Hyperglycemia Alters Expression of Cerebral Metabolic Genes after Cardiac Arrest

ARTICLE IN PRESS

Hyperglycemia Alters Expression of Cerebral Metabolic Genes after Cardiac Arrest Rickard Per Fredrik Lindblom, MD, PhD,*,† Maria Molnar, MD, PhD,‡ Charlotte Israelsson, PhD,§ Belinda Röjsäter, MSc,‖ Lars Wiklund, MD, PhD,‡ and Fredrik Lennmyr, MD, PhD*,†

Background: Survivors of cardiac arrest often experience neurologic deficits. To date, treatment options are limited. Associated hyperglycemia is believed to further worsen the neurologic outcome. The aim with this study was to characterize expression pathways induced by hyperglycemia in conjunction with global brain ischemia. Methods: Pigs were randomized to high or normal glucose levels, as regulated by glucose and insulin infusions with target levels of 8.5-10 mM and 4-5.5 mM, respectively. The animals were subjected to 5-minute cardiac arrest followed by 8 minutes of cardiopulmonary resuscitation and direct-current shock to restore spontaneous circulation. Global expression profiling of the cortex using microarrays was performed in both groups. Results: A total of 102 genes differed in expression at P < .001 between the hyperglycemic and the normoglycemic pigs. Several of the most strongly differentially regulated genes were involved in transport and metabolism of glucose. Functional clustering using bioinformatics tools revealed enrichment of multiple biological processes, including membrane processes, ion transport, and glycoproteins. Conclusions: Hyperglycemia during cardiac arrest leads to differential early gene expression compared with normoglycemia. The functional relevance of these expressional changes cannot be deduced from the current study; however, the identified candidates have been linked to neuroprotective mechanisms and constitute interesting targets for further studies. Key Words: Cerebral—ischemia-reperfusion—gene expression—glucose— hyperglycemia—microarray—pigs. © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved.

From the *Department of Cardiothoracic Surgery and Anaesthesia, Uppsala University Hospital, Uppsala, Sweden; †Department of Surgical Sciences, Section of Thoracic Surgery; ‡Department of Surgical Sciences, Section of Anaesthesiology and Intensive Care; §Department of Neuroscience, Developmental Neuroscience; and ‖Department of Medical Sciences, Science for Life Laboratory, Uppsala University, Uppsala, Sweden. Received September 14, 2017; accepted November 26, 2017. Rickard Per Fredrik Lindblom and Maria Molnar had equal contribution. Grant support: Grants were obtained from the Laerdal Foundation and Erik, Karin & Gösta Selander’s Foundation. The funders had no role in study design, data collection and analysis, decisionto publish, or preparation of the manuscript. Conflict of interest: The authors have not disclosed any potentialconflicts of interest. Address correspondence to Maria Molnar, MD, PhD, Department of Surgical Sciences, Section of Anaesthesiology and Intensive Care, Uppsala University Hospital, SE-751 85 Uppsala, Sweden. E-mail:[email protected]. 1052-3057/$ - see front matter © 2017 National Stroke Association. Published by Elsevier Inc. All rights reserved. https://doi.org/10.1016/j.jstrokecerebrovasdis.2017.11.036

Journal of Stroke and Cerebrovascular Diseases, Vol. ■■, No. ■■ (■■), 2017: pp ■■–■■

1

ARTICLE IN PRESS R.P.F. LINDBLOM ET AL.

2

Introduction Cerebral ischemia after cardiac arrest (CA) is common and life-threatening, and those who survive often experience significant cerebral dysfunction. The treatment options are limited and aim at optimizing cerebral perfusion and oxygenation complemented with induced therapeutic hypothermia.1,2 Concomitant hyperglycemia is frequently perceived after CA3 and worsens neurologic outcome.4,5 Glycemic control is therefore an important clinical goal6 among other efforts to attenuate the neurologic damage after CA. The development of neuronal injury following global ischemia is likely exacerbated by the reperfusion that follows once blood flow is restored to the ischemic area.7 The absence of cerebral blood flow initiates a cascade of molecular events such as anaerobic glycolysis, lactate acidosis, free radical production, altered cell signaling, inflammatory processes, and activation of gene expression.8 Paradoxically, when replenishment of oxygen occurs, the cerebral tissue damage can be augmented by means of increased production of reactive oxygen and nitrogen species, that is, oxidative stress.9 Concomitant acute hyperglycemia further aggravates the ischemia-reperfusion injury in synergy with mediators of oxidative stress10; however, the causative upstream molecular mechanisms are unknown. In a previous report from our laboratory, a subtle but significant increased level of S100β was identified in hyperglycemic pigs following 5 minutes of CA.11 The aim of the current study was to improve the understanding of the pathogenesis behind how concomitant hyperglycemia can augment the injury after cerebral ischemia. The results could generate hypothesis for the development of new therapeutic options to prevent or minimize secondary neuronal damage after cerebral ischemia. To that end we performed a global transcriptome analysis of brains from hyperglycemic and normoglycemic pigs after CA.

Methods

(.04 mg/kg) was given, after which vein catheters were placed in the ears for intravenous anesthesia and fluid administration. Airway access was through a tracheotomy after 20 mg of morphine, and the animals were mechanically ventilated (Servo I, Macquet, Solna, Sweden) with a respiratory frequency of 25 per minute using volumecontrolled mode with a FiO2 of .30, I-to-E ratio of 1:2. Tidal volumes were adjusted to maintain normocapnia of 5.0-5.5 kPa. An infusion of pentobarbital (Pentothal, 8 mg/ kg/h), morphine (.5 mg/kg/h), and pancuronium bromide (Pavulon, Reading Laboratories, .25 mg/kg/h) was used to maintain anesthesia. Volume loss through ventilation and diuresis was compensated for by Ringer’s acetate (30 mL/kg) during the first hour; thereafter, bolus doses of 100 mL were given on demand to treat hypovolemia (mean arterial pressure < 70 mm Hg) along with tachycardia and increased respiratory variations in blood pressure. Under anesthesia, CA was induced by applying alternating current, resulting in ventricular fibrillation allowed to continue for 5 minutes followed by 8 minutes of cardiopulmonary resuscitation, then direct-current defibrillation was applied to restore spontaneous circulation. Hemodynamic parameters were recorded. Blood was sampled for glucose and protein S100β analyses. These results have been presented.11 After 180 minutes of observation period, the piglets were euthanized under deep anesthesia and the brain was removed and stored in −70°C until usage.

RNA Preparation Isolation of total RNA from 12 pigs (HG = 6 and NG = 6) was performed by quickly dissecting approximately 100 mg of the frontoparietal cortex, and instantly putting the brain tissue in homogenizing buffer containing β-mercaptoethanol according to manufacturer’s protocol (Qiagen Inc., Valencia, CA). The tissue was then immediately homogenized using a Polytron homogenizer and total RNA isolated by RNeasy Mini kit (Qiagen) with absorbance determined at 260 and 280 nm.

Animals and Surgery The experimental procedure was approved by the Uppsala Ethical Committee for Animal Research (C19/ 8) and has been described.11 In brief, 12 triple-breed male piglets, obtained from a single provider (10-12 weeks old, 22-29 kg), were randomized to high or normal glucose levels, as regulated by glucose and insulin infusions with target levels of 8.5-10 mM (hyperglycemic group [HG]) and 4-5.5 mM (normoglycemic group [NG]). There was no weight difference between the groups. The animals were acquired from a local farmer and transported individually to the operation facility, 1 animal per day. Anesthesia was induced and maintained as described.11 An intramuscular injection of tiletamine-zolazepam (Zoletil, Reading Laboratories, Carros, France, 6 mg/kg), xylazine (Rompun, Reading Laboratories, 2.2 mg/kg), and atropine

Microarray Expression Analysis RNA quality was evaluated using the Agilent 2100 Bioanalyzer system (Agilent Technologies Inc, Palo Alto, CA). A total of 250 ng of total RNA from each sample was used to generate amplified and biotinylated sensestrand cDNA from the entire expressed genome according to the GeneChip WT PLUS Reagent Kit User Manual (P/N 703174 Rev. 1, Affymetrix Inc., Santa Clara, CA). GeneChip ST Arrays (GeneChip Porcine Gene 1.0 ST Array) were hybridized for 16 hours in a 45°C incubator, rotated at 60 rpm. According to the GeneChip Expression Wash, Stain and Scan Manual (P/N 702731 Rev. 3, Affymetrix Inc., Santa Clara, CA), the arrays were then washed and stained using the Fluidics Station 450 and finally scanned using the GeneChip Scanner 3000 7G.

ARTICLE IN PRESS HYPERGLYCEMIC CEREBRAL EXPRESSION PATHWAYS

Microarray Data Analysis The raw data were normalized in the free software Expression Console, provided by Affymetrix (http:// www.affymetrix.com), using the robust multi-array average method.12,13 Principal component analysis was applied to visualize the data, more specifically using the “princomp” function in R (http://www.r-project.org). This revealed 1 highly aberrant sample (HG), probably from technical reasons of tissue handling, which was therefore excluded from the analysis (Supplementary Fig S1). After excluding the outlier sample, normalization was performed again as described above. Subsequent analysis of the gene expression data from the remaining 11 samples was carried out in the freely available statistical computing language R (http://www.r-project.org), using packages available from the Bioconductor project (www.bioconductor.org). To search for the differentially expressed genes, between the HG and the NG groups, an empirical Bayes moderated t-test was applied,14 employing the “limma” package with the robust method.15 To address the problem with multiple testing, the P-values were adjusted using the method of Benjamini and Hochberg.16 The adjusted P-values correspond to the false discovery rate (FDR). Heat maps were generated in Genesis 1.0 using the normalized data for each sample. The data were adjusted by using the command “mean center genes,” and hierarchical clustering was performed according to Genesis 1.0 Operation Manual (Austria, 2001, v1.0). The microarray data are available in Minimal Information About a Microarray Experiment-compliant format at the Gene Expression Omnibus database (http://www.ncbi.nlm.nih .gov/geo) under accession code GSE70107.

Quantitative Reverse-Transcriptase PCR The same RNA as extracted above was used. Total RNA (10 ng) was analyzed, and measurements were repeated twice using duplicated and triplicated microwells (25 µL reaction volume). Transcriptional changes were detected using the following primer pairs (reference sequence number and Oligo # at Sigma-Aldrich): SLCO1A2 (XM_003481678, 8018395300-10/0 and 10/1) and SLC5A11 (NM_001110422, 8018395300-50/0 and 50/1). The iScript One-Step RT-PCR Kit with SYBR Green (Bio-Rad Laboratories, Inc., Hercules, CA) was used. Reverse transcription was performed at 50°C during 10 minutes. Thereafter, quantitative reverse-transcriptase polymerase chain reaction (qRT-PCR) was initiated by a hot-start at 95°C during 5 minutes followed by 36 cycles (95°C 10 seconds, 60°C 30 seconds) using a CFX96 thermal cycler (Bio-Rad). Melting curves were acquired by temperature increments of .5°C from 55.0°C to 94.5°C. Two housekeeping genes were used to normalize the levels of mRNA expression of the studied transcripts; glyceraldehyde 3-phosphate dehydrogenase (NM_001206359, 801839530040/0 and 40/1) and RPL13A (NM_001244068, 8018395300-

3

30/0 and 30/1). Normalized expression levels were calculated with the Bio-Rad CFX manager v1.6 (BioRad). See supplementary Table S1 for primer sequences.

Statistics Unpaired t-tests with Welch’s correction (because of different sample size) calculated with GraphPad Prism 5.0 (San Diego, CA) were carried out on the qRT-PCR data; results are represented as mean ± standard error of the mean. A P-value less than .05 was considered statistically significant for the qRT-PCR data. The significance levels of the microarray analysis are reported with exact P-values; for detailed description, see the microarray data analysis section.

Results A previous study identified elevated levels of S100β as a surrogate marker of increased brain injury following CA in conjunction with hyperglycemia compared with normoglycemic CA.11 To assess potential underlying pathways, global expressional profiling of the frontoparietal cortex of the piglets from the NG and HG groups was performed. The expression of more than 19,000 transcripts was studied using microarrays. Analysis revealed significant differences in expression of 102 transcripts at P less than .001 (Table 1). A P-value less than .001 was chosen as P less than .05 or P less than .01 was considered too liberal. All genes with P value less than .001 fulfilled the criteria of FDR less than .20. No regard to fold change was taken, as the time frame for the experiment was only 3 hours after CA and not all genes can be expected to be fully up- or downregulated in this period. A closer look at the genes most strongly differentially regulated between the 2 experimental conditions identified several that are involved in pathways of glucose transport and metabolism, for instance, solute carrier family 5, member 11 (SLC5A11), a sodium/glucose co-transporter (P = 4.3 × 10−7) and 3-hydroxybutyrate dehydrogenase type 1 (BDH1) P = 1.5 × 10−5. SLC5A11 was expressed higher in NG pigs, whereas levels of BDH1 were higher in HG animals (Fig 1, A,B). Another member of the membrane transporter molecules was also among the genes that differed most in expression between the groups, solute carrier organic anion transporter family member 1A2 (SLCO1A2) (P = 8.33 × 10−6) (Fig 1, C). Contactin 3 (CNTN3), a gene that codes for a neuronal membrane associated with synapses, was the transcript that differed most strongly between the groups (P = 3.2 × 10−8) and was higher expressed in HG (Fig 1, D). There were also 2 microRNAs among the top genes, miR-664 with higher expression in NG (P = 1.5 × 10−5) and miR9-2 with higher expression in HG (P = 3.3 × 10−5) (Fig 1, E,F). To assess whether the identified genes could be linked together in related biologic processes, the transcripts were

ARTICLE IN PRESS R.P.F. LINDBLOM ET AL.

4

Table 1. List of genes differentially regulated between hyperglycemic and normoglycemic pigs

Probe set ID

Gene symbol

15220755 15268890

CNTN3 SLC5A11

15189936 15315339 15293513

— LRRC1 SLCO1A2

15331225 15201759 15215274 15222161

LOC100736569 MIR664 — BDH1

15238925

LOC100152050

15274237 15199852 15201868

— — CNIH3

15264626

PRKACA

15293494

KCNJ8

15266042 15320910

MIR9-2 -

15249098

PIGU

15270648 15275318

— LIMS1

15310071 15208814 15218470 15247199 15267935

LOC100736626 EVI2B LOC100153274 R3HDML ARPC1B

15274191 15335059

TMC7 RAB33A

15232937

LOC100524529

15264294 15213654 15218677

LOC100523948 C17ORF107 CHAF1B

15233920 15234692 15243752

— FGFR2IIIC LPCAT1

15253268 15254652 15273633 15305675 15319065

LOC100625033 LOC100524665 LOC100521314 RHBDL2 PRIMA1

Gene description

P-value

FDR

Fold change HG versus NG

Contactin 3 Solute carrier family 5 (sodium/glucose cotransporter), member 11 — Leucine-rich repeat containing 1 Solute carrier organic anion transporter family, member 1A2 Membrane cofactor protein-like microRNA mir-664 — 3-Hydroxybutyrate dehydrogenase, type 1 Ectonucleotide pyrophosphatase/ phosphodiesterase 6 — — Cornichon family AMPA receptor auxiliary protein 31 Protein kinase, cAMP-dependent, catalytic, alpha Potassium inwardly rectifying channel, subfamily J, member 8 microRNA mir-9-2 ribosomal phosphoprotein large PO subunit Phosphatidylinositol glycan anchor biosynthesis, class U — LIM and senescent cell antigen-like domains 1 Uncharacterized protein C6orf154-like Ecotropic viral integration site 2B Putative RNA-binding protein 11-like Peptidase inhibitor R3HDML-like Actin related protein 2/3 complex, subunit 1B, 41 kDa — RAB33A, member RAS oncogene family Transmembrane protein C10orf57 homolog Olfactory receptor 18-like — Chromatin assembly factor 1, subunit B (p60) — FGF receptor 2IIIc Lysophosphatidylcholine acyltransferase 1 Uncharacterized LOC100625033 Olfactory receptor 5AN1-like Olfactory receptor 7A10-like Rhomboid, veinlet-like 2 (Drosophila) Proline-rich membrane anchor 1-like

3.15E−08 4.39E−07

.00060158 .00419651

1.377127693 .812191522

1.68E−06 1.71E−06 8.34E−06

.00817258 .00817258 .02651421

1.212463253 .709336208 .67190532

8.52E−06 1.47E−05 1.38E−05 1.53E−05

.02651421 .02651421 .02651421 .02651421

1.271569598 .767806203 .843971964 1.191675164

1.20E−05

.02651421

.714057686

1.05E−05 1.83E−05 2.05E−05

.02651421 .02795572 .02795572

1.233398987 .7372534 1.310535006

2.01E−05

.02795572

1.144594826

2.80E−05

.03574455

.868794906

3.28E−05 6.86E−05

.0391727 .07718569

1.25166833 .516542134

7.76E−05

.08240634

.848192901

9.86E−05 9.60E−05

.08978935 .08978935

1.248278713 .8965796

9.58E−05 .000121003 .000112984 .000123731 .000117576

.08978935 .09281698 .09281698 .09281698 .09281698

1.180457282 .822612426 1.542090952 1.127053349 .856558931

.000128991 .000131043

.09281698 .09281698

.573562195 .854196108

.000140804

.0928531

.852888925

.000137231 .000198946 .000217305

.0928531 .10389371 .10389371

.662440805 .837193852 1.180347248

.000177515 .000171008 .000207795

.10389371 .10389371 .10389371

.806024622 .846624607 1.156908606

.000203442 .0002006 .000199768 .000203703 .000212395

.10389371 .10389371 .10389371 .10389371 .10389371

.87118969 .864260125 .883708391 .544204419 .726737768 (continued on next page)

ARTICLE IN PRESS HYPERGLYCEMIC CEREBRAL EXPRESSION PATHWAYS

5

Table 1. (continued) Probe set ID

Gene symbol

15321357

-

15213609 15256928 15187496

KIF1C ANGPTL6 TMX3

15190372 15214679

GSN LRRC3B

15236337

SCN2A

15236865

PPP1R1C

15259825 15266944 15185817

— — —

15201325 15245935

PRKCQ SPTLC3

15254894 15327661 15308233 15230912 15243012 15335161 15226165

LOC100519182 SYPL1 — GLTP SSC.88132 SLC9A6 ABCB10

15316472

CHRNB4

15292087

TENC1

15191711 15193187 15199282

— UBE2J1 ABL1

15216214 15239611 15262627 15330724 15319523 15185965

CNTN4 — ANO3 MIR100 — GNB5

15190575 15229921 15284729

— — CASQ1

15189789

ANP32B

15293702 15210723 15198991 15237690 15247903

— — ANGPTL2 CXCR2 DLC1

Gene description

P-value

FDR

Fold change HG versus NG

Fibronectin type-III domain-containing protein C4orf31-like Kinesin family member 1C Angiopoietin-like 6 Thioredoxin-related transmembrane protein 3 Gelsolin Leucine-rich repeat-containing protein 3B-like Sodium channel protein type 2 subunit alpha-like Protein phosphatase 1 regulatory subunit 1C-like Protocadherin beta-11-like — E3 ubiquitin-protein ligase NEDD4like Protein Kinase C, Theta1 Serine palmitoyltransferase, long chain base subunit 3 — Synaptophysin-like protein 1-like — Glycolipid transfer protein — Sodium/hydrogen exchanger 6-like ATP-binding cassette sub-family B member 10, mitochondrial-like Neuronal acetylcholine receptor subunit beta-4-like Tensin like C1 domain containing phosphatase — Ubiquitin-conjugating enzyme E2, J1 c-abl oncogene 1, non-receptor tyrosine kinase Contactin 4 — Anoctamin 3 microRNA mir-100 — Guanine nucleotide-binding protein subunit beta-5-like — — Calsequestrin 1 (fast-twitch, skeletal muscle) Acidic (leucine-rich) nuclear phosphoprotein 32 family, member B — — Angiopoietin-like 2 Chemokine (C-X-C motif) receptor 2 Rho GTPase-activating protein 7-like

.000210193

.10389371

1.322275024

.000225929 .000230553 .000244173

.10497834 .10497834 .10859445

.83329702 .822885883 1.139642163

.000272865 .000286306

.1117412 .1117412

.802425068 1.490075026

.000279606

.1117412

1.155585225

.000267734

.1117412

.60413182

.000283913 .000258564 .000344729

.1117412 .1117412 .13152698

1.123813429 .755913797 .815446235

.000378191 .000378267

.13152698 .13152698

.658701627 1.473109745

.000373933 .000364174 .00035256 .00038691 .000403659 .000400594 .000417929

.13152698 .13152698 .13152698 .13212981 .13309626 .13309626 .13320796

1.103349501 .817493121 .709488346 .87543261 1.163105002 1.134479317 .868195562

.000416329

.13320796

.79225057

.00042751

.13402785

.887311801

.000452368 .000463306 .000472878

.13758822 .13758822 .13758822

.863713545 1.119565031 .782377177

.000492171 .000496423 .000488369 .000495566 .000460673 .000543385

.13758822 .13758822 .13758822 .13758822 .13758822 .14636184

1.230775971 .781164991 1.200157314 .814531498 1.220002848 1.095269019

.000536331 .000563893 .000569278

.14636184 .14913536 .14913536

.637828308 .789734004 .677855955

.000603067

.15179071

.804428656

.000600093 .000603226 .000663643 .00067586 .000654056

.15179071 .15179071 .1576238 .1576238 .1576238

1.366036168 1.12982242 .803989647 .878996553 .906134056 (continued on next page)

ARTICLE IN PRESS R.P.F. LINDBLOM ET AL.

6

Table 1. (continued) Probe set ID

Gene symbol

15249846

KCNG1

15269030 15236387 15223644 15301767 15266293

TMC7 — MFAP3L DHODH NUDT12

15326964 15244064 15325447 15183222

— LMBRD2 — AGPAT4

15248312

LOC100154727

15256661 15307966

EPOR LOC100153328

15313482 15199584

— PCSK6

15201338 15203173

PRKCQ FAM208B

15236408

B3GALT1

15275546

INPP4A

15276692 15283549 15283551 15336315

— ENV ENV FUNDC1

Gene description

P-value

FDR

Fold change HG versus NG

Potassium voltage-gated channel, subfamily G, member 1 Transmembrane Channel-Like 71 — Microfibrillar-associated protein 3-like — Nudix (nucleoside diphosphate linked moiety X)-type motif 12 — LMBR1 domain containing 2 — 1-Acylglycerol-3-phosphate O-acyltransferase 4 LAMP family protein C20orf103 homolog Erythropoietin receptor Butyrophilin subfamily 1 member A1like — Proprotein convertase subtilisin/kexin type 6 Protein kinase C, Theta Family with sequence similarity 208, Member B UDP-Gal:betaGlcNAc beta 1,3galactosyltransferase, polypeptide 1 Inositol polyphosphate-4-phosphatase, type I — — — FUN14 domain-containing protein 1-like

.000638381

.1576238

1.394515567

.000656139 .000669157 .000692839 .000685492 .00071706

.1576238 .1576238 .15773643 .15773643 .15945408

.636356404 1.532720668 .870372896 1.158196788 .903103007

.000709501 .00073134 .000810242 .000869546

.15945408 .1607604 .17608039 .18105118

.817357616 1.138608772 1.089686608 .734977792

.000857613

.18105118

1.188457706

.000880299 .000880452

.18105118 .18105118

.845037794 1.118432321

.000864701 .000968263

.18105118 .18153975

.899275874 .605241751

.000910459 .000914533

.18153975 .18153975

.647436654 1.145079286

.000903867

.18153975

1.163947391

.000952391

.18153975

1.107341716

.000923657 .000967974 .000967974 .000932707

.18153975 .18153975 .18153975 .18153975

.819162176 1.092459084 1.092459084 .727789876

The expression of more than 19,000 transcripts was studied. Analysis revealed significant differences in expression of 102 transcripts at P-value less than .001. The false discovery rate (FDR) corresponds to the adjusted P-values correspond. Fold change (FC) more than 1.0 indicates upregulation in hyperglycemic animals and FC less than 1.0 indicates higher expression in normoglycemic animals.

analyzed using the DAVID functional annotation tool.17,18 This revealed significant enrichment for several biological pathways, for instance membrane processes (P = 2.1 × 10−6) and ion transport (P = .0012), as already suggested by the presence of several solute carrier genes among the most differentially regulated genes (Table 2 and Supplementary Table S2). Also, glycosylation site (P = .0015) and glycoproteins (P = .0026) were processes that were enriched among the identified genes, which verified that glucose-related pathways were differentially activated between the conditions (Table 2 and Supplementary Table S2). Heat maps created by hierarchical clustering of the normalized gene expression for each transcript were created to further visualize the data (Fig 2).

Other biological processes that were identified among the identified genes using the enrichment analysis were regulation of actin cytoskeleton organization (P = .0021) and actin filament-based process (P = .0023), suggesting that also intracellular or structural processes were differentially activated (Table 2 and Supplementary Table S2).

qRT-PCR Confirmation of Selected Targets To verify the results from the microarray analysis, expression of SLC5A11 and SLCO1A2 was assessed using qRT-PCR. It is known that expression of house-keeping genes can be affected by cerebral ischemia.19 We therefore analyzed the microarray expression of several of the commonly used house-keeping genes to see whether they

ARTICLE IN PRESS HYPERGLYCEMIC CEREBRAL EXPRESSION PATHWAYS

7

Figure 1. Expression of several of the most strongly differentially regulated transcripts. Solute carrier family 5, member 11 (SLC5A11) was higher expressed in pigs in the NG (A), whereas 3-hydroxybutyrate dehydrogenase type 1 (BDH1) levels were higher in animals in the HG (B). Solute carrier organic anion transporter family, member 1A2 (SLCO1A2), was alike SLC5A11 upregulated in NG compared with HG (C). Contactin 3 (CNTN3) was the transcript that differed most strongly between the groups and was expressed higher in HG (D). Two microRNAs were among the top genes, mir-664 with higher expression in NG (E) and mir9-2 with higher expression in HG (F). Abbreviations: HG, hyperglycemic group; NG, normoglycemic group.

were affected by our experimental conditions. This suggested significant differences in expression of PPIA and YWHAZ, P-value less than .05, as well as borderline significance of HPRT1 (P = .059) between the experimental conditions (Table 3). Based on this, we chose RPL13A (ribosomal protein L13a) and glyceraldehyde 3-phosphate dehydrogenase. The expression data obtained from qRTPCR demonstrated a similar pattern as that observed from the microarray expression, with higher expression of both SLCO1A2 and SLC5A11 in the normoglycemic than in the hyperglycemic group (Fig 3). However, the P-value was just below significance for both targets: P = .055 for SLCO1A2 and P = .14 for SLC5A11 (Fig 3).

Discussion In the current study, significant differences in cerebral metabolic gene regulation were identified in pigs after hyperglycemic compared with normoglycemic CA. The prime objective of this study was to identify early transcriptional reactions in the brain to increase the understanding of the mechanisms behind exacerbated neuronal damage seen after hyperglycemic CA. To this end, microarray expressional profiling is a powerful tool. Analysis of large microarray data sets is complex and includes the problem with multiple testing. We therefore chose P-value less than .001 as an initial alpha level and

ARTICLE IN PRESS R.P.F. LINDBLOM ET AL.

8

Table 2. The most enriched biological pathways among the strongest differentially regulated genes Category

Term

Count

P-value

Genes

List total

Pop hits

Pop total

SP_PIR_KEYWORDS UP_SEQ_FEATURE SP_PIR_KEYWORDS SP_PIR_KEYWORDS UP_SEQ_FEATURE

Membrane Transmembrane region Transmembrane Ion transport Glycosylation site:N-linked (GlcNAc,,,) GO:0032956~regulation of actin cytoskeleton organization GO:0032970~regulation of actin filament-based process Glycoprotein

36 28 28 8 23

.00000214 .00016000 .00017885 .00123981 .00153535

See Table 3 See Table 3 See Table 3 See Table 3 See Table 3

56 56 56 56 56

6256 4911 4973 578 4129

19235 19113 19235 19235 19113

4

.00211980

See Table 3

40

89

13528

4

.00233018

See Table 3

40

92

13528

23

.00257764

See Table 3

56

4318

19235

GOTERM_BP_FAT GOTERM_BP_FAT SP_PIR_KEYWORDS

The term “Category” refers to the database where information regarding the transcripts was collected. The term “Term” refers to the biological process the identified transcripts are classified to belong to. The “Count” column is how many transcripts in the uploaded list belong to this category. The P-value is a modified Fisher exact P-value (here 8 decimals shown). The genes in each cluster are shown in supplementary Table 1. The “List total” column describes how many of the transcripts there are in the list. In most databases, 56 of the total of 102 transcripts could be classified into any category, but in the” GOTERM_BP_FAT” database a smaller number of transcripts are categorized (40). This also differs in the “Pop(ulation) Hits” column, which describes how many genes from this category there are in the organism in total, which also varies between the databases or categories. The “Pop(ulation) Total” column describes how many genes there are in total in the organism that are categorized in the respective database or category used.

corrected for multiple testing as described. The cutoff FDR less than .20 was set based on the short time frame of the experiment and the relatively mild intervention in 2 randomized and strictly controlled groups. Expression of SLC5A11, the 11th member of the sodium/ glucose transport family SLC5, was significantly downregulated in HG compared with animals in the NG. These transporters contrast to plasma membrane cotransporters (GLUTs), as the SLC5 transporters exert active transport of molecules including sugars along

the sodium electrochemical gradient.20 SLC5A11 is particularly responsible for transporting the glucose 6-phosphatase derivate D-chiro-inositol.21 The SLC5 proteins play a role in preserving intracellular glycemic levels in case of substrate deficiency, and thereby contribute to endogenous neuroprotection during ischemia.22 The current finding that SLC5A11 was down-regulated in HG group may in this context lend support to our previous results that suggest aggravated brain injury in the HG group.11

Table 3. Commonly used house-keeping genes and their performance in the current model Gene symbol PPIA YWHAZ

HPRT1 B2M OAZ1 GAPDH GUSB RPL13A

Gene description

Average expression HG

Average expression NG

P-value

FDR

Peptidylprolyl isomerase A (cyclophilin A) Tyrosine 3-monooxygenase/ tryptophan 5-monooxygenase activation protein, zeta polypeptide Hypoxanthine phosphoribosyltransferase 1 Beta-2-microglobulin Ornithine decarboxylase antizyme 1 Glyceraldehyde-3-phosphate dehydrogenase Glucuronidase, beta Ribosomal protein L13a

12.802216

12.72533667

.01861755

.43972441

12.572802

12.44946167

.03937599

.50633654

10.160888

10.006867

.05953402

.53728949

11.30297 11.339868

11.36565333 11.31777167

.25816238 .2803523

.72521403 .74273995

13.34372

13.32460333

.48489109

.85808087

8.5473995 8.372405

.61706279 .6928749

.91440245 .93712202

8.6020448 8.4270888

Peptidylprolyl isomerase A/cyclophilin A (PPIA), tyrosine 3-monooxygenase/tryptophan 5-monooxygenase activation protein, zeta polypeptide (YWHAZ), and hypoxanthine phosphoribosyltransferase 1 (HPRT1) vary significantly in expression between the HG and the NG animals and are therefore unsuitable to use.

ARTICLE IN PRESS HYPERGLYCEMIC CEREBRAL EXPRESSION PATHWAYS

9

Figure 2. The pattern of expression for the 102 genes with P-value less than .001. A heat map illustrating the pattern of expression for the 102 genes with P-value less than .001 was constructed, with hierarchical clustering depicting the internal relationships between the genes. Red color indicates upregulation and green color downregulation. The more intense the color, the stronger the difference. Transcript ID and gene symbols are in the right column. (Color version of figure is available online.)

ARTICLE IN PRESS 10

R.P.F. LINDBLOM ET AL.

Figure 3. Expression of SLCO1A2 and SLC5A11 assessed with qRT-PCR. Expression of SLCO1A2 and SLC5A11 was quantified using qRT-PCR. This demonstrated a suggested higher expression of both SLCO1A2 (A) and SLC5A11 (B) in the normoglycemic as compared with the hyperglycemic group. However, the P-value was just below significance for both targets, P = .055 for SLCO1A2 and P = .15 for SLC5A11. Abbreviation: qRT-PCR, quantitative reverse-transcriptase polymerase chain reaction.

BDH1 was significantly upregulated in hyperglycemic compared with normoglycemic animals. The encoded mitochondrial enzyme D-3-hydroxybutyrate dehydrogenase oxidizes β-hydroxybutyrate to acetoacetate, a ketone body that can be absorbed by brain. Ketone bodies decrease the need for glucose during starvation, and when the relationship between cerebral blood flow and glucose metabolism is mismatched,23 the brain adapts to ketone metabolism.24 This may provide ischemic neuroprotection,25 as supported by experimental studies where intravenous β-hydroxybutyrate decreased cerebral infarct volume.26 Being upregulated in the HG group, this mechanism would theoretically ameliorate the ischemic injury. Expression of SLCO1A2, a gene that encodes the organic anion transporting polypeptide (OATP), was significantly downregulated in hyperglycemic compared with normoglycemic animals. The OATPs participate in the regulation of solute exchange between compartments. OATP1A2, encoded by SLCO1A2, is found especially in frontal cortex, hippocampus, but also in capillary endothelial cells.27 Besides substrates, such as substance P and vasoactive intestinal peptide, SLCO1A2 has a very high transport activity for dehydroepiandrosterone sulfate. Dehydroepiandrosterone sulfate in turn inhibits the GABAA receptors,28 which may indirectly provide neuroprotection. Referring to this mechanism, downregulation of SLCO1A2 would theoretically contribute to the ischemic damage. The gene that differed most significantly between the groups was CNTN3, contactin-3, an immunoglobulin superfamily member, expressed uniquely in subsets of neurons, for instance in the cerebral cortex.29 The contactins are involved in developmental processes like neural cell migration, neurite overgrowth, and axonal guidance.30 Members of the contactin family have been linked to autism spectrum disease,31 but otherwise little is known about CNTN3 in human disease.

Two differentially regulated microRNAs were also among the findings, the brain-specific miR-9 and miR-664. These were significantly up- and downregulated, respectively, in hyperglycemic compared with normoglycemic pigs. Although the complex field of microRNA responses is beyond the scope of this study, it is interesting that altered regulation of microRNAs has been demonstrated in early phase of reperfusion,32 compatible with the frame of the present study. In addition to the microarray analysis, qRT-PCR was performed to measure mRNA levels of SLCO1A2 and SLC5A11. The expression patterns were like the microarray findings, that is, suggestive of relatively higher expression in the normoglycemic group. With P-values of .055 and .14 for SLCO1A2 and SLC5A11, respectively, the discrepancies did not reach the stipulated alpha level, but the qRT-PCR data tend to align with rather than contradict the microarray data. We cannot exclude that factors such as freezing and thawing could have influenced the data distribution, and for this reason, a larger number of samples may be required in future analyses. The study is based on early gene expression data, and further elaboration warrants prolonged experiments, proteomic analyses, and ultimately outcome end points. However, among the 19,000 transcripts analyzed, there was an overt overrepresentation among the most changed transcripts that were related to glucose, organic transport, and ketone metabolism. Furthermore, this pattern appeared in response to a brief and moderate difference of 4.5 mM glucose in rigorously controlled experiments. The knowledge is scarce on how initial glycemic control affects the course of CA, but the present data show that immediate transcriptional changes follow brief and limited glycemic alterations. The literature provides theoretical connections between neuroprotection and the affected genes, which makes them highly interesting targets for further studies.

ARTICLE IN PRESS HYPERGLYCEMIC CEREBRAL EXPRESSION PATHWAYS

Conclusions Hyperglycemia, compared with normoglycemia, during CA leads to differential early gene expression associated with glucose metabolism or transport, but also other mechanisms of possible pathogenic importance. Whether the expression differences represent compensatory mechanisms or early steps in a pathologic cascade cannot be deduced from this study. Acknowledgments: Microarray analysis and expression analysis was performed by Array and Analysis Facility, Science for Life laboratory at Uppsala biomedical Centre (BMC), Husargatan 3, SE-751 23 Uppsala, Sweden. We thank Malin Olsson for technical assistance.

Appendix: Supplementary Material Supplementary data to this article can be found online at doi:10.1016/j.jstrokecerebrovasdis.2017.11.036.

References 1. Nolan JP, Cariou A. Therapeutic hypothermia and coronary angiography are mandatory after out-of-hospital cardiac arrest: yes. Intensive Care Med 2014;40:1027-1029. doi:10.1007/s00134-014-3327-3. 2. Koenig MA. Brain resuscitation and prognosis after cardiac arrest. Crit Care Clin 2014;30:765-783. doi:10.1016/ j.ccc.2014.06.007. 3. Calle PA, Buylaert WA, Vanhaute OA. Glycemia in the post-resuscitation period. The Cerebral Resuscitation Study Group. Resuscitation 1989;17(Suppl):S181-S188, discussion S199-206. 4. Nielsen N, Sunde K, Hovdenes J, et al. Adverse events and their relation to mortality in out-of-hospital cardiac arrest patients treated with therapeutic hypothermia. Crit Care Med 2011;39:57-64. doi:10.1097/CCM.0b013e3181 fa4301. [Published Online First: 2010/10/21]. 5. Beiser DG, Carr GE, Edelson DP, et al. Derangements in blood glucose following initial resuscitation from in-hospital cardiac arrest: a report from the national registry of cardiopulmonary resuscitation. Resuscitation 2009;80:624-630. doi:10.1016/j.resuscitation.2009.02.011. 6. Sonneville R, Vanhorebeek I, den Hertog HM, et al. Critical illness-induced dysglycemia and the brain. Intensive Care Med 2014;doi:10.1007/s00134-014-3577-0. 7. Allen BS, Buckberg GD. Studies of isolated global brain ischaemia: I. Overview of irreversible brain injury and evolution of a new concept—redefining the time of brain death. Eur J Cardiothorac Surg 2012;41:1132-1137. doi:10 .1093/ejcts/ezr315. [Published Online First: 2012/03/09]. 8. Sutherland BA, Minnerup J, Balami JS, et al. Neuroprotection for ischaemic stroke: translation from the bench to the bedside. Int J Stroke 2012;7:407-418. doi:10.1111/j.1747-4949.2012.00770.x. 9. Kleikers PW, Wingler K, Hermans JJ, et al. NADPH oxidases as a source of oxidative stress and molecular target in ischemia/reperfusion injury. J Mol Med 2012;90:1391-1406. doi:10.1007/s00109-012-0963-3. 10. Bemeur C, Ste-Marie L, Montgomery J. Increased oxidative stress during hyperglycemic cerebral ischemia. Neurochem Int 2007;50:890-904. doi:10.1016/j.neuint.2007.03.002.

11 11. Molnar M, Bergquist M, Larsson A, et al. Hyperglycaemia increases S100beta after short experimental cardiac arrest. Acta Anaesthesiol Scand 2014;58:106-113. doi:10.1111/ aas.12209. 12. Irizarry RA, Hobbs B, Collin F, et al. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 2003;4:249-264. doi:10.1093/biostatistics/4.2.249. 13. Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection. Proc Natl Acad Sci USA 2001;98:31-36. doi:10.1073/pnas.011404098. 14. Smyth GK. Linear models and empirical Bayes methods for assessing differential expression in microarray experiments. Stat Appl Genet Mol Biol 2004;3:Article 3.doi:10.2202/1544-6115.1027. 15. Smyth GK. In: Gerntleman R, Carey V, Dudoit S, et al., eds. Bioinformatics and computational biology solutions using R and bioconductor. New York: Springer, 2005. 16. Benjamini Y, Hochberg Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J R Stat Soc Series B Stat Methodol 1995;289-300. 17. da Huang W, Sherman BT, Lempicki RA. Systematic and integrative analysis of large gene lists using DAVID bioinformatics resources. Nat Protoc 2009;4:44-57. doi:10.1038/nprot.2008.211. 18. Dennis G Jr, Sherman BT, Hosack DA, et al. DAVID: database for annotation, visualization, and integrated discovery. Genome Biol 2003;4:P3. [Published Online First: 2003/05/08]. 19. Gubern C, Hurtado O, Rodriguez R, et al. Validation of housekeeping genes for quantitative real-time PCR in in-vivo and in-vitro models of cerebral ischaemia. BMC Mol Biol 2009;10:57. doi:10.1186/1471-2199-10-57. [Published Online First: 2009/06/18]. 20. Wright EM, Loo DD, Hirayama BA. Biology of human sodium glucose transporters. Physiol Rev 2011;91:733-794. doi:10.1152/physrev.00055.2009. 21. Tsai LJ, Hsiao SH, Tsai LM, et al. The sodium-dependent glucose cotransporter SLC5A11 as an autoimmune modifier gene in SLE. Tissue Antigens 2008;71:114-126. doi:10.1111/j.1399-0039.2007.00975.x. 22. Yu AS, Hirayama BA, Timbol G, et al. Regional distribution of SGLT activity in rat brain in vivo. Am J Physiol Cell Physiol 2013;304:C240-C247. doi:10.1152/ ajpcell.00317.2012. 23. Pulsinelli WA, Levy DE, Duffy TE. Regional cerebral blood flow and glucose metabolism following transient forebrain ischemia. Ann Neurol 1982;11:499-502. doi:10.1002/ ana.410110510. 24. McIlwain H. Metabolic adaptation in the brain. Nature 1970;226:803-806. 25. Prins ML. Cerebral metabolic adaptation and ketone metabolism after brain injury. J Cereb Blood Flow Metab 2008;28:1-16. doi:10.1038/sj.jcbfm.9600543. 26. Suzuki M, Suzuki M, Sato K, et al. Effect of betahydroxybutyrate, a cerebral function improving agent, on cerebral hypoxia, anoxia and ischemia in mice and rats. Jpn J Pharmacol 2001;87:143-150. 27. Gao B, Vavricka SR, Meier PJ, et al. Differential cellular expression of organic anion transporting peptides OATP1A2 and OATP2B1 in the human retina and brain: implications for carrier-mediated transport of neuropeptides and neurosteroids in the CNS. Pflugers Arch 2014;doi:10.1007/s00424-014-1596-x.

ARTICLE IN PRESS 12 28. Maninger N, Wolkowitz OM, Reus VI, et al. Neurobiological and neuropsychiatric effects of dehydroepiandrosterone (DHEA) and DHEA sulfate (DHEAS). Front Neuroendocrinol 2009;30:65-91. doi:10.1016/j.yfrne.2008.11.002. 29. Yoshihara Y, Kawasaki M, Tani A, et al. BIG-1: a new TAG-1/F3-related member of the immunoglobulin superfamily with neurite outgrowth-promoting activity. Neuron 1994;13:415-426. 30. Mohebiany AN, Harroch S, Bouyain S. New insights into the roles of the contactin cell adhesion molecules

R.P.F. LINDBLOM ET AL. in neural development. Adv Neurobiol 2014;8:165194. 31. Chen J, Yu S, Fu Y, et al. Synaptic proteins and receptors defects in autism spectrum disorders. Front Cell Neurosci 2014;8:276. doi:10.3389/fncel.2014.00276. [Published Online First: 2014/10/14]. 32. Dharap A, Bowen K, Place R, et al. Transient focal ischemia induces extensive temporal changes in rat cerebral microRNAome. J Cereb Blood Flow Metab 2009;29:675-687. doi:10.1038/jcbfm.2008 .157.