mass Spectrom.,
2004, 18, 523–528.
We thank the National Science Foundation (MCB-1052557), the
Chemical Sciences, Geosciences and Biosciences Division, 15 W.-N. Wu and C.-H. Huang, Chin. Pharm. J., 2006, 58,
Office of Basic Energy Sciences, Office of Science, U.S. Depart- 41–55.
ment of Energy (DE-FG-0397ER20259) and the G. Thomas 16 W. R. Dunstan and T. A. Henry, J. Chem. Soc., Trans., 1898,
and Anita Hargrove Center for Plant Genomic Research, for 73, 209–226.
generous financial support. MALDI-MS based imaging analysis 17 F. B. Power, Am. J. Pharm., 1878, 50, 369–370.
was performed on an instrument acquired through a Major 18 F. B. Power, Chem. News, 1898, 78, 26–27.
Research Instrumentation grant (DBI-1229749) from the 19 S. A. Wani, K. W. Shah and M. A. Ahmad, Int. J. Biol. Pharm.
National Science Foundation. Bioinformatics analysis and gene
cloning/recombinant protein expression work was supported by 20 S. A. Wani, M. Ashfaq, K. W. Shah and D. Singh, J. Curr.
the NSF MCB-1052557 grant and the G. Thomas and Anita Chem. Pharm. Sci., 2012, 2, 125–128.
Hargrove Center for Plant Genomic Research; metabolite iso- 21 D. K. Liscombe, B. P. MacLeod, N. Loukanina, O. I.
Res., 2012, 3, 684–686.
lation, identification and in situ localization (using MALDI-IMS)
Nandi and P. J. Facchini, Phytochemistry, 2005, 66,
was supported by the DOE, with instrumentation being from
1374–1393.
2848 | Mol. BioSyst., 2014, 10, 2838--2849
This journal is ©The Royal Society of Chemistry 2014

View Article Online
Paper
Molecular BioSystems
22 Y.-D. Kim, S.-H. Kim, C. H. Kim and R. K. Jansen, Biochem. 27 S. M. Mohamed, E. M. Hassan and N. A. Ibrahim, Nat. Prod.
Syst. Ecol., 2004, 32, 291–301. Res., 2010, 24, 1395–1402.
23 W. Wang, Z.-D. Chen, Y. Liu, R.-Q. Li and J.-H. Li, Syst. Bot., 28 T. M. Hung, J. P. Lee, B. S. Min, J. S. Choi, M. K. Na,
2007, 32, 731–742.
24 A. J. Broomhead and P. M. Dewick, Phytochemistry, 1990, 29,
3831–3837.
25 R.-W. Jiang, J.-R. Zhou, P.-M. Hon, S.-L. Li, Y. Zhou, L.-L. Li,
W.-C. Ye, H. X. Xu, P.-C. Shaw and P. P.-H. But, J. Nat. Prod.,
2007, 70, 283–286.
X. F. Zhang, T. M. Ngoc, I. S. Lee and K. H. Bae, Biol. Pharm.
Bull., 2007, 30, 1157–1160.
29 T. M. Hung, M. K. Na, B. S. Min, X. F. Zhang, I. S. Lee,
T. M. Ngoc, P. T. Thuong, D. E. Sok and K. H. Bae, Planta
Med., 2007, 73, 1281–1284.
30 E. Ku¨peli, M. Kosar, E. Yesilada and K. H. C. Baser, Life Sci.,
2002, 72, 645–657.
¨
26 S. Konig, K. Feussner, A. Kaever, M. Landesfeind, C. Thurow,
˜
P. Karlovsky, C. Gatz, A. Polle and I. Feussner, New Phytol., 2014, 31 J. B. I. de la Pena, H. L. Lee, S. Y. Yoon, G. H. Kim, Y. S. Lee
202, 823–837.
and J. H. Cheong, J. Nat. Med., 2013, 67, 814–821.
This journal is © The Royal Society of Chemistry 2014
Mol. BioSyst., 2014, 10, 2838--2849 | 2849