Microarray and functional cluster analysis implicates transforming growth factor beta1 in endothelial cell dysfunction in a swine hemorrhagic shock model.

Abstract

Trauma leading to massive hemorrhage results in widespread tissue hypoxia, anaerobic metabolism, and production of inflammatory cytokines and oxidative molecules injurious to the vascular endothelium. Although trauma-related endothelial cell pathophysiology has been extensively studied, very little is known regarding gene transcriptional changes that occur during the event, particularly in endothelia. Thus, we employed fluorescent microarray analysis of gene transcription to elucidate critical pathways and gene products involved in endothelial dysfunction.
A trauma-hemorrhage/shock (T-H/S) model mimicking the physiologic changes seen in human trauma was performed on 10 Yorkshire swine, consisting of 35% blood volume hemorrhage followed by 6 h of full resuscitation. Aortic endothelium was analyzed by microarray and functional clusters were identified through the use of Database for Annotation, Visualization, and Integrated Discovery (DAVID) software.
Injured swine developed profound acidosis, coagulopathy, massive resuscitative fluid requirements, and microscopic changes of ischemia/reperfusion injury. While 1007 transcripts were down-regulated, 529 transcripts were up-regulated. DAVID functional clustering analysis revealed 21 significantly altered biological processes that were grouped into 12 distinct functional categories. The transforming growth factor beta (TGFβ) family of genes was the most interrelated. In addition, vascular endothelial growth factor (VEGF) signaling members and leukocyte chemoattractants were also altered.
Our model identified two major signaling pathways, TGFβ and VEGF, which undergo early transcriptional changes in injured endothelial cells. Our results suggest that TGFβ and VEGF may play a crucial role in the development of endothelial cell injury leading to increased vascular permeability during shock-trauma.

Full text

Page 1

ASSOCIATION FOR ACADEMIC SURGERY
Microarray and Functional Cluster Analysis Implicates Transforming
Growth Factor Beta1 in Endothelial Cell Dysfunction in a Swine
Hemorrhagic Shock Model
Marlin Wayne Causey, M.D.,* Zachary S. Hoffer, M.D., Ph.D.,† Seth L. Miller, M.D.,* Laurel J. Huston, M.S.,‡
Steve A. Satterly, M.D.,* Matthew Martin, M.D.,* and Jonathan D. Stallings, Ph.D.‡,1
*Department of Surgery, Madigan Army Medical Center, Tacoma, Washington; †Department of Pathology, Madigan Army Medical
Center, Tacoma, Washington; and ‡Department of Clinical Investigation, Madigan Army Medical Center, Tacoma, Washington
Submitted for publication October 19, 2010
Background. Trauma leading to massive hemor-
rhage results in widespread tissue hypoxia, anaerobic
metabolism,andproductionofinflammatorycytokines
andoxidativemoleculesinjurioustothevascularendo-
thelium. Although trauma-related endothelial cell
pathophysiology has beenextensively studied, very lit-
tle is known regarding gene transcriptional changes
thatoccurduringtheevent,particularlyinendothelia.
Thus, we employed fluorescent microarray analysis of
gene transcription to elucidate critical pathways and
gene products involved in endothelial dysfunction.
Materialsand Methods. A
shock(T-H/S)model mimicking
changes seen in human trauma was performed on 10
Yorkshire swine, consisting of 35% blood volume
hemorrhagefollowedby6hoffullresuscitation.Aortic
endothelium was analyzed by microarray and func-
tional clusters were identified through the use of Data-
base for Annotation, Visualization, and Integrated
Discovery (DAVID) software.
Results. Injured swine developed profound acido-
sis, coagulopathy, massive resuscitative fluid require-
ments,and microscopic
reperfusioninjury.While
down-regulated, 529 transcripts were up-regulated.
DAVID functional clustering analysis revealed 21 sig-
nificantly altered biological processes that were
grouped into 12 distinct functional categories. The
transforming growth factor beta (TGFb) family of
genes was the most interrelated. In addition, vascular
trauma-hemorrhage/
the physiologic
changes
1007
of ischemia/
were transcripts
endothelial growth factor (VEGF) signaling members
and leukocyte chemoattractants were also altered.
Conclusions. Ourmodel identified twomajor signal-
ing pathways, TGFb and VEGF, which undergo early
transcriptional changes in injured endothelial cells.
Our results suggest that TGFb and VEGF may play
a crucial role in the development of endothelial cell in-
juryleadingtoincreasedvascularpermeabilityduring
shock-trauma.
Published by Elsevier Inc.
Key Words: acidosis; critical care; swine; microarray;
hemorrhage; shock; trauma; endothelial dysfunction;
TGFb; VEGF.
INTRODUCTION
Following severe trauma, ‘shock syndrome’ develops
as the result of hemodynamic instability, coagulopathy,
decreased oxygen delivery, decreased tissue perfusion,
and cellular hypoxia [1, 2]. These physiologic derange-
ments ultimately lead to multiorgan failure (MOF),
a systemic inflammatory process that often leads to vi-
tal organs failure and death [3]. Supportive care to pre-
vent or reverse the onset of hypoxemia, hypovolemia,
and shock requires the immediate control of bleeding,
followed by surgical intervention and treatment in an
intensivecareunit(ICU)[4,5].Thus,currentstrategies
ofresuscitationemphasize
measures that rely on blood product replacement, in-
travenous crystalloids, reversal of coagulopathy, pre-
vention of hypothermia, and vasopressors to promote
tissue perfusion, prevent hypoxia, and minimize addi-
tional blood loss [4, 5]. With advances in resuscitation
science, however, more patients are surviving severe
injuries unheard a decade ago, but are prone to longer
early andaggressive
1To whom correspondence and reprint requests should be
addressed at Madigan Army Medical Center, Department of Clinical
Investigation, Fort Lewis, WA 98431. E-mail: jonathan.stallings@us.
army.mil.
0022-4804/$36.00
Published by Elsevier Inc.
120
Journal of Surgical Research 170, 120–132 (2011)
doi:10.1016/j.jss.2011.01.014

Page 2

ICU stays and protracted recovery with long-term
disability [6].
Mechanisms underlying the pathogenesis of vital or-
gan injury and onset of MOF are poorly understood.
Severalstudies supporttheideathatgut-derivedbacte-
rial endotoxin released in combination with hypoxia/re-
perfusion injury promote leukocyte activation [7, 8].
Indeed, inflammatory cytokines, nitric oxide, and reac-
tive oxygen species all seem to play a significant role in
hemorrhagic shock [8, 9], and resuscitative efforts can
inadvertently potentiate processes that cause vital
organ damage [9, 10]. To date, there are few proven
therapies to prevent or treat MOF.
Theendotheliumisattheforefrontoftrauma,hemor-
rhagicshock,andorgandamage.Undernormalcircum-
stances, the endothelium forms an intact barrier to
circulatingmolecules,whileallowingtransportoffluids
and nutrients to the tissues [17]. Significant trauma,
which causes local and/or global ischemia and reperfu-
sion injury during resuscitation, damages vascular en-
dothelial cells and alters active nutrient transport,
vascular permeability, immune response, and coagula-
tion. Despite aggressive fluid resuscitation, endothelial
cell dysfunction often becomes irreversible [11, 12]. In-
deed, endothelial dysfunction plays an important and
early role in the development of long term systemic or-
gan failures, particularly failures of the liver, lungs,
andkidneys.Thus,maintenanceofendothelialfunction
after traumatic injury is an obvious target of early
pharmacologic therapy as preservation of endothelial
cellular function should maximize circulating fluid
volume and cardiac output and minimize the metabolic
demands of reperfused organs.
Swine models are extensively used to examine the
physiologic impact and treatments of major trauma
because of their similar anatomy and physiology to hu-
mans. However, one limitation of many swine trauma
models has been the inability to reliably reproduce all
the metabolic derangements seen in critical trauma pa-
tients. Healthy swine have a very robust ability to com-
pensate for hemorrhage and shock, and they resist
developing the degree of acidosis and coagulopathy
seen in severely injured people. Our model of 35% blood
volume hemorrhage with reperfusion injury closely
replicates the most severe physiologic alterations seen
in critical trauma patients and the ensuing ‘shock
syndrome’.
Changes in genomic transcriptional activity invoked
inanimalmodelsofsystemicshockcanbestudiedusing
gene microarray and cluster software analysis technol-
ogies. These technologies can identify genes that signif-
icantly deviate from their basal transcriptional rate
during systemic shock and organize them into func-
tional clusters on the basis of their putative functions.
Thus, we employed fluorescent microarray technology
to discover major target genes whose expression was
significantly altered secondary to hypoxia/ischemia re-
perfusion injury in our shock model of endothelial dam-
age. We then utilized the Database for Annotation,
Visualizationand Integrated Discovery(DAVID) Bioin-
formatics Resources 6.7 at the National Institute of Al-
lergyandInfectiousDisease(NIAID)[13,14]toidentify
endothelial functional clusters and gene members ac-
tive during trauma, hemorrhage, and shock. To our
knowledge, this is the first study to combine microarray
technology with DAVID cluster analysis to identify
gene families related to endothelial dysfunction in
a swine model of hemorrhagic shock/trauma. Further-
more, we examined the physiologic derangements and
histologic findings that support the validation of our
severe hemorrhage shock model.
MATERIALS AND METHODS
Surgical Model and Histology
Following Institutional Animal Care and Usage Committee ap-
proval, 10 Yorkshire swine were purchased from a USDA approved
Swine Research Facility and housed in accordance with the guide
for the care and use of Laboratory Animals [15]. Animals involved
in this study were maintained in accordance with the ‘Guide for the
Care and Use of Laboratory Animals’ published by the National Re-
search Council/Institute of Laboratory Animal Research (ILAR). On
the day of surgery, the swine were anesthetized, intubated, and vas-
cular access was obtained to continuously monitor blood pressure in
the common carotid and pulmonary arteries, and a 9 Fr jugular ve-
nous catheter allowed for intravenous fluid and medication adminis-
tration. A midline laparotomy permitted placement of supra-pubic
bladder and 6 Fr inferior vena cava catheter and aortic cross clamp.
Swine were then divided into sham and traumatic shock groups.
The sham (or control) group underwent laparotomy, placement of
all monitoring lines with a total anesthesia time equal to that of the
experimental group, thereby making the only altered variables hem-
orrhage and cross clamp. Swine in the hemorrhagic shock group un-
derwent 35% blood volume hemorrhage (24.5 mL/kg) through an
inferior venacava catheterover 20 min.After the hemorrhagic phase,
a 50 min ischemic phase began with placement of a supraceliac aortic
cross clamp, followed by 6 h of intensive care resuscitation (Fig. 1A).
Resuscitation was done using normal saline and titration of vasopres-
sorstomaintainmeanarterialbloodpressuregreaterthan40mmHg.
Continuous arterial, pulmonary, and central venous pressures were
recorded,andbloodsamples forlaboratoryanalysisweredrawnevery
15 min during the crossclamp setup and application and every 30 min
during resuscitation with similar blood draws in the sham group as
well. Thus, all things being equal, we reasoned that differences in
gene expression were attributable to injury rather than elements of
the lengthy surgical procedure. Histology was performed by standard
methodology, and two pathologists independently evaluated each tis-
sue section for microscopic evidence of hypoxia/ischemia reperfusion
injury.
Isolation of Endothelial RNA and Porcine Genome Microarray
The aorta was dissected free prior to euthanasia and four atrau-
matic vascular cross clamps were applied, two above and two below
the site of the supraceliac aorta crossclamp before dividing the vessel
to harvest its endothelium. To ensure isolation of only aortic endothe-
lial cells, the opened aorta was cleansed of residual blood with washes
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
121

Page 3

of sterile ice-cold phosphate buffer, followed by gentle scraping of the
endothelium with a sterile scalpel and several washes with lysis
buffer from a High Pure RNA isolation kit (Roche Diagnostics
GmbH, Mannheim, Germany) [16]. Total RNA concentration and
quality was confirmed by Nanodrop ND-100 spectrophotometry
(Nanodrop Technologies, Inc, Wilmington, DE) and Agilent 2100 Ex-
pert SW Eukaryote Total RNA Nano Series II (Agilent Technologies,
Inc., Santa Clara, CA) before undergoing single round amplification
using a MessageAmp III RNA Amplification kit (Ambion, Austin,
TX). Finally, the amplified RNA (cRNA) was biotinylated, frag-
mented, and probe hybridized to an Affymetrix GeneChip (Affyme-
trix, Santa Clara, CA) (Fig. 1B).
Affymetric GeneChip Porcine Genome hybridization raw pixel
analysis was performed at a commercial facility (Benaroya Institute,
Seattle, WA). Raw pixel files were then analyzed using Affymetrix
Gene Chip Operating Software, (GCOS ver. 1.4, Affymetrix). Back-
ground signals normally range from 20 to 100 (pixel intensity, arbi-
trary units) for arrays when scanned with a GeneChip Scanner
3000 (Affymetrix), and our array backgrounds ranged from 34.67 to
38.59. Eukaryotic Hybridization Controls, comprising a known mix-
ture of biotin-labeled cRNA transcripts applied to an array, were
dose-dependent and ensured linearity of our array signals. As an in-
directassessment ofRNAhybridizationsignalstrength,wecompared
the sum of the pig GeneChip signals to the sum of the eukaryotic hy-
bridization control signals, which consistently showed that that our
RNA hybridization signals were robust (data not shown).
Statistical Analysis
Physiologic and laboratory parameters were compared using re-
peated measures correlation over the measured time intervals and
Student’s t-test was computed to analyze the differences in the final
measurements collected at resuscitation h 6 using PASW ver, 18.0
(PASW, Chicago, IL). Distribution of raw microarray data, coefficient
of variance, determination of thresholds, and additional t-tests were
performed using Excel 2007 (Microsoft Corporation, Seattle, WA).
Single and Batch GeneChip analysis was performed by GCOS. The
GCOSscalingstrategysoftwaresettingsforgenecomparisonanalysis
included a target signal value of 500, a scaling probe array type, and
a normalization of 1.0. gene detection (Present, Marginal, Absent)
was based on a computed discrimination score of 0.8 (the ratio for
each probe pair) and t value of 0.015. Gene detection P value was de-
termined by a one-sided Wilcoxon’s signed rank test, which was also
usedinthecomparisonanalysistodeterminebiologicalstatisticalsig-
nificance. Each probe set signal was then compared with the baseline
array signal and a Change P value was determined by GCOS. To in-
tuitively verify these statistics, genes that were determined to be sta-
tistically different by GCOS analysis (a 6 1.5-fold change in gene
expression) were visually confirmed by examining heat maps of raw
mean raw pixel intensities between corresponding experimental
and control swine (Fig. 1C). Lastly, lists including the up and down
regulated pig gene were annotated [17] to their corresponding human
counterparts using DAVID Bioinformatics Resources 6.7 NIAID,
National Institutes of Health, as previously described [13, 14].
RESULTS
Hemorrhagic Shock Model and Extent of Injury
Ten age, weight (control: 45.24 6 1.30 versus experi-
mental: 45.88 6 1.57 kg, P ¼ 0.76), and length-matched
(control: 116.4 6 1.69 versus experimental: 119 6 1.89
cm, P ¼ 0.33) male swine were used in this study.
Acidosis was demonstrated; pH was significantly de-
creased (Fig. 2A) in the injured group (trauma-hemor-
rhage/shock; T-H/S) compared with controls (6.97 6
0.03 versus 7.45 6 0.03, P ? 0.001) and lactate was sig-
nificantly increased (Fig. 2B) by 5.2 fold (12.7 6 0.7 ver-
sus 2.44 6 0.5 mmol/L, P < 0.001). In the experimental
group, coagulopathy was demonstrated by a greater
than two fold decrease in fibrinogen (Fig. 2C) and
a greater than two fold increase in international
normalized ratios (Fig. 2D) (2.5 6 0.9 versus 0.9 6
0.04, P ¼ 0.014). Hemodynamic monitoring showed
hemorrhagic shock, defined by profound tachycardia
FIG. 1.
icantly up and down-regulated genes present on microarray.
(A) Schematic of surgical procedure. (B) Schematic of tissue harvest and microarray techniques. (C) Algorithm for detecting signif-
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
122

Page 4

FIG. 2.
(solid line) and experimental (dashed line) pigs, respectively. (E), (F) Mean Cardiac output and total peripheral resistance in controls (black
bars) and experimental pigs (white bars) calculated as the mean across 6 h of resuscitative efforts, respectively. (G), (H) Total resuscitative
efforts to include the epinephrine and fluid requirements during in controls (black bars) and experimental pigs (white bars). Error bars ¼ stan-
dard error (n ¼ 5).
(A), (B), (C), (D) Physiologic measurements of pH, lactate, plasma fibrinogen, and international normalization ratios in control
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
123

Page 5

(experimental group: 163 6 6.4 versus control group:
86.266.2beatspermin,datanotshown)andincreased
mean cardiac output (Fig. 2E) throughout the 6-h re-
suscitation (7.8 6 1.1 versus 3.8 6 0.8 L/min, P ¼
0.042). In the experimental pigs, total peripheral resis-
tance (Fig. 2F, TPR equals mean arterial pressure
divided by cardiac output) was significantly decreased.
Vasopressor and intravenous fluid requirements were
dramatically increased in the T-H/S groups during the
resuscitative period (Fig. 2G, H) to maintain MAPs >
40 mm Hg. Hemorrhagic shock swine required 5.26-
fold more intravenous fluids than control swine (20 6
2.3 L versus 3.8 6 0.8 L, P < 0.001), although urine
output was dramatically decreased (410 6 121 mL
versus 1345 6 377 mL, P ¼ 0.046, data not shown).
Histologic Analysis
Microscopic evidence of hypoxic/ischemic tissue in-
jury was identified in the majority of cross-clamped an-
imals, with the most significant changes seen in liver,
including focal zone-three centrilobular necrosis and
scattered apoptotic cells indicating that reperfusion
injury occurred (Fig. 3A). Additionally, extensive
microvacuolar changes and sinusoidal dilation were
observed and occasional diffuse acute inflammatory
infiltrates could be found. More subtle liver findings
included hepatocyte pallor. All of these histologic
changes are consistent with other studies that examine
swine liver following hypoxic/ischemic times of 90 to
240 min [18–20]. A few animals showed renal changes
consistent with ischemia/hypoxia injury, to include
cast nephropathy, proximal tubular cell degeneration,
and focal acute inflammation (Fig. 3B). Consistent
with clinical observations that transplanted human
kidneys tolerate ischemic times up to 24 h, swine kid-
ney was more resistant to hypoxia/ischemia than liver
in that we did not see frank necrosis. Additional histo-
logic findings included pallor of pancreatic acinar cells
(not shown), tip necrosis of small bowel villi (Fig. 3C),
and focal loss of aortic endothelial cells (Fig. 3D).
Microarray Analysis
Validation was performed to ensure that technical
variation between chips was statistically insignificant.
Calculation of the mean pixel intensity was determined
for each quartile, ensuring uniformity between Gene-
Chips (data not shown). A total of 24,123 transcripts
were tested on the GeneChip, and 12,643 (52.4%) tran-
scripts were present in all five control swine and thus
were thefocus ofamore
(Fig. 1C). After normalization, a change algorithm
exhaustiveanalysis
FIG.3.
pigs. (A’) Asterisks show centrilobular necrosis in liver lobules in an experimental pig. (A) Normal liver in a control pig. (B) Kidney from an
experimental pig, black arrowheads showing detached apoptotic cells, granular casts, and vacuolar changes in proximal convoluted tubules.
(B’) Kidney from a control pig. (C) Ischemic changes of small bowel villi. Gray arrowheads show loss of goblet cells at the villus tips, whereas
black arrowheads show goblet cell preservation at the bases, where oxygen tension is generally higher. Note the dense acute inflammatory in-
filtrate lamina propria of villus tips. (D) Cross section of pig aorta showing damaged endothelium. Arrowheads show scattered remaining en-
dothelial cell nuclei. Inset shows low power magnification of full thickness aorta wall with relative preservation of the tunica media and
adventitia.
Histologicevidenceofischemia/reperfusioninjuryinliver,kidney,smallbowel,andaorticendotheliumincontrolandexperimental
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
124

Page 6

was used to determine whether transcripts were
decreased, moderately decreased, not changed, moder-
ately increased, or increased compared with the base-
line array. Among the control chips, 5868 transcripts
(46.4% of 12,643 present) demonstrated no change
within any of the control endothelium, and 9851
(77.9% of 12,643 present) transcripts were within one
standard deviation of no change values. No change
transcripts and those within one standard deviation
were then compared with their corresponding tran-
scripts on chips from experimental pigs, and genes
that underwent significant changes in expression in
the experimental group (as determined by GCOS
change algorithm to be increased or decreased) were
the focus of a more exhaustive analysis.
In the experimental group, 1536 gene transcripts
(529 up and 1007 down) underwent significant change
according to GCOS change algorithm criteria and
were selected for analysis by Cluster 3.0 (Lawrence
Berkeley National Laboratory, Berkley, CA) to be orga-
nized into a gene dendrogram using TreeView (Boston,
MA). However, before proceeding with our functional
cluster and dendrogram analysis, we qualitatively con-
firmed the strength of the GCOS statistical data by or-
ganizing genes selected by the change algorithm into
a visual heat map. As shown in Figure 4, the heat
map (red ¼ higher expression, and green ¼ lower ex-
pression) demonstrated significant differences in gene
expression between control and experimental groups,
whereas within-group differences were minimal. Sta-
tistically significant up and down regulated pig genes
were annotated [17] and matched to their correspond-
ing human ENSEMBL ID and queried with the DAVID
knowledge database for functional clustering analysis
[13, 14]. A total of 1297 corresponding human
ENSEMBL IDs were identified, and 46 had no known
human counterpart. On medium stringency (default
DAVID software setting), 351 clusters were identified,
of which only 33 clusters reached an enrichment score
of 1.5 or greater. Of these 33 clusters, Gene Ontology
(GO) and Kyoto Encyclopedia of Genes and Genomes
(KEGG) pathway members were consolidated into 12
common cellular biological processes (Fig. 4). Due to
the redundant nature of annotations, some of the iden-
tified genes were consolidated into several overlapping
biological processes and, thus, each functional cluster
was more rigorously filtered by the software into seven
non-overlapping cellular pathways. Inflammation and
adhesionpathwayscontainedthemostnumerousmem-
bers (123 genes), with actin cytoskeleton, motion, and
migration (100), phosphorylation (95), apoptosis (92),
and steroid hormone response (67) being present in de-
creasing frequency. Angiogenesis and blood vessel (32)
and cell junction (11) had fewer members but were
particularly relevant to endothelial tissue.
Among these genes, we identified several that were
highly related to multiple functional clusters (Table
1). Transforming growth factor beta-2 (TGFb2) was
the most inter-related gene, belonging to eight clusters.
TGFB1 and thrombospondin 2 (THBS1) belonged to
seven clusters. Monocyte chemoattractant protein-1
(CCL2, also known as MCP-1) belonged to six clusters.
Nonmetastatic cells 1 (NME1), fibroblast growth factor
(FGF2), and prostaglandin synthase G/H 2 (PTGS2)
belonged to five clusters. In addition, 13 genes were
identified in four clusters, and 36 genes overlapped
three clusters. Some of the notable members include
TGFBR3, integrin A6 (ITGA6), selectin E (SELE),
ITGB3, FLT1 (also known as VEGFR1), SOD2, BMP-
2, ITGAV, and IL8. Since the functional clusters in-
flammation, actin cytoskeleton, apoptosis, steroid hor-
mone, and angiogenesis are particularly important to
the discussion below, a complete list of genes that
are up-regulated and down-regulated is displayed in
Table 2.
DISCUSSION
Physiologic monitoring demonstrated that our swine
model accurately and reliably reproduces the patho-
physiology associated with severe trauma (see Fig. 2).
Well-known compensatory mechanisms are immedi-
ately triggered in response to HS, including barorecep-
tor signaling in the aortic arch and carotid bodies [21].
In response to these signals, powerful neuroendrocrine
reflexes increase cardiac output and peripheral resis-
tance and shunt blood from low priority tissues like
muscle and bowel to the brain and heart [22]. When
these reflexes are exhausted, decompensated shock
with paradoxically increased peripheral vasodilatation
occurs, probably due to unopposed actions of inducible
endothelial and tissue nitric oxide [22]. As seen in
critically-injured humans, compensatory changes in
our model included increased cardiac output. Further,
despite heavy use of epinephrine to maintain MAP,
eventually total peripheral resistance dropped, making
evident that our injured pigs reached decompensated
hemorrhagic shock.
Like critically-injured humans, our injured pigs
switchedtoanaerobicmetabolismanddevelopedsignif-
icant lactic acidosis (Fig. 2). Although vasopressor use
eventually caused some degree of lactic acidosis in our
control animals (data not shown), and these metabolic
derangements were more severe and occurred after
cross-clamp release in our experimental group (see
hour time points between 1.75 and 3 h in Fig. 2A–D),
suggesting that severe hemorrhagic and ischemia/re-
perfusion injury, and not spurious use of vasopressors,
caused the severe decompensatory changes seen in the
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
125

Page 7

experimental group. Decompensated shock was further
complicated by coaugulopathies associated with fibri-
nogenemia [22]. While vigorous fluid resuscitation
may cause some degree of dilutional fibrinogenemia,
consistentlyincreasedINRs
experimental pigs experienced similar disseminated
intravascular coagulopathies to those observed in
critically-injured patients. When combined with irre-
futable histologic evidence of hemorrhagic shock in
the liver, kidney, intestine, and endothelium (Fig. 3),
it is fairly evident that our pig model recapitulates the
anatomic and physiologic changes seen shock/trauma.
Since the endothelium is an important line of defense
against physiologic insult and essential in protecting
organs from metabolic toxins, we examined changes
inendothelialgeneexpression.WeusedthePorcineGe-
nome Microarray, determined the corresponding En-
sembl annotation, and utilized DAVID to elucidate
functional biological processes involved by the injury
(Fig. 1). In addition, we identified genes that were the
most highly interrelated among all of the functional
suggestthatour
clusters present (Table 1). DAVID functional clustering
analysis allowed us to determine that the TGFb
signaling pathway was related to the most (eight) func-
tional clusters (Table 1). This suggests that the TGFb
pathway is central to many of the changes in gene ex-
pression that occurred in our shock pig model (Table
1: TGFb 2 (–1.82), TGFb 1 (þ1.86), THBS1 (þ1.65),
TGFbR3 (–2.55), and BMP-2 (þ1.54).
TGFb1isexpressedinendothelialcells[23]andplays
an important role in extracellular matrix production,
regulation of cell growth, differentiation, migration,
and apoptosis [24]. Production of TGFb1 requires cleav-
age of a large precursor molecule (44 kDa) to form a ma-
ture 25 kDa molecule that binds to LAP to form a 100
kDa small latent complex [25]. After interaction with
TGFb1bindingproteins,a290kDalatentcomplexisse-
creted [26], which interacts with furin-like proprotein
convertases,plasmin,thrombospondin
þ1.65), urokinase-type plasminogen activator (MRC2,
þ2.58), and mannose-6-phosphate/insulin-like growth
factor II [27], particularly after arterial injury.
(THBS1,
FIG.4.
levels. Color gradient to the right of the heat map is in arbitrary units. Table (right) showing 12 functional clusters with enrichment scores ?
1.5,relatedcellularbiologicalprocesses,GO,andKEGGterms(seetheResultssectionforexplanation),uniquegenemembers,andPvaluesfor
each cluster analysis.
Heatmap(left)ofmicroarray datacorroboratingGCOSstatisticaldata.Redshowshigherandyellowshowslowergenesexpression
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
126

Page 8

TABLE 1
Gene Members Belonging to Multiple DAVID Functional Categories
Human
SymbolPorcine IDENSEMBL ID
Final Fold
ChangeInflammatory
Actin
Cytoskeleton Phosphorylation Apoptosis
Steroid
Hormone Endocytosis
Ribosome
Processes Translation Angiogenesis
Vitamin
Response Lysosome
Cell
Junction
TGFB2
TGFB1
THBS1
CCL2
NME1
PTGS2
GHR
FGF2
APP
TGFBR3
ASNS
ITGA6
MAP1B
KRAS
VAV3
SORT1
PTPRK
PRKCA
CITED2
CDK5R1
TNFRSF12A Ssc.1864.1.A1_at
SCINSsc.31102.1.A1_at
SELE Ssc.16297.1.S2_at
IGF1Ssc.25004.1.S1_at
ITGB3 Ssc.44.1.S1_at
SP100 Ssc.7207.3.A1_at
CTTNBP2Ssc.18707.1.A1_at
TRIP6Ssc.24386.1.S1_a_at ENSG00000087077
ACP5Ssc.575.1.S1_at
FLT1
Ssc.3872.1.S1_at
CXCL12Ssc.7243.1.A1_at
AATFSsc.5258.1.A1_at
GAB1Ssc.11005.1.A1_at
SOD2Ssc.3706.1.S1_at
BCL6Ssc.15518.1.A1_at
RTN4Ssc.13822.1.A1_at
LEPRSsc.816.1.S1_at
CTGFSsc.8562.2.S1_a_at
SPP1Ssc.101.1.S1_at
BMP2 Ssc.4190.1.S1_at
CDO1 Ssc.7106.1.S1_at
PPT1 Ssc.11457.1.A1_at
AP3B1Ssc.18606.1.A1_at
MYH10Ssc.20017.1.A1_at
ITGAVSsc.15932.1.S1_at
FGD4Ssc.26819.1.A1_at
VAV1Ssc.5910.1.A1_at
CYR61Ssc.9720.1.A1_at
PTPRFSsc.2337.1.S1_at
RHOBSsc.10226.1.A2_at
FOXO1Ssc.11251.1.A1_at
ITGA5 Ssc.16663.1.S1_at
TNFSF13Ssc.26879.2.S1_at
Ssc.10287.1.A1_at
Ssc.76.3.S1_a_at
Ssc.924.3.A1_at
Ssc.657.1.A1_at
Ssc.19546.1.S1_at
Ssc.23994.1.A1_at
Ssc.15583.1.S1_at
Ssc.17772.1.S1_at
Ssc.9950.1.A1_at
Ssc.1176.1.A1_at
Ssc.11609.1.A1_at
Ssc.25207.1.A1_at
Ssc.27009.1.A1_at
Ssc.29092.1.A1_at
Ssc.12682.1.A1_at
Ssc.12492.1.A1_at
Ssc.16444.1.S1_at
Ssc.13637.1.A1_at
Ssc.9718.1.A1_at
Ssc.21036.1.S1_at
ENSG00000092969
ENSG00000105329
ENSG00000137801
ENSG00000108691
ENSG00000011052
ENSG00000073756
ENSG00000112964
ENSG00000138685
ENSG00000142192
ENSG00000069702
ENSG00000070669
ENSG00000091409
ENSG00000131711
ENSG00000133703
ENSG00000134215
ENSG00000134243
ENSG00000152894
ENSG00000154229
ENSG00000164442
ENSG00000176749
ENSG00000006327
ENSG00000006747
ENSG00000007908
ENSG00000017427
ENSG00000056345
ENSG00000067066
ENSG00000077063
?1.82
1.86
1.65
3.03
1.51
1.69
1.66
1.75
?1.66
?2.55
1.72
?1.64
1.57
1.63
?1.95
?1.64
?1.98
2.00
?2.63
?1.81
1.66
?4.10
1.53
?3.33
2.30
?2.49
?1.74
1.51
?1.75
1.54
?2.00
1.65
?1.50
1.62
?1.53
1.54
?1.56
1.56
?1.95
1.54
?1.85
?1.52
1.63
?1.51
1.82
?2.37
?2.04
2.12
1.63
1.53
?1.67
1.52
?2.03
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXX
XX
XX
X
X
XXX
XX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XXX
X
X
X
X
X
X
X
X
X
X
XX
X
X
X
X
X
X
XX
XX
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
X
XX
X
X
X
X
X
X
X
X
X
X
XX
ENSG00000102575
ENSG00000102755
ENSG00000107562
ENSG00000108270
ENSG00000109458
ENSG00000112096
ENSG00000113916
ENSG00000115310
ENSG00000116678
ENSG00000118523
ENSG00000118785
ENSG00000125845
ENSG00000129596
ENSG00000131238
ENSG00000132842
ENSG00000133026
ENSG00000138448
ENSG00000139132
ENSG00000141968
ENSG00000142871
ENSG00000142949
ENSG00000143878
ENSG00000150907
ENSG00000161638
ENSG00000161955
X
X
XX
X
X
X
XX
XXX
XXX
X
X
X
X
X
X
X
XX
X
X
XX
X
X
X
X
X
XX
XX
XX
XX
X
X
X
X
XX
X
X
X
X
XXX
XX
X
X
X
X
XX
XX
XX
XXX
XX
X
X
XX
(Continued)
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
127

Page 9

After activation and cofactor binding, TGFb1 signaling
occurs through a complex cascade of events mediated
by the SMAD (SMAD6, þ1.56) signaling pathway,
which itself can regulate TGFb1 gene expression [28].
Furthermore, TGFb1 increases DNA synthesis when
other potent mitogens are present, such as fibroblastic
growth factor (FGF-2, þ1.75), epidermal growth factor
(EGF), connective tissue growth factor (CTGF, þ1.56)
[29], and vascular endothelial growth factor (VEGF,
þ1.37*see below). The large number of cell signaling
processes linked to the TGFb signaling pathway sug-
gestsacentral roleduring endothelialdamage,perhaps
contributing to increased vascular permeability.
Indeed, during acute lung injury, TGFb1 increases
lung capillary permeability, resulting in pulmonary
edema and vascular remodeling, processes that are
likely mediated through the family of integrin mole-
cules expressed on endothelial cells [30, 31]. Increased
vascular permeability could start aviscous cycle by eas-
ing the passage of inflammatory cytokines and immune
cells into an already inflamed vesselwall and end organ
[32]. In vitro studies have shown that TGFb1 induces
alterations in endothelial cell morphology within 24 h
and reduces vascular endothelial monolayer barrier in-
tegrity as early as 3 h, changes that might contribute to
third-spacing of intravenously administered fluids
during resuscitation and loss of endothelial cellular in-
tegrity [33]. These studies, in combination with our in
vivo swine data, lead us to hypothesize that inhibition
of the TGFb signaling pathway may attenuate vascular
permeability and prevent organ damaging fluid-
overload by reducing third-spacing of intravenous
fluids needed to maintain tissue perfusion. By pharma-
cologically targeting the early intracellular pathways,
downstream activation and inhibition of adverse
cellular transcript may be potentiated.
In addition to increased vascular permeability, cellu-
lar hypoxia profoundly decreases energy dependent-
transcription and protein synthesis. Not surprisingly,
1007 genes demonstrated a significant decrease in ex-
pression, whereas only 529 genes were increased. To
overcome deficits in energy-dependent transcription,
hypoxia-inducible factor (HIF) coordinates changes in
gene transcription. HIF is a highly conserved protein
that is post-translationally regulated (i.e., destruction
of the a subunits) in response to hypoxic conditions.
RNA interference (knockdown) studies have shown
that decreased EGLN1 (–1.74) activity is sufficient to
activate HIF [34]. CITED2 (–2.63) encodes for a protein
that inhibits transactivation of HIF1A-induced genes
by competing with binding of HIF1a to p300-CH1.
More than a hundred genes appear to be transcrip-
tional targets of HIF, many of which are mitogens
that regulate erythropoiesis, angiogenesis, and metab-
olism. For example, HIF induces expression of VEGFs,
TABLE 1
(Continued)
Human
Symbol
Porcine ID
ENSEMBL ID
Final Fold
Change
Inflammatory
Actin
Cytoskeleton Phosphorylation Apoptosis
Steroid
Hormone Endocytosis
Ribosome
Processes Translation Angiogenesis
Vitamin
Response Lysosome
Cell
Junction
YWHAZ
Ssc.9681.1.A1_at
ENSG00000164924
1.69
X
X
X
PTK2
Ssc.6996.1.A1_at
ENSG00000169398
?1.76
X
X
X
IL8
Ssc.658.1.S1_at
ENSG00000169429
2.88
X
X
X
ASL
Ssc.20648.1.S1_at
ENSG00000169910
?1.59
X
X
X
PTPRM
Ssc.27106.1.A1_at
ENSG00000173482
?1.53
X
X
X
FOSL1
Ssc.3391.1.S1_at
ENSG00000175592
2.07
X
X
X
RYR2
Ssc.14352.1.S1_at
ENSG00000198626
?2.55
X
X
X
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
128

Page 10

TABLE 2
Functional Cluster Members of Inflammation, Actin Cytoskeleton, Apoptosis, Steroid Hormone, and Angiogenesis
Inflammation Actin cytoskeleton ApoptosisSteroid hormone Angiogenesis
CCL2 (3.03)
IL8 (2.88)
ITGB3 (2.30)
CYR61 (2.12)
MGLL (1.93)
LAMB1 (1.88)
S100A9 (1.88)
CD93 (1.88)
TGFB1 (1.86)
ITGAV (1.82)
OLR1 (1.77)
PXK (1.76)
FGF2 (1.75)
S100A8 (1.72)
CDK6 (1.70)
YWHAZ (1.69)
TNFRSF12A (1.66)
THBS1 (1.65)
AATF (1.65)
PAPSS2 (1.63)
AP3B1 (1.63)
PTPRF (1.63)
SOD2 (1.62)
CXCL6 (1.58)
MAP1B (1.57)
CTGF (1.56)
AMIGO2 (1.55)
LMAN1 (1.55)
BMP2 (1.54)
ARHGAP5 (1.53)
RHOB (1.53)
SELE (1.53)
ITGA5 (1.52)
JMJD3 (1.51)
TRIP6 (1.51)
NME1 (1.51)
FERMT1 (1.51)
PLAA (1.51)
P2RY12 (-5.07)
BLNK (-4.99)
VSIG4 (-3.99)
IGF1 (-3.33)
CCR5 (-3.18)
COL11A1 (-2.79)
LYZ (-2.78)
IL18 (-2.78)
CCR1 (-2.71)
CD163 (-2.67)
C1QC (-2.63)
CITED2 (-2.63)
TLR9 (-2.60)
ITGB2 (-2.52)
C3 (-2.52)
AMBN (-2.47)
PLXNC1 (-2.47)
C1QB (-2.46)
CADM1 (-2.45)
CYBB (-2.40)
AMICA1 (-2.40)
CNTN3 (-2.34)
CTNNAL1 (-2.33)
PGM5 (-2.30)
PCDHB16 (-2.23)
C1QA (-2.16)
ADAM12 (-2.12)
PKD2 (-2.08)
NFRKB (-2.07)
VAV1 (-2.04)
PARP4 (-2.02)
CXCL12 (-2.00)
PTPRK (-1.98)
ITGA8 (-1.96)
ITIH4 (-1.96)
SPP1 (-1.95)
VAV3 (-1.95)
GPNMB (-1.92)
IL15 (-1.90)
LAMA2 (-1.89)
THRA (-1.88)
CDO1 (-1.85)
CFD (-1.85)
NCAM1 (-1.83)
TGFB2 (-1.82)
LY96 (-1.82)
CDK5R1 (-1.81)
PCDH15 (-1.79)
GRHL3 (-1.78)
PPFIA2 (-1.76)
CNTN1 (-1.73)
SERPINC1 (-1.73)
LAMA1 (-1.73)
MUC4 (-1.70)
CCL2 (3.03)
IL8 (2.88)
ITGB3 (2.30)
SPAG9 (2.01)
PRKCA (2.00)
DBN1 (1.98)
LAMB1 (1.88)
S100A9 (1.88)
TGFB1 (1.86)
VEGFC (1.82)
MACF1 (1.75)
FGF2 (1.75)
PTGS2 (1.69)
SERPINE2 (1.66)
TNFRSF12A (1.66)
THBS1 (1.65)
KRAS (1.63)
PLXNA2 (1.61)
C14orf104 (1.59)
KLHL1 (1.58)
MAP1B (1.57)
CTGF (1.56)
B3GNT1 (1.55)
RHOU (1.55)
FLT1 (1.54)
PRR5 (1.54)
RTN4 (1.54)
VASP (1.54)
ARHGAP5 (1.53)
SELE (1.53)
ITGA5 (1.52)
HMGCR (1.52)
TRIP6 (1.51)
YWHAE (1.51)
RALA (1.51)
MYO1E (1.50)
RANBP1 (1.50)
EPB49 (1.50)
CLIC4 (1.50)
SCIN (-4.10)
IGF1 (-3.33)
EPB41L3 (-2.96)
LCP1 (-2.78)
CITED2 (-2.63)
TGFBR3 (-2.55)
ANK3 (-2.52)
ITGB2 (-2.52)
SP100 (-2.49)
SLIT2 (-2.41)
FGD4 (-2.37)
DMD (-2.27)
0 (-2.20)
LIMCH1 (-2.20)
MYCBP2 (-2.14)
GAS1 (-2.12)
WASF3 (-2.10)
EPHA4 (-2.04)
EFNA5 (-2.04)
TNFSF13 (-2.03)
CXCL12 (-2.00)
PTPRK (-1.98)
VAV3 (-1.95)
TNS3 (-1.94)
PRC1 (-1.91)
MYLIP (-1.90)
LAMA2 (-1.89)
TGFB2 (-1.82)
KIF5C (-1.82)
CDK5R1 (-1.81)
SSH2 (-1.77)
ABLIM1 (-1.76)
PTK2 (-1.76)
CTTNBP2 (-1.74)
LAMA1 (-1.73)
PCNT (-1.72)
BBS2 (-1.72)
ERMN (-1.71)
ANTXR1 (-1.70)
SGCD (-1.70)
DAAM1 (-1.68)
APP (-1.66)
PDLIM3 (-1.65)
FUT8 (-1.65)
RACGAP1 (-1.65)
ITGA6 (-1.64)
MAP2 (-1.63)
TMOD1 (-1.60)
CNN3 (-1.57)
CEP250 (-1.56)
ARHGDIB (-1.55)
EPB41L2 (-1.53)
PTPRM (-1.53)
BCL6 (-1.53)
CCL2 (3.03)
PIM2 (2.39)
ANGPTL4 (2.15)
FOSL1 (2.07)
PRKCA (2.00)
CYCS (1.88)
TGFB1 (1.86)
MYC (1.80)
IER3 (1.80)
EEF1E1 (1.78)
FGF2 (1.75)
ASNS (1.72)
PTGS2 (1.69)
YWHAZ (1.69)
GHR (1.66)
THBS1 (1.65)
AATF (1.65)
PSME3 (1.65)
EIF5A (1.65)
KRAS (1.63)
PTPRF (1.63)
PLCG2 (1.63)
SOD2 (1.62)
PHLDA1 (1.62)
INHBA (1.60)
SMAD6 (1.57)
AMIGO2 (1.55)
RTN4 (1.54)
GRIK2 (1.53)
HSPA9 (1.51)
NME1 (1.51)
BAG1 (1.51)
YWHAE (1.51)
POLB (1.51)
BID (1.51)
SCIN (-4.10)
IGF1 (-3.33)
CITED2 (-2.63)
PDE3A (-2.61)
RYR2 (-2.55)
IFI16 (-2.54)
CADM1 (-2.45)
TP53INP1 (-2.39)
FGD4 (-2.37)
CHD8 (-2.27)
TNFAIP8 (-2.23)
XRCC2 (-2.21)
PRODH (-2.19)
GAS1 (-2.12)
DAPK1 (-2.11)
MOAP1 (-2.09)
VAV1 (-2.04)
TNFSF13 (-2.03)
BIRC6 (-2.02)
TIAM2 (-2.00)
VAV3 (-1.95)
PIK3CG (-1.90)
MITF (-1.89)
SIRT1 (-1.84)
TGFB2 (-1.82)
CDK5R1 (-1.81)
CAT (-1.76)
APAF1 (-1.71)
MAP3K5 (-1.70)
SGMS1 (-1.69)
MCF2L (-1.68)
FOXO1 (-1.67)
ING3 (-1.66)
APP (-1.66)
SORT1 (-1.64)
NME6 (-1.64)
PECR (-1.62)
SLC25A4 (-1.61)
AIFM2 (-1.61)
BNIP3L (-1.60)
TRIO (-1.58)
TP53I3 (-1.58)
MAPK9 (-1.58)
PROC (-1.55)
SOS2 (-1.55)
TRIM35 (-1.55)
IFI6 (-1.55)
KCNH8 (-1.53)
BCL6 (-1.53)
BTG2 (-1.52)
PPT1 (-1.52)
PRDX5 (-1.51)
NGFRAP1 (-1.51)
HDAC1 (-1.50)
CCL2 (3.03)
GSTM3 (2.68)
CYR61 (2.12)
FOSL1 (2.07)
PRKCA (2.00)
EIF4EBP1 (1.96)
TGFB1 (1.86)
MYC (1.80)
CYP1B1 (1.73)
LDLR (1.73)
RCAN1 (1.72)
ASNS (1.72)
PTGS2 (1.69)
WFDC1 (1.69)
GHR (1.66)
THBS1 (1.65)
DNAJC25 (1.65)
TOR1B (1.64)
KRAS (1.63)
PLCG2 (1.63)
RARA (1.62)
ADFP (1.61)
EIF2C2 (1.59)
MAP1B (1.57)
ZFP106 (1.57)
ABCB4 (1.55)
CTSC (1.53)
SELE (1.53)
TJP2 (1.52)
NME1 (1.51)
NR4A3 (1.51)
PDE3A (-2.61)
RYR2 (-2.55)
TGFBR3 (-2.55)
SP100 (-2.49)
C1QB (-2.46)
ERO1L (-2.42)
MGST1 (-2.21)
BCKDHB (-2.21)
NDUFS4 (-2.11)
ALDH1A2 (-2.09)
SPP1 (-1.95)
KAT2B (-1.89)
THRA (-1.88)
ABCC5 (-1.86)
CDO1 (-1.85)
ASAH1 (-1.83)
TGFB2 (-1.82)
LY96 (-1.82)
PTPN2 (-1.80)
ACP5 (-1.75)
CTTNBP2 (-1.74)
FOXO1 (-1.67)
PTCH1 (-1.65)
SORT1 (-1.64)
PRKACB (-1.62)
PHIP (-1.61)
ASL (-1.59)
IDH1 (-1.58)
PRKAR2B (-1.57)
LEPR (-1.56)
BTG2 (-1.52)
GAB1 (-1.50)
TTC3 (-1.50)
ZCCHC11 (-1.50)
GLUL (-1.50)
DNAJB2 (-1.50)
IL8 (2.88)
ANGPTL4 (2.15)
CYR61 (2.12)
GJC1 (2.01)
NCL (1.86)
ITGAV (1.82)
VEGFC (1.82)
FGF2 (1.75)
TNFRSF12A (1.66)
THBS1 (1.65)
NARG1 (1.60)
MIB1 (1.57)
CTGF (1.56)
ZMIZ1 (1.55)
FLT1 (1.54)
RTN4 (1.54)
RHOB (1.53)
MYO1E (1.50)
FOXF1 (-2.86)
IL18 (-2.78)
CITED2 (-2.63)
TGFBR3 (-2.55)
SLIT2 (-2.41)
TNFSF13 (-2.03)
CXCL12 (-2.00)
MKL2 (-1.86)
TGFB2 (-1.82)
PTK2 (-1.76)
FIGF (-1.75)
EFNB2 (-1.75)
FOXO1 (-1.67)
LEPR (-1.56)
(Continued)
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
129

Page 11

which are critical regulators of wound healing,
angiogenesis, and vessel homeostasis by inhibiting en-
dothelial cell apoptosis. The mitogen VEGF (first iden-
tified as vascular permeability factor, VPF) is clinically
linked to angiogenesis during tumorigenesis [35]. Dur-
ing angiogenesis, VEGF works in concert with other
growth factors, such as basic FGF, TGFb1, platelet-
derived growth factors (PDGFC, –1.88), and angiopoie-
tins (ANGPTL4, þ2.15; CYR61 þ 2.12) [36, 37]. During
hypoxia, however, other growth factors control VEGF
expression (2); both FGF-2 and TGFb1 induce endothe-
lial cell VEGF expression, which in turn potentiates the
angiogenicactivity ofFGF-2invitro and invivo.Insup-
port of the idea that the VEGF signaling cascade is ac-
tive during ischemia/hypoxia, we showed a moderate
increase in VEGFR1 during trauma (also known as
FLT1, þ1.54), and although the fold increase for the li-
gand to VEGFR1 did not reach our significance thresh-
old of 1.5 (the ligand VEGF, þ1.37), in light of the
studies mentioned above we believe VEGF signaling
through cognate receptors like VEGFR1 is a sentinel
molecular event during trauma and tissue reperfusion.
The results of DAVID analysis indicate that in-
creased expression of certain TGFb members occurs
as a pathologic response to massive bleeding, and may
lead to the hemodynamic instability, coagulopathy, de-
creased oxygen delivery, decreased tissue perfusion,
and cellular hypoxia. Concomitant transcriptional acti-
vation of the HIF pathways may recruit VEGF signal-
ing cascades that are critical to mitigating the onset of
endothelial dysfunction. The results of this study, as
with other microarray approaches, are intended for
hypothesis-generation rather than testing causality.
Through the functional analysis ofthese highly interre-
lated cellular signaling pathways, potential mitigation
of endothelial barrier disruption during trauma man-
agement is indeed plausible. Indeed, it is well known
that maintenance of vascular integrity is essential in
trauma management, and minimizing endothelial
damage at the genetic and molecular level through tar-
geted therapy will become central to the proper man-
agement of trauma patients in the not too distant
future [38]. As evidenced by this study, webelieve those
targetsinclude,butarenotlimitedto,TGFbandVEGF.
Finally, we show that it is possible, using DAVID anal-
ysis, to identify a handful of active signaling pathways
critical to the induction of endothelial dysfunction
among thousands present on an array,which facilitates
the identification of novel pharmaceutical targets that
might be useful in managing trauma patients. It is im-
portant to note, however, that it is unknown whether
the TGFb and VEGF pathways play important roles
in human models of hemorrhage shock. Indeed, future
experiments are necessary to test whether modulation
TABLE 2
(Continued)
Inflammation
Actin cytoskeleton
Apoptosis
Steroid hormone
Angiogenesis
EDIL3 (-1.59)
ANTXR1 (-1.70)
SLIT3 (-1.52)
PCDH7 (-1.58)
CCDC80 (-1.70)
CETN3 (-1.51)
CD47 (-1.56)
SGMS1 (-1.69)
MYH10 (-1.51)
PROC (-1.55)
IL1RAP (-1.66)
GAB1 (-1.50)
MDK (-1.55)
APP (-1.66)
DLG1 (-1.50)
ARHGDIB (-1.55)
PCDHGC5 (-1.65)
KIFAP3 (-1.54)
CXCL9 (-1.64)
ADORA3 (-1.54)
ITGA6 (-1.64)
PTPRM (-1.53)
SDK1 (-1.63)
BCL6 (-1.53)
BOC (-1.62)
ITGAE (-1.51)
LRRN2 (-1.62)
PRDX5 (-1.51)
MAGI1 (-1.61)
MYH10 (-1.51)
F5 (-1.59)
DLG1 (-1.50)
C8B (-1.59)
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
130

Page 12

of these pathways and their members mitigate end or-
gan damage and/or failure due to hemorrhage shock.
Major trauma leads to widespread physiologic
changes that include endothelial dysfunction. Endothe-
lial dysfunction leads to loss of vascular integrity and
contributes to end organ damage. Further complicating
ourunderstandingofendothelialdysfunctionareamyr-
iad of circulating cytokines, metabolic byproducts, and
recently discovered genes that are turned on and off
during hypoxia/ischemia. Our microarray data suggest
a critical role for TGFb and, perhaps, VEGF pathways
in endothelial dysfunction and identifies other mem-
bers of their functional clusters that may contribute
to impaired vascular integrity. Though it is likely that
clinically relevant organ dysfunction occurs through
several different pathways, pharmaceuticals discov-
ered through functional analysis of microarray data
may minimize endothelial cell damage in trauma
patients when reperfusion recirculates high concentra-
tions of metabolic byproducts.
ACKNOWLEDGMENTS
Theauthorsthankthe LaboratoryAnimalResearchServicestaffat
Madigan Army Medical Center, Major Patrick J. Canchola, DVM,
Brian Theis, John Schaphorst, Juan Tercero, David Montano, Karen
Gibeaut, and Shelly Spahn-Bridges for their incredible work with the
surgical model. M. J. Dehart provided significant technical support in
preparingtissueforanalysis.TheauthorsalsothankDr.JamesHem-
pel for helping evaluate the histology in this study. The views ex-
pressed are those of the authors and do not reflect the official policy
of the Department of the Army, the Department of Defense, or the
U.S. Government.
REFERENCES
1. AngeleMK,SchneiderCP, Chaudry IH. Bench-to-bedside review:
Latest results in hemorrhagic shock. Crit Care 2008;12:218.
Rossaint R, Cerny V, Coats TJ, et al. Key issues in advanced
bleeding care in trauma. Shock 2006;26:322.
Jarrar D, Chaudry IH, Wang P. Organ dysfunction following
hemorrhageand sepsis: Mechanisms
approaches (Review). Int J Mol Med 1999;4:575.
Moore FA, McKinley BA, Moore EE, et al, the Inflammation and
the Host Response to Injury Collaborative Research. Inflamma-
tion and the Host Response to Injury, a large-scale collaborative
project: Patient-oriented research core–standard operating
procedures for clinical care. III. Guidelines for shock resuscita-
tion. J Trauma 2006;61:82.
Spahn DR, Cerny V, Coats TJ, et al. Management of bleeding
following major trauma: A European guideline. Crit Care
2007;11:R17.
Cotton BA, Guy JS, Morris JA Jr, et al. The cellular, metabolic,
and systemic consequences of aggressive fluid resuscitation
strategies. Shock 2006;26:115.
Ayala A, Ertel W, Chaudry IH. Trauma-induced suppression of
antigen presentation and expression of major histocompatibility
class II antigen complex in leukocytes. Shock 1996;5:79.
Chaudry IH, Zellweger R, Ayala A. The role of bacterial translo-
cation on Kupffer cell immune function following hemorrhage.
Prog Clin Biol Res 1995;392:209.
2.
3.
andtherapeutic
4.
5.
6.
7.
8.
9.HarbrechtBG, Billiar TR.The role of nitric oxide in Kupffer cell-
hepatocyte interactions. Shock 1995;3:79.
Hierholzer C, Billiar TR. Molecular mechanisms in the early
phase of hemorrhagic shock. Langenbecks Arch Surg 2001;
386:302.
Ba ZF, Wang P, Koo DJ, et al. Attenuation of vascular endothe-
lial dysfunction by testosterone receptor blockade after trauma
and hemorrhagic shock. Arch Surg 2001;136:1158.
Wang P, Ba ZF, Chaudry IH. Endothelial cell dysfunction occurs
very early following trauma-hemorrhage and persists despite
fluid resuscitation. Am J Physiol 1993;265:H973.
Dennis G Jr, Sherman BT, Hosack DA, et al. DAVID: Database
for Annotation, Visualization, and Integrated Discovery. Ge-
nome Biol 2003;4:P3.
Huang da W, Sherman BT, Lempicki RA. Systematic and inte-
grative analysis of large gene lists using DAVID bioinformatics
resources. Nat Protoc 2009;4:44.
Clark JD, Baldwin RL, Bayne KA, et al. Guide for the care and
use of laboratory animals. Washington, D.C.: National Academy
Press, 1996.
Passerini AG, Polacek DC, Shi C, et al. Coexisting proinflamma-
tory and antioxidative endothelial transcription profiles in a dis-
turbed flow region of the adult porcine aorta. Proc Natl Acad Sci
USA 2004;101:2482.
Tsai S, Cassady JP, Freking BA, et al. Annotation of the affyme-
trix porcine genome microarray. Anim Genet 2006;37:423.
Brockmann JG, August C, Wolters HH, et al. Sequence of reper-
fusion influences ischemia/reperfusion injury and primary graft
function following porcine liver transplantation. Liver Transpl
2005;11:1214.
MonbaliuD,LibbrechtL,DeVosR,etal.Theextentofvacuolation
innon-heart-beating porcine donor liver grafts prior to transplan-
tation predicts their viability. Liver Transpl 2008;14:1256.
van Wagensveld BA, van Gulik TM, Gelderblom HC, et al. Pro-
longed continuous or intermittent vascular inflow occlusion dur-
ing hemihepatectomy in pigs. Ann Surg 1999;229:376.
Ursino M, Magosso E. Short-term autonomic control of cardio-
vascular function: A mini-review with the help of mathematical
models. J Integr Neurosci 2003;2:219.
Peitzman AB, Billiar TR, Harbrecht BG, et al. Hemorrhagic
shock. Curr Probl Surg 1995;32:925.
Ghosh J, Murphy MO, Turner N, et al. The role of transforming
growthfactorb1inthevascularsystem.CardiovascPathol2005;
14:28.
Massague J, Gomis RR. The logic of TGF-b signaling. FEBS Lett
2006;580:2811.
Lawrence DA. Latent-TGF-b: An overview. Mol Cell Biochem
2001;219:163.
Maeda S, Dean DD, Gomez R, et al. The first stage of transform-
ing growth factor b1 activation is release of the large latent com-
plex from the extracellular matrix of growth plate chondrocytes
by matrixvesicle stromelysin-1 (MMP-3).CalcifTissue Int 2002;
70:54.
Bobik A. Transforming growth factor-bs and vascular disorders.
Arterioscler Thromb Vasc Biol 2006;26:1712.
Agrotis A, Kalinina N, Bobik A. Transforming growth factor-b,
cell signaling and cardiovascular disorders. Curr Vasc Pharma-
col 2005;3:55.
Hua X, Liu X, Ansari DO, et al. Synergistic cooperation of TFE3
and smad proteins in TGF-b-induced transcription of the plas-
minogen activator inhibitor-1 gene. Genes Dev 1998;12:3084.
Budinger GR, Chandel NS, Donnelly HK, et al. Active trans-
forming growth factor-b1 activates the procollagen I promoter
in patients with acute lung injury. Intensive Care Med 2005;
31:121.
Pittet JF, Griffiths MJ, Geiser T, et al. TGF-b is a critical medi-
ator of acute lung injury. J Clin Invest 2001;107:1537.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
31.
CAUSEY ET AL.: TGFb MEMBERS ALTERED IN HEMORRHAGE SHOCK
131

Page 13

32. Budhiraja R, Tuder RM, Hassoun PM. Endothelial dysfunction
in pulmonary hypertension. Circulation 2004;109:159.
HurstVI, GoldbergPL, MinnearFL, et al.Rearrangement ofad-
herens junctions by transforming growth factor-b1: Role of con-
traction. Am J Physiol 1999;276:L582.
Berra E, Benizri E, Ginouves A, et al. HIF prolyl-hydroxylase 2
isthekeyoxygensensorsettinglowsteady-statelevelsofHIF-1a
in normoxia. EMBO J 2003;22:4082.
Cebe-Suarez S, Zehnder-Fjallman A, Ballmer-Hofer K. The role
of VEGF receptors in angiogenesis; complex partnerships. Cell
Mol Life Sci. 2006;63:601.
33.
34.
35.
36. Pepper MS. Transforming growth factor-b: Vasculogenesis, an-
giogenesis, and vessel wall integrity. Cytokine Growth Factor
Rev 1997;8:21.
Benjamin LE, Hemo I, Keshet E. A plasticity window for blood
vessel remodeling is defined by pericyte coverage of the
preformed endothelial network and is regulated by PDGF-B
and VEGF. Development 1998;125:1591.
van Bruggen N, Thibodeaux H, Palmer JT, et al. VEGF antago-
nism reduces edema formation and tissue damage after ische-
mia/reperfusion injury in the mouse brain. J Clin Invest 1999;
104:1613.
37.
38.
JOURNAL OF SURGICAL RESEARCH: VOL. 170, NO. 1, SEPTEMBER 2011
132