Progressive histone alterations and proinflammatory gene activation: consequences of heme protein/iron-mediated proximal tubule injury

Abstract

Rhabdomyolysis (Fe)-induced acute renal failure (ARF) causes renal inflammation, and, with repetitive insults, progressive renal failure can result. To gain insights into these phenomena, we assessed the impact of a single episode of glycerol-induced rhabdomyolysis on proinflammatory/profibrotic [TNF-α, monocyte chemoattractant protein-1 (MCP-1), and transforming growth factor-β1 (TGF-β1)] gene expression and the time course of these changes. CD-1 mice were studied 1–7 days after glycerol injection. Normal mice served as controls. RNA polymerase II (Pol II) binding to the TNF-α, MCP-1, and TGF-β1 genes, “gene-activating” histone modifications [histone 3 lysine 4 (H3K4) trimethylation (H3K4m3) and histone 2 variant H2A.Z], and cognate mRNA levels were assessed. Results were contrasted to changes in anti-inflammatory heme oxygenase-1 (HO-1). Glycerol produced severe ARF (blood urea nitrogen ∼150–180 mg/dl) followed by marked improvement by day 7 (blood urea nitrogen ∼40 mg/dl). Early increases in TNF-α, MCP-1, and TGF-β1 mRNAs, Pol II gene binding, and H3K4m3/H2A.Z levels were observed. These progressed with time, despite resolution of azotemia. Comparable early HO-1 changes were observed. However, HO-1 mRNA normalized by day 7, and progressive Pol II binding/histone alterations did not occur. Fe-mediated injury to cultured proximal tubule (HK-2) cells recapitulated these in vivo results. Hence, this in vitro model was used for mechanistic assessments. On the basis of these studies, it was determined that 1) the H3K4m3/H2A.Z increases are early events (i.e., they precede mRNA increases), 2) subsequent mRNA elevations reflect transcription, not mRNA stabilization (actinomycin D assessments), and 3) increased transcription, per se, helps sustain elevated H2A.Z levels. We conclude that 1) Fe/glycerol-induced tubular injury causes sustained proinflammatory gene activation, 2) decreasing HO-1 expression, as reflected by mRNA levels, may facilitate this proinflammatory state, and 3) gene-activating histone modifications are early injury events and progressively increase at selected proinflammatory genes. Thus they may help sustain a proinflammatory state, despite resolving ARF.

iron-mediated oxidative stress can be a critical determinant of diverse forms of acute renal failure (ARF) and chronic renal disease (2, 3, 28, 29, 35). Extrarenal (e.g., filtered myoglobin) (43) or intrarenal (e.g., cytochrome P-450s) (2, 3) Fe sources can be involved. Once liberated into a “catalytic” form, free radical generation culminates in lipid, protein, and DNA oxidation, organellar dysfunction (e.g., in mitochondria) (30), and, ultimately, apoptotic and/or necrotic cell death. “Proof of concept” for this sequence of events has been gleaned from the glycerol model of rhabdomyolysis-induced ARF. After the induction of myohemoglobinuria, proximal tubular cell heme Fe loading and oxidative injury culminate in marked tubular damage and severe ARF (28, 43). Attenuation of this ARF model by Fe-chelating agents (e.g., deferoxamine) and by antioxidants underscores the pathogenic role of Fe (42). Because of the pathogenic role of Fe in heme protein-induced ARF, it is possible to use clinically applicable Fe-containing compounds, e.g., FeSO₄ or intravenous Fe formulations [e.g., Fe sucrose (FeS) or Fe gluconate (FeG)], to “model” this form of injury in vitro (42–46).

In addition to direct free radical-mediated cytotoxicity, Fe can also indirectly contribute to tissue damage by activating secondary proinflammatory pathways. For example, many proinflammatory [e.g., TNF-α and monocyte chemoattractant protein-1 (MCP-1)] and profibrotic [e.g., transforming growth factor-β1 (TGF-β1)] cytokines are highly redox sensitive and, thus, are induced by prooxidant states (9, 31, 38–40). The resultant inflammatory response serves as a “positive-feedback loop,” evoking ongoing tissue damage beyond the initial injury phase (37). Conversely, Fe can also stimulate countervailing anti-inflammatory pathways, most notably, the induction of the heme-degrading enzyme heme oxygenase 1 (HO-1), which has potent antioxidant properties (27, 31, 38, 39). Indeed, it may be the balance between these pro- and anti-inflammatory influences that ultimately determines whether ongoing tissue injury results.

Increased gene transcription has generally been assumed to be responsible for Fe-induced increases in renal TNF-α, MCP-1, TGF-β1, and HO-1 protein levels, because corresponding increases in their cognate mRNAs have been observed (15, 18, 24–26, 28, 41, 44). However, these observations leave a number of important issues unresolved. 1) Is it possible that posttranscriptional mRNA stabilization helps support TNF-α, MCP-1, TGF-β1, and HO-1 mRNA levels after renal injury by decreasing their degradation rates (8, 13, 20)? 2) If increased transcription is a primary event in TNF-α, MCP-1, TGF-β1, and HO-1 mRNA induction, is there a corresponding increase in RNA polymerase II (Pol II) binding to their cognate genes? This would be expected, given that Pol II is the critical enzyme that drives transcription (16, 21). 3) It has recently been demonstrated that renal injury may induce gene-activating histone modifications (24–26, 45). By loosening chromatin structure, increased Pol II binding results (17, 19, 22). However, whether heme Fe-mediated oxidant stress evokes such histone changes has not been addressed. 4) If histone modifications are induced by Fe-mediated oxidant injury, are these simply delayed reflections of that injury or potential determinants of it? Also, are such changes relatively durable, or are they merely transient postinjury events? 5) Gene-activating histone alterations can increase transcription rates and/or be secondary consequences of it (e.g., due to DNA unwinding with secondary nucleosomal repositioning) (4–7, 10 22, 23, 36). This relationship within injured renal cortex has not been defined. The following study was undertaken to gain insights into each of these issues. Toward these ends, the glycerol-ARF model and direct Fe (FeS/FeG)-mediated HK-2 cell injury models were employed.

METHODS

Glycerol Model of Rhabdomyolysis-Induced ARF

ARF model.

Male CD-1 mice (35–45 g body wt; Charles River Laboratories, Wilmington, MA), maintained under routine vivarium conditions with free access to food and water, were used for all in vivo experiments. The protocols were approved by the Institutional Animal Care and Use Committee. Hypertonic glycerol (50% solution, 10 ml/kg im) was injected in equally divided doses into each hindlimb of mice under brief isoflurane anesthesia. Littermate-matched mice, maintained under the same vivarium conditions but without glycerol injection, served as controls. At 1, 2, or 7 days after glycerol injection, the mice (and their controls) were deeply anesthetized with pentobarbital sodium (50 mg/kg ip), the abdominal cavity was opened, and a blood sample was obtained by venipuncture from the inferior vena cava for blood urea nitrogen (BUN) analysis. Both kidneys were then immediately resected and iced, and the cortices were dissected. At least seven glycerol-treated and seven control mice were studied at each time point.

Kidney Tissue Analysis

mRNA analyses.

Total RNA was extracted from renal cortical samples using the RNeasy Plus kits according to the manufacturer's instructions (Qiagen, Valencia, CA) (45). The mRNAs for TNF-α, MCP-1, TGF-β1, and HO-1 were assessed in control and glycerol-treated kidneys on days 1, 2, and 7 by RT-PCR (45, 47, 49) and expressed relative to simultaneously determined GAPDH mRNA (multiplexed reactions).

Chromatin immunoprecipitation assay.

As a gauge of in vivo gene transcription, Pol II binding to exon 1 of the TNF-α, MCP-1, TGF-β1, and HO-1 genes was assessed by chromatin immunoprecipitation (ChIP) assay, as previously described (32, 33, 45). Briefly, chromatin samples were obtained from fragments of formalin-fixed cortical tissues and immunoprecipitated with anti-Pol II antibody (catalog no. GTX25408, Gene Tex, Irvine, TX), the immunoprecipitates were subjected to real-time (quantitative) PCR to quantify exon 1 sequences, and the results are expressed relative to the amount of chromatin protein added to the reaction.

As a gauge of histone modifications at these four target genes, methylation of histone 3 lysine 4 (H3K4) was assessed using an antibody that recognizes mono-, bi-, and trimethylation (H3K4m1,2,3; catalog no. 04-791, Millipore, Temecula, CA). To determine whether these results were associated with a specific increase in H3K4m3, a gene-activating modification, H3K4m3 was probed in glycerol-treated and control kidney chromatin samples at day 7 using a specific antibody directed at this moiety (45). Finally, levels of the H2 variant H2A.Z, which is also associated with active gene transcription, was assessed by ChIP analysis (26, 45).

Cell Culture Experiments

Fe-mediated oxidant stress is the prime mediator of the glycerol-induced myohemoglobinuric model of ARF (28, 43). Hence, the following cell culture experiments were undertaken to directly assess the impact of Fe-mediated injury on proximal tubular cell TNF-α, MCP-1, TGF-β1, and HO-1 gene expression and associated chromatin responses.

mRNA assessments.

Human proximal tubular (HK-2) cells, maintained in T75 Costar flasks in keratinocyte serum-free medium and supplemented with pituitary extract (34), were used for all cell culture experiments. After trypsinization, they were seeded into six-well Costar plates. At ∼6 h after seeding, the wells were divided equally into three groups: 1) continued control incubation in keratinocyte serum-free medium, 2) incubation with FeG (Ferrlecit, Watson Pharmaceuticals, Corona, CA), or 3) incubation with FeS (Venofer, American Regent, Shirley, NY). The elemental Fe dose used with both preparations was 100 μg/ml. These experiments were repeated on three separate occasions. After 18-h incubations, the cells were recovered by scraping, and total RNA was extracted with the RNeasy Plus kit (see above) and assayed for TNF-α, MCP-1, TGF-β1, and HO-1 mRNAs by RT-PCR (26, 46, 47). The results are expressed relative to simultaneously obtained GAPDH product.

mRNA stability assessments: impact of actinomycin D.

Cells were incubated for 18 h in six-well plates under control conditions or after the addition of FeS or FeG (see above). After the 18-h exposures, the cells in each group were incubated for an additional 6 h with or without actinomycin D (to inhibit new mRNA synthesis; 2 μg/ml). The cells were then harvested for mRNA analysis. The percentage of mRNA remaining in actinomycin D-incubated cells is expressed as a percentage of that in the paired groups of non-actinomycin D-incubated cells.

Fe effects on HK-2 cell Pol II binding and H3K4m3 and H2A.Z expression.

HK-2 cells were cultured as described above and subjected to overnight FeS or control incubations. On the following morning, chromatin was isolated and probed for Pol II, H3K4m3, and H2A.Z at exon 1 of the TNF-α, MCP-1, and TGF-β1 genes (26, 45). Unlike the in vivo experiments in which HO-1 changes were assessed at exon 1, the short exon 1 length of human HO-1 required assessments at the longer exon 3 to obtain suitable quantitative PCR results.

Impact of transcription on H2A.Z and H3K4m3 expression: exploration with actinomycin D.

Changes in histone expression can be a cause or a result of increased gene transcription (6, 7, 10, 11, 14, 17, 19). Thus we assessed whether transcription, per se, impacts Fe-induced H2A.Z and H3K4m3 changes at the four target genes. Four groups of HK-2 cells were prepared (n = 3 per treatment): 1) overnight control incubation, 2) incubation with actinomycin D, 3) incubation with FeS, or 4) incubation with FeS + actinomycin D. On the following morning, chromatin was extracted and assayed for H2A.Z and H3K4m3 at the four test genes, as described above.

Temporal relationship between Pol II recruitment, histone modifications, and mRNA increases.

All the above cell culture experiments were conducted when Fe had already increased the levels of the four test mRNAs (18 h). We next tested whether histone modifications could be detected “early,” i.e., before the above-mentioned mRNA increases. Cells were maintained under control or FeS incubation conditions for 6 h. Then the cells were extracted for mRNA and ChIP analyses, as described above (n = 4 for mRNAs, n = 6 for ChIP analyses). ChIP data were analyzed as a percentage of input [(detected amount of target exon ÷ total target exon added) × 100%] (26), because a constant amount of chromatin protein was added in these short-term experiments.

Cell viability assessments.

To assess the impact of FeS incubations on cell viability, HK-2 cells were plated into 24-well Costar plates and incubated for 18 h under control conditions or with 100 μg/ml Fe, as described above. At the completion of the incubation, lethal cell injury was assessed by determining the percentage of lactate dehydrogenase release (n = 4) or by measuring cellular uptake of the tetrazolium dye 3(4,5-dimethylthiazol-2-yl)2,5-diphenyltetrazolium bromide (MTT), as previously described (n = 4) (34, 48). MTT uptake directly correlates with the number of viable cells; conversely, reduced MTT uptake reflects a decrease in viable cell number (due to an antiproliferative response or apoptotic/necrotic cell death).

Calculations and Statistics

Values are means ± SE. Statistical analyses were performed by unpaired Student's t-test. If multiple comparisons were made, Bonferroni's correction was applied. In each of the ChIP analyses, Pol II levels or histone marks at a negative control gene (β-globin, exon 1, for the in vivo experiments; β-actin, exon 1, for the HK-2 cell experiments) were also assessed (24–26).

RESULTS

In Vivo Experiments

Severity of ARF.

Glycerol injection induced severe ARF, as denoted by severe azotemia on days 1 and 2 [BUN 156 ± 15 and 194 ± 12 mg/dl, respectively, P < 0.001 vs. control (i.e., 28 ± 1 mg/dl)]. However, by day 7, a significant recovery of renal function was observed (BUN 41 ± 9 mg/dl, P = not significant vs. control).

TNF-α, MCP-1, TGF-β1, and HO-1 mRNAs and Pol II Binding to Cognate Genes

mRNAs.

As shown in Figs. 1–3, progressive increases in TNF-α, MCP-1, and TGF-β1 mRNAs were observed over the course of the 1-wk experiments. Thus, by day 7, which corresponded with substantial recovery of renal function, TNF-α, MCP-1, and TGF-β1 mRNA values were ∼10, 60, and 5 times higher than their time-matched control values. Anti-inflammatory HO-1 mRNA was also increased at 24 and 48 h (Fig. 4). However, by day 7, HO-1 mRNA had returned to control levels. Thus the proinflammatory mRNAs were theoretically left relatively unopposed.Figs.2

Fig. 1.

Time course of renal cortical TNF-α mRNA levels and RNA polymerase II (Pol II) binding to the TNF-α gene. After glycerol injection, TNF-α mRNA was markedly elevated and continued to increase over the 7 days of observation. A corresponding progressive increase in Pol II binding to exon 1 of the TNF-α gene was also observed. Open bars, controls; solid bars, glycerol-treated mice.

Fig. 3.

Time course of renal cortical transforming growth factor-β1 (TGF-β1) mRNA levels and Pol II binding to the TGF-β1 gene. Similar to TNF-α and MCP-1, TGF-β1 mRNA levels continued to increase over the 7 days after glycerol injection. Results were paralleled by progressive increases in Pol II binding to exon 1 of the TGF-β1 gene. Open bars, controls; solid bars, glycerol-treated mice.

Fig. 4.

Time course of renal cortical heme oxygenase-1 (HO-1) mRNA expression and Pol II binding to the HO-1 gene. At 1 and 2 days after glycerol injection, HO-1 mRNA was markedly increased compared with control values. By day 7, mRNA levels had returned to control values. Increased Pol II binding was also observed at each time point. Unlike the Pol II binding results for TNF-α, MCP-1, and TGF-β1, no further increase in Pol II binding to the HO-1 gene was observed on day 7. Open bars, controls; solid bars, glycerol-treated mice. NS, not significant.

Fig. 2.

Time course of renal cortical monocyte chemoattractant protein-1 (MCP-1) mRNA expression and Pol II binding to the MCP-1 gene. MCP-1 mRNA was markedly increased by 1 day after glycerol injection and continued to increase over the course of the experiments. mRNA increases were paralleled by a progressive rise in Pol II binding to exon 1 of the MCP-1 gene. Open bars, controls; solid bars, glycerol-treated mice.

Pol II binding.

The time courses of Pol II binding to the start exon of each of the above-mentioned genes are depicted in Figs. 1–4. Progressive increases in Pol II binding were observed at the TNF-α, MCP-1, and TGF-β1 genes (Figs. 1–3), thereby paralleling the levels of their cognate mRNAs, as described above. Increased Pol II binding to HO-1 exon 1 was also observed at each time point (Fig. 4). However, in contrast to TNF-α, MCP-1, and TGF-β1 genes, Pol II binding to HO-1 was not progressive over the 7-day experiments.

Changes in Histone Expression at the TNF-α, MCP-1, TGF-β1, and HO-1 Genes

Histone variant H2A.Z.

As shown in Figs. 5–8, by 24–48 h after glycerol injection, H2A.Z levels approximately doubled at the start exon of each of the four genes. Despite the resolving azotemia, these elevated levels persisted (or, in the case of TNF-α and MCP-1, increased) throughout the 7 days of observation.Figs.6–7

Fig. 5.

Time course of histone 2 variant H2A.Z and histone 3 lysine 4 (H3K4) methylation at exon 1 of the TNF-α gene. A progressive increase in levels of H2A.Z was observed 1, 2, and 7 days after glycerol injection. H3K4 was probed with an antibody that recognizes mono-, di-, and trimethylation, and a progressive increase in the sum of these 3 moieties was observed over the 7 days of observation. Open bars, controls; solid bars, glycerol-treated mice.

Fig. 8.

Time course of H2A.Z expression and H3K4 methylation at the HO-1 gene. Significant increases in H2A.Z and H3K4m1,2,3 were observed 1 and 7 days (but not 2 days) after glycerol injection. Open bars, controls; solid bars, glycerol-treated mice.

Fig. 6.

Time course of H2A.Z and H3K4 methylation at exon 1 of the MCP-1 gene. Similar to TNF-α, progressive increases in H2A.Z and H3K4 methylation were observed after glycerol treatment at the MCP-1 gene. Open bars, controls; solid bars, glycerol-treated mice.

Fig. 7.

Time course of H2A.Z and H3K4 methylation at exon 1 of the TGF-β1 gene. At each time point, a significant increase in H2A.Z and H3K4 methylation was observed in the glycerol-treated animals at the start exons of the TGF-β1 gene. Unlike TNF-α and MCP-1, increases at TGF-β1 reached maximal or near-maximal levels on day 1 and were not progressive thereafter. Open bars, controls; solid bars, glycerol-treated mice.

H3K4 methylation.

As a broad screen for H3K4 methylation, an antibody that recognizes mono-, di-, or trimethylation (H3K4m1,2,3) was used to probe exon 1 of the four test genes, and the results are depicted in Figs. 5–8. The results were highly analogous to those described above for H2A.Z: 1) increases were observed at each test gene by day 1; 2) increases were also seen at days 2 and 7, and 3) in the case of TNF-α and MCP-1, progressive increases were observed over the course of the 7-day experiment.

H3K4 trimethylation at target genes.

Given that increased amounts of H3K4 methylation (i.e., some combination of mono-, di-, or trimethylation) were observed at each of the target genes, we next assessed whether at least some of this increase reflected trimethylation, which facilitates gene transcription. These experiments were restricted to day 7, inasmuch as this was the time of maximal H3K4m1,2,3 expression. As shown in Fig. 9, substantial increases in H3K4m3 were observed at exon 1 at each of the four test genes.

Fig. 9.

H3K4m3 7 days after glycerol treatment. To assess whether H3K4m3, in particular, contributed to the overall increase in H3K4 methylation observed with the anti-H3K4m1 + m2 + m3 antibody, samples were probed with an antibody specific for H3K4m3 (a gene-activating histone mark). Assessments were restricted to kidney samples obtained on day 7. H3K4m3 was increased at each of the 4 genes. Open bars, controls; solid bars, glycerol-treated mice.

β-Globin gene assessments.

Relatively low levels of Pol II binding were observed at exon 1 of the β-globin gene, approximating 10–20% of levels in the four test genes. This low level of binding presumably reflects that β-globin within kidney is a “silent” gene. Regarding H2A.Z and H3K4 methylation marks, no significant differences were observed between the control and glycerol-treated mice at 1, 2, or 7 days after glycerol injection, with a single exception: an 80% increase (P < 0.05 vs. controls) 7 days after glycerol injection.

Cell Culture Experiments

18-h mRNA responses to Fe challenge.

As shown in Fig. 10, after 18-h exposures, FeG and FeS increased the mRNAs for HO-1, MCP-1, TNF-α, and TGF-β1, thereby recapitulating the in vivo postglycerol results. The mRNA increases were greater with FeS than FeG, which is consistent with prior observations that FeS is more readily endocytosed by HK-2 cells than is FeG (48).

Fig. 10.

mRNA levels in HK-2 cells incubated for 18 h under control conditions or subjected to 18 h of Fe [Fe gluconate (FeG) or Fe sucrose (FeS)] challenge. Both Fe challenges raised each of the mRNA levels. The increase was greater with FeS than FeG, consistent with the fact that FeS gains greater intracellular HK-2 access than does FeG.

Assessments of mRNA stability after 18 h of FeG and FeS treatment.

Figure 11 depicts the percentages of HO-1, MCP-1, TNF-α, and TGF-β1 mRNAs that remained in control and FeG- and FeS-treated HK-2 cells after 6 h of actinomycin D exposure. Two findings are noteworthy: 1) there were marked differences in the amounts of mRNA reductions with actinomycin D exposure: in the control incubated cells, ∼95%, ∼50%, ∼25%, and ∼10% reductions were observed in TNF-α, HO-1, MCP-1, and TGF-β1 mRNAs, respectively (indicating progressively longer half-lives); and 2) neither FeG nor FeS exposure altered these mRNA reductions. Stated differently, neither Fe formulation prolonged mRNA half-lives, i.e., increased their stability.

Fig. 11.

Effects of Fe on mRNA stability in HK-2 cells incubated overnight under control conditions or with FeG or FeS. Half of the cells in each group were then treated for 6 h with actinomycin D (AD) to inhibit new mRNA synthesis. Thus, changing mRNA levels in the presence of actinomycin D reflects mRNA degradation rates. Percentage of mRNA remaining in actinomycin D-treated cells was then calculated as follows: (mRNA in actinomycin-treated cells ÷ amount of mRNA in control incubated cells) × 100%. Extent of mRNA reductions in control cells (C) greatly varied (TNF-α manifested the greatest decline, followed in order by HO-1, MCP-1, and TGF-β1 mRNAs). In no instance, did Fe pretreatment affect these results (indicating no Fe-mediated mRNA stabilization).

Fe effects on Pol II recruitment: 18-h assessments.

The FeS-induced increases in TNF-α, MCP-1, HO-1, and TGF-β1 mRNAs were associated with increased Pol II recruitment to each of these genes (Fig. 12). Thus, along with the actinomycin D data, this finding further substantiates the idea that the mRNA elevations reflect increased gene transcription.

Fig. 12.

Pol II binding to TNF-α, MCP-1, TGF-β1, and HO-1 genes in HK-2 cells incubated overnight under control conditions or with FeS. In each instance, Fe significantly increased Pol II binding to each test gene.

FeS effects on H3K4m3 and H2A.Z expression: 18-h assessments.

A correlate of the increase in Pol II binding to the TNF-α, MCP-1, and HO-1 genes consisted of significant increases in modified histone H3 (H3K4m3; Fig. 13) and the histone variant H2A.Z (Fig. 14). Increased expression of H3K4m3 and H2A.Z was also seen at the TGF-β1 gene, although these changes did not achieve statistical significance (P < 0.15).

Fig. 13.

H3K4m3 levels at TNF-α, MCP-1, TGF-β1, and HO-1 genes in HK-2 cells. Cells were treated as described in Fig. 12 legend and then assayed for H3K4m3 levels. Fe significantly increased the H3K4m3 histone mark compared with control incubated cells at each gene.

Fig. 14.

H2A.Z levels at TNF-α, MCP-1, TGF-β1, and HO-1 genes in HK-2 cells. Cells were treated as described in Fig. 12 legend and then assayed for H2A.Z levels at each test gene. Fe significantly increased the H2A.Z histone variant at each gene compared with control incubated cells.

β-Actin gene assessments.

Fe treatment did not significantly increase Pol II, H2A.Z, or H3K4m3 at the β-actin gene, which served as a negative control.

Effects of Transcription on Fe-Induced H2A.Z and H3K4m3 Expression

H2A.Z.

As shown in Fig. 15, actinomycin D treatment reduced H2A.Z levels by ≥50% at the TNF-α, MCP-1, and TGF-β1 gene in the absence and presence of the Fe challenge. In contrast, actinomycin D had no significant effect on H2A.Z at the HO-1 gene in the presence or absence of Fe. Thus these data imply that active transcription induced or was a prerequisite for H2A.Z increases at the proinflammatory, but not the anti-inflammatory, HO-1 gene.

Fig. 15.

Effects of actinomycin D on HK-2 cell H2A.Z expression in the presence and absence of FeS. Under control incubation conditions, actinomycin D reduced H2A.Z expression at TNF, MCP-1, and TGF-β1 genes by approximately one-half. Actinomycin also significantly blunted the extent to which the Fe challenge raised H2A.Z expression at these 3 genes. Conversely, actinomycin failed to suppress control H2A.Z levels at the HO-1 gene, although there was a trend (P < 0.07) toward H2A.Z suppression in the presence of Fe.

H3K4m3.

In contrast to H2A.Z, actinomycin D had no significant effect on H3K4m3 expression at any of the test loci. This was true under control incubation conditions or with FeS exposure (data not shown). This lack of effect on H3K4m3 indicates that the actinomycin D-induced changes on H2A.Z were relatively specific (vs. a nonspecific actinomycin effect).

Temporal Relationship Between Pol II, Recruitment, Histone Modifications, and mRNA Increases

Pol II.

As shown in Table 1, within 6 h of Fe addition to HK-2 cells, a significant increase in Pol II binding was observed at the TNF-α, MCP-1, and HO-1 genes. An increased amount of Pol II was also detected at the TGF-β1 gene, although this was at the borderline of statistical significance (P = 0.06).

Table 1.

Effects of short-term (6 h) FeS incubation on Pol II, H2A.Z, and H3K4m3 at four test genes and on their cognate mRNA levels

	Pol II	H2A.Z	H3K4m3	mRNA
TNF-α
Controls	1.0 ± 0.2	2.3 ± 0.2	32 ± 1	1.0 ± 0.03
FeS	1.8 ± 0.2	3.5 ± 0.3	42 ± 3	0.5 ± 0.1
P	<0.03	<0.01	<0.01	<0.001
MCP-1
Controls	1.2 ± 0.1	7.3 ± 0.3	46 ± 2	1.3 ± 0.1
FeS	2.1 ± 0.2	9.6 ± 0.7	59 ± 4	1.2 ± 0.01
P	<0.05	<0.02	<0.015	NS
TGF-β1
Controls	2.3 ± 0.4	6.2 ± 0.6	43 ± 2	3.3 ± 0.5
FeS	3.4 ± 0.4	9.1 ± 0.8	60 ± 4	3.2 ± 0.1
P	0.06	<0.01	<0.001	NS
HO-1
Controls	1.0 ± 0.1	1.5 ± 0.1	26 ± 1	0.9 ± 0.05
FeS	2.2 ± 0.2	2.3 ± 0.2	38 ± 2	13.3 ± 0.8
P	<0.001	<0.01	<0.001	<0.001

Values are means ± SE. HK-2 cells were maintained under control conditions or challenged for 6 h with Fe sucrose (FeS), and RNA polymerase II (Pol II) binding and histone 3 lysine 4 trimethylation (H3K4m3) and histone 2 variant H2A.Z levels were analyzed by chromatin immunoprecipitation and expressed as percentage of DNA input (26). mRNA-to-GAPDH ratios were also determined. FeS induced significant increases in Pol II recruitment and modified histone enrichment at each of the 4 target genes. In the case of TNF-α, monocyte chemoattractant protein-1 (MCP-1), and transforming growth factor-β1 (TGF-β1) genes, increases in Pol II binding and H3K4m3 preceded detectable increases in their cognate mRNAs. Conversely, FeS induced an early increase in heme oxygenase-1 (HO-1) mRNA, which corresponded to increased Pol II/histone changes. NS, not significant.

Histones.

After the 6-h Fe exposure, there were significant increases in H2A.Z and H3K4m3 levels at each of the four assessed genes. Thus these changes corresponded to increased Pol II binding.

mRNAs.

Although increases in Pol II binding and histone modifications were observed at the TNF-α, MCP-1, and TGF-β1 genes, there were no corresponding increases in their cognate mRNA levels. This indicates that these changes preceded complete production and processing of the mRNAs (i.e., the histone changes were early events). Conversely, an ∼13-fold increase in HO-1 mRNA was observed after 6 h of Fe treatment, indicating rapid HO-1 gene transcription.

FeS effects on cell viability.

The above-noted changes in HK-2 cell mRNA, Pol II, and histone expression developed with only minimal lethal cell injury as assessed by percentage of lactate dehydrogenase release (14 ± 0.1 and 7 ± 0.1% for FeS and controls, respectively, P < 0.001). The more sensitive MTT assay showed no apparent evidence of lethal cell injury: identical MTT uptake values were observed for the control and the FeS group (0.15 ± 0.002 colorimetric units for both groups).

DISCUSSION

It has previously been established that the glycerol model of ARF has as a pathogenic component a secondary renal inflammatory response (37). Additionally, Nath et al. (29) demonstrated that when rats are challenged with repetitive glycerol injections, progressive interstitial inflammation, fibrosis, and renal insufficiency result. A surprising finding of the present study was the robustness and durability of this proinflammatory state. At 24–48 h after glycerol injection, corresponding to the apparent height of renal dysfunction (BUN ∼150–200 mg/dl), approximately fourfold elevations of TNF-α, MCP-1, and TGF-β1 mRNAs were observed. Subsequently, TNF-α, MCP-1, and TGF-β1 mRNA increased progressively, despite resolving ARF. This suggests that these three genes, first activated by the initial injury phase, sustained molecular changes that supported ongoing gene activity. As will be discussed below, injury-induced stable alterations in gene-activating histone modifications could help explain this sequence of events. In contrast to TNF-α, MCP-1, and TGF-β1 mRNAs, the initial HO-1 mRNA elevations completely normalized by day 7 (as would have been predicted given the resolving ARF). Thus, when all four mRNAs are considered together, a state of increasing proinflammatory (TNF-α and MCP-1)/profibrotic (TGF-β1) mRNAs, with a reciprocal change in anti-inflammatory HO-1 mRNA, suggests a shifting balance toward a proinflammatory state.

An increase in tissue mRNA levels can arise from increased transcription and/or posttranscriptional mRNA stabilization, thereby prolonging mRNA half-life (1, 8, 13, 20). Thus it is difficult to ascertain which of these two processes initiated and maintained the in vivo mRNA elevations, as discussed above. Assessments of Pol II binding to target genes can help resolve this issue, given that Pol II is the enzyme that drives transcription and its binding to gene loci serves as a semiquantitative index of transcription rates (16, 21, 24, 26). Given these considerations, we assessed Pol II binding to each of the four target genes. At 1 and 2 days after ischemia, comparable increases in Pol II binding were observed at the start exons of the TNF-α, MCP-1, TGF-β1, and HO-1 genes. The degree of Pol II binding to TNF-α, MCP-1, and TGF-β1 increased with time, thereby corresponding to the stepwise increases in their mRNAs. In the case of HO-1, Pol II binding appeared to peak at day 2 but remained elevated by day 7, when HO-1 mRNA levels had returned to control values. The reason for this apparent discrepancy in Pol II-HO-1 gene binding, despite a normalized HO-1 mRNA, remains unknown. However, it implies a downstream “block” in mRNA transcription, e.g., due to counterregulatory histone(s) changes, somewhere along the gene.

Two important caveats need to be considered when the above-described Pol II binding results are interpreted. 1) Although the glycerol-ARF model primarily induces injury to proximal tubules, the Pol II data were gathered using whole renal cortex. Thus neither the Pol II results nor, for that matter, the mRNA changes can definitively be attributed to proximal tubule cell events. 2) Although the Pol II data imply that increased transcription plays the dominant role in the mRNA increases, it remains possible that an increase in mRNA stabilization also contributed. Therefore, to gain more definitive information regarding these two potential caveats, cultured proximal tubule cells were exposed to an 18-h Fe challenge to 1) confirm that oxidant injury does, indeed, increase TNF-α, MCP-1, TGF-β1, and HO-1 gene expression in proximal tubule cells and 2) directly test whether such increases might arise from increased mRNA stability. After 18 h of incubation with FeS or FeG, significant increases were observed in each of the four test mRNAs. The extent of these increases was greater with FeS than with FeG, undoubtedly reflecting the greater uptake of FeS by HK-2 cells (46). When control and Fe-treated cells were subsequently incubated with actinomycin D for 6 h (to stop transcription), the same percent reductions in each of the four test mRNAs were observed in the Fe-treated and control groups. This indicates comparable mRNA degradation rates and, hence, a lack of mRNA stabilization. Thus the present studies clearly support the conclusions derived from the in vivo experiments: 1) Fe does, indeed, evoke TNF-α, MCP-1, TGF-β1, and HO-1 mRNA increases within proximal tubules, and 2) increased gene transcription, rather than mRNA stabilization, is the dominant mechanism that produces these results.

Gene activation in response to extra- or intracellular stimuli is often associated with specific changes in chromatin structure. These changes can serve as markers, as well as mediators, of gene activation. The latter action arises from the ability of gene-activating histone alterations to loosen chromatin structure, thereby facilitating transcription factor recruitment and Pol II binding to promoter and transcribed regions (10–12, 14, 23). A prominent histone alteration in response to activating stimuli is methylation of the fourth lysine of histone H3, culminating in the production of mono-, di-, or trimethylated H3K4 (H3K4m1, H3K4m2, or H3K4m3, respectively). H3K4m3 is one of the best-studied covalent modifications that denote, and/or facilitate, an increased transcriptional state (10–13, 23, 26). We previously documented increased H3K4m3 levels at the 3-hydroxy-3-methylglutaryl-CoA reductase gene in postischemic ARF (25), and this change corresponded with increased gene activity (cholesterol synthesis). We also previously observed increased levels of H3K4m3 at proinflammatory genes in the setting of maleate (ATP depletion-induced) ARF (49). To assess whether H3K4 methylation also occurs in an oxidant model of ARF, we probed chromatin samples from glycerol-treated kidneys with an antibody that recognizes H3K4 mono-, di-, and trimethylation. Over a 1-wk time course, increased levels were observed at each of the four test genes, but not at the β-globin (negative control) gene. To test whether this overall increase in H3K4 methylation was specifically associated with an increase in gene-activating H3K4m3, the latter was probed at 7 days of glycerol treatment. Again, increased expression was observed. Finally, to test whether the above-described in vivo changes reflected, at least in part, a direct Fe effect on proximal tubule cells, FeS-treated HK-2 cells were probed for H3K4m3, and, again, significant increases at each of the four test genes were observed. Thus it appears that increased histone methylation is one genomic consequence of Fe-mediated renal injury.

A second type of histone modification, enrichment of specific gene loci with histone variants, can also result from tissue injury. Unlike methylation, which reflects covalent histone modification via enzymatic (methyltransferase) activity, an increase in histone variant expression typically reflects nucleosomal repositioning/histone exchange from other genomic sites. The latter pathway can be a secondary response to Pol II-driven transcription, which relaxes gene structure and, thus, allows repositioning. Enrichment with H2A.Z, a gene-activating H2 histone variant, is one such chromatin modification. Indeed, we previously documented increased H2A.Z expression in the setting of postischemic ARF (45). The present study tested whether Fe-mediated in vitro and/or in vivo injury might produce this same result. Indeed, at 1–2 days after induction of glycerol-mediated ARF, dramatic increases in H2A.Z were observed at each of the four test genes. In the cases of TNF-α and MCP-1 (but not TGF-β1 or HO-1), these increases were progressive with time, as assessed at day 7. Furthermore, Fe-mediated HK-2 cell injury also raised H2A.Z levels. Thus, when these data are interpreted in light of the above-noted H3K4m3 results, we conclude that at least two potential gene-activating histone modifications, arising via two different pathways (methylation and histone exchange), are consequences of Fe-mediated injury and rhabdomyolysis-induced ARF. That they increase with time at the TNF-α and MCP-1 genes, but not at the HO-1 gene, suggests their potential to increase inflammation into, and potentially beyond, the early ARF recovery phase.

Four questions were suggested to us from the above-described findings, and each was subsequently explored with the Fe-challenged HK-2 cell injury model. 1) We questioned whether the histone modifications were simply nonspecific consequences of lethal proximal tubular cell injury, given that marked proximal tubular necrosis is a hallmark of glycerol-induced ARF (8, 9). However, the HK-2 cell culture data seemingly exclude this possibility, given the marked H3K4m3 and H2A.Z increases at each test gene in the absence of tubular cell death (e.g., MTT uptake data). 2) Because histone modifications can be markers, as well as facilitators, of increased gene transcription, we questioned whether the H3K4m3 and H2A.Z increases simply denoted increased TNF-α, MCP-1, TGF-β1, and HO-1 transcriptional states. The answer to this question also appears to be no, because after 6 h of Fe exposure, increases in H3K4m3 and H2A.Z were observed at the TNF-α, MCP-1, and TGF-β1 genes before increases in their corresponding mRNAs. This strongly implies that these histone changes are early chromatin events, and, hence, they are able to participate in early transcriptional activation. 3) Because H3K4m3 or H2A.Z increases can contribute to Pol II recruitment or be a consequence of it, we questioned whether inhibiting Pol II elongation along the test genes might inhibit rising histone levels. To make this assessment, cells were incubated under control and Fe-challenged conditions in the presence and absence of actinomycin D (which was added to block Pol II extension). As shown in Fig. 15, actinomycin D markedly decreased H2A.Z levels at the TNF-α, MCP-1, and TGF-β1 genes, but not at the HO-1 gene, under control conditions and with Fe exposure. In contrast, actinomycin D had no effect on H3K4m3 expression at any of these test genes in the presence or absence of the Fe challenge. Thus these results support the assertion that Pol II activity can, indeed, enhance H2A.Z at proinflammatory genes, most likely as a result of its repositioning from other genomic sites. Conversely, that actinomycin D did not influence H3K4m3 levels at any of the assessed genes is consistent with the concept that injury-induced H3K4m3 elevations are a result of increased methyltransferase activity. 4) Most critical is whether the observed chromatin changes actually facilitated ongoing gene transcription, despite resolving ARF. To answer this question, it would be necessary to inhibit, or reverse, these histone alterations, either singly or in combination, in a gene/exon-specific fashion and then assess whether decreased gene transcription results. To our knowledge, no existing method can induce such site-specific changes. Thus, until such a technique is available, a mechanistic association between specific histone modifications and associated transcription rates remains a hypothesis. However, the observations that increasing TNF-α, MCP-1, and TGF-β mRNAs were correlated with increasing H3K4/H2A.Z levels at their cognate genes and that these histone changes can precede mRNA increases in HK-2 cells suggest a mechanistic link.

In conclusion, this study indicates that rhabdomyolysis-induced ARF evokes progressive increases in renal cortical proinflammatory/profibrotic gene expression, during which time early increases in HO-1 mRNA rapidly resolve. That these changes occur during a period of resolving azotemia suggests that the initial renal insult “sets the stage” for a progressive proinflammatory state. The early induction of, and progressive increases in, gene-activating histone modifications (H3K4m3 and H2A.Z) may contribute to this process via their ability to “open” the chromatin structure and, thus, enhance gene transcription rates. Fe-mediated HK-2 cell injury faithfully recapitulates most of the glycerol-induced ARF findings. Thus it provides a useful tool for probing involved mechanisms. On the basis of such in vitro studies, we conclude that 1) increased gene transcription, not mRNA stabilization, is the dominant mechanism responsible for the observed mRNA increases, 2) gene-activating H2A.Z and H3K4m3 increases are early pathogenic events (i.e., preceding mRNA accumulation), and 3) increased Pol II activity/gene transcription, per se, can enhance H2A.Z levels. The ultimate role of these histone changes in proinflammatory gene expression remains to be defined. Nevertheless, the present data suggest a new paradigm whereby early renal injury evokes gene-activating histone changes, which are self-perpetuating over time, facilitate ongoing gene transcription, and, by facilitating proinflammatory gene expression, may potentially contribute to delayed resolution of ARF or, possibly, to progressive renal disease. If this is true, then this paradigm suggests new potential targets in our efforts to mitigate diverse forms of renal disease.

GRANTS

This work was supported by National Institute of Diabetes and Digestive and Kidney Diseases Grants R37 DK-38432 and R21-DK-083315-01.

DISCLOSURES

No conflicts of interest are declared by the authors.