The events occurring early in the burn wound trigger a sequence of local and systemic responses that influence cell and tissue survival and, consequently, wound healing and recovery. Using high-density oligonucleotide arrays we identified gene expression patterns in skin samples taken from a region of injury in the burn rat model. The associated genomic events include the differential expression of genes involved in cell survival and death, cell growth regulation, cell metabolism, inflammation, and immune response. The functional gene cluster detected and their time appearance matched the time sequence known to occur in burn wound healing.
- gene expression profile
- wound healing
- rat burn model
thermal injury is one of the most severe forms of trauma that affects the organism both locally and systemically. Localized burns affect the skin and impair its ability to act both as a protective barrier to the environment and as an immune organ (1). Physiological responses after a thermal injury include changes in cellular protection mechanisms, local and systemic inflammation, and reperfusion injury, which are all a major consequence of abrupt cellular energy depletion and cell damage (19, 22, 27). Initial local events may result in a cascade of responses, such as hypermetabolism, prolonged catabolism, organ dysfunction, and prolonged immunodeficiency (5,7, 12, 21, 28).
Burn wound healing is a complex process consisting of an early phase of abrupt energy depletion and necrosis, followed by a two-stage inflammatory phase, delayed cell death, formation of granulation tissue, matrix formation, and remodeling (11, 18, 24). Inflammation-mediated delayed cell death occurs at the wound borders and in surrounding tissue. Increased accumulation of macrophages and fibroblasts at the wound site leads to extracellular matrix deposition and angioneogenesis to form granulation tissue. Surrounding epithelial cells proliferate and migrate to gradually cover the open wound surface. Increased extracellular matrix deposition, especially collagen, and remodeling of the newly formed connective tissue persist for several months after injury (11, 18, 24). In the burn wound, this cascade of events not only takes place locally at the wound site, but also systemically via release of inflammatory mediators (14, 16).
DNA microarray analysis has changed the way biological events are investigated. Now systematic and global approaches are applied to investigate physiological mechanisms in health and disease. Differing from the traditional approach of sequentially studying individual gene products, a complete set of activated genes is now analyzed in toto. The major advantage of this application lies in reducing confounding events and circumstances and allows the characterization of interactions among different cellular pathways that so far are considered separately. Analyses at the genomic level indicate that events involved in biological processes can only be understood in the context of the activity of several thousand genes and only a composite assessment can determine the true significance of signaling events and structural entities (6).
The focus of the present study is to identify local responses to a thermal injury and the initial cellular responses through gene expression patterns in the burn wound in an effort to define the associated genomic events.
MATERIALS AND METHODS
Adult male Sprague-Dawley rats (Harlan Sprague Dawley, Houston, TX), weighing 350–375 g, were housed in wire bottom cages in a temperature-controlled room with a 12:12-h light-dark cycle. Rats were acclimatized to their environment for 7 days. All received a liquid diet of Sustacal (Mead Johnson Nutritionals, Evansville, IN) and water ad libitum throughout the study.
Anesthesia was by intraperitoneal injection of pentobarbital sodium (50–90 mg/kg) and buprenorphine-tartrate (0.1–1.0 mg/kg). Rats were shaved and randomized into burn or sham control groups. The burn group received a 40% total body surface area full-thickness scald burn on the dorsum and ventrum as previously described (14). Burned rats were resuscitated with 60 ml/kg lactated Ringer, placed on a warming pad, and, once awake, returned to their cages. Sham-burned animals received anesthesia and equivalent handling without a scald burn. Six rats in each group were killed at 2, 6, 24, and 240 h after burn. Skin (∼1 cm2) was harvested from the wound borders or corresponding sites in sham-burned animals and immediately frozen in liquid nitrogen and stored at −70°C for analysis.
This study was approved by the Animal Care and Use Committee of the University of Texas Medical Branch (UTMB), Galveston, TX, and followed the guidelines established by the National Research Council.
Total RNA extraction.
Total RNA was isolated from skin samples by acid guanidinium thiocyanate-phenol-chloroform extraction using TRI Reagent (Molecular Research Center, Cincinnati, OH). This method was based on the single-step method of RNA isolation described by Chomczynski and Sacchi (3). Samples were homogenized in TRI Reagent on ice, and total RNA was extracted following the manufacturers' instructions. Purified RNA was quantified by UV absorbance at 260 and 280 nm and stored in 25-μg aliquots at −70°C for DNA microarray analyses.
Gene array analysis.
Twenty-five micrograms of purified total RNA were transcribed into cRNA, purified, and then used as templates for in vitro transcription of biotin-labeled antisense RNA. All protocols followed the recommendations of the manufacturer (Affymetrix, Santa Clara, CA). Twenty micrograms of biotinylated antisense RNA preparation was fragmented, assessed by gel electrophoresis, and placed in a hybridization mixture containing four biotinylated hybridization controls (BioB, BioC, BioD, and Cre). Samples were hybridized to an identical lot of Affymetrix microarrays for 16 h. Microarrays were washed and stained using the instrument's standard Eucaryotic GE Wash 2′ protocol and antibody-mediated signal amplification. The images from the scanned microarrays were processed with Afffymetrix GeneChip Analysis Suite 3.2. Images from each microarray were scaled and adjusted to an average intensity value for all arrays of 1,500. Scaled average difference values and absolute call data from each microarray were exported to text files and used for statistical analysis. In vitro transcription and chip hybridization were performed in collaboration with the UTMB Genomic Core Facility.
The data analysis of genomic data was carried out as microarray validation, cluster analysis of transcription profiles (using Manhattan distances and looking at all genes expressed), identification of genes expressed only in one group (burned or unburned skin), identification of genes expressed in both groups at significantly different levels, and the temporal analysis of expression profiles.
Identification of present and absent genes.
The presence or absence of each probe within the group was determined according to the Affymetrix algorithm. A probe was considered present if its absolute call was present for at least two members of the group containing three samples. Otherwise the probe was regarded as absent, and the respective gene was considered absent in that group.
For microarray validation, data were clustered to detect gross discrepancies among different array data. Data of an array would be discarded if only a small fraction of probes was present compared with the remaining arrays in the group. The degree of similarity or dissimilarity among transcription profiles was tested using clustering methods (2, 9, 23, 26). The clustering of arrays was performed by SPSS software using unsupervised clustering methods (Pearson, Euclidian, and Manhattan hierarchical clustering) without artificially imposing their number. This distance is defined between two points, x = (x 1, x 2, . … xn ) andy = (y 1, y 2, . … yn ), inn-dimensional spaces asd(x,y) = Σ (x i− y i), here n = 8,799. All the above three clustering procedures yielded similar results. To further assess the significance of outcome, we performed correlation analysis of all the present probes. If a large deviation in the number of present calls or if the correlation coefficient among samples within one group was <0.85, the data set would be discarded (see Fig. 1). All data sets used for further analysis met these criteria.
Identification of genes with significantly different expression.
Further analysis focused on genes that were present in both burned and unburned groups. The within-group average of the expression was calculated for each group, and comparison was made between groups. This is done by calculating the expression difference for each gene and listing those that show a larger than twofold increase or decrease in activity compared with the unburned control group. Outliers were eliminated using a “jack-knife” procedure in which an entry is discarded if its value falls outside the three standard deviations calculated from the other two remaining entries. Because of the limited sensitivity of Affymetrix chips in detecting low-abundant mRNAs, we applied the t-test on data with present absolute call. Only statistically significant expression differences as determined were retained (8). Considering the small sample number per group, the power of the t-test was computed and results were discarded, even in light of a significant difference, if the power was <0.8. After establishing a list of genes with significantly changed expression, they were organized into categories depending on their physiological function.
Validation of microarray results using RPA.
To determine local mRNA expression of pro- and anti-inflammatory cytokines, multiple probe RNase protection assays (RPA) (RiboQuant Pharmingen, San Diego, CA) were performed using a multiprobe template set (rCk-1) containing probes for IL-1α, IL-1β, IL-2, IL-3, IL-4, IL-6, IL-10, TNF-α, TNF-β, and IFN-γ. α-32P-labeled antisense RNA probes from standard and rCK-1 templates were generated and hybridized with the sample (target) RNA. After the digestion of free probes and other single-stranded RNA with RNases, protected probes were purified and resolved on denaturing polyacrylamide gels and developed on film. Identification of expressed mRNA species was achieved by correlating the presence of bands to the expected fragment length. Signal intensity of α-32P-labeled probes was measured by phosphorimaging (Molecular Dynamics), and the quantity of each mRNA species was determined based on the intensity of the protected probe fragments. Templates for the analysis of L32 and GAPDH housekeeping genes are included to allow assessment of total RNA levels for normalizing samples and technique errors. Levels of the cytokine mRNA expression were normalized to the housekeeping gene L32.
Correlation analysis of the normalized expression levels of 8,799 probe pairs detected after hybridization of three equally treated burn skin samples with three Affymetrix microarrays per group showed correlation coefficients of 0.98 to 0.99 (Fig.1). This indicates very small interanimal variations and permits the use of average expression values to be used as representative data for the analysis of burn-induced changes in gene expression patterns.
Hierarchical cluster analysis.
Hierarchical cluster analysis of the gene expression profiles of burned skin at 2, 6, and 24 h after burn differed significantly from unburned skin. At 240 h, the transcription profiles of burned skin grouped closer with the profile of unburned skin (Fig.2).
To identify burn-induced changes at different time points, ratios of expression levels (burn/control) were calculated (Fig.3). Transcriptional profiles at 2 and 6 h postburn appear to be nearly identical. The transcription profiles at 24 h and 240 h differed from the 2- and 6-h profiles.
Identification of changes in gene expression after burn.
Genes present only in burned skin are listed in Table1. Gene expression analysis revealed 85, 48, 120, and 21 genes in burned skin at 2, 6, 24, 240 h, respectively. The corresponding numbers of expression sequence tags (ESTs) are indicated (Fig.4). Twenty genes were expressed at both 2 and 6 h, 34 genes were expressed at 2, 6, and 24 h. Ten genes were present in burned skin at all time points and included IL-1β#M98820 , growth-related oncogene gro # D11445 , tyrosine kinase p72 syk # U21683 , cyclin E # D14015 , JAK2 # AJ000557 , and Doc2A # U70779 . Thus initial activation of overall gene expression was followed by increased activation at 6 and 24 h, returning to basal levels 240 h after injury (Fig. 4). Genes not detected in burned skin but present in normal skin are indicated in Table 2 and Fig. 5.
Suppressed expression of 49 genes was seen in burned skin at 2 h after burn, 39 genes at 6 h, 34 genes at 24 h, and 50 genes at 240 h compared with unburned skin. However, only three genes, aminopeptidase N #AF03989, brain serine protease-1#AJ00564, and cytochrome p-450e#J00728 , were completely absent at 2, 6, and 24 h in burned skin. Ten genes were absent at 2 and 6 h, 3 at 6 and 24 h, and 7 at 24 and 240 h. No single gene present in unburned skin was absent in burned skin at all times tested.
The transcription profile of burned skin shows an increasing expression of genes during the first 24 h, followed by a decrease atday 10 consistent with the transient induction of gene activity at the wound site (Fig. 6). The number of genes that are no longer expressed after injury compared with normal skin does not decrease over time, suggesting that by the 10th day, the healing process is ongoing.
Validation of microarray results.
The expression of IL-1β #M98820 in burned skin was confirmed by measuring mRNA using an RPA, which correlated well with the microarray findings. Similarly, IL-6 #M26744 expression at 24 h and its absence at 10 days corresponded to mRNA values detected by RPA (Fig. 7). IL-1α, IL-2, IL-3, IL-4, IL-10, and TNF-β could not be detected in samples by RPA, consistent with our microarray findings.
Dynamic of gene expression over time.
There were 2,703 genes expressed at all times in both burned skin and control groups. We calculated their expression ratios (burn/control) to identify expression patterns over time. Genes that increased or decreased twofold in at least one time point compared with control were selected. This “filter” eliminated 781 genes (including 338 EST), and the remaining 443 genes were used for further analysis.
In burned skin, 85 genes showed above threefold or greater increases and 54 genes showed threefold or greater decreases in expression compared with normal skin. These genes can be grouped according to function and their involvement in physiological processes, such as cell survival and death, inflammation, cell growth, and differentiation (Table 3; Fig.8). The temporal distribution of these genes is shown in Fig. 9. Four of these genes initially decreased in expression, followed by more than a threefold increase at 10 days.
The sequence of burn wound healing virtually starts with the injury itself. The concept of injury zones in burn wounds described by Jackson (15) consists of three zones with different degrees of tissue damage. The central “zone of coagulation” is the site of irreversible tissue necrosis; adjacent to it is an area with ischemic tissue at risk for necrosis (zone of stasis), and a peripheral zone characterized by reactive vasodilatation and inflammation (zone of hyperemia). From a surgical standpoint, the intermediate region of injury is the one of greatest clinical interest, as it harbors the potential to undergo tissue necrosis as well as full recovery. We therefore took the biopsy for analysis in the region of the wound border (zone of stasis).
Immediately after injury, cascading cellular and humoral events are activated leading to capillary leakage or local bleeding and coagulation. The associated deposition of fibrin provides an early matrix for immigrating cells and entraps platelets, which in turn secrete growth factors (PDGF, TGF-β, etc.), to attract and activate fibroblasts, endothelial cells, macrophages, and immune-competent cells (13).
During the early phase of wound healing, platelets secrete a number of direct or indirect angiogenic substances such as TGF-β, PDGF, TNF-α, and basic FGF, all of which stimulate the proliferation and in-growth of capillaries. In full-thickness wounds, the proliferating and migrating epithelium arises from the wound border. The rate of epithelial coverage is modulated by growth factors that stimulate proliferation and chemotaxis, and the presence and production of matrix metalloproteinases and plasminogen activator to facilitate migration through the ECM degradation (24). About 1 wk after injury, fibroblasts, attracted by macrophages and the secreted growth factors, migrate into the wound and begin collagen production (4). Type I and III collagens are predominantly synthesized to form new extracellular matrix. With maturing of newly formed scar at the closed wound site, specific matrix metalloproteinases are produced by keratinocytes, fibroblasts, granulocytes, and macrophages to remodel the deposited extracellular matrix and to facilitate further cell migration.
Gene expression changes.
Examination of the cDNA microarray data shows small interanimal differences in all the groups examined over time after burn trauma (0.98 < r < 0.99). Thus this reflects not only significance of the measured expression values (P < 0.05), but also high statistical power.
Cluster analyses results were consistent with the known sequence of physiological events associated with burn trauma in that there was a prompt and an early robust response in gene expression. Cluster analyses results at postburn day 10 were consistent with the onset of recovery that signifies a return to baseline values of gene expression.
Examination of present vs. absent gene expression over time showed early transient stimulation of cytokines and other genes encoding inflammatory mediators (Table 1); for example, IL-1β#M98820 , IL-6 #M26744 , inducible nitric oxide synthase #U03699 , IL-1β converting enzyme#U84410 , prostacyclin synthase #U53855 , and COX-2#S67722 . Not surprisingly, a large number of cell survival-promoting genes and proliferative regulators were activated within a few hours after burn trauma (Table 1); for example, p53#X13058 , heat shock protein 70 (hsp70) #Z27118 , and N-ras/p21 #X68394 . These signal transduction changes agree with known burn-induced changes in IGF-I and IGF-I binding proteins after burn trauma (17).
The early activation of cell cycle and proliferation-related genes, such as cyclin C #D14013 and E #D13015 , osteoprotegerin #U94330 and topoisomerase #D14046 suggests that these are likely to be initiators of the wound healing process, a response not foreseen. The sheer magnitude of the number of genes whose expression is affected speaks to the overwhelming nature of the force of the trauma.
Interestingly, examination of genes whose expression decreases in skin shortly after burn trauma (Table 2) shows a preponderance of genes associated with neuronal, neuroendocrine, and metabolic functions. This suggests that their absence is likely due to functional and structural loss of neurons and supporting glia, and a delayed reinnervation associated with inflammation and other events required for wound healing.
When taking a closer look at the complex pattern of significantly increased gene expression over time after burn trauma (Table 3, Fig.8), it becomes clear that the affected genes play a role in cell growth and differentiation, cell signaling, stress response, energy metabolism, immune response, cell survival and migration, inflammation, or protein synthesis or are structural proteins. As an early response to the inflicted tissue damage and necrosis, an increased expression of cell stress-related genes (e.g., hsp70 # L16764 , hsp60 # X54793 , metallothioneine 1 and 2 #M11,794, oxidative stress inducible protein tyrosine phosphatase # S81474 ), and of genes related to inflammation, innate immune response, and transcriptional and translational mechanisms (Complement C1 # X71127 , NGF induced factor A # AF023087 , Calprotectin # L18948) was seen within 2 h after injury. At the same time, gene expression of structural proteins (Collagen # AJ005394 , α-tropomyosin # M60666 ) decreased. Within 2 to 6 h, transcription factors, immediate early genes, cell proliferation, and cell survival-associated genes were affected. Their activation, in many instances, appears to be over by 10 days (Table 3). Inflammation-related genes and stress response genes remained the mainstay of upregulated gene expression for the first 24 h in agreement with earlier reports (10). Genes involved in cell migration, such as membrane type matrix metalloproteinase # X83537 and versican V3 # AF072892 , appear to be activated later. By 10 days after burn, the expression pattern shifted to increased expression of structural, cellular, and extracellular matrix proteins, genes involved in the specific immune response, and regulation of cellular function. Not surprisingly, there was a repression of genes playing significant roles in energy metabolism, such as pyruvate dehydrogenase #U10357 , phosphoglyceromutase #Z17319 , or cytochrome oxidase#U40836 , indicating metabolic failure of the involved tissue, which was resolved in 10 days. This is consistent with reports showing a central role for neutrophil migration and immune activation in burns in rats (25).
All the described events of differential gene expression correlated well with distinct gene families typically identified by gene cluster analysis. These genes include cellular- and intercellular mediators, such as inflammatory and anti-inflammatory cytokines, hormones, stress-response genes, cell-cycle genes, pro- and anti-apoptotic genes, gene products of cellular metabolism, genes coding for structural proteins, and extracellular matrix products and their proteinases. The assignations from DNA microarray analyses typically correspond to known sequelae of cellular and molecular events associated with burns and sequential and coordinated wound healing. However, there were some interesting exceptions that point to an activation of some clusters of genes at earlier times than previously thought and a persistence in the expression of genes previously believed to display transient responses, such as the inflammatory mediators gro #D11445 and IL-1β #M98820 .
The changes in gene expression in the burn wound can be correlated to known events in the wound healing process after burn. These events are coordinated and follow a well-defined sequence (Fig.10). The initial injury leads to energy depletion and cellular damage, which may lead to tissue necrosis. Acute stress genes are initially upregulated to prevent further tissue damage and trigger restoration of tissue homeostasis. At the same time, inflammation and nonspecific host defense mechanisms are activated. The activation of restorative mechanisms, such as growth regulatory genes and genes encoding for the production of structural proteins, occurs at later times.
Gene array assays are not as sensitive as the validation procedures typically used to measure mRNA levels. In these experiments we relied on the use of RPAs that have one order of magnitude higher level of sensitivity than Affymetrix arrays in terms of detecting a specific low abundance mRNA, a serious caveat when considering signaling molecules with low copy number under control or sham situations, such as several of the cytokines. To overcome this caveat we employed fairly rigid standards of present vs. absent conditions in the analyses of the microarray analyses as described in materials and methods. For the RPA analyses, the lower threshold of sensitivity is in the picogram range, whereas for the microarray analyses, although it is hard to determine based on our experience in a number of paradigms (20), it is more in the nanomolar range.
In summary, we were able to show a correlation of gene expression in skin after burn trauma with known sequences and events present in wound healing. These gene families included physiological cellular- and intercellular mediators, such as pro- and anti-inflammatory cytokines, hormones, stress-response genes, cell-cycle genes, pro- and anti-apoptotic genes, genes of cell metabolism, genes coding for structural proteins, and extracellular matrix products and their proteinases. A number of genes were present at all times in burned skin that were completely absent in unburned skin. These genes encompassed genes involved in cytokine and inflammatory signaling, cell cycle genes, and genes regulating cell growth and proliferation. The understanding of the activation process of these genes early in the sequelae of skin undergoing thermal trauma may suggest several novel therapeutic approaches to ameliorate the local tissue damage, protect tissue at risk of necrosis, and improve tissue recovery.
This study was supported by Shriners Hospitals for Children Grants 8660 and 8490 and National Institutes of Health Grants 1P50GM60338–1 and 5RO1GM572903.
Address for reprint requests and other correspondence: M. R. K. Dasu, Shriners Hospitals for Children, 815 Market St., Galveston, TX 77550 (E-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2002 the American Physiological Society