BRG1 Increases Transcription of Proinflammatory Genes in Renal Ischemia

Ischemia-Reperfusion Induces Increasingly Sustained Expression of TNF-α and MCP-1 mRNA in Renal Cortex

Acute kidney injury increases expression of TNF-α and MCP-1 mRNA in the kidney.^4,10,22,23 We used reverse transcriptase–PCR (RT-PCR) to follow renal cortical TNF-α and MCP-1 mRNA levels soon after that injury and during the subsequent week. By 2 h of reperfusion, there was an increase in the MCP-1 mRNA levels, whereas TNF-α was not yet significantly increased, but at both 3 and 7 d after ischemia, marked elevations of both mRNAs were observed (Figure 1).

Ischemia-Reperfusion Induces Sustained Chromatin Changes and Co-recruitment of BRG1 and Pol II along the TNF-α and MCP-1 Genes in Renal Cortex

The aforementioned ischemia-reperfusion (I/R)-induced increases in renal cortical TNF-α and MCP-1 mRNA levels could reflect enhanced Pol II recruitment to these loci, leading to increased transcription rates.¹⁰ We used the matrix chromatin immunoprecipitation (ChIP) platform^24,25 to profile Pol II levels along these genes (Figures 2 and 3). At 2 h after acute injury, the levels of Pol II at the start (P < 0.003) and end exon (P < 0.05) of TNF-α (Figure 2A) were higher in postischemic kidneys, compared with contralateral controls. Like the mRNA levels (Figure 1), the I/R-induced Pol II changes persisted for at least 1 wk after the injury. The levels of Pol II at a 3′ flanking region 5 kb from the end of the TNF-α gene were low and not different between injured and contralateral kidneys (Figure 2B), indicating the specificity of the ChIP assay. Similar results were obtained along the MCP-1 gene locus (Figure 3). The ChIP analysis of Pol II density along these genes suggests that the I/R-induced TNF-α and MCP-1 mRNA increases reflected, at least in part, increased transcription.

Activation of gene expression is often associated with specific changes in chromatin structure that are important for induction of transcription.^12,26–29 Trimethylation of lysine 4 of histone H3, H3K4m3, is one of the best studied covalent modifications that mark an open chromatin.³⁰ Like Pol II, H3K4m3 levels are highest near the transcription start sites.^12,28,31 We previously described increased H3K4m3 levels along several genes in renal cortex from different mouse models of ARF.^10,25 Recently, an increase in H3K4m3 was also reported in response to hypoxia in hepatocytes.³² In this study, H3K4m3 levels at the first and last exons of the TNF-α gene were higher in postischemic kidneys, compared with contralateral controls (P < 0.003 and P < 0.01; respectively; Figure 2A). Furthermore, H3K4m3 in the intergenic 3′ 5-kb flanking region was very low and not different between the clamped and contralateral kidneys (Figure 2B), serving as a negative internal control. Interestingly, the differences between the injured and control kidneys remained at 3 but not 7 d after I/R. Similar results were obtained along the MCP-1 gene locus (Figure 3). These results suggest that whereas transcription of TNF-α and MCP-1 genes remains enhanced in the injured kidney, the classical H3K4m3 mark of open chromatin does not remain higher. This observation suggests that there is evolution of epigenetic processes to maintain genomic access after injury.

The H2A.Z histone variant is highly enriched at sites of active transcription,³³ and the substitution of H2A.Z for the canonical H2A histone is increased with stimulation of transcription.²⁹ At transcribed loci, the H2A.Z levels are highest at the transcription start site.³³ Two hours after I/R, H2A.Z levels at the TNF-α gene were higher in the I/R kidneys versus their contralateral controls (Figure 2A). Unlike H3K4m3, these differences were sustained for 7 d. The H2A.Z density in the intergenic 3′ 5-kb flanking region were low and not different between the clamped and control contralateral kidneys (Figure 2B; negative internal control). Similar results were obtained along the MCP-1 gene locus (Figure 3). These results provide further evidence that (1) in response to I/R, the chromatin environments encompassing the TNF-α and MCP-1 genes are altered and (2) these differences persist for some (H2A.Z) but not all (H3K4m4) epigenetic marks for at least 1 wk after injury induction.

The ATP-dependent SWI/SNF multisubunit chromatin remodeling complexes are important in the regulation of transcription.^13,29 With respect to our results (Figures 2 and 3) and those obtained from yeast^14,34 to mammals,^29,35 SWI/SNF complexes play a role in the replacement of H2A with H2A.Z. SWI/SNF complexes contain either the BRG1- or the BRM-ATPase catalytic subunit and several BRG-associated factors.¹³ The induction of the erythropoietin gene in response to hypoxia depends on BRG1 recruitment to this locus.²¹ Recruitment of BRG1 is also required for the induction of heme oxygenase 1 gene transcription in response to oxidative stress.²⁰ BRG1 recruitment also plays a role in the transcriptional regulation of inflammatory mediators, including TNF-α^36,37; therefore, we next assessed BRG1 levels along the TNF-α locus in response to I/R. These studies revealed higher levels of BRG1 at the first and last TNF-α exons in injured kidneys, compared with contralateral controls. These effects were observed as early as 2 h after I/R and were sustained for 1 wk. The BRG1 levels in the intergenic 3′ 5-kb flanking region were low and not different between the clamped and contralateral kidneys (Figure 2B). Similar results were obtained along the MCP-1 gene locus (Figure 3). Thus, the time course of increased levels of BRG1 along the TNF-α and MCP-1 loci is similar to that observed for Pol II and H2A.Z. Matrix ChIP observations suggest that similar epigenetic mechanisms underlie the increased transcription of TNF-α and MCP-1 genes in response to renal ischemia (Figure 1). These observations also suggest that the inducible recruitment of BRG1 plays a role in the induction of these genes in response to renal ischemia.

Simulated I/R in Cultured Proximal Tubule Cells Alters Chromatin and Enhances Recruitment of Pol II and BRG1 along TNF-α and MCP-1 Genes

The combination of antimycin A (AA) and 2-deoxyglucose (DOG) depletes intracellular ATP, and this has been used to simulate oxygen deprivation in cultured cells.^23,38 For example, AA treatment stimulates TNF-α mRNA and protein synthesis in HK-2 cells,²³ recapitulating the in vivo proinflammatory mediator expression in response to ischemia. Thus, this system is a potentially useful model to define epigenetic and transcriptional mechanisms that promote cytokine/chemokine expression in response to hypoxic stress. We used matrix ChIP assay^24,25 to compare Pol II, BRG1, and chromatin profiles along TNF-α and MCP-1 genes in untreated and in reversibly ATP-depleted HK-2 cells (AA + DOG for 4 h, followed by 2 h of ATP recovery [AA + DOG washout]; Figure 4). As in the case of renal ischemia (Figures 2 and 3), ATP depletion in HK-2 cells increased Pol II density along both the TNF-α (Figure 4A) and MCP-1 (Figure 4B) genes. This observation suggests that the reversible ATP depletion–induced cytokine/chemokine mRNA increases in HK-2 cells (Figure 5) is, at least in part, transcription dependent. As in the case of in vivo I/R (Figures 2 and 3), the ATP-mediated increase in Pol II density along TNF-α and MCP-1 genes was associated with higher levels of H3K4m3 and H2A.Z at these loci. In addition, there was an injury-mediated recruitment of BRG1 to these genes, which may play a role in the increased expression of TNF-α and MCP-1 in response to ATP depletion. This possibility was tested in the next series of experiments.

Knockdown of BRG1 Attenuates Injury-Induced TNF-α and MCP-1 mRNA Expression in Proximal Tubule HK-2 Cells

We used small interfering RNA (siRNA) knockdown to examine the role of BRG1 in the aforementioned changes (Figure 5). ATP depletion increased BRG-1 mRNA expression. Transfection of cells with BRG1 siRNA decreased BRG1 mRNA levels by approximately 75% in both control (P < 0.004) and ATP-depleted (P < 0.002) cells (Figure 5). In response to ATP depletion, there was a >2× increase in TNF-α (P < 0.02) and MCP-1 (P < 0.02) mRNAs in cells transfected with noncomplementary (NC) siRNA. In cells transfected with siRNA BRG1 for 24 h, the constitutive and inducible levels of both TNF-α and MCP-1 mRNAs were greatly decreased. In contrast, there was no change in β-actin mRNA expression in response to injury, and the levels of β-actin transcripts were not different between NC and BRG1 siRNA-transfected cells. Consistent with previous studies,^{20,29,36,37,39} these results suggest that the BRG1 chromatin remodeler regulates expression of inducible subsets of genes. The increased expression of TNF-α and MCP-1 genes in response to renal ischemia (Figure 1) that is associated with increased co-recruitment of Pol II and BRG1 to these loci (Figures 2 and 3) suggests that BRG1 plays a role in injury-induced transcription. This point was addressed next.

Knockdown of BRG1 Attenuates Injury-Induced Recruitment of Pol II to TNF-α and MCP-1 Genes in Proximal Tubule HK-2 Cells

Reversible ATP depletion caused a more than three-fold increase in the density of BRG1 at the first and last exons of the TNF-α gene (P < 0.001 and P < 0.001, respectively; Figure 6). There was 40 to 80% decrease in BRG1 density at the TNF-α locus in cells transfected with BRG siRNA under baseline conditions. Furthermore, the ATP depletion–mediated BRG1 increase was prevented. Similar effects of BRG1 siRNA were seen at the MCP-1 gene (Figure 7). Together with the renal I/R data (Figures 2 and 3), this experiment indicates that ischemia increases recruitment of the BRG1 chromatin remodeler to the TFN-α and MCP-1 genes in proximal tubule cells.

In agreement with the ChIP results from postischemic kidneys (Figure 2), reversible ATP depletion increased Pol II density at the first (P < 0.08) and last (P < 0.001) exons of the TNF-α gene (Figure 6). BRG1 knockdown reduced both the constitutive and the inducible recruitment of Pol II to the TNF-α gene. Similar results were obtained along the MCP-1 gene (Figure 7). In sum, these results suggest that the increased expression of TNF-α and MCP-1 transcripts in response to reversible ATP depletion in HK-2 cells (Figure 5) is, at least in part, transcriptionally mediated and that this effect depends on the recruitment of BRG1 to these genes. That BRG1 is also recruited to TNF-α and MCP-1 genes in vivo (Figures 2 and 3) suggests that BRG1 plays a role in expression of TNF-α and MCP-1 genes after in vivo renal I/R (Figure 1). Unlike the effects on Pol II, changes in the levels of BRG1 recruitment to the TNF-α (Figure 6) and MCP-1 (Figure 7) genes had no effects on H3K4m3. This suggests a hierarchically ordered series of epigenetic events that enhance chromatin access to transcribing Pol II, as suggested in other studies.^12,14,17,40 The lack of BRG1 effect on the level of H3K4m3 suggests that in this chain of events, H3K4m3 changes may operate upstream and are independent of BRG1.¹⁷

In Vivo I/R Protocol

Male CD 1 mice (Charles River Laboratories, Wilmington, MA; 30 to 35 g), maintained under routine vivarium conditions and subjected to institutional animal care and use committee–approved protocols, were used for all experiments. In brief, mice were anesthetized and subjected to a midline abdominal incision under sterile conditions. Left renal ischemia was induced with an atraumatic microvascular clamp applied to the renal pedicle. After 30 min of vascular occlusion, the clamp was released and reperfusion of the entire kidney was assessed visually (by loss of global cyanosis). A 30-min ischemic insult was selected for study because it induces moderately severe but reversible ischemic kidney damage. At variable times after ischemia, the mice were re-anesthetized and the kidneys were resected and processed for analyses. With the unilateral ischemia protocol, azotemia is not predicted. As previously documented,²³ the right nonischemic, contralateral (control) kidney recapitulates what is seen in sham-operated kidneys and, hence, served as an internal control.

In Vitro “Chemical Ischemia” (ATP depletion) Experimental Protocol

The HK-2 immortalized proximal tubule cell culture maintained in keratinocyte serum-free medium was seeded into T75 flasks, as described previously.³⁸ Culture conditions included 100 U/ml penicillin plus 25 μg/ml streptomycin. For experimentation, the cells were seeded into six-well Costar plates and then studied approximately 18 h later.

To assess the effects of reversible ATP depletion in HK-2 cells, we used AA-induced mitochondrial inhibition (7.5 μM) + DOG-induced (20 mM) inhibition of glycolysis, as described previously.^23,38 After 4 h of AA + DOG exposure, ATP depletion was reversed by removal of the cell culture medium, washing with fresh medium, and then allowing a 2-h recovery period.

siRNA Knockdown Experiments

siRNA knockdowns were done using siRNA Gene Silencing Santa Cruz Protocol (http://www.scbt.com/protocols.html). Briefly, cells were grown in six-well plates to 30 to 50% confluence. Lipofectamine 2000 (Invitrogen, Carlsbad, CA) and either BRG1 siRNA (sc-29827; Santa Cruz Biotechnology, Santa Cruz, CA) or NC siRNA (sc-37007; Santa Cruz Biotechnology) Transfection Reagents was diluted separately with siRNA Transfection Medium (Opti-MEM I; Invitrogen). After 5 min of incubation, the diluted siRNA and Lipofectamine 2000 were combined and the mixture was incubated for another 30 min at room temperature. The mixture was then overlain onto the medium bathing cells in the wells. After 12 to 24 h, the overlay was aspirated and replaced with fresh culture medium for additional 24 to 48 h before mRNA and ChIP experiments.

RNA Extraction

TRIzol was used to purify RNA according to the manufacturer's protocol (Invitrogen). Excess TRIzol was used to separate clearly the RNA and DNA phases. Total extracted RNA was dissolved in 20 μl of sterile water and stored at −80°C. RNA concentrations were measured with a spectrophotometer (Bio Mate3; Thermo Fisher Scientific, Waltham, MA), and the 260/280 ratio of RNA was >1.7.

RT-PCR

First-strand cDNA was synthesized by priming 1 μg of total RNA with 10 μM random hexamers (Promega, Madison, WI) by heating at 65°C for 10 min and snap-cooling on ice. Reverse transcription (37°C for 1 h) was performed in the presence of 10 mM each of dATP, dCTP, dTTP, and dGTP (Invitrogen), 4 μl of 5× first-strand buffer (Invitrogen), 0.1 M dithiothreitol (Invitrogen), 200 U of Maloney-murine leukemia virus reverse transcriptase (Invitrogen), and 20 U of RNase inhibitor (Invitrogen). After the reaction, the sample was heated at 94°C for 5 min and cooled on ice; after addition of 180 μl of water, samples were stored at −80°C. RT cDNA samples were measured either with competitive (Figure 1) or real-time PCR (Figure 5). List of RT-PCR primers is shown in Table 1.

Sonication of Renal Cortex Chromatin

Approximately 25 mg of minced renal cortex was fixed with formaldehyde (final concentration 1.42% in PBS for 15 min; 22°C) and then quenched with 125 mmol/L glycine (5 min, 22°C). The cross-linked tissues were then extensively washed with PBS (4°C). For shearing the chromatin, the washed cross-linked tissue pellets were resuspended in IP buffer.⁴⁹ The shearing was done using either Masonic 3000 microprobe (1 ml of IP buffer, six rounds of sonication power 5, 15 s, on ice)^49,50 or Diagenode Bioruptor (100 μl of IP buffer, 30 rounds for 30 s on/30 s off, high power, 4°C). The suspension was cleared by centrifugation at 12,000 × g (10 min at 4°C), and the supernatant, representing sheared chromatin, was aliquotted and stored at −80°C.

Microplate-Based Matrix ChIP Platform

ChIP assays were done using the Matrix ChIP platform in 96-well polystyrene high-binding capacity microplates as described previously.^24,25,49 ChIP DNA samples were assayed by quantitative PCR. All PCR reactions were run in triplicate. PCR primers were designed using the Primer3 software (http://frodo.wi.mit.edu/). At least four samples were run for each determination. PCR calibration curves were generated for each primer pair from a dilution series of total mouse or human sheared genomic DNA. The PCR primer efficiency curve was fit to cycle threshold versus log (genomic DNA dilutions) using an r² best fit. DNA concentration values for each ChIP and input DNA samples were calculated from their respective average cycle threshold values. Final results are expressed as percentage input DNA.²⁴ Matrix ChIP PCR primers are shown in Table 2 and the list of antibodies in Table 3.

Statistical Analysis

All results are presented as means ± 1 SD. Statistical comparisons were made by unpaired t test (significance judged by P < 0.05). When multiple statistical comparisons were made, the Bonferroni correction was applied.