Abstract
ABSTRACT. Abnormal activation of protein kinase C-β isoforms in the diabetic state has been implicated in the development of diabetic nephropathy. It is thus plausible that DNA sequence differences in the protein kinase C-β1 gene (PRKCB1), which encodes both βI and βII isoforms, may influence susceptibility to nephropathy. Nine single-nucleotide polymorphisms (SNP) in PRKCB1 were tested for association with diabetic nephropathy in type I diabetes mellitus, by using both case-control and family-study designs. Allele and genotype distributions of two SNP in the promoter (−1504C/T and −546C/G) differed significantly between case patients and control patients (P < 0.05). These associations were particularly strong with diabetes mellitus duration of <24 yr (P = 0.002). The risk of diabetic nephropathy was higher among carriers of the T allele of the −1504C/T SNP, compared with noncarriers (odds ratio, 2.54; 95% confidence interval, 1.39 to 4.62), and among carriers of the G allele of the −546C/G SNP (odds ratio, 2.45; 95% confidence interval, 1.37 to 4.38). Among individuals with diabetes mellitus duration of ≥24 yr, these two SNP were not associated with diabetic nephropathy. These positive findings were confirmed by using the family-based transmission disequilibrium test. The T-G haplotype, with both risk alleles, was transmitted more frequently than expected from heterozygous parents to offspring who developed diabetic nephropathy during the first 24 yr of diabetes mellitus. It is concluded that DNA sequence differences in the promoter of PRKCB1 contribute to diabetic nephropathy susceptibility in type I diabetes mellitus. E-mail: andrzej.krolewski@joslin.harvard.edu
Received December 18, 2002. Accepted April 26, 2003.
Diabetic nephropathy is a major determinant of the excess morbidity and premature death associated with type I diabetes mellitus (1,2⇓). Although clinical studies have clearly demonstrated that prolonged hyperglycemia is an important risk factor for the development of this complication (3,4⇓), epidemiologic and family studies suggest that genetic susceptibility also plays a critical role (5,6⇓). Therefore, identification of the susceptibility genes has become the focus of intensive research efforts (7). These investigations have predominantly used the candidate gene approach, which entails examining genes whose protein products have been implicated in the pathogenic mechanisms underlying diabetic nephropathy.
Among these mechanisms, there is ample evidence supporting the hypothesis that abnormal activation of protein kinase C (PKC) in the diabetic state is important for the development of diabetic nephropathy (8,9⇓). This activation of PKC has been observed in rat mesangial cells cultured under high-glucose conditions and in glomeruli of diabetic rats (10,11⇓). The PKC family consists of at least 12 isoforms, which are characterized by their dependence on lipids and Ca2+ for kinase activity (12). Interestingly, renal injury induced by high glucose levels is associated with the preferential activation of specific PKC isoforms, including both isoforms of PKC-β (termed PKC-βI and PKC-βII), which are activated in diabetic glomeruli (13,14⇓). Moreover, use of the pharmacologic agent LY333351, which preferentially inhibits PKC-βI and PKC-βII over other PKC isoforms, can ameliorate glomerular hyperfiltration, albuminuria (13), and mesangial expansion in rats with streptozotocin-induced diabetes mellitus (15). This inhibition of PKC-β also prevents upregulation of gene expression of transforming growth factor-β1 and matrix molecules in diabetic kidneys (14). These results strongly suggest that PKC-β may play an important role in the development of diabetic nephropathy.
Taking into consideration these observations, it is plausible that the PKC-β1 gene (PRKCB1), which encodes both PKC-βI and PKC-βII (16), may be an important susceptibility gene for diabetic nephropathy. To test this hypothesis, we sought to identify common DNA sequence differences in the promoter and coding regions of PRKCB1. These DNA sequence differences were then examined for association with diabetic nephropathy among patients with type I diabetes mellitus. A population-based approach was used, with a large case-control study design, and the results from the case-control study were verified in a family-based study using the transmission disequilibrium test (TDT) (17).
Materials and Methods
Study Groups
The Committee on Human Subjects of the Joslin Diabetes Center approved the protocol and informed consent procedures for this genetic study of diabetic nephropathy. Our laboratory has been investigating the genetic determinants underlying diabetic nephropathy for the past decade. As part of our ongoing studies, we continually recruited young adults (<45 yr of age) with type I diabetes mellitus from among the patients attending the Joslin Clinic (Boston, MA). This recruitment effort, which occurred in 1991 to 2000, was previously described in detail (18). Diabetes mellitus was classified as type I if it was diagnosed before the age of 30 yr and continuous treatment with insulin began within 1 yr after diagnosis. At the time of this investigation, 451 diabetic individuals (220 control subjects and 231 case patients) had been recruited and were included in this study. Case patients were defined as Caucasian patients with type I diabetes mellitus and persistent proteinuria or end-stage renal disease (ESRD) attributable to diabetic nephropathy. Persistent proteinuria was defined on the basis of positive findings with reagent strips (≥2+ results with Multistix strips; Bayer Corp., Diagnostics Division, Elkhart, IN) for two of three successive urinalysis samples or a urinary albumin/creatinine ratio of ≥250 mg/g (men) or ≥355 mg/g (women). Among these patients, 45.0% had ESRD as a result of diabetic nephropathy; the remaining patients demonstrated persistent proteinuria. Patients were classified as control subjects if they had exhibited type I diabetes mellitus for ≥15 yr and the albumin/creatinine ratio was <17 mg/g (men) or <25 mg/g (women) in at least two of three spot urine specimens obtained during the 2-yr period preceding enrollment in the study. Patients with microalbuminuria or intermittent proteinuria were excluded from the study.
Effect of Duration of Diabetes Mellitus
As we recently demonstrated, the duration of diabetes mellitus is an important factor to be controlled in genetic studies of diabetic nephropathy (19). A susceptibility allele would be more strongly associated with affected individuals with a short duration, rather than a long duration, of diabetes mellitus. In this analysis, the duration of diabetes mellitus was defined as the time elapsed (in years) between the diagnosis of type I diabetes mellitus and (1) the initiation of dialysis for case patients with ESRD, (2) the time of enrollment in the study for case patients with persistent proteinuria, or (3) the time of the most recent examination for control subjects. For this analysis, the study group was divided into two groups, on the basis of the median duration of diabetes mellitus among case patients (<24 and ≥24 yr). We estimated that most case patients with diabetes mellitus of <24-yr duration developed persistent proteinuria during the first 15 to 20 yr of diabetes mellitus, whereas most case patients with diabetes mellitus of ≥24-yr duration developed persistent proteinuria during the third or fourth decade of diabetes mellitus. Stratification according to the median duration (24 yr) proved to be informative in our analysis of another candidate gene (20).
Family-Based Study
In addition to the case-control study design, we used the family-based study design referred as the TDT. Like the case-control study, the TDT assesses the association between a specified allele and a disease phenotype. However, in contrast to the case-control study, which may yield false-positive results because of unrecognized population stratification in the patient sample, the TDT is robust against such confounding (17). To perform the TDT, it is necessary to know the genotypes of both parents and their offspring who manifests the phenotype of interest. In this study, the phenotype of interest is the presence of diabetic nephropathy. If a parent is heterozygous for any locus, then transmission of a particular allele to an offspring is expected to be 50% if the allele is not associated with the phenotype (null hypothesis). However, if that allele is associated with increased risk for the phenotype, then excess transmission (>50%) of the specific allele to affected offspring is expected. Conversely, deficient transmission is expected if a protective allele is associated with decreased risk for diabetic nephropathy. For the family-based TDT, genomic DNA from both parents of 154 offspring with diabetic nephropathy was available for this study.
Screening of PRKCB1 for Common DNA Sequence Differences
PRKCB1 spans approximately 375 kb on human chromosome 16p11.2 (21) and consists of 18 coding exons. For systematic identification of common DNA sequence differences in this gene, all exons (including exon/intron boundaries) were sequenced for eight Caucasian subjects with type I diabetes mellitus (four case patients and four control patients). In addition, 2.2 kb of the promoter region upstream of exon 1 was searched for common DNA sequence differences. This promoter region was previously implicated in the regulation of PRKCB1 gene expression (22).
GenBank (http://www.ncbi.nlm.nih.gov/) was queried to identify DNA sequence entries corresponding to PRKCB1 genomic DNA, and those sequences were used to design PCR primers. Accession numbers for those sequences are AC002302 (promoter region and exons 1 to 3), AJ002799 (exon 4), AJ002800 (exon 5), and AC002299 (exons 6 to 18). PCR products were sequenced with an ABI 377 automated DNA sequencer using dye terminator chemical processes (Perkin Elmer Biosystems, Foster City, CA). Polymorphisms in the promoter region were noted according to their distances upstream from the transcriptional start site of PRKCB1, as previously reported (23); this site corresponded to position 83722 in GenBank entry AC002302.
PCR Amplification and Genotyping
Genomic DNA was extracted from peripheral lymphocytes by using a phenol/chloroform method. Nine single-nucleotide polymorphisms (SNP), including six in the promoter region (denoted −1504C/G SNP, −1440G/T SNP, −546C/G SNP, −348A/G SNP, −287C/T SNP, and −238C/G SNP) and three in the exons (exon 1C/A SNP, exon 10C/T SNP, and exon 16C/T SNP), were analyzed (Figure 1). PCR (25-μl reaction volume) was typically performed with 20 ng of genomic DNA, using 0.6 U of Taq polymerase (PGC Scientific Corp., Frederick, MD), 200 μM levels of each dNTP, and 10 nM levels of the forward and reverse primers, in the presence of 1.5 mM MgCl2, for 40 cycles. Cycling parameters included denaturation at 95°C for 45 s, annealing at PCR fragment-specific temperatures for 45 s (Table 1), and extension at 72°C for 90 s, with final extension at 72°C for 10 min.
Figure 1. Genomic structure of PRKCB1 (approximately 375 kb). Exons (thick vertical lines) and common DNA sequence differences (vertical arrows) are indicated. Single-nucleotide polymorphisms (SNP) in exons 1, 10, and 16 are located in the open reading frame, but they do not cause missense amino acid substitutions. Markers located in the promoter and exon 1 were in linkage disequilibrium, with pairwise D′ values ranging from >0.50 to 1.00. Exon 10C/T and exon 16C/T SNP were in strong linkage disequilibrium with each other (D′ = 0.88) but not with the others (D′ < 0.10). The diagram is not drawn to scale.
Table 1. PCR primers and ASO probes for detection of PRKCB1 SNPa
For amplification of the promoter fragment containing the SNP at positions −546, −348, −287, and −238, as well as the fragment containing exon 1C/A SNP, a nested PCR method was used, because this part of the gene is rich in G+C content (>80%). PCR amplification was first performed in a 25-μl reaction mixture containing 20 ng of genomic DNA, 200 μM levels of each dNTP, 1.5 mM MgCl2, 0.5 U of Pfu turbo polymerase (Stratagene, La Jolla, CA), 10× PCR buffer, and 10 nM levels of each primer (first set of forward and reverse primers) (Table 1). An initial denaturation step at 97°C for 5 min was followed by 40 cycles of denaturation at 97°C for 30 s, annealing at 59°C for 30 s, and elongation at 72°C for 90 s. The PCR ended with 10 min of incubation at 72°C. The second (nested) PCR amplification was performed in a similar manner, except that 1 μl of the first PCR product was used as the template for PCR amplification with the second (nested) set of primers (Table 1).
With the exception of the exon 10C/T SNP, genotyping for all polymorphisms was performed by hybridization with allele-specific oligonucleotide probes (Table 1) (18). For the exon 10C/T SNP, genotyping was performed by digestion of the 404-bp PCR product with the restriction endonuclease SmaI (New England Biolabs, Beverly, MA). The digestion mixture was then separated by electrophoresis through a 1.5% agarose gel. When the T allele was present, the restriction site was absent and the intact PCR product migrated as a single band. When the C allele was present, the SmaI site was created, and the PCR product was observed as two bands (237 and 167 bp) after digestion.
Real-Time RNA Quantitation
To examine the relationship between risk genotypes and expression of PKC-βI and PKC-βII, we used Epstein-Barr virus-immortalized lymphoblasts (24) that had been previously established in our laboratory. DNA from the 57 cell lines was genotyped for −1504C/T and −546C/G SNP. Three cell lines homozygous for the risk haplotype (−1504T/T and −546G/G) and four cell lines homozygous for the protective haplotype (−1504C/C and −546C/C) were selected for functional studies. The cell lines were recovered from liquid nitrogen storage and cultured in RPMI 1640 growth medium containing 15% fetal calf serum, 1 mM glutamine, 105 IU/liter penicillin, 100 mg/liter streptomycin, and either 5 mM (low) or 25 mM (high) glucose. The same batch of serum was used throughout the study. The RPMI 1640 medium was buffered with 24 mM NaHCO3 (pH 7.4, in 95% air/5% CO2). The cells were incubated at 37°C and were fed by doubling of the volume of the complete growth medium every 3 to 4 d until they reached a volume of 10 ml, with an approximate density of 106 cells/ml. One day before harvesting, the cells were fed with complete growth medium. The cell pellets were washed several times with phosphate-buffered saline and were used for RNA extraction. Total RNA was extracted and treated with DNase by using a commercially available kit (Qiagen, Valencia, CA). Real-time RNA quantitation was performed with TaqMan One-Step RT-PCR Master Mix, in conjunction with an ABI Prism 7700 sequence detection system (Applied Biosystems, Foster City, CA), according to the recommendations of the manufacturer. Each data point was measured in triplicate. PKC-β isoforms were designated as described by Kubo et al. (25). PKC-βI mRNA quantitation was performed with the primers AACTTCGACAAAGAGTTCACCAGA (forward) and GGTCCAAGTTCATGATGAAGAGTTT (reverse), together with the TaqMan probe 5′-FAM-AGCCTGTGGAACTGACCCCCACTGA-TAMRA-3′. PKC-βII mRNA detection was performed with the primers TGCTGAAAACTTCGACCGATT (forward) and TTCCTGGTCGGGAGGTGTT (reverse), with 5′-FAM-TCACCCGCCATCCACCAGTCC-TAMRA-3′ as the probe. PKC-β mRNA levels were normalized to endogenous glyceraldehyde-3-phosphate dehydrogenase levels, as assessed with glyceraldehyde-3-phosphate dehydrogenase TaqMan PDAR controls (Applied Biosystems). The validity of these assays was confirmed by ensuring that the targets (PKC-βI and PKC-βII) and glyceraldehyde-3-phosphate dehydrogenase exhibited similar PCR efficiencies, as demonstrated by a constant ΔCT value (CT,Target − CT,Control) for a range of at least 3 log units of RNA concentrations (1000-fold difference); a regression coefficient of <0.1 was obtained when log input RNA amount was plotted against ΔCT. In this assay, a higher ΔCT value corresponds to a smaller amount of target mRNA.
Statistical Analyses
The study groups were compared by using chi-squared and t tests for categorical and continuous variables, respectively (SAS for Windows, version 6.12; SAS Institute Inc., Cary, NC). Odds ratios and their confidence intervals were used to estimate relative risks (26). Logistic regression models (SAS for Windows, version 6.12) were used to adjust for differences between case patients and control patients in the distributions of glycemic control (hemoglobin A1c levels) and BP and to test diabetes mellitus duration as a modifier of the effect of genotype on the risk of diabetic nephropathy. In these models, diabetes mellitus duration, hemoglobin A1c levels, and systolic and diastolic BP were divided at their medians for the group of case patients and were entered into the model as indicator variables.
For the TDT, excessive deviation from the 50% transmission expected with the null hypothesis was evaluated with McNemar’s test (17). Transmission of two-locus haplotypes in TDT trios was determined with the tdt2 program in Genehunter (version 2.1) (http://www.fhcrc.org/labs/kruglyak/Downloads/index.html). Haplotype frequencies were estimated by using the EH program (27).
Results
Identification of Common DNA Sequence Differences
We sequenced 2.2 kb of the promoter region and all exons (including exon/intron boundaries) of PRKCB1 for eight subjects (16 chromosomes). A total of nine DNA sequence differences were identified, including six SNP in the promoter region and three SNP in the coding exons (Figure 1). None of the three coding SNP caused missense amino acid substitutions.
Case-Control Comparisons
Characteristics of control patients and case patients are presented in Table 2. Gender proportions were comparable for the two groups. Although control patients were 1 yr older than case patients at the time of diabetes mellitus diagnosis (P < 0.01), their diabetes mellitus duration was longer (P < 0.01). Hemoglobin A1c levels and systolic and diastolic BP at the time of examination were significantly higher for case patients, compared with control patients (P < 0.001 for each parameter). Forty-five percent of case patients had already developed ESRD as a result of diabetic nephropathy.
Table 2. Selected clinical characteristics of subjects with type I diabetes mellitus, according to nephropathy statusa
All nine SNP were genotyped in the case-control study. Genotype distributions did not significantly deviate from those expected with the Hardy-Weinberg equilibrium for either group. For four SNP, namely, −1504C/T (P = 0.01), −546C/G (P = 0.003), −348A/G (P = 0.03), and −287C/T (P = 0.04), the minor allele was significantly more frequent among case patients than control patients (data not shown). The difference for the −546C/G SNP remained statistically significant even after the most conservative multiple-comparison adjustment (Bonferroni correction). With a more realistic adjustment that took into account the strong linkage disequilibrium among the markers (i.e., their lack of independence), the differences for the other three SNP remained significant or borderline significant. For two of the four SNP (−1504C/T and −546C/G), genotype distributions were also significantly different between case patients and control patients (Table 3). Heterozygous and homozygous carriers of the minor allele were more frequent among case patients with nephropathy than among control patients with normoalbuminuria; this pattern is consistent with a dominant mode of inheritance. Carriers of the T allele of the −1504C/T SNP demonstrated increased odds of diabetic nephropathy, compared with noncarriers (odds ratio, 1.55; 95% confidence interval, 1.06 to 2.28). Similarly, carriers of the G allele of the −546C/G SNP demonstrated an increased risk of diabetic nephropathy, compared with noncarriers (odds ratio, 1.74; 95% confidence interval, 1.19 to 2.52). For both SNP, the frequency of risk allele carriers was increased among case patients regardless of whether the case patients exhibited proteinuria at the time of examination or had already developed ESRD secondary to diabetic nephropathy (data not shown).
Table 3. Genotype distributions of PRKCB1 SNP, according to nephropathy status
Effect of Duration of Diabetes Mellitus
The duration of diabetes mellitus is an important effect modifier of genetic susceptibility to diabetic nephropathy (19). Simulation studies performed by our group demonstrated that focusing on individuals who developed diabetic nephropathy early in the course of diabetes mellitus could significantly improve the power of case-control and TDT studies to detect alleles contributing to susceptibility to diabetic nephropathy (19). Therefore, for this study we stratified the study group into two groups according to the median diabetes mellitus duration among case patients (24 yr), one group with a mean diabetes mellitus duration of 19 yr and the other with a mean duration of 31 yr. The two groups were similar with respect to all characteristics presented in Table 2 except for age at the time of diabetes mellitus diagnosis, which was 14 yr for those with <24-yr duration and 9 yr for those with longer diabetes mellitus duration. As in the total group, control patients demonstrated a slightly older age at diagnosis and longer diabetes mellitus duration, compared with case patients.
The frequencies of carriers of the T risk allele of the −1504C/T SNP and carriers of the G risk allele of the −546C/G SNP among case patients and control patients are presented separately in Table 4 for patients with diabetes mellitus duration of <24 yr and those with diabetes mellitus duration of ≥24 yr. For patients with diabetes mellitus duration of ≥24 yr, these two SNP were not significantly associated with diabetic nephropathy. However, in the group with diabetes mellitus duration of <24 yr, these two SNP were strongly associated with diabetic nephropathy. Specifically, carriers of the T risk allele of the −1504C/T SNP demonstrated a considerable personal risk of diabetic nephropathy, compared with noncarriers (odds ratio, 2.54; 95% confidence interval, 1.39 to 4.62) (Table 4). Similar results were obtained for the −546C/G SNP, with carriers of the G risk allele being at greater risk of diabetic nephropathy, compared with noncarriers (odds ratio, 2.45; 95% confidence interval, 1.37 to 4.38) (Table 4).
Table 4. Association of risk alleles for SNP at −1504C/T and −546C/G with diabetic nephropathy, stratified by median diabetes mellitus durationa
To test the significance of the risk differences associated with the risk alleles of these two SNP in short- and long-duration diabetes mellitus, we examined all 231 case patients and 220 control patients with logistic regression models that included an indicator variable for diabetes mellitus duration of <24 yr and another indicator variable for carrier status. In the model for the T risk allele of the −1504C/T SNP, the term for interaction between genotype and duration of diabetes mellitus was statistically significant (P = 0.02). In the corresponding model for the G risk allele of the −546C/G SNP, the difference in risk associated with the G risk allele was not significant (P = 0.13). To adjust for differences in the distributions of hemoglobin A1c levels and BP, additional models were used. Hemoglobin A1c levels and systolic and diastolic BP (coded as indicator variables if above the median) were all associated with significantly increased risk of diabetic nephropathy, regardless of the duration of diabetes mellitus. Moreover, with the inclusion of these important covariates in the models, the estimates of the effects of the risk alleles increased slightly and both became statistically significant (P = 0.01 for each risk allele; data not shown).
Haplotype Analysis
Linkage disequilibrium between the −1504C/T and −546C/G SNP was very strong. Lewontin’s D value (a commonly used measure of linkage disequilibrium) was 0.91 (28). The frequencies of the four haplotypes formed by these two SNP were estimated separately among case patients and control patients, according to the duration of diabetes mellitus (Table 5). The haplotype without either risk allele, i.e., C-C, predominated in all groups; among patients with diabetes mellitus of <24-yr duration, however, it was less frequent among case patients than among control patients. The deficit was made up by an increased frequency of the haplotype carrying both risk alleles, T-G, which was almost twice as frequent among case patients as among control patients (24% versus 13%). The remaining two haplotypes, with one risk allele each (C-G and T-C), were infrequent and did not differ in frequency between the case patients and control patients. Among patients with diabetes mellitus duration of ≥24 yr, the distributions of haplotypes in the two study groups were comparable.
Table 5. Estimated frequencies of haplotypes formed by SNP at −1504C/T and −546C/G for case and control patients, stratified by diabetes mellitus duration
TDT
Theoretically, the positive findings in our case-control study might have been attributable to unrecognized population stratification in our patient sample. To substantiate the case-control results, we examined whether these findings could be confirmed with the family-based TDT. Because the positive findings in the case-control study were stronger for patients with a short duration of diabetes mellitus (<24 yr), TDT data were analyzed after stratification according to diabetes mellitus duration. In trios in which the diabetes mellitus duration of the offspring with diabetic nephropathy was <24 yr, transmission of the T-G risk haplotype was 70%, significantly exceeding the 50% expected with the null hypothesis (P = 0.03) (Table 6). In trios in which the diabetes mellitus duration of the offspring with diabetic nephropathy was ≥24 yr, transmission of the risk haplotype was 35%. Although this transmission does not quite represent a statistically significant decrease from the expected 50%, it is significantly less than the 70% transmission of this haplotype in shorter-duration trios (Table 6).
Table 6. Transmission of the risk T-G haplotype formed by −1504C/T and −546C/G SNP from heterozygous parents to offspring with diabetic nephropathy, stratified by diabetes mellitus duration
Real-Time RNA Quantitation
The mRNA levels for the PKC-βI and PKC-βII isoforms were measured (as relative expression levels in comparison with GAPDH mRNA, see Methods) in lymphoblasts from individuals homozygous for the T-G risk haplotype (n = 3) or the C-C protective haplotype (n = 4) (Table 7). For PKC-βI, mRNA levels were not influenced by growth in high-glucose versus low-glucose medium, regardless of PRKCB1 genotype. For PKC-βII, however, mRNA levels in lymphoblasts from individuals homozygous for the risk haplotype were significantly reduced with growth in high-glucose medium, in comparison with the level of expression in the same cells grown in low-glucose medium (5.78 ± 1.13, compared with 5.09 ± 1.21, P = 0.04). In contrast, PKC-βII mRNA levels in lymphoblasts from individuals homozygous for the protective C-C haplotype were reduced in response to hyperglycemia but the difference was not statistically significant.
Table 7. Real-time quantitation of PKC-βI and -βII mRNA levels in lymphoblasts, according to genotype and cell culture conditionsa
Discussion
In addition to hyperglycemia (3), genetic susceptibility is recognized as an important factor influencing the development of diabetic nephropathy (7). In this study, we generated substantial evidence supporting the hypothesis that polymorphisms in PRKCB1, which encodes PKC-βI and PKC-βII (16), were associated with diabetic nephropathy. Our findings were derived from both population- and family-based studies. With these two different but complementary approaches, a consistent pattern of association between two SNP (−1504C/T and −546C/G) in PRKCB1 and diabetic nephropathy was evident. This study thus represents the first implicating PRKCB1 as a susceptibility gene for diabetic nephropathy in type I diabetes mellitus.
We currently do not know how these SNP act to influence the expression of PRKCB1. Although both SNP are located in the promoter and presumably could influence PRKCB1 gene transcription, the functional roles of the −1504C/T and −546C/G SNP are unknown. In silico analyses indicate that the SNP may lie in potential binding sites for transcription factors. For instance, −1504C/T SNP is located in a potential binding site for the transcription factor Sp1 (D.P.K. Ng, L. Wyrwicz, and A.S. Krolewski, unpublished observations); diabetes mellitus-induced glycosylation of Sp1 may be involved in the pathogenesis of diabetic complications (29). However, without direct evidence from functional studies, it is premature to conclude that −1540C/T and −546C/G SNP are the true functional polymorphisms underlying the observed associations. Although we were thorough in ensuring that all regions of PRKCB1 with known functional properties were screened for common DNA sequence differences, this approach might not have been exhaustive, because additional regulatory elements that have not been experimentally characterized might exist.
Nevertheless, to gain preliminary insights into this issue, we examined whether PKC-βI and PKC-βII mRNA levels in lymphoblasts were influenced by the presence of the T-G risk haplotype or the C-C protective haplotype. Although PKC-βI levels remained largely unchanged, exposure to high glucose concentrations decreased PKC-βII levels, compared with levels observed with low glucose concentrations. This reduction was primarily observed in T-G cells, with little change in C-C cells. Because increased activity of PKC-β isoforms has been implicated in diabetic nephropathy (8), our preliminary findings were somewhat unexpected. Despite the positive result, we are cautious in drawing a firm conclusion, because this functional study was conducted with a limited number of cell lines. More comprehensive efforts will be required to confirm these initial findings and to fully delineate how DNA sequence differences at this locus affect the complex molecular mechanisms that control PRKCB1 gene expression (22,23,30,31⇓⇓⇓). Reporter gene assays may be considered in such future endeavors, but interpretation of the results may not be straightforward. For instance, it is distinctly possible that the functional polymorphisms responsible for the genetic association could be located in regulatory regions of PRKCB1 that have not yet been experimentally identified. Therefore, the absence of any observed effect in experiments in vitro might be attributable to a failure to transfect a relevant gene fragment into the cells being investigated.
It is noteworthy that the associations reported here were more significant among patients with diabetes mellitus of shorter (<24-yr), rather than longer (≥24-yr), duration. The cutoff value of 24 yr was determined a priori, on the basis of previously recorded information on the median diabetes mellitus duration (20), as described in Materials and Methods. The rationale for this stratification was to identify patients with shorter diabetes mellitus duration at the onset of proteinuria, because we previously demonstrated that this group was ideal for the identification of nephropathy genes (19). Lacking a direct estimate of the duration of diabetes mellitus at the onset of proteinuria, we chose to use the diabetes mellitus duration at the time of enrollment for patients with proteinuria and the duration at the initiation of dialysis for those with ESRD. Therefore, our covariate diabetes mellitus duration reflects a time point approximately one decade after the onset of proteinuria. However, even with this imperfect measurement, we observed a very striking pattern of allelic transmission (Table 6). Specifically, the <24-yr duration group exhibited excess transmission of the T-G haplotype, whereas the ≥24-yr duration group exhibited deficient transmission of the same haplotype. This is precisely the pattern we would expect for a broad class of genetic models of diabetic nephropathy in which the duration of diabetes mellitus is important (19).
An alternative explanation for the observed duration effect is related to possible mortality effects; carriers would have to be subject to a higher risk of death (possibly resulting from ESRD), which would lead to selective depletion of genotypes. Arguing against this explanation, however, is the fact that the frequencies of carriers were similar among case patients regardless of whether they demonstrated proteinuria or ESRD at the time of enrollment.
In conclusion, our investigation (with both case-control and TDT study designs) provided clear consistent evidence that PRKCB1 polymorphisms are associated with diabetic nephropathy, with carriers having considerably greater personal risk of diabetic nephropathy, compared with noncarriers. The findings were strengthened when our analysis focused on patients with diabetes mellitus duration of <24 yr. Such findings are consistent with the hypothesis that the subset of patients who develop diabetic nephropathy after a short duration of diabetes mellitus is most likely to include patients with the heaviest genetic burden for this complication. Finally, in view of the ongoing phase II clinical trial evaluating the efficacy of the PKC-β inhibitor LY333531 in preventing diabetes mellitus-induced albuminuria, we propose that our findings may have significant implications for interpretation of the eventual results of that study and future phase III studies. Importantly, genetic susceptibility attributable to the PRKCB1 locus could be a reason for any potential failure of that drug to achieve target clinical outcomes among poor responders.
Acknowledgments
This research was supported by National Institutes of Health Grants DK41526 and DK53534. Dr. Araki was supported by a grant from Manpei Suzuki Diabetes Foundation (Japan). Dr. Ng is the recipient of a Biomedical Sciences International Fellowship from the Agency for Science, Technology, and Research (Singapore). Drs. Araki and Ng contributed equally to this work.
- © 2003 American Society of Nephrology
References
Jump to section
More in this TOC Section
Cited By...
- Vascular endothelial growth factor stimulates protein kinase C{beta}II expression in chronic lymphocytic leukemia cells
- Polymorphisms of the Protein Kinase C-{beta} Gene (PRKCB1) Accelerate Kidney Disease in Type 2 Diabetes Without Overt Proteinuria
- Assessment of 115 Candidate Genes for Diabetic Nephropathy by Transmission/Disequilibrium Test
- Pathogenetic Mechanisms of Diabetic Nephropathy
- Scrutiny of the Glutamine-Fructose-6-Phosphate Transaminase 1 (GFPT1) Locus Reveals Conserved Haplotype Block Structure not Associated With Diabetic Nephropathy