Comparative Medicine Vol 56, No 6 Copyright 2006 December 2006 by the American Association for Laboratory Animal Science Pages 482-486

Validation of High-throughput Methods for Measuring and Urinary Albumin Concentrations in Mice

Susan Grindle,1 Cheryl Garganta,2 Susan Sheehan,1 Joe Gile,1 Andree Lapierre,1 Harry Whitmore,1 Beverly Paigen,1 and Keith DiPetrillo1,†,*

Chronic kidney disease is a substantial medical and economic burden. Animal models, including mice, are a crucial component of kidney disease research; however, recent studies disprove the ability of methods to accurately quantify plasma levels, an established marker of kidney disease, in mice. Therefore, we validated autoanalyzer methods for measuring blood urea nitrogen (BUN) and urinary albumin concentrations, 2 common markers of kidney disease, in samples from mice. We used high-performance liquid chromatography to validate BUN concentrations measured using an autoanalyzer, and we utilized mouse albumin standards to determine the accuracy of the autoanalyzer over a wide range of albumin concentrations. We observed a significant, linear correlation between BUN concentrations measured by autoanalyzer and high-performance liquid chroma- tography. We also found a linear relationship between known and measured albumin concentrations, although the autoanalyzer method underestimated the known amount of albumin by 3.5- to 4-fold. We confirmed that plasma and urine constituents do not interfere with the autoanalyzer methods for measuring BUN and urinary albumin concentrations. In addition, we verified BUN and albuminuria as useful markers to detect kidney disease in aged mice and mice with 5/6-nephrectomy. We conclude that autoana- lyzer methods are suitable for high-throughput analysis of BUN and albumin concentrations in mice. The autoanalyzer accurately quantifies BUN concentrations in mouse plasma samples and is useful for measuring urinary albumin concentrations when used with mouse albumin standards.

Abbreviations: B6, C57BL6/6J; BUN, blood urea nitrogen; CKD, chronic kidney disease; D2, DBA2/J; RDRP, Renal Disease Research Plan

Chronic kidney disease (CKD) is an important medical prob- renal function in mice will provide a key tool for forward genetic lem in the United States, affecting roughly 25 million people at strategies to identify genes underlying CKD, an important goal an estimated cost of $22 billion.6 In light of the substantial health set forth in the RDRP. risks associated with CKD, the National Institutes of Diabetes and High-throughput phenotyping of renal function and damage Digestive and Kidney Diseases, in conjunction with the Council is facilitated by clinical chemistry , which can rap- of American Kidney Societies, devised the Renal Disease Research idly quantify the 3 principal clinical markers of kidney function Plan (RDRP) as a roadmap to improve the prevention, diagnosis, or damage: 1) glomerular filtration rate estimated from plasma and treatment of CKD. Several of the recommendations outlined and urine creatinine concentrations;4 2) blood urea nitrogen in the RDRP stress the need for better animal models of kidney (BUN) concentration;1 and 3) the ratio of albumin concentration disease, such as mutant mouse strains and novel mouse mod- to creatinine concentration in the urine.4 Because glomerular fil- els to study the progression of kidney disease. The RDRP also tration rate cannot be estimated reliably or measured easily in recommends new tools for identifying genes underlying kidney small animal models, such as mice, plasma creatinine levels have disease in current animal models. These mouse models represent been used as an alternative marker of kidney function for animal an important component of research into the onset and progres- studies. However, recent evidence confirms that chromagens sion of CKD because the chromosomal regions underlying kid- present in mouse plasma interfere with the Jaffe method com- ney disease in rodents often predict the location of genetic factors monly used in autoanalyzers to measure creatinine concentra- influencing kidney disease in humans.3 Common strategies for tions.2 This problem hinders high-throughput analysis of plasma identifying genes underlying disease in mice include mutagenesis creatinine concentrations in mouse studies. Although BUN is and quantitative trait locus analysis, but these strategies require not a sophisticated marker of renal function, BUN concentration large-scale phenotyping protocols dependent on high-throughput is an easy way to investigate renal function in mice and is ame- methods. Validation of high-throughput methods for measuring nable to high-throughput analysis required of primary pheno- typic screens in genetic studies. Therefore, we sought to validate Received: 10 Mar 2006. Revision requested: 4 Aug 2006. Accepted: 20 Aug 2006. the commonly used autoanalyzer methods for measuring BUN 1The Jackson Laboratory, Bar Harbor, Maine; 2Department of Genetics, Yale University and urinary albumin concentrations in mice as high-throughput School of Medicine, Biochemical Disease Detection Laboratory, New Haven, Connecticut. methods for assessing kidney function and damage, respectively, †Present address: Novartis Institutes for BioMedical Research, East Hanover, New Jersey. to facilitate forward genetic strategies to identify genes underly- *Corresponding author. Email: [email protected] ing CKD. 482 Validation of BUN and urinary albumin measurements in mice

Materials and Methods Column eluent was derivatized with ninhydrin and the absor- Animals. We collected urine and blood samples from 7- to 10- bance monitored at 570 nm. Urea eluted at 4.9 min, and the inter- wk-old, male DBA2/J (D2), A/J, and C57BL/6J (B6) mice and nal standard eluted at 6 min. Between samples, the column was from separate groups of 10- to 14-mo-old and 20- to 24-mo-old washed at 70 pC for 2 min with 1% lithium hydroxide, pH 13. The D2 mice. All mice were obtained from The Jackson Laboratory. concentration of urea was determined by comparison of the ratio Blood (approximately 150 Nl) was collected by retro-orbital bleed- of peak areas (urea:internal standard) with a standard curve of 5 ing through heparin-coated capillary tubes into microcentrifuge to 80 mg urea/dl (3 to 27 mg BUN/dl) in 70% (v/v) horse tubes containing 5 Nl of 200 NM EDTA, and plasma was isolated (previously dialyzed against phosphate buffered saline). The coef- by centrifuging blood samples at 20,200 × g for 10 min. Spot urine ficient of variation of this method in the Yale Biochemical Disease samples were collected from individual mice into microcentrifuge Detection Laboratory over the last 5 y has been less than 5%. tubes over the course of several mornings. Measuring urinary albumin concentrations. To validate the auto- Additional groups of A/J and D2 mice underwent 5/6-nephrec- analyzer method for quantifying mouse albumin concentrations, tomy as a model of kidney disease. Mice were anesthetized with we purchased mouse albumin standards from Kamiya Biomedi- 250 mg/kg tribromoethanol injected intraperitoneally. Fur on the cal Company (Seattle, WA) and Sigma (St Louis, MO). Series of left and right sides was shaved from the last rib to the iliac crest, mouse albumin standards were measured on the autoanalyzer. and skin was disinfected with povidone iodine and 70% ethanol. Albumin was detected with a goat anti-human albumin antibody To remove the left kidney, a 5- to 8-mm incision through the skin to form antigen–antibody complexes, which increase the turbidity was made parallel to the last rib midway between the last rib and of the sample. The autoanalyzer measured the change in absor- iliac crest. A 4- to 7-mm incision then was made in the abdominal bance at 380 nm, which is proportional to the concentration of wall, and the kidney was elevated through the incision to expose albumin in the sample. the renal artery and vein. A single ligature was placed around the Measuring urinary creatinine concentrations. Urinary creatinine renal artery, renal vein, and ureter, and the kidney was excised. concentrations were measured on an autoanalyzer (Synchron CX5 The sutured vasculature was returned to the abdominal cavity, Clinical Chemistry Analyzer, Beckman Coulter, Brea, CA; reagent the abdominal incision was sutured with absorbable suture, and kit 442760) using the Jaffe method, where creatinine combines the skin incision was closed with wound clips. The mouse was re- with picric acid to form a creatinine-picrate complex that changes positioned, and the right kidney was elevated from the abdomen the absorbance of the solution. Dunn and colleagues have recent- through the same procedure as for the left kidney. Epinephrine ly demonstrated that this method accurately measures urinary 2 was applied topically to the right kidney, and 2/3 of the kidney creatinine concentrations in mice. was removed from the poles and dorsal portions. The kidney was Statistics. Prism software (GraphPad Software, San Diego, CA) returned to the abdomen and the abdominal and skin incisions was used for all statistical analyses, including correlation of BUN were closed as described for the left kidney. Mice were checked concentrations between HPLC and autoanalyzer methods, as well daily to ensure proper healing and normal behavior and were as linear regression of albumin standards. Mean values were ana- administered postoperative analgesia according to animal wel- lyzed by Student t tests or analysis of variance with Student New- fare guidelines of The Jackson Laboratory. Urine samples were man-Keuls post-test as indicated in the figure legends. collected on days 3 through 7 after surgery, and plasma samples were collected 3 and 10 d after surgery. The Animal Care and Use Results Committee at The Jackson Laboratory approved all procedures. We compared BUN concentrations measured by both auto- Measuring plasma BUN concentrations. Plasma samples from analyzer and HPLC in replicate plasma samples taken from 7- each mouse were separated into 2 aliquots, and BUN concentra- to 10-wk-old mice (n  41 mice from 3 strains) and from 7- to tions were measured on an autoanalyzer (Synchron CX5 Clinical 10-wk-old A/J and D2 mice that underwent 5/6-nephrectomy Chemistry Autoanalyzer, Beckman Coulter, Brea, CA) and by (n  33 plasma samples taken from 17 mice 3 and 10 d after 5/6- high-performance liquid chromatography (HPLC). For the auto- nephrectomy). We observed a significant, linear correlation (r2 analyzer, BUN concentration (reagent kit 442750, Beckman Coul-  0.9355, P  0.0001) between the methods (Figure 1 A). This ter) was measured by an enzymatic rate method, whereby urea correlation was linear between 15 and 96 mg/dl, although the first was hydrolyzed to and carbon dioxide. Glutamate manufacturer’s instructions report a linear range extending as dehydrogenase then converted ammonia and B-ketoglutarate to low as 0.2 mg/dl. glutamate, with the concomitant oxidation of reduced C-nico- To verify that BUN is an effective marker of diminished kidney tinamide adenine dinucleotide to C-nicotinamide dinucleotide. function, we performed 5/6-nephrectomies on D2 mice and mea- The autoanalyzer measured the change in absorbance at 340 nm, sured plasma BUN concentrations at 3 and 10 d after surgery. As which is directly proportional to the concentration of urea nitro- expected, 5/6-nephrectomized mice exhibited significantly (P  gen in the sample. 0.05) elevated BUN concentrations, reflecting reduced kidney For HPLC, urea concentration was determined using an amino function, at both 3 and 10 d after surgery (Figure 1 B). We also acid analyzer (model 6300, Beckman, Palo Alto, CA) and post- observed a significant linear correlation r( 2  0.9532, P  0.0001) 5 column derivatization with ninhydrin. Plasma samples (40 Nl) between BUN concentrations measured on the autoanalyzer and were precipitated with 0.1 volume 35% (w/v) sulfosalicylic acid by HPLC in the subset of plasma samples from nephrectomized and centrifuged for 4 min at 14,400 t g. The supernatant (30 Nl) mice. These findings show that plasma constituents from mice was combined with internal standard (0.2 mM glucosaminic acid with reduced renal function do not interfere with the autoana- in Li-S buffer, pH 2.2; cat #338084, Beckman, Palo Alto, CA), and lyzer method of measuring BUN concentration. 50 Nl was injected onto a Beckman 10-cm lithium column. Urea In addition to validating the autoanalyzer method for measur- was separated from other early-eluting compounds using 20 ml/h ing BUN, we also tested the autoanalyzer method for quantifying Li-A buffer (1% lithium citrate, 0.05% lithium chloride, pH 2.8). urinary albumin concentration. To do this, we used commercially 483 Vol 56, No 6 Comparative Medicine December 2006

A the autoanalyzer method was not accurate based on calibration with human albumin. However, accurate quantification of mouse albumin is possible by calculating the albumin concentrations of experimental samples by linear regression of a standard curve generated with mouse albumin standards. To determine whether biologic constituents of urine interfere with the autoanalyzer method for detecting albumin, we diluted mouse albumin standards purchased from Kamiya in urine (1:1 [v/v] dilution of standard into pooled urine) collected from either B6 mice (no detectable albumin in the pooled urine sample) or 5/6-nephrectomized D2 mice (measured albumin concentration in the pooled urine sample, 0.73 mg/dl). Dilution of the standards in B6 urine did not alter the linear relationship between known and measured albumin concentrations (r2  0.9911, n  3; Figure 2 C). Similarly, dilution of the standards in urine from 5/6-nephrec- tomized D2 mice did not interfere with the autoanalyzer method for measuring urinary albumin concentrations; we observed a linear correlation between known and measured albumin con- centrations (r2  0.9943, n  3; Figure 2 D) offset by the basal level of albumin in the pooled urine sample. We also observed a sig- nificant P( 0.0001) linear correlation, offset by the basal albumin B concentration in the pooled urine sample, between known and measured albumin concentrations using Sigma mouse albumin standards diluted 1:10 into urine from approximately 1-y-old D2 mice (r2  0.9970; n 3; data not shown). These findings show that the autoanalyzer method is not adversely affected by the presence of mouse urine constituents, even in a mouse model of decreased renal function with albuminuria. We also validated the utility of these methods for detecting CKD in mice. D2 mice exhibit microalbuminuria (urinary albu- min to creatinine ratio greater than 30 mg/g) at 10 wk of age, so we tested for progressive kidney disease by measuring plasma BUN concentrations and urinary albumin to creatinine ratios in 10- to 14- and 20- to 24-mo-old D2 mice. We found a significant (P  0.05) increase in plasma BUN concentrations (Figure 3 A) and urinary albumin to creatinine ratios (Figure 3 B) in older mice compared to 10-wk-old mice, demonstrating that BUN and urinary albumin to creatinine ratios are suitable markers for de- tecting CKD in mice.

Discussion The study of the genetic basis of CKD in mice is limited by the analytic methods available for evaluating kidney function and Figure 1. Validation of autoanalyzer method for measuring plasma BUN damage. Elevated plasma creatinine levels have routinely been concentrations. (A) BUN concentrations were measured by HPLC and autoanalyzer in duplicate plasma samples derived from 7- to 10-wk-old used as a marker of reduced kidney function in animal studies. male DBA2/J, A/J, and C57BL/6J mice (n  41 mice total) and from However, high-throughput analysis of plasma creatinine con- 5/6-nephrectomized male DBA2/J and A/J mice (n  33 plasma samples centrations by an autoanalyzer is not feasible because plasma taken from 17 mice 3 and 10 d after surgery). Symbols represent BUN chromagens interfere with the assay.2 Although HPLC analysis concentrations in individual plasma samples. (B) BUN concentrations accurately quantifies plasma creatinine concentrations, it is more measured in 7- to 10-wk-old male DBA/2J mice with and without 5/6- costly and less amenable to high-throughput analysis than is us- nephrectomy (Nephrx). Bars represent the mean q standard error of 14 ing an autoanalyzer. Here we present evidence that chromagens (DBA/2J) or 12 (DBA/2J Nephrx) mice. *, P  0.001 versus DBA/2J, as in mouse plasma do not interfere with the autoanalyzer method determined by analysis of variance. for quantifying BUN concentrations. Because BUN concentration increases as kidney function declines, plasma BUN is a good al- available mouse albumin standards. We observed a linear rela- ternative to creatinine as a high-throughput screen for evaluating tionship between known and measured albumin concentrations kidney function in mice. using mouse albumin purchased from either Kamiya (r2  0.9973, In addition to BUN as a marker for kidney function, the ratio n  12; Figure 2 A) or Sigma (r2  0.9959, n  3; Figure 2 B). This of urinary albumin concentration to creatinine concentration is relationship was linear between 0.1 and 51.6 mg/dl at least. The commonly used as an indicator of kidney damage in animal stud- autoanalyzer method underestimated the known amount of al- ies. Our results show that the autoanalyzer method, designed to bumin by 3.5- to 4-fold, so although it exhibited good precision, detect human albumin with an antibody, also detects mouse albu- 484 Validation of BUN and urinary albumin measurements in mice

A B

CD

Figure 2. Validation of autoanalyzer method for measuring urinary albumin concentrations. Mouse albumin standards purchased from Kamiya (A) and Sigma (B) were measured by autoanalyzer and compared to the known concentrations of the standards. Mouse albumin standards purchased from Kamiya were diluted 1:1 in urine from C57BL/6J mice (no detectable albumin); (C) or urine from 5/6-nephrectomized DBA/2J mice (measured albumin concentration in pooled urine sample, 0.73 mg/dl); (D). Symbols represent the mean q standard error of 12 (A) or 3 (B–D) replicates of each concentration. min. Although the autoanalyzer method was linear across a range nine ratios can be performed on autoanalyzers. of albumin concentrations, accurate quantification of albumin Our findings support the use of plasma BUN concentrations concentrations in mouse samples will require linear regression and urinary albumin to creatinine ratios for high-throughput using mouse albumin standards, which are commercially avail- evaluation of kidney function and damage in mice. Although able. In addition, Dunn and colleagues have demonstrated good measurement of plasma creatinine concentrations by HPLC will linear correlation and accuracy between HPLC and autoanalyzer be useful for detailed studies evaluating glomerular filtration rate methods for measuring urinary creatinine concentrations,2 mean- in small sample sizes, plasma BUN and urinary albumin concen- ing that high-throughput analysis of urinary albumin to creati- trations provide easy, reliable markers to assess CKD in large- scale studies using mice.

485 Vol 56, No 6 Comparative Medicine December 2006

AB

Figure 3. Progression of kidney disease in DBA/2J mice. (A) Comparison of plasma BUN concentrations in 10-wk-old versus 1- and 2-y-old, male DBA/2J mice. Bars represent mean q standard error of 14 (10-wk-old), 22 (1-y-old), or 5 (2-y-old) mice. (B) Comparison of urinary albumin to creatinine ratios in 10-wk-old versus 1- and 2-y-old, male DBA/2J mice. Albumin concentrations were calculated by linear regression using mouse albumin standards. Bars represent mean q standard error of 21 (10-wk-old), 20 (1-y-old), or 4 (2-y-old) mice. †, P  0.05 versus 1- and 2-y-old DBA/2J mice, as determined by analysis of variance. *, P  0.05 versus 2-y-old DBA/2J mice, as determined by analysis of variance.

3. Korstanje R, DiPetrillo K. 2004. Unraveling the genetics of chronic Acknowledgments kidney disease using animal models, Am J Physiol Renal Physiol We thank Simon WM John (The Jackson Laboratory, Bar Harbor, ME) 287(3):F347–352. for providing 20- to 24-mo-old mice for this study. 4. National Kidney Foundation. 2002. K/DOQI clinical practice guidelines for chronic kidney disease: evaluation, classification, and stratification. Am J Kidney Dis39(Suppl 1):S1–246. References 5. Shapira EBM, Miller JB, Africk DK. 1989. Biochemical genetics: 1. Duarte CG, Preuss HG. 1993, Assessment of renal function—glo- a laboratory manual. Oxford (UK): Oxford University Press. p merular and tubular. Clin Lab Med 13(1):33–52. 89–93. 2. Dunn SR, Qi Z, Bottinger EP, Breyer MD, Sharma K. 2004. Utility 6. US Renal Data System (USRDS). 2003. USRDS 2003 annual data of endogenous creatinine clearance as a measure of renal function report. In: Atlas of end-stage renal disease in the United States. in mice. Kidney Int 65(5):1959–1967. Bethesda (MD): National Institute of Diabetes and Digestive and Kidney Diseases.

486