Modifier Locus on Mouse Chromosome 3 for Renal Vascular Pathology in AT1A Receptor-Deficiency
We previously showed that the phenotype of mice with targeted disruption of the gene encoding the AT1A receptor (Agtr1a), the major murine AT1 receptor isoform, is strongly influenced by recessive genetic modifiers derived from the C57BL/6 or 129 inbred strains. To further evaluate the genetic modifiers on the C57BL/6 background, we performed backcrosses between F1(C57BL/6×129) and C57BL/6 Agtr1a−/− mice and analyzed the progeny, focusing on the development of structural lesions in the renal vasculature. In affected animals, these lesions are characterized by medial thickening of small arteries and arterioles in the kidney that are reminiscent of vascular lesions in patients with nephrosclerosis. Among 180 consecutive progeny, 170 (94%) survived to completion of the study. On masked pathological examination at age 8 months, 86 had intermediate to severe vascular lesions whereas 84 had no detectable lesions. Based on a hypothetical model of a single recessive modifier locus arising from the C57BL/6 background, the observed proportion of affected animals among the backcross progeny was not statistically different from that predicted by χ2 analysis (51% versus 50%; P=0.88). We next performed genomic microsatellite analysis in a subset of 121 backcross progeny using a panel of markers spanning ≈15 cM intervals across the mouse genome. By 2-point analysis, we found a region spanning 5 cM on chromosome 3, with significant linkage to the development of renal vascular lesions (LOD score: 3.3 to 3.8).
The renin-angiotensin system (RAS) plays a key role in the regulation of blood pressure and sodium and fluid homeostasis. Dysregulation of the RAS can contribute to the pathogenesis of a variety of disorders, including hypertension, congestive heart failure, and diabetic nephropathy. In humans, polymorphisms in the renin, angiotensin, ACE, and AT1 genes have been identified. Some of these are associated with the development of cardiovascular diseases.
Gene-targeting experiments have provided new insights into physiological roles for genes in the RAS. For example, mice that are unable to generate angiotensin II such as angiotensinogen (Agt), angiotensin-converting enzyme (Ace), or double-renin (Ren1Ren2) knockout mice have similar phenotypes characterized by reduced survival, low blood pressure, and abnormal kidney structure.1–5 A virtually identical phenotype is seen in mice with combined deletions of the two AT1 receptor isoforms (Agtr1a and Agtr1b).6,7 Thickening of arterioles and interlobular arteries observed only in the renal vasculature,8 characterized by hypertrophy and/or hyperplasia of the media, and associated with focal areas of inflammation develop in these animals. This pathology closely resembles the vascular lesions seen in hypertensive nephrosclerosis, but it occurs in animals with blood pressures (BP) that are 20 mm Hg below normal.
While the major features of the phenotypes of Agt−/−, Ace−/−, and combined Agtr1a−/−/Agtr1b−/− mouse lines produced independently by several investigative groups have many similarities,2,3,7,9–11 there is substantial variation in the reported phenotypes between lines with virtually identical null mutations on mixed genetic backgrounds. Some of this variability includes differences in the severity of postnatal mortality and the degree of hypotension. These differences probably reflect the heterogeneity of the genetic backgrounds of these lines, indicating the potential for background genes to modify functions of the RAS.
In our initial analysis of Agtr1a−/− mice on mixed genetic backgrounds, we found that kidney histopathology was normal in most animals with the exception of hypertrophy of the juxtaglomerular apparatus and occasional mesangial expansion.6,12 However, among isolated litters of Agtr1a−/− animals, we noted severe abnormalities of kidney structure, including vascular lesions that were typical of those that we had observed in Agt−/− and Agtr1a−/−/Agtr1b−/− mice.3,10 This suggested the presence of genes in the 129 and/or C57BL/6 backgrounds that could substantially modify the phenotype of AT1A receptor-deficiency. The purpose of this study was to characterize and localize these modifying genes.
To systematically examine the effects of genetic background, Agtr1a−/− mice were generated on 2 inbred backgrounds, 129/J and C57BL/6. Inbred Agtr1a−/− 129 mice were produced by breeding our original chimeras12 with 129+/+ mice and intercrossing the resulting Agtr1a+/− heterozygotes. To generate inbred Agtr1a+/− C57BL/6 animals, the Agtr1a null mutation was backcrossed onto the C57BL/6 background for >6 generations. These mice were intercrossed to produce Agtr1a−/− C57BL/6 mice with a genetic background expected to be >90% from C57BL/6. Agtr1a genotypes were determined by Southern blot analysis.12 Animals were bred and maintained at the Durham Veterans Affairs Medical Center under NIH guidelines.
Generation of Intercrosses and Backcrosses of Agtr1a−/− Mice
F1(129×C57BL/6) Agtr1a−/− mice were generated by mating inbred 129 Agtr1a+/− and inbred C57BL/6 Agtr1a+/− mice. F1(129×C57BL/6) Agtr1a−/− mice aged 2 to 6 months were crossed to generate F2 Agtr1a−/− mice. In a separate cross, F1(129J×C57BL/6) Agtr1a−/− mice were crossed with C57BL/6 Agtr1a−/− mice.
The number of Agtr1a−/− animals that survived to weaning at 21 days was compared with that predicted.
Evaluation of Kidney Histomorphology
At age 8 months, progeny from all crosses were euthanized and the kidneys were harvested and fixed in formalin. This age was chosen to ensure accurate vascular phenotyping as these lesions become more severe with age. The fixed kidneys were paraffin-embedded, sectioned, and stained with periodic acid Schiff (PAS). Vascular abnormalities were assessed by a renal pathologist (A.F.) who was masked to the experimental genotypes. Animals with detectable renal vascular abnormalities were classified as affected whereas animals with no detectable abnormalities were classified as unaffected (Mendelian trait). Vascular lesions were graded as a quantitative trait (QT) on a “severity” scale of normal, mild, moderate, or severe. Mild lesions were defined as rare arteries/arterioles with mild to moderate thickening caused by vascular smooth muscle cell hyperplasia/hypertrophy. Moderate lesions were defined as up to ≈25% of vessels with these hyperplastic/hypertrophic lesions. Severe lesions were defined as involvement of >25% of vessels with moderate to severe vascular hyperplasia/hypertrophy.
Blood Pressure Measurement
Systolic blood pressure (SBP) was recorded in conscious mice aged 4 to 6 months using a computerized tail-cuff system (Visitech Systems) that was validated and described previously.13
Eighty-nine animals from the C57BL/6 backcross were screened using 121 microsatellite markers (Applied Biosystems) spaced at ≈15-cM intervals throughout the 19 mouse autosomes. The PCR reactions were amplified in Gen Amp PCR System 9700 thermal cycler (Applied Biosystems). The amplified DNA samples were run on an ABI PRISM 3100 Sequencer, and data analysis was performed using GeneScan Analysis 3.1 software (Applied Biosystems).
Genetic distances were obtained from public databases (Whitehead Institute and Mouse Genome Database). The full set of 121 markers was used to score 89 C57BL/6 backcross progeny. An additional 11 markers were used to narrow the region of interest in 119 animals. Genotyping errors were assessed by evaluation for adherence to Mendelian inheritance and for the presence of unlikely double-recombinant events. The order of markers across the genome was determined using available databases and the internal order that resulted in the least number of double recombinants.
Linkage and quantitative trait loci (QTL) analyses were performed for the vascular phenotype by considering the trait as “affected” (Mendelian) or “severity” (QT normal, mild, moderate, severe), respectively, using Map Manager QTXb17 software.14 For the Mendelian trait “affected,” linkage was tested using G-statistic for independence.15 For the QTL, single-locus association tests16 were performed between each marker and the vascular phenotype. Simple interval mapping was performed at 1-cM intervals. Permutation tests were performed at 1-cM intervals for 10 000 permutations to establish linkage thresholds. Suggestive (P=0.63), significant (P=0.05), and highly significant (P=0.001) were established using guidelines suggested by Lander and Kruglyak17 and correspond to the 37th, 95th, and 99.95th percentiles, respectively. Likelihood ratio statistic (LRS) value was divided by 4.6 to obtain the equivalent logarithm of the odds (LOD) score.
Survival of Agtr1a−/− Mice Is Affected by Genetic Background
We evaluated survival of 3-week-old Agtr1a−/− weanlings generated from heterozygous matings of inbred 129 and C57BL/6 Agtr1a+/− and compared it with Mendelian predictions. Shown in Figure 1, from heterozygote Agtr1a+/− matings, the expected percentage of Agtr1a−/− mice is 25%. The percentage of viable 129 Agtr1a−/− weanlings was reduced to only 10% of predicted (2 observed versus 20 predicted; P<0.001). The percentage of C57BL/6 Agtr1a−/− mice surviving to weaning was similarly reduced to only 30% of predicted (17 observed versus 60 predicted; P<0.001). The reduction in survival in the inbred lines is similar to that of Agt−/− mice within our colony. To determine inheritance pattern of the survival phenotype, we next evaluated survival in progeny of 129 Agtr1a+/−×C57BL/6 Agtr1a+/− intercross. In contrast to the parental strains, survival of F1(129×C57BL/6) Agtr1a−/− mice was not different from predicted (57 observed versus 51 expected; P=not significant). Thus, on 2 distinct genetic backgrounds, the absence of AT1A receptors results in diminished survival. This survival disadvantage is abrogated in F1 animals, suggesting a recessive pattern of inheritance.
Kidney Histomorphology in Inbred Agtr1a−/− Mice
In the 129 and C57BL/6 Agtr1a−/− lines (n=6 per line), diffuse renal vascular lesions were present. These lesions were characterized by extensive medial hypertrophy/hyperplasia, often associated with perivascular inflammation (data not shown). There were no consistent differences in the character of vascular lesions between the 129 and C57BL/6 Agtr1a−/− lines. AT1A receptor-deficiency on both inbred backgrounds was associated with pathology of the interlobular vessels and severe hypoplasty of the inner medulla. These abnormalities are virtually identical to those seen in angiotensin-II-deficient mice.3,7,9,10,18 By contrast, in the F1(129 ×C57BL/6) Agtr1a−/− mice, we found that with the exception of juxta-glomerular apparatus hypertrophy, kidney histomorphology was essentially normal.
The results of these studies clearly demonstrate that there are genetic modifiers on the 129 and C57BL/6 backgrounds that dramatically alter the phenotype of AT1A receptor-deficiency. The survival and normal kidney phenotypes of the F1(129×C57BL/6) Agtr1a−/− mice indicate the existence of at least 1 distinct recessive modifier from each parental strain which, when homozygous, can worsen the Agtr1a−/− phenotype.
Segregation of Survival and Kidney Vascular Lesion Traits in F1 Intercrosses
Based on the characterization of the inbred and F1 Agtr1a−/− lines, we hypothesized that there are 2 recessive modifiers, one from the C57BL/6 strain and a different one from the 129 strain, that confer susceptibility to perinatal mortality and to the development of renal vascular lesions. To test this hypothesis, we performed F1 intercrosses and evaluated the prevalence of the survival and kidney traits in Agtr1a−/− mice.
Based on our hypothetical model and the observed survival rate in the inbred lines, the predicted mortality rate of F2 Agtr1a−/− progeny is 42%. Among 70 consecutive F2 progeny (F ×F1) (Figure 2), only 4 pups died before weaning, giving a mortality rate of 6%. This was significantly less than that predicted by our 2-locus model, suggesting that there are multiple modifying loci for the survival trait. Of those that survived, all lived to age 8 months, at which time their kidney histomorphology was examined. Thirty-three of 66, or 50%, had typical renal vascular pathology (Figure 3), whereas the remaining animals were unaffected. Among the affected, there was a range of severity of vascular pathology from mild to moderate to severe. The observed proportion of F2 animals affected with vascular pathology was not significantly different from the 44% predicted (P=0.39), thus supporting a model of 2 distinct recessive modifier loci, one from each strain, that confer susceptibility to vascular pathology.
Blood Pressure Phenotype of Affected and Unaffected F2Agtr1a−/− Mice
As part of the characterization of the F2 Agtr1a−/− progeny, we also measured their BP. Among the affected animals, there was a gradation in severity of vascular lesions from mild to moderate to severe. Because of the potential for abnormal renal function to alter BP regulation, the few animals with severe vascular pathology were excluded from the analysis. Blood pressure in all F2 Agtr1a−/− animals was significantly lower than that in wild-type controls (data not shown). However, among the F2 Agtr1a−/− mice, BP was significantly different between affected and unaffected animals. SBP was significantly lower in affected mice with intermediate severity (mild or moderate), which had a mean SBP of 72±1.8 mm Hg (n=21) compared with unaffected mice with a mean SBP of 80±2.3 mm Hg (n=27) (P=0.009) (Figure 4).
Segregation of Survival and Kidney Vascular Lesion Traits in F1×C57BL/6 Backcross
Because C57BL/6 Agtr1a−/− mice are more robust breeders than the 129 Agtr1a−/− mice, we first focused on the modifiers from the C57BL/6 parental strain. We performed backcrosses between F1(129J×C57BL/6) Agtr1a−/− mice and C57BL/6 Agtr1a−/− mice, generating 180 consecutive C57BL/6 backcross progeny. Among these (Figure 5), only 10 died before weaning, giving a mortality rate of 5%, which was significantly less than the 46.5% predicted (P<0.0001), again suggesting that the survival trait is affected by multiple modifying loci. Of the 170 that survived, 86 animals were affected with renal vascular pathology, ranging from mild, moderate, and severe, and 84 animals had normal kidney vasculature. The proportions observed were virtually identical to those predicted (P=0.878) by a 1-locus model of a single recessive modifier on the C57BL/6 background that confers susceptibility to vascular pathology.
To map the gene that modifies the renal vascular phenotype in the C57BL/6 background, we initially screened 89 animals from the C57BL/6 backcross using 121 informative microsatellite markers spaced ≈15 cM across the genome. For the Mendelian trait “affected,” we identified a region on chromosome 3 spanning 15 cM that approached significant evidence for linkage with LOD scores ranging from 2.0 to 2.4. We examined an additional 30 animals with 11 more markers on chromosome 3 and identified marker D3Mit345 with a LOD score of 3.8 that was significant for linkage. This marker was flanked by D3Mit103 (LOD 3.7) and D3mit57 (LOD 3.3) (Figure 6a), defining a region that spans 5 cM.
We also performed a separate analysis for QTL linked to the trait “severity” using simple interval mapping. This analysis also identified, in the same interval, marker D3Mit345 with the highest LRS 17.8 that approaches highly significant (Figure 6b), flanked by D3Mit103 (LRS 16.7) and D3Mit57 (LRS 15.5). When the LRS values were converted to obtain equivalent LOD scores, the QTL mapping curve was almost identical to that of LOD scores obtained for the Mendelian trait “affected” (Figure 6c). We also performed χ2 analysis based on the expected recombination fraction of 50% for any 2 loci. Shown in Figure 6d, Mendelian inheritance was preserved and there was no evidence of genotyping error because 59 of the 119 progeny were homozygous B6/B6 and the other 59 of the progeny were heterozygous 129/B6 as expected. However, of the affected animals (mild, moderate, and severe), 61% were homozygous B6/B6, whereas of the normal animals, only 17% were homozygous B6/B6 at marker D3Mit345. This frequency distribution of genotypes versus phenotypes was significantly different than expected (P=0.01). Our genome-wide screen found no significant evidence for linkage of either the trait “affected” or the trait “severity” to any other chromosomal segments.
We demonstrate the strong effects of genetic background on the survival and renal vascular phenotypes in the inbred 129 and C57BL/6 and the F1 Agt1a−/− lines. Thus in an experimental system, background genes can significantly modify a phenotype. While the potential influence of background genes on phenotype complicates the interpretation of any gene-targeting experiment, it does not invalidate the usefulness of the technique. To the contrary, it provides an approach for identifying powerful genetic modifiers that may otherwise be unrecognized.
Our data suggest that there are multiple genetic modifiers that influence the survival phenotype of Agtr1a−/− mice, whereas there is one recessive modifying locus on the 129 background and a separate one on the C57BL/6 background that have strong effects on their vascular phenotype. Two lines of evidence suggest that the survival phenotype is separate and independent of the renal vascular phenotype in Agtr1a−/− mice. First, the proportions of viable weanlings in the F2 and C57BL/6 backcross are significantly different than predicted by our 2-locus or 1-locus model, respectively, whereas the proportions of animals in both crosses with renal vascular lesions are virtually identical to that predicted. Second, most of the F2 and C57BL/6 backcross Agtr1a−/− progeny survive despite having renal vascular pathology similar to that of the inbred Agtr1a−/− lines that have very poor survival. The causes of early mortality in the inbred Agtr1a−/− lines remain difficult to determine. Because only a small fraction of pups die before weaning in the F2 and C57BL/6 backcrosses, it would not be feasible to identify the modifier loci for survival using this methodology. We therefore first focused on identification of the modifier locus for the vascular phenotype on the C57BL/6 background.
The pathogenesis of the renal vascular lesions seen in angiotensin-II-deficient and AT1 receptor-deficient mice is not understood. Because stimulation by angiotensin II causes growth of vascular smooth muscle cells in culture, and because inhibition of ACE prevents smooth muscle cell hyperplasia after vascular injury,19,20 the appearance of renal vascular hypertrophy in hypotensive animals lacking angiotensin II or AT1 receptor was surprising. Several growth factors, including PDGF-A and IGF-1, have been shown to be upregulated at the mRNA and protein levels in renal tissues of combined Agtr1a−/−/Agtr1b−/− mice.21 However, a direct link between these growth factors and the vascular lesions is unclear. Alternatively, it is possible that these lesions arise from compensatory mechanisms to restore blood pressure in these animals. In F2 Agtr1a−/− mice, we found that the affected animals had significantly lower SBP than unaffected animals. Using the genetic approach, our strategy provides an opportunity for insights into the pathogenesis and identification of genetic differences that facilitate the development of these lesions.
Our genomic screen for the modifying locus of renal vascular lesion on the C57BL/6 background in Agtr1a−/− mice yielded a locus significant for linkage to the vascular phenotype on chromosome 3. This locus, D3Mit345, flanked by D3Mit106 and D3Mit57, lies in the 55 cM position on chromosome 3, spanning 5 cM, and has a moderate number of known and novel genes. Several lines of evidence support that this segment of chromosome 3 contains the modifier locus that confers susceptibility to renal vascular hypertrophy in Agtr1a−/− mice. Our analyses using the Mendelian trait affected and the QTL severity yielded near 100% correlation between the LOD scores and LOD score equivalent, respectively, in this region. These two different methods of analyses resulted in the same locus linked to the vascular phenotype. Furthermore, simple χ2 analysis of the frequency distribution of genotypes versus phenotypes at this locus also showed that this was significantly different than expected (P=0.01). Additionally, in a small group of F2 progeny, a separate and independent cross, our screen for the C57BL/6 modifying locus also revealed LOD scores of 1.6 to 2.5 in a region that overlaps with this region in our C57BL/6 backcross. Additional polymorphic markers and recombinant events may be required to narrow this region to allow for identification of the specific gene responsible for susceptibility to renal vascular pathology.
Because simultaneous disruption of the Agtr1b along with the Agtr1a gene was sufficient to produce these renal vascular lesions,10 the Agtr1b gene was a potential candidate gene for renal vascular pathology in Agtr1a−/− mice. Thus, it is noteworthy that the chromosomal segment to which we established linkage excludes the Agtr1b gene that lies further upstream in the 7.6 cM position on chromosome 3. No component of the RAS lies in this region.
Although the AT1A receptor-deficient mouse model, a hypotensive model, may not have immediate applicability to human hypertension and its consequences, the renal vascular lesions in our Agtr1a−/− mouse lines have striking resemblance to those seen in human hypertensive nephrosclerosis. Epidemiological studies in humans have shown that there are marked differences in risks and character of cardiovascular and renal diseases between different races and ethnic groups. Although the reasons for these differences are not understood, it is likely that some of the disparities may have a genetic basis. Moreover, because of its key role in blood pressure regulation and in the development of end-organ injury, variation in RAS genes may account for variation in cardiac or renal diseases between different human populations. Genetic strategies in mouse models of disease, such as the one we are reporting here, may shed light on genetic mechanisms responsible for disease variation in humans.
Significant racial differences in RAS physiology have already been well-documented. Hypertensive black patients more frequently have lower plasma renin activity compared with hypertensive white patients.22,23 In addition, it has been suggested that ACE inhibitors are less efficacious as antihypertensives in black patients with high blood pressure compared with white patients.24 A precise genetic basis for these functional differences has not been identified.
Within the RAS, polymorphisms in the renin, angiotensin, ACE, and AT1 genes have been identified in human populations. Some of these have been associated with hypertensive, cardiovascular or renal diseases in certain populations. Our findings in this study suggest that there are genetic modifiers that lie outside of the RAS that can modulate its functions in health and disease. Using the AT1A receptor-deficient mouse, we identified a monogenic inheritance of a modifier locus from the C57BL/6 genetic background that confers susceptibility to renal vascular hypertrophy.
The study we report represents a different approach to the use of knockout mice in physiological investigations. Because of the key role of the RAS in normal physiological regulation and in the pathogenesis of cardiovascular disease, further understanding of its genetic regulation has obvious relevance to a number of fields. Furthermore, the general paradigm of using mice with gene deletions to uncover strong genetic modifiers should prove useful for identifying and characterizing unknown gene functions in other systems.
We thank Ellen Donnert for dedicated work in tissue preparations, and Jaime Ramirez and Kamie Snow for assistance with genotyping. We also thank Dr Howard Rockman for helpful comments and suggestions. This work was supported by the National Institutes of Health (grants DK02941, HL56122, HL69230, DK56942, DK44757).
- Received September 29, 2003.
- Revision received November 3, 2003.
- Accepted December 3, 2003.
Kim HS, Krege JH, Kluckman KD, Hagaman JR, Hodgin JB, Best CF, Jennette JC, Coffman TM, Maeda N, Smithies O. Genetic control of blood pressure and the angiotensinogen locus. Proc Natl Acad Sci U S A. 1995; 92: 2735–2739.
Tsuchida S, Matsusaka T, Chen X, Okubo S, Niimura F, Nishimura H, Fogo A, Utsunomiya H, Inagami T, Ichikawa I. Murine double nullizygotes of the angiotensin type 1A and 1B receptor genes duplicate severe abnormal phenotypes of angiotensinogen nullizygotes. J Clin Invest. 1998; 101: 755–760.
Yababa M, Umemura S, Kihara M, Tamura K, Nyui N, Ishigami T, Ishii M, Kiuchi Y, Yagami K, Tanimoto K, Sugiyama F, Fukamizu A, Murakami K. Altered development of vasculature in the kidney but not in other main organs in angiotensinogen-deficient mice. Hypertension. 1996; 28: 543A.
Niimura F, Labosky PA, Kakuchi J, Okubo S, Yoshida H, Oikawa T, Ichiki T, Naftilan AJ, Fogo A, Inagami T, Hogan BLM, Ichikawa I. Gene targeting in mice reveals a requirement for angiotensin in the development and maintenance of kidney morphology and growth factor regulation. J Clin Invest. 1995; 96: 2947–2954.
Oliverio M, Kim H, Ito M, Le T, Audoly L, Best C, Hiller S, Kluckman K, Maeda N, Smithies O, Coffman T. Reduced growth, abnormal kidney structure, and AT2-mediated blood pressure regulation in mice lacking both AT1A and AT1B receptors for angiotensin II. Proc Natl Acad Sci U S A. 1998; 95: 15496–15501.
Krege JH, Kim HS, Moyer JS, Jennette JC, Peng L, Hiller SK, Smithies O. Angiotensin-converting enzyme gene mutations, blood pressures, and cardiovascular homeostasis. Hypertension. 1997; 29: 150–157.
Ito M, Oliverio MI, Mannon PJ, Best CF, Maeda N, Smithies O, Coffman TM. Regulation of blood pressure by the type 1A angiotensin II receptor gene. Proc Natl Acad Sci U S A. 1995; 92: 3521–3525.
Krege J, Hodgin J, Hagaman J, Smithies O. A computerized system for measuring blood pressure in mice. Hypertension. 1995; 25: 1111–1115.
Meer JM, Cudmore RH, Jr., Manly KF. Map Manager QTX17. Roswell Park Cancer Institute.
Garcia-Dorado A, Gallego A. On the use of the classical tests for detecting linkage. J Heredity. 1992; 83: 143–146.
Berk BC, Vekshtein V, Gordon HM, Tsuda T. Angiotensin II-stimulated protein synthesis in cultured vascular smooth muscle cells. Hypertension. 1989; 13: 305–314.
Powell JS, Clozel JP, Muller RK, Kuhn H, Hefti F, Hosang M, Baumgartner HR. Inhibitors of angiotensin-converting enzyme prevent myointimal proliferation after vascular injury. Science. 1989; 245: 186–188.
Kilcoyne MM, Thomson MD, Branche G, Williams M, Garnier C, Chiles B, Soland T. Characteristics of hypertension in the black population. Circulation. 1974; 50: 1006–1013.
Materson BJ, Reda DJ, Cushman WC, Massie BM, Freis ED, Kochar MS, Hamburger RJ, Fye C, Lakshman R, Gottdiener J, Ramirez, EA, Henderson WG. Single-drug therapy for hypertension in men. A comparison of six antihypertensive agents with placebo. The Department of Veterans Affairs Cooperative Study Group on Antihypertensive Agents. Published erratum appears in N Engl J Med. 1994;330:1689. N Engl J Med. 1993; 328: 914–921.