Midkine, a heparin-binding protein, is increased in the diabetic mouse kidney postmenopause

Abstract

Estrogen is thought to protect against the development of chronic kidney disease, and menopause increases the development and severity of diabetic kidney disease. In this study, we used streptozotocin (STZ) to induce diabetes in the 4-vinylcyclohexene diepoxide (VCD)-treated mouse model of menopause. DNA microarrays were used to identify gene expression changes in the diabetic kidney postmenopause. An ANOVA model, CARMA, was used to isolate the menopause effect between two groups of diabetic mice, diabetic menopausal (STZ/VCD) and diabetic cycling (STZ). In this diabetic study, 8,864 genes of the possible 15,600 genes on the array were included in the ANOVA; 99 genes were identified as demonstrating a >1.5-fold up- or downregulation between the STZ/VCD and STZ groups. We randomly selected genes for confirmation by real-time PCR; midkine (Mdk), immediate early response gene 3 (IEX-1), mitogen-inducible gene 6 (Mig6), and ubiquitin-specific protease 2 (USP2) were significantly increased in the kidneys of STZ/VCD compared with STZ mice. Western blot analysis confirmed that Mdk and IEX-1 protein abundance was significantly increased in the kidney cortex of STZ/VCD compared with STZ mice. In a separate study, DNA microarrays and CARMA analysis were used to identify the effect of menopause on the nondiabetic kidney; VCD-treated mice were compared with cycling mice. Of the possible 15,600 genes on the array, 9,142 genes were included in the ANOVA; 20 genes were identified as demonstrating a >1.5-fold up- or downregulation; histidine decarboxylase and vanin 1 were among the genes identified as differentially expressed in the postmenopausal nondiabetic kidney. These data expand our understanding of how hormone status correlates with the development of diabetic kidney disease and identify several target genes for further studies.

diabetes is one of the most costly diseases afflicting developed countries (1). Approximately one-third of diabetic patients die of end-stage renal disease (37) as a result of progressive renal damage and hypertension. Nondiabetic renal disease progresses more slowly in premenopausal women than age-matched men (30); however, this difference disappears after menopause (5). Many studies, in humans and animals, suggest that estrogen is also protective against the development of diabetic kidney disease (1, 4, 6, 14, 23, 31, 33, 38). For example, when diabetic female rats were ovariectomized (OVX), several markers of diabetic kidney disease, such as glomerulosclerosis, tubular interstitial fibrosis, and urine albumin excretion, were significantly increased compared with cycling intact diabetic females. Estrogen replacement in the OVX rats improved these parameters (23).

OVX is the most commonly used animal model of menopause. However, we recently reported that the 4-vinylcyclohexene diepoxide (VCD)-treated mouse model of menopause can be used to study how the loss of estrogen influences the development of type 1 and type 2 diabetes. The VCD model of menopause involves repeated daily injections of mice with VCD to induce gradual ovarian failure (21). Before menopause, mice injected with VCD enter a period analogous to human perimenopause, in which cycles gradually become longer and more irregular and estrogen levels fall. After ovarian failure, the residual ovarian tissue continues to secrete androgens (24), similar to postmenopausal human ovaries (13, 19, 41).

Combining the streptozotocin (STZ) model of type 1 diabetes with the chemically induced mouse model of menopause (VCD), we previously demonstrated that the development of diabetic kidney damage was exacerbated in the menopausal compared with the cycling female (15). The aim of the present study was to identify changes in gene expression that may contribute to this increased progression of diabetic kidney damage in the menopausal female. We hypothesized that genes that exhibit differential expression in the menopausal diabetic kidney, compared with the diabetic kidney, may be potential markers of early kidney damage that may not have been previously associated with diabetic kidney damage. To test this hypothesis, mRNA was isolated from the kidney cortices of menopausal diabetic and cycling diabetic mice after 6 wk of hyperglycemia, and gene expression was analyzed using DNA microarrays. Targets for further study were confirmed using real-time PCR and Western blot analysis. In conjunction with the microarray study to determine changes in gene expression in diabetic kidneys of menopausal vs. cycling mice, we also used kidneys from nondiabetic cycling and menopausal mice. These data identify changes in gene expression due solely to estrogen loss, and not as a result of diabetes. The data presented here expand on our understanding of how changes in reproductive hormone status correlate with the development of diabetic kidney disease, and we identify several candidate genes for cell signaling studies.

METHODS

Animals.

Female B₆C₃F₁ mice were housed in standard cages in the animal facility of the Arizona Health Sciences Center under National Institutes of Health guidelines; the animals had ad libitum access to regular food and water. All protocols were approved by the Institutional Animal Care and Use Committee at the University of Arizona.

Induction of ovarian failure and diabetes.

Ovarian failure and diabetes were induced as previously described (15). Briefly, VCD (catalog no. V3630, Sigma) was administered via intraperitoneal injection at a dose of 160 mg/kg body wt using a dosing standard of 2.5 ml/kg body wt for 15 consecutive days to induce gradual ovarian failure (21, 25). Sesame oil was used as vehicle control. Progression into ovarian failure was monitored by daily vaginal cytology. Diestrus for 10 consecutive days was considered indicative of ovarian failure (21). Diabetes was induced by intraperitoneal injection of STZ (catalog no. S0130, Sigma) at a dose of 75 mg/kg body wt using a dosing standard of 0.2 ml/22 g body wt to 4-h-fasted mice for 3 consecutive days. STZ was administered during periovarian failure (immediately following VCD dosing and analogous to human perimenopause) or 2 wk after ovarian failure for the menopausal groups. Mice were killed 6, 8, or 12 wk after STZ injection. Mice were fasted for 4 h prior to death, at which point blood was collected, and blood glucose was measured using the CardioCheck PA blood testing device (catalog no. 2568, HealthCheck Systems) with CardioCheck glucose test strips (catalog no. l2556, HealthCheck Systems).

RNA isolation.

Kidneys were separated into cortex and medulla, and RNA was isolated from renal cortex using the RNeasy Mini Kit according to the manufacturer's protocol (Qiagen) for isolation from tissue. DNA contamination was eliminated during the isolation procedure with a 15-min DNase incubation. RNA was quantified using a spectrophotometer (model ND1000, Nanodrop, Wilmington, DE).

Microarray protocol.

Full methodology for RNA amplification, cDNA production, and microarray slide preparation, hybridization, and analysis has been previously published (26). Total amplified RNA was used for microarray analysis. RNA was amplified using the MessageAmp kit according to the manufacturer's protocol (Ambion). As a template for each amplification reaction, 2.5 μg of total RNA from each individual mouse were used. Amplified RNA was then reverse-transcribed to cDNA with amino allyl incorporation using the EndoFree RT kit according to the manufacturer's protocol (Ambion). Amino allyl-modified cDNA was purified using PCR purification columns (catalog no. 28104, Qiagen) according to the manufacturer's protocol. The modified cDNA was labeled with Alexa dyes via the free amine modification.

Microarray slides for this study were prepared in the Genomics Research Laboratory at the University of Arizona using the National Institute on Aging mouse 15K clone set (http://lgsun.grc.nia.nih.gov/cDNA/15k/html). Full methodology for microarray production and hybridization protocols has been previously published (26). Labeled modified cDNA in hybridization buffer was loaded onto a slide and set to hybridize at 42°C for ≥16 h. Then, after a short wash in the hybridization solution, the slide was removed and dipped in 0.1× saline-sodium citrate to remove any residual nonhybridized cDNA. The slide was dried and analyzed using the arrayWORx CCD-based microarray scanner (Applied Precision, Issaquah, WA), which is capable of multichannel fluorescence scanning.

Microarray analysis.

Microarray data were reduced and analyzed by CARMA, a custom-designed software package for analysis of microarray (10). The hybridization scheme was based on the interwoven loop design, as described previously (16, 26). Each RNA sample was hybridized to an individual array four times, and each gene clone was spotted twice per array; a total of eight independent measurements per gene per sample were obtained. An ANOVA model was used to isolate the effect of menopause while taking into account experimental effects (array, dye, spot). The two channels for each array were normalized using intensity- and location-dependent Lowess regression. Data were first transformed using a Loglin function, which performs a logarithmic transformation for higher intensities and a linear transformation for lower intensities (10). An ANOVA was performed on all genes that were measured on a minimum of five of the eight opportunities for at least one sample. Genes were considered to be differentially expressed between groups (i.e., STZ vs. VCD/STZ and control vs. VCD) if P < 0.05 (by ANOVA). Genes exhibiting <1.5-fold up- or downregulation between the experimental groups were excluded from future analysis. The results were submitted to the Gene Expression Omnibus.

Promoter analysis.

The 500 base pairs immediately upstream of the transcription start site were obtained from Ensembl (www.ensembl.org) for genes identified by the STZ vs. VCD/STZ microarray study as significantly differentially expressed. Using the Alibaba2 online transcription factor binding site prediction software, we identified promoter response elements for estrogen, progesterone, and androgens according to the TRANSFAC classification of transcription factor binding sites.

Real-time PCR.

Real-time PCR was performed as previously described (26). Briefly, 2.5 μg of RNA were reverse-transcribed with the murine leukemia virus reverse transcriptase enzyme, and the resulting cDNA was diluted 1:25 to an approximate final concentration of 8 ng/μl. Each real-time PCR contained 5 μl of SYBR Green Master Mix, 1 μl of water, 2 μl of diluted cDNA, and 5 pmol each of forward and reverse primer in a total volume of 10 μl. Each reaction was performed in triplicate at 95°C for 5 min, then 95°C for 15 s and 60°C for 30 s for 40 cycles. The RotorGene RG3000 (Corbett Research, San Francisco, CA) sequence detection system was used. Primers were designed to the 3′ end of genes of interest using Primer3 software (34) and are listed in Table 1. Cycle threshold (C_t) values were used to calculate the expression levels of genes of interest relative to the expression of β-actin mRNA measured in parallel samples. Analysis was performed as described previously (20), and results are presented as mean fold change on a base 2 logarithmic scale.

Table 1.

Real-time PCR primers for genes identified by microarray

			Primer
Gene Name	Abbreviation	GenBank Reference No.	Forward	Reverse
Carnitine O-octanoyltransferase	Crot	NM_023733	GCATTTCCATTTGAACCTAATTG	ATAGCACCCTTGAAGCGAAC
CDC28 protein kinase regulatory subunit 2	Cks2	NM_025415	AAGTGCAGCTGGGATCATCT	GCAGTTGCATTTGACTGAGC
Immediate early response 3	IEX-1	NM_133662	GGGTAACACTCCGTCTTCCA	TACTAGGCGACCCCAGACAG
Midkine	Mdk	NM_010784	GTCAATCACGCCTGTCCTCT	CAAGTATCAGGGTGGGGAGA
ERBB receptor feedback inhibitor 1/mitogen inducible gene 6	Mig6	NM_133753	TACACCCATTAAAAGCTGCCC	TGGTGGTTAGAAGCTCTACCTC
Ubiquitin-specific protease 2	Usp2	NM_198092	GAACCAGCAAGCTCACAACA	GTTCCGGAGTGATTGGACAC

Protein isolation.

The protocol for protein isolation from kidneys has been previously published (2). Kidneys were separated into cortex and medulla, and the cortex was homogenized in 3 ml of ice-cold isolation solution (250 mM sucrose and 10 mM triethanolamine, pH 7.6, containing 1 μg/ml leupeptin and 0.1 mg/ml phenylmethylsulfonyl fluoride) using a tissue homogenizer (Omni 1000 fitted with a micro-sawtooth generator) at three-quarters maximum speed for three 20-s intervals. Total protein concentrations were measured using the Pierce bicinchoninic acid kit (catalog no. PI23227, Fisher), and samples were solubilized in Laemmli sample buffer at 60°C for 12 min.

Western blotting.

Semiquantitative immunoblotting was carried out as described previously (3). To confirm that protein loading of the gels was equal, preliminary 12% polyacrylamide gels were stained with Coomassie blue. Images of gels were taken using the Bio-Rad Universal Hood II, and densitometry was performed on representative bands to ensure equal loading (<10% variation relative to the mean) using Quantity One software. Proteins were separated on 12% polyacrylamide gels by SDS-PAGE and transferred to polyvinylidene difluoride (2-μm pores) membrane electropheretically using the Mini Trans-Blot Cell (Bio-Rad, Hercules, CA). Membranes were blocked with Odyssey blocking buffer for 1 h at room temperature and then probed overnight at 4°C with the respective primary antibodies diluted in Odyssey blocking buffer with 0.2% Tween 20. Membranes were washed and exposed to secondary antibodies for 1 h at room temperature. Bands were visualized via the fluorescent tag on the secondary antibodies using a LI-COR Odyssey Imager and Odyssey version 2.1 software. The fluorescence intensities of individual bands were determined using Odyssey version 2.1 software. To facilitate comparisons, intensity values were normalized, such that the mean value of the diabetic group is defined as 100%.

Antibodies.

Immediate early response gene 3 (IEX-1) primary antibody (catalog no. HM1182) was purchased from Hypromatrix, Mdk primary antibody (catalog no. AF-258-PB) from R & D Systems; 800-nm fluorescently tagged anti-mouse secondary antibody (catalog no. 927-32212) from LI-COR, and 680-nm fluorescently tagged anti-goat secondary antibody (catalog no. A-21109) from Invitrogen.

Reagents.

Odyssey blocking buffer (catalog no. 927-40000) was purchased from LI-COR; Coomassie blue (catalog no. LC6025) and murine leukemia virus reverse transcriptase enzyme (catalog no. 28025013) from Invitrogen; SYBR Green Master Mix (catalog no. 600581) from Stratagene; Alexa dyes for microarray studies (catalog nos. A-20002-546 and A-20006-647) from Molecular Probes; saline-sodium citrate solution (catalog no. 9763), MessageAmp kit (catalog no. 1750), and EndoFree kit (catalog no. 1740) from Ambion; and RNeasy Mini Kit (catalog no. 74104), DNase (catalog no. 79254), and PCR purification columns (catalog no. 28104) from Qiagen.

Statistics.

Real-time PCR and Western blot data were analyzed using Student's t-test. Results are presented as means ± SE. Microarray data were analyzed by ANOVA to determine statistical significance via CARMA, a custom-designed software package for analysis of microarray (10). In all tests, P < 0.05 was considered significant.

RESULTS

Effect of menopause on blood glucose levels.

Diabetes was induced by STZ injection 2 wk after a mouse entered ovarian failure (analogous to human menopause). At 6 wk after STZ injection, blood glucose levels were higher in menopausal diabetic than cycling diabetic mice (15). There was no difference in blood glucose levels between nondiabetic control and menopausal mice (15). Here we extended this model to 8 and 12 wk after STZ injection. There was no significant difference in blood glucose levels between cycling diabetic and menopausal diabetic mice after 8 wk (333 ± 52 and 416 ± 69 mg/dl for STZ and STZ/VCD, respectively, P > 0.05) or 12 wk (311 ± 21 and 298 ± 10 mg/dl for STZ and STZ/VCD, respectively, P > 0.05) of diabetes.

Microarray analysis.

Total RNA was isolated from kidney cortex of cycling diabetic and menopausal diabetic mice 6 wk after STZ injection. Microarray analysis was used to identify genes that may be regulated by menopausal status in the diabetic kidney. After hybridization, 8,864 of the possible 15,600 genes were included in the microarray analysis. Genes were considered to be differentially expressed between cycling diabetic and menopausal diabetic mice if P < 0.05 (by ANOVA) with a ≥1.5-fold change. A total of 99 genetic sequences representing 66 known genes and 22 unknown sequences were identified as differentially expressed between kidney cortex of cycling diabetic and kidney cortex of menopausal diabetic mice. [Details of individual measurements, gene name (if known), and links to the National Institute on Aging and GenBank databases are available in Supplemental Data A in Supplemental Material for this article, available online at the Journal website.]

In a second microarray experiment, samples from control and nondiabetic menopausal mice were compared to identify genes whose expression is altered by changes in hormone status absent the context of diabetes. In this microarray, 9,142 genes were included in the analysis, and 20 genetic sequences representing 14 known genes and 5 unknown sequences were identified as differentially expressed between control and nondiabetic menopausal samples. Genes that were identified as changing in both sets of arrays were excluded from further analysis based on the rationale that changes in their expression were due solely to changes in hormone status regardless of diabetic status. Two genes, aldehyde dehydrogenase 1a1 and T-cell immunoglobulin and mucin domain containing 2, met this criterion. [Details of individual measurements, gene name (if known), and links to the National Institute on Aging and GenBank databases are available in Supplemental Data B.]

Real-time PCR confirmation of selected genes from microarray.

To confirm the microarray results, we used the same amplified RNA samples (6-wk time point) that were used for the microarray analysis to perform real-time PCR for a randomly selected subset of the genes. We previously demonstrated that amplified RNA is a good measure of the relative gene expression in total RNA samples (26). We used actin as an internal standard for the real-time PCR analysis, as it was unchanged in the microarray data. Microarray data identified CDC28 protein kinase regulatory subunit 2a (Cks2) and carnitine O-octanoyltransferase (Crot) as significantly decreased in abundance in the kidneys of menopausal diabetic compared with cycling diabetic mice; real-time PCR confirmed that both genes were significantly decreased (1.6-fold decrease for Cks2 and 1.7-fold decrease for Crot). In contrast, ubiquitin-specific protease 2 (USP2), IEX-1, Mdk, and ERBB receptor feedback inhibitor 1/mitogen-inducible gene 6 (Mig6) were identified as significantly increased in abundance in the kidneys of menopausal diabetic compared with cycling diabetic mice, and these increases were confirmed by real-time PCR (1.7-fold for USP2, 1.9-fold for IEX-1, 1.7-fold for Mdk, and 2.4-fold for Mig6). Figure 1 summarizes the microarray and real-time PCR data for these randomly selected genes.

Fig. 1.

SYBR Green real-time PCR confirmation of selected genes from diabetic vs. menopausal diabetic microarray. Values are expressed as mean relative fold change in renal cortex of menopausal diabetic mice normalized to cycling diabetic mice. Assays for actin were run in parallel on each sample for normalization of data. Significance was determined by Student's t-test: *P < 0.05 vs. cycling diabetic [STZ (streptozotocin)]. VCD, 4-vinylcyclohexene diepoxide; Cks2, CDC28 protein kinase regulatory subunit 2; Crot, carnitine O-octanoyltransferase; USP2, ubiquitin-specific protease 2; IEX-1, immediate early response 3; Mdk, midkine; Mig6, ERBB receptor feedback inhibitor 1/mitogen-inducible gene 6.

Microarray data from kidneys of control and menopausal mice identified histidine decarboxylase (Hdc) as significantly decreased in abundance in the kidneys of menopausal compared with control (cycling) mice; this result was confirmed by real-time PCR (1.64-fold decrease for Hdc). In contrast, vanin 1 (Vnn1), peroxisomal Δ³,Δ²-enoyl-coenzyme A isomerase (Peci), and 3-hydroxy-3-methylglutaryl-coenzyme A synthase 2 (Hmgcs2) were identified as significantly increased in abundance in the kidneys of menopausal compared with control (cycling) mice, and these data were confirmed by real-time PCR (1.7-fold for Vnn1, 1.8-fold for Peci, and 9-fold for Hmgcs2). Figure 2 summarizes the microarray and real-time PCR data for these randomly selected genes.

Fig. 2.

SYBR Green real-time PCR confirmation of selected genes from control vs. menopausal microarray. Values are expressed as mean relative fold change in renal cortex of menopausal mice normalized to cycling nondiabetic mice. Assays for actin were run in parallel on each sample for normalization of data. Significance was determined by Student's t-test: *P < 0.05 vs. cycling control (Control). Hdc, histidine decarboxylase; Vnn2, vanin 1; Peci, peroxisomal Δ³,Δ²-enoyl-coenzyme A isomerase; Hmgcs2, 3-hydroxy-3-methylglutaryl-coenzyme A synthase 2.

Western blot analysis for selected genes.

To confirm that the increases in mRNA abundance translated to increases in protein expression, Western blot analyses were performed for Mdk and IEX-1 at 6, 8, and 12 wk following STZ injection.

Mdk is a heparin-binding growth factor that has been suggested to play a role in the development of diabetic kidney damage in mice (17, 18). Mdk is an ∼13- to 14-kDa protein (40) that migrates as a 17- to 18-kDa protein because of its highly basic nature (29). It is also capable of forming a dimer, which migrates at ∼29–31 kDa (12, 28). As shown in Fig. 3, Mdk protein expression was significantly increased in the kidney cortex of menopausal diabetic compared with cycling diabetic mice after 8 wk of diabetes: 100 ± 6 and 142 ± 11% for Mdk monomer for STZ and STZ/VCD, respectively (P < 0.05) and 100 ± 9 and 176 ± 13% for Mdk dimer for STZ and STZ/VCD, respectively (P < 0.05).

Fig. 3.

Mdk protein abundance in kidney cortex. A: immunoblot of Mdk protein examining abundance in kidney cortex homogenates of cycling diabetic and menopausal diabetic mice. B: densitometry of immunoblot. Values are normalized, and values from cycling diabetic mice are set at 100 to facilitate comparison. *P < 0.05 vs. cycling diabetic (STZ).

IEX-1 is a regulator of proliferation and apoptosis (43). As shown in Fig. 4, IEX-1 protein expression was significantly increased in the kidney cortex of menopausal diabetic mice compared with cycling diabetic mice following 12 wk of diabetes: 100 ± 11 and 146 ± 11% for STZ and STZ/VCD, respectively (P < 0.05).

Fig. 4.

IEX-1 protein abundance in kidney cortex. A: immunoblot of IEX-1 protein examining abundance in kidney cortex homogenates of cycling diabetic and menopausal diabetic mice. B: densitometry of immunoblot. Values are normalized, and values from cycling diabetic mice are set at 100 to facilitate comparison. *P < 0.05.

Real-time PCR in perimenopausal/diabetic mice.

Microarray data identified gene candidates as differentially expressed between the renal cortex of menopausal diabetic and cycling diabetic mice. We next studied expression of these same genes in the renal cortex of perimenopausal diabetic and cycling diabetic mice.

As shown in Fig. 5, mRNA expression of Crot was decreased 1.25-fold (P < 0.05) in perimenopausal diabetic compared with cycling diabetic mice. mRNA expression of USP2 was increased 1.59-fold (P < 0.05) and IEX-1 was increased 1.96-fold (P < 0.05) in perimenopausal diabetic compared with cycling diabetic mice; these increases were similar to those identified in the menopausal diabetic mice. There was no significant difference in the mRNA expression levels of Cks2 or Mdk in perimenopausal diabetic compared with cycling diabetic mice.

Fig. 5.

SYBR Green real-time PCR of selected genes in diabetic vs. perimenopausal diabetic samples. Values are expressed as mean relative fold change in renal cortex of perimenopausal diabetic mice normalized to cycling diabetic mice. Assays for actin were run in parallel on each sample for normalization of data. Significance was determined by Student's t-test: *P < 0.05 vs. cycling diabetic (STZ).

Promoter analysis of genes identified as differentially expressed in menopausal diabetic kidneys.

To assess putative binding sites for transcription factors in the 5′-flanking region of genes identified as differentially expressed between diabetic and menopausal diabetic kidneys, we carried out promoter analyses for estrogen (ERE), progesterone (PRE), and androgen response elements (ARE) using Alibaba2, a transcription factor binding site prediction software. Table 2 reports the putative hormone response elements identified in the 5′-flanking regions. No AREs were identified in any of the genes examined. In contrast, we identified EREs or PREs in 31 of the 99 genes.

Table 2.

Identification of EREs and PREs

Gene Name	Abbreviation	GenBank Reference No.	Ratio	ERE	PRE
Phosphoribosyl pyrophosphate amidotransferase	PPAT	BC023841	2.10	2
Immediate early response 3	IEX-1	NM_133662	1.94		1
Arginyl aminopeptidase (aminopeptidase B)	Rnpep	NM_145417	1.69	1	1
Midkine	Mdk	NM_010784	1.68	2
CDC42 small effector 2	Cdc42 se2	NM_178626	1.67	1
Nucleolin	Ncl	NM_010880	1.65	3
Josephin domain containing 3	Josd3	NM_029248	1.63	1	1
BAI1-associated protein 2-like 1	Baiap2l1	NM_025833	1.60	2
Heat shock protein 8	Hspa8	NM_031165	1.55		1
Transducer of ERBB2, 2	Tob2	NM_020507	1.54	2
Mus musculus RIKEN cDNA 4933427D14 gene		NM_028963	1.54	1
Ellis van Creveld syndrome 2 homolog (human)	Evc2	NM_145920	1.53	1
Dystonin	Dst	NM_134448	1.52	1
LPS-responsive beige-like anchor	Lrba	NM_030695	1.52	1
Disabled homolog 2 (Drosophila)	Dab2	NM_023118	1.51	1
M-phase phosphoprotein 9, transcript variant 5	Mphos9	XM_001002130	1.51	2	1
G elongation factor, mitochondrial 1	Gfm	NM_138591	1.51	3
BCL2/adenovirus E1B interacting protein 1, NIP1	Bnip1	NM_172149	−1.52	2
Acyl-CoA dehydrogenase, medium chain	Acadm	NM_007382	−1.53	1
ATP-binding cassette, subfamily C (CFTR/MRP), member 3	Abcc3	NM_029600	−1.54	1	1
Mitochondrial ribosomal protein L9	Mrpl9	NM_030116	−1.55	1
tRNA splicing endonuclease 34 homolog (SEN34, Saccharomyces cerevisiae)	Tsen34	NM_024168	−1.56	2
Eukaryotic translation initiation factor 3, subunit D	Eif3c	NM_018749	−1.56		2
Mitochondrial ribosomal protein L36	Mrpl36	NM_053163	−1.61	1
Neuroguidin, EIF4E binding protein	Ngdn	NM_026890	−1.70	2
Vacuolar protein sorting 45 (yeast)	Vsp45	NM_013841	−1.75	1
Chloride channel, nucleotide-sensitive, 1A	Clns1a	NM_023671	−1.77	1
Solute carrier family 16 (monocarboxylic acid transporters), member 1	Scl16a1	NM_009196	−1.84	1
COMM domain containing 5	Comm5	NM_025536	−1.89		1
Carnitine O-octanoyltransferase	Crot	NM_023733	−1.92		1

Conserved binding motifs for estrogen, progesterone, and androgen response elements (ERE, PRE, and ARE) were identified in the 5′-flanking regions of genes identified as differentially expressed by microarray between kidneys of diabetic and menopausal diabetic mice. Table lists gene name and abbreviations, GenBank reference numbers, ratio change in expression identified by microarray, and number of predicted EREs or PREs in the first 500 bp upstream of transcription start site. No AREs were detected.

DISCUSSION

In a previous study, we combined the VCD mouse model of menopause with STZ-induced type 1 diabetes and reported that renal damage develops more rapidly and/or severely in post-ovarian failure (menopausal) diabetic than cycling diabetic mice.

In this study, microarray analysis was used to determine the gene expression changes that occur postmenopause in the diabetic kidney. The aim of this study was to identify gene candidates involved in the rapid progression of diabetic kidney damage following changes in estrogen-to-androgen ratios. We identified 99 genes that were differentially expressed between kidneys from cycling diabetic and menopausal (VCD-treated) diabetic mice. EREs and/or PREs were identified in 31 of the 99 genes.

In a second study, we used microarray analysis to determine the gene expression changes that occur postmenopause in the nondiabetic kidney. 20 genes were identified as differentially expressed between kidneys from control (cycling) and menopausal mice. We used real-time PCR to confirm the changes in mRNA expression for several genes from both microarray studies. Furthermore, we used Western blot analysis to confirm that changes in gene expression translated to changes in protein expression.

Our previous study demonstrated that renal cell proliferation was significantly increased in menopausal diabetic compared with diabetic mice (15). Here, using microarray analysis, we identified the differential expression of several genes associated with cell proliferation and/or apoptosis. As such, these represent candidate genes that may contribute to renal cell proliferation in diabetes. For example, IEX-1 was significantly increased in the renal cortex of menopausal diabetic compared with cycling diabetic mice. IEX-1 regulates cell proliferation and apoptosis in response to stress in a variety of tissues (43), although no previous studies have identified IEX-1 regulation in diabetic kidneys. Similarly, Cks2 and Mig6 were significantly increased in the menopausal diabetic kidney. Mig6 is known to suppress proliferation in skin cells (7). Cks2 is involved in regulation of the cell cycle, although its exact role remains unclear (32).

Data presented here identified a significant increase in Mdk expression in the kidneys of menopausal diabetic compared with diabetic mice. Mdk, a heparin-binding protein associated with inflammation and tissue repair, has previously been identified as increased in expression in kidneys from diabetic mice; however, our study suggests that the expression of Mdk is enhanced when the onset of diabetes occurs during menopause. Mdk is thought to promote macrophage infiltration in the diabetic kidney (18); indeed, several studies have shown that knockdown of Mdk (mdk^−/− mice) prevents the development of kidney damage. For example, knockdown of Mdk prevented the onset of kidney disease in diabetes (18), ischemia-reperfusion (35, 36), and 5/6 nephrectomy-induced hypertension (11). Additional studies using mesangial cell cultures from mdk^−/− mice demonstrated that the expression of TGFβ was lower, suggesting that loss of Mdk may prevent kidney damage by decreasing expression of TGFβ (18). Consistent with our previous study that demonstrated exacerbated diabetic kidney damage in menopausal diabetic mice, Mdk expression was increased in menopausal diabetic compared with cycling diabetic mice.

Approximately one-third of the genes identified by microarray as differentially expressed between menopausal diabetic and cycling diabetic mice contained one or more predicted EREs or PREs in the 5′-flanking region of their respective promoters. On further analysis, several genes identified in this study have previously been demonstrated to be regulated by progesterone or estrogen signaling pathways. Mdk expression is known to be increased by estrogen in cultured murine endometrial cells (44), and two predicted EREs were identified in the 5′-flanking sequence of Mdk. Similarly, the promoter of Abcc3 (ATP-binding cassette, subfamily C, member 3) contains predicted EREs and PREs; Abcc3 expression in the mouse kidney was identified as decreased following OVX and increased by 17β-estradiol replacement (22). These examples demonstrate the validity of our study's aim, which was to identify novel gene candidates that may correlate with alterations in reproductive hormone levels and the development of diabetic kidney damage.

Interestingly, none of the genes analyzed contained any predicted AREs. However, several of the genes identified by our microarrays have previously been shown to be regulated by androgens in other cell types or tissues. For example, USP2, the expression of which is increased in human and rat proliferative nephropathies (42), has also been identified as regulated by androgens; USP2 expression is increased in prostate cancer cells by testosterone (9). USP2 was significantly increased in the menopausal diabetic kidneys. Similarly, Hdc is suppressed by testosterone in the kidneys of female mice (27). Our data demonstrated that Hdc was significantly decreased in menopausal diabetic kidneys.

The VCD-treated mouse model of menopause is a powerful model of rodent menopause, because mice retain ovarian tissue, which continues to secrete androgens as they transition from a perimenopausal to a postmenopausal state (25). Postmenopausal females have an increased risk of many diseases, and several risk factors have been identified as originating in the perimenopausal period. For example, longer cycle length, which occurs during perimenopause, is correlated with an increase in body mass index and levels of C-reactive protein (24). Perimenopause is also associated with an increase in serum triglycerides and an increase in diastolic blood pressure compared with premenopausal status (8). Adiponectin, a hormone that positively modulates glucose disposal and insulin sensitivity, is present in lower levels in perimenopausal than premenopausal women (39).

With this in mind, we examined the expression levels of several candidate genes in the kidneys of perimenopausal diabetic mice and compared them with diabetic age-matched mice. Genes were selected from the results of the microarray study in diabetic menopausal mice. Similar to the data in menopausal diabetic mice, Crot, USP2, and IEX-1 expression levels were significantly increased in kidneys of mice made diabetic during perimenopause. These data suggest that further analysis of perimenopause in the context of the development of diabetic kidney disease is important to further our understanding of how hormone status contributes to disease progression.

In summary, we have demonstrated that the gradual transition into menopause is associated with changes in gene expression in the diabetic kidney. We have identified several candidate genes for further study, including Mdk, a heparin-binding protein associated with inflammation, which was increased in expression in the menopausal diabetic kidney. Our data emphasize the utility of the VCD model of menopause to study disease processes, specifically the development of diabetic kidney disease across the menopausal transition.

GRANTS

This work was funded by National Institutes of Health Grants RO1 DK-073611 (H. L. Brooks), RO1 AG-021948 (P. B. Hoyer), and P30 ES-006694; a National Science Foundation Integrative Graduate Education and Research Traineeship Fellowship in Genomics (M. K. Diamond-Stanic); and the Achievement Rewards for College Scientists Foundation (M. K. Diamond-Stanic).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.