The aim of this study was to compare gene expression profiles of leukocytes from blood (white blood cells; WBCs) and spleen harvested at an early time point after injury or sham injury in mice subjected to trauma/hemorrhage, burn injury, or lipopolysaccharide (LPS) infusion at three experimental sites. Groups of injured or LPS-infused animals and sham controls were killed at 2 h after injury and resuscitation, blood and spleen were harvested, and leukocyte populations were recovered after erythrocyte lysis. RNA was extracted from postlysis leukocyte populations. Complementary RNA was synthesized from each RNA sample and hybridized to microarrays. A large number (500–1,400) of genes were differentially expressed at the 2-h time point in injured or LPS-infused vs. sham animals. Thirteen of the differentially expressed genes in blood, and 46 in the spleen, were upregulated or downregulated in common among all three animal models and may represent a common, early transcriptional response to systemic inflammation from a variety of causes. The majority of these genes could be assigned to pathways involved in the immune response and cell death. The up- or downregulation of a cohort of 23 of these genes was validated by RT-PCR. This large-scale microarray analysis shows that, at the 2-h time point, there is marked alteration in leukocyte gene expression in three animal models of injury and inflammation. Although there is some commonality among the models, the majority of the differentially expressed genes appear to be uniquely associated with the type of injury and/or the inflammatory stimulus.
a central objective of the National Institute of General Medical Sciences (NIGMS)-funded Large-Scale Collaborative Research Program, “Inflammation and the Host Response to Injury,” is to compare gene expression profiles after traumatic injury, thermal injury, or infusion of bacterial lipopolysaccharide (LPS). We plan to make sequential observations by microarray analysis of messenger RNA (mRNA) abundance in circulating leukocytes from burn and trauma patients, and in murine models of thermal and traumatic injury, and to compare the effects of injury with those induced by LPS infusion in both normal volunteers and in uninjured mice. We hope to delineate differences and similarities between alterations in leukocyte gene expression after the two types of injury and compare patient responses with those of the mouse injury models. Similar comparisons will be made between the response to LPS infusion in mice and in human volunteers.
As an initial step in this project, we elected to compare leukocyte gene expression profiles in groups of mice undergoing the two forms of injury, or LPS infusion, with animals subjected to sham procedures at an early time point, 2 h postinjury. Observations at this early time were considered appropriate because of clinical and experimental evidence indicating that an inflammatory response is already initiated at this point after major traumatic or thermal injury and after exposure to LPS (4, 6, 7, 13, 14, 16–18). In addition, we wished to compare the gene expression profiles of circulating leukocytes with those from the spleen in the same animals to begin to address the phenomenon of compartmentalization of the immune system. A further goal of this study was to determine the effects of two standard red blood cell lysis buffers on the recovery of total leukocytes and major subsets from the blood of injured and LPS-infused animals. Although we had already established and tested protocols for anesthesia and blood drawing and processing at the three participating laboratories and had determined that the two lysis buffers yielded equivalent results in blood drawn from normal mice at the three sites (8), we wished to confirm the effectiveness of one or both methods of leukocyte isolation in injured mice, to establish a protocol for use in future studies. We demonstrate that the two lysis methods yield equivalent, but hardly ideal, results. We further show that while the three models of injury and inflammation induce a significant number of commonly expressed genes, the majority of differentially expressed genes are specific for the individual injury models.
MATERIALS AND METHODS
Male C57BL/6J mice were purchased from The Jackson Laboratory (Bar Harbor, ME). The mice were maintained in accredited animal facilities at the University of Michigan, the University of Alabama at Birmingham, and the Brigham and Women's Hospital, Harvard Medical School, in accordance with the guidelines of the National Institutes of Health and the respective universities. The study was approved by the Institutional Animal Use and Care Committees of each institution. The mice were acclimated for at least 1 wk before use in these experiments at 8 wk of age.
On the morning of the experiment, groups of 12 mice undergoing the injury protocols were given inhalation anesthesia with isofluorane and were then subjected to 25% total body surface area scald burn or trauma/hemorrhage (T/H). After burn injury, mice were resuscitated with 1 ml of normal saline, injected intraperitoneally (ip) (12). T/H consisted of laparotomy followed by withdrawal of sufficient blood from an arterial line to decrease and maintain mean arterial blood pressure at 35 mmHg for 90 min, after which these mice were resuscitated with Ringer lactate (4 times the shed blood volume) as previously described (2,15). Groups of 12 mice also underwent sham burn or sham T/H procedures under anesthesia. Mice receiving LPS, 10 ng of Escherichia coli 0113 (a generous gift of Dr. H. Shaw Warren, Boston, MA), in 200 μl of PBS or saline control by ip injection were not anesthetized. Two hours after all procedures, mice were anesthetized with isofluorane and were exsanguinated by cardiac puncture.
Blood was collected in syringes containing EDTA. A 20-μl aliquot of each sample was then used for a total and differential leukocyte count using the Hemavet instrument (Drew Scientific, Oxford, CT). The remainder of the blood sample was transferred to a 5-ml centrifuge tube, and an equal volume of PBS was added. The mixture was then centrifuged for 15 min at 100 g. The platelet-rich plasma was then carefully removed, and the remaining blood cell solution was added to 5 ml of commercial EL red blood cell (RBC) lysis buffer (Qiagen, Valencia, CA) or 5 ml of ACK lysis buffer (prepared by adding 8.29 g NH4Cl, 1 g KHCO3, and 0.1 ml of 0.5 M disodium EDTA to 1,000 ml of sterile pyrogen-free H2O; all purchased from Sigma Chemicals, St. Louis, MO). The tubes were mixed gently by inversion for 1 min and then centrifuged at 300 g for 10 min. The supernatant was removed, and the cell pellet was loosened by light agitation. PBS solution was added to make up the original volume, and another 20-μl sample was removed for Hemavet analysis. The sample tube was then centrifuged again for 10 min at 300 g, the supernatant was decanted, and the cell pellet was resuspended by vortexing before the addition of 1 ml of RLT buffer (Qiagen), supplemented with 10 μl β-mercaptoethanol/ml RLT buffer. The pellet was then homogenized in the RLT buffer by aspirating and injecting the sample through a 1-ml syringe, fitted with a 20-gauge needle. The homogenized cell suspension was transferred to a 2-ml RNase-free microcentrifuge tube, which was shipped in dry ice to the Department of Surgery at Washington University, St. Louis, for microarray hybridization and further analysis.
At the time of exsanguination, the spleens were removed under sterile conditions, and cell suspensions were prepared by mincing the tissue on sterile wire mesh gauze and collecting the cells in PBS. The same RBC lysis procedure with ACK buffer was then carried out with the spleen cell suspensions. After RBC lysis, the splenocytes were placed in RLT buffer, homogenized, and shipped in dry ice to St. Louis.
The Washington University site extracted cellular RNA and performed the microarray analyses. Total RNA was isolated using the manufacturer's (Qiagen) protocol. Isolated RNA was quantified using UV spectroscopy at 260 nm, and quality was assessed using an Agilent 2100 BioAnalyzer with the RNA NanoChip (Agilent, Andover, MA). RNA quality was judged on an arbitrary scale of 1–4, with 4 being the highest.
Usually 1–2 μg of total RNA were used to make double-stranded cDNA using a dT-T7 promoter primer, RT, and subsequent DNA polymerase in accordance with the manufacturer's directions (Affymetrix, Santa Clara, CA). Purified double-stranded cDNA was then utilized as a template to create biotinylated complementary RNA (cRNA). The labeled cRNA target was quantified and fragmented, and 15 μg were used for Affymetrix GeneChip hybridization. The complete target preparation protocol was performed according to the manufacturer's instructions (Affymetrix). Labeled targets were hybridized to Affymetrix MOE430A microarrays for 16 h at 45°C and washed according to Affymetrix standard protocols. The Affymetrix scanner output cell files were uploaded to the computational analysis group at Massachusetts General Hospital and at the University of Florida, Gainesville.
Real-Time PCR Protocol
One-microgram quantities of the spleen RNA preparations used for the microarrays were also subjected to RT-PCR using primers specific for a representative cohort of those genes identified by microarray analysis as differentially expressed in common among the three injury models and also classified as belonging to the “immune response” or “cell death” pathways identified by Ingenuity Pathway Analysis (IPA), described below. To validate further the microarray findings, RT-PCR was performed using primers for additional genes in the immune response and cell death pathways that were not shown to be differentially expressed in common by microarray analysis. The primer sets utilized for the above genes and for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) as a housekeeping gene control were obtained from Invitrogen (Carlsbad, CA) and can be found as Supplemental Materials (available at the Physiological Genomics web site).1 cDNA was synthesized using the SuperScript III RT System (Invitrogen) according to the manufacturer's instructions. RT-PCR was carried out on the Gene Amp 5700 instrument (PE Biosystems, Foster City, CA) with the quantitative PCR (QPCR) SYBR Green detection reagent (AB Gene, Rochester, NY) as previously reported. (11) Cycle threshold (CT) values for each gene were determined, and fold induction compared with GAPDH was calculated using the 1/Δ CT method. (11) RT-PCR results were recorded for each gene as fold change vs. sham for each of the three animal models. PCR products were evaluated by dissociation curves and by 1.0% agarose gel electrophoresis to confirm single amplicons of the appropriate molecular size.
The significance of the differences in leukocyte yields and subset percentages between the two lysis buffers at each experimental site was determined by a two-tailed unpaired t-test. The correlation between the results of microarray and RT-PCR studies of gene expression was determined by linear regression analysis. The Prism 4.0 statistical package (GraphPad Software, San Diego, CA) was utilized. P ≤ 0.05 was considered significant.
Gene Expression Analysis
Microarray data were initially processed, normalized, and modeled using the dChip v1.3 (8, 9) software package. Hybridizations deemed outliers by dChip were eliminated from subsequent statistical analysis.
For unsupervised analysis, probe sets with a coefficient of variation >0.5 across the entire data set were identified and subjected to hierarchical cluster analysis using average linkage clustering and 1 − Pearson correlation coefficient as the distance metric.
For each model of injury and LPS infusion, lists of probe sets differentially expressed between sham and injured or infused animals were then derived. These lists were also compared across the three models. Furthermore, lists were compared to determine which probes were differentially expressed in common between blood leukocyte (white blood cell; WBC) and splenocyte samples. For determining whether a probe set was differentially expressed between two groups, the following criteria were used: 1) P value ≤ 0.05 (for rejecting the hypothesis that the mean expression of the probe set in the two groups is equal), 2) absolute difference in mean expression across two groups is >100, 3) the lower 90% confidence bound on the fold change between two groups is >1.2, and 4) there is at least one gene array where the probe set was detected as “present” in either group (1).
To assess the statistical significance of the derived gene lists in the comparisons between the sham and injured groups, the four criteria for identifying differentially expressed genes were applied to the gene expression data sets with permuted array labels. In each of the 100 permutations (applied separately to each of the comparisons), the number of falsely identified genes was obtained. The false discovery rate (FDR) was defined as the median number of the genes falsely identified as differentially expressed in the 100 permutations divided by the number found using the four criteria. To determine the effect of the four criteria on the number of genes found to be differentially expressed, each criterion was changed, leaving the three remaining criteria the same, and the number of differentially expressed genes in each of the groups was determined.
Ingenuity Gene Network Analysis
To identify significant interactions and pathways from the experimental data sets, we used the IPA software (Ingenuity Systems, Mountain View, CA). The IPA uses the Ingenuity Pathways Knowledge Base (IPKB), which is a curated database of biological networks consisting of millions of individually modeled, peer reviewed pathway relationships. The following steps were performed to generate the resulting data. 1) Genes identified as significant were overlaid on the genomic network from the database and labeled as “Focus” genes, a gene subset having direct interaction(s) with other genes in the database. 2) Connections for each Focus gene were calculated by the percentage of its connections to other significant genes. The initiation and growth of the pathways proceeded from genes with the highest specificity of connections, and pathways of highly interconnected genes were identified by statistical likelihood using the equation given the number of genes in the genomic network, N, of which G are focus genes, for a pathway of s genes, f of which are focus genes. C(n,k) is the binomial coefficient. Pathways with a score >4, corresponding to P < 0.0001, were combined to form a composite network representing the underlying biology of the process. Gene symbols were colored according to the temporal behavior of their expression (red indicating increased and green indicating reduced expression). A more detailed description of this process can be found at the following web site: http://www.ingenuity.com.
Blood Lysis Results
The mean total and differential leukocyte counts obtained at each of the three laboratory sites for the two lysis buffers used are shown in Table 1. It is apparent that there is substantial loss of WBCs with both lysis techniques at two of the three laboratories. Differential counts show that a substantial (P < 0.05) fractional loss of neutrophils (polymorphonuclear neutrophils; PMNs) occurred in the burn-injured and LPS-infused animals. These animals, in contrast to those undergoing trauma/hemorrhage, had increased percentages of PMNs in the 2-h postinjury samples. These putatively activated PMNs may have been more susceptible to lysis. Therefore, with the exception of one skewed result after ACK lysis in the trauma/hemorrhage model, postlysis WBC populations were very similar across the three sites. There was no significant difference between the results obtained with the two lysis buffers, and RBC lysis was uniformly effective.
Analysis of RNA Yields
As shown in Table 2, all blood samples from all three institutions and all models of injury yielded adequate quantities of RNA for microarray analyses. Agilent profiles showed all RNA samples to be of good quality, grades 3 and 4, with grade 4 being the highest quality.
As had been the case for gene expression profiles from WBCs of unmanipulated mice of the same strain at the three laboratories (8), the method of RBC lysis had no significant effect on gene expression, and leukocyte samples obtained after both EL and ACK lysis of blood from individual experimental groups were analyzed together.
Figure 1 is an unsupervised cluster analysis of 2,460 probe sets with a coefficient of variation >0.5 in injured and sham animals from all three models. It is apparent that the results from the various injury and sham groups' splenocytes cluster together in almost all instances. However, there is more overlap in the blood (WBC) results, particularly in the LPS model. A list of the 2,460 probe sets with a coefficient of variation >0.5 is available as Supplemental Materials.
Burn Injury Model Analysis
Of a total of 23 samples hybridized to microarrays, 3 microarrays were deemed to be outliers by initial dChip evaluation and were removed from further analysis. The remaining 20 microarrays corresponded to 10 samples obtained from sham and 10 samples from injured mice. Of a total of 500 probe sets that were identified as differentially expressed in burn-injured vs. sham mice, 315 were upregulated and 185 were downregulated.
Of a total of 12 samples hybridized to microarrays, 1 sample was identified as an outlier and was removed from further analysis. The remaining 11 microarrays correspond to 5 samples obtained from sham and 6 from burn animals. Of a total of 428 probe sets identified as differentially expressed in burn vs. sham samples, 274 probe sets were upregulated and 154 downregulated.
Trauma/Hemorrhage Model Analysis
Of a total of 22 samples hybridized to microarrays, 11 were obtained from injured and 11 from sham mice. Of the 1,461 probe sets that were identified as differentially expressed between injured vs. sham mice, 1,293 were upregulated and 168 were downregulated.
Of a total of 11 samples hybridized to microarrays, 6 were obtained from sham and 5 from injured mice. Of the 1,283 probe sets identified as differentially expressed between injured vs. sham animals, 577 probes were upregulated and 706 downregulated.
LPS Model Analysis
After removal of 1 outlier microarray from further consideration, 22 microarrays were analyzed, 10 of which were hybridized from samples obtained from mice that received LPS and 12 of which were from mice that received saline control treatment. Six hundred fifty-six probes were differentially expressed in the groups receiving LPS compared with saline. Of these 656 probes, 339 were upregulated and 317 were downregulated.
Of a total of nine samples hybridized to microarrays, four were obtained from animals receiving LPS and five from those receiving control saline injection. Three hundred probe sets were identified as differentially expressed in mice receiving LPS vs. saline control animals. Of these 300 probe sets, 262 were upregulated and 38 were downregulated.
Comparisons of Models
Comparisons of microarrays performed on WBCs and splenic leukocyte samples from injured and sham-injured mice in the three models are illustrated in Table 3. In general, it is apparent that, of the probe sets differentially expressed in common in each model for both blood and spleen samples, the vast majority were up- or downregulated in common. However, in all three models, a majority of probe sets differentially expressed in blood or spleen were not differentially expressed in common.
Differential gene expression in relation to sham in blood and spleen in each animal model was compared with differential gene expression in each of the other two models as illustrated in Tables 4 and 6. While a considerable number of genes were up- and downregulated in common between any two models in both blood and spleen, there were also a sizeable number of probe sets that were up- or downregulated disparately between the two injury models and between the burn model and the LPS model. However, of the genes differentially expressed in common between the trauma-hemorrhage model and the LPS model, only one probe set was disparately regulated.
A list of the genes differentially expressed in injured mice vs. sham in each model in blood and spleen and those differentially expressed in common between any two of the three models in blood or spleen, along with a notation as to up- and downregulation, can be found as Supplemental Materials.
Figure 2 shows Venn diagrams of genes differentially expressed in WBCs and in splenocytes among the three models. It is apparent that only 28 genes were differentially expressed in WBCs in all three models. As shown in Table 5⇓, 13 probe sets were upregulated in common among the three models, and none was downregulated in common. Fifty-five genes were differentially expressed in the spleen in all three injury models. Table 7 shows that 37 probe sets were upregulated and 9 probe sets were downregulated in common among the three models.
The FDRs did not exceed 1% of the identified genes in any subcategory as shown in Table 8, confirming that the gene lists derived are truly differentially expressed rather that identified by chance. When the condition that the P value of the t-test comparing gene expressions between sham and injured groups be ≤0.05 was removed, the gene lists did not change significantly. When the condition that a gene was detected as present on at least one array was strengthened so the gene had to be detected as present on at least 50% of the arrays in at least one of the two compared groups, the resulting gene lists did not decrease by more than a few genes. When the condition that the lower 90% bound of the fold change exceed 1.2 was changed to have fold change exceed 1.5, the number of differentially expressed genes decreased on average by 38% in the comparisons of blood samples and by 26% in the comparisons of spleen samples. When the condition that the absolute difference in the gene expressions of the two compared groups exceed 100 was removed, the number of differentially expressed genes increased on average by 50% in comparisons of blood samples and by 100% in the comparisons of spleen samples. For example, in the intersection of the three injury models in blood samples, 36 genes were identified as differentially expressed as opposed to 28 when the original criteria were used. Similarly, in spleen samples, the intersection of the gene lists differentially expressed in the three injury models increased from 55 to 116.
The IPKB was used for functional analysis of the genes upregulated or downregulated in common among the three animal models for both blood and spleen.
IPA indicated that, of the 13 probe sets upregulated in common among the three models in blood, 6 probe sets were involved in the immune response pathway (BCL2L11, CCR1, CD14, HDC, IL1R2, and SLC11A1).
Among the 46 probe sets up- or downregulated in common in the spleen among the three models, 17 genes identified by these probe sets were found to be involved in the immune response pathway (CCND1, CCR2, CD14, CEBPB, CLECSF9, CXCL2, ETS2, GP49B, HDC, HHEX, IL6ST, MAIL, RGS1, SOCS3, STAT3, TGM2, and THBS1). Eighteen genes were involved in the cell death pathway (BCL2L11, BIRC2, CCND1, CCR2, CD14, CEBPB, ETS2, IL6ST, LMNB1, MAIL, NFKBIA, RNF19, SATB1, SERPINB9, SOCS3, STAT3, TGM2, and THBS1). It is apparent that several genes were identified as belonging to both pathways. Pathways of highly interconnected genes were identified by statistical likelihood, and pathways corresponding to a P value <0.0004 were combined to form a composite network representing the underlying biology. Because a larger number of genes were up- or downregulated in common in the spleen, the two major pathways potentially influenced by a majority of these genes are shown in Fig. 3. Genes were color coded according to their expression, with red indicating increased and green indicating decreased expression. Gene products are depicted in their anticipated locations in a representative cell. It is apparent that there was upregulation of gene expression for a chemokine (CXCL2) and thrombospondin ordinarily found in the extracellular space, for receptors for LPS and IL-6 located at the plasma membrane, for apoptosis-facilitating molecules and inhibitors of cytokine signaling usually present in the cytoplasm, and for several transcription factors in the nucleus with established roles in the activation of genes involved in inflammation.
To confirm the validity of the microarray findings with regard to genes up- or downregulated in common among all three models, the same spleen RNA samples used for the microarrays were subjected to RT-PCR. Again, spleen RNA was used because a greater number of genes were differentially expressed in common among the three models by microarray in the spleen than WBC from the blood. Primers were selected for 11 representative genes of the 17 identified by microarray as differentially expressed in common and classified as belonging to the immune response pathway by the above-mentioned IPA analysis. Similarly, RT-PCR was performed to look for expression of 12 representative genes of the 18 expressed in common by microarray and classified as belonging to the cell death pathway. The results are shown in Fig. 4. For purposes of comparison, fold change of mRNA abundance by RT-PCR in samples from injured vs. sham for the three models is plotted along with the Affymetrix analyses, where the mean difference in signal intensity of probe sets from injured vs. sham is also plotted as fold change. It is apparent that there is good correspondence between the RT-PCR findings and the microarray results. The validity of this finding is further supported by the analysis shown in Fig. 5, where regression plots are shown comparing the mean fold change in gene expression in the two pathways across all three models by Affymetrix analysis with that observed by RT-PCR for the same genes in the same mRNA specimens. The correlation coefficients of 0.7278 and 0.7985 clearly indicate a high degree of similarity in the results obtained by the two methodologies.
Finally, RT-PCRs were done with probes for 23 other genes in the immune response pathway and 23 genes in the cell death pathway that were not found to be significantly up- or downregulated in common among the three models by microarray analysis. Neither was there much evidence of a common pattern of expression of these genes across the three models by RT-PCR (r2 = 0.2262) (Fig. 6). Taken together, the results of these RT-PCR assays indicate that the microarray results in this experiment give an accurate picture of genes up- or downregulated in common among the three models.
It should be noted that for each animal model, the differentially expressed genes were determined by comparing injured or endotoxin-infused animals with simultaneously studied shams. This appeared to be the appropriate way to analyze the data, since the sham groups were treated quite differently at the three sites, e.g., shams underwent anesthesia and shaving of the dorsal skin, which was exposed to room temperature water in the case of the burn injury model; shams were subjected to anesthesia, exposure of the femoral vessels, and laparotomy in the case of the trauma/hemorrhage model. Thus the management of sham animals at each of the sites, while appropriate for the model of injury or inflammation being studied, nevertheless resulted in differences in sham gene expression. (Data are available as Supplemental Materials).
The results of this study of altered gene expression in leukocytes from blood and a major secondary lymphoid organ in two animal models of clinically relevant forms of injury demonstrate that, at an early postinjury time point, there is a substantial alteration in gene expression when samples of blood and splenic leukocytes obtained from injured animals are compared with those obtained from sham animals. Of the 22,690 probe sets represented on the Affymetrix MOE430-A microarray platform used in these experiments, between 500 and 1,400 genes were differentially expressed in the two injury models when compared with sham controls. A similar large alteration in gene expression was seen at 2 h after the infusion of LPS, a procedure often used as a surrogate for sepsis (14). Others have reported similar, extensive, early alterations in mRNA expression in liver and lungs early after LPS infusion (4, 18) and in the skin early after burn injury in mice (19).
At present, there is no consensus on the criteria that should be used for identifying differentially expressed genes in any particular situation. We selected a four-element criterion used in previous studies (1); this criterion yielded a list of genes that were highly statistically significant, in that <1% of the genes in the list would be expected to be false-positive genes. Changing two components of this criterion produced a negligible change in genes found, but two components were important. These components were concerned with absolute bounds on the ratio and with the absolute difference of expression levels. Making the ratio more stringent would decrease the number of genes by ∼25%, and making the absolute difference less stringent would significantly increase the number of genes. The sensitivity of differentially expressed gene determinations to the criteria used reinforces the need to make gene expression arrays available to the scientific community, as is done in this report, so that other investigators can use different criteria for analysis based on their own scientific questions.
The Ingenuity Pathways Knowledge Base was used for functional analysis of the genes upregulated or downregulated in common among the three animal models in leukocyte samples from both blood and spleen (13 in the blood and 46 in the spleen). Slightly less than one-half of the probe sets were found to be involved in the immune response pathway, and a roughly equal number were involved in the cell death pathway. These genes up- and downregulated in common among the three models of injury and sepsis very likely represent a common early transcriptional response to systemic proinflammatory signals from a variety of causes. The validity of the microarray data used to draw this conclusion was assessed by RT-PCR analysis of the same spleen RNA samples used for the microarrays from all three sites. Primers were selected for 11 of 17 genes up- or downregulated in common by microarray in the immune response pathway and for 12 of 18 genes regulated in common by microarray in the cell death pathway. RT-PCR results confirmed the microarray findings with a high coefficient of correlation.
In comparing the burn and trauma/hemorrhage forms of injury with one another, it is apparent that there were a number of genes that were up- and downregulated in common, 106 in the blood and 129 in the spleen. However, a substantial number, 69 in the blood and 28 in the spleen, were regulated in opposite fashion by the two forms of injury. Similar disparities were noted when the comparing burn injury model with the model of sepsis. Thus, at an early time point after injury or exposure to a bacterial product, there is some commonality of the transcriptional response. However, there remain major alterations in gene expression that are unique to each type of inflammatory insult.
The comparison between the samples from the blood and spleen in each of the two injury models and the LPS infusion model reveals a very substantial similarity in the up- or downregulation of the genes differentially expressed in the two cell populations. This supports the idea that there is some commonality in the gene expression profile between circulating blood leukocytes and those in a major secondary lymphoid organ at an early time point after injury or LPS infusion. However, it is also clear that a majority of the genes that were differentially expressed in each animal model were not differentially expressed in common in both blood and spleen, thus supporting the idea of compartmentalization of the early response to inflammatory stimuli.
The present study has the advantage that sufficient numbers of microarrays were performed on cRNA samples from both injured and sham animals to permit suitable statistical analysis. However, this experiment has focused on a single time point after injury or LPS infusion, and sequential observations in the three animal models planned in future experiments will be needed to resolve the question of whether some of the differences observed in gene expression among the models merely represent differences in the kinetics of the response to the two forms of injury and to the LPS stimulus. A better understanding of the relationship of altered gene expression in the three models to known immunological and inflammatory pathways should also result from such sequential data.
Moreover, these initial results were obtained from analysis of whole leukocyte populations from blood and spleen. Clearly, there were no marked differences in the percentage of any cell type in the postlysis WBC populations from the blood among the three sites. However, the loss of PMNs in postlysis samples precludes assessment of the contribution of this important cell type to changes in leukocyte gene expression after injury. This also highlights a problem inherent in all studies of WBCs in the mouse, since commonly used RBC lysis techniques eliminate substantial numbers of PMNs. We have, therefore, begun a project to develop a lysis protocol for mouse blood that we hope will permit preservation of PMNs in future experiments. Moreover, it is likely that analyses of gene expression in individual leukocyte subsets, planned in future experiments, will reveal important new information on the effects of injury and LPS on immune cell types known to mediate both inflammation and the host defense.
The information obtained from such a systematic evaluation of the three animal models can also be compared with the results of an analysis of the gene expression profiles in circulating WBCs withdrawn sequentially from patients with serious thermal or traumatic injury and from volunteers receiving LPS. Studies will continue over the next two years from clinical investigators in the “Inflammation and the Host Response to Injury” Project (http://www.gluegrant.org).
This work was supported by NIGMS Grant 1U54-GM-62119-03.
As noted in the author line, there are additional participating investigators in the Large-Scale Collaborative Research Program, “Inflammation and the Host Response to Injury”: Paul E. Bankey, Timothy R. Billiar, Steven E. Calvano, David G. Camp II, Celeste Campbell-Finnerty, George Casella, Mashkoor A. Choudhry, Ronald W. Davis, Asit De, Constance Elson, Bradley Freeman, Richard L. Gamelli, Nicole S. Gibran, Douglas L. Hayden, Brian G. Harbrecht, David N. Herndon, Jureta W. Horton, John Lee Hunt, Jeffrey Johnson, Matthew B. Klein, Stephen F. Lowry, Ronald V. Maier, Philip H. Mason, Grace P. McDonald-Smith, Bruce A. McKinley, Carol Miller-Graziano, Michael N. Mindrinos, Joseph P. Minei, Lyle L. Moldawer, Ernest E. Moore, Frederick A. Moore, Avery B. Nathens, Grant E. O'Keefe, Laurence G. Rahme, David A. Schoenfeld, Michael B. Shapiro, Martin G. Schwacha, Geoffrey M. Silver, Richard D. Smith, John Storey, Ronald G. Tompkins, Mehmet Toner, H. Shaw Warren, and Michael A. West.
We wish to thank Nochiketa Mohanty, Shunhua Hu, Adam Delisle, Marissa Miller, and Shannon Copeland for valuable technical assistance with these studies.
↵* B. Brownstein and T. Logvinenko contributed equally to this study.
↵1 The Supplemental Material for this article is available online at http://physiolgenomics.physiology.org/cgi/content/full/00213.2005/DC1.
Article published online before print. See web site for date of publication (http://physiolgenomics.physiology.org).
Address for reprint requests and other correspondence: J. A. Lederer or J. A. Mannick, Dept. of Surgery, Brigham and Women's Hospital, 75 Francis St., Boston, MA 02115 (e-mail:or ).
- Copyright © 2006 the American Physiological Society