Transcriptional profiling and inhibition of cholesterol biosynthesis in human T lymphocyte cells by the marine toxin azaspiracid.

Abstract

Azaspiracid-1 (AZA-1) is a marine biotoxin reported to accumulate in shellfish from several countries, including eastern Canada, Morocco, and much of western Europe, and is frequently associated with severe gastrointestinal human intoxication. As the mechanism of action of AZA-1 is currently unknown, human DNA microarrays and qPCR were used to profile gene expression patterns in human T lymphocyte cells following AZA-1 exposure. Some of the early (1 h) responding genes consisted of transcription factors, membrane proteins, receptors, and inflammatory genes. Four- and 24-h responding genes were dominated by genes involved in de novo lipid biosynthesis of which 17 of 18 involved in cholesterol biosynthesis were significantly up regulated. The up regulation of synthesis genes was likely in response to the ca. 50% reduction in cellular cholesterol, which correlated with up regulated protein expression levels of the low-density lipoprotein receptor. These data collectively detail the inhibition of de novo cholesterol synthesis, which is the likely cause of cytotoxicity and potentially a target pathway of the toxin.

Full text

Page 1

This article was published in an Elsevier journal. The attached copy
is furnished to the author for non-commercial research and
education use, including for instruction at the author’s institution,
sharing with colleagues and providing to institution administration.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright

Page 2

Author's personal copy
Transcriptional profiling and inhibition of cholesterol biosynthesis in human
T lymphocyte cells by the marine toxin azaspiracid
Michael J. Twinera,⁎, James C. Ryana, Jeanine S. Moreya, Kent J. Smithb, Samar M. Hammadc,
Frances M. Van Dolaha, Philipp Hessd, Terry McMahond, Masayuki Satakee,
Takeshi Yasumotof, Gregory J. Doucettea
aMarine Biotoxins Program, Center for Coastal Environmental Health and Biomolecular Research,
NOAA/National Ocean Service, 219 Fort Johnson Road, Charleston, SC 29412, USA
bDepartment of Biochemistry and Molecular Biology, Medical University of South Carolina, Charleston, SC 29401, USA
cDepartment of Cell Biology and Anatomy, Medical University of South Carolina, Charleston, SC 29401, USA
dBiotoxin Chemistry, Marine Institute, Rinville, Oranmore, Ireland
eGraduate School of Agricultural Science, Tohoku University, Sendai, Japan
fJapan Food Research Laboratory, Tama Laboratory, Nagayama, Tama, Tokyo,
and Okinawa Health Biotechnology Research Development Center, Gushikawa, Okainawa, Japan
Received 6 March 2007; accepted 20 October 2007
Available online 11 January 2008
Abstract
Azaspiracid-1 (AZA-1) is a marine biotoxin reported to accumulate in shellfish from several countries, including eastern Canada, Morocco, and
much of western Europe, and is frequently associated with severe gastrointestinal human intoxication. As the mechanism of action of AZA-1 is
currentlyunknown,humanDNAmicroarraysandqPCRwereusedtoprofilegeneexpressionpatternsinhumanTlymphocytecellsfollowingAZA-1
exposure. Some of the early (1 h) responding genes consisted of transcription factors, membrane proteins, receptors, and inflammatory genes. Four-
and 24-h responding genes were dominated by genes involved in de novo lipid biosynthesis of which 17 of 18 involved in cholesterol biosynthesis
were significantly up regulated. The up regulation of synthesis genes was likely in response to the ca. 50% reduction in cellular cholesterol, which
correlated with up regulated protein expression levels of the low-density lipoprotein receptor. These data collectively detail the inhibition of de novo
cholesterol synthesis, which is the likely cause of cytotoxicity and potentially a target pathway of the toxin.
© 2007 Elsevier Inc. All rights reserved.
Keywords: Azaspiracid-1; Cholesterol; Fatty acid; Gene expression; Harmful algal bloom; Immunoblot; Low-density lipoprotein receptor; T lymphocyte cells;
Microarray; Shellfish toxin
Azaspiracids (AZA) are polyether marine toxins (Fig. 1) that
were first detected in mussels (Mytilus edulis) from Ireland
following a 1995 outbreak of gastrointestinal illness and have
since been detected in other bivalve species, including oysters
(Crassostrea gigas, Ostrea edulis), scallops (Pecten maximus),
clams (Tapes philippinarum), cockles (Cardium edule), and
razor clams (Ensis siliqua) [1–3]. Cases of AZA intoxication
and/or contaminated shellfish have been documented in several
other European countries, including the United Kingdom,
Norway, The Netherlands, France, Spain, and Italy [4–7] and
more recently in eastern Canada (M. Quilliam, personal
communication) as well as Morocco [8]. In Europe, a regulatory
limit of 160 μg AZA-1 equivalents/kg whole shellfish flesh has
been set to protect human health (Regulation EC 853/2004).
Recently,fromJulytoDecember2005,levelsofAZAexceeding
this regulatory limit were detected in mussels from production
areas along the west coast of Ireland [9] and in some production
areas AZA levels remained above the regulatory limit for more
than 9 months.
Following human consumption of AZA-contaminated shell-
fish, there is a rapid onset of symptoms that include nausea,
vomiting, severe diarrhea, and stomach cramps [10,11]. Al-
though the human symptoms resemble those of diarrhetic
Available online at www.sciencedirect.com
Genomics 91 (2008) 289–300
www.elsevier.com/locate/ygeno
⁎Corresponding author. Fax: +1 843 762 8700.
E-mail address: Mike.Twiner@noaa.gov (M.J. Twiner).
0888-7543/$ - see front matter © 2007 Elsevier Inc. All rights reserved.
doi:10.1016/j.ygeno.2007.10.015

Page 3

Author's personal copy
shellfish poisoning (DSP), the illness has subsequently been
named azaspiracid poisoning (AZP) [12,13], reflecting the
causative toxin family. Murine intraperitoneal injections induce
symptoms such as fatigue, paralysis, labored breathing, and
death as soon as 35 min postinjection [10,11]. Pathological
effects include histological alterations in the liver, pancreas,
spleen, and necrotic lymphocytes in the thymus. Experiments in
which mice were chronically dosed with AZA-1 (oral admin-
istration of 5–50 μg/kg) demonstrated that AZA-1 exposure
induces gastrointestinal inflammation, including accumulation
of fluid, necrosis, and edema in the lamina propria of the mid-
intestinaltract,aswellas fusion and shortening of villi. Necrosis
of T and B lymphocytes was also documented in the spleen
and thymus, as well as fatty changes in the liver, hyperplasia of
the epithelial lining in the stomach, and tumors in the lungs
[5,13,14]. Due to the incidence of lung tumors, AZA-1 is cur-
rently being investigated as a potential carcinogen, particularly
since it has been shown recently to act as a teratogen in fish
embryos [15].
Studies to determine the mechanism of action of AZA have
been conducted by several investigators using in vitro tech-
niques. Although AZP symptoms in humans are similar to
those of DSP, AZA-1 does not alter the activities of protein
phosphatase-1 (PP1) or PP2A [16], both known targets for DSP
toxins, indicating a different mechanism of action for AZA.
Cytotoxicity testing of seven mammalian cell lines has shown
that AZA-1 is cytotoxic to all tissue types examined (lymphoid,
kidney, lung, neuronal, and pituitary cells), with EC50values
in the low nanomolar range (0.9–16.8 nM) [16]. Interestingly,
immune-type cells appeared to be most sensitive to the cy-
totoxic effects of AZA-1, correlating with in vivo pathological
observations [5,14,17]. In particular, the morphology and
cytoskeleton of a human T lymphocyte cell line (Jurkat) were
distinctly affected by AZA-1. Similar changes have also been
documented by Roman et al. [18], in that AZA-1 reduced cel-
lular F-actin content in a nonapoptotic manner following the
elevation of cytosolic calcium and cAMP levels. Recent work
by Vilarino et al. [19] has also revealed effects of AZA-1
toward cytoskeletal components, while Ronzitti et al. [20]
reported that AZA-1 causes fragmentation of E-cadherin, a
membrane protein involved in cell–cell adhesion. To investi-
gate further the potential mechanism(s) of action of AZA-1, we
employed a whole-genome expression microarray to assess the
differential expression of N37,000 genes at three time points
following the exposure of T lymphocyte cells to AZA-1 over
24 h. A pathway of interest was identified from the gene ex-
pression data and confirmed by qPCR, immunoblots, and
substrate analysis.
Results
To assess the response of T lymphocyte cells to AZA-1,
gene expression in Jurkat cells following exposure to AZA-1
(10 nM) was compared to basal gene expression in control cells
exposed to equivalent amounts of the methanolic vehicle at 1,
4, and 24 h. As illustrated in Fig. 2, this concentration had
significant effects on both cell morphology and cytotoxicity.
After 24 h of AZA-1 exposure, the altered morphology of
T lymphocyte cells, including cell rounding and retraction of
pseudopodia, was characteristic of a cytotoxic response. This
was confirmed by detecting the release of the cytosolic enzyme
glucose-6-phosphate dehydrogenase, which isalso corroborated
by our previous study [16].
Signature genes and clusters
Features from the composite gene expression microarrays
were classified as differentially expressed, or “signature,” if
they possessed a p value ≤0.01 using the Rosetta error model
and weighted averaging. An overview of the signature feature
gene data can be found in Supplementary Table 1. Signature
features with an absolute differential expression of N1.5
(=0.58 log2ratio value) and p value V10−4were analyzed
further and resulted in three distinct clusters containing a total of
437 differentially expressed genes (Fig. 3). Cluster 1 (highly up
regulated features) contained 92 genes, cluster 2 (moderately up
regulated) contained 202 genes, and cluster 3 (moderately down
regulated) contained 143 genes. Details of these features are
shown in Supplementary Tables 2, 3, and 4, respectively. At
least 38 genes within these clusters are involved directly in lipid
regulation (i.e., transcription factors, enzymes for cholesterol
or fatty acid synthesis, elongation, modification, etc.). The
majority of these genes were from cluster 1 (33 of 38), and 5
genes were from cluster 2. Table 1 illustrates the details of these
lipid-related genes in which it is clear that most were sig-
nificantly up regulated at the 4- and 24-h time points, but were
not differentially expressed at 1 h. In some instances, multiple
yet different probes on the array query the same gene; as a
result, some genes appear more than once in Table 1. These data
are only a subset of the gene expression array data; however, all
raw gene expression data are available at Gene Expression
Omnibus (GEO; http://www.ncbi.nlm.nih.gov/geo/; GEO Ac-
cession No. GSE5346).
Although clusters 1 and 2 were dominated by up regulated
lipid-related genes, cluster 1 also included genes related to
glucose metabolism (i.e., insulin-induced gene 1), cell signaling
(i.e., C/EBP-induced protein), and cell growth (i.e., cyclin G2)
Fig. 1. Structure of azaspiracid-1 (AZA-1).
290M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 4

Author's personal copy
(Supplementary Table 2). Cluster 2 contained many genes
related to glucose metabolism (i.e., pyruvate dehydrogenase
kinase), cell signaling (i.e., interferon-stimulated gene, inter-
cellular adhesion molecule 2, G-protein-coupled receptors,
transcription factors), ion transport (i.e., ATPases, potassium
channel), and cell death/apoptosis (i.e., BCL2-interacting
protein 3, p53-inducible nuclear protein 1) (Supplementary
Table 3). The typically down regulated genes within cluster 3
were represented by genes involved in cell signaling (i.e.,
glucagon receptor, WNT frizzled homolog 9, neuronatin), cell
growth (i.e., cyclin B2, cyclin-dependent kinase inhibitor 3),
stress response (i.e., heat shock proteins), and translation (i.e.,
mitochondrial ribosomal proteins) (Supplementary Table 4). It
shouldalsobenotedthatpreliminarygeneexpressionanalysisof
cellsexposedto1nMAZA-1alsoelicitedsimilartranscriptional
patterns (data not shown).
Fig. 3. Clustered trend set. T lymphocytes were exposed to AZA-1 (10 nM) for 1, 4, and 24 h and differentially expressed genes identified. Genes were clustered using
a K-means algorithm with Euclidean distance measurements (fold change ≥1.5 or ≤−1.5; pV10−4) and resulted in 437 differentially expressed genes. Three clusters
were identified: cluster 1 (green) contained 92 genes, cluster 2 (blue) contained 202 genes, and cluster 3 (red) contained 143 genes. Thirty-eight genes (33 from cluster
1; 5 from cluster 2) were involved in lipid regulation (see Table 1). Data are plotted as ratio log2values.
Fig. 2. Effects of azaspiracid-1 on cell morphology andcytotoxicityof human Tlymphocyte cells.Cells were continuouslyexposedto either (A) methanol(0.1% final)
or (B) AZA-1 (10 nM) before the photomicrographs were taken. (C) Cytotoxicity was determined with increasing concentrations of AZA-1 (10−11to 10−7M) for 24 h
before cellular lysis was assessed via release of glucose-6-phosphate dehydrogenase (G6PD). G6PD release was quantified and normalized as a function of total G6PD
(100% cytotoxicity). Cytotoxicity data are means±SE in triplicate wells for three independent experiments. Controls were treated with equivalent amounts of the
methanol vehicle. An EC50value of 1.1 nM was determined by a variable slope sigmoidal regression analysis using GraphPad Prism software (version 4.00).
291 M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 5

Author's personal copy
qPCR
Eleven genes involved with cholesterol biosynthesis were
selected for verification by qPCR, and an α-tubulin-like gene
(GenBank Accession No. NM_145042) was chosen for
normalization purposes. The α-tubulin-like gene exhibited no
significant changes in expression in either the microarray or the
qPCR analysis (data not shown). The changes (i.e., magnitude
and direction) in many of the cholesterol biosynthesis genes
detected by microarrays were confirmed by qPCR (Fig. 4).
Pearson's correlation coefficient between the microarray and
the qPCR gene expression ratios were for all data 0.9173,
n=24, p≤0.0001; for 4 h data 0.7538, n=12, p=0.0046; and
for 24 h data 0.9436, n=12, p≤0.0001. In addition, the
direction of regulation was conserved for 23 of 24 (96%) genes.
In the one instance in which the direction was not conserved
(i.e., PMVK), the fold change was less than 1.2 by both
methods. For additional qPCR and microarray comparisons
from this study, refer to Morey et al. [21].
Mapping gene expression patterns for the cholesterol
biosynthesis pathway using GenMapp
Analysis of the gene expression data by MappFinder yielded
a list of pathways and/or cellular processes sorted by Z score
that were identified as having a higher proportion of dif-
ferentially expressed genes than expected. Fatty acid, lipopro-
tein, and cholesterol pathways were calculated to have some of
the highest Z scores (≥4.0) for the 4- and 24-h time points (data
not shown). Gene expression values at 4 and 24 h have been
plotted onto the cholesterol biosynthesis pathways for visuali-
zation (Fig. 5). At 4 h following the addition of AZA-1, 14 of
Table 1
Differentially expressed genes involved in lipid regulation
Accession or
Agilent probe No.
Sequence description1 h4 h24 h Cluster
Ratiop valueRatiop value
2.90×10−6
3.43×10−21
2.75×10−40
b1.0×10−45
2.21×10−19
b1.0×10−45
1.47×10−6
0.84
2.59×10−15
4.00×10−5
0.14
0.03
1.87×10−12
7.71×10−12
5.57×10−28
5.19×10−8
1.50×10−4
1.54×10−7
3.00×10−24
0.38
3.00×10−5
0.03
2.00×10−5
2.24×10−10
0.14
0.55
2.62×10−42
0.09
4.26×10−27
8.30×10−19
4.24×10−31
0.07
2.30×10−4
0.02
0.90
6.71×10−14
1.12×10−12
1.80×10−7
Ratiop value
7.69×10−23
b1.0×10−45
2.14×10−34
8.46×10−9
6.41×10−25
b1.0×10−45
b1.0×10−45
1.00×10−4
b1.0×10−45
b1.0×10−45
b1.0×10−45
b1.0×10−45
8.62×10−12
b1.0×10−45
2.77×10−41
b1.0×10−45
2.10×10−32
2.48×10−8
3.07×10−12
4.45×10−9
9.21×10−12
5.02×10−17
1.29×10−16
b1.0×10−45
5.02×10−10
4.42×10−9
1.01×10−43
9.00×10−5
1.21×10−6
b1.0×10−45
2.43×10−9
2.64×10−27
b1.0×10−45
6.06×10−15
1.32×10−6
2.70×10−6
2.00×10−5
2.58×10−38
NM_014762
NM_000859
BC000297
AK095492
NM_001360
NM_005891
AF356877
NM_024090
NM_006579
NM_002004
A_32_P95223
XM_291508
NM_004462
NM_013402
NM_013402
NM_004265
NM_016371
NM_016371
NM_004508
AK096769
NM_002340
AK096769
NM_002340
NM_000527
NM_181726
NM_002461
X75311
NM_022776
NM_003129
NM_139164
NM_005063
NM_005063
AF132203
NM_004176
NM_004599
NM_006745
NM_006918
BC012333
24–Dehydrocholesterol reductase (DHCR24)
HMG–coenzyme A reductase (HMGCR)
HMG–coenzyme A synthase 1 (HMGCS1)
HMGCS1
7–Dehydrocholesterol reductase (DHCR7)
Acetyl–coenzyme A acetyltransferase 2 (ACAT2)
ACAT2
ELOVL family, fatty acid elongation (ELOVL6)
Emopamil–binding protein (sterol isomerase) (EBP)
Farnesyl diphosphate synthase (FDPS)
FDPS
FDPS
Farnesyl diphosphate farnesyltransferase 1 (FDFT1)
Fatty acid desaturase 1 (FADS1)
FADS1
Fatty acid desaturase 2 (FADS2)
Hydroxysteroid (17–β) dehydrogenase 7 (HSD17B7)
HSD17B7
Isopentenyl–diphosphate ä isomerase (IDI1)
Lanosterol synthase (LSS)
LSS
LSS
LSS
Low–density lipoprotein receptor (LDLR)
LDLR–related–binding protein (LRP2BP)
Mevalonate (diphospho) decarboxylase (MVD)
Mevalonate kinase (MVK)
Oxysterol–binding protein–like 11 (OSBPL11)
Squalene epoxidase (SQLE)
START domain–containing 4 (STARD4)
Stearoyl–CoA desaturase (SCD)
SCD
SCD
Sterol reg. element–binding transcription factor 1 (SREBF1)
Sterol reg. element–binding transcription factor 2 (SREBF2)
Sterol–C4–methyl oxidase–like (SC4MOL)
Sterol–C5–desaturase (SC5DL)
SC5DL
1.02
1.06
1.03
1.02
1.00
1.08
0.98
1.20
1.03
0.98
1.03
0.99
1.00
0.88
1.08
1.02
1.02
1.16
1.01
0.94
0.95
1.01
1.10
1.10
0.99
0.83
0.96
0.99
1.18
1.13
0.97
1.04
1.11
1.01
0.92
1.23
1.07
0.96
0.83
0.40
0.76
0.83
0.98
0.41
0.67
0.50
0.70
0.70
0.79
0.94
0.95
0.22
0.48
0.81
0.92
0.31
0.94
0.89
0.53
0.98
0.34
0.07
0.96
0.16
0.59
0.90
0.01
0.33
0.76
0.80
0.29
0.93
0.49
0.10
0.60
0.68
1.81
2.74
2.78
4.22
2.26
2.03
1.79
1.08
1.49
1.46
1.22
1.27
1.61
1.48
1.74
1.37
1.75
2.03
2.24
1.61
2.19
1.86
2.22
2.22
1.15
1.14
1.92
1.18
2.37
1.99
1.95
1.47
1.75
1.23
0.99
2.95
2.14
1.99
4.13
5.66
5.04
5.21
5.42
4.83
5.48
3.02
3.63
4.57
4.33
3.96
2.45
3.16
4.02
3.24
2.47
2.52
4.25
4.85
4.90
4.33
7.08
3.47
1.87
3.02
3.61
1.75
2.92
5.15
2.99
9.26
3.93
1.88
1.70
3.00
2.80
2.88
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
2
1
1
1
1
1
1
2
1
1
2
1
1
1
1
1
2
2
1
1
1
T lymphocyte cells were exposed to AZA–1 (10 nM) for 1, 4, and 24 h and differentially expressed genes from Fig. 3 that are involved with lipid regulation are shown
here. Multiple differentially expressed values for the same gene may be given due to multiple probes made against different locations on a single sequence for a given
transcript.
292M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 6

Author's personal copy
the 16 genes illustrated in the cholesterol biosynthesis pathway
were significantly up regulated (between 1.5- and 4.2-fold) for
at least one of the gene isoforms (Fig. 5). At 24 h, 15 of the 16
genes (ranging between 2.3- and 7.1-fold) were significantly up
regulated. Genes that encode ACAT2 and LDLR were also
significantly up regulated at both time points. Although not
illustrated, many of the genes involved in fatty acid synthesis
were also significantly up regulated at 4 and 24 h, in addition to
INSIG1, INSIG2, SREBP1, and SREBP2 (see Supplementary
Tables 2 and 3).
Immunoblot analysis of LDLR
T lymphocyte cells exposed to AZA-1 were assessed for
differential expression of the low-density lipoprotein receptor
(LDLR) by immunoblot analysis. Although control cells dis-
playedlittlechangeintheexpressionofLDLRoverthe24-htime
course, cells exposed to AZA-1 (1 or 10 nM) had significantly
higher levels of LDLR in a time- and concentration-dependent
manner (Fig. 6). LDLR in the 1 nM AZA-1 treatments was
significantly up regulated at 12 h (∼2.5-fold), whereas in the
Fig. 4. Quantitative PCR validation of arrays. Eleven selected genes from the lipid biosynthesis pathways were validated by qPCR at (A) 4 and (B) 24 h. Array and PCR
expression data are expressed as log2ratio. The Pearson's correlation coefficient between the microarray and the qPCR gene expression ratios were for all data 0.9173,
n=24, p≤0.0001; for 4 h data 0.7538, n=12, p=0.0046; and for 24 h data 0.9436, n=12, p≤0.0001. Full gene names and accession numbers are shown in Fig. 5.
293M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 7

Author's personal copy
10 nM treatments it was up regulated at 4 h (∼1.5-fold). Data for
the 10 nM treatments at 24 h were not available because of
cytotoxicity and an insufficient amount of protein available for
loading onto the gel. Serum-starved cells (−FBS) were used as
positive controls for up regulated LDLR (data not shown).
Quantitative analysis of cellular cholesterol
T lymphocyte cells were exposed to AZA-1 (1 or 10 nM) for
4, 12, or 24 h for quantification of free and esterified cellular
cholesterol. Control cells normalized to ng/μg protein displayed
some reductions in total cellular cholesterol levels over the
period of the experiment, likely due to limiting extracellular
serumascellnumbersincreased(Fig.7).Totalcholesterollevels
(mean±SE, n=3) in the controls were 9.1±0.4 ng cholesterol/
μg protein or 891±16 fg/cell. Treatments with AZA-1 caused
significant (pb0.001) reductions in cellular cholesterol levels to
4.8±0.2 ng cholesterol/μg protein (or 737±85 fg/cell), for the
1 nM treatment, and 4.8±0.4 ng cholesterol/μg protein (or 711±
16 fg/cell) for the 10 nM treatment. Differences in free or
esterified cholesterol were not detected between controls and
treatments, in which esterified cholesterol consistently
accounted for 10 to 15% of the total cholesterol content (data
not shown).
Discussion
Azaspiracidaccumulationinshellfishanditspotentialhuman
health risk to seafood consumers is an ongoing issue in Europe
and appears to be an emerging problem in African and North
American waters. However, our lack of understanding of the
AZA-1 mechanism of action has hindered our ability to assess
both its acute and its chronic healtheffects. This study is the first
toassesstheinvitroeffectsofAZA-1ongeneexpression,which
in turn, has demonstrated conclusively the inhibitory effects of
AZA-1 on de novo cholesterol biosynthesis.
Biological pathway analysis following T lymphocyte
cell exposure to AZA-1
Analysis of gene expression patterns using whole human
genome microarrays revealed several biological pathways that
appear to be targeted when T lymphocyte cells are exposed to
AZA-1. These pathways include, but are likely not limited to,
cholesterol and fatty acid synthesis, the insulin/glucagon path-
way, the WNT signaling pathway, and other pathways that are
involved in inflammation, ion channel activities, the cytoskele-
ton, and cell growth/division.
Cholesterol and fatty acid synthesis pathway
Many of the genes identified in clusters 1 and 2 are neces-
sary for cholesterol biosynthesis and were up regulated at the
4- and 24-h time points. AZA-1 elevated transcriptional levels
of 15 of 16 genes involved in the synthesis of cholesterol from
acetyl-CoA, which is similar to inhibition of the cholesterol
biosynthesis pathway caused by statins. Skeletal muscle cells
exposed to various statins over 24 h were shown to up regulate
many cholesterol synthesis genes (i.e., LSS, HMGCR, SQLE,
HMGCS1) [22]. The remarkable clustering of up regulated
genes of similar function and regulation by AZA-1 prompted
further experimental examination of LDLR.
Found within the membrane of many cell types, LDLR is
typically involved in regulating the level of cholesterol in the
blood. This receptor binds with and internalizes low-density
lipoproteins delivering cholesterol for cellular use and storage.
Following internalization, LDLRs are recycled back to the outer
membrane. LDLRs play a critical role in regulating cholesterol
in which their translocation and transcriptional expression
are inversely proportional to intracellular cholesterol levels.
HMGCR inhibitors such as statins [23,24] are well known to
lower cellular cholesterol levels with a concomitant elevation in
LDLR activity [25]. LDLR protein expression is clearly shown
to be up regulated in a time- and concentration-dependent
fashion by AZA-1 relative to the control cells.
ProteinLDLRupregulationisconsistentwithboththeLDLR
transcriptional response and the decreased levels of cellular
cholesterol induced by AZA-1. Inhibition of total cellular cho-
lesterol synthesis was clearly illustrated by an almost 50%
decrease from time-matched control cells for each AZA-1
concentration. The total cholesterol level in controls cells was
∼8.5 ng cholesterol/μg protein, whereas in both 24-h treatments
levels had dropped to ∼4.6 ng cholesterol/μg protein. Although
future experimentation will be needed to assess further the
effects of additional AZA-1 concentrations on cholesterol levels
inJurkatcells,theeffectsofAZA-1onthecholesterolbiosynthetic
pathway appear to be consistent with those of statins and inhi-
bition of de novo cholesterol synthesis. Although speculative,
a possible site of inhibition may involve phosphomevalonate
kinase (PMVK; GenBank Accession No. NM_006556), the lone
geneinthispathwaythatwasnotupregulatedbyAZA-1.PMVK
catalyzes the phosphorylation of mevelonate-5-P into mevalo-
nate-5-PPandhasbeenshowntobetranscriptionallycontrolledin
response to cellular sterol levels [26]; however, there may be a
posttranscriptional control not detectable by the current gene
expression study.
Genes required for enzymatic synthesis of fatty acids were
also up regulated in a time-dependent manner at 4 and 24 h. In
addition to the fatty acid synthesis genes, a variety of other
sterol regulatory genes such as SREBP and INSIG1 (insulin-
inducedgene1)wereupregulated,potentiallyelicitingcontrolof
the fatty acid and/or cholesterol synthesis pathways [27–31].
INSIG1 is essential for feedback regulation of cholesterol syn-
thesis [27]. Similarly, sucrose-induced vacuolation of human
fibroblast cell lines has been shown to up regulate many of the
same genes relating to fatty acid and cholesterol synthesis [32]
that were observed in the present study. These authors suggested
that cholesterol, fatty acid, and vesicle-trafficking genes are up
regulated as a function of cellular swelling and increased cellular
demand for sterol and lipid membrane components. Although
this mechanism of up regulated lipid gene expression cannot be
excluded, cell swelling does not appear to be a common charac-
teristic of AZA-1-induced cytotoxicity (see Fig. 2 and Twiner
et al. [16]). Mutated fibroblasts lacking a functional NADPH
steroid dehydrogenase-like (NSDHL) gene, which is necessary
294 M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 8

Author's personal copy
Fig. 5. Effects of azaspiracid-1 on cholesterol biosynthesis gene expression. T lymphocyte cells were exposed to AZA-1 (10 nM) for 4 and 24 h. Each box represents a
gene isoform capable of expressing a protein involved in the biosynthesis of cholesterol from acetyl-CoA. Each box is labeled with an abbreviated gene name in bold
italicandthenumberrepresentsthefoldchangerelativetothetime-matchedcontrolexpression(foldchange=1).Normalfontrepresentsthesubstrates.Geneexpression
levels are categorized as significant (red) or not significant (gray) accordingto fold change(≥1.5-fold) and p value (V104). Biochemical pathwaymaps were generated
using Gene Map Annotator and Pathway Profiler 2.0 (GenMapp). Gene names are HMGCS1, 3-hydroxy-3-methylglutaryl–coenzyme A synthase 1; HMGCR, 3-
hydroxy-3-methylglutaryl-coenzymeAreductase;MVK,mevalonatekinase;PMVK,phosphomevalonatekinase;MVD,mevalonate(diphospho)decarboxylase;IDI1,
isopentenyl-diphosphate δ-isomerase 1; FDPS, farnesyl diphosphate synthetase; SQLE, squalene epoxidase; LSS, lanosterol synthase; CYP51A1, cytochrome P450,
family 51, subfamily A, polypeptide 1 (lanosterol 14 α-demethylase); SC4MOL, sterol-C4-methyl oxidase-like; NSDHL, NAD(P)-dependent steroid dehydrogenase;
SC5DL, sterol-C5-desaturase-like; DHCR7, 7-dehydrocholesterol reductase; DHCR24, 24-dehydrocholesterol reductase; LDLR, low-density lipoprotein receptor;
ACAT, acetyl-coenzyme A acetyltransferase 2.
295M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 9

Author's personal copy
for cholesterol synthesis, revealed a high degree of up regulated
cholesterol and fatty acid synthesis gene expression [31], not
unlike the data presented herein. Similarly, cytochrome P450
knockout mice were shown to have elevated expression levels of
multiple lipid synthesis genes [33], for which cytochrome en-
zyme oxidation is one of the final steps in cholesterol synthesis.
The transcription factor family of CCAAT/enhancer binding
proteins (C/EBP) are key regulators of adipogenesis (i.e., fatty
acid and cholesterol synthesis) and contain a highly conserved
DNA-binding region that is very similar to that of SREBPs [34].
C/EBPs are known to bind to a promoter region of the LDLR
gene, inducing enhanced expression of this cholesterol receptor
[35,36]. The induced expression of C/EBP by AZA-1 may
therefore have enhanced the up regulation of fatty acid and
cholesterol synthesis pathways observed in this study.
Alterationofthelipidbiosynthesispathwayisconsistentwith
reports of altered secondary messenger signaling [18], cytotoxi-
city [16], and developmental retardation [15] following AZA-1
exposure.Consideringthedependenceofmembranefunctionon
cholesterol levels [37], the effects of AZA-1 on particular mem-
brane proteins such as claudins [9] and cadherins [20] reported
by others are not unexpected. In the current study, the tight-
junction protein claudin 5 was found to be down regulated at the
1-h time point (see supplementary data on the GEO database).
Insulin and glucagon signaling pathway
Many of the early responding (i.e., 1 h) genes tended to
involve insulin and/or glucagon signaling pathways. Cellular
energy stores, such as glucose, are well described modulators of
these pathways [38], in which glucagon receptor expression is
positively regulated by glucose and negatively regulated by
glucagon, as well as agents that increase intracellular cAMP
[39]. In human lymphocytes similar to the cells used in this
study, AZA-1 was shown to increase intracellular Ca2+and
cAMPlevels[18],potentiallystimulatingglucagonand/orcAMP-
dependent pathways. The glucagon receptor gene (GCGR), in
addition to neuronatin (NNAT), 6-phosphofructo-2-kinase/
fructose-2,6-bisphosphatase (PFKFB3), and α-1,4-galactosyl-
transferase (A4GALT), are each involved with the glucagon/
insulin pathways and were all down regulated at 1 h following
exposure to AZA-1. The down regulation of neuronatin, a pro-
teolipid membrane protein that responds positively to glucose-
mediated insulin secretion [40], suggests stimulation of an
insulin-opposing pathway. Similarly, PFKFB3, whose expres-
sion is also stimulated by insulin [41], was down regulated. The
observed down regulation of CBP/MORF (cAMP-responsive
element binding protein/monocytic leukemia zinc-finger pro-
tein-related factor), which is under partial cAMP control, can
result in stimulated expression of lipogenic genes such as fatty
acid synthase (FAS) and ATP-citrate lyase (ATP-CL) [42].
WNT signaling pathway
The WNTcell signaling pathway is a receptor-mediated path-
waythattightlycontrolscell-to-cellcommunicationandisknown
to be expressed in T lymphocyte cells [43]. As a regulator of cell
proliferation and differentiation, the primary component of the
WNT pathway is often, but not exclusively, β-catenin, a tran-
scriptional cofactor with T cell factor (TCF)/lymphoid enhancer
factor[44].The Tcell factor TCF7L2-HMGboxandCBP,which
werebothdownregulated byAZA-1,interactwith β-cateninand
have been implicated in transcriptional activation of the WNT
pathway [45]. Interestingly, the Ras-related GTPase ARL7 and a
receptor for the WNT pathway, Frizzled-9 (FZD9), were con-
currently down regulated at 1 h and may represent gene ex-
pression feedback control following initial activation by AZA-1.
LDLR has been shown to play an essential role in ligand-
mediated WNT receptor signaling [46]. These findings may be
intricately associated with the observations of Ronzitti et al.
[20], revealing effects of AZA-1 on E-cadherins, which are
known to affect the WNT pathway [44].
Fig. 6. Effectofazaspiracid-1onlevelsofLDLR.Immunoblotanalysiswasused
to evaluate the expression of LDLR in detergent extracts of azaspiracid-treated
T lymphocyte cells. Equal amounts of protein (40 μg) were separated by SDS–
PAGE under reducing conditions. A representative immunoblot scan at 120 kDa
is illustrated. Scanning densitometry was performed using NIH ImageJ and
graphical data (means±SE) are representative of three independent experiments.
Data from the 10 nM treatment at 24 h were not available (n.a.) because of
cytotoxicity and insufficient amount of protein to be loaded.
difference at pb0.001 relative to corresponding time-matched controls.
*Significant
Fig. 7. Effect of azaspiracid-1 on cellular cholesterol levels. T lymphocyte cells
were exposed to AZA-1 (1 or 10 nM) for 0, 4, 12, and 24 h. Control cells were
exposed to equivalent amount of methanol (0.1% v/v). Data (means±SE, n=3)
are expressed as total cellular cholesterol (free + ester) per unit of cellular
protein.*Significant difference at pb0.001 relative to time-matched controls.
296 M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 10

Author's personal copy
Inflammatory, ion channel, cytoskeletal, and cell growth
and division pathways
Exposure of cells and organisms to toxic agents often elicits
an initial inflammatory response. Inflammatory genes such as
the calcium-binding protein S100A4, tissue plasminogen
activator, and cytochrome c oxidase (COX) IV were each
down regulated at the 1-h time point after treatment with AZA-
1. However, the calcium-binding protein S100A4 not only is
involved in inflammation [47], but also enhances metastasis via
effects on the cytoskeleton, apoptotic pathways, and plasmino-
gens [48]. Although it does not appear that AZA-1 causes
cytotoxicity via an apoptotic mechanism in lymphocyte cells
(unpublished observation), it is clear that AZA-1 does induce
dramatic effects on cytoskeletal elements [16,18].
Cytoskeletal changes induced by AZA-1 may also affect ion
transport in cells as evidenced by up regulation of the chloride
channel 6 (CLCN6) gene. CLCN expression was shown to be
elevated during stress-induced periods of cellular swelling
[32,49] and inflammation [50]. These voltage-gated chloride
channels are known to aid in cholesterol trafficking [32] and ion
balance. Chloride channels are co-localized with sarco/endo-
plasmic-reticulum Ca2+pumps [51], which may account for the
various effects of several AZA congeners on cytosolic calcium
ion levels [18,52,53]. The effects of AZA-1 on CLCN6 gene
expression might also be manifested as alterations in bioelec-
trical activity of spinal cord neurons that appears to involve the
GABA neurotransmitter system [54] and may be linked directly
to GABA receptor gating of chloride flux [55].
Cyclin G2 (CCNG2), an important gene for cell cycle mitotic
control[56]knowntobeatargetforthetumorsuppressorprotein
p53, a regulator of cell growth and apoptosis [57], was sig-
nificantly up regulated at 4 and 24 h. While p53 gene expression
was not affected significantly by AZA-1, 5- to 14-fold up regu-
lation of CCNG2 gene expression was observed in lymphocytes
during periods of growth inhibition induced by dexamethasone,
a known inducer of apoptosis [58]. Jurkat lymphocyte T cell
growth inhibition induced by AZA-1, independent of cyclin G2
up regulation, has been clearly demonstrated, in which cyto-
toxicity and membrane disruption are protracted responses,
typically after ≥18 h of continuous exposure [16].
Gene expression summary and conclusions
Themajorityofdifferentiallyexpressedgenesidentifiedinthis
study tend to encode proteins withsimilar and/or related pathway
functionality. T lymphocyte cells exposed to AZA-1 initially
decreased the expression of transcription factor genes, genes
related to membrane function and ion homeostasis, and key
receptors and membrane proteins of the glucagon and WNT
pathway. With continued AZA-1 treatment, dramatic and
coordinated up regulation of nearly all cholesterol and fatty acid
synthesis genes as wellas the LDLRprotein was observed, likely
inresponsetoreduced levelsof cellular cholesterol.Interestingly,
the glucagon/insulin, WNT, and cholesterol/fatty acid synthesis
pathways are highly integrated based on their overlapping
functionality, thereby suggesting that the effects elicited by
AZA-1 may be via the toxin binding to a single, common target.
The T lymphocyte cells employed herein, a sensitive cell
type identified previously from in vivo studies [5,14,17],
clearly exhibited differential expression of the cholesterol
biosynthesis pathway in concert with reduced intracellular
levels of cholesterol and up regulated expression of the LDLR
protein. While the mechanism of action for AZA-1 remains
uncertain, other investigators have proposed the interaction of
AZA-1 with membrane proteins such as claudins [9] and
cadherins [20]. Our data are consistent with this argument, as
many of these same genes showed differential expression.
Altered levels of such membrane proteins could, in turn, be
expected to affect both lipid biosynthesis gene expression and
cytoskeletal rearrangement as observed in our current and past
studies [16]. The data presented herein not only will serve to
guide future hypothesis-driven investigations aimed at identi-
fying the molecular target of AZA-1, but also can be used to
develop exposure biomarkers and/or to assess the therapeutic
potential of AZA-1.
Materials and methods
Azaspiracid purification
Azaspiracid was extracted from 2 kg of mussels (M. edulis) that were
collectedin 1996 fromKillary Harbour, on the west coast of Ireland,and in 1999
from Bantry Bay, on the southwest coast of Ireland. Toxins were extracted and
purified in 2001 (N93% pure by NMR and showed b1% impurity of other AZA
subtypes/congeners), as previously described [16].
Cell lines and cytotoxicity
Human Jurkat E6-1 T lymphocyte cells (American Type Culture Collection
TIB-152; Manassas, VA, USA) were grown as described in Twiner et al. [16].
Cytotoxicity was determined using the Vybrant cytotoxicity assay kit (Cat. No.
V-23111; Invitrogen, Carlsbad, CA, USA) as per the manufacturer's instructions.
Photomicrographs of unfixed cells were taken using an Axiovert S100 epifluo-
rescencemicroscope(CarlZeiss,Inc.,Thornwood,NY,USA).Foreachexperiment
(n=3 biological replicates), between 1.2 and 10.6×106cells were resuspended in
fresh RPMI medium supplemented with FBS (10%) and allowed to acclimate for
N12 h prior to the addition of AZA-1 (1 or 10 nM final) or equivalent amounts of
methanolic vehicle (0.1% final). Cells were harvested for RNA extractions at 1, 4,
and 24 h and for protein and cholesterol extractions at 4, 12, and 24 h.
RNA processing
Immediately following centrifugation at 1000 ×g for 5 min, cells were
disrupted by resuspending in 1 ml Tri-Reagent (Molecular Research Center,
Inc., Cincinnati, OH, USA). All samples were processed according to the
manufacturer's protocol. Total RNA was resuspended in DEPC water, purified
with a Qiagen RNeasy column (Valencia, CA, USA), and quantified by UV–Vis
spectroscopy. The RNA was then qualified on an Agilent 2100 Bioanalyzer
(Palo Alto, CA, USA) to confirm the yield of high-quality RNA.
RNA labeling and array hybridization
Four hundred nanograms of total RNA from each time-matched control and
treatment sample was amplified separately and labeled with either Cy3- or Cy5-
conjugated CTP (Perkin–Elmer, Boston, MA, USA) with a low input linear
amplification kit (Agilent Technologies) according to the manufacturer's pro-
tocol. After labeling and cleanup, amplified RNA was quantified by UV–Vis
spectroscopy. One microgram each of Cy3- and Cy5-labeled targets were
combined and hybridized to an Agilent whole human genome oligonucleotide
microarray (Cat. No. G4112A) array for 17 h at 60°C. After hybridization, arrays
297 M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 11

Author's personal copy
were washed consecutively in solutions of 6×SSPE with 0.005% N-lauroylsar-
cosine and 0.06×SSPE with 0.005% N-lauroylsarcosine for 1 min each at room
temperature, followed by a 30-srinse in Agilent stabilization anddrying solution.
Threebiologicalreplicates,oneofwhichwasusedinadyeswap,wereperformed
at the 4- and 24-h time points.
Microarray analysis
Microarrays were imaged using an Agilent microarray scanner. Images were
extracted with Agilent Feature Extraction software version A7.5.1 and data
analyzedwiththeRosetta Resolver7.0geneexpressionanalysissystem(Rosetta
Informatics, Seattle, WA, USA). Using a rank consistency filter, features were
subjected to a combination linear and lowess normalization algorithm. Based on
the Rosetta error modeldesignedfor the Agilent platform,a composite array was
generated at each time point, in which the data for each feature underwent a
weighted averaging based on feature quality in the triplicate arrays making up
the composite. A list of “signature” gene features was then generated for each
time point from the composite array by p value sorting and absolute differential
expression ≥1.5-fold (≥0.58 and ≤−0.58 log2ratio values). Gene feature
clusters in the trend plots were grouped using a K-means algorithm with
Euclidean distance measurements for genes with a p value b10−4.
Quantitative real-time PCR
One microgram of total RNA was reverse transcribed using Ambion's
RETROscript kit (Austin, TX, USA) with oligo(dT) primers for the two-step
qPCR assays.Gene-specific primers (Supplementary Table 5) wereusedtoamplify
message by qPCR using an ABI 7500 with ABI SYBR green master mix (Foster
City, CA, USA). All 25 µl qPCRs were run in triplicate and contained 1× SYBR
green PCR master mix, 400 nM gene-specific primers, and 1 μl of reverse-tran-
scribed sample. Statistical analyses were performed using JMP software (version
5.1.2;SASInstitute,Inc.,Cary,NC,USA).Forfurtherdetails,seeMoreyetal.[21].
Analysis of microarray and qPCR correlation
The correlation of qPCR with microarray data was analyzed using JMP
software (version 5.1.2; SAS Institute, Inc.). All calculations of correlation were
performed on the log2ratio value from the composite array and the correspond-
ing mean log2ratio from qPCR analyses. Because the data were normally
distributed Pearson's correlation calculation was used. One-way ANOVAs were
then used to determine the relationship between the observed correlations. An α
value of 0.05 was used for all tests.
Analysis of cellular signaling pathway alterations
TheGeneMapAnnotatorandPathwayProfiler(GenMAPP)andMAPPFinder
analysis softwares were used to examine various cellular signaling pathways for
AZA-specific alterations in microarray gene expression [59]. MAPPFinder uses
Gene Ontology (GO)-based annotations for displaying and analyzing the gene
expression data in the GO hierarchy [60]. GO terms identified in MAPPFinder
with high Z scores were then used to direct efforts toward signaling pathways of
interest illustrated in GenMAPP. Z scores were calculated by subtracting the
expectednumberofdifferentiallyexpressedgenesfromtheobservednumberand
then dividing by the standard deviation of the observed number of genes. A
positiveZscoreindicatedthatthereweremoregenesdifferentiallyexpressedina
GO term/pathway than would be expected by random chance. All gene expres-
sion data, regardless of relative fold change or p value, were loaded into the
GenMAPP program, in which the data were plotted onto specific biological
signaling pathways for visualization. Gene expression was visualized using a
color coding system, whereby genes were differentially colored according to up
or down regulation, fold change (≤−1.5 or ≥1.5), and p value (pV10−4).
Immunoblot analysis of LDLR
Cells were pelleted and lysed using ice-cold Extraction Buffer (Stressgen
Cat. No. 80–1526) containing a protease inhibitor mixture (EDTA-free; Roche
Diagnostics Cat. No. 10481700) and 0.1 mM PMSF. Sampleswere incubated on
ice for 30 min and then cleared by centrifugation at 15,000 ×g for 10 min at 4°C.
Protein concentrations of the extractswere determined by the BCA protein assay
(Pierce). Equal amounts of protein (30 to 40 μg) were run under reducing
conditions on NuPAGE 3–8% Tris–acetate gels (Invitrogen) and transferred to
pure nitrocellulose membranes. Following transfer, membranes were incubated
in 0.2% Ponceau-S in 1 M acetic acid to ensure even loading. Membranes were
blocked by incubation in Tris-buffered saline containing 5% nonfat dry milk and
0.1% Tween 20 for 1 h at room temperature, washed three times in Tris-buffered
saline, and incubated overnight with the primary antibody at 4°C. Membranes
were washed three times with Tris-buffered saline and incubated with the
corresponding horseradish peroxidase-conjugated anti-IgG for 1 h at room
temperature. LDLR was probed using a biotin-linked antibody (Abcam Cat. No.
ab50296) with NeutrAvidin (Pierce Cat. No. 31000) biotin-binding, HRP-
conjugated protein. Detection of proteins was achieved by the Western
Lightning ECL system (Perkin–Elmer LAS, Inc., Cat. No. NEL102). For
relative protein quantification, representative blots were scanned using an Epson
Perfection 2450 scanner and analyzed using NIH ImageJ version 1.38x.
Significant differences between means were assessed by an ANOVA followed
by a Tukey multiple t-test where pb0.001 were considered significant.
Quantitative analysis of cellular cholesterol by gas
chromatography
Cells (2×106) from triplicate experimental flasks were pelleted and washed
twice in 10 ml phosphate-buffered saline. Cellular cholesterol was extracted and
assayed as previously described by Ishikawa et al. [61]. Briefly, pellets were
extracted with 5 ml of extraction solution (3 ml hexane, 2 ml 2-propanol)
including 5 μg stigmasterol (Sigma Cat. No. S2424) as internal standard for 45
to 60 min. Samples were centrifuged at 1500 ×g for 10 min and supernatants
were collected and divided into two equal volumes for separate extraction of
“free cholesterol” and “total cholesterol” and then dried under nitrogen. Free
cholesterol samples were immediately resuspended in chloroform (200 μl) and
prepared for GC injection. Total cholesterol samples were resuspended in 100 μl
tetramethylammonium hydroxide (Sigma Cat. No. 334901; 25% in 2-propanol)
and incubated at 80°C for 20 min. Samples were vortexed and 50 μl of
tetrachloroethylene/methylbutyrate (1:3; Sigma Cat. Nos. 270393 and 246093)
solution and 200 μl Milli-Q water were added. Samples were centrifuged at
2500 ×g and the lipid-soluble fractions were collected for GC injection.
CholesterolquantificationwascarriedoutonanAgilentTechnologiesModel
6890N gas chromatograph system outfitted with a Model DB-17 capillary
column (J&W Scientific Cat. No. 125–1712). Total and free cholesterol data
werenormalizedtocellnumber(hemocytometerandCoultercountercounts)and
cellular protein. Significant differences between means were assessed by an
ANOVA followed by a Tukey multiple t-test where pb0.001 were considered
significant.
Acknowledgments
Funding for the Marine Institute, Ireland, was obtained from
the Irish National Development Plan under Marine Research
Strategic Project ST-02-02, Azaspiracids Standards and Tox-
icology. M.J.T. was supported by a National Research Council
Associateship Award through NOAA/NOS/NCCOS/CCEHBR.
S.M.H. was supported by NIH Grant HL079274 and the South
Carolina COBRE in Lipidomics and Pathobiology (P20
RR17677 from NCRR). The authors thank Arjen Gerssen for
his help with ChemDraw and Dr. Richard Klein and Charlyne
Chassereau for their technical assistance with cholesterol gas
chromatography.
Appendix A. Supplementary data
Supplementary data associated with this article can be found,
in the online version, at doi:10.1016/j.ygeno.2007.10.015.
298M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 12

Author's personal copy
References
[1] P. Hess, T. McMahon, D. Slattery, D. Swords, G. Dowling, M. McCarron,
D. Clarke, W. Gibbons, J. Silke, M. O'Cinneide, Use of LC-MS testing to
identify lipophilic toxins, to establish local trends and interspecies
differences and to test the comparability of LC-MS testing with the
mouse bioassay: an example from the Irish Biotoxin Monitoring Prog-
ramme 2001. Molluscan Shellfish Safety, Proceedings of the Fourth In-
ternational Conference on Molluscan Shellfish Safety, Consellería de
Pesca e Asuntos Marítimos da Xunta de Galicia, and Intergovernmental
Oceanographic Commissionof UNESCO, Santiagode Compostella, 2003,
pp. 57–65.
[2] A. Furey, C. Moroney, A.B. Magdalena, M.J.F. Saez, M. Lehane, K.J.
James, Geographical, temporal, and species variation of the polyether
toxins, azaspiracids, in shellfish, Environ. Sci. Technol. 37 (2003)
3078–3084.
[3] P.Hess,T.McMahon,D.Slattery,D.Swords,G.Dowling,M.McCarron,D.
Clarke, L. Devilly, W. Gibbons, J. Silke, M. O'Cinneide, Biotoxin chemical
monitoring in Ireland 2001, Proceedings of the Second Irish Marine Science
Biotoxin Workshop, Marine Institute, Dublin, 2001, pp. 8–18.
[4] M. Satake, K. Ofuji, K.J. James, A. Furey, T. Yasumoto, New toxic event
caused by Irish mussels, in: B. Reguera, J. Blanco, M.L. Fernandez, T.
Wyatt (Eds.), Harmful Algae, Xunta de Galicia and Intergovernmental
Oceanographic Commissionof UNESCO, Santiagode Compostella, 1998,
pp. 468–469.
[5] E. Ito, M. Satake, K. Ofuji, M. Higashi, K. Harigaya, T. McMahon, T.
Yasumoto, Chronic effects in mice caused by oral administration of
sublethal doses of azaspiracid, a new marine toxin isolated from mussels,
Toxicon 40 (2002) 193–203.
[6] K.J. James, A. Furey, M. Lehane, H. Ramstad, T. Aune, P. Hovgaard, S.
Morris, W. Higman, M. Satake, T. Yasumoto, First evidence of an
extensive northern European distribution of azaspiracid poisoning (AZP)
toxins in shellfish, Toxicon 40 (2002) 909–915.
[7] A.B. Magdalena, M. Lehane, S. Krys, M.L. Fernandez, A. Furey, K.J.
James, The first identification of azaspiracids in shellfish from France and
Spain, Toxicon 42 (2003) 105–108.
[8] H. Taleb, P. Vale, R. Amanhir, A. Benhadouch, R. Sagou, A. Chafik, First
detection of azaspiracids in mussels in north west Africa, J. Shellfish Res.
25 (2006) 1067–1070.
[9] P. Hess, P. McCarron, N. Rehmann, J. Kilcoyne, T. McMahon, G. Ryan,
P.M. Ryan, M.J. Twiner, G.J. Doucette, E. Ito, T. Yasumoto, Isolation
and purification of azaspiracids from naturally contaminated materials,
and evaluation of their toxicological effects—final project report ASTOX
(ST/02/02), Marine Institute: Marine Environment & Health Series,
2007, p. 129.
[10] T. McMahon, J. Silke, Re-occurrence of winter toxicity, Harm. Algae
News 17 (1998) 12.
[11] T. McMahon, J. Silke, Winter toxicity of unknown aetiology in mussels,
Harm. Algae News 14 (1996) 2.
[12] K. Ofuji, M. Satake, T. McMahon, J. Silke, K.J. James, H. Naoki, Y.
Oshima, T. Yasumoto, Two analogs of azaspiracid isolated from mussels,
Mytilus edulis, involved in human intoxication in Ireland, Nat. Toxins 7
(1999) 99–102.
[13] E. Ito, K. Terao, T. McMahon, J. Silke, T. Yasumoto, Acute pathological
changes in mice caused by crude extracts of novel toxins isolated from
Irish mussels, in: B. Reguera, J. Blanco, M.L. Fernandez, T. Wyatt
(Eds.), Harmful Algae, Xunta de Galicia and Intergovernmental
Oceanographic Commission of UNESCO, Santiago de Compostella,
1998, pp. 588–589.
[14] E. Ito, M. Satake, K. Ofuji, N. Kurita, T. McMahon, K. James, T.
Yasumoto, Multiple organ damage caused by a new toxin azaspiracid,
isolated from mussels produced in Ireland, Toxicon 38 (2000) 917–930.
[15] J.R. Colman, M.J. Twiner, P. Hess, T. McMahon, M. Satake, T. Yasumoto,
G.J.Doucette,J.S.Ramsdell,Teratogeniceffectsofazaspiracid-1identified
by microinjection of Japanese medaka (Oryzias latipes) embryos, Toxicon
45 (2005) 881–890.
[16] M.J. Twiner, P. Hess, M.-Y. Bottein Dechraoui, T. McMahon, M.S.
Samons, M. Satake, T. Yasumoto, J.S. Ramsdell, G.J. Doucette, Cytotoxic
and cytoskeletal effects of azaspiracid-1 on mammalian cell lines, Toxicon
45 (2005) 891–900.
[17] E. Ito, M.O. Frederick, T.V. Koftis, W. Tang, G. Petrovic, T. Ling, K.C.
Nicolaou, Structure toxicity relationships of synthetic azaspiracid-1 and
analogs in mice, Harmful Algae News 5 (2006) 586–591.
[18] Y. Roman, A. Alfonso, M.C. Louzao, L.A. de la Rosa, F. Leira, J.M.
Vieites, M.R. Vieytes, K. Ofuji, M. Satake, T. Yasumoto, L.M. Botana,
Azaspiracid-1, a potent, nonapoptotic new phycotoxin with several cell
targets, Cell Signal. 14 (2002) 703–716.
[19] N. Vilarino, K.C. Nicolaou, M.O. Frederick, E. Cagide, I.R. Ares, M.C.
Louzao, M.R. Vieytes, L.M. Botana, Cell growth inhibition and actin
cytoskeleton disorganization induced by azaspiracid-1 structure–activity
studies, Chem. Res. Toxicol. 19 (2006) 1459–1466.
[20] G. Ronzitti, P. Hess, N. Rehmann, G.P. Rossini, Azaspiracid-1 alters the
E-cadherin pool in epithelial cells, Toxicol. Sci. 95 (2007) 427–435.
[21] J. Morey, J. Ryan, F. Van Dolah, Microarray validation: factors influencing
correlation between inkjet printed oligonucleotide microarrays and real-
time PCR, Biol. Proced. Online 8 (2006) 175–193.
[22] S. Morikawa, T. Murakami, H. Yamazaki, A. Izumi, Y. Saito, T.
Hamakubo, T. Kodama, Analysis of the global RNA expression profiles
of skeletal muscle cells treated with statins, J. Atheroscler. Thromb. 12
(2005) 121–131.
[23] W.H. Frishman,R.C. Rapier, Lovastatin: an HMG-CoAreductase inhibitor
for lowering cholesterol, Med. Clin. North Am. 73 (1989) 437–448.
[24] A. Endo, M. Kuroda, Y. Tsujita, ML-236A, ML-236B, and ML-236C, new
inhibitorsofcholesterogenesisproducedbyPenicilliumcitrinium,J.Antibiot.
(Tokyo) 29 (1976) 1346–1348.
[25] M. Cassader, G. Ruiu, R. Gambino, N. Alemanno, F. Veglia, G. Pagano,
Hypercholesterolemiainnon-insulin-dependentdiabetesmellitus:different
effect of simvastatin on VLDL andLDL cholesterol levels, Atherosclerosis
99 (1993) 47–53.
[26] L.M. Olivier, K.L. Chambliss, K.M. Gibson, S.K. Krisans, Characteriza-
tion of phosphomevalonate kinase: chromosomal localization, regulation,
and subcellular targeting, J. Lipid Res. 40 (1999) 672–679.
[27] L.J. Engelking, G. Liang, R.E. Hammer, K. Takaishi, H. Kuriyama, B.M.
Evers, W.-P. Li, J.D. Horton, J.L. Goldstein, M.S. Brown, Schoenheimer
effect explained—feedback regulation of cholesterol synthesis in mice
mediated by Insig proteins, J. Clin. Invest. 115 (2005) 2489–2498.
[28] D.Eberle,B.Hegarty,P.Bossard,P.Ferre,F.Foufelle,SREBPtranscription
factors: master regulators of lipid homeostasis, Biochimie 86 (2004)
839–848.
[29] B.-L. Song, N.B. Javitt, R.A. DeBose-Boyd, Insig-mediated degradation
of HMG CoA reductase stimulated by lanosterol, an intermediate in the
synthesis of cholesterol, Cell Metab. 1 (2005) 179–189.
[30] J.D. Horton, J.L. Goldstein, M.S. Brown, SREBPs: activators of the
completeprogramofcholesterolandfattyacidsynthesisintheliver,J.Clin.
Invest. 109 (2002) 1125–1131.
[31] D. Cunningham, D. Swartzlander, S. Liyanarachchi, R.V. Davuluri, G.E.
Herman, Changes in gene expression associated with loss of function of
the NSDHL sterol dehydrogenase in mouse embryonic fibroblasts, J. Lipid
Res. 46 (2005) 1150–1162.
[32] A. Helip-Wooley, J.G. Thoene, Sucrose-induced vacuolation results in
increased expression of cholesterol biosynthesis and lysosomal genes,
Exp. Cell Res. 292 (2004) 89–100.
[33] X.Wang,M.Chamberlain,O.Vassieva,C.Henderson,C.Wolf,Relationship
between hepatic phenotype and changes in gene expression in cytochrome
P450 reductase (POR) null mice, Biochem. J. 388 (2005) 857–867.
[34] L. Weber, M. Boll, A. Stampfl, Maintaining cholesterol homeostasis:sterol
regulatory element-binding proteins, World J. Gastroenterol. 10 (2004)
3081–3087.
[35] N.Sekar,J.D.Veldhuis,InvolvementofSp1andSREBP-1aintranscriptional
activation of the LDL receptor gene by insulin and LH in cultured porcine
granulosa–luteal cells, Am. J. Physiol., Endocrinol. Metab. 287 (2004)
E128–E135.
[36] E.D. Rosen, C.J. Walkey, P. Puigserver, B.M. Spiegelman, Transcriptional
regulation of adipogenesis, Genes Dev. 14 (2000) 1293–1307.
[37] D. Lambert, C.A. O'Neill, P.J. Padfield, Depletion of Caco-2 cell
cholesterol disrupts barrier function by altering the detergent solubility
299M.J. Twiner et al. / Genomics 91 (2008) 289–300

Page 13

Author's personal copy
and distribution of specific tight-junction proteins, Biochem. J. 387 (2005)
553–560.
[38] S.-H. Koo, L. Flechner, L. Qi, X. Zhang, R.A. Screaton, S. Jeffries, S.
Hedrick, W. Xu, F. Boussouar, P. Brindle, H. Takemori, M. Montminy, The
CREB coactivator TORC2 is a key regulator of fasting glucose metabolism,
Nature 437 (2005) 1109–1111.
[39] N. Abrahamsen, K. Lundgren, E. Nishimura, Regulation of glucagon
receptor mRNA in cultured primary rat hepatocytes by glucose and cAMP,
J. Biol. Chem. 270 (1995) 15853–15857.
[40] K. Chu, M.-J. Tsai, Neuronatin, a downstream target of BETA2/NeuroD1
in the pancreas, is involved in glucose-mediated insulinsecretion, Diabetes
54 (2005) 1064–1073.
[41] L. Riera, A. Manzano, A. Navarro-Sabate, J.C. Perales, R. Bartrons,
Insulin induces PFKFB3 gene expression in HT29 human colon
adenocarcinoma cells, Biochim. Biophys. Acta 1589 (2002) 89–92.
[42] S. Herzig, S. Hedrick, I. Morantte, S.-H. Koo, F. Galimi, M. Montminy,
CREBcontrolshepatic lipidmetabolismthroughnuclearhormonereceptor
PPAR-γ, Nature 426 (2003) 190–193.
[43] D.J.VanDenBerg,A.K.Sharma,E.Bruno,R.Hoffman,Roleofmembersof
the Wnt gene family in human hematopoiesis, Blood 92 (1998) 3189–3202.
[44] W.J. Nelson, R. Nusse, Convergence of Wnt, β-catenin, and cadherin
pathways, Science 303 (2004) 1483–1487.
[45] R. Nusse, WNT targets: repression and activation, Trend Genet. 15 (1999)
1–3.
[46] X. Zeng, K. Tamai, B. Doble, S. Li, H. Huang, R. Habas, H. Okamura, J.
Woodgett, X. He, A dual-kinase mechanism for Wnt co-receptor
phosphorylation and activation, Nature 438 (2005) 873–877.
[47] R. Eckert, A. Broome, M. Ruse, N. Robinson, D. Ryan, K. Lee, S100
proteins in the epidermis, J. Invest. Dermatol. 123 (2004) 23–33.
[48] S.C. Garrett, K.M. Varney, D.J. Weber, A.R. Bresnick, S100A4, a mediator
of metastasis, J. Biol. Chem. 281 (2006) 677–680.
[49] N. Comes, X. Gasull, A. Gual, T. Borras, Differential expression of the
human chloride channel genes in the trabecular meshwork under stress
conditions, Exp. Eye Res. 80 (2005) 801–813.
[50] Y.-P. Dai, S. Bongalon, W.J. Hatton, J.R. Hume, I.A. Yamboliev, ClC-3
chloride channel is upregulated by hypertrophy and inflammation in rat
and canine pulmonary artery, Br. J. Pharmacol. 145 (2005) 5–14.
[51] G. Buyse, D. Trouet, T. Voets, L. Missiaen, G. Droogmans, B. Nilius, J.
Eggermont, Evidence for the intracellular location of chloride channel
(ClC)-type proteins: co-localization of ClC-6a and ClC-6c with the sarco/
endoplasmic-reticulum Ca2+pump SERCA2b, Biochem. J. 330 (1998)
1015–1021.
[52] A. Alfonso, Y. Roman, M.R. Vieytes, K. Ofuji, M. Satake, T. Yasumoto,
L.M. Botana, Azaspiracid-4 inhibits Ca2+entry by stored operated channels
in human T lymphocytes, Biochem. Pharmacol. 69 (2005) 1627–1636.
[53] Y. Roman, A. Alfonso, M.R. Vieytes, K. Ofuji, M. Satake, T. Yasumoto,
L.M.Botana, Effects of azaspiracids2 and 3 on intracellular cAMP, [Ca2+],
and pH, Chem. Res. Toxicol. 17 (2004) 1338–1349.
[54] K.V. Kulagina, M.J. Twiner, P. Hess, T. McMahon, M. Satake, T.
Yasumoto, J.S. Ramsdell, G.J. Doucette, W. Ma, T.J. O'Shaughnessy,
Azaspiracid-1 inhibits bioelectrical activity of spinal cord neuronal
networks, Toxicon 47 (2006) 766–773.
[55] A.Gulacsi,C.R.Lee, A.Sik, T. Viitanen, K.Kaila,J.M.Tepper, T.F.Freund,
Cell type-specific differences in chloride-regulatory mechanisms and
GABAAreceptor-mediated inhibition in rat substantia nigra, J. Neurosci.
23 (2003) 8237–8246.
[56] M.C. Horne, G.L. Goolsby, K.L. Donaldson, D. Tran, M. Neubauer, A.F.
Wahl, Cyclin G1 and cyclin G2 comprise a new family of cyclins with
contrasting tissue-specific and cell cycle-regulated expression, J. Biol.
Chem. 271 (1996) 6050–6061.
[57] T. Jacks, R.A. Weinberg, Cell-cycle control and its watchman, Nature 381
(1996) 643–644.
[58] M.C. Horne, K.L. Donaldson, G.L. Goolsby, D. Tran, M. Mulheisen, J.W.
Hell, A.F. Wahl, Cyclin G2 is up-regulated during growth inhibition and B
cell antigen receptor-mediated cell cycle arrest, J. Biol. Chem. 272 (1997)
12650–12661.
[59] K.D. Dahlquist, N. Salomonis, K. Vranizan, S.C. Lawlor, B.R. Conklin,
GenMAPP, a new tool for viewing and analyzing microarray data on
biological pathways, Nat. Genet. 31 (2002) 19–20.
[60] S. Doniger, N.Salomonis, K. Dahlquist,K. Vranizan, S. Lawlor, B.Conklin,
MAPPFinder: using Gene Ontology and GenMAPP to create a global gene-
expression profile from microarray data, Genome Biol. 4 (2003) R7.
[61] T.T.Ishikawa,J.MacGee,J.A.Morrison,C.J.Glueck,Quantitativeanalysis
of cholesterol in 5 to 20 μl of plasma, J. Lipid Res. 15 (1974) 286–291.
300 M.J. Twiner et al. / Genomics 91 (2008) 289–300