Fresh Frozen Plasma Modulates Brain Gene Expression in a Swine Model of Traumatic Brain Injury and Shock: A Network Analysis

Accepted Manuscript Fresh Frozen Plasma Modulates Brain Gene Expression in a Swine Model of Traumatic Brain Injury and Shock: A Network Analysis Martin Sillesen, MD, Phd, Theodore Bambakidis, Msc, Simone Dekker, MD, Yongqing Li, MD, PhD, Hasan B. Alam, MD, FACS PII:

S1072-7515(16)31515-0

DOI:

10.1016/j.jamcollsurg.2016.09.015

Reference:

ACS 8496

To appear in:

Journal of the American College of Surgeons

Received Date: 20 August 2016 Revised Date:

7 September 2016

Accepted Date: 20 September 2016

Please cite this article as: Sillesen M, Bambakidis T, Dekker S, Li Y, Alam HB, Fresh Frozen Plasma Modulates Brain Gene Expression in a Swine Model of Traumatic Brain Injury and Shock: A Network Analysis, Journal of the American College of Surgeons (2016), doi: 10.1016/j.jamcollsurg.2016.09.015. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT 1

Fresh Frozen Plasma Modulates Brain Gene Expression in a Swine Model of Traumatic Brain Injury and Shock: A Network Analysis

RI PT

Martin Sillesen, MD, Phd1,2, Theodore Bambakidis, Msc3, Simone Dekker, MD3, Yongqing Li, MD, PhD,3 Hasan B Alam, MD, FACS3

SC

1) Department of Surgical Gastroenterology, Copenhagen University Hospital, Rigshospitalet, Denmark

M AN U

2) Institute for Inflammation Research, Copenhagen University Hospital, Rigshospitalet, Denmark

3) Department of Surgery, University of Michigan, Ann Arbor, MI

TE D

Disclosure Information: Nothing to disclose.

Abstract presented at the American College of Surgeons 101st Annual Clinical Congress, Scientific Forum, Chicago, IL, October 2015.

EP

Support: The study was funded by a grant from the US Army Medical Research Material

AC C

Command (T00521959), to Dr Alam.

Correspondence address: Hasan B. Alam, MD Norman Thompson Professor of Surgery Head of General Surgery Section, University of Michigan Health System 2920 Taubman Center/5331, 1500 E. Medical Center Drive Ann Arbor, MI 48109-5331 [email protected]

Short title: Fresh Frozen Plasma Effect on Gene Expression

ACCEPTED MANUSCRIPT 2

ABSTRACT BACKGROUND: Resuscitation with Fresh Frozen Plasma (FFP) decreases brain lesion size and swelling in a swine model of Traumatic Brain Injury (TBI) and hemorrhagic shock (HS). We

RI PT

hypothesized that brain gene expression profiles following TBI+HS would be modulated by FFP resuscitation.

STUDY DESIGN: 15 swine underwent a protocol of TBI and HS, two hours of shock followed

SC

by resuscitation with FFP, Normal Saline (NS) or hetastarch (Hextend, HEX, 5/group). After 6 hours, brain RNA was isolated and hybridized onto a Porcine Gene ST 1.1 microarray. Weighted

M AN U

Gene Correlation Network Analysis (WGCNA) was used to identify clusters of highly coexpressed genes. Principal Component Analysis identified cluster eigenvectors, indicating overall direction and magnitude of cluster gene expression. Using linear regression, cluster eigenvectors were associated with treatment as well as brain lesion size and swelling. Results

TE D

were post-hoc corrected using False Detection Rate (FDR). Relevant gene clusters were subjected to pathway analysis using the Reactome tool. RESULTS: Network analysis identified 322 gene expression clusters (total of 12.462 co-

EP

expressed genes). FFP resuscitation (but not NS or HEX) was positively associated with two distinct gene clusters (termed A and B), comprising 493 genes. Gene expression in both clusters

AC C

was negatively associated with brain swelling, and cluster B was also negatively associated with lesion size. Pathway analysis revealed an upregulation of genes involved in metabolic and platelet signaling as well as collagen formation, and down regulation of inflammation. CONCLUSIONS: FFP resuscitation in this model was associated with downregulation of inflammatory pathway genes, and expression of gene clusters mapping to increased metabolic and platelet signaling, which in turn was reversely associated with brain swelling.

ACCEPTED MANUSCRIPT 3

INTRODUCTION Trauma, including Traumatic Brain Injury (TBI) and hemorrhagic shock (HS) is well known to induce clinically relevant perturbations in gene expression patterns of multiple cell lines (1-3).

RI PT

Preclinical studies have indicated that these patterns may be further modulated by the choice of resuscitation strategy, thus emphasizing the important clinical role of gene expression following injury (4). While previous studies have focused on the detrimental effects of crystalloid

SC

resuscitation on both outcomes (5) and gene expression (4) following trauma, little is known of the effects of blood products on gene transcription following trauma, TBI and HS. Balanced

M AN U

resuscitation strategies employing high ratios of Fresh Frozen Plasma (FFP) to packed red blood cells (PRBC) forms for the mainstay of current trauma resuscitation paradigms, with multiple studies indicating a survival benefit of this regimen in trauma patients with critical bleeding (68). Although less well studied, this effect may extend to patients with traumatic brain injury

TE D

(TBI) (9) as suggested by preclinical studies (10, 11).

While the effect of FFP on coagulation derangements following trauma is usually perceived as the main mechanism of action, multiple preclinical studies have suggested that FFP exerts its

EP

protective effects through multiple alternate pathways. Studies have thus suggested a protective effect on the endothelium (12), platelets (13), cellular damage (14) as well as metabolic

AC C

derangements (15). With such a plethora of different effects attributable to FFP resuscitation, it is highly likely that this resuscitation strategy modulates cellular regulation relatively upstream, such as gene transcription.

Whether the effect of FFP resuscitation can be detected at the level of the gene transcriptome is unknown, and constitutes the focus of this study. Using our previously validated large animal model of combined TBI and hemorrhagic shock (HS), we hypothesized that FFP resuscitation

ACCEPTED MANUSCRIPT 4

would be associated with modulation of brain gene expression profiles compared with normal saline (NS) and colloids (Hextend, HEX), and that these gene alterations could further be

METHODS

RI PT

associated with clinically relevant endpoints: brain lesion size and swelling.

This study adhered to the guidelines stipulated in the Animal Welfare act as well as federal

SC

statues regarding animal experiments. All experiments were performed under the supervision of veterinarian and were approved by the Institutional Animal Care and Use Committee This study

M AN U

represents secondary use of biomaterial retrieved from a previously published study (10).

Animal model

This porcine model of combined TBI and HS has previously been described in detail (10).

TE D

Briefly, 15 female Yorkshire swine (42.50kg, Tufts Veterinary School, Grafton, MA) were utilized for this study. Animals were anesthetized using isoflurane and cannulations of the left femoral and internal jugular veins as well as right and left femoral arteries were performed using

EP

a cutdown technique. A 20-mm bur hole was made on the right side of the skull in order to expose the dura. This was used for the creation of TBI using a computer controlled cortical

AC C

impact device as previously described(10). An additional 2mm bur hole located 10mm lateral and 10mm anterior to the bregma was used for insertion of a combined Intracranial Pressure (ICP) and Brain oxygenation measurement device (Integra Lifesciences, Plainsboro, NJ). Following the cortical impact, animals were subjected to a controlled 40% hemorrhage followed by two hours of shock, targeting a mean arterial pressure of 35mmHg. Animals were then resuscitated (n=5/group) with FFP, NS or HEX. Volumes of FFP and HEX equaled that of the

ACCEPTED MANUSCRIPT 5

shed blood whereas NS was three times the volume of shed blood. FFP was procured from healthy porcine donors as previously described (10). Following resuscitation, animals were

RI PT

observed under anesthesia for 6 hours prior to sacrifice.

Tissue sampling and RNA preparation

Tissue sampling and RNA preparation has been previously described in detail (16). Briefly,

SC

immediately upon sacrifice, brains were harvested and sliced into 5mm sections in the coronal plane. Brain slices were incubated in 2% 2,3,5 triphenyltetrazolium chloride (Sigma Chemical

M AN U

Co, St. Louis, MO) in order to stain viable tissue. Lesion size was then calculated by the use of the ImageJ software package (National Institutes of Health). The degree of brain swelling was assessed by comparing the ipsilateral and contralateral hemisphere volumes as previously described (10). Thirty milligram pieces of tissues were obtained from the penumbra of the injury

TE D

inferior to the lesion, and homogenized. RNA was then extracted from the samples (RNeasy mini kit, Qiagen, Valencia, CA) and prepared for microarray analysis with the GeneChip WT Plus Reagent kit (Affymetrix Inc, Santa Clara, CA) according to manufacturer instructions. RNA was

EP

then reverse transcribed to synthesize complementary cDNA as previously described (16), prior to hybridization onto an Affymetrix porcine Gene ST 1.1 microarray (Affymetrix Santa Clara,

AC C

CA). For each gene, the expression values were calculated using a multiarray average. This consisted of background correction, quantile normalization and summarization of the expression values (16, 17).

Weighted gene correlation network analysis (WGCNA)

ACCEPTED MANUSCRIPT 6

Network analysis was performed using the WGCNA r-package (18). Briefly, the WGCNA algorithm creates a gene co-expression network by correlating the expression of each gene on the microarray with all other genes on the array. This co-expression network is then transformed into

RI PT

an equivalent weighted matrix of connection strengths by a power function (soft-thresholding level). A hierarchical clustering algorithm is then utilized for the identification of clusters of coexpressed genes.

SC

Following analysis of network topology, a soft thresholding level of 6 was utilized for the

creation of the weighted matrix. Minimal cluster size was set to 10 genes. Following clustering,

M AN U

principal component analysis was used to extract the principal component (cluster eigenvector) of each expression cluster. This entity thus denotes the overall direction and magnitude of the expression of the genes comprising the cluster in question.

TE D

Associations of expression clusters with resuscitation regimens and clinical outcomes. Using linear regression, cluster eigenvectors were associated with resuscitation regimens (FFP, NS and HEX). Between group comparisons of directions and magnitudes of cluster eigenvectors

EP

was done using ANOVA. P-values were post-hoc corrected using False Detection Rate (FDR) as proposed by Benjamini and Hochberg (19). Minimal level for declaring significance for the

AC C

resulting FDR q-values was set at <0.05. In order to access the clinical relevance of gene expression, the above identified expression clusters were subjected to regression analysis associating cluster eigenvectors with clinically relevant outcome parameters; Brain lesion size and swelling. In order to elucidate a potential hemodynamic effect of resuscitation regimens on gene expression, eigenvectors were furthermore associated with relevant vital signs and hemodynamic parameters, including mean arterial pressure (MAP), cardiac output (CO),

ACCEPTED MANUSCRIPT 7

intracranial pressure (ICP) and brain oxygenation. An Area Under the Curve (AUC) value was calculated for all these parameters in order to extract a single value representing the changes over the time course of the experiment. p Values for these correlations with clinical outcomes were

RI PT

not post-hoc corrected for the FDR due to the limited number of comparisons.

Intramodular analysis

SC

Clusters with a significant association to FFP resuscitation were subjected to intramodular

analysis in order to identify central genes in the relevant cluster as well as genes with a strong

M AN U

association with FFP resuscitation. Gene Significance (GS) for FFP resuscitation, defined as the absolute value of the correlation between the gene expression and presence or absence of FFP resuscitation, was calculated according to a method previously described (18). Identification of genes with a central role in the cluster in question were done through quantification of Module

TE D

Membership (MM), defined as the absolute correlation between gene expression and the cluster eigenvector(18). The top 10 genes with the strongest association with the identified clusters (highest MM) as well as FFP resuscitation (Highest GS) were identified and their function

EP

investigated using human equivalent Entrez gene information (Entrez online database,

AC C

www.ncbi.nlm.gov/gquery).

RESULTS

Physiological effects of injury and resuscitation The effects of injury and resuscitation on physiological parameters have previously been described in detail(10). Overall, the model had a 10% mortality rate, but all animals that survived the 2 hour shock phase survived resuscitation and were included in the experiment. No

ACCEPTED MANUSCRIPT 8

differences in MAP, Heart rate (HR), CO, ICP or brain oxygenation were observed at baseline or following the two-hour shock phase. Following injury and resuscitation, animals resuscitated with FFP had lower MAP compared with HEX, but this effect was not sustained throughout the

RI PT

observation period and no differences were observed prior to sacrifice. No differences in MAP were observed between FFP and NS. CO, HR, ICP and brain oxygenation did not significantly differ between FFP and NS/HEX at any time point following resuscitation.

SC

Arterial blood gas (ABG) analysis showed comparable baseline values between groups for pH, P02, PC02, lactate and hemoglobin. Following resuscitation and the 6-hour observation period, no

M AN U

difference between FFP and NS/HEX resuscitated animals could be identified. For actual values, the reader is kindly referred to the primary study(10).

Associations between gene expression clusters and resuscitation method

TE D

WGCNA identified 322 clusters of co-expressed genes, comprising a total of 12.462 of the 27.558 (45%) probes on the microarray.

Regression analysis of cluster eigenvectors, indicating the overall direction and magnitude of the

EP

gene expression cluster, identified an association between increased cluster eigenvectors of two clusters of co-expressed genes and FFP resuscitation (Cluster A: 439 genes q-value 0.02, Cluster

AC C

B: 54 genes, q-value 0.04). No association between co-expression clusters and resuscitation with NS or HEX could be identified. Between group comparisons (Figure 1) indicated that FFP resuscitation resulted in higher expression of genes in cluster A compared with HEX (0.28±0.10 vs. -0.26±0.12, q=0.02), but no difference compared with NS (0.28±0.10 vs. -0.03±0.20, q=0.20). For genes in cluster B (Figure 1), FFP was associated with higher cluster expression

ACCEPTED MANUSCRIPT 9

compared with both NS (0.32±0.11 vs. -0.18±0.13, q=0.03) and HEX (0.32±0.11 vs. -0.13±0.17, q=0.04).

RI PT

Associations between gene expression clusters, clinical outcomes and vital signs

Main results of the regression analysis, associating gene expression clusters A and B with

clinical outcomes (Brain lesion size, swelling and ICP) as well as vital signs (MAP, CO and

SC

Brain Oxygenation) are summarized in table 1 and selected values are graphically depicted in Figure 2. Increased cluster gene expression (positive cluster eigenvectors) in cluster A and B

M AN U

were negatively associated with brain swelling (Figure 2, β-coefficient -13.0±6.3%, p=0.05 and 16.8±5.5%, p=0.009 respectively). Brain lesion size was negatively associated with increased expression of genes in cluster B (Figure 2, β-coefficient -1800.7mm3±598.3, p=0.01), but not cluster A (Table 1 and Figure 2).

TE D

ICP was negatively associated with eigenvectors of expression cluster B, but no association was identified with cluster A. MAP, CO and brain oxygenation were not associated with cluster

EP

eigenvectors.

Intramodular analysis: Identifying central cluster genes

AC C

Figure 3 shows the correlations plots for Gene Significance (GS) for FFP resuscitation vs. Module Membership (MM) of expression clusters A and B respectively. There was a high correlation between GS for FFP resuscitation and MM in both cluster A (r=0.8, p<1*10-200) and cluster B (r=0.72, p=1.8*10-19), indicating that central genes in clusters A/B were also tightly associated with FFP resuscitation. The 10 top genes with the highest GS for FFP resuscitation and MM for clusters A and B respectively are listed in table 2.

ACCEPTED MANUSCRIPT 10

Pathway analysis Results of the pathway analysis are shown in Figure 4. Upregulated genes in cluster A

RI PT

significantly mapped to several pathways, notably involving prostacyclin signaling (-log10 FDR 13), glucagon signaling (-log10 FDR 9), aquaporin mediated transport (-log10 FDR 7) as well as platelet homeostasis (-log10 FDR 4). Down regulated genes in Cluster A mapped to pathways

SC

involved in Tumor Necrosis factor α signaling (-log10 FDR 4), Death Receptor Signaling (-log10 FDR 3), Extrinsic pathway (-log10 FDR 3). Furthermore, multiple pathways involved in

M AN U

interleukin 3,5 and 7 as well as GM-CSF signaling was significantly enriched. Upregulated genes in Cluster B mapped to pathways involved in aspects of collagen formation and assembly as well as p38MAP kinase events (-log10 FDR 2). No significant mapping of down regulated

DISCUSSION

TE D

genes to pathways could be identified for cluster B.

In this study, we demonstrate that resuscitation with FFP was associated with modulation of

EP

brain gene expression profiles compared with NS and HEX. Modulated gene clusters furthermore had clinically relevant associations with level of brain swelling (Clusters A and B),

AC C

lesion size (Cluster B) as well as ICP (Cluster B). In contrast, no correlation between expression clusters and the selected vital signs could be demonstrated. Taken together, these results suggest that the previously observed beneficial effects of FFP resuscitation on brain lesion size and swelling may be driven by changes at the level of the brain transcriptome. Furthermore, these effects seem to be independent of the differential hemodynamics effects of the resuscitation fluids, suggesting that mechanisms outside of restoration of the intravascular volume may be in

ACCEPTED MANUSCRIPT 11

play. This concept is furthermore strengthened by the observation of a strong association between gene significance for FFP resuscitation and gene module membership, indicating that the expression of a number of genes altered by FFP resuscitation may also play a relevant role in

RI PT

the development of brain swelling and lesion size following injury. While the lesion size is to a large extend governed by the mechanical forces of the initial insult, these results also suggest that factors other than sheer mechanics may govern final lesion size. These findings are in line with

SC

previous observations from the same animal model indicating that the addition of valproic acid, a potent modulator of the immune response, to both HEX and FFP further reduced lesion size(20,

M AN U

21).

Furthermore, it is interesting to note that while both HEX and FFP have been shown to reduce brain swelling in previous models(10), only FFP was associated with brain transcriptomic changes in this model. This could suggest that HEX exerted its protective effect mainly through

TE D

changes in plasma osmolality and subsequent starling-forces mediated conservation of fluid in the vasculature. As FFP is a complex fluid containing a plethora of different proteins, it could be speculated that pertubations of immunological and transciptomic changes were induced through

EP

yet undiscovered mechanisms.

Although transcriptomic and epigenetic changes have previously been associated with trauma

AC C

and resuscitation (1, 4, 22), little is known of the effects of crystalloids and artificial colloids compared to FFP on gene expression profiles.. Pathway analysis results from the current study suggests that the previously observed protective effects of FFP following brain injury (10) may in part be mediated through an upregulation of genes involved in prostacyclin and platelet signaling, glucagon and insulin regulation, aquaporin transport as well as collagen formation and biosynthesis.

ACCEPTED MANUSCRIPT 12

Prostacyclin produced by endothelial cells play a pivotal role in microvascular dilation and platelet homeostasis. Prostacyclin thus serves as a potent platelet inhibitor, inducing a reduction in both platelet activation and aggregation to the vessel wall. As several studies have identified a

RI PT

presumably hyperactivation-driven dysfunction of platelets following TBI (23, 24), it could be speculated that the previously reported ill effects of artificial resuscitation fluids can be mitigated by the protective effect of FFP resuscitation on platelet function following TBI and shock(13),

SC

was in part mediated through increased prostacyclin signaling. This again could theoretically be associated with reduced microvascular thrombosis and conservation of microvascular flow and

M AN U

oxygen delivery to the injured areas of the brain.

Down-regulated genes in cluster A showed a strong association with multiple inflammation related pathways, such as Tumor Necrosis Factor α (TNF-α) and death receptor signaling as well as modulation of interleukins 3,5 and 7. As platelets play an immune-modulatory role that far

TE D

extends beyond the constraints of the coagulation system (25), the observation in this study of a concurrent down regulation of genes (cluster A) involved in platelet signaling could either be a driver or a direct effect of this reduced inflammatory drive.

EP

Interestingly, both preclinical and clinical studies have indicated a potential protective effect of prostacyclin treatment following TBI (26, 27), including a reduction in circulating inflammatory

AC C

markers such as interleukin 6 (IL-6) (27). This concept is supported by several reports indicating a down regulation of genes involved in endothelial activation and permeability following FFP resuscitation (28) as well as a preservation of the endothelial glycocalyx layer (12, 29). Taken together, these results suggest that FFP may in part exert its protective effects through genes involved in endothelial and platelet stabilization following injury. Whether the observed perturbations of inflammation signaling is secondary to these events or the results of primary

ACCEPTED MANUSCRIPT 13

actions of FFP cannot be readily deduced from these results. Regardless, it would seem that FFP resuscitations modulates local inflammatory signaling pathways at the site of the injury. Our results do, however, also suggest that mechanisms outside of platelets, endothelium and

RI PT

inflammation may be modulated by TBI and FFP resuscitation. Indeed, multiple different metabolic pathways, including glucagon and insulin signaling, were associated with FFP

resuscitation in this model. In line with this, previous studies have indicated alterations in

SC

glucose metabolism and glycolysis following TBI (30, 31). Although the therapeutic

implications of these findings are yet to be clearly defined, glucagon like peptide (GLP-1)

M AN U

signaling has also been associated with preservation of the blood brain barrier (BBB) and improved sensorimotor function in animals treated with GLP-1 following TBI (32). Furthermore, FOXO1A signaling, inhibited by genes upregulated in cluster B, has been implicated in gluconeogenesis, glycogenolysis as well as insulin signaling. Interestingly, upregulation of the

TE D

FOXO1A gene has also been associated with BBB disruption following subarachnoid hemorrhage (33).It could thus be speculated that FFP resuscitation modulated brain metabolism either through preservation of the microvascular flow and consequent improved micronutrient

EP

supply or through other yet to be elucidated mechanisms. In line with this, we have previously reported superior brain metabolic profiles and conservation of mitochondrial energy production

AC C

in swine resuscitated with FFP compared with normal saline (15). The observation of an association between FFP resuscitation and up regulation of genes involved in collagen synthesis and turnover is less clear. As type IV collagen is a major constituent of the blood brain barrier, it could be speculated that the increase in genes involved in collagen synthesis represent an increase in reparative mechanisms following injury. The observed inverse association between brain lesion size and swelling with up regulation of genes in cluster B

ACCEPTED MANUSCRIPT 14

strengthens this possibility. As others have not previously demonstrated these associations, this does however remain speculative. The potential protective effect of FFP resuscitation on the BBB is, however, strengthened by the observation of an inverse association between level of

RI PT

brain swelling and up regulation of genes in both clusters A and B. Aside from the potential effects of collagen synthesis observed in cluster B, Cluster A was also significantly enriched for pathways with a potential role in BBB integrity, namely aquaporin signaling. As such,

SC

suppression of aquaporins plays a central role in BBB disruption and the emergence of brain edema following TBI(34).

M AN U

Several limitations exist in this study and should be acknowledged. First, the controlled laboratory setting as well as the species difference may not yield complete clinical translatability. Numerous aspects inherent to most laboratory animal models, including fixed volume hemorrhage, anesthesia and TBI injury creation modality may impact on results and reduce

TE D

clinical translatability(35-37). As such, the observed results should ideally be reproduced across other species although validation in humans appear less realistic due to the obvious ethical concerns involved in securing samples from the injured brain. Secondly, it should be noted that

EP

although brain swelling differed between groups, this model only induced a rise in ICP that did not reach clinically relevant levels. Although this could bring the clinical relevance of the model

AC C

into question, the lack of change is likely related to the short observation period (6 hours) employed here. As such, the observed changes in gene transcriptomics should be interpreted with this in mind. Third,, it should be noted that the observed differences are the results of comparisons of FFP versus other resuscitation fluids. Whether the observed results are a direct effect of FFP or a combination of this with a detrimental effect of the comparative resuscitation fluid cannot readily be deduced. Also, comparisons with other resuscitation fluids (eg. albumin)

ACCEPTED MANUSCRIPT 15

may have yielded other results. In line with this, it is also important to acknowledge the relatively small sample size of this study. As such, minor pertubations in hydrostatic or oncotic pressures induced by the different resuscitation fluids may impact on the observed results

RI PT

without differences in surrogate markers such as ICP or brain oxygenation. We thus cannot

deduce whether the observed differences could be due to differences in fluid composition or direct immunological effects of FFP proteins.

SC

Also, the need for cross species validation is underlined by the fact that the porcine genome and transcriptome is less well studied than its murine, rodent and human counterparts. As many

M AN U

porcine genes thus have less well-defined functions, this again impedes the clinical translatability. Furthermore, the utilization of different gene expression pathway analysis tools may have yielded other results.

Finally, FFP is a complex fluid with more than a 1000 proteins, and the precise mechanisms by

TE D

which FFP and artificial fluids exert their effects on the transcriptome can’t be determined in this model. Thus, the results presented here should be seen as hypothesis generating and should serve as the basis for subsequent, targeted studies into the mechanistic effects of FFP resuscitation.

EP

Owing to the above-mentioned considerations, care should be taken before translating these results to a clinical scenario before validation in other models (including lower species). The key

AC C

finding in this study is thus not so much the individual genes affected, as the knowledge of the porcine genome remains limited. Rather, the fact the FFP resuscitation incurred transciptomic changes strongly associated with clinically relevant outcomes (brain lesion size and swelling) suggests that the effects of FFP resuscitation extend beyond hemodynamics and the coagulation system. These results thus corroborate numerous previous studies and underline the need for further studies in order to increase our understanding of the mechanistic effects underlying the

ACCEPTED MANUSCRIPT 16

protective effects of FFP in trauma. Furthermore, these and previous results add to the growing debate of the optimal prehospital resuscitation fluid and suggest that plasma based resuscitation

AC C

EP

TE D

M AN U

SC

RI PT

may be preferable to colloids in both the civilian and military setting.

ACCEPTED MANUSCRIPT 17

REFERENCES 1.

Chen H, Alam HB, Querol RI, et al. Identification of expression patterns associated

with hemorrhage and resuscitation: integrated approach to data analysis. J Trauma 2006;60:701-

2.

RI PT

723; discussion 723-724.

White TE, Ford GD, Surles-Zeigler MC, et al. Gene expression patterns following

unilateral traumatic brain injury reveals a local pro-inflammatory and remote anti-inflammatory

3.

SC

response. BMC Genomics 2013;14:282.

Desai KH, Tan CS, Leek JT, et al. Dissecting inflammatory complications in

M AN U

critically injured patients by within-patient gene expression changes: a longitudinal clinical genomics study. PLoS Med 2011;8:e1001093. 4.

Alam HB, Stegalkina S, Rhee P, Koustova E. cDNA array analysis of gene

expression following hemorrhagic shock and resuscitation in rats. Resuscitation 2002;54:195-

5.

TE D

206.

Kasotakis G, Sideris A, Yang Y, et al. Aggressive early crystalloid resuscitation

adversely affects outcomes in adult blunt trauma patients: an analysis of the Glue Grant database.

6.

EP

J Trauma Acute Care Surg 2013;74:1215-1221; discussion 1221-1222. Holcomb JB, Tilley BC, Baraniuk S, et al. Transfusion of plasma, platelets, and red

AC C

blood cells in a 1:1:1 vs a 1:1:2 ratio and mortality in patients with severe trauma: the PROPPR randomized clinical trial. JAMA 2015;313:471-482. 7.

Brown LM, Aro SO, Cohen MJ, Trauma Outcomes G, Holcomb JB, Wade CE, et

al. A high fresh frozen plasma: packed red blood cell transfusion ratio decreases mortality in all massively transfused trauma patients regardless of admission international normalized ratio. The J Trauma 2011;71:S358-363.

ACCEPTED MANUSCRIPT 18

8.

Zink KA, Sambasivan CN, Holcomb JB, et al. A high ratio of plasma and platelets

to packed red blood cells in the first 6 hours of massive transfusion improves outcomes in a large multicenter study. Am J Surg 2009;197:565-570; discussion 570. Peiniger S, Nienaber U, Lefering R, et al. Balanced massive transfusion ratios in

RI PT

9.

multiple injury patients with traumatic brain injury. Crit Care 2011;15:R68. 10.

Jin G, DeMoya MA, Duggan M, et al. Traumatic brain injury and hemorrhagic

SC

shock: evaluation of different resuscitation strategies in a large animal model of combined insults. Shock 2012;38:49-56.

Imam A, Jin G, Sillesen M, et al. Fresh frozen plasma resuscitation provides

M AN U

11.

neuroprotection compared to normal saline in a large animal model of traumatic brain injury and polytrauma. J Neurotrauma 2015;32:307-313. 12.

Kozar RA, Peng Z, Zhang R, et al. Plasma restoration of endothelial glycocalyx in

TE D

a rodent model of hemorrhagic shock. Anesth Analg 2011;112:1289-1295. 13.

Sillesen M, Johansson PI, Rasmussen LS, et al. Fresh frozen plasma resuscitation

attenuates platelet dysfunction compared with normal saline in a large animal model of

EP

multisystem trauma. J Trauma Acute Care Surg 2014;76:998-1007. 14.

Sillesen M, Jin G, Oklu R, et al. Fresh-frozen plasma resuscitation after traumatic

AC C

brain injury and shock attenuates extracellular nucleosome levels and deoxyribonuclease 1 depletion. Surgery 2013;154:197-205. 15.

Hwabejire JO, Imam AM, Jin G, et al. Differential effects of fresh frozen plasma

and normal saline on secondary brain damage in a large animal model of polytrauma, hemorrhage and traumatic brain injury. J Trauma Acute Care Surg 2013;75:968-974; discussion 974-975.

ACCEPTED MANUSCRIPT 19

16.

Dekker SE, Bambakidis T, Sillesen M, et al. Effect of pharmacologic resuscitation

on the brain gene expression profiles in a swine model of traumatic brain injury and hemorrhage. J Trauma Acute Care Surg 2014;77:906-912; discussion 912. Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and summaries of

RI PT

17.

high density oligonucleotide array probe level data. Biostatistics 2003;4:249-264.

Langfelder P, Horvath S. WGCNA: an R package for weighted correlation network

analysis. BMC Bioinformatics 2008;9:559. 19.

SC

18.

Benjamini Y, Drai D, Elmer G, et al. Controlling the false discovery rate in

20.

M AN U

behavior genetics research. Behav Brain Res 2001;125:279-284.

Jin G, Duggan M, Imam A, et al. Pharmacologic resuscitation for hemorrhagic

shock combined with traumatic brain injury. J Trauma Acute Care Surg 2012;73:1461-1470. 21.

Imam AM, Jin G, Duggan M, et al. Synergistic effects of fresh frozen plasma and

Surgery 2013;154:388-396. 22.

TE D

valproic acid treatment in a combined model of traumatic brain injury and hemorrhagic shock.

Alam HB, Shults C, Ahuja N, et al. Impact of resuscitation strategies on the

2008;76:299-310.

Castellino FJ, Chapman MP, Donahue DL, et al. Traumatic brain injury causes

AC C

23.

EP

acetylation status of cardiac histones in a swine model of hemorrhage. Resuscitation

platelet adenosine diphosphate and arachidonic acid receptor inhibition independent of hemorrhagic shock in humans and rats. J Trauma Acute Care Surg 2014;76:1169-1176. 24.

Sillesen M, Johansson PI, Rasmussen LS, et al. Platelet activation and dysfunction

in a large-animal model of traumatic brain injury and hemorrhage. J Trauma Acute Care Surg 2013;74:1252-1259.

ACCEPTED MANUSCRIPT 20

25.

Ostrowski SR, Johansson PI. Rethinking platelet function: thrombocytopenia

induced immunodeficiency in critical illness. Med Hypotheses 2011;77:798-802. 26.

Bentzer P, Mattiasson G, McIntosh TK, et al. Infusion of prostacyclin following

RI PT

experimental brain injury in the rat reduces cortical lesion volume. J Neurotrauma 2001;18:275285. 27.

Wahlstrom MR, Olivecrona M, Ahlm C, et al. Effects of prostacyclin on the early

SC

inflammatory response in patients with traumatic brain injury-a randomised clinical study. Springerplus 2014;3:98.

Potter DR, Baimukanova G, Keating SM, et al. Fresh frozen plasma and spray-

M AN U

28.

dried plasma mitigate pulmonary vascular permeability and inflammation in hemorrhagic shock. J Trauma Acute Care Surg 2015;78:S7-S17. 29.

Torres LN, Sondeen JL, Ji L, et al. Evaluation of resuscitation fluids on endothelial

TE D

glycocalyx, venular blood flow, and coagulation function after hemorrhagic shock in rats. J Trauma Acute Care Surg 2013;75:759-766. 30.

O'Connell MT, Seal A, Nortje J, et al. Glucose metabolism in traumatic brain

EP

injury: a combined microdialysis and [18F]-2-fluoro-2-deoxy-D-glucose-positron emission tomography (FDG-PET) study. Acta neurochirurgica Supplement 2005;95:165-168. Bergsneider M, Hovda DA, Shalmon E, et al. Cerebral hyperglycolysis following

AC C

31.

severe traumatic brain injury in humans: a positron emission tomography study. J Neurosurg 1997;86:241-251. 32.

Hakon J, Ruscher K, Romner B, Tomasevic G. Preservation of the blood brain

barrier and cortical neuronal tissue by liraglutide, a long acting glucagon-like-1 analogue, after experimental traumatic brain injury. PLoS One 2015;10:e0120074.

ACCEPTED MANUSCRIPT 21

33.

Li Z, Liang G, Ma T, et al. Blood-brain barrier permeability change and regulation

mechanism after subarachnoid hemorrhage. Metab Brain Dis 2015;30:597-603. 34.

Blixt J, Svensson M, Gunnarson E, Wanecek M. Aquaporins and blood-brain

2015;1611:18-28.

Bellamy RF, Maningas PA, Wenger BA. Current shock models and clinical

correlations. Ann Emerg Med 1986;15:1392-1395. 36.

SC

35.

RI PT

barrier permeability in early edema development after traumatic brain injury. Brain Res

Clausen F, Hillered L. Intracranial pressure changes during fluid percussion,

M AN U

controlled cortical impact and weight drop injury in rats. Acta Neurochir (Wien) 2005;147:775780; discussion 780. 37.

Mondello S, Shear DA, Bramlett HM, et al. Insight into pre-clinical models of

traumatic brain injury using circulating brain damage biomarkers: operation brain trauma

AC C

EP

TE D

therapy. J Neurotrauma 2016;33:595-605.

ACCEPTED MANUSCRIPT 22

Table 1. Associations between Gene Expression Clusters A and B and Clinical Outcomes Cluster A β-coefficient ±

Cluster B p Value

SE -1180.1±707.0

0.12

-1800.7±598.3

0.01*

-0.13±0.06

0.05*

-0.17±0.06

0.009*

-0.0001±0.0001

0.50

-0.0002±0.0001

0.18

0.001±0.0008

0.21

-0.0006±0.0008

0.52

0.0001±0.0002

0.52

6.70×10-5±0.0001

0.70

-0.0005±0.0002

0.05*

mm3 Brain swelling,

pressure, AUC Cardiac output,

oxygenation, AUC Intracranial pressure, AUC

-0.0002±0.0002

EP

Brain

TE D

AUC

M AN U

% Mean arterial

p Value

SE

SC

Brain lesion size,

β-coefficient ±

RI PT

Outcomes

0.40

AC C

Mean arterial pressure, cardiac output, brain oxygenation and intracranial pressures over the course of the experiment were calculated as the area under the curve (AUC) for the respective measures.

*Significant.

ACCEPTED MANUSCRIPT 23

Table 2. Top10 Genes with the Highest Association with Fresh Frozen Plasma Resuscitation (Gene Significance) Clusters A and B Membership (Module Membership) Human gene function summary

Beta glucosidase

Carbohydrate transport and metabolism Scaffold protein in multiple signaling pathways Unknown Negative regulator of autophagy and endocytic trafficking Stabilization of p53/TP53 Member of zinc finger DHHC domain-containing protein family. May function as palmitoyltransferase Mobilizes intracellular calcium, second messenger to various stimulation Protein coding Docking and fusion of synaptic vesicles Beta oxidation and transport of long-chain fatty acids into mitochondria

Neurabin-2 like

SGSM2 LOC100153714 LOC100155348 ZDHHC8

Small G protein signaling modulator 2 Run domain Beclin-1 interacting and cysteine-rich containing protein BRI3-binding protein-like Zinc finger, DHHC-type containing 8

INPP5A

Inositol polyphosphatase-5-phosphatase

PRR14L STX16

Proline rich 14-like Syntaxin 16

CPT1C

Carnitine palmitoyltransferase

TE D

Cluster B TTBK2

M AN U

LOC100516875

Tau tubulin kinase 2

Serine/arginine splicing factor 2 SIX homeobox 4

LOC100739045

Glycine dehydrogenase, mitochondrial like Beta glucosidase

SMYD3

Nuclear receptor subfamily 6, group A, member 1 SET and MYND Domain containing 3

LOC100512626

ORM1-like protein 3-like

LOC100522896 SLC26A6

Tectin3-like Solute carrier family 26, member 6

AC C

NR6A1

EP

SRSF12 SIX4

GBA2

RI PT

Title

SC

Gene Cluster A GBA2

Phosphorylation of tau and tubulin proteins Protein coding Transcription factor with potential role in differentiation or maturation of neuronal cells Degradation of glycine Carbohydrate transport and metabolism Nuclear hormone receptor family. May be involved in neurogenesis Histone methyltransferase, functions in RNA polymerase II Negative regulator of sphingolipid synthesis Filament forming protein Transport of chloride, oxalate, sulfate and bicarbonate

ACCEPTED MANUSCRIPT 24

FIGURE LEGENDS Figure 1. Comparisons of cluster eigenvectors, a vector indicating the overall direction and magnitude of the genes in (A) gene cluster A and (B) gene cluser B, between different

RI PT

resuscitation fluids. FFP, fresh frozen plasma; HEX, hetastarch; NS, normal saline.

Figure 2. Graphical depiction of associations between gene eigenvectors for expression (A)

SC

clusters A and (B) cluster B, with clinical outcomes: brain swelling and brain lesion size.

M AN U

Figure 3. Plot of gene significance (defined as the absolute value of the correlation between the gene expression and presence or absence of fresh frozen plasma [FFP] resuscitation) vs module membership (defined as the absolute correlation between gene expression and the cluster eigenvector) in (A) cluster A and (B) cluster B. The observed high degree of correlation indicates

resuscitation.

TE D

that central genes in the identified module module also were highly associated with FFP

AC C

EP

Figure 4. Pathway analysis of up- and down-regulated genes in clusters A and B.

ACCEPTED MANUSCRIPT 25

Precis In this study, we investigated the effect of fresh frozen plasma resuscitation after traumatic brain

RI PT

injury and shock in brain gene expression in a swine model, and found that resuscitation after traumatic brain injury is associated with alterations in gene expression profiles, including

AC C

EP

TE D

M AN U

SC

inflammatory, metabolic, and platelet signaling pathways.

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT

EP

TE D

S

M AN U

AC C

EP

TE D

M AN U

SC

RI PT

ACCEPTED MANUSCRIPT