Dissection of a genetic locus influencing renal function in the rat and its concordance with kidney disease loci on human chromosome 1q21

Abstract

Previously, we conducted a genome scan on a population derived from the Dahl salt-sensitive hypertensive (S) and the spontaneously hypertensive rat (SHR) using urinary albumin excretion (UAE) as our primary measure of renal function. We identified 10 quantitative trait loci (QTL) linked to several renal and/or cardiovascular traits. In particular, linkage and subsequent congenic strain analysis demonstrated that the loci on chromosome 2 had a large and significant effect on UAE compared with the S rat. The present work sought to characterize the chromosome 2 congenic strain [S.SHR(2)] by conducting a time-course analysis (week 4–20), including evaluating additional renal parameters, histology, electron microscopy, and gene expression/ pathway analysis. Throughout the time course the congenic strain consistently maintained a threefold reduction in UAE compared with S rats and was supported by the histological findings of significantly reduced glomerular, tubular and interstitial changes. Gene expression/pathway analysis performed at week 4, 12, and 20 revealed that pathways involved in cellular assembly and organization, cellular movement, and immune response were controlled differently between the S and congenic. When all the data are considered, the chromosome 2 congenic appears to attenuate renal damage primarily through an altered fibrotic response. Recombinant progeny testing was employed to reduce the QTL to ~1.5 cM containing several interesting candidate genes. The concordance of this rat QTL with renal disease loci on human chromosome 1q21 demonstrate that elucidating the causative gene and mechanism of the rat QTL may be of particular importance for understanding kidney disease in humans.

CHRONIC KIDNEY DISEASE (CKD) is a worldwide healthcare problem with increasing incidence and prevalence (38). It is characterized by a gradual loss of kidney function, usually over years and can culminate in renal failure. Additionally, CKD is an important risk factor for the development of cardiovascular disease and overall mortality. Interestingly, most CKD cases are not associated with primary renal disease, but with systemic conditions like diabetes and hypertension (38, 55). Essential hypertension can cause renal injury, but renal susceptibility genes most likely determine occurrence and severity of renal damage (54).

It has been well established, through familial studies, that CKD has a strong genetic component (2, 18, 19, 49). Investigation of congenital and familial forms of kidney disease has led to the identification of genes required for proper functioning of the glomerular filtration barrier (7). These studies have provided insight into mechanisms of proteinuria and glomerulosclerosis. However, these mutations have not been linked to CKD in the general population, demonstrating that CKD is most certainly a complex trait and as such makes genetic studies in humans more difficult (7). On the other hand, the rat provides a particularly fertile model to study disease because it overcomes some of the limitations associated with using human subjects. There are many well-defined inbred rat strains currently being used to study the genetics of CKD (5, 35, 37, 45–47, 51). In particular, the Dahl salt-sensitive rat (S) was selectively bred as a model to study the genetics of salt-sensitive hypertension (11, 12). However, the S rat also develops progressive proteinuria, severe glomerular, vascular, and tubular lesions, besides the development of hypertension, and consequently can serve as a model of hypertension related renal disease (23, 52).

Recently, a genetic analysis of renal and cardiovascular traits was conducted using the S and the spontaneously hypertensive rat (SHR) by our group (20) and others (42, 48). Linkage analysis identified quantitative trait loci (QTL) on multiple chromosomes (1, 2, 6, 8–11, 13, and 19) for urinary protein excretion (UPE) and/or urinary albumin excretion (UAE) with variable time-course patterns (20). A second study found that several QTL (chromosome 2, 11, and 19) were influenced by salt-loading, presumably a result of changes in blood pressure in response to the high-salt diet (21). Congenic strain analysis established that the QTL on chromosome 2 had a major influence on renal function in the S rat (21).

The first aim of the present work was to characterize the chromosome 2 congenic strain by conducting a time-course analysis and establishing onset and progression of renal disease compared with both parental strains. We employed a comprehensive approach, including evaluation of additional renal parameters, histology, electron microscopy, gene expression analysis, and gene pathway analysis to characterize the strain. A second aim was to employ recombinant progeny testing (RPT) to reduce the QTL to a small genomic region to aid in gene identification.

MATERIALS AND METHODS

Animals

The Dahl salt-sensitive (SS/Jr or S), spontaneously hypertensive rat (SHR/NHsd or SHR), and S.SHR(2) congenic strain were maintained in our animal facility at the University of Toledo, Health Science Campus (UTHSC, formerly the Medical University of Ohio). All experiments had approval of our Institution Animal Care and Use Committee. At 4 wk of age, a group of age-matched males S (n = 50), S.SHR(2) (n = 50), and SHR (n = 50) were weaned to a low-salt diet (0.3% NaCl, TD7034; Harlan Teklad, Madison, WI) and studied for several renal and cardiovascular traits at 4, 5, 6, 8, 12, 16, and 20 wk of age.

At each time point a subset of animals (n = 6 to 12) was euthanized (overdose of pentobarbital sodium) for the collection tissue. Kidney samples were processed for histological, electron microscopy, and microarray analysis. Serum samples were obtained to measure blood parameters. Additionally body, heart, and kidney weights were also measured.

For RPT, a large F1[S × S.SHR(2)] × S backcross population (n = 971) was developed to establish recombinant animals within the introgressed region on chromosome 2. Male S.SHR(2) animals were crossed to S females to produce F1[S × S.SHR(2)] animals. The F1 rats were backcrossed to the S to produce the F1[S × S.SHR(2)] × S population. Additionally, urine parameters were collected on this population to reduce the confidence interval for the QTL region. The 971 rats were handled in six blocks of ~162 animals. Rats in each block were closely age matched, but between blocks the average ages differed by a few days. This allowed urine collections to proceed on all rats at approximately the same ages. Block effects, if any, were removed statistically before further analysis.

Phenotyping

Blood pressure

Blood pressure (BP) was also collected using a telemetry system (Data Sciences International, St. Paul, MN). At 61–63 days of age a transmitter was surgically implanted in S (n = 8) and S.SHR(2) congenic (n = 8) animals. Surgery was performed under 1.5% isoflurane anesthetic in 100% O₂. The probe body was placed subcutaneously in the flank, the probe catheter was inserted into the femoral artery, and the tip of the probe was advanced to the lower abdominal aorta as done previously (26). Data on systolic BP, diastolic BP, mean BP, pulse pressure, and heart rate were collected. Readings for each BP parameter was collected for a 24-h period (5-min time intervals for 10 s) at 8, 12, 16, and 20 wk of age.

Urine and blood parameters

To collect urine, animals were kept in metabolism cages (Lab Products, Seaford, DE) for 24 h with free access to water. Sodium azide was added to the collection vials for a final concentration of ~0.01% in the urine as a preservative. UPE was determined colorimetrically using pyrogallol red/molybdate complex (Quantimetrix, Redondo Beach, CA). UAE was determined by rat specific albumin EIA kit (SPI-bio). UPE and UAE are expressed as mg/24 h. Urine creatinine was determined by the Jaffe method (Cayman Chemical, Ann Arbor, MI). Following 24-h urine collection, blood was obtained by cardiac puncture. Blood parameters (Table 1) were determined by standard methods using an Alfa Wassermann ACE automated chemistry analyzer (BioReliance, Rockville, MD).

Table 1.

Comparison of blood and urine parameters for S, S.SHR(2), and SHR at week 20

	S	S.SHR(2)	SHR
Weight, g
Body weight	421±11.1	435±10.0	339±9.3*
Heart weight	1.3±0.02	1.3±0.02	1.3±0.01
Total kidney weight	2.8±0.11†	2.6±0.07	2.6±0.04
Blood
Total protein	6.7±0.13	6.4±0.17	6.4±0.12
Creatinine, mg/dl	0.54±0.01	0.53±0.02	0.4±0.02*
Blood urea nitrogen, mg/dl	33.4±1.50†	29.7±0.49	29.3±0.76
Total cholesterol, mg/dl	114.9±4.3†	97.3±1.7	47.3±7.6*
Triglycerides, mg/dl	106.1±8.5†	90.9±7.5	63.0±7.5*
Sodium, meq/l	149.6±2.36†	142.8±1.05	144.2±0.60
Potassium, meq/l	4.6±0.13†	5.4±0.22	5.3±0.21
Bicarbonate (CO2), meq/l	26.6±1.09†	23.1±0.47	23.5±0.16
Urine
Urine volume	14.2±1.86	22.3±5.1	14.2±3.4
Urine pH	6.6±0.07	6.8±0.07	7.2±0.10*
Creatinine clearance, ml·min−1·g TKW−1	0.75±0.018	0.78±0.025	0.80±0.027
Sodium excretion, meq/day	0.43±0.03	0.45±0.02	0.82±0.05*

Values are means ± SE; n = 6 per group. Heart weight and total kidney weight were adjusted for differences in body weight between the groups. Creatinine clearance was normalized to total kidney weight (TKW) to adjust for differences in animal size.

^*P < 0.05 vs. Dahl salt-sensitive rat (S) or chromosome 2 congenic strain [S.SHR(2)],

^†P < 0.05 vs. S.SHR(2) or spontaneously hypertensive rats (SHR).

Histology

Kidneys were fixed in 10% neutral buffered formalin, embedded in paraffin, cut into 3-μm sections, and stained with hematoxylin and eosin or Masson’s trichrome. Two central longitudinal sections from each kidney (n = 6, each group) were examined in a blinded fashion. Percent glomerular injury was established by using the number of glomeruli exhibiting glomerulosclerosis and/or mesangial expansion divided by the total number of glomeruli evaluated, on average 1,000 per group. Percent interstitial injury was determined by evaluation of slides stained with Masson’s trichrome. We evaluated 20 random regions of each slide using Image Pro 5.1 (Media Cybernetics) to quantify the percent fibrosis (blue staining) compared with background. Vascular and tubular injury was evaluated separately on a semiquantitative scale from 0 (normal) to 4 (severe). Vascular compartments were assessed for vessel wall thickening, sclerotic involvement, and vasculitis/perivascular inflammation. Tubules were evaluated for the presence of necrosis, hydropic change, and/or tubular casts.

Electron microscopy

For transmission electron microscopy, a 1-cm-square region of kidney cortex was obtained from each animal (n = 3, each group). Samples were fixed in 3% glutaraldehyde for 1.5 h and washed three times for 10 min each with 0.2 M sodium cacodylate buffer, postfixed for 1 h with 1% OsO₄ followed by 1 h with saturated uranyl acetate. Dehydration was carried out by a graded series of chilled ethanol solutions (30–100%) and a final dehydration with 100% acetone. Cortical samples were infiltrated overnight in Spurr’s resin (Electron Microscope Sciences) and ultrathin sections were obtained and collected on copper 200-mesh support grids. Sections were poststained with uranyl acetate and lead citrate, and examined using a Philips CM 10 transmission electron microscope.

Microarray and Analysis

DNA microarray analysis was performed using Affymetrix Gene-chip Rat Genome 230 2.0 array at three time points. The chip contains 31,000 probe sets from >30,000 transcripts on one array. Three male S and three male S.SHR(2) congenic were selected at random from each group at weeks 4, 12, and 20. Kidney was cut into ~0.5-cM-size cubes, suspended in RNAlater (Ambion, Austin, TX) and stored overnight at 4°C. RNA was extracted using TRIzol reagent (Invitrogen, Carlsbad, CA) and purified using Mini RNeasy kit (Qiagen, Valencia, CA) according to manufacturer’s protocols. RNA quality was assessed by an OD260/280 ratio >2.0 and visually by ethidium bromide staining on an agarose gel.

Biotinylated cRNA was synthesized from 10 μg of total RNA using the One-Cycle Target Labeling Kit (Affymetrix, Santa Clara, CA) as directed by the user manual. cRNA quality for each sample was assessed by hybridization of 5 μg of adjusted kidney cRNA to Affymetrix Genechip Test3 array. Subsequently, 15 μg of adjusted kidney cRNA was hybridized to the Genechip Rat 230 2.0 array. Hybridized chips were automatically washed, stained, and scanned at the UTHSC Genomics core facility using Affymetrix equipment. Data obtained from these gene expression studies are deposited in the Gene Expression Omnibus (GEO) database (http://www.ncbi.nlm.nih.gov/geo/) with the GEO accession number GSE6208.

Microarray analysis was performed using the commercially available GeneSifter.net software platform (http://www.genesifter.net). Expression values were generated by probe level analyses using GC-RMA (robust multiarray averaging with an adjustment for GC content). The GC-RMA method converts the intensities from multiple probes from a probe set into a single expression value and uses both perfect-match and mismatch probe information. The data are normalized by quantile normalization and log₂ transformed. Following the normalization process, differentially expressed gene were statistically identified by t-test using two methods at each time point: 1) family-wise error rate procedure, P < 0.05 and fold-change ± 1.5 or greater; and 2) a more stringent procedure, using Benjamini and Hochberg FDR (false discovery rate) which corrects for multiple comparison, using P < 0.05, and fold-change ± 1.5 or greater. Additionally, a two-way ANOVA using a P < 0.01 was performed to identify strain × time interaction effects between the S and S.SHR(2) from week 4 to week 20.

Gene networks and functional analysis were generated through the use of Ingenuity Pathways Analysis (Ingenuity Systems, www.ingenuity.com). Gene identifiers and corresponding expression values from data sets analyzed at week 4, 12, 20 or the strain × time interaction from week 4 to 20 from GeneSifter.net was uploaded into in the application. Each gene identifier was mapped to its corresponding gene object in the Ingenuity Pathways Knowledge Base. These genes, called focus genes, were overlaid onto a global molecular network developed from information contained in the Ingenuity Pathways Knowledge Base. Canonical pathways analysis identified the pathways from the Ingenuity Pathways Analysis library of canonical pathways that were most significant to the data set. The significance of the association between the data set and the canonical pathway was measured in two ways: 1) a ratio of the number of genes from the data set that map to the pathway divided by the total number of genes that map to the canonical pathway; and 2) A Fisher’s exact test was used to calculate a P value determining the probability that the association between the genes in the dataset and the canonical pathway is explained by chance alone.

Expression differences for Cct3 and Sfrp2 were confirmed by quantitative real-time PCR using SYBR-green dye chemistry on a Bio-Rad MyIQ Real-Time PCR machine (n = 6 each strain). Gene expression between the congenic strain and the S control were evaluated using the comparative method (ΔΔCt) of gene expression using β-actin expression as the control gene (22).

Genotyping

Genomic DNA was obtained by tail biopsy from F1[S × S.SHR(2)] × S backcross and RPT animals and prepared using Wizard SV 96 Genomic DNA kit (Promega, San Luis Obispo, CA). Genotyping was done by multiplexing (4 primer sets per PCR reaction) using a fluorescent-based approach on a Beckman Coulter CEQ8000 XL capillary sequencer. In brief, each primer is labeled with a fluorescent dye (dye D4-PA; Beckman Coulter, Fullerton, CA), which produces a PCR product that can be detected using the CEQ system. Given only one fluorescent dye was utilized, primers for multiplex PCR were selected to vary in size and allow multiple PCR products to be identified during any given CEQ run. A total of 24 markers spaced at ~5-cM intervals on chromosome 2 were used to identify recombinant animals in the F1[S × S.SHR(2)] × S backcross population, which were subsequently used for RPT.

PCR reactions were prepared as follows in a 10-μl reaction: 1× buffer containing 1.5 mM MgCl₂ (Promega, Madison, WI), 0.2 mM dNTPs (Sigma, St. Louis, MO), 2.5 pmol D4-PA each forward primer (Sigma-Proligo, Boulder, CO), 2.5 pmol each reverse primer, and 0.25 U Taq DNA polymerase (Promega). PCR amplification was performed as follows: an initial step at 95°C for 3 min and continued for 35 cycles of 94°C for 40 s, 55°C for 40 s, 72°C for 1 min. After amplification, the PCR products were diluted 1:10 in water, 1–3 μl of diluted multiplex PCR product was added to 40 μl of deionized formamide containing DNA size standard (Beckman Coulter), loaded on CEQ sequencer, and then analyzed using CEQ fragmentation analysis software.

Statistical Analysis

Linkage analysis and QTL localization were performed using Map Manager QTX (http://www.mapmanager.org) (34). The approximate confidence internal of the chromosome 2 QTL peak was estimated using the bootstrap method of the QTX program. S, S.SHR(2), and SHR time-course data was evaluated by one-way ANOVA followed by post hoc multiple comparisons using Tukey’s test (SPSS, Chicago, IL). RPT data were evaluated by an independent t-test using SPSS. All data are presented as means ± SE.

RESULTS

Characterization of S, S.SHR(2), and SHR

A time course for UPE, UAE, and BP of each group of rats is presented in Fig. 1. No detectable difference in UPE (Fig. 1A) was observed between the groups at week 4. Starting at week 5 and continuing through week 20, the S rat had significantly higher UPE compared with either the S.SHR(2) congenic or SHR (P < 0.0001). The S.SHR(2) congenic consistently maintained at least a twofold reduction in UPE compared with the S rat. The S rat rapidly progressed beyond the “proteinuria” threshold (>20 mg/24 h) between week 6 and 8, while the congenic reached this point between week 8 and 12. The SHR never progressed beyond this threshold. At week 20 S rat UPE was 270 ± 22.9 mg/24 h compared with 92.3 ± 8.3 in the congenic. Differences between groups for UAE (Fig. 1B) were similar to UPE, although the congenic maintained a greater fold reduction (3-fold) compared with the S rat. Despite the large difference between the S and congenic with respect to UPE, no detectable difference in systolic BP (Fig. 1C) or associated parameters, such as diastolic BP, pulse pressure, or heart rate was observed (data not shown).

Fig. 1.

Time course for urinary protein excretion (UPE, A); urinary albumin excretion (UAE, B); and systolic blood pressure (SBP, C) in male Dahl salt-sensitive (S), chromosome 2 congenic strain [S.SHR(2)], spontaneously hypertensive (SHR) rats. A schematic of the congenic strain is shown in A, inset. The black bar designates the extent of introgressed SHR alleles on the S background. The open region on each end of the congenic segment represents the recombination interval. In A, the dashed line represents the threshold for “proteinuria” (>20 mg/24 h). Rats were maintained on low-salt diet (0.3% NaCl) for entire course of the experiment. The number of animals at each time varied from n = 50 per group (week 4) to n = 12 per group (week 20). SBP was measured by telemetry (n = 8 per group). Mean values ± SE.

Table 1 shows additional experimental parameters between the groups at week 20. Body weights (BW) between the S and S.SHR(2) congenic were similar but were significantly greater than the SHR. No significant difference in heart weight (adjusted for BW differences) between the groups were observed, but kidneys from the S rat were significantly larger than either the congenic or SHR. Plasma sodium and potassium levels were significantly different in the S compared with either the congenic or SHR. Blood urea nitrogen was also higher in the S rat. Both total cholesterol and triglycerides levels were also significantly decreased in the congenic compared with the S, a common feature of improved renal function. Creatinine clearance, normalized to kidney weight, was not different between the groups, although plasma creatinine was significantly reduced in the SHR.

Kidney from S, S.SHR(2), and SHR were examined and evaluated at week 4, 12, and 20 for glomerular, tubular, vascular, and interstitial damage (Fig. 2). Representative histological images for week 20 animals are shown in Fig. 3. As early as week 4, small, but significant interstitial changes were observed between the S and the congenic (Fig. 2B) preceding any significant difference in UPE (Fig. 1). At week 12, glomerular damage of the congenic was significantly lower than the S, although both exhibited focal glomerulosclerosis. Tubular changes were significantly attenuated (~1.5-fold) in the congenic compared with the S (Fig. 2C). No significant vascular changes were seen among any of the groups. At week 20, the S rat had a significantly greater number of damaged glomeruli compared with the congenic (64% vs. 40%). Most notable was that the S rat experienced a significant increase in fibrosis from week 12 (9.2% ± 0.89) to 20 (17.5% ± 1.79), while there was essentially little change in the congenic (4.5% ± 0.24 vs. 6.3% ± 0.44). The SHR showed similar glomerular pathology to the congenic at week 12. Again, no significant vascular changes were seen between groups.

Fig. 2.

Time-course histological examination of male S, S.SHR(2), and SHR rats. Percent glomerular injury (A), percent interstitial injury (B), tubular injury score (C), vascular injury score (D). Bar graphs show the effect of the S.SHR(2) congenic (C) strain compared with both parental strains, S and SHR. Glomerular and interstitial injury were evaluated quantitatively (see MATERIALS AND METHODS). Tubular and vascular injury was graded on a semiquantitative scale from 0 (normal) to 4 (severe). Kidneys from 6 animals were evaluated at each time point. ND, not detectable. a, Significantly different from S.SHR(2) and SHR at P < 0.05; b, significantly different from S and S.SHR(2) at P < 0.05. P values are from a 1-way analysis of variance followed by post hoc multiple comparisons using Tukey’s test. Error bars are SE.

Fig. 3.

Representative light [×40 hematoxylin & eosin (H&E), ×20 Masson’s trichrome] and electron microscopy (EM, ×3,000) images of week 20 low-salt kidney samples from S, S.SHR(2), and SHR. S rats exhibit glomerulosclerosis and mesangial expansion (thin arrow), arterial wall thickening (open arrow), protein casts (PC), and significant fibrosis (thick arrow) compared with S.SHR(2) and SHR. Kidneys from SHR are essentially normal. *Normal arterial vessel. EM of S kidney illustrated significant glomerular disease including evidence of effacement of foot processes, focal mesangial matrix increase, and diffuse mesangial electron-dense deposits. The S.SHR(2) congenic demonstrated significantly less pathology compared with the S. The SHR appeared essentially normal, although there was some evidence of minimal effacement of foot processes (considering all data).

An ultrastructural examination of kidney from S, S.SHR(2), and SHR confirmed the light microscopy data and provided more specific detail on pathological changes between the groups (representative images are shown in Fig. 3 only for week 20). At week 4, little or no pathology was observed between groups. Glomeruli appeared to be within normal limits regarding general appearance of discrete foot processes of the visceral epithelial cells. At week 12, significant differences were seen between the S and congenic. Changes were focal in nature and did not involve all glomeruli. Little or no pathology was seen in the SHR. At week 20, significant glomerular disease including evidence of visceral cell effacement, significant glomerulosclerosis, focal mesangial matrix increase, and diffuse mesangial electron dense deposits were observed in the S (Fig. 3). S.SHR(2) kidneys were also found to have focal visceral cell effacement, focal glomerulosclerosis, focal mesangial matrix increase, and focal mesangial electron dense deposits; however, the pathologic changes were not nearly as severe as the S. SHR had some glomerular pathology including minimal effacement of foot processes, focal mesangial matrix increase, and rare mesangial electron dense deposits.

Expression Profiling and Pathway Analysis

Gene expression profiling was performed at week 4, 12, and 20 using kidney from the S and S.SHR(2) congenic to: 1) identify expression differences of positional candidates within the QTL region; 2) correlate temporal gene expression changes between the S and congenic with degree of renal damage; and 3) identify biochemical pathways potentially involved in the attenuated renal damaged observed in the congenic. Table 2 gives a summary of the number of genes found to be differentially expressed throughout the genome and on chromosome 2, striated by fold-change and statistical significance. On average, depending on fold-change threshold and statistical tests, ~40% of differentially expressed genes map within the congenic interval. A total of 727 protein-coding genes (www.ensembl.org) reside in the congenic interval and are represented on the Affymetrix Genechip, of these, anywhere from 48 or 6.6% (week 4) to 100 or 13.7% (week 20) of genes were found to be differentially expressed.

Table 2.

Number of differentially expressed genes in kidney between S and S.SHR(2) striated by fold-change and statistical significance

Analyisis Criteria	Time	FWER, P < 0.05		FDR, P < 0.05
		All Chromosomes	RN02	All Chromosomes	RN02
Fold-change ±1.5	week 4	168	48	4
	week 12	109	50	72	41
	week 20	360	100	152	53
Fold-change ±1.8	week 4	62	26	1
	week 12	54	30	42	27
	week 20	105	44	54	30
Fold-change ±2.0	week 4	35	18	1
	week 12	44	25	35	22
	week 20	58	28	35	21

FWER, family-wiseerror rate; FDR, false discovery rate.

A detailed description of genes found to be differentially expressed and that map within the congenic interval are shown in Table 3. Genome-wide analyses of differentially expressed genes are shown in Supplemental Table S1.¹ A total of 37 genes/expressed sequence tags (EST) on chromosome 2 (fold-change ± 2, FDR, P < 0.05) were observed to be differentially expressed between the S and S.SHR(2) throughout the time course. Seven of these genes (Table 3, nos. 31–37) were consistently downregulated at all three time points. In general, genes that were observed to be differentially expressed at one time point were also found to be differentially expressed at an additional time point (serving to confirm the expression data). However, there were many genes observed in entire data set (Supplemental Table S1) where expression differences were time specific.

Table 3.

Time course of genes differentially expressed in kidney between S and S.SHR(2) located on chromosome 2

Number	Accession#	Gene Name	Fold Change			Description
			Week 4	Week 12	Week 20
1	AI501417	Mgst2?	+2.4			EST
2	BE107853		+5.4	+6.4		EST
3	BF415556		+5.5	+4.7		EST
4	BG374101		+3.6	+2.7		EST
5	AA859982		+2.6	+2.2		EST
6	AW141832			+2.1		similar to RIKEN cDNA 2310050P13 (LOC310808)
7	BF551322			+2.2		similar to CG5805-PA (LOC365841)
8	BG379771			+2.8		EST
9	BF553729	Wnt2b			+2.7	EST
	AI555464	Wnt2b			+2.4	EST
10	BE097574	Kcnab1			+2.4	EST
11	AA819288				+2.3	retinoic acid receptor responder protein 1
12	BF411765	Celsr2			+2.1	cadherin EGF LAG seven-pass G-type receptor 2
13	AI715140				+2.0	similar to histone protein Hist2 h3c1 (LOC310679)
14	BF284175		+2.3	+2.4	+2.2	similar to group XII-1 phospholipase A2 (LOC362039)
15	BF396545	Sfrp2*	+2.4			secreted frizzled-related protein 2
16	AI511405		−13.2	−6.3		similar to hypothetical protein FLJ22693 (LOC310467)
17	BF522861		−2.1	−2.1
18	BF388224		−7.4	−8.7
19	AI454310		−2.0	−2.5		EST
20	BF402235	Sars1		−2.1		EST
21	AI172217			−2.9		similar to protein beta-galactosidase, alpha
22	U86635	Gstm5		−2.1		glutathione S-transferase, mu 5
23	NM_021584			−2.0	−2.8	activity/neurotransmitter-induced early gene protein 4
24	BF290410				−2.2	EST
25	BE120370				−2.1	EST
26	NM_024157	Cfi			−4.0	complement factor I
27	NM_012889	Vcam1			−2.1	vascular cell adhesion molecule 1
28	AF367210	Il7			−2.1	interleukin 7
29	NM_031120	Ssr3			−2.9	TRAP-complex gamma subunit
30	U05675	Fgb			−2.3	fibrinogen, beta polypeptide
31	BI288681		−7.3	−6.7	−3.8	transcription activator of D-serine dehydratase?
32	BE100973		−5.8	−8.7	−8.4	Rattus norvegicus transcribed sequences
33	AI454769		−4.3	−4.8	−3.9	Rattus norvegicus transcribed sequences
34	AI043891	Cct3*	−2.8	−2.8	−2.0
35	AI232217		−4.2	−3.7	−2.2	Rattus norvegicus transcribed sequences
36	NM_031154	Gstm3	−5.1	−4.7	−3.4	glutathione S-transferase, mu 3(Yb3)
37	M28241	Gstm1	−3.2	−3.8	−3.6	glutathione S-transferase, mu 1

A positive fold change means that the transcript was expressed higher in the S.SHR(2) congenic compared with the S. Conversely, a negative value means that the S transcript was expressed at a higher level than the congenic. Transcripts found to have a fold-change ±2.0 or greater are included here; see Supplemental Table S1 for a complete gene list. Statistical calculations were performed by 2 methods (see MATERIALS AND METHODS FOR MORE DETAIL): P value (FWER) and adjusted P value (FDR using Benjamani and Hochberg correction for multiple corrections). Fold change data shown are all significant at least P < 0.05. For exact P value see Supplemental Table S1.

^*Both of these genes map to the reduced quantitative trait locus (QTL) interval, differential expression was confirmed using quantitative real-time PCR. Cct3 was found to be differentially expressed at all 3 time points, with a fold-change ranging from −1.20 to −1.35 (at least P < 0.05). Sfrp2 was observed to be differentially expressed at week 4 (confirming the Affymetrix data) and also at week 12 and 20, with a fold-change ranging from +1.60 to +1.79 (at least P < 0.05).

To gain insight into the major biological processes and pathways linked to genes differentially expressed between the S and S.SHR(2) congenic, the gene expression data was evaluated using the Ingenuity Pathway Analysis program (see MATERIALS AND METHODS for details). Multiple networks were discovered in each of the three datasets (week 4, 12, and 20). Cellular movement, cellular growth and proliferation, and connective tissue development and function were recurring top scoring functions from week 4 to 20 (Supplemental Table S2 and Supplemental Fig. S1). Additionally, canonical pathways (which are the most relevant pathways based on the entire dataset) for Wnt/β-catenin signaling, amyloid processing, and antigen presentation were identified in week 4, Wnt/β-catenin signaling pathway and glutathione metabolism in week 12, and nitric oxide signaling in week 20.

A two-way ANOVA was performed on the time-course microarray data to identify genes that were significantly different between strains [S and S.SHR(2)] and over time [week 4, 12, and 20], i.e., strain × time interaction. A total of 186 genes were identified using fold-change ± 1.5 or greater and a P < 0.01 (Supplemental Table S3). Approximately one-third of these genes were not previously observed to be different when examined at each time point individually. Top functions identified involved cellular assembly, cellular movement, and immune related functions (Supplemental Fig. S2).

Fine Mapping the Renal Function Locus

RPT was employed to refine the genomic region containing the QTL responsible for the observed difference in renal function between the S and S.SHR(2) congenic. The basis of RPT is to find recombinants in an interval derived from the S.SHR(2) congenic, to propagate these recombinants, and then to measure the phenotypes of their progeny to determine the genotypic value of each recombinant. A large backcross population (n = 971) was developed from the S.SHR(2) congenic strain and the S rat to identify recombinant animals and to reduce the confidence interval for the QTL region. Bootstrapping analysis of the original linkage analysis (backcross population, n = 276) (20) localized the 95% confidence interval (CI) to a ~14-cM region delimited approximately by markers D2Rat220 and D2Rat50 (Fig. 4). The new backcross population derived from the S.SHR(2) congenic reduced the 95% CI to ~1.5 cM, flanked by D2Arb11 and D2Rat284. A total of 20 recombinant animals were selected from this population for RPT. All recombinant families were evaluated for UPE at 6 wk of age for an initial survey of the families (Fig. 4). Recombinant families 1–4 (shown in red) all demonstrated a significant reduction (40–60%) in UPE, that is, animals that carried the recombinant chromosome (congenic-like) had significantly lower UPE compared with nonrecombinant littermates (S-like). Families 5–17 (shown in yellow) showed no significant effect on UPE, which eliminates the region above D2Arb11, and below D2Rat108, from significantly contributing to the refined QTL. Families 18–20 (also shown in red) demonstrated a significant effect on UPE.

Fig. 4.

Fine mapping of renal function quantitative trait loci (QTL) using recombinant progeny testing (RPT). The physical map of rat chromosome 2 is shown on the left. The solid bars to the right of the physical map indicate the extent of the SHR-donor regions for each recombinant family. The open region represents the recombination interval. UPE was measured at week 6. The solid black bar denotes the location of the QTL based on the RPT. The 95% confidence interval (CI) obtained from the original linkage analysis (20) and the present study is shown to the right of the figure. Red bars denote that recombinant animals (congenic-like) had significantly (P < 0.05) lower UPE compared with nonrecombinant (S-like) littermates. Yellow bars denote there was not a significant difference in UPE between recombinant and nonrecombinant littermates. The number of animals tested from each recombinant family ranged from 16 to 29.

Figure 5 shows an enlargement of the QTL region and the four recombinant families (4, 6, 17, and 18) important for delimiting the QTL. To confirm the data obtained from the initial survey (Fig. 4), additional animals (n = 43–84) were tested from each family of the four critical families. UPE was evaluated in these families at 6 and 12 wk of age to determine if there was a sustainable effect on UPE in older animals (Fig. 5). Families 4 and 18 demonstrated a significant effect on UPE at week 6, which continued through week 12. For family 4 (week 12), the average UPE for nonrecombinant animals was 95.3 ± 4.94 mg/24 h vs. 56.70 ± 4.45 mg/24 h for recombinant animals. For family 18, the average UPE for nonrecombinant animals was 95.3 ± 4.84 mg/24 h vs. 54.70 ± 4.04 mg/24 h for recombinant animals. No significant effect on UPE was observed for family 5 or 17 at either week 6 or 12, demonstrating that the QTL lies in the interval flanked by D2Rat46 and D2Rat230 (Fig. 5).

Fig. 5.

Enlargement of the QTL region and the 4 recombinant families important in delimiting the QTL region. The physical map of rat chromosome 2 is shown on the left, with an enlargement of the physical map of the QTL region to the right. Map distances are in base pairs (www.ensembl.org, Ensembl v38, Apr 2006). UPE was measured at week 6 and week 12. Data for “UPE Effect” below each bar are UPE of recombinant rats (congenic-like) minus the UPE of nonrecombinant (S-like) littermates. A negative value indicates that recombinant rats had lower UPE than nonrecombinant rats, and when this is significant this indicates that the QTL is in the congenic interval. The numbers of animals tested are shown below the effect. P values are from an independent t-test. *P < 0.0001; NS is not significant. Data are means ± SE.

The refined QTL region spans 1.5 cM or ~5.0 Mb, which contains 64 known and/or predicted genes. The locations of these genes are depicted in Fig. 5 (additional information provided in Supplemental Table S4). Of the 37 differentially expressed genes found to map on RNO2 (Table 3), only two of these map to the refined QTL region. They are secreted frizzled-related protein 2 (Sfrp2) and chaperonin containing TCP1, subunit 3 (Cct3).The region also contains several candidate genes based on known function or renal involvement including, Neph1, Ras-like protein in brain 25 (Rab25), and IQ motif containing GTPase activating protein 3 (Iqgap3).

DISCUSSION

For this study we sought to investigate differences in onset and progression of renal disease in the Dahl S, S.SHR(2) congenic, and SHR using a time course. Our data clearly demonstrate that the locus on chromosome 2 has a major and sustained ability to attenuate renal damage. Not only was the number of damaged glomeruli less in the congenic, but the degree of damage was also significantly less. This is supported by the finding that albumin consistently composed a smaller proportion of the total amount of protein in the congenic compared with the control S. That is to say, a molecule the size (and charge) of albumin will only significantly compose the ultrafiltrate once the glomerular basement membrane/slit diaphragm is sufficiently compromised. This phenomenon also serves to explain why the presence of microalbuminuria is highly correlated with progression of kidney disease in humans. As early as week 4, significant interstitial changes were observed between the S and the congenic that preceded any significant difference in UPE. This suggests that the QTL may attenuate renal damage primarily through controlling fibrosis, as opposed to controlling glomerular permeability. This is further supported by the fact the congenic experienced increased glomerular damage from week 12 to 20, with no significant change in interstitial injury. Additionally, no significant difference in systolic BP was observed between the S and congenic. These data demonstrate that the reduced renal damage in the congenic is not simply a consequence of lower BP compared with the S suggesting the underlying mechanism of the QTL is unlikely to be mediated through BP regulation.

In recent years, several groups have used a combined QTL and gene expression approach to prioritize candidate genes for further study within a QTL region (27, 36, 59), while others have successfully used the approach to expedite the identification of disease-causing genes (1, 17, 28, 44). More recently, two groups have utilized a large-scale integrated approach using either a panel of recombinant inbred strains (25) or chromosome substitution strains (33) with gene expression profiling to gain insight into genes and pathways underlying disease processes. To this end, we sought to utilize gene expression analysis to identify differential expressed genes linked to the QTL and establish biochemical pathways responsible for the renal protective effect of the congenic strain.

Network analysis of differentially expressed genes consistently identified cellular movement and cellular growth and proliferation as top functions associated with gene networks at each week. Additionally, networks linked to genes that were significantly different between strains [S and S.SHR(2)] and over time (week 4, 12, and 20) were cellular assembly and organization, cellular movement, and immune response. The individual analysis provided a “snapshot” of genes differentially expressed at each time point, while the second analysis (strain × time) placed the networks in context of what occurred over time and identified additional genes of interest. The purpose of conducting the gene expression analysis at week 4 before the onset of detectable differences in UPE was to address the issue of “cause and effect.” The assumption being that networks identified at the early time point are more likely to reflect the cause and not the effect of differences in renal function such as networks observed at later time points (week 12 and 20).

At week 4 (and week 12) the Wnt/β-catenin signaling pathway was identified as the most significant pathway based on the entire dataset. Sfrp2 and Wnt2b are genes that participate in this pathway, were differentially expressed, and mapped within the QTL interval (Table 3). Wnts encode secreted proteins that bind members of the Frizzled receptor family, which regulates cell proliferation and differentiation (3, 39). Conventional Wnt signaling is mediated by stabilization of β-catenin, which then translocates into the nucleus and interacts with T-cell factor/ lymphoid enhancing factor family to alter gene expression (39). β-Catenin also mediates the interplay of adherence junctions with the actin cytoskeleton through E-cadherin (39). The Wnt/β-catenin pathway has been implicated in playing in a role in renal fibrosis by driving tubular cells to undergo an epithelial-mesenchymal transition (EMT) into activated fibroblasts (53).

Renal fibrosis represents the pathological event most correlated with loss of renal function considering all forms of renal disease (32, 60). So, does proteinuria in itself cause renal fibrosis? An attractive explanation is that albumin being the principal component of ultrafiltered protein during proteinuria causes tubular cell injury. This hypothesis has been supported by several in vitro studies (15, 61). Another more interesting hypothesis involves the translocation of growth factors accompany proteinuria [including insulin-like growth factor 1; hepatocyte growth factor, and transforming growth factor-β (TGF-β)] into the tubular fluid causing activation of tubular cells and recruitment of circulating fibrocytes (24). The “activated” tubular cells secrete chemokines [namely, monocyte chemoat-tract protein-1 (CCL2) and RANTES (CCL5)] that activate macrophages to secrete TGF-β, promoting tubular EMT and producing myofibroblasts (24). These highly differentiated myofibroblasts cells produce extracellular matrix and in coordination with matrix degrading enzymes result in fibrosis.

The gene expression/network analysis, along with the histology data, seems to support a hypothesis by which the congenic experiences reduced kidney damage primarily through controlling fibrosis. While the congenic does have significantly less UPE compared with the S, it still develops proteinuria and accompanying kidney pathology. This is not surprising given that the congenic still contains other S susceptibility alleles at other QTL (on chromosome 1, 6, 8, 9, 13, and 19, which act to increase UPE). Consistent with the above hypothesis, growth factors make it into the tubular fluid of both the S and S.SHR(2) congenic strain. However, the congenic responds differently to these growth factors and associated signaling pathways, attenuating fibrosis, which results in less renal scarring and associated glomerular damage.

RPT was used as an alternative to conventional substitution analysis to reduce the QTL interval. RPT has been suggested as one approach for fine mapping QTL with large and dominant effects (13), both of which are the case for the chromosome 2 QTL (21). Slight variations of this approach (RPT using congenic strains or interval specific congenic strains) have been utilized to substantially narrow QTL regions to small intervals, mostly in mice (8, 9, 17, 31, 40). The main benefit over conventional substitution mapping is rapid development and testing of many recombinant animals and the ability to minimize environmental noise (the phenotypic effect of the recombinant chromosomes is compared with nonrecombinant littermates, instead of between control and experimental groups). A limitation of this approach is that small and/or additive effects may go undetected because the effect of only one allele (heterozygous animals) is evaluated. That is to say, a recombinant family that exhibits no phenotypic difference between genotypes does not necessarily imply it doesn’t contain a QTL. It just means an observable effect may require both alleles to be present.

After one round of RPT, the QTL that contributes the largest and most dominant effect on UPE was significantly reduced to ~1.5 cM, eliminating many of the genes found to be differentially expressed as causative to this QTL. The region still contains many more genes than are feasible to comprehensively evaluate without further work. However, the region does contain several interesting candidate genes, two of which were differentially expressed. These were Sfrp2 and Cct3. SFRPs act as soluble modulators of Wnt signaling, so Sfrp2 could play a role in the proposed QTL mechanism discussed earlier. Cct3 belongs to a protein complex (composed of eight subunits) involved in protein folding of actin and tubulin, suggesting a possible role in maintaining cytoskeleton integrity (6). Additionally, CCT promotes the activation of the anaphase promoting complex, a complex in which inversin (INVS) also interacts (6). Mutations in INVS cause nephronophthisis type 2 (NPHP2), an autosomal recessive form of cystic kidney disease (41). Interestingly, besides kidney cysts, a prominent feature of the disorder is the development of renal interstitial fibrosis (41).

The most interesting gene, Neph1, is structurally related and interacts with Nephrin (NPHS1), a protein that localizes to the slit-diaphragm, and was found to be mutated in congenital nephrotic syndrome type 1 (29). Additionally, mice deficient in Neph1 develop severe proteinuria and die within the first weeks of life (16). Rab25 belongs to a family of small GTPases and are involved in intracellular vesicular transport and trafficking (50). Recently, a null mutation in Rab38 (family member of Rab25) has been shown to be a strong candidate for the development of proteinuria in the fawn-hooded hypertensive rat by potentially altering the tubular reuptake of filtered protein (43). The fact that neither Neph1 nor Rab25 was differentially expressed or contained coding sequence variants between the S and congenic (data not shown) most likely eliminate them being causative to the QTL. Lastly, Iqgap3 belongs to a family of proteins that are integral components of cytoskeletal regulation, which has been linked to E-cadherin-mediated cell-cell adhesion and β-catenin-mediated transcription (4).

Of importance to the current study is its relevance to human kidney disease. Rat chromosome 2 around the QTL is homologous to both human chromosome 1 and 4 (Fig. 6). The break point (between 178.523 and 179.114 Mb) occurs around the middle of the rat QTL. Interestingly, only human chromosome 1 has been linked to renal disease, whereas human chromosome 4 (homologous region to rat QTL) has not been linked to renal disease. An autosomal dominant form of medullary cystic kidney disease (MCKD1) was mapped to chromosome 1q21 using several kindreds (30, 56, 58). The disease is characterized by tubulointerstitial nephropathy that causes renal salt-wasting and end-stage renal failure (30). Cysts are common but not always present. The presence of proteinuria is variable. Recently, high-resolution haplotype analysis was performed on 16 kindreds with MCKD1 (57). Mutational analysis of 37 genes identified variations in three different genes, including CCT3 (57). Another group mapped an autosomal dominant form of progressive renal failure and hypertension to the same region using a large Israeli family (10). Clinically, all affected members present with hypertension with varying degrees of renal pathology (glomerulosclerosis, interstitial fibrosis, and tubular atrophy). A third study, which conducted a genome scan for renal function among hypertensive individuals (HyperGen study), found linkage of creatinine clearance to the same region on chromosome 1 (14).

Fig. 6.

Comparative map showing overlap of renal susceptibility loci between rat and human. The physical map of the rat QTL is shown on the left. The region in human that is homologous to the rat QTL lies on both human chromosome 1 and 4. Map distances are in base pairs (www.ensembl.org, Ensembl v38, Apr 2006). Adult onset nephropathy and hypertension (10); MCKD1, medullary cystic kidney disease (30, 56, 58); and creatinine clearance (HyperGen study) (14).

The colocalization of multiple renal disease loci clearly demonstrates the importance of this region in renal function. The fact that there are disparate etiologies to these disorders suggests that the gene(s) that underlie each disorder are not the same. However, the possibility does exist that a common gene could underlie all the disorders. Multiple alleles of the “gene” could have a distinct effect on the observed phenotype or the same allele could simply be influenced by the genetic background of the individual, resulting in a distinct disorder. Regardless, colocalization of these renal disease loci provides a means to prioritize the analysis of genes in the QTL for study until subsequent RPT can help delimit the QTL further. So, if one assumes the same gene underlie all the different phenotypes, the QTL can be reduced to 600 Kb containing 18 genes (Fig. 6, Supplemental Table S4).

A limitation of the current study relates to the relatively large size of the congenic interval used to conduct the gene expression analysis. Only two of the 37 differentially expressed genes mapped to the refined QTL. A logical question to ask is to what extent do expression differences in the other 35 differentially expressed genes (or even sequence variants in genes outside the refined QTL) influence the major gene networks observed in this study? That is, are the observed biochemical pathways unrelated to the attenuated renal damage in the S.SHR(2) congenic? Unfortunately, this question can only really be answered once the causative gene is identified and its mechanism understood. So, while the gene expression/ network analysis does provide a clue to a potential mechanism and a starting point for further investigation, the data should be approached with some caution.

We have attempted to use a combined approach using congenic strain analysis (RPT) and gene expression/network analysis to expedite the identification of the gene causative to the QTL on chromosome 2. The concordance of the rat QTL with several human renal disease loci demonstrate that elucidating the causative gene and mechanism of the rat QTL may be of particular importance for understanding human kidney disease.