Reversibility of Structural and Functional Damage in a Model of Advanced Diabetic Nephropathy

Leptin Replacement, but Not RAAS Inhibition or Treatment with Hydralazine, Rapidly Reverses Diabetes, Obesity, and Manifestations of DN in BTBR ob/ob Mice

In these experiments, treatment began at 18 weeks of age, when DN was well established, and continued for 6 weeks. BTBR ob/ob mice have significantly elevated blood glucose levels and body weight compared with BTBR wild-type (WT) littermates. Leptin replacement results in rapid return to normoglycemia that is sustained and a significant decrease in body weight. Enalapril, losartan, or hydralazine treatment had no effect on body weight or blood glucose level (Table 1 and Supplemental Figure 1).

BTBR ob/ob mice develop progressive albuminuria, detected as early as 8 weeks of age, progressively increasing through 18 weeks of age. After 6 weeks of leptin replacement or enalapril treatment, beginning at week 18, albuminuria was reduced, with reduction most marked in mice receiving leptin replacement. Losartan treatment also resulted in reduced albuminuria, but the values were not statistically significant. Albuminuria did not decrease with hydralazine treatment (Table 1).

Urine albumin-to-creatinine ratio also decreased significantly in leptin-, enalapril-, and losartan-treated mice compared with untreated BTBR ob/ob control mice. Hydralazine treatment did not significantly reduce albumin-to-creatinine ratio (Table 1).

Serum BUN was significantly decreased in the leptin replacement group compared with the 24-week control BTBR ob/ob mice (Table 1). Enalapril, losartan, and hydralazine treatment did not result in significant changes in BUN levels (Table 1). Serum creatinine levels were significantly improved in BTBR ob/ob mice with leptin replacement. Creatinine levels improved in all other treatment modalities compared with 24-week-old control BTBR ob/ob (Table 1), but this did not reach statistical significance.

Insulin levels in BTBR ob/ob mice were significantly elevated at 18 weeks compared with those in BTBR WT mice. After 6 weeks of leptin treatment, insulin levels dropped to nearly normal levels (Supplemental Figure 2).

Serum leptin levels did not significantly differ between BTBR WT mice and leptin-treated BTBR ob/ob mice (mean ± SEM, 5078.67±1125.55 versus 3583.07±966.81 pg/ml), confirming that leptin administration achieved physiologic levels. Untreated control BTBR ob/ob mice had no detectable leptin.

Diabetes-induced renal hypertrophy was evidenced by gradually increasing kidney weight from 18-week-old to 24-week-old BTBR ob/ob mice. All treatments decreased kidney weight, but these changes were statistically significant only in the leptin replacement group (Table 1).

Mesangial matrix expansion was quantitated by measurement of matrix stained for collagen IV and by silver methenamine. Leptin replacement significantly reduced the mesangial matrix accumulation seen in BTBR ob/ob mice to levels equivalent to those of nondiabetic BTBR WT mice. Enalapril, losartan, and hydralazine did not significantly decrease the mesangial matrix expansion, although treatment with enalapril and losartan prevented further increase in mesangial matrix beyond that already present in 18-week-old BTBR ob/ob mice (Table 2 and Figures 1 and 2).

Mesangiolysis results in both expansion of mesangial regions, with loss of normally compact silver-staining mesangial matrix demonstrable by light microscopy, and corresponding areas of lucency demonstrable by electron microscopy (Figure 3). Mesangiolysis disrupts sites where capillary loops are anchored to the mesangium, leading to marked dilatation/ballooning of the untethered capillary loops. Both mesangiolysis and the consequent capillary wall dilatation can be detected in silver methenamine–stained histologic sections of BTBR ob/ob mice (Figure 2A).^10,12 Leptin replacement significantly reduced the extent of mesangiolysis compared with 24-week-old BTBR ob/ob control mice, unlike treatment with enalapril, losartan, and hydralazine (Table 2 and Figure 2B).

Leptin replacement reduced the thickening of glomerular capillary basement membranes in BTBR ob/ob mice compared with untreated BTBR ob/ob mice at 24 weeks (169.15±1.78 versus 181.36± 3.76 nm), but this reduction did not reach statistical significance in the limited number of mice used for this analysis (n=3 each) (Figure 3, B and C).

The number of Mac-2 expressing monocyte/macrophages infiltrating glomeruli was reduced in all treatment groups except the hydralazine group (Table 2).

BTBR ob/ob mice develop mild tubulointerstitial fibrosis quantifiable with picrosirius red staining.¹⁰ Leptin, enalapril, losartan, and hydralazine treatment significantly reduced interstitial fibrosis in these mice compared with 24-week-old control BTBR ob/ob mice (Table 2).

Leptin Replacement, but Not RAAS Inhibition, Results in Restoration of Podocyte Density and Podocyte Number

Progression of DN is characterized by significant loss of podocyte density, quantified as Wilms tumor 1 (WT-1)–expressing cells divided by mean glomerular volume. Using the method of Sanden et al., which involves thick and thin paraffin sections immunostained with WT-1, we previously demonstrated that BTBR ob/ob mice with DN show marked reduction in the total number and density of podocytes per glomerulus compared with BTBR WT controls at 20 weeks of age.¹⁰ In the current study, podocyte density was restored in leptin-treated mice compared with BTBR ob/ob control mice and was similar to that in WT mice of similar age (Figure 4). In contrast, treatment with enalapril, losartan, or hydralazine did not significantly alter podocyte density compared with 24-week-old BTBR ob/ob mice (Figure 4, A and B). Use of this method of podocyte counting showed no significant differences in total podocyte numbers. Glomerular volume was significantly reduced in leptin-treated BTBR ob/ob mice compared with 24-week-old control BTBR ob/ob mice (Figure 4B), contributing to the observed changes in podocyte density. Treatment with enalapril, losartan, and hydralazine did not significantly change glomerular volume. Residual podocytes in 24-week-old BTBR ob/ob mice show partial effacement of foot processes overlying capillary basement membranes, which are largely, but not completely, restored to normal in leptin-treated mice compared with BTBR WT mice (Figure 3C).

In recognition of controversies in the literature over methods for enumerating podocytes, along with recent data demonstrating that WT-1 may not be specific for podocytes because other kidney cells (including parietal epithelial cells [PECs]) may express this marker, immunohistochemical staining was performed on serial histologic sections to mark p57-positive cells. Linear tracing methods have previously shown p57 to be both a sensitive and a specific marker of podocytes in murine glomeruli.¹³ Taniguchi et al. demonstrated the utility of p57 as a podocyte-specific marker by using nephrin-Cre green fluorescent protein (GFP) mice that express GFP under the podocyte-specific nephrin promotor. In both control and diseased mice treated with an antipodocyte antibody, the number of podocytes immunohistochemically stained positive for p57, WT-1, and GFP were similar.¹³ With use of established stereologic methods (fractionator/dissector),^14,15 total podocyte numbers (p57-positive glomerular cells) were counted in BTBR WT, BTBR ob/ob, and BTBR ob/ob with leptin treatment. Total podocyte numbers were reduced in 24-week-old BTBR ob/ob mice compared with BTBR WT mice (50.4±2.8 versus 61.1±2.7 p57-positive podocytes per glomerulus; P=0.03), and leptin treatment restored those numbers to WT mouse levels (60.2±2.4 p57-positive podocytes per glomerulus; P=0.82 versus BTBR WT and P=0.03 versus BTBR ob/ob) (Figure 5). These findings parallel our measurements showing restoration of podocyte density.

As one further method of validation, one BTBR ob/ob and one BTBR WT mouse from the p57-stained groups were used to obtain serial 1-micron-thick plastic embedded sections (“thick” sections as used for electron microscopy), stained with toluidine blue, and podocyte profiles were counted using published stereologic methods (Supplemental Figure 3).^14,15 The numbers obtained from this counting did not significantly differ from the numbers obtained using the p57-immunostained serial sections. However, as illustrated in Supplemental Figure 3, this method has limitations in its accuracy when substantial mesangial expansion is present. Such expansion obscures the anatomic identification of podocytes based solely on their location. For this reason, this approach was not used in additional mice.

Leptin Replacement Increases WT-1 Expression and Proliferation of PECs

WT-1– and Ki67-positive PECs were quantitated in all groups (Figure 6). Leptin treatment of BTBR ob/ob mice resulted in a significant increase of WT-1 expression in PECs lining the Bowman capsule (Figure 6B). Double-label immunohistochemistry showed individual cells expressing both the podocyte marker WT-1 and the PEC marker claudin-1 (Figure 6). There was a trend toward increased numbers of Ki67-positive, proliferating PECs in the leptin-treated mice compared with 18- and 24-week-old BTBR ob/ob mice (Ki67-positive PECs/glomerulus: leptin-treated BTBR ob/ob, 0.118; 18-week-old BTBR ob/ob, 0.04083; 24-week-old BTBR ob/ob, 0.0718; hydralazine-treated BTBR ob/ob, 0.09314; enalapril-treated BTBR ob/ob, 0.09163; losartan-treated BTBR ob/ob, 0.088; 24-week-old BTBR WT, 0.02375; P=0.10), although this trend did not reach statistical significance. Proliferation of podocytes (WT-1–expressing cells) located within the glomerular tuft and not on the Bowman capsule was not present. p57-positive PECs were seen in only the leptin-treated mice but were much less prevalent than WT-1–positive PECs (Figure 5).

Mesangial Expansion and Podocyte Loss Are Associated with Accumulations of Reactive Oxygen Species

Accumulation of superoxide was demonstrated in glomeruli of BTBR ob/ob mice by staining tissue sections with dihydroethidium (DHE) (Figure 7). DHE produces 2-hydroxyethidium when oxidized by O₂⁻, which appears as red fluorescent nuclear staining.¹⁶ After 8 weeks with a control saline pump, BTBR ob/ob mice at 20 weeks and at 28 weeks showed markedly increased DHE staining compared with BTBR WT mice (Figure 7). Treatment with leptin reversed the glomerular accumulation of superoxide (Figure 7). Double staining with DHE and WT-1 demonstrated that many of the cells positive for superoxide are podocytes (Figure 7). DHE staining of the hydralazine, enalapril, and losartan groups showed very elevated accumulation of reactive oxygen species only in the enalapril-treated mice; levels in the hydralazine and losartan treatment groups did not significantly differ from those in the WT mice (Supplemental Figure 4).

Lipid-related oxidative stress was also increased in glomeruli of BTBR ob/ob mice with DN, demonstrated by increased immunostaining for 4-hydroxynonenal (4-HNE) (Supplemental Figure 5). Similar to the results with DHE, glomeruli of BTBR ob/ob mice at 20 weeks and at 28 weeks show increased 4-HNE, but in this case the increase was localized to the mesangium. BTBR WT mice at 20 weeks and leptin-treated BTBR ob/ob mice at 28 weeks showed little or no glomerular accumulation of 4-HNE (Supplemental Figure 5).

BTBR ob/ob Mice Do Not Develop Hypertension

As previously reported, BTBR ob/ob mice are hypotensive relative to BTBR WT and heterozygous BTBR ob^+/−mice, probably as a direct consequence of leptin deficiency.¹⁰ The systolic pressures of BTBR ob/ob mice at 24 weeks were significantly lower than those of BTBR WT mice: 89.8±4.32/66.5±3.31 versus 120±8.08/83.71±5.95 mmHg (P<0.05). Mice treated for 6 weeks with enalapril, losartan, and hydralazine had BP similar to those of untreated BTBR ob/ob mice (enalapril, 84.72±9.38/53.63±9.7 mmHg; losartan, 80.26±5.01/55.2±8.14 mmHg; hydralazine, 81.25±2.75/50±4.18 mm Hg; P=NS).

BTBR ob/ob Mice Express ObRa within Glomeruli

RNA isolated from whole cortex and from isolated glomeruli was used to perform RT-PCR using primers to both the short form of the leptin receptor, ObRa, and the long form, ObRb. Expression of ObRa was detectable in BTBR ob/ob and BTBR WT mice in both cortex and isolated glomeruli, whereas there was no detectable expression of ObRb (Figure 8). We further showed by immunohistochemistry that in leptin-treated BTBR ob/ob mice, phosphorylated-Stat 3 (pStat3) is upregulated in intrinsic renal cells in both glomeruli and tubular epithelium (Figure 8), an indication that there is possible leptin-mediated signaling through the leptin receptor ObRa. Double immunolabeling of tissue sections with the PEC marker claudin-1 or the podocyte marker nephrin and pStat3 showed de novo expression of pStat3 in both of these cell types in leptin-treated mice but not untreated diabetic mice (Supplemental Figure 6).

Animals

The experimental protocol was reviewed and approved by the Animal Care Committee of the University of Washington.

BTBR ob/ob mice are a well studied model of type 2 diabetes,^41–43 and the nephropathy that develops in these mice has been characterized previously.¹⁰ Mice were maintained in specific pathogen-free facility with a temperature-controlled room, regulated with 12-hour light/dark cycle and free access to water and food (standard chow diet: Picolab Rodent Diet 20, Brentwood, MO) containing 0.2% glucose, 0.33% Na, 20% protein, 4.5% fat, and 11.9% calories from fat.

Experimental Design

Six cohorts of mice were studied: (1) BTBR ob/ob mice that received subcutaneous implantation of osmotic mini-pumps (Alzet 2006, Durect Corp., Cupertino, CA) containing leptin at age 18 weeks (n=7); leptin was released from these pumps at a rate of 0.15 µl per hour at a dose of 250 pM per day; (2) BTBR ob/ob mice undergoing inhibition of the RAAS by administration of the angiotensin-converting enzyme inhibitor enalapril (Sigma-Aldrich, St. Louis, MO) at 125 mg/L in drinking water, a dose equal to 30 mg/kg body weight per day, previously established to be effective in mice⁴⁴ (n=9); (3) BTBR ob/ob mice undergoing inhibition of the RAAS by the angiotensin II type 1 receptor blocker losartan (Merck, White House Station, NJ) at 100 mg/L in drinking water, equal to 25 mg/kg body weight per day (n=7); (4) a control group of BTBR ob/ob mice treated with the antihypertensive drug hydralazine (Sigma-Aldrich), 200 mg/L in drinking water, equal to 50 mg/kg body weight per day (n=7); (5) a control group of 24-week-old BTBR ob/ob mice that did not undergo therapeutic interventions (n=6); and (6) a control group of normoglycemic 24-week-old BTBR WT mice (n=5). All treatments were initiated at 18 weeks of age and continued for 6 weeks until 24 weeks of age.

In separate pilot studies, osmotic pumps (Alzet 2004, Durect Corp.) containing leptin or saline were implanted in BTBR ob/ob mice at 20 weeks and continued to 28 weeks of age, with pumps being replaced after 4 weeks (n=3). Kidneys from control 20-week-old BTBR WT and BTBR ob/ob mice and saline- or leptin-treated BTBR ob/ob mice at 28 weeks were used for reactive oxygen species studies.

Blood Chemistry

Blood samples were collected by retro-orbital bleeding at time of euthanasia. Blood glucose levels were monitored weekly using an ACCU-CHEK Ativa Kit (Roche Diagnostics, Indianapolis, IN) beginning at 18 weeks to establish baselines for each mouse. BUN was measured using a liquid urea nitrogen reagent set (Pointe Scientific, Inc., Canton, MI). Serum creatinine levels were measured by an HPLC-based method, as recommended by the Animal Models of Diabetic Complications Consortium¹² at the Yale University Mouse Metabolic Phenotyping Center (MMPC) metabolic testing core. Serum insulin levels were measured using an ELISA-based assay by the University of Washington MMPC metabolic core. Leptin levels were measured in serum using a commercial ELISA kit, according to the manufacturer’s instructions (R&D Systems, Minneapolis, MN).

Urine Measurements

Timed (6-hour) urine collections were obtained beginning at the start of treatment and weekly thereafter. Urine samples were evaluated for proteinuria using the albumin-to-creatinine ratio, as previously described.¹⁰ Albuminuria was measured using mouse albumin ELISA quantitation (Bethyl Laboratories, Montgomery, TX) and creatinine using the Creatinine Companion kit (Exocell) according to the protocols of the manufacturer.

Body weight was measured weekly.

BP Measurement

BP was measured using the CODA 6 noninvasive tail-cuff system (Kent Scientific, Torrington, CT) on conscious mice, as previously described.^10,44 BP was measured at the start of treatment at 18 weeks and before euthanasia at 24 weeks. The accuracy of BP measurements obtained using the CODA 6 have been validated in several studies that used comparisons to BP measurements done by invasive telemetry.^45,46

Tissue Collection and Histologic and Electron Microscopic Studies

Kidney tissue was obtained at euthanasia. Tissues were divided and portions fixed in 10% neutral buffer formalin, in methyl Carnoy solution or 1/2× Karnovsky solution, then processed and embedded in paraffin or Eponate resin using routine protocols.⁴⁷ Formalin-fixed, paraffin-embedded kidney tissue was sectioned and stained with hematoxylin and eosin, periodic acid-Schiff, silver methenamine, and picrosirius red. Karnovsky-fixed tissue was used for electron microscopic study and measurements of basement membrane thickness. Portions of kidney were snap-frozen and stored at −80°C and used for immunofluorescence studies as previously described.⁴⁸ In a subset of animals, magnetic beads (Dynabeads M-450 Tosylactivated beads, Life Technologies, Grand Island NY) were used to isolate glomeruli as published.⁴⁹

Immunohistochemistry

Four-micron sections of formalin or methyl Carnoy-fixed, paraffin-embedded tissue were immunostained as described previously.^10,48,50,51 Antibodies used for immunohistochemistry were (1) goat anti–collagen IV (Southern Biotechnology, Birmingham, AL); (2) rat anti-Mac-2, a marker of monocytes/macrophages (Cedarlane, Hornby, Ontario, Canada);⁵² (3) mouse anti–α-smooth muscle actin (α-SMA), clone 1A4 (Sigma, St. Louis, MO);⁵³ (4) rabbit anti-Ki67, a marker of proliferating cells (clone SP6, Neomarkers, Fremont, CA); (5) rabbit polyclonal antibody to WT-1 protein, a specific marker of podocyte nuclei in glomeruli (Santa Cruz Biotechnology, Santa Cruz, CA); (6) rabbit monoclonal antibody to WT-1 protein (Epitomics, Burlingame, CA); (7) rabbit polyclonal anti-p57 (Santa Cruz Biotechnology); (8) rabbit anti-HNE, a marker of reactive oxygen species production (Alpha Diagnostics, San Antonio, TX); and (9) phospho-STAT3 (Tyr705) XP rabbit mAb (Cell Signaling, Danvers, MA). Double immunohistochemistry was performed by first performing WT-1, nephrin, or claudin-1 immunohistochemistry and development with Vector SG (Vector Laboratories, Burlingame, CA), resulting in a black/gray reaction product. Slides were then blocked with 3% hydrogen peroxide and donkey antirabbit antibody (Jackson Immunoresearch, West Grove, PA), followed by immunohistochemistry for claudin-1 (Abcam, San Francisco, CA) or pSTAT3 and development with ImmPACT DAB substrate (Vector Laboratories) to give a brown reaction product. Secondary reagents used for immunohistochemistry were ImmPRESS antirabbit horseradish peroxidase (HRP) (Vector Laboratories), ImmPRESS antigoat HRP (Vector Laboratories), MM HRP-Polymer kit (Biocare, Concord, CA), and Polink-1 HRP rat NM kit (Golden Bridge International, Mukilteo, WA).

Tissues frozen in optimal cutting temperature were sectioned at 5 μm and stained with dDHE, a second marker of accumulation of reactive oxygen species, in Kreb HEPES buffer (20 mM HEPES, pH 7.4; 128 mM NaCl, 2.5 mM KCl, 2.7 mM CaCl2, 1 mM MgCl2, 16 mM glucose) for 30 minutes at 37°C. Glomerular DHE-positive staining was scored semiquantitatively on a scale of 0 (no staining) to 3 (robust glomerular cell staining). Sections used for double staining were first incubated with rabbit anti–WT-1 (Santa Cruz), followed by AlexaFluor 488–conjugated antirabbit (Invitrogen, Carlsbad, CA) and then incubated in DHE as detailed above.

Negative controls for immunohistochemistry included both substitution of the primary antibody with an isotype-matched irrelevant immunoglobulin or antisera from the same species and substitution with PBS.

Quantitative Analysis of Glomerular and Interstitial Lesions

To quantitate glomerular changes for each animal, 15 (0.1 mm²) sections were systematically digitized using a 40× microscope objective and examined. Glomeruli were consecutively encountered by moving the microscope stage with an S-shape path.

Glomerular mesangial expansion, matrix accumulation/regression, and mesangial cell activation were established using three complementary methods: (1) mesangial area occupied by silver methenamine–stained matrix, (2) mesangial area occupied by matrix immunostained to detect type IV collagen accumulation, and (3) immunostaining to detect mesangial cell expression of α-SMA. The cross-sectional glomerular tuft area (independent of the urinary space) was quantified, as was the total area positive for matrix, collagen IV, or α-SMA by computer image analysis (Image Pro Plus, Media Cybernetics, Bethesda, MD) for each glomerular cross-section as previously described.⁴⁸

Mesangiolysis was determined by assessing the silver methenamine–stained glomeruli and scoring at least 50 glomeruli per animal. Mesangiolysis was scored as present when lucency and dissolution of the normally compact silver-staining mesangial matrix was present, and/or there was dilatation/ballooning of adjacent capillary loops.

Picrosirius red–stained sections were used to evaluate concurrent interstitial fibrosis. Stained tissue sections were photographed under polarized light to achieve maximum brightness, and the percentage positive interstitial staining per area was quantified using Image Pro Plus image.¹⁰

Ki67-positive proliferating cells, WT-1–positive cells, and Mac-2–expressing monocytes/macrophages within glomeruli were quantified by counting positive cells at a magnification of 40× in 50 consecutive glomeruli, as previously described.¹⁰ Additionally, Ki67- and WT-1–positive cells within the parietal cell layers lining the Bowman capsule were separately counted in a minimum of 50 glomerular cross-sections and expressed as average number of cells per parietal layer of the Bowman capsule.

Estimation of Glomerular Volume and Podocyte Density

The thick- and thin-section method of Sanden et al.⁵⁴ was used to evaluate podocyte density using WT-1 antibody as a marker of podocyte nuclei. Briefly, two histologic slides of formalin-fixed kidney (one 3 µm thick and one 9 µm thick) were stained with WT-1 antibody to label glomerular podocytes. Images from 50 consecutive glomeruli from 3-μm and 9-μm sections were photographed with a digital camera under 40× magnification. Images were then imported into Image Pro Plus software for counting WT-1–expressing cells (exclusive of WT-1–expressing parietal cells as described above) and morphometric analysis.

The mean glomerular volume was derived with the Weibel formula⁵⁵ using the glomerular cross-sectional area of the 50 consecutively measured glomerular profiles, as previously described.¹⁰ The value for glomerular volume per WT-1–positive nucleus was calculated and designated as GV/P. Calculation of the number of WT-1–positive nuclei per glomerulus and GV/P was adjusted to reflect the calibrated thickness difference of the microtome used to cut the tissue sections, as outlined by Sanden et al.⁵⁴ This then allowed us to calculate the mean number of WT-1–expressing podocytes per glomerulus tuft, equal to the mean glomerular volume divided by GV/P.

To additionally validate podocyte counting, we collected 100 serial 3-micron paraffin sections, placing 4 sections per slide from BTBR ob/ob (n=4), BTBR WT (n=4), and BTBR ob/ob leptin-treated mice (n=4). Every other slide was stained with anti-p57 antibody to label podocytes, and the whole slide was scanned using a Bioimagene iScan Coreo (Ventana Medical Systems, Tucson, AZ). A minimum of 10 glomeruli per animal were counted, comparing section 1 to section 2, section 2 to section 1, section 3 to section 4, and section 4 to section 3 on each slide. If a positively stained podocyte was seen in one section but not the adjacent section, it was counted. Glomeruli were chosen so that the entire glomerulus was counted beginning when it first was apparent on the slide until it disappeared. Additionally, one BTBR ob/ob and one BTBR WT mouse from the above groups were processed and embedded for electron microscopy, and serial 1-micron-thick sections were cut and stained with toluidine blue for counting of podocyte profiles according to published fractionator/dissector methods.^14,15

RNA Preparation and PCR

Isolation of RNA, reverse transcription, and PCR from total cortical and glomerular RNA were prepared as described previously.^56,57 Primers used for PCR were ObRa-F 5′-ACA CTG TTA ATT TCA CAC CAG AG-3′, ObRa-R 5′-AGT CAT TCA AAC CAT TAG TTT AGG-3′, ObRb-F 5′-GTG TGA GCA TCT CTC CTG GAG-3′, ObRb-R 5′-ACC ACA CCA GAC CCT GAA AG-3′.

Statistical Analyses

All data are expressed as mean ± SEM. Statistical analysis of the data for multiple groups was performed by one-way ANOVA, the Tukey-Kramer multiple comparisons test, and two-tailed t test with Mann-Whitney test, using the GraphPad Prism 5.0 program for Windows (GraphPad, La Jolla, CA). Differences with P<0.05 were considered significant.