Chromatin changes trigger laminin genes dysregulation in aging kidneys

Expression of laminin genes in young and old kidneys

To distinguish between common versus genotype-specific changes in laminin gene expression during aging, we used F344 and FBN-F1 rat lines, two established model systems supported by the National Institute of Aging (NIA). As life spans of these two rat lines are substantially different, we compared young animals (4 months old, 4 mo) to animals at the age corresponding to their median life span (50% survival), which is close to 24 months for F344 rats and 32 months for FBN-F1 rats [14]. 28 mo F344 rats (10% survival) were also used to examine progression of changes with aging.

The results of RT qPCR analysis of laminin transcript levels in F344 and FBN-F1 rat kidneys are shown in Figure 1A. Lama2, a4, a5, b2, c1, and c3 were highly expressed in young animals, with no change in old animals. In contrast, laminin genes expressed at very low levels in young animals – Lama3, b3, and c2 – were induced in old animals, with the largest changes seen in the Lamc2 transcript. Increased expression of Lamc2 in old F344 kidneys was also clearly detectable at the protein level (Figure 1B). The increase of Lamc2 in aging kidneys was associated with a matching increase in the density of RNA polymerase II (Pol II) recruited to the gene (Figure 1C), implicating transcriptional activation. Next, we explored the possibility of epigenetic dysregulation driving increased Lamc2 transcription.

Increases in laminin transcript levels in aging kidneys are matched by decreases in the level of H3K27m3 histone modification at the corresponding genes

To define the role of epigenetic processes in the deregulation of laminin gene expression in aging kidneys, we used Matrix ChIP assay [15] with antibodies to several histone modifications. Rat Lamc2 gene is located on chromosome 13 between Nmnat2 and Lamc1 genes (Figure 2A). Within this genomic region, Lamc1 gene is well expressed in young kidneys and shows no changes with aging. In contrast, Lamc2 is a low-abundance transcript in young kidneys, and its expression is significantly induced in old kidneys independent of genetic background (Figure 1A and 2B). Nmnat2 gene expression was not detected in kidneys of any age group (Figure 2B). Previous analysis showed that in different cell types and tissues, Lamc1 gene contains active histone marks, whereas Nmnat2 and Lamc2 reside within a silenced domain [16]. Thus, the Nmnat2-Lamc2-Lamc1 locus seems to be an attractive model of aging-associated chromatin changes in sections of the mammalian genome containing arrays of neighboring silenced and active domains.

ChIP analysis of histone modification profiles revealed little or no changes in H3K4m3 density at Lamc1, Lamc2 and other laminin genes in old compared to young kidneys (not shown). There were also some decreases in H3K9m3, H3K36m3, and H3K79m2 levels along all tested loci (including laminin and control genes) which were not correlated with changes in corresponding transcript levels (not shown). It has previously been reported that aging is associated with the loss of silencing mark H3K27m3 at selected gene loci [17]. In agreement with these observations, the density of this mark was substantially decreased in old kidneys at the induced Lamc2 gene (Figure 2C). Similar decreases were found in both rat lines, indicating that this effect is not strain-specific (Figure 2C, compare 24mo F344 to 32mo FBN-F1). These observations suggest that the loss of H3K27m3 and the upregulation of Lamc2 gene were causally-related events in aging kidneys. Nuclear levels of H3K27m3 modification, as estimated by Western blot analysis, were not altered in F344 and were even increased in FBN-F1 old compared to young kidneys (Figure 3) showing that the loss of this histone mark was gene-specific rather than global.

Given the genomic proximity of Nmnat2, Lamc2 and Lamc1 genes, we examined the relationship between the mRNA expression level and H3K27m3 density at their gene promoters (Figure 2D). Interestingly, this analysis revealed that, at H3K27m3 density above a certain “threshold” level, no transcript could be detected. In contrast, below this threshold, transcript levels were inversely proportional to the H3K27m3 density.

Changes in Ezh2 protein

Ezh2, a component of PRC2 complex, is a histone methyltransferase (KMT6A) for H3K27 [18]. It has been reported that cellular levels of Ezh2 protein decrease during aging in vivo and senescence in vitro [17,19,20]. In contrast, we have previously shown that in old F344 rat kidneys, levels of Ezh2 mRNA and protein were higher than in young animals [7]. Western blot analysis confirmed these observations in F344 kidneys (Supplementary Figure S1). In longer-lived FBN-F1 animals, there was no difference in the cellular levels of Ezh2 between old and young kidneys (Supplementary Figure S1). Other factors are likely involved in the loss of H3K27m3 at selected genes in aging kidneys.

We wondered if the loss of H3K27m3 at the Lamc2 gene during aging was caused by a decrease in Ezh2 protein recruitment to that locus (rather than a decrease in cellular levels). Results of ChIP analysis show that in old kidneys, levels of Ezh2 protein along the locus were actually higher than those of young kidneys (Supplementary Figure S2), a difference which is not statistically significant. At Cdkn2a gene, a site with previously-described high levels of Ezh2 protein, which serves as a positive control [19], aging was also associated with increased Ezh2 density (not shown). To estimate enzymatic efficiency of Ezh2 along the Lamc1-Lamc2 locus, we divided H3K27m3 density by Ezh2 level. The H3K27m3/Ezh2 ratio was lower in old animals along the length of the entire locus (Supplementary Figure S3).

5mC changes at Lamc2 in aging kidneys

5mC changes during aging were described in different organisms [21–23], and 5mC levels measured at several CpG loci predict human age with very high precision [24,25]. Therefore, along with histone modification changes in aging kidneys, we also estimated 5mC levels at Lamc2 and control genes in DNA purified from the same tissues by using MeDIP assay. This analysis revealed significant losses of 5mC at the 5’ end of Lamc2 gene but no changes at the 5’ end of Lamc1 gene, in old compared to young kidneys (Figure 4). Next, we examined the functional significance of H3K27m3 and 5mC losses for Lamc2 expression.

The effect of Ezh2 and Dnmt inhibitors on Lamc2 expression in kidneys cells in vitro

The inverse correlation between Lamc2 mRNA levels and H3K27m3/5mC density at the gene (Figure 2D and Figure 3) suggested that these modifications mediate the silenced state of this gene, and that the increase in Lamc2 expression during aging is caused, at least in part, by the loss of H3K27 methylation and/or 5mC density at the gene. To test this suggestion, we first used an unspecific Ezh2 inhibitor, 3-deazaneplanocin (DZNep) [26]. This compound is a potent inhibitor of S-adenosylhomocysteine (AdoHcy) hydrolase [27,28]. Accumulated AdoHcy, in turn, is a competitive inhibitor of methyl-transferases, including those specific to histones and most likely DNA. It has been shown that DZNep in micromolar concentrations specifically inhibits methylation of H3K27 and activates PRC2-silenced genes [26,29], whereas its effect on DNA methylation has not been reported. Therefore, we used this inhibitor to examine its effect on H3K27m3 levels and Lamc2 expression in vitro.

In DZNep experiments we used human kidney cells (HEK293) where Lamc2 gene is not expressed. Treatment with 5µM DZNep for 48 hours slowed down the cell division rate but had little effect on cell survival (not shown). Western blot analysis revealed a decrease in global H3K27m3 levels in cells treated with DZNep compared to the controls (Figure 5). In agreement with previous observations [26,29], little or no changes were found in H3K9m3 levels at Lamc2 gene in DZNep-treated cells, indicating a selective effect on H3K27m3 histone modification (Supplementary Figure S4). RT qPCR analysis revealed substantial increases in Lamc2 transcript level after DZNep treatment, whereas no changes were observed in control Lamc1 and Gapdh transcript levels (Figure 5C).

Next, we used a more specific inhibitor of Ezh2, GSK126 [30]. This compound substantially decreases cellular levels of H3K27m3 in cells (Figure 5B). However, unlike DZNep, GSK126 treatment has no effect on Lamc2 expression (Figure 5C). These results indicate that H3K27m3 plays little or no role in the maintenance of Lamc2 silencing; therefore, loss of this modification does not result in gene upregulation. We conclude that DZNep treatment increased Lamc2 expression in vitro through inhibition of other than Ezh2 methyl-transferase(s). DNA methylation, a likely candidate, was tested next.

5-aza-2′-deoxycytidine, DAC, is a selective inhibitor of DNA methyl transferases [31]. DAC exposure has also been shown to substantially shorten the lifespan of cells [32]. HEK293 treatment with DAC for 48 hrs decreased DNA methylation levels at Lamc2 gene (Figure 5A) and activated its transcription to the same extent as aging in vivo (compare Figure 5C). Consistent with our suggestion of the role of 5mC loss in the upregulation of Lamc2, DZNep treatment significantly decreased 5mC levels at Lamc2 gene, whereas GSK126 had no effect (Figure 5A).

Animal tissues

Kidneys from ad lib fed F344 (4, 24, and 28 months old, mo) and FBN-F1 (4 and 32 mo) males were used in these studies. Frozen tissues were received from the National Institute of Aging rodent tissue collection and stored at -80°C. Six animals in each group were examined. For further analysis, frozen kidneys were cut into radial segments (~1/8 part of the entire organ) while kept on liquid nitrogen. From each kidney, three fragments similarly representing major kidney compartments, cortex and medulla, were used in protein, RNA and ChIP analyses, respectively.

Cell culture experiments

Human kidney cell line HEK293 was used in these studies. Cells were grown in DMEM supplemented with glutamine, penicillin, streptomycin, and 10% fetal bovine serum. Cells were treated with inhibitors as specified in figure legends. RNA purification, reverse transcription, and MeDIP assays were done as described below for kidney tissue.

Western blot analysis

Cytoplasmic, nuclear and histone-enriched protein fractions were purified from kidney tissue fragments as described in [45]. Protein concentration was measured using the Bradford method (Pierce). Equal amounts of protein were loaded on 10% SDS gels for detection of Ezh2 protein, or 13% gels for histone modification analysis. Proteins were transferred on PVDF membrane. After blocking with TBST buffer (Tris-HCl, pH 7.5, 10mM; NaCl, 150 mM; Tween 20, 0.05%) containing 5% BSA, membranes were incubated with indicated primary antibodies in TBST/1% BSA buffer for one hour at room temperature, or overnight at 4°C. After washes, secondary antibodies conjugated to alkaline phosphatase (Bio Rad) were added for 1 hour at room temperature. Membranes were washed with TBST and developed with BCIP/NBT substrate. Antibodies to H3K27m3 were from Millipore; others were as listed below.

Reverse transcription

For transcript level analysis, RNA purified from rat kidneys was treated with RNase-free DNase I, 1 U per 10 µg of RNA in 20 µl reactions, (Epicentre Technologies, Madison, WI) for 15 min at 37°C, deproteinized with phenol/chloroform mixture and precipitated with ethanol. One microgram of DNA-free RNA was reverse transcribed by Superscript II (100 U, Invitrogen, Gaithersburg, MD.) with random hexanucleotide primer mixture (1µM) in 10-µl final volume for 1 hr at 42°C. Reaction was stopped by mixing with 90µl TE (Tris-HCl 10mM, pH8.0, EDTA 1mM) and incubation at 95ºC for 5 min. RT mixtures were further diluted ten times with TE and analyzed by real time PCR with gene-specific sets of primers. Transcript levels were normalized to Gapdh mRNA levels.

Chromatin immunoprecipitation, MeDIP, and real-time PCR analysis

ChIP was done in 96-well plates as previously described [15]. The following antibodies were used, to RNA-polymerase II (anti-CTD, clone 4H8, or anti-Pol II N terminus, both Santa Cruz), histone H3 (ab1791, Abcam), H3K9m3 (ab8898, Abcam), H3K27m3 (ab6002, Abcam; 07-449, Millipore), H3K9/14Ac (Diagenode), Ezh2 (39933, Active Motif), Kdm6a (ab36938, Abcam). Mock IP was done without added antibodies. Precipitated DNA was purified in 100μl final volume. Input DNA was purified from 10% of the amount of tissue extract used in IPs. ChIP assays were repeated at least three times. Chromatin analysis with Ezh2 antibodies was done using Fast ChIP protocol [46]. Histone modification and Pol II levels at DNA sites of interest were calculated as detailed in [15]. ChIP data were expressed as a percent of input DNA. Results of ChIP and gene expression analyses were evaluated statistically by using analysis of variance with age (young versus old) as experimental factor.

5mC antibodies used in MeDIP are specific to 5-methylated cytosines in a single stranded DNA; thus, before immuno-precipitation, DNA from animal tissues was melted by boiling. Specifically, DNA purified from cells/tissues was diluted to 0.5 ml with TE buffer, treated with ultrasound for 30 sec (Bronson sonifier, equipped with a microtip), precipitated with ethanol, washed once with 70% ethanol, dried, and dissolved in 10 μl of TE buffer. Before immunoprecipitation (IP), DNA samples were boiled for 5 min, and chilled on ice. 0.5 μg of DNA was used in one IP reaction. MeDIP was done in 96-well plates as described [47,48]. Monoclonal antibodies to 5mC (Mouse clone 33D3, Aviva, San Diego, CA), 0.3 μg per IP reaction, were used. Mock IP was done without added antibodies. qPCR analysis of precipitated DNA was done with gene-specific primers. 10% of the amount of input DNA used in IPs was analyzed in parallel by qPCR to estimate efficiency of IP.

PCR reaction mixture contained 2.5μl 2X SYBR Green PCR master mix (SensiMix, Quantace), 2μl DNA template and 0.2μl primers (10 μM each) in 5μl final volume in 384-Well Optical Reaction Plate (Applied Biosystems). Amplification (three steps, 40 cycles), data acquisition and analysis were carried out using the 7900HT Real Time PCR system and SDS Enterprise Database software (Applied Biosystems). PCR reactions were run in triplicate. Standard dilutions of genomic DNA (for genomic targets) or dilutions of pooled RT reactions (for cDNA targets) were included in each PCR run. Sequences of primers used in these studies are available upon request.