Characterization of Perceived Hyperoxia in Isolated Primary Cardiac Fibroblasts and in the Reoxygenated Heart* □ S

Under normoxic conditions, p O 2 ranges from 90 to < 3 torr in mammalian organs with the heart at (cid:1) 35 torr (5%) and arterial blood at (cid:1) 100 torr. Thus, “normoxia” for cells is an adjustable variable. In response to chronic moderate hypoxia, cells adjust their normoxia set point such that reoxygenation-dependent relative elevation of p O 2 results in perceived hyperoxia. We hypothesized that O 2 , even in marginal relative excess of the p O 2 to which cells are adjusted, results in the activation of specific O 2 -sensitive signal transduction pathways that alter cellular phenotype and function. Thus, reperfusion causes damage to the tissue at the focus of ischemia while triggering remodeling in the peri-infarct region by means of perceived hyperoxia. We reported first evidence demonstrating that perceived hyperoxia triggers the differentiation of cardiac fibroblasts (CF) to myofibroblasts by a p21-dependent mechanism (Roy, S., Khanna, S., Bickerstaff, A. A., Subramanian, S. V., Ata-lay, M., Bierl, M., Pendyala, S., Levy, D., Sharma, N., Venojarvi, M., Strauch, A., Orosz, C. G., and Sen, C. K. (2003) Circ. Res. 92, 264–271). Here, we employed to compare the difference between pair-matched normal- perfused and I-R myocardial tissue. A p value (cid:2) 0.05 was considered statistically significant. Statistics related to DNA microarray data proc- essing are described above under “Data Analysis.”

Cellular oxygen (O 2 ) concentrations are maintained within a narrow "normoxic" range to circumvent the risk of oxidative damage from excess O 2 (hyperoxia) and of metabolic demise from insufficient O 2 (hypoxia) (1). pO 2 ranges from 90 to below 3 torr in mammalian organs under normoxic conditions with arterial pO 2 of about 100 torr or ϳ14% O 2 (2). Thus, "normoxia" for cells is an adjustable variable that is dependent on the specific localization of the cell in organs and functional status of the specific tissue. O 2 sensing is required to adjust to physiological or pathophysiological variations in pO 2 . Current work in this field is almost exclusively focused on the study of hypoxia. Reoxygenation, on the other hand, has been mostly investigated in the context of oxidative injury.
Ischemia in the heart results in a hypoxic area containing a central focus of near-zero O 2 pressure bordered by tissue with diminished but nonzero O 2 pressures. These border zones extend for several millimeters from the hypoxic core, with the O 2 pressures progressively increasing from the focus to the normoxic region (3). Moderate hypoxia is associated with a 30 -60% decrease (ϳ1-3% O 2 ) in pO 2 (4). During chronic hypoxia in the heart, cells adjust their normoxic set point such that the return to normoxic pO 2 after chronic hypoxia is perceived as relative hyperoxia. Cardiac fibroblasts (CF) 1 are mainly responsible for the synthesis of major extracellular matrix in the myocardium including fibrillar collagen types I and III and fibronectin. More than 90% of the interstitial cells of the myocardium are fibroblasts (5), which actively cross-talk with myocytes to determine the quantity and quality of extracellular matrix. We have recently observed that CF, isolated from adult murine ventricle, cultured in 10 or 21% O 2 (high O 2 , relative to the pO 2 to which cells are adjusted in vivo), compared with 3% O 2 (mildly hypoxic), exhibit reversible growth inhibition and a phenotype indicative of differentiation. We reported that marginal relative elevation in pO 2 , compared with pO 2 to which cells are adjusted during chronic moderate hypoxia, serve as a signal to trigger CF differentiation and tissue remodeling. p21 Waf1/Cip1/Sdi1 was identified as a key mediator of signaling triggered by perceived hyperoxia (6). Thus, while acute insult during reperfusion may be lethal to cells localized at the focus of insult, sudden relative elevation of O 2 tension in the surrounding peri-infarct ischemic tissue is expected to trigger phenotypic changes in the surviving fibroblasts that may be associated with tissue remodeling. Our current objective is 2-fold: (i) characterize the genomic responses to perceived hyperoxia in isolated cardiac fibroblasts using a DNA microarray approach and (ii) seek first in vivo evidence testing whether p21 expression is induced in the heart recovering from ischemia-reoxygenation.

Cardiac Fibroblast Isolation and Culture
Experiments were performed using primary CF isolated from adult (5-6 weeks old) mouse ventricle using procedures described previously (6).
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

GeneChip TM Probe Array Analysis
To identify sets of perceived hyperoxia sensitive genes in CF, we utilized the GeneChip TM approach (7,8). CF were isolated and cultured in a moderately hypoxic condition at 3% O 2 (6). Five days after isolation, cells were split and exposed to either elevated (20% O 2 ; widely interpreted as in vitro normoxia) or the same (3%) O 2 tension for 20 h. Cells were harvested and the total RNA was extracted using the RNeasy kit (Qiagen). Targets were prepared for microarray hybridization according to previously described protocols (7,8). To assess quality, the samples were hybridized for 16 h at 45°C to GeneChip TM test arrays. Satisfactory samples were hybridized to the mouse genome arrays (U74Av2) for the screening of over 12,000 genes and expressed sequence tags. The arrays were washed, stained with streptavidin-phycoerythrin, and were then scanned with the GeneArray scanner (Agilent Technologies) in our own facilities. To allow for statistical treatment, data were collected from three experiments.

Data Analysis
Raw data were collected and analyzed using Affymetrix Microarray Suite, version 5.0 (MAS) and Data Mining Tool, version 2.0 (DMT) software. Additional processing of data was performed using dChip software (9). A detailed analysis scheme has been illustrated in Fig. 3. Statistical (t test) and comparison analysis were the two approaches utilized to identify differentially expressed genes (7). t test was performed using DMT on absolute files generated from MAS. Genes that significantly (p Ͻ 0.05) changed (increased or decreased) in the perceived hyperoxia (20% O 2 ) group compared with the 3% O 2 group were selected. Next, dChip (version 1.3, Harvard University) software was used to further filter genes using following criteria: (i) fold change Ͼ1.2; (ii) t test, p Ͻ 0.05; and (iii) present call in all experimental (20% O 2 ) samples for up-regulated genes and present call in all base-line (3% O 2 ) samples for down-regulated genes. Using comparison analysis in MAS, nine pairwise comparisons were generated from three experiments of both 3 and 20% O 2 groups. Average fold changes were calculated for both up-or down-regulated genes. Genes with 100% (9 out of 9 pairs) concordance in pairwise comparisons were selected. For data visualization, genes filtered using the statistical (t test) approach were subjected to hierarchical clustering using dChip (version 1.3) software. Functional categorization and pathway construction were performed using the following software/web resources: Gene Ontology Data Mining Tool (Affymetrix), KEGG (Kyoto Encyclopedia of Genes and Genomes), Gen-MAPP (10), DAVID (Data Base for Annotation, Visualization, and Integrated Discovery Verification) (11), and LocusLink (Swiss-Prot). Microarray data were verified using real-time PCR assay.
mRNA Quantitation mRNA were quantified by real-time PCR assay using doublestranded DNA binding dye SYBR Green-I as described previously (6). The primer set used for individual genes are listed in A and B, after isolation, CF were either (i) fixed 4 h after isolation or (ii) cultured at 5 or 20% O 2 for 8 days. A, for morphologic comparisons with freshly isolated CF fixed 4 h after isolation, CF cultured at various O 2 tensions for 8 days were trypsinized and reseeded. Cells were fixed 4 h after reseeding. The reseeding was performed to obtain comparable cell spreading in freshly isolated CF versus CF maintained in various O 2 environments for 8 days. Cells were stained with phalloidin (actin, red) and nuclei were stained with 4Ј,6-diamidino-2-phenylindole (blue). Imaging was performed using a Zeiss microscope. Scale bar, 20 m. B, cell area quantification was performed using Axiovision software. C, quantification of p21 mRNA expression using real-time PCR. p21 data presented have been normalized for glyceraldehyde-3-phosphate dehydrogenase (GAPDH).
FIG. 2. CF cultured in three-dimensional collagen matrices mimic the effect of oxygen observed in standard CF monolayer culture. Three-dimensional culture of CF in collagen gel was performed as described under "Materials and Methods." A, after isolation CF were cultured in either 5 or 20% O 2 for 5 days. Cells were infected with adenoviral vectors carrying green fluorescent protein (Ad.GFP) gene followed by culturing the in three-dimensional collagen matrix. Images were collected 72 h after culture in three-dimensional collagen gels. B, for gene expression studies, cells were harvested from collagen gel after 72 h using collagenase digestion. RNA was extracted and p21 expression was determined using real-time PCR. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. and microscopy (Zeiss Axiovert 200M) were performed as described previously (6).

Culture of CF in Three-dimensional Collagen Gels
Collagen lattices were prepared using type I collagen from rat tail tendon as described previously (12). CF were suspended in collagen (1.25 mg/ml) and aliquoted into culture plates. Collagen matrices were incubated in a cell culture incubator for 45 min followed by overlying the gel with Dulbecco's modified Eagle's medium containing 10% serum. Matrices were gently released after 24 h from the underlying culture dishes with a spatula and allowed to float in basal medium. Collagenase digestion was performed to harvest cells from collagen gel for gene expression studies.

Survival Model for Coronary Artery Occlusion and Reperfusion
Sprague-Dawley rats weighing 250 -300 g were subjected to ischemia-reperfusion (I-R) of the heart. Parallel groups of rats were used as sham-operated controls. The studies were approved by the Animal Protection Committee of The Ohio State University. Rats were anesthetized, intubated, and mechanically ventilated on a positive pressure respirator with room air. The body temperature was maintained at 36 -37°C with a heated small animal operating table. A left thoracotomy was performed via the fifth intercostal space to expose the heart. A 30-min occlusion of left anterior descending coronary artery was followed by reperfusion. Laser Doppler flow measurement was used to verify I-R. Upon successful reperfusion, the thorax was closed, and negative thoracic pressure was re-established for survival. The rats were killed 2-7 days after reperfusion. For tissue harvesting, the infarcted I-R area was visualized under a dissecting microscope. Tissue samples were collected from infarcted (I-R) site as well as from a part of the left ventricle unaffected by I-R. As additional controls, tissues were collected from a location corresponding to the I-R and unaffected sites in sham-operated rats. These rats were exposed to all surgical procedures except I-R.

Histology
Histochemistry-Formalin-fixed tissues were embedded in paraffin and sectioned (4 m) followed by hematoxilin and eosin staining.

In Vivo Hyperoxia Exposure
Male C57BL6 mice (5 weeks) were exposed to pure oxygen (hyperoxia) environment for 36 h. Control mice were maintained under similar conditions in room air (normoxia). p21 mRNA levels in myocardial tissue were evaluated by real-time PCR.

Statistics
The effect of perceived hyperoxia on p21 mRNA expression in hearts subjected to ischemia and reperfusion was analyzed using the Statistical Package for the Social Science, (Chicago, IL). Wilcoxon test was FIG. 3. GeneChip TM data analysis scheme. GeneChip TM data analysis scheme used to identify perceived hyperoxia-sensitive genes in three experiments (A, B, C) where cardiac fibroblasts were exposed to 3 or 20% O 2 for 20 h. Data processing was primarily performed using MAS and DMT software. Additional data filtration was performed using dChip using the following criteria: FIG. 4. Cluster images illustrating genes sensitive to perceived hyperoxia. For a clear graphic display of perceived hyperoxia-sensitive genes, the t test was performed on data from repeated experiments (A, B, C) involving 3 and 20% O 2 groups. The genes that significantly (p Ͻ 0.05) changed between the two groups compared were selected and subjected to hierarchial clustering using dChip software as described in the legend to Fig. 3. Red to green gradation in color represent higher to lower expression signal. A and B, up-regulated (A) or down-regulated (B) genes in 20% O 2 compared with 3% O 2 group. employed to compare the difference between pair-matched normalperfused and I-R myocardial tissue. A p value Ͻ 0.05 was considered statistically significant. Statistics related to DNA microarray data processing are described above under "Data Analysis."

RESULTS
Under conditions of systemic normoxia, heart cells receive a limited supply of O 2 representing less than 10% (13)(14)(15). We have directly determined that mouse heart ventricular pO 2 is in the range of 5% (6). Thus, in this work 3, 5, and 20% O 2 are referred to as conditions representing mild hypoxia, normoxia, and perceived hyperoxia, respectively. Previously we have reported that the culture of isolated CF under room air (20% O 2 ) conditions results in major phenotypic changes including increased cell size and differentiation to myofibroblasts. In addition, perceived hyperoxia induced the expression of p21 (6). Our current results with isolated cardiac fibroblasts demonstrate that the phenotype of freshly isolated CF may be preserved under standard culture conditions by controlling the ambient O 2 condition to match the physiological pO 2 to which the cells are adjusted to in vivo. Maintenance of CF at 5% O 2 preserved cellular phenotype and did not result in the induction of the perceived hyperoxia-sensitive gene p21 (Fig. 1). Recently it has been identified that the study of fibroblasts in three-dimensional collagen matrices represents a powerful approach to model in vivo biology (12). Therefore, we sought to investigative whether the perceived hyperoxia induced phenotypic changes manifest during three-dimensional culture of CF. Indeed, both cell size as well as p21 expression was clearly increased in CF cultured in three-dimensional collagen matrices (Fig. 2). Under conditions of perceived hyperoxia, CF grown in collagen matrices exhibit increased contractile properties indicative of differentiation to myofibroblasts (6). Thus, our findings from our standard culture model are consistent with results obtained from three-dimensinal culture of CF.
In our previous study, we employed quantitative single-gene analysis approaches to determine that perceived hyperoxia influences genes related to cell cycle and differentiation (6). To obtain a comprehensive understanding of the genome-wide effects of perceived hyperoxia, we conducted a DNA microarray analysis of O 2 -sensitive genes in CF. CF grown at 3 or 5% O 2 exhibits comparable phenotype and growth pattern. Thus, we sought to compare the transcriptome of CF grown at 3% O 2 with that of CF grown at 20% O 2 . The objective was to identify sets of O 2 -sensitive genes that would help us understand the biology of perceived hyperoxia. Three independent experiments were conducted so that appropriate data treatment could be possible. The data analysis design is illustrated in Fig. 3. Out of 12,488 probe sets screened, 5071 were present in CF at 3% O 2 . Exposure of CF to 21% O 2 resulted in the detection of 5574 genes/expressed sequence tags as being present. Perceived hyperoxia resulted in the induction of 216 genes and down-regulation of 85 genes (Fig. 4). To obtain the biological significance of these findings, the genes sensitive to perceived hyperoxia were subjected to biological process categorization using Gene Ontology Data Mining (Affymetrix), GenMAPP, and DAVID. Fig. S1 illustrates the major functional categories identified using these approaches. Out of the 216 up-regulated probe sets uploaded, 109 had annotations for Gene Ontology biological process (Fig. S1A). For the 85 genes/expressed sequence tags down-regulated in response to perceived hyperoxia, 49 had annotations for Gene Ontology biological process (Fig. S1B). It was clear that among the annotated genes up-regulated in response to perceived hyperoxia, genes related to cell growth/ maintenance and cell communications represent the largest functional categories. This finding is consistent with our previous results suggesting that perceived hyperoxia may play a key role in CF differentiation and tissue remodeling in the post-reoxygenation peri-infarct tissue. Next, genes sensitive to perceived hyperoxia were analyzed for cellular compartmentalization of their products using Gene Ontology Data Mining and DAVID. The major compartments affected are illustrated (Fig.  S2). The total volume of up-regulated genes was roughly 2.5- The design includes isolation of CF followed by culturing then in 3% O 2 for 5 days. After 5 days cells are split and then exposed to either 3 or 20% O 2 for 20 h. B, After isolation CF were cultured in either 3 or 20% O 2 for 5 days. After splitting, 3 and 20% O 2 group were exposed to 3 and 20% O 2 , respectively, for next 3 days. p21, cyclin-dependent kinase inhibitor 1A; Gja, gap junction membrane channel protein ␣ 1; Wig1, wild-type p53-induced gene 1; Rbm3, RNA binding motif protein 3; Tob1, transducer of ErbB-2.1; Klf4, Krü ppel-like factor 4. Fold changes are indicated in parentheses.
fold higher than the down-regulated genes. Growth arrest and differentiation are largely driven by nuclear gene products. Consistently, genes with products associated with nuclear function were observed to be a large group sensitive to perceived hyperoxia. Of interest, genes encoding cytoskeletal proteins were specifically induced in response to perceived hyperoxia (Fig. S2). This finding is consistent with our claim that perceived hyperoxia potently influences cellular phenotype.
To validate our DNA microarray results, the expression of select perceived hyperoxia sensitive genes were independently determined using real-time PCR. Using this approach, we typically find that the microarray approach underestimates the magnitude of change while reliably detecting the direction of change (Fig. 5A). To test the robustness of the findings, we conducted independent experiments with a more long-term O 2 exposure design (Fig. 5B). These efforts led to successful verification of findings obtained from the experiments conducted for the microarray study (Fig. 5A). To obtain an understanding of how the perceived hyperoxia-sensitive genes are related with respect to signaling pathways, the candidate genes were mapped onto known pathways associated with cell cycle/death and cytoskeletal changes. GenMAPP, KEGG, and Gene Ontology tools were used to reconstruct the pathways. The relevant signaling pathways that were derived using this approach identified growth inhibition and fibroblast differentiation as two related outcomes that are induced by perceived hyperoxia (Fig. 6).
To document the incidence of perceived hyperoxia in vivo, we developed a survival surgery model involving ischemia-reoxygenation of the rat heart. The objective was to examine the post-reoxygenation tissue for the induction of p21, a key perceived hyperoxia-sensitive gene (6). The affected site was histologically characterized for reoxygenation injury and inflammation (Fig. 7). Staining of the affected site revealed increased presence of p21 in post-reoxygenation tissue (Fig. 8A). This finding was verified by real-time PCR (Fig. 8B). To obtain proof of principle that p21 expression in the heart is indeed sensitive to hyperoxia in vivo, male C57BL6 mice were exposed to a pure oxygen environment for 36 h. Control mice were maintained under similar conditions in room air. Evaluation of mRNA levels in myocardial tissue using real-time PCR showed multifold increase in the expression of p21 following hyperoxic insult (Fig. 8).

DISCUSSION
Cardiac remodeling leads to changes in the extracellular matrix, and in some cases to cardiac fibrosis, a process involving myofibroblasts as an active component. Myofibroblasts are phenotypically distinct. They are larger and more stellate, proliferate slowly, and have the most prominent microfilament arrays. We have recently proposed that reoxygenation of an ischemic site, in addition to being a trigger for injury, may induce tissue remodeling. We observed that exposure to a higher pO 2 , relative to which cells are adjusted, induces the differentiation of CF to myofibroblasts. This is of significant relevance to cardiac tissue remodeling in the face of reoxygen-FIG. 6. Pathway construction based on GeneChip TM expression data. To obtain insights on perceived hyperoxia-induced changes in cell growth and differentiation on a pathway sense, the results of GeneChip TM analysis were mapped onto known pathways associated with cell cycle/death and cytoskeletal changes. GenMAPP, KEGG, and Gene Ontology were used to create the pathways. Genes shown in red ellipse are candidates identified using GeneChip TM assay that were up-regulated in 20% O 2 compared with 3% O 2 . Green ellipses are genes that were down-regulated under conditions mentioned above. The expressions of candidates shown in red ellipse with blue outline have been independently verified using either real-time PCR or ribonuclease protection assay (6). APAF, apoptotic protease activating factor; Bmp, bone morphogenetic protein; BAX, Bcl2-associated X protein; BID, BH3 interacting domain death agonist; Catn, catenin; Cappa, capping protein ␣; CASP, caspage; ccng, cyclin G; Cdc61, cell division cycle; CDK, cyclin-dependent kinase; CDKN1A, cyclin-dependent kinase inhibitor 1A (p21); Clu, clusterin; Cx43, gap junction membrane channel protein; GADD, growth arrest and DNA damage-inducible; IQGAP1, IQ motif containing GTPase activating protein 1; Inhba, inhibin ␤-A; Klf, Krü ppel-like factor 4 (gut); Lsp1, lymphocyte specific 1; MAPK, mitogen-activated protein kinase; Mdm2, transformed mouse 3T3 cell double minute 2; N-Cdh, cadherin 2; Pfn, profilin 2; PXN, paxillin; TGF, transforming growth factor; Tob, transducer of ErbB-2.1; TP53, transformation-related protein 53; Vcl, vinculin; Wig, wild-type p53-induced gene 1. ation (6). In the absence of robust in vivo models for chronic low flow we have turned toward the study of isolated CF, key players in tissue remodeling. The culture of isolated CF under conditions of O 2 -rich room air results in a clear switch of phenotype compared with the phenotype of freshly isolated CF. In this work we report direct evidence demonstrating that solely by maintaining the O 2 ambience of the culture condition to a level closely matching in vivo ventricular pO 2 , such culturedependent switch of phenotype may be prevented. These findings further support the contention that changes in O 2 environment is a cue that CF sensitively responds to. Fibroblast biology in three-dimensional collagen matrices closely models in vivo situations (12). Under such conditions, cells exhibit a distinct phenotype referred to as dendritic fibroblasts (12). Our observation that perceived hyperoxia induces the expression of p21, increase in cell size, and differentiation to myofibroblast (6) has been confirmed in CF grown in a three-dimensional matrix. Indeed, myofibroblasts have been detected abundantly in healing myocardial reperfused infarcts (16).
Application of the high density DNA microarray approach coupled with current bioinformatics tools enabled us to identify sets of O 2 -sensitive genes in CF and categorize them into functional groups. In addition, the candidate genes have been categorized with respect to the cellular localization of their products. That low O 2 ambience serves as a cue to trigger angiogenesis is a well accepted notion. We have previously demonstrated that the sensing of O 2 environment is not limited to hypoxia but extends to perceived hyperoxia as well (6). Our current report constitutes first evidence demonstrating that the culture of cells, isolated from an organ, at room air condition results in specific and significant genome-wide changes. As we strive to improve in vitro approaches to study cell biology relevant to in vivo conditions, a careful consideration of the O 2 -ambience is clearly warranted. Furthermore, the results provide a comprehensive definition to perceived hyperoxia establishing that cell communication and cell growth/maintenance are indeed the primary outcomes affected. Both of these processes are tightly linked to cellular differentiation and tissue remodeling (6,17). Recently it has been proposed that FIG. 7. Histologic evaluation of myocardium following ischemia-reperfusion. Occlusion (30 min) of left anterior descending coronary artery was followed by reperfusion for 2 or 7 days. Tissue samples were collected from infarcted (I-R) or normally (normal) perfused ventricular myocardium. A and B, hematoxylin and eosin staining of the myocardium. Note the infiltration of inflammatory cells and damaged myocardium. C, presence of macrophages (green, ED-1 antibody staining) at the I-R site.
FIG. 8. Activation of p21 in myocardial tissue following ischemia/reperfusion and hyperoxic insult. A and B, occlusion (30 min) of left anterior descending coronary artery was followed by reperfusion for 7 days. Tissue samples were collected from infarcted (I-R) or normally (control) perfused ventricular myocardium. Samples from control and I-R sites were also collected from sham-operated animals as additional control samples. A, immunolocalization of p21 protein (brown) in myocardium following ischemia (30 min) and reperfusion (7 days). Upper panel, ϫ200 magnification; lower panel, ϫ400 magnification. B, quantification of p21 mRNA expression using realtime PCR; paired data (mean Ϯ S.E.; n ϭ 14) shown from animals underwent ischemia (30 min) and reperfusion (7 days). *, p Ͻ 0.05 significantly different compared with its pair-matched normal-perfused tissue. C, male C57BL6 mice were exposed to 100% oxygen (hyperoxia) environment for 36 h. Control mice were maintained in room air (normoxia). p21 mRNA levels in myocardial tissue were evaluated by real-time PCR. Data for individual animals have been presented to demonstrate tight uniformity of the findings. GAPDH, glyceraldehyde-3-phosphate dehydrogenase. many of the signaling pathways that control cellular decisions related to tissue remodeling are regulated by nuclear interactions of cell cycle proteins (18). Indeed, our results demonstrate that nuclear proteins represent one of the largest categories of products encoded by genes sensitive to perceived hyperoxia. Using the perceived hyperoxia-sensitive gene sets that we have identified previously (6), and the sets identified in the current study, our pathway reconstruction results characterize the specific effects of perceived hyperoxia on cell cycle.
Select candidate genes that have been verified using quantitative real-time PCR include p21 (6), Wig-1, Gja, Rbm3, Tob1, and Klf4. The transcription factor Krü ppel-like factor 4 (Klf4) supports p53-dependent cell cycle arrest (19) and differentiation (20). Tob is a member of the p53-independent, as well as -dependent, anti-proliferative family genes (21,22). A critical role of Tob in checking cell cycle is suggested by findings that mice lacking Tob are predisposed to a higher incidence of cancer (23). The nuclear glycine-rich ribonucleoprotein RNA binding motif protein 3 (Rbm3) (24) has translation regulatory functions, the cardiovascular significance of which remains to be characterized. Gap junctions are widely distributed structures that mediate communication between cells. The channels that allow passage of small molecules between adjacent cells are made up of connexins that are encoded by a family of related genes. In the heart, connexins ensure electric and metabolic coupling between cells. Cell-cell communication is a key process during differentiation and typically it is down-regulated during cell proliferation. Gap junction gene A1 (Gja1) or connexin 43 is essentially required for coronary artery development and patterning (25). Thus, a central remodeling role of this gene in the healing tissue is expected. In humans, defect in connexin 43 result in visceroatrial heterotaxia (26). Wig-1 represents a stress-responsive p53-induced growth-inhibitory zinc finger protein (27). It is nuclear localized and binds to doublestranded RNA (28). Double-stranded RNA binding proteins are emerging as evolutionarily conserved key regulatory mechanisms modulating fine aspects of gene expression including translation, RNA editing, and stability (29). p21 supports remodeling of tissues injured by oxygenation (30). Our previous work identified p21 as a key mediator of the effects of perceived hyperoxia (6). Here we present first evidence demonstrating the induction of p21 in post-reoxygenated recovering myocardial tissue. Other responses to perceived hyperoxia include transforming growth factor-␤ activation and differentiation of fibroblasts to myofibroblasts (6). Both of these have been identified in the myocardial tissue recovering from reoxygenation injury (16,31). More recent studies present transforming growth factor-␤ as an important mediator of postreperfusion healing (32), consistent with our proposed role of perceived hyperoxia in facilitating tissue remodeling. That low oxygen ambience serves as a cue to trigger angiogenesis is a well accepted notion. Our studies establish that the sensing of oxygen environment is not limited to hypoxia. Our studies demonstrate that in addition to being a trigger for injury as is widely recognized, reoxygenation insult has a built-in component of tissue remodeling induced by perceived hyperoxia. It is plausible that in the event of minor injuries that go unnoted in our daily lives, such remodeling serves to heal as a physiological reparative mechanism. Understanding of this physiological healing response may offer novel opportunities to manage more intense reoxygenation injuries.