Sage Journals: Discover world-class research

Abstract

This study determined whether stroke and other types of insults produced a gene expression profile in blood that correlated with the presence of neuronal injury. Adult rats were subjected to ischemic stroke, intracerebral hemorrhage, status epilepticus, and insulin-induced hypoglycemia and compared with untouched, sham surgery, and hypoxia animals that had no brain injury. One day later, microarray analyses showed that 117 genes were upregulated and 80 genes were downregulated in mononuclear blood cells of the “injury” (n = 12) compared with the “no injury” (n = 9) animals. A second experiment examined the whole blood genomic response of adult rats after global ischemia and kainate seizures. Animals with no brain injury were compared with those with brain injury documented by TUNEL and PANT staining. One day later, microarray analyses showed that 37 genes were upregulated and 67 genes were downregulated in whole blood of the injury (n = 4) animals compared with the no-injury (n = 4) animals. Quantitative reverse transcription–polymerase chain reaction confirmed that the vesicular monoamine transporter-2 increased 2.3- and 1.6-fold in animals with severe and mild brain injury, respectively, compared with no-injury animals. Vascular tyrosine phosphatase-1 increased 2.0-fold after severe injury compared with no injury. The data support the hypothesis that there is a peripheral blood genomic response to neuronal injury, and that this blood response is associated with a specific blood mRNA gene expression profile that can be used as a marker of the neuronal damage.

Keywords

Microarray Neuronal injury markers White blood cells Ischemia Stroke Brain hemorrhage

An inflammatory, immune response occurs after all types of brain injury including ischemia, status epilepticus, hemorrhage, hypoglycemia, or trauma. Several clinical trials of stroke have targeted this inflammatory response (Furuya et al., 2001; Iadecola and Alexander, 2001). Although white blood cell functions have been investigated in brain injury, it was not known if gene expression of white blood cells might also provide an indicator of neuronal cell death. Using microarrays to assess the gene expression profiles of mononuclear blood cells, we found that stroke, seizures, hypoxia, and hypoglycemia produced gene expression profiles characteristic of each condition (Tang et al., 2001). The present study was designed to test whether there is a common blood gene expression profile that correlates with neuronal injury that occurs in all of these conditions.

There are many causes of neuronal injury. Cerebral infarction and hemorrhage can be imaged using computed tomography and magnetic resonance imaging. The diffuse neuronal cell death produced by global cerebral ischemia, status epilepticus, hypoglycemia, and degenerative diseases, however, cannot be imaged. This prompted measurements of neuronal and glial proteins in serum and CSF that are released from injured cells, including neuron-specific enolase, the glial S-100B protein, and myelin basic protein. Though increased serum levels of these proteins have been reported after stroke (Missler et al., 1997), cardiac arrest (Rosen et al., 1998), head trauma (Pleines et al., 2001), and cardiopulmonary bypass (Basile et al., 2001), their utility as injury markers for seizures and hypoglycemia-induced brain injury has been controversial (Buttner et al., 1999; DeGiorgio et al., 1999; Suzuki et al., 1999). After injury, however, many cytokines, chemokines, and other immuno-regulatory molecules are expressed by neurons, glia, blood vessels, and white cells (del Zoppo et al., 2000; Feuerstein et al., 1998). We postulated that there is a peripheral immune response to the occurrence of neuronal death but independent of the cause of the injury, and that this would produce a specific blood gene expression profile.

Two experiments addressed this hypothesis. First, the genomic expression of blood mononuclear cells in animals with brain injury (ischemia, hemorrhage, seizures, and hypoglycemia) was compared with that in animals without injury (sham surgery, untouched controls, and hypoxia). Second, the whole blood genomic expression in animals with neuronal death produced by global ischemia and kainate-induced seizures and documented by TUNEL or PANT staining was compared with that in animals with no neuronal death after global ischemia and kainate seizures. The data show blood gene expression profiles that correlate with neuronal injury in brain.

MATERIALS AND METHODS

Mononuclear cell experiment

This experiment was performed as previously described (Tang et al., 2001). Briefly, adult male Sprague-Dawley rats (Harlan, Indianapolis, IN, U.S.A.) were subjected to ischemic stroke, hemorrhagic stroke, kainate-induced seizures, insulin-induced hypoglycemia, hypoxia, sham surgery, or assigned as untouched controls. Twenty-four hours later, rats were killed and blood was drawn. Total RNA was isolated from mononuclear cells and surveyed with triplicate Affymetrix U34A arrays (Affymetrix, Santa Clara, CA, U.S.A.). Three separate samples per group were hybridized to three separate arrays (n = 21; Fig. 1).

FIG. 1.

Genes upregulated (red) and downregulated (green) in blood monocytes of animals with brain injury (left side—BH1, BH2, BH3, IG2, IG3, BI1, BI3, BI2, IG1, K1, K3) compared with animals without brain injury (right side—C2, H2, S1, S3, C3, S2, H1, H3, K2, C1). The plot shows the hierarchical clustering of 197 regulated genes (y-axis) from blood monocytes of 21 different rat samples (x-axis). Twenty-four hours after brain hemorrhage (BH), insulin-induced hypoglycemia (IG), brain ischemia (BI), kainate-induced seizures (K), 8% hypoxia for 6 h (H), sham surgery (S) or being assigned as untouched controls (C), adult rats were killed and mononuclear cells were separated. Total RNA was isolated and gene expression assessed with Affymetrix U34A microarrays (3 arrays/group). An “injury” group of animals that included BH, BI, K, and IG subjects was compared with a “no injury” group that included C, S, and H subjects. A nonparametric Wilcoxon-Mann-Whitney test was used to screen genes that are differentially expressed between injury samples and no-injury samples. A Benjamini and Hochberg false discovery rate of <0.3 was used as a significance threshold. For each of 197 genes that met the threshold, the raw expression data were normalized to the median value of 21 measurements if the median value was greater than 100. Hierarchical clustering was performed with GeneSpring software. The color bar indicates the normalized expression level. For genes with low expression values at the bottom of the figure, because the median of the 21 measurements was less than 100 and the expression data were normalized to 100, most appear as a green color. Genes in the black box are upregulated genes, some of which are shared between one no-injury sample (C3) and injury samples.

Whole blood experiments

RNA isolated from whole blood is from mononuclear cells and polymorphonuclear cells and may include RNA from red cells and platelets. Compared with mononuclear cells, using whole blood produces less technical variability, since no cell separation is required, and could be more advantageous for genomic studies. Two additional groups of animals were used for microarray studies of whole blood. Animals were subjected to either four-vessel global cerebral ischemia or they were injected intraperitoneally with kainic acid (KA). Twenty-four hours later, blood was drawn via cardiac puncture. Animals were then killed and their brains were removed, frozen, and processed for TUNEL and PANT staining. A total of 20 rats were used: 10 for kainate-induced seizures and 10 for global ischemia. Four RNA samples in each group were placed on Affymetrix U34A microarrays, whereas all samples were used for reverse transcription–polymerase chain reaction (RT-PCR) confirmation.

Global cerebral ischemia. A total of 21 Sprague-Dawley rats were used in the global ischemia study. The four-vessel occlusion method was used with modifications (Pulsinelli and Brierley, 1979).

Kainate-induced seizures. A total of 14 adult, male Sprague-Dawley rats were used. After a subcutaneous injection of 10 mg/kg KA (Sigma Chemical, St. Louis, MO, U.S.A.) dissolved in 0.9% sterile saline, animals were returned to their home cages with food and water available ad libitum. The onset of seizures usually occurred 30 minutes after KA injection, generally persisted for several hours, and stopped spontaneously. The seizures were monitored and scored according to Racine (1972).

RNA isolation. RNA from whole blood was isolated with TRIZOL LS Reagent (Invitrogen, Carlsbad, CA, U.S.A.) immediately after the blood draw. All RNA samples were purified further using an RNeasy mini kit (QIAGEN, Valencia, CA, U.S.A.). RNA was quantified using spectrophotometry and stored at −80°C for microarray and RT-PCR studies.

Detection of DNA strand breaks by nick end-labeling. The terminal deoxynucleotidyl transferase–mediated dUTP nick end-labeling (TUNEL) technique can detect double DNA strand breaks by enzymatically labeling the free 3′-OH termini with modified nucleotides. TUNEL staining was done with ApopTag Peroxidase In Situ Apoptosis Detection kit (Intergen Company, Purchase, NY, U.S.A.) according to the manufacturer's instructions.

Detection of single DNA strand breaks by nick-translation. PANT labeling was performed on fresh-frozen sections using a modified method (Chen et al., 1997).

The TUNEL- and PANT-positive cells were detected and counted using MCID software (Imaging Research, St. Catherines, Ontario, Canada). The numbers of cells per mm² were counted on at least four sections in at least eight fields from each subject through the neocortex, entorhinal cortex, and the hippocampus.

Microarray analysis

Sample labeling, hybridization, and image scanning were carried out as described in the Affymetrix Expression Analysis Technical Manual. At least 10 μg total RNA from each sample was used for microarray study. The Affymetrix rat U34A array contains more than 7,000 known genes and 1,000 expressed-sequence tags. Only genes that had relatively higher expression (called “present” on at least one array), however, were used for subsequent analysis, using GeneSpring software (Silicon Genetics, Redwood City, CA, U.S.A.).

Analysis of the mononuclear cell gene expression data. The ischemic stroke, hemorrhagic stroke, kainate-induced seizures, and insulin–glucose treatment samples were designated as “injury” samples (n = 12), whereas untouched controls, sham surgery, and hypoxia samples were designated as “no injury” (n = 9). A nonparametric Wilcoxon-Mann-Whitney test was used to screen genes that were differentially expressed between injury and no-injury samples. A Benjamini and Hochberg false discovery rate lower than 0.3 was used as a significance threshold (which means 30% of genes identified by this procedure are expected to pass the threshold by chance). This procedure provides a balance between discovery of significant genes and protection against false positives, and it is widely used for multiple-testing correction.

Analysis of the whole blood gene expression data. The rats subjected to kainate injections or to global ischemia were classified as having “injury” or “no injury” based on the results of TUNEL and PANT staining. No-injury animals had no TUNEL- or PANT-stained cells on any brain section. A nonparametric Wilcoxon-Mann-Whitney test was again used to screen genes differentially expressed between injury (n = 4) and no-injury samples (n = 4) and a Benjamini and Hochberg false discovery rate lower than 0.3 was used as a significance threshold. The four injury rats were further segregated into “severe injury” (n = 2) and “mild injury” (n = 2) based on the number of dead cells determined by TUNEL and PANT staining. Mild injury was defined as 1 to 5 TUNEL and PANT cells/mm² on average in the neocortex, entorhinal cortex, and hippocampus. Severe injury was defined as >5 cells/mm² on average in neocortex, entorhinal cortex, and hippocampus. The Significance Analysis For Microarrays (SAM) program was used to select the genes from whole blood that correlated with the degree of neuronal injury (Tusher et al., 2001).

Hierarchical clustering analyses were performed on the genes that passed the statistical thresholds. A standard correlation coefficient of 0.95 was used as the measure for significant statistical similarity. The branching behavior of the tree was controlled using a separation ratio setting of 0.5 and a minimum distance setting of 0.001.

Quantitative reverse transcription–polymerase chain reaction

Taqman RT-PCR was performed on two selected genes using the 5700 Sequence Detection System (Applied Biosystems, Foster City, CA, U.S.A.). All primers and probes were designed using Primer Express 2.0 (Applied Biosystems). A one-step RT-PCR was performed according to Taqman One-Step RT-PCR Master Mix Reagents Kit protocol (Applied Biosystems). Normalization to GAPDH was performed to account for variability in the initial concentration and quality of total RNA and in the conversion efficiency of the reverse transcription reaction.

RESULTS

Genomic responses of mononuclear blood cells to neuronal injury

There were a total of 5,084 genes expressed at relatively high levels in mononuclear white blood cells and that had at least one “present call” by the Affymetrix software. Of these, 197 genes were differentially expressed in the mononuclear white blood cells in the injury group compared with the no-injury group of animals. The focal ischemia produces death of neurons and glia, whereas intracerebral hemorrhage, prolonged kainate seizures, and prolonged hypoglycemia produce isolated neuronal cell death (Auer et al., 1984; Matsushita et al., 2000; Zhang et al., 1997). This contrasts to pure hypoxia (Simon, 1999) as well as sham controls and untouched controls that have no isolated cell death and have no TUNEL- or PANT-stained cells in brain (Chen et al., 1997). Therefore, the 197 genes may serve as potential candidate markers of neuronal injury in mononuclear blood cells. Among these 197 genes, 117 were upregulated and 80 were downregulated in the injury samples compared with the no-injury samples. Using the expression data of these regulated genes, a hierarchical cluster analysis discriminated the injury samples from no-injury samples (Fig. 1). The only two samples that did not seem to cluster with the appropriate groups were one kainate-treated animal (K2) and one untouched animal (C1; right two columns; note cluster dendrogram for these above). The expression pattern of the kainate K2 animal, however, resembled the pattern of the injury animals, and the expression pattern of the control C1 animal was similar to that of the no-injury animals. To confirm the results of hierarchical clustering, we adopted the approach of “leave-one-out” cross-validation, which takes one sample out and uses the remaining samples to build an algorithm and predict the category of the remaining sample (Golub et al., 1999). This approach successfully predicted the category of the 21 samples as either injury or no injury.

It is noteworthy that the majority of the 197 genes were regulated within a twofold range. Genes that are regulated more than twofold are listed in Table 1.

TABLE 1.

Genes that were identified as upregulated or downregulated more than two fold in the mononuclear cell experiments and in the whole blood experiments

		Systematic no.	Genbank no.	Fold change	Description
Mononuclear cell	Upregulated	AF023621_at	AF023621	2.7	Sortilin
		M55636_at	NM_013170	2.5	Heat-stable enterotoxin receptor/guanylyl cyclase
		L10336_at	L10336	2.3	Guanine nucleotide-releasing protein
		rc_AI136396_at	M69056	2	Farnesyltransferase beta subunit
		AF045464_s_at	NM_013215	2.3	Aflatoxin B1 aldehyde reductase
		rc_AA849036_at	NM_017090	2.4	Gucy1a3, guanylate cyclase 1, soluble, alpha 3
		rc_AI043796_s_at	NM_013031	2.1	S1c18a2, solute carrier family 18 A2 (vesicular monoamine transporter 2)
		L00603_at	NM_013031	2.1	Vesicular monoamine transporter
		U40790_at	NM_017269	2.2	Vascular protein tyrosine phosphatase-1 rDEP-1
	Downregulated	U17254_g_at	NM_024388	2.1	Immediate early gene transcription factor NGFI-B mRNA
		L41685mRNA_at	L41685	2.5	RATORFC (clone REM3) ORF
		U39875_at	NM_024139	2.8	EF-hand Ca²⁺-binding protein p22
Whole blood	Upregulated	U05014_g_at	U05014	5.4	PHAS-I
		M28259cds_at	M28259	2.8	Fibronectin
		U09631_at	NM_017238	2.2	VIP2, vasoactive intestinal peptide receptor
		L00603_at	NM_013031	2.0	Vesicular monoamine transporter
		J05132_s_at	J05132	2.1	3-Methylcholanthrene–inducible truncated UDP-glucuronosyltransferase
	Downregulated	X61295cds_s_at	X61295	2.8	L1 retroposon, ORF2 mRNA (partial)

There is variability in the patterns of gene expression within subjects for both groups. In particular, the K2 (kainate, second sample) animal had fewer upregulated genes (red) compared with the other subjects in the injury group. Since TUNEL and PANT staining were not performed on these animals, it is not known if this animal had a different degree of neuronal injury compared with the other subjects. It is also notable that one no-injury animal (C3) had a pattern that was different from the other no-injury animals in the group (Fig. 1). It shared some upregulated injury-group genes (some genes in the box, Fig. 1). Though the significance of these genes is unclear, they may be induced in blood in response to injury or stress that was not otherwise apparent in this control subject C3, possibly due to an unrecognized infection.

Whole blood experiments

Kainic acid–induced seizures and neuronal injury. Systemic injection of KA-induced clinical seizures in all animals studied. Of the 14 rats injected with KA, 10 animals survived the seizures and recovered, whereas 4 died after displaying continuous generalized convulsions for 2 to 3 h.

Among rats with seizure scores of 4 or 5, we found significant PANT- and TUNEL-positive cells within the amygdala and entorhinal cortex (Fig. 2, right panel) that accounted for almost half of the neuron population in that region, followed by smaller percentages of positive cells found in the thalamus and the hippocampal CA-3 region (not shown). We did not detect any TUNEL- or PANT-positive cells in rats with a seizure score of 3 or less (Fig. 2, left panel).

FIG. 2.

PANT staining is shown in rat brain after severe (right panel) kainate-induced seizures compared with no PANT staining after brief, mild seizures (left panel). Adult rats were administered kainic acid (10 mg/kg) intraperitoneally and the severity of their seizures was scored. They were allowed to survive for 24 h, at which time they were killed and their brains processed for both PANT and TUNEL staining followed by methyl green counterstaining. Animals with mild seizures (n = 4) and a seizure score of 3 or less had no detectable PANT- (left panel) or TUNEL (not shown)-stained cells in the entorhinal cortex (left panel) or any other brain regions. Animals that had severe seizures (n = 6) with a seizure score of 4 or more had many PANT- (right panel) and TUNEL (not shown)-stained nuclei in the amygdala and entorhinal cortex (right panel), as well as in the thalamus, hippocampal CA-3 region, and the cortex (not shown). The PANT and TUNEL staining indicate nuclear DNA damage and probably represent dead cells.

Transient global cerebral ischemia and neuronal injury. Of the 21 rats subjected to the four-vessel-occlusion global cerebral ischemia, 8 rats died during or shortly after the surgery. Three of the surviving 13 rats developed seizures and were excluded from the study. Three animals experienced respiratory depression unrelated to anesthesia after common carotid artery occlusion, requiring the ischemic time to be decreased below the 20-minute benchmark. Thus, the ischemic time for all surviving rats ranged from 7.5 to 20 minutes. This variable ischemic time, however, did not affect the experimental analysis because the goal of this was to correlate the differential gene expression profile and the severity of neuronal injury due to the global cerebral ischemia. Short durations of global ischemia did not produce any TUNEL- or PANT-stained cells anywhere in the brain. Moreover, long durations of global ischemia produced TUNEL- and PANT-stained cells in many brain regions at 24 h after reperfusion, the most numbers of cells being in the caudate-putamen, cortex, and less so in the hippocampus.

Genomic responses of whole blood to neuronal injury

Among 5,084 genes examined, 104 genes showed differential expression between injury and no-injury animals. Since the injury and no-injury animals were generally treated identically for each type of injury, it is likely that they would have similar levels of systemic stress. Therefore, we propose that these 104 genes represent whole blood markers of neuronal injury. Genes regulated greater than twofold in whole blood are listed at the bottom of Table 1. A hierarchical cluster analysis successfully classified 8 samples from whole blood into 2 clusters with one cluster consisting of the 4 injury samples and the other cluster consisting of 3 no-injury samples (Fig. 3). The remaining no-injury sample, although not automatically clustered, showed a similar expression profile to other no-injury samples (Fig. 3).

FIG. 3.

Genes upregulated (red) and downregulated (green) in whole blood of adult rats 24 h after global ischemia (I) or seizures (S) in a group of four animals that had brain “injury” compared with a group of four animals that had no brain injury—“no injury.” The plot shows hierarchical clustering of 104 regulated genes (y-axis) from whole blood of 8 different rats (x-axis). Rats were killed 24 h after global ischemia or kainate-induced seizures and their brains were processed for both PANT and TUNEL staining. Total RNA from whole blood was isolated and gene expression assessed using Affymetrix U34A microarrays. The brain samples were classified as injury or no injury. Injury-group animals (I2, S2, I1, S1) had multiple PANT- and TUNEL-stained nuclei in brain, including entorhinal cortex, hippocampus, and neocortex. No-injury animals (I3, S4, S3, I4) had no PANT- or TUNEL-stained cells in brain. A nonparametric Wilcoxon-Mann-Whitney test was used to screen genes that are differentially expressed between injury samples and no-injury samples. A Benjamini and Hochberg false discovery rate of <0.3 was used as a significance threshold. For each of 104 genes that met the threshold, the raw expression data were normalized to the median value of 8 measurements if the median value was greater than 100. Hierarchical clustering was performed with GeneSpring software. Genes that show similar expression patterns across different treatments cluster together. The color bar indicates normalized expression levels.

The next analysis determined whether there were genes that correlated with the degree of neuronal injury. A total of nine genes were significantly regulated as determined by the SAM program (Table 2). These genes showed the highest expression levels in severe-injury samples, slightly elevated expression in mild-injury samples, and lowest expression in no-injury samples (Fig. 4A). A hierarchical cluster analysis of these genes separated eight samples into two major clusters (Fig. 4B). One cluster consisted of two severe-injury samples, whereas the other cluster was further separated into two subclusters. One subcluster included two mild-injury samples, and the other subcluster consisted of three no-injury samples. Although the remaining no-injury sample was not automatically clustered, it showed similar expression profile to the other no-injury samples (Fig. 4B).

TABLE 2.

Genes with expression levels that were correlated with the degree of injury in the whole blood experiments as determined using SAM software

Systematic no.	Genbank no.	Description
L00603_at	NM_013031	Vesicular monoamine transporter
X63253cds_s_at	X63253	Serotonin transporter
rc_AI101103_s_at	NM_012663	Vamp2, vesicle-associated membrane protein (synaptobrevin 2)
S81353_s_at	S81353	Sulfated glycoprotein-1 (prosaposin)
L25785_at	NM_013043	TSC-22, transforming growth-factor beta–stimulated clone 22
J05132_s_at	J05132	Rat 3-methylcholanthrene–inducible truncated UDP-glucuronosyltransferase
rc_AI007857_at	NM_019387	HGF-regulated tyrosine kinase substrate
M28259cds_at	M28259	Fibronectin
rc_AA893280_at	BG673602	ESTs, moderately similar to adipose differentiation–related protein

SAM, significance of microarray analysis; HGF, hepatocyte growth factor; EST, expressed sequence tag.

FIG. 4.

Expression profiles of 9 genes in whole blood of rats that correlate with the degree of brain injury. Twenty-four hours after global ischemia (I) or kainate-induced seizures (S), adult rats were killed and their brains were processed for TUNEL and PANT staining. Total RNA from whole blood was isolated and gene expression was assessed with Affymetrix U34A microarrays. The brain samples were classified as showing “severe injury” (S1, S2), “mild injury” (I1, I2) or “no injury” (I3, I4, S4, S3) according to the results of TUNEL and PANT staining. A Significance of Microarray Analysis (SAM) program was used to identify 9 genes whose expression in whole blood correlated with the degree of brain injury. (A) Graphical representation of the expression values of the 9 genes that correlated with the severity of brain injury. The raw expression data were normalized to the median value of 8 measurements. (B) Hierarchical clustering of the 9 genes identified in whole blood that had low levels of expression in animals with no injury (I3, I4, S4, S3), somewhat higher levels of expression in animals with mild injury (I2, I1), and had the highest expression levels in animals with severe injury (S1, S2).

Many of the regulated genes shown in Tables 1 and 2 have potential brain- or vascular-related functions. For example, the vesicular monoamine transporter (Tables 1 and 2) accumulates cytosolic monoamines into synaptic vesicles and is expressed in dopamine neurons in the striatum (Miller et al., 1999). Its expression in striatal dopaminergic neurons may regulate susceptibility to neuronal damage (Wang et al., 1997). Considering the report that dopamine levels are significantly increased after brain trauma (Levitt et al., 1995), the elevation of vesicular monoamine transporter-2 (VMAT-2) may represent an adaptive mechanism of lymphocytes to decrease the systemic levels of dopamine after neuronal injury produced by stroke, seizures, and trauma.

Fibronectin, a glycoprotein that exists in blood plasma and extracellular matrix, plays a crucial role in a variety of morphogenetic processes such as cell adhesion, migration, and signal transduction (Owens and Cimino, 1982). Of interest is the report that plasma fibronectin–deficient mice show increased neuronal apoptosis and larger infarcts after focal cerebral ischemia (Sakai et al., 2001).

Prosaposin is a precursor of activator proteins for sphingolipid hydrolases (saposin) (O'Brien and Kishimoto, 1991). In human blood, leukocytes contain mainly saposin, whereas plasma contains mainly precursor forms, and platelets show many forms (Sano et al., 1989). It is reported that prosaposin protects against focal and global ischemia (Lu et al., 2000) and reduces cavity formation after stab wounds in rat brain (Hozumi et al., 1999).

Quantitative reverse transcription–polymerase chain reaction

Taqman quantitative RT-PCR was performed on 2 selected genes using 20 whole blood samples with each sample duplicated (total n = 40): severe-injury samples (n = 12), mild-injury samples (n = 6), no-injury samples (n = 22) (Fig. 5). The results show a 2.15-fold increase of VMAT-2 messenger RNA (mRNA) expression after 24 h in peripheral blood in both KA-induced seizure and global cerebral ischemia models. Rat density-enhanced phosphatase-1 (rDEP-1) mRNA levels showed similar increases in injured rats compared with uninjured, though the differences were less pronounced in global cerebral ischemia: 1.49- and 1.38-fold increase in KA-induced seizures and global ischemia, respectively. When the kainate and global ischemia data were combined, the expression level of VMAT-2 showed excellent correlation with the number of dead cells in brain: highest expression occurred with severe-injury samples, moderate expression with the mild-injury samples, and lowest expression in no-injury samples (Fig. 5). In contrast, the expression of rDEP-1 was also increased in the severe-injury samples but did not show upregulation in mild-injury compared with no-injury samples (Fig. 5).

FIG. 5.

Reverse transcription–polymerase chain reaction quantification of whole blood vesicular monoamine transporter (VMAT-2) RNA and rat density-enhanced phosphatase-1 (rDEP-1), a vascular tyrosine phosphatase 1 RNA. Adult rats were killed 24 h after global cerebral ischemia or kainate-induced seizures, whole blood RNA was isolated, and the brains were processed for TUNEL and PANT staining. Based on the TUNEL and PANT staining, animals were then divided into three groups, those showing “no injury” (n = 22 samples), “mild injury” (n = 6 samples) or “severe injury” (n = 12 samples). The normalized ratio for each gene in each group, plotted on the y-axis, was calculated as the average expression of the gene in each condition normalized to the GAPDH control for each sample. An ANOVA showed an effect of injury on expression for VMAT-2 (P < 0.0009) and rDEP-1 (P < 0.003), and a post hoc Fisher test showed that VMAT-2 was significantly greater in severe-injury (P < 0.0002) and mild-injury (P < 0.01) groups compared with the no-injury group. In addition, rDEP-1 was significantly greater in the severe-injury (P< 0.0005) compared with the no-injury group. **P < 0.001; *P < 0.01.

An ANOVA showed an effect of injury on expression for VMAT-2 (P < 0.0009) and rDEP-1 (P < 0.003), and a post hoc Fisher test showed that VMAT-2 was significantly greater in severe-injury (P < 0.0002) and mild-injury (P < 0.01) compared with the no-injury group. In addition, rDEP-1 was significantly greater in the severe-injury (P < 0.0005) compared with the no-injury group.

DISCUSSION

We previously reported that several disease models in rats including ischemia, hemorrhage, seizures, hypoglycemia, and hypoxia each produce unique, characteristic genomic expression patterns in blood (Tang et al., 2001). In contrast, the present study shows a common gene expression pattern in peripheral blood 1 d after various types of experimental neuronal injuries. This suggests that gene expression profiling of peripheral white blood cells could be used to assess the occurrence of neuronal injury associated with various diseases.

Several underlying mechanisms likely explain the genomic responses of peripheral white cells to neuronal injury. One mechanism likely involves the interaction between peripheral immune cells and dead cells in brain. There is an inflammatory, immune response after various brain injuries. The recruitment of peripheral immune cells through the blood–brain barrier and subsequent inflammatory response may exacerbate brain injury early and contribute to repair later (Raivich et al., 1999). Neutrophils appear in brain a few hours after focal ischemia due to middle cerebral artery occlusion (Kochanek and Hallenbeck, 1992). Both T-lymphocytes and macrophages infiltrate brain by 12 h after middle cerebral artery occlusion and persist for days (Schroeter et al., 1994). After global cerebral ischemia, resident microglial cells and macrophages proliferate throughout hippocampus (Liu et al., 2001). An increase of neutrophils was found in both perivascular and brain parenchymal regions after reperfusion (Bottiger et al., 1998). Excitotoxic lesions produced by injection of KA or N-methyl-d-aspartate recruited neutrophils, macrophages, and some lymphocytes around vessels and in the surrounding parenchyma, with levels peaking as early as 8 h after injection (Bolton and Perry, 1998). According to the clonal selection theory, each lymphocyte is genetically committed to express a unique antibody or receptor. Later, on encountering external antigens, the cells respond by clonal expansion and differentiation (Jerne, 1985). Therefore, it is plausible that various clones of lymphocytes will be selected and expanded after exposure to previously hidden antigens, and the changes in the cumulative transcriptional pattern of blood could reflect the specific genomic programs of selected clones that are responding to the neuronal injury.

In contrast, various signaling molecules related to the immune response are released into serum and CSF in response to neuronal injury (Fassbender et al., 1994; Jankowsky and Patterson, 2001). The induction of cytokine secretion may be the underlying mechanism for upregulation of rDEP-1 expression in the monocyte blood experiments. rDEP-1 is a homolog gene of human DEP-1/human protein tyrosine phosphate-η (Borges et al., 1996), which has inhibitory effects on the immune system and specifically inhibits activation of lymphocytes (Tangye et al., 1998). It is likely that the increased expression of rDEP-1 may partly be due to T-cell activation induced by a serum surge of cytokines caused by acute neuronal injury.

Another mechanism that may contribute to the genomic responses of white cells is the release of neurotransmitters after brain injury. Peripheral white blood cells possess neurotransmitter receptors and transporters (Amenta et al., 2001; Dorsam et al., 2000; Kohm and Sanders, 2000; Lombardi et al., 2001; Stefulj et al., 2000), which may play important roles in the immunomodulation by the CNS. For example, dopamine interacts directly with its D₂ and D₃ receptors on human T cells and helps the trafficking and extravasation of T cells across blood vessels and tissue barriers (Levite et al., 2001). Catecholamines can also trigger the secretion of interleukin-10 from monocytes through a β-adrenoreceptor–mediated pathway, which causes changes in leukocyte distribution and lymphocytic apoptosis (Stevenson et al., 2001). Our finding that two monoamine transporters, VMAT-2 and serotonin transporter, along with a vesicle-associated membrane protein, synaptobrevin 2, are upregulated in white cells underscores the possible role of neurotransmitter systems in linking CNS injury and peripheral white blood cell genomic responses.

It should be noted that many genes were not shared between the mononuclear and whole blood experiments. There are several explanations for this. One is that the statistical thresholds were not extremely stringent, and hence several false-positive genes would be expected in both groups. A separate biologic problem in these studies is that the expression profiles are the cumulative expression of all the cells used for each experiment. All subsets of lymphocytes and monocytes were included in the mononuclear cell study. For the whole blood study, the RNA was even more heterogeneous, being derived from all cells in blood. It is possible that only a small fraction of the white blood cells respond to neuronal injury, whereas other elements in the system do not respond, but their presence could obscure changes in one cell type. Therefore, surveying whole blood or even just mononuclear cells may not be sensitive enough for identifying patterns of diagnostic value. Genomic profiles of specific subtypes of white blood cells may be required.

With the explosion of sequence information and the advances in genomics and proteomics, it may be possible to diagnose and classify clinical diseases at the molecular level. Transcriptional profiling with complementary DNA or oligonucleotide microarrays has been used to diagnose and classify cancer samples, to evaluate the response to chemotherapy, and to predict prognosis (Alizadeh et al., 2000; Armstrong et al., 2002; Golub et al., 1999; Lu et al., 2000; Ramaswamy et al., 2001). The application of this powerful technology, however, to most neurologic diseases will likely be limited by the unavailability of brain tissue. The many receptors and signal transduction pathways in peripheral blood cells may provide a potential alternative or surrogate for transcriptional profiling of diseases of brain and other organs (Tang et al., 2001, 2002). The present finding that there are rodent blood genomic markers of neuronal injury supports this approach. Compared with using a single or several biochemical markers, using the cumulative pattern of many genes is probably less susceptible to the influences of environmental factors and concurrent diseases and therefore could be more specific. It remains to be seen, however, whether blood genomics will be successful in human disease diagnosis, prognosis, or treatment evaluation.

Footnotes

Abbreviations used

References

Alizadeh

Eisen

Davis

Lossos

Rosenwald

Boldrick

Sabet

Tran

Powell

Yang

Marti

Moore

Hudson

Jr Lu

Lewis

Tibshirani

Sherlock

Chan

Greiner

Weisenburger

Armitage

Warnke

Staudt

(2000) Distinct types of diffuse large B-cell lymphoma identified by gene expression profiling. Nature 403:503–511

Amenta

Bronzetti

Cantalamessa

El-Assouad

Felici

Ricci

Tayebati

(2001) Identification of dopamine plasma membrane and vesicular transporters in human peripheral blood lymphocytes. J Neuroimmunol 117:133–142

Armstrong

Staunton

Silverman

Pieters

den Boer

Minden

Sallan

Lander

Golub

Korsmeyer

(2002) MLL translocations specify a distinct gene expression profile that distinguishes a unique leukemia. Nat Genet 30:41–47

Auer

Wieloch

Olsson

Siesjo

(1984) The distribution of hypoglycemic brain damage. Acta Neuropathol (Berl) 64:177–191

Basile

Fusi

Conti

Paniccia

Trefoloni

Pracucci

Di Carlo

Noferi

Carbonetto

Pretelli

Calamai

Vaccari

Abbate

Inzitari

(2001) S-100 protein and neuron-specific enolase as markers of subclinical cerebral damage after cardiac surgery: Preliminary observation of a 6-month follow-up study. Eur Neurol 45:151–159

Bolton

Perry

(1998) Differential blood-brain barrier breakdown and leucocyte recruitment following excitotoxic lesions in juvenile and adult rats. Exp Neurol 154:231–240

Borges

Seifert

Grant

Hart

Disteche

Edelhoff

Solca

Lieberman

Lindner

Fischer

Lok

Bowen-Pope

(1996) Cloning and characterization of rat density-enhanced phosphatase-1, a protein tyrosine phosphatase expressed by vascular cells. Circ Res 79:570–580

Bottiger

Schmitz

Wiessner

Vogel

Hossmann

(1998) Neuronal stress response and neuronal cell damage after cardiocirculatory arrest in rats. J Cereb Blood Flow Metab 18:1077–1087

Buttner

Lack

Jager

Wunsche

Kuhn

Muller

Przuntek

Postert

(1999) Serum levels of neuron-specific enolase and s-100 protein after single tonic-clonic seizures. J Neurol 246:459–461

10.

Chen

Jin

Chen

Pei

Kawaguchi

Greenberg

Simon

(1997) Early detection of DNA strand breaks in the brain after transient focal ischemia: Implications for the role of DNA damage in apoptosis and neuronal cell death. J Neurochem 69:232–245

11.

DeGiorgio

Heck

Rabinowicz

Gott

Smith

Correale

(1999) Serum neuron-specific enolase in the major subtypes of status epilepticus. Neurology 52:746–749

12.

del Zoppo

Ginis

Hallenbeck

Iadecola

Wang

Feuerstein

(2000) Inflammation and stroke: Putative role for cytokines, adhesion molecules and iNOS in brain response to ischemia. Brain Pathol 10:95–112

13.

Dorsam

Voice

Kong

Goetzl

(2000) Vasoactive intestinal peptide mediation of development and functions of T lymphocytes. Ann N Y Acad Sci 921:79–91

14.

Fassbender

Rossol

Kammer

Daffertshofer

Wirth

Dollman

Hennerici

(1994) Proinflammatory cytokines in serum of patients with acute cerebral ischemia: Kinetics of secretion and relation to the extent of brain damage and outcome of disease. J Neurol Sci 122:135–139

15.

Feuerstein

Wang

Barone

(1998) The role of cytokines in the neuropathology of stroke and neurotrauma. Neuroimmunomodulation 5:143–159

16.

Furuya

Takeda

Azhar

McCarron

Chen

Ruetzler

Wolcott

DeGraba

Rothlein

Hugli

del Zoppo

Hallenbeck

(2001) Examination of several potential mechanisms for the negative outcome in a clinical stroke trial of enlimomab, a murine anti-human intercellular adhesion molecule-1 antibody: A bedside-to-bench study. Stroke 32:2665–2674

17.

Golub

Slonim

Tamayo

Huard

Gaasenbeek

Mesirov

Coller

Loh

Downing

Caligiuri

Bloomfield

Lander

(1999) Molecular classification of cancer: Class discovery and class prediction by gene expression monitoring. Science 286:531–537

18.

Hozumi

Hiraiwa

Inuzuka

Yoneoka

Akiyama

Tanaka

Kikugawa

Nakano

Tsuji

O'Brien

(1999) Administration of prosaposin ameliorates spatial learning disturbance and reduces cavity formation following stab wounds in rat brain. Neurosci Lett 267:73–76

19.

Iadecola

Alexander

(2001) Cerebral ischemia and inflammation. Curr Opin Neurol 14:89–94

20.

Jankowsky

Patterson

(2001) The role of cytokines and growth factors in seizures and their sequelae. Prog Neurobiol 63:125–149

21.

Jerne

(1985) The generative grammar of the immune system. Science 229:1057–1059

22.

Kochanek

Hallenbeck

(1992) Polymorphonuclear leukocytes and monocytes/macrophages in the pathogenesis of cerebral ischemia and stroke. Stroke 23:1367–1379

23.

Kohm

Sanders

(2000) Norepinephrine: A messenger from the brain to the immune system. Immunol Today 21:539–542

24.

Levite

Chowers

Ganor

Besser

Hershkovits

Cahalon

(2001) Dopamine interacts directly with its D₃ and D₂ receptors on normal human T cells, and activates β1 integrin function. Eur J Immunol 31:3504–3512

25.

Levitt

Cook

Simon

Williams

(1995) Biochemical markers of cerebral injury in patients with minor head trauma and ethanol intoxication. Acad Emerg Med 2:675–680

26.

Liu

Bartels

Sharp

(2001) Microglia/macrophages proliferate in striatum and neocortex but not in hippocampus after brief global ischemia that produces ischemic tolerance in gerbil brain. J Cereb Blood Flow Metab 21:361–373

27.

Lombardi

Dianzani

Miglio

Canonico

Fantozzi

(2001) Characterization of ionotropic glutamate receptors in human lymphocytes. Br J Pharmacol 133:936–944

28.

Otero

Hiraiwa

O'Brien

(2000) Neuroprotective effect of retro-inverso prosaptide D5 on focal cerebral ischemia in rat. Neuroreport 11:1791–1794

29.

Matsushita

Meng

Wang

Asahi

Moskowitz

(2000) Evidence for apoptosis after intercerebral hemorrhage in rat striatum. J Cereb Blood Flow Metab 20:396–404

30.

Mller

Erickson

Perez

Penland

Mash

Rye

Levey

(1999) Immunochemical analysis of vesicular monoamine transporter (VMAT2) protein in Parkinson's disease. Exp Neurol 156:138–148

31.

Missler

Wiesmann

Friedrich

Kaps

(1997) S-100 protein and neuron-specific enolase concentrations in blood as indicators of infarction volume and prognosis in acute ischemic stroke. Stroke 28:1956–1960

32.

O'Brien

Kishimoto

(1991) Saposin proteins: Structure, function, and role in human lysosomal storage disorders. Faseb J 5:301–308

33.

Owens

Cimino

(1982) Synthesis of fibronectin by the isolated perfused rat liver. Blood 59:1305–1309

34.

Pleines

Morganti-Kossmann

Rancan

Joller

Trentz

Kossmann

(2001) S-100 β reflects the extent of injury and outcome, whereas neuronal specific enolase is a better indicator of neuroinflammation in patients with severe traumatic brain injury. J Neurotrauma 18:491–498

35.

Pulsinelli

Brierley

(1979) A new model of bilateral hemispheric ischemia in the unanesthetized rat. Stroke 10:267–272

36.

Racine

(1972) Modification of seizure activity by electrical stimulation. Part II: Motor seizure. Electroencephalogr Clin Neurophysiol 32:281–294

37.

Raivich

Bohatschek

Kloss

Werner

Jones

Kreutzberg

(1999) Neuroglial activation repertoire in the injured brain: Graded response, molecular mechanisms and cues to physiological function. Brain Res Brain Res Rev 30:77–105

38.

Ramaswamy

Tamayo

Rifkin

Mukherjee

Yeang

Angelo

Ladd

Reich

Latulippe

Mesirov

Poggio

Gerald

Loda

Lander

Golub

(2001) Multiclass cancer diagnosis using tumor gene expression signatures. Proc Natl Acad Sci US A 98:15149–15154

39.

Rosen

Rosengren

Herlitz

Blomstrand

(1998) Increased serum levels of the S-100 protein are associated with hypoxic brain damage after cardiac arrest. Stroke 29:473–477

40.

Sakai

Johnson

Murozono

Sakai

Magnuson

Wieloch

Cronberg

Isshiki

Erickson

Fassler

(2001) Plasma fibronectin supports neuronal survival and reduces brain injury following transient focal cerebral ischemia but is not essential for skin-wound healing and hemostasis. Nat Med 7:324–330

41.

Sano

Hineno

Mizuno

Kondoh

Ueno

Kakimoto

Inui

(1989) Sphingolipid hydrolase activator proteins and their precursors. Biochem Biophys Res Commun 165:1191–1197

42.

Schroeter

Jander

Witte

Stoll

(1994) Local immune responses in the rat cerebral cortex after middle cerebral artery occlusion. J Neuroimmunol 55:195–203

43.

Simon

(1999) Hypoxia versus ischemia. Neurology 52:7–8

44.

Stefulj

Jernej

Cicin-Sain

Rinner

Schauenstein

(2000) mRNA expression of serotonin receptors in cells of the immune tissues of the rat. Brain Behav Immun 14:219–224

45.

Stevenson

Westermann

Liebmann

Hortner

Rinner

Felsner

Wolfler

Schauenstein

(2001) Prolonged α-adrenergic stimulation causes changes in leukocyte distribution and lymphocyte apoptosis in the rat. J Neuroimmunol 120:50–57

46.

Suzuki

Toribe

Goto

Kato

Futagi

(1999) Serum and CSF neuron-specific enolase in patients with West syndrome. Neurology 53:1761–1764

47.

Tang

Aronow

Sharp

(2001) Blood genomic responses differ after stroke, seizures, hypoglycemia, and hypoxia: Blood genomic fingerprints of disease. Ann Neurol 50:699–707

48.

Tang

Aronow

Wagner

Sharp

(2002) Genomic responses of the brain to ischemic stroke, intracerebral haemorrhage, kainate seizures, hypoglycemia, and hypoxia. Eur J Neurosci 15:1937–1952

49.

Tangye

Phillips

Lanier

de Vries

Aversa

(1998) CD148: A receptor-type protein tyrosine phosphatase involved in the regulation of human T cell activation. J Immunol 161:3249–3255

50.

Tusher

Tibshirani

Chu

(2001) Significance analysis of microarrays applied to the ionizing radiation response. Proc Natl Acad Sci U S A 98:5116–5121

51.

Wang

Gainetdinov

Fumagalli

Jones

Bock

Miller

Wightman

Caron

(1997) Knockout of the vesicular monoamine transporter 2 gene results in neonatal death and supersensitivity to cocaine and amphetamine. Neuron 19:1285–1296

52.

Zhang

Gelowitz

Lai

Boulton

(1997) Gradation of kainic acid-induced rat limbic seizures and expression of hippocampal heat shock protein-70. Eur J Neurosci 9:760–769

Blood Genomic Expression Profile for Neuronal Injury

Abstract

Keywords

MATERIALS AND METHODS

Mononuclear cell experiment

Whole blood experiments

Microarray analysis

Quantitative reverse transcription–polymerase chain reaction

RESULTS

Genomic responses of mononuclear blood cells to neuronal injury

Whole blood experiments

Genomic responses of whole blood to neuronal injury

Quantitative reverse transcription–polymerase chain reaction

DISCUSSION

Footnotes

Abbreviations used

References