Loss of the Endothelial Glycocalyx Links Albuminuria and Vascular Dysfunction

Abstract

Patients with albuminuria and CKD frequently have vascular dysfunction but the underlying mechanisms remain unclear. Because the endothelial surface layer, a meshwork of surface-bound and loosely adherent glycosaminoglycans and proteoglycans, modulates vascular function, its loss could contribute to both renal and systemic vascular dysfunction in proteinuric CKD. Using Munich-Wistar-Fromter (MWF) rats as a model of spontaneous albuminuric CKD, multiphoton fluorescence imaging and single-vessel physiology measurements revealed that old MWF rats exhibited widespread loss of the endothelial surface layer in parallel with defects in microvascular permeability to both water and albumin, in both continuous mesenteric microvessels and fenestrated glomerular microvessels. In contrast to young MWF rats, enzymatic disruption of the endothelial surface layer in old MWF rats resulted in neither additional loss of the layer nor additional changes in permeability. Intravenous injection of wheat germ agglutinin lectin and its adsorption onto the endothelial surface layer significantly improved glomerular albumin permeability. Taken together, these results suggest that widespread loss of the endothelial surface layer links albuminuric kidney disease with systemic vascular dysfunction, providing a potential therapeutic target for proteinuric kidney disease.

Albuminuria and CKD are potent indicators of long-term vascular health and survival. Progressive albuminuria and deteriorating renal function both independently predict increased risk of cardiovascular disease^1,2 and are associated with widespread vascular dysfunction, such as impaired macrovascular flow-dependent dilation,^3–5 and increased microvascular fluid⁶ and albumin⁷ flux. Both flow-dependent dilation and microvascular permeability depend upon intact endothelial cell function, and generalized endothelial dysfunction has been proposed to underlie the widespread vascular complications apparent in those with CKD and albuminuria.⁸ Generalized endothelial dysfunction is also considered important in the development of atherosclerosis,^9,10 which contributes to increased cardiovascular disease risk in these individuals.¹¹ Mechanisms underlying the relation between renal disease and systemic endothelial cell dysfunction remain incompletely understood,^12,13 but a defect in the endothelial surface layer (ESL), which is common to all blood vessels, has been suggested as a mechanistic link between widespread vascular dysfunction and albuminuric kidney disease.^12–14

Glycosaminoglycans, proteoglycans, and adsorbed plasma components form a surface-bound (“endothelial glycocalyx”) and loosely adherent matrix covering the luminal surface of all vascular endothelial cells, termed the ESL.¹⁵ As the primary interface between blood and vessel walls, the ESL regulates vital endothelial cell functions under physiologic and pathophysiologic circumstances. In large conduit vessels, ESL removal limits nitric oxide bioavailability,^16,17 resulting in blunted shear-stress–dependent vasodilation¹⁶ that resembles the impaired flow-dependent dilation observed in albuminuric individuals.⁵ In addition, inhibition of hyaluronan synthesis diminishes ESL depth and accelerates atherosclerosis.¹⁸ ESL loss may therefore result in, as well as result from,¹⁹ large vessel dysfunction and disease.

The ESL also regulates microvascular function, including mechanotransduction, leukocyte adhesion, and the permeability of exchange vessels (capillaries and postcapillary venules).^20,21 The ESL is a cross-linked matrix that provides resistance to fluid movement, thereby contributing to the water permeability (hydraulic conductivity: L_P) of the vessel wall.²⁰ In addition, the regular arrangement²² and negative charge²³ of endothelial glycocalyx fibers permits sieving of macromolecules (e.g., albumin), such that the ESL also contributes to the proportion of albumin molecules “reflected” by the vessel wall back into plasma (reflection coefficient).²⁴ Because the ESL covers the major pathways for fluid and solute movement in vessels with continuous endothelium (e.g., luminal opening of intercellular clefts in mesenteric microvessels) and fenestrated endothelium (e.g., fenestrae of glomerular capillaries), the ESL is a shared regulator of permeability coefficients in these vessel types that are otherwise very different.²⁵ Selective removal of ESL components with a variety of enzymes (e.g., neuraminidase²⁶) increases flux across glomerular endothelial cell monolayers,²⁶ intact glomerular capillary walls,^27,28 coronary arterioles,²⁹ and mesenteric microvessels.^30,31 The ESL therefore regulates glomerular and extrarenal microvessel permeability, and ESL changes result in quantifiable changes in microvascular permeability.

Acknowledging the difficulties of measuring ESL in humans,³² widespread ESL loss has been reported in patients with type 1³³ and type 2³⁴ diabetes, and the development of microalbuminuria in patients with diabetes results in additional reductions in whole body ESL volume.³³ ESL loss from mouse glomerular capillaries occurs shortly after adriamycin administration, in association with albuminuria and increased glomerular albumin clearance.³⁵ However, both hyperglycemia³⁶ and some nephrotoxic agents³⁷ directly suppress ESL constituent synthesis by endothelial cells. The possibility that the ESL is lost in spontaneous, chronic, albuminuric kidney disease, and the potential importance of this loss in mediating the widespread vascular dysfunction that frequently occurs in albuminuria and kidney disease, have not been assessed.

Given that vascular dysfunction after both ESL removal and development of kidney disease are strikingly similar, and given the prognostic importance of widespread vascular dysfunction in albuminuria and renal impairment, we sought to test the hypothesis that spontaneous development of proteinuric kidney disease is associated with ESL loss and deranged vascular function (microvascular permeability) in both renal and extrarenal microvessels. We therefore examined Munich-Wistar-Fromter (MWF) rats, an inbred strain that develop proteinuria, hypertension, and renal failure as they age.

Results

Older MWF rats, compared with both healthy age-matched Wistar rats and healthy young MWF rats, displayed typical features of proteinuric kidney disease,³⁸ including increased urine protein/creatinine and albumin/creatinine ratios, hypertension, elevated plasma creatinine (Table 1), FSGS, glomerulomegaly, and flattened tubular epithelium (Supplemental Figure 1).

Table 1: Animal phenotype

Increased Microvascular Permeability in Older MWF Rats

Mesenteric microvessel hydraulic conductivity (L_P) (Figure 1A) and glomerular K_f (L_PA) (Figure 1B) were both approximately 2.3-fold higher in older (proteinuric) MWF rats (both P<0.05; unpaired t test). Both glomerular volume–corrected L_PA (L_PA/V_i [min⁻¹ · mmHg⁻¹] 1.5±0.1 versus 1.0±0.1, P<0.01, unpaired t test) and surface area–corrected L_PA (L_PA/A [cm · min⁻¹ · mmHg⁻¹] 4.8±0.4 versus 2.7±0.3; P<0.001; unpaired t test) remained higher in older MWF rats, suggesting elevated L_P of glomerular and mesenteric capillary walls in older MWF rats. Higher mesenteric L_P and glomerular L_PA/V_i values were significantly correlated (Pearson r=0.58; P<0.01) (Figure 1C).

Figure 1: Widespread increase in microvascular permeability in old MWF rats with proteinuric kidney disease. (A) Mean (±SEM) hydraulic conductivity (L _P) of mesenteric microvessels was significantly elevated in MWF rats (black bar; n=7 animals and vessels), compared with healthy Wistar rats (open bar; n=15 animals and vessels). (B) Mean (±SEM) glomerular K _f, normalized for glomerular volume (L _P A/V _i) was significantly elevated in glomeruli from MWF rats (black bar; n=58 glomeruli from 10 animals) compared with glomeruli isolated from healthy Wistar rats (open bar; n=24 glomeruli from 6 animals). (C) Glomerular L _P A/V _i and mesenteric L _P values from the same animal were positively correlated (Pearson r=0.58; P<0.05) (open circles, Wistar rats; filled circles, MWF rats). (D) There was a significant, negative relation between mesenteric microvessel reflection coefficient (mesenteric σ) and urinary protein/creatinine ratio in MWF rats (Spearman r=−0.71; P<0.05). (E) Glomerular albumin sieving coefficient (Θ_albumin) was low in healthy young MWF rats (n=11 glomeruli in 3 animals), and significantly elevated in old MWF rats with proteinuric kidney disease (n=20 glomeruli in 5 animals). *P<0.05 (unpaired t test). **P<0.01 (unpaired t test).

Mesenteric microvessel reflection coefficients (σ) in older MWF rats (0.95±0.04; n=7) and healthy, age-matched Wistar rats (0.93±0.02; n=7) were not significantly different, but older MWF rats with greater proteinuria had lower σ (Spearman r=–0.71; P<0.05) (Figure 1D). As an index of glomerular macromolecular permeability, the glomerular albumin sieving coefficient (Θ_albumin), measured with multiphoton microscopy,^39–41 averaged 0.00065±0.00003 in healthy young MWF rats, and was significantly elevated in older MWF rats (0.0015±0.00016; P<0.01; unpaired t test) (Figure 1E). Animals with more albuminuria had higher glomerular Θ_albumin (Pearson r=0.83; P<0.05).

Altered ESL in Nonrenal Microvessels in Older MWF Rats

An intense, linear FITC-WGA lectin signal was present along the luminal border of mesenteric microvessel walls in healthy young MWF rats (Figure 2A). Neuraminidase, which enzymatically removes the ESL,²⁶ removed this WGA signal (Figure 2B). This linear, marginal FITC-WGA lectin signal in mesenteric microvessels was substantially reduced in older MWF rats (Figure 2C). This baseline pattern in older MWF rats (Figure 2C) resembled the pattern after enzymatic ESL removal in healthy rats (Figure 2B). Furthermore, neuraminidase did not alter the pattern of FITC-WGA lectin labeling of mesenteric microvessels in older MWF rats (Figure 2D). These findings indicate loss of ESL from nonrenal microvessels in these animals with proteinuric kidney disease.

Figure 2: ESL of mesenteric microvessels is compromised in old MWF rats. Confocal microscopy images of FITC-WGA lectin (green) labeling ESL (esl), alongside simultaneous DIC imaging of microvessel anatomy in a healthy young MWF rat (A and B) and an old MWF rat (C and D). Linear labeling of ESL at the vessel margin in a young MWF rat (Ai and Aii) is removed by neuraminidase (Bi and Bii). ESL is not apparent in mesenteric microvessels of old MWF rats before (Ci and Cii) or after (Di and Dii) neuraminidase. Under baseline conditions (open bars), ESL covers most of wall of mesenteric microvessels (E), and is approximately 0.6 µm deep (F), in healthy rats (old Wistar rats, n=111 vessels in 3 rats; young MWF rats, n=47 vessels in 3 rats). ESL coverage and depth are both reduced by neuraminidase (black bar) in these healthy rats. In contrast, ESL is significantly sparser (E) and shallower (F) in old MWF rats under baseline conditions (n=165 vessels in 4 rats), and is not reduced any further by neuraminidase. *P<0.05 versus both healthy rat groups (one-way ANOVA, Bonferroni); ^# P<0.05 versus baseline values (one-way ANOVA, Bonferroni); ns, not significant versus preneuraminidase values (one-way ANOVA, Bonferroni). esl, ESL; vl, vessel lumen; i, interstitium; vw, vessel wall. Scale bar, 10 microns.

Mesenteric microvessel ESL coverage (Wistar rats: 90%±1%; young MWF rats: 83%±11%) (Figure 2E) and depth (Wistar rats: 0.65±0.03 µm; young MWF rats: 0.57±0.07 µm) (Figure 2F) in healthy animals were significantly greater than in older MWF rats (detectable coverage: 25%±9%; depth: 0.20±0.05 µm; both P<0.001). In healthy animals, neuraminidase significantly reduced mesenteric microvessel ESL depth and coverage (all P<0.05; ANOVA) (Figure 2, E and F), but did not alter mesenteric microvessel ESL depth and coverage in older MWF rats. Reduced ESL depth and coverage is compatible with increased permeability of mesenteric microvessels in older MWF rats.

Altered Glomerular ESL in Older MWF Rats

Similar patterns of ESL labeling were also observed in glomerular capillaries of healthy young MWF rats (Figure 3), with linear FITC-WGA labeling of the margin of glomerular capillaries under baseline conditions (Figure 3A). WGA labeled the ESL, instead of (or in combination with) labeling of other GAG- and sialoprotein-rich components of the glomerular capillary wall, as demonstrated by alexa-594-WGA labeling at the luminal border of lucifer yellow-loaded glomerular endothelial cells (Figure 3B). Neuraminidase removed the linear FITC-WGA label at the margin of glomerular capillaries (Figure 3C) covering the luminal surface of endothelial cells (Figure 3D). Neuraminidase reduced glomerular ESL coverage (83%±2% to 31%±5%; P<0.005; paired t test) (Figure 3E). As with mesenteric microvessels, glomerular capillaries in older MWF rats did not exhibit linear FITC-WGA labeling of the ESL (Figure 3F), exhibited less coverage with glomerular ESL (P<0.05 versus young MWF; one-way ANOVA, Bonferroni), and no significant changes were observed after neuraminidase treatment (Figure 3G) (baseline coverage 39%±1%, postneuraminidase 32%±2%; P>0.05; paired test) (Figure 3H). Animals with proteinuric kidney disease therefore exhibit ESL loss from both renal and nonrenal microvessels.

Figure 3: ESL of glomerular capillaries is compromised in older MWF rats. (A) Under baseline conditions, multiphoton microscopy imaging reveals a linear label (white arrows) of FITC-WGA lectin (green) at the margin of superficial glomerular capillaries containing fluorescently labeled albumin (red). (B) Again under baseline conditions, linear labeling with the same WGA lectin (but tagged with the red fluorophore alexa-594) is on the luminal surface of lucifer yellow-labeled glomerular endothelial cells (green), confirming that the WGA lectin labels glomerular ESL. (Bii) High-magnification image of the rectangular region outlined in Bi. (C) After neuraminidase treatment, the majority of glomerular capillary wall sites do not exhibit linear glomerular ESL label (open arrowheads), although some linear label is still apparent (white arrows). (D) Imaging of the same optical section of the same glomerulus shown in C after neuraminidase treatment. Most of the ESL label has been removed as observed at low magnification (open arrowheads in Di) and high magnification (broken lines in Di). (E) Under baseline conditions (open circles), the majority of glomerular capillary wall sites in young MWF rats exhibit ESL labeling. Neuraminidase treatment of these healthy animals (black circles) significantly reduces glomerular ESL labeling. (F) In old MWF rats, multiphoton microscopy imaging of FITC-WGA lectin (green) reveals very little linear labeling (white arrow) at the margin of superficial glomerular capillaries containing fluorescently labeled albumin (red). The majority of glomerular capillary wall sites do not display linear labeling (open arrowhead). (G) After neuraminidase treatment of an old MWF rat, there is little change in the pattern of WGA lectin labeling (labeling of the brush border of tubular epithelium confirms the presence of detectable FITC-WGA lectin). (H) ESL coverage in old MWF rats under baseline conditions (open circles) is less than in young rats (compared with 3E), and is not further reduced by neuraminidase (closed circles. Each circle represents mean ± SEM values from 4 to 10 glomeruli in individual animals. ^$ P<0.05 versus baseline (one-way ANOVA, Bonferroni); ^# P<0.05 (one-way ANOVA, Bonferroni); ns, P>0.05 versus baseline (one-way ANOVA, Bonferroni). gc, glomerular capillaries; ec, endothelial cell; esl, ESL; *, extracapsular capillary; pus, peripheral urinary space; P, podocyte (negatively labeled by lucifer yellow); t, tubular epithelium. Scale bar, 10 µm in A, C, F, and G; 5 µm in Bi, Bii, Di, and Dii.

ESL Loss Contributes to Altered Permeability in Older MWF Rats

To assess the functional relevance of ESL loss in older MWF rats, we examined changes in glomerular Θ_albumin after neuraminidase-induced ESL removal, because these measurements can uniquely be coupled with real-time ESL imaging, before and after neuraminidase, in multiple vessels in the same animal. Low values of Θ_albumin in healthy young MWF rats were significantly elevated by application of neuraminidase (0.00040±0.00010 to 0.00096±0.00020; P<0.05; paired t test) (Figure 4A), mirroring neuraminidase-induced glomerular ESL removal in these animals (Figure 3E). Θ_albumin in healthy young MWF rats after neuraminidase was similar to Θ_albumin in older MWF rats before neuraminidase (Figure 4A).

Figure 4: Loss of glomerular ESL contributes to increased glomerular albumin sieving coefficient, and modification of ESL improves glomerular albumin sieving coefficient, in old MWF rats. (A) In healthy young MWF rats, glomerular albumin sieving coefficient values (Θ_albumin) of individual glomeruli (open circles; open square represents mean ± SEM) were elevated after exposure to neuraminidase in every glomerulus (black circles; black square represents mean ± SEM). ^# P<0.05 versus baseline (paired t test). In contrast, higher Θ_albumin values in old MWF rats (open circles; open square represents mean ± SEM) were not significantly elevated any further after neuraminidase exposure (black circles; black square represents mean ± SEM) ns, P>0.05 versus baseline (paired t test). (B) Binding of WGA lectin to the ESL (hashed bars) significantly reduced Θ_albumin in old MWF rats (black bar). A similar, nonsignificant trend toward reduced Θ_albumin after lectin binding was also observed in young MWF rats (open bar). *P<0.05; ns, P>0.05 versus baseline (one-way ANOVA, Bonferroni). (C) Θ_albumin for individual glomeruli (gray circles, young MWF; black circles, old MWF) were significantly correlated with the percentage of glomerular capillary wall sites with ESL cover in the same glomerulus (Pearson r=−0.59; P<0.005).

In contrast, Θ_albumin in older MWF rats (0.00082±0.00016) was not significantly different after neuraminidase treatment (0.00097±0.00012; P>0.05; paired t test) (Figure 4A). Coupled with observations that neuraminidase does not exacerbate the loss of ESL already apparent in older MWF rats (Figure 3H), this indicates that loss of ESL contributes to altered Θ_albumin in these animals with proteinuric kidney disease.

Moreover, modification of the ESL, by adsorption of acutely intravenously injected WGA lectin, reduced glomerular albumin permeability in older MWF rats. Glomerular Θ_albumin decreased from 0.0015±0.0002 in older MWF rats under baseline conditions to 0.00092±0.00007 in older MWF rats treated with WGA lectin (Figure 4B) (P<0.05; one-way ANOVA, Bonferroni). A similar pattern of reduced Θ_albumin after lectin binding to the ESL, although not statistically significant, was also observed in young MWF rats (lectin binding: 0.00043±0.00009; baseline conditions: 0.00065±0.00004; P>0.05; one-way ANOVA).

Coupled examinations also revealed that the degree of albuminuria in individual animals correlated with the degree of ESL cover of both mesenteric (Pearson r=−0.87; P<0.005) and glomerular (Pearson r=–0.94; P<0.001) capillaries in the same animal: higher degrees of albuminuria were associated with significantly sparser coverage of microvessel walls with ESL. In individual glomeruli, ESL coverage and Θ_albumin were also significantly correlated: glomeruli with less ESL cover had higher Θ_albumin (Figure 4C).

Discussion

We report widespread disruption of microvascular permeability and ESL in MWF rats with spontaneous established proteinuric kidney disease, using physiologic and multiphoton microscopy measurements of single microvessels in vivo, which allow unparalleled resolution of the structure and function of the microcirculation.^39,42 Changes in both permeability and ESL in these animals mirrored the consequences of enzymatic removal of ESL in healthy animals. In addition, application of the same enzyme (neuraminidase) that successfully removed labeled ESL in healthy animals was ineffective in altering either the structure or function of glomerular capillaries in animals with proteinuric kidney disease. Modification of the ESL in old MWF rats with proteinuric kidney disease significantly reduced glomerular albumin permeability. These findings indicate that ESL loss contributes to increased microvascular permeability in this model of proteinuric kidney disease.

Widespread ESL loss has been reported in individuals with diabetes,^33,34 albeit recognizing the difficulties of measuring the ESL in humans.³² Acute hyperglycemia directly suppresses ESL component synthesis³⁶ and reduces ESL volume in healthy individuals.⁴³ ESL loss may therefore represent a specific consequence of diabetes. However, individuals with diabetes and albuminuria exhibit greater loss of ESL (despite equivalent glycemic control³³ and the predicted additional elevations in systemic microvessel permeability,⁴⁴ suggesting that the development of renal dysfunction may also result in widespread ESL damage. These reports are in accord with our findings of widespread loss of ESL and permeability defects in a model of proteinuric kidney disease, and support previous contentions^12–14 that alterations in ESL may provide a functional link between kidney disease and extrarenal manifestations of vascular dysfunction.

Increased permeability of nonrenal vessels (Figure 1) was considered the most likely explanation for the increased limb capillary filtration coefficient (product of L_P and available surface area) demonstrated in patients with nephrotic syndrome,⁶ and may contribute to the development and perpetuation of edema in nephrotic states.^6,45 We and others have shown that enzymatic removal of ESL in mesenteric microvessels of healthy animals increases L_P,^30,31 as observed in older MWF rats with reduced ESL depth and cover, although undoubtedly the endothelial glycocalyx is only one important layer of the composite barriers presented by capillary walls.⁴⁶ Widespread, functionally relevant loss of ESL in albuminuric conditions may also result in large blood vessel dysfunction and disease. Both removal of ESL and albuminuria result in a nitric-oxide–dependent disruption of flow-dependent vasodilation^5,16; and loss of ESL in proteinuric kidney disease may contribute to this aspect of vascular dysfunction. In atheroma-prone vessels, ESL limits LDL entry into the vessel wall,⁴⁷ is thinner at atherosclerosis-prone sites,⁴⁸ and is lost in response to maneuvers that accelerate atherosclerosis (e.g., high-fat diet).⁴⁸ Pharmacologic disruption of ESL accelerates atherosclerosis.¹⁸ We speculate that disruption of ESL in proteinuric kidney disease also contributes to the accelerated atherosclerosis that affects these individuals. Techniques for examining the ESL in humans are being developed,^32,33 and identifying whether the phenomenon demonstrated here also occurs in humans with proteinuric kidney disease and low-grade albuminuria is an important question.

Our observations confirm previous reports that the ESL is a functionally important layer of the healthy glomerular filtration barrier.^26–28,49 Jeansson et al. recently reported an 80% reduction in glomerular ESL depth, coupled with a 3.5-fold increase in Θ_albumin, shortly after induction of adrimaycin nephropathy in mice.³⁵ These changes are larger than in our study, which is in keeping with our finding that greater loss of glomerular ESL results in a larger increase in Θ_albumin. Moreover, we report that glomerular ESL remains diminished 10–20 weeks after the spontaneous onset of albuminuria in older MWF rats,⁵⁰ in the context of a number of clinical features reminiscent of human proteinuric kidney disease. In addition, our demonstration that neuraminidase was incapable of removing more ESL or elevating Θ_albumin further in older MWF rats suggests that this loss of ESL continues to make an important contribution to increased Θ_albumin in these animals. In addition to the observed changes in the ESL, however, it is likely that damage to other layers of the glomerular filtration barrier is present in these glomeruli with FSGS, and occurs in response to neuraminidase treatment,⁵¹ thereby contributing to the functional defects of this multilayered barrier in these settings. However, modification of the ESL, through adsorption of lectins and other ligands (e.g., ferritin), alters microvascular function in other organs,^52,53 and the observation that binding of WGA lectin to the ESL significantly improved glomerular albumin permeability in old MWF rats provides a preliminary indication that modification of the ESL may be an effective therapeutic strategy to reduce albumin leak across the glomerular capillary wall in kidney disease. Because the ESL also regulates endothelial responses to hemodynamic stimuli and regulates nitric oxide bioavailability,⁵⁴ and supplementation of the ESL can be achieved with endogenous and therapeutic agents,^31,55 modification of the ESL may improve a number of aspects of glomerular function in kidney disease.

This study extends previous reports of multiphoton imaging of ESL^56,57 to the intravital setting, allowing real-time examination of the ESL, including before and after intervention, in multiple organs of the same animal, and coupling these measurements with assessment of physiology. The approach appears sufficiently sensitive to detect changes in the ESL induced by enzymatic degradation and disease states. The depth of ESL in healthy animals (0.57–0.65 µm) is comparable with results obtained with electron microscopy,^58–60 cell exclusion,⁶¹ lectin binding,⁶² and particle velocimetry,⁶³ although significant variability within and between techniques, and between organ beds, is reported.⁶⁴ The dimensions of the ESL approach the optical resolution of fluorescence microscopy, however, and current limitations precluded coupling of microvascular physiology and glycocalyx anatomy assessments in real time in cannulated vessels (required for measurement of mesenteric permeability coefficients), in contrast to our ability to couple measurements of structure and function before and after neuraminidase in glomeruli.

Increased glomerular L_P in old MWF rats with proteinuric kidney disease has also been noted in micropuncture studies.⁵⁰ Low Θ_albumin values in healthy, nonalbuminuric, young MWF rats in this report (approximately 0.0006) are also comparable with Θ_albumin values obtained with other techniques in healthy animals in vivo,^27,65,66 but are significantly lower than previous estimates made with multiphoton microscopy.^40,41,67 Although the magnitude of Θ_albumin varies with technical factors,^39,41,68 it was possible to detect differences in Θ_albumin between experimental groups (young versus older MWF rats), before and after ESL disruption (neuraminidase) and binding (lectin), and to discern significant relations between increasing Θ_albumin and the magnitude of albuminuria, and between Θ_albumin and ESL coverage. However, we note that Θ_albumin doubled in older MWF rats, whereas final urine albumin concentrations were approximately 10-fold higher. Focal areas of high glomerular clearance of albumin, either from a small number of individual glomeruli or from leaky areas within a single glomerulus,⁴² could substantially increase the amount of albumin in final urine. In addition, although the magnitude of tubular handling of albumin has been extensively debated,⁶⁹ renal tubules do reabsorb significant quantities of filtered albumin.^70,71 We note flattened tubular epithelium in these older MWF rats (Supplemental Figure 1), and altered glomerular and tubular handling of albumin may coexist in this model of proteinuric kidney disease.

In summary, we have extended recent advances in multiphoton imaging techniques, in combination with accurate measurements of microvascular permeability, to demonstrate widespread loss of ESL and altered permeability in an animal model of proteinuric kidney disease. We have also shown that modification of the ESL can improve glomerular albumin permeability in this model of proteinuric kidney disease. These findings have important implications for understanding the mechanisms of both altered glomerular permeability and widespread vascular dysfunction in proteinuric kidney disease.

Concise Methods

Animal Models

Male MWF rats, examined at 4–6 months of age (“older MWF”), were compared with either healthy age-matched male Wistar rats (mesentery studies) or healthy young (aged 4–5 weeks) male MWF rats (“young MWF”). Two groups of control rats were included to accommodate potential strain-related (Wistar versus older MWF rats) and age-related (young MWF versus older MWF rats) differences in ESL. In addition, although young MWF exhibit the superficial glomeruli required for multiphoton imaging of Θ_albumin and glomerular ESL, their small size before the development of albuminuria (90±13 g) precluded mesenteric microvessel cannulation studies. Conversely, age-matched Wistar rats were suitable for mesenteric microvessel cannulation studies, but these rats do not have superficial glomeruli for renal imaging studies. In total, 29 MWF rats (7 young; 22 older) and 21 Wistar rats were included in this study.

Wistar and MWF rats were purchased from Harlan; MWF rats were subsequently bred in-house at the University of Southern California (USC) (Los Angeles, CA). Rats were kept under environmentally controlled conditions with a 12-hour/12-hour light/dark cycle at 20–22°C, were fed a standard rat diet, and had free access to food and water. Permeability studies (other than Θ_albumin) were conducted at the University of Bristol (UB) (Bristol, UK), and imaging studies were conducted at USC. All experiments were carried out in accordance with national regulations, including the Guide for the Care and Use of Laboratory Animals of the US National Institutes of Health (NIH) as well as UK Home Office legislation. Local approval was obtained from the USC Institutional Animal Care and Use Committee as well as the UB local ethical committee.

Phenotype Analyses

Rats were housed in metabolic cages (Techniplast) for collection of random urine samples. Urine albumin was measured with a rat-specific albumin ELISA (Exocell, Philadelphia, PA), urine protein with spectrophotometric analysis of pyrogallol molybdenum-protein complexes (Thermo Fisher Scientific, Waltham, MA), and urine glycosaminoglycans by spectrophotometry of Alcian blue-urine glycosaminoglycan complexes formed in acidic solution.⁷² To account for differences in urine volume, urine creatinine concentration was determined in the same sample using either an enzymatic colorimetric assay (UB) (Konelab T-Series 981845; Thermo Fisher Scientific, Finland), or ELISA (USC). Results obtained with the two techniques were identical (UB: 65.49±7.28 mg/dl; USC 65.51±7.06 mg/dl; P>0.99, unpaired t test). Systolic BP was measured using a noninvasive specialized tail cuff and pulse transducer (ADInstruments; Visitech Systems).

Structural Analyses

Kidney samples from Wistar and older MWF rats were formalin fixed, embedded in paraffin, sectioned (5 µm), mounted onto glass slides, dewaxed (Histoclear; RA Lamb, Eastbourne, UK), and rehydrated through graded ethanol solutions (100%, 90%, and 70% vol/vol). Mounted sagittal kidney sections were then stained with hematoxylin and eosin and imaged using a Nikon Eclipse E600FN Microscope, Coolpix 995 digital camera, and a DN-100 digital imaging system (Nikon Instruments). Average glomerular volume (V_MEAN) in each animal was determined using the method described by Pagtulanan et al.⁷³ Glomerular images were replicated in Adobe Photoshop CS3 (Adobe Systems Inc, San Jose, CA) and the area (Aμm²) calculated using Image J software (NIH, Bethesda, MD). In each animal glomerular area (A) was measured in at least 30 profiles from at least three different sections. Glomerular volume was calculated from the area measurements using Equation 1:

where β (the shape coefficient for spheres: the idealized shape of glomeruli) equals 1.38, and k (the size distribution coefficient) equals 1.1.^73,74

Microvascular Permeability

Mesenteric Permeability

The hydraulic conductivity (L_p – water permeability) and the effective oncotic pressure (σΔπ − protein permeability) were measured in individually perfused rat mesenteric microvessels in Wistar and older MWF rats using a well characterized²⁴ modification of the Landis technique reported by Michel et al.⁷⁵ Anesthesia was induced by intramuscular injection of fentanyl, fluanisone, and midazolam, with maintenance subcutaneous doses titrated to maintain suppression of reflex responses. The mesentery was teased out over a transparent quartz pillar and transilluminated, and images were recorded on videotape via an inverted bright-field microscope (Leica DM IL HC Fluo). The mesentery was continuously superfused with warmed (37°C) physiologic mammalian Ringer solution (all in mm): NaCl 132, KCl 4.6, MgSO₄ 1.27, CaCl₂ 2, NaHCO₃ 25, D-glucose 5.5, Hepes acid 3.07, and 1.9 Hepes sodium salt, pH corrected to7.45±0.02 with 0.115 m NaOH if required.

Individual mesenteric microvessels were cannulated using a beveled glass micropipette, pressurized (30–70 cmH₂0) via an attached, adjustable manometer and perfused with physiologic Ringer solution containing BSA at a concentration (C) of 50 mg · ml⁻¹ (solution oncotic pressure [π_c] calculated as 27.2 cmH₂O⁷⁶), and containing rat erythrocytes as flow markers. For each pressure, the microvessel was transiently occluded downstream of the cannulation site. No vessel was occluded >5 times in the same site to avoid measurements being made on damaged portions of capillary. The rate of movement of erythrocytes within the occluded portion of vessel represents fluid flux across the vessel wall between the erythrocyte and the occluder. The radius of the vessel (r), velocity of the marker cells approximately 2 seconds after occlusion (dL/dt) and the distance between the marker cell and the occlusion site (L) were measured from recorded images off-line using Apple iMovie and Image J software. The rate of fluid efflux per unit area (J_v/A) was calculated using Equation 2:

The rate of fluid efflux during each occlusion (J_v/A) was plotted against the imposed capillary hydrostatic pressure, and a linear regression line applied to the data points. Assuming the capillary hydrostatic pressure was equal to that in the manometer, and the interstitial hydrostatic pressure was zero,⁷⁷ the slope of the regression line describes L_p, and the abscissal intercept describes the effective oncotic pressure of the luminal solution, σΔπ.⁷⁸ The reflection coefficient of the vessel wall to albumin (σ) was calculated from the relation σΔπ = σ²π_c.⁷⁸

Glomerular K_f

The glomerular K_f (hydraulic conductivity-surface area product: L_pA) was determined using an oncometric technique initially reported by Savin and Terreros⁷⁹ and modified by Salmon et al.⁸⁰ After measurement of mesenteric microvessel permeability, rats were culled by cervical dislocation and kidneys removed. Glomeruli were isolated in precooled mammalian Ringer solution supplemented with 10 mg · ml⁻¹ BSA using a standard sieving technique. Individual glomeruli were loaded onto the tip of a suction micropipette within a flow controlled closed system. Initial glomerular volume (V_i) was determined off-line from the glomerular profile. Swift exchange of the surrounding solution from 10 mg · ml⁻¹ BSA to 80 mg · ml⁻¹ BSA sets up an oncotic pressure gradient across the glomerular capillary wall, creating an absorptive force that draws fluid out of the glomerulus, resulting in a reduction in glomerular volume. The rate of glomerular volume change represents the rate at which fluid moves across the glomerular filtration barrier (J_v). Glomerular volume measurements were made from individual video images immediately before and after exchange of bathing solution. Glomerular L_pA (nl · min⁻¹ · mmHg⁻¹) was calculated from the quotient of the rate of glomerular volume change and applied oncotic pressure, as shown in Equation 3:

Microvascular Imaging Preparation

Animals were anesthetized for imaging studies with thiobutabarbital (Inactin: 65 mg/ml intraperitoneally), and the right carotid artery and trachea of older MWF rats were cannulated with PE-50 and PE-160 tubing, respectively (PE-10 and PE-90 were used in young MWF rats). Core body temperature was maintained at 36±0.5°C throughout the procedure with a homeothermic device, and BP was maintained at postinduction baseline values throughout the experiment with an infusion of 0.9% saline solution, supplemented with the volume required for delivery of fluorophores. The mesentery was exteriorized via a left flank incision, and imaged through a no. 1 glass coverslip during continuous superfusion with physiologic Ringer solution. After mesenteric microvessel imaging, the mesentery was carefully returned to the abdominal cavity, and the left kidney exteriorized through the same incision. When necessary, a ligature was applied to one end of the incision to maintain the external position of the kidney without compromising the blood supply of the kidney. The kidney was stabilized in a custom-designed holder placed above the same coverslip. Images of mesenteric microvessels after neuraminidase were obtained by reversing the order of imaging (i.e., kidney then mesentery). All images were acquired using an inverted microscope (Leica TCS SP5 AOBS MP confocal microscope system; Leica Microsystems, Heidelberg, Germany) using an HCX PL APO 63×/1.3NA glycerol CS objective (Leica).

Glomerular Albumin Sieving Coefficient

Rat serum albumin (A6414; Sigma, St. Louis, MO) was labeled with atto-550 NHS Ester (51146; Sigma) according to the manufacturer’s instructions. Labeled albumin was separated from free dye by gel permeation chromatography (PD-10 column; Sigma), followed by repeated and extensive centrifugation across a 30-kD molecular weight cutoff membrane (Pall, Ann Arbor, MI).

Fluorescence excitation at 820 nm wavelength was achieved with a Chameleon Ultra-II multiphoton laser (Coherent, Santa Clara, CA), and the emitted, nondescanned fluorescent light was detected by two external photomultipliers (green and red channels) after passage through a FITC/TRITC filter block (Leica). Photodetector gain (red: 775V), amplifier offset (−25%), and system output (16-bit) were fixed for all images in which measurements of Θ_albumin were made. The relative positions of up to seven glomeruli per animal were recorded, and background images were obtained, before delivery of labeled albumin, such that individual glomeruli could be identified for repeated imaging before and after intervention (e.g., infusion of neuraminidase).

Labeled albumin (approximately 1 ml/kg body wt of approximately 1.5 mg/ml solution; selected after preliminary experiments with the quick look-up table setting to achieve plasma fluorescence intensity [I_F^plasma] that did not saturate the detection system) was delivered via the carotid cannula, and allowed to equilibrate for approximately 10 minutes before imaging. Up to 20 sequential images were collected over 15–60 s from superficial optical sections of surface glomeruli, to minimize the consequences of light scattering in deeper renal tissue. An area of the urinary space was recorded as the locus for urinary space fluorescence intensity measurements (I_F^US) after confirming that there were no capillaries (from which out-of-focus fluorescence could be collected) in the same x-y position but at different depths.

Image analysis was performed with LAS AF software (Leica). Mean I_F^US was recorded in a region of interest in the previously identified urinary space area. The region of interest was created as large as possible to obtain a representative average value of I_F^US in a single optical section. The average value of these readings across the 20 sequential frames was recorded, from which background values were subtracted. For I_F^plasma measurements, obtaining a representative measurement is most reliably achieved by recording I_F^plasma at the margin of glomerular capillaries, where scattering and absorption of both exciting and emitted light by erythrocytes is minimized. I_F^plasma was recorded at the margin of four glomerular capillaries in a single optical section, selected from the 20 sequential images because of equal intensity across capillaries and minimal interference from erythrocytes, and background values were again subtracted. The lower limit of detection of the system was calculated as 4.65 × SD of I_F^US in all background images, divided by mean I_F^plasma in all experimental images.⁴¹ The albumin sieving coefficient was calculated as I_F^US/ I_F^plasma, and the mean ± SEM values for an individual glomerulus were calculated from the four estimates of I_F^plasma made in that glomerulus.

ESL Imaging

FITC-labeled wheat germ agglutinin (FITC-WGA) lectin (from Triticum vulgaris; 6.25 mg/kg body wt; L4895; Sigma, MO), administered via the carotid cannula approximately 30 minutes before imaging, was used to label N-acetyl glucosamine (a component of heparan sulfate and hyaluronan glycosaminoglycans) and N-acetyl neuraminic acid (a major sialic acid present on the endothelial cell surface) oligosaccharide moieties on the endothelial cell surface.²⁶ N-acetyl glucosamine residues are present in both the surface-bound “glycocalyx” region and loosely adherent “cell coat” region of the ESL.¹⁵ For mesenteric imaging studies, Argon (488 nm; 38% power) and He/Ne (543 nm; 62% power) lasers were used for single photon excitation. Emitted fluorescent signal was collected with 512×512-pixel resolution in 148×148-µm images with 16-bit intensity output, using an internal (de-scanned) detector (gain 950 V, offset –6%) with bandwidth set at 500–535 nm (FITC) using an AOBS bandwidth modification system (Leica), alongside differential interference contrast (DIC) detection. With these settings, the optical (approximately 0.24–0.29 µm) and digital (0.29 µm) resolution of the system were matched. For glomerular studies, FITC-WGA lectin was excited at the same 820 nm wavelength used for excitation of labeled albumin, by a Chameleon Ultra-II multiphoton laser (Coherent, Santa Clara, CA), and the emitted, nondescanned fluorescent light emitted by FITC-WGA and atto-550-albumin were detected simultaneously by two external photomultipliers (green: gain 1100 V, offset –2.5%, other settings and red channel as above) with the help of a FITC/TRITC filter block (Leica). For both kidney and mesenteric tissue, all identifiable microvessel networks were imaged.

Images were analyzed with LAS AF software (Leica). For mesenteric studies, DIC images of microvessels were used to identify the line of greatest contrast between the endothelial cell border (dark) and apposed vessel lumen (bright) as the luminal border of endothelial cells (Supplemental Figure 2). To determine mean ± SD background FITC fluorescence signal, 10 regions of interest were placed within each of the vessel lumen and mesenteric interstitium. A rectangular region of interest (1 μm perpendicular to vessel wall × 2–5 μm parallel to vessel wall) was placed at the edge of the vessel lumen (i.e., with one edge abutting the luminal border of endothelial cells), as defined above. FITC fluorescence signal intensity in this vessel margin region that was greater than the mean + 2 SDs of both the plasma and interstitial signal was taken to indicate the presence of FITC-WGA labeling material at the vessel wall, interpreted as presence of ESL. Ten replicate measurements were made in each vessel.

Depth of the ESL in mesenteric microvessels was determined by placing a line of interest across the vessel wall. The position of the pixels with the greatest difference in DIC signal intensity was identified as the luminal border of the endothelial cell (as defined above). To ensure consistency, the darkest pixel (endothelial cell) was used as the vessel wall reference point. The distance between this reference point and the position of the peak FITC signal was taken as the depth of the ESL.

For glomerular studies, only ESL coverage could be measured, because no glomerular endothelial cell label was routinely available. Glomerular anatomy precluded DIC imaging, and the use of two fluorophores for coupled structure-function studies required both available photodetectors and precluded routine use of a fluorescent label for endothelial cells. The presence of ESL was determined as described above (vessel wall signal exceeds mean + 2 SDs of vessel lumen signal) at four equally spaced loci around the margin of three randomly selected capillary loops in every identifiable glomerulus.

Enzymatic Removal of ESL

Neuraminidase (N2876 from Clostridium perfringens [C. welchii], type V, lyophilized powder; molecular mass of approximately 65 kD), an enzyme that cleaves sialic acid residues of oligosaccharides and glycoproteins and has been shown to remove the ESL from the surface of glomerular endothelial cells,^26,81 was prepared and stored according to the manufacturer’s instructions to maintain activity. Neuraminidase was administered via the carotid cannula over 5–10 minutes at a low dose of 1.6 IU/kg body wt.^81,82 Studies in healthy control rats and older MWF rats were interspersed to ensure consistency of enzyme action.

Statistical Analyses

Unpaired t tests and one-way ANOVA were used to compare permeability coefficient values of ESL measurements between groups; reflection coefficient values cannot exceed 1 and were therefore analyzed with nonparametric statistics (Mann–Whitney test). Correlations were assessed with Pearson’s (parametric) or Spearman’s (nonparametric) method. Multiple vessels were analyzed in each animal: results from individual vessels were pooled for each animal, and the number of animals studied was used for statistical comparisons. Glomerular L_PA measurements were the exception to this rule (as justified in Salmon et al.⁸⁰): the number of glomeruli studied was compared between groups. Where permeability coefficient or ESL measurements were made in the same vessel before and after neuraminidase, paired t tests were used for comparisons. A P value <0.05 was considered statistically significant.

Disclosures

None.

This work was supported by the British Heart Foundation (FS/05/114/19959 to J.K.F.), Medical Research Council and Academy of Medical Sciences (G0802829 to A.H.J.S.) and National Institute of Diabetes and Digestive and Kidney Diseases Grant DK64324 (to J.P.P.).

Part of this work was presented in abstract form at the American Society of Nephrology Kidney Week 2011, held November 8–13, 2011, in Philadelphia, Pennsylvania.