Recent studies of the sieving of serum albumin in the rat kidney using a two-photon microscope suggested that the glomerular sieving coefficient (GSC) of albumin is 0.034, much higher than earlier micropuncture determinations. In the present study, we critically evaluated the use of the two-photon microscope to measure the GSC of albumin in the Munich-Wistar rat in vivo. The albumin GSC averaged 0.004 (SD 0.004), n = 34 glomeruli, when determined with a Zeiss two-photon microscope system and 0.002 (SD 0.002), n = 5, when determined with an Olympus two-photon microscope system. These values are close to the lower limit of detection of GSC, which we estimate to be ∼0.001–0.003. We identified several factors that were likely responsible for the higher albumin GSCs reported earlier using two-photon microscopy. These include animal conditions (i.e., low glomerular filtration rate) and failure to recognize the role of out-of-focus fluorescence in contaminating the fluorescence signal from the urinary space of Bowman's capsule. We observed that hypothermia plus dehydration or a low blood pressure led to an increased albumin GSC. High levels of illumination (high laser outputs) resulted in a falsely elevated albumin GSC. Use of external, instead of internal, photodetectors resulted in an exaggerated albumin GSC because of greater collection of out-of-focus fluorescence. In conclusion, the albumin concentration in the glomerular filtrate of the normal rat, determined by two-photon microscopy, is exceedingly low (5–10 mg/dl).
- glomerular permeability
- multiphoton microscopy
- rat kidney
the kidney glomeruli filter the plasma at a high rate, but produce a filtrate that is generally thought to be nearly protein-free. The exact concentration of plasma proteins, such as serum albumin, in the glomerular filtrate is highly controversial. A recent study by Russo et al. (38) in Munich-Wistar rats suggested that renal albumin filtration is 50 times higher than the lowest values previously reported (43). Such a high rate of albumin filtration implies that the kidney tubules normally reabsorb proteins at a high rate, because the final urine contains little protein. Russo et al. suggested that most of the reabsorbed (endocytosed) albumin was returned intact to the blood, i.e., transcytosed. These new findings, if confirmed, would represent a paradigm shift in our thinking of how the kidney handles albumin, and they have engendered considerable discussion (7, 8, 10, 16, 18, 31, 35–37).
Russo et al. used a new approach, two-photon in vivo microscopy, to measure the glomerular sieving coefficient (GSC) of albumin in anesthetized rats. GSC depends on the glomerular permeability to albumin and the rate of filtration of water. Albumin GSC was calculated from the albumin concentration (fluorescence) in the urinary space of Bowman's capsule divided by the albumin concentration (fluorescence) in the glomerular blood plasma. The two-photon microscopy approach is appealing because of its directness, but its use in determining very low sieving coefficients has not been critically evaluated heretofore. Earlier estimates of glomerular sieving of albumin were based on kidney micropuncture, most often involving tubular fluid collections with extrapolation of values to Bowman's space (13, 15, 22, 28, 39, 43), and, on occasion, direct collection from Bowman's space (33).
In the present study, we reexamined the determination of GSC of fluorescently labeled rat serum albumin (RSA) using two-photon microscopy, and critically examined factors that could explain the high GSC reported by Russo et al. We found that the GSC of albumin averaged 0.004 using internal photodetectors on a Zeiss two-photon microscope and 0.002 using internal photodetectors on an Olympus two-photon microscope, values far below Russo's value of 0.034, and closer to values estimated from micropuncture studies. We identified several probable reasons why high albumin GSCs could be reported by two-photon microscopy in in vivo imaging studies.
The extent to which serum albumin is filtered is of major importance in renal physiology and pathophysiology. Excessive filtration of albumin and other plasma proteins may result in renal damage and appears to be a factor in the progression of chronic renal disease (1, 5, 6, 9, 44, 45, 47).
Animals and surgical procedures.
Experiments were done on 30 Munich-Wistar rats of both sexes from a colony, maintained at our school, derived from rats obtained from Simonsen Laboratories (Gilroy, CA). The same rat source was used by Russo et al. The Simonsen rats were originally selected for surface glomeruli, but have fewer surface glomeruli than the Munich-Wistar Frömter rat substrain. Rats were usually deprived overnight of food, but not water, before experiments, and in 14 rats urine was collected in a metabolic cage for measurement of urine protein excretion. Urine proteins were determined by the method of Lowry et al. (23) after filtration of the urine through filter paper and precipitation of urine proteins with trichloroacetic acid.
The rat was weighed and then anesthetized with 50 mg/kg body wt pentobarbital sodium ip. Surgical procedures included a tracheostomy, cannulation of the right femoral artery (for measurement of blood pressure and blood sampling) and femoral vein (for intravenous administration of fluids), and exposure of the left kidney by a flank incision. During surgery, we administered isotonic saline in an amount equal to 1% of body weight to improve the circulating plasma volume. This was followed by constant intravenous infusion of isotonic saline at 1.27 ml/h using a syringe pump. Arterial blood pressure was monitored using a Statham P23Db transducer and Beckman Dynograph recorder. An arterial blood sample (0.25 ml) was collected for measurement of total plasma protein concentration in seven rats. In three rats, we measured the glomerular filtration rate (GFR) of the imaged (left) kidney, using the clearance of polyfructosan, a synthetic inulin, as described previously (40).
The majority of experiments were done using an upright Zeiss Axioplasm-2 microscope. An adjustable warming plate was placed underneath the animal board. Rectal temperature was monitored, and the warming plate temperature was adjusted so as to keep the animal's temperature close to 37°C (observed range 36.2–37.8°C). The left kidney was supported by a micropuncture cup, with a coverglass lightly touching the upper surface.
In nine experiments, we used an inverted Olympus microscope. In this case, the left kidney was exteriorized through a small flank incision and the animal rested on top of the kidney (12); this configuration is similar to that used by Russo et al. The kidney was immersed in a dish of isotonic saline that was warmed by a stainless steel coil that contained circulating, heated water. The temperature under the kidney was measured at the end of four experiments and averaged 36.5°C (SD 0.1). Animal preparation and conditions when using the Olympus microscope were otherwise identical to those used with the Zeiss microscope system.
Experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, and protocols were approved by the Indiana University School of Medicine Animal Care and Use Committee.
Fluorescence images were most often acquired with a Zeiss SLM 510 Meta confocal/multiphoton microscope system. Observations were made using a Zeiss ×63 water-immersion objective (numerical aperture 1.2). Fluorescence excitation, at a wavelength of 800 nm, was provided by a titanium-sapphire laser (Spectraphysics, Mountain View, CA). Images were collected using the internal photodetectors, with the pinhole open to its maximum. The photodetectors were preceded by a 545-nm long-pass beam splitter and by 500- to 550 (“green channel”)- and 565- to 615-nm (“red channel”) band pass filters, coated with infrared blocker. We set the photodetector gain at 750 V in both channels, and adjusted the black level by changing the amplifier offset while viewing the kidney. Once these initial photodetector settings were made, they were not changed.
We identified one to four surface glomeruli and collected background images at laser outputs between 10 and 70%. The fluorescently labeled RSA, ∼1 mg in 0.5 ml isotonic saline, was then administered intravenously and after waiting 10 min to achieve stable conditions, the same glomeruli were imaged. In six experiments we gave both “green” (Alexa fluor 488)- and “red” (Alexa fluor 568)-labeled RSA, so that we could compare simultaneously two different RSA preparations (crystalline vs. Fraction V) in the same glomeruli. In these experiments, we always gave the green fluorescent molecule first so that we could correct for crossover of green fluorescence into the red channel. Crossover of red fluorescence into the green channel is negligible.
Images were also collected using an Olympus two-photon microscope and a ×60 water-immersion objective (numerical aperture 1.2). We used both internal and external photodetectors with this microscope system. The configuration for collecting the fluorescence emission with internal detectors (also used on the Zeiss 2-photon microscope) is identical to that used in confocal microscopy. The fluorescence signal returns to the scanning mirrors, i.e., it is “descanned” before passing through a pinhole on its way to the internal photodetectors. With external detectors, the fluorescence signal from the sample is redirected by a dichroic mirror toward detectors that are placed outside the scanning unit. As a result, the emission signal is not descanned by the scanning mirrors and, therefore, the external detector used in this configuration is called “nondescanned detector (NDD).” The external photodetectors on the Olympus microscope are GaAs detectors, and emission light does not pass through any elements of the scan head; consequently, the efficiency of light detection is higher than with internal detectors. The output of the Olympus system was a 12-bit intensity (gray level) scale, whereas the Zeiss system output was an 8-bit scale. Other conditions (excitation wavelength, laser outputs, mirrors and filters, etc.) were similar to those used with the Zeiss two-photon microscope. In these experiments, we gave the red fluorescent albumin first and gave green fluorescent compounds (fluoresceinated dextran 2,000,000 or fluoresceinated thyroglobulin) later. This approach did not allow the simultaneous measurement of fluorescence in red and green channels, but the same glomeruli were imaged sequentially, first with the red RSA and ∼20–30 min later with the green fluorescent probe. To correct for instrument drift, a sequence of blank images (laser light blocked) was collected seconds before collecting kidney images.
To reduce possible photodamage, we did not collect movies or image stacks, and limited the number of images collected from a single glomerulus. The glomeruli were selected using visible (blue) light from a mercury arc lamp. To reduce the possibility that readings in Bowman's space might be contaminated by fluorescence from glomerular capillary blood plasma, we most often focused on the largest diameter of the glomerulus, so that we could take readings in Bowman's space with no glomerular capillaries immediately above or below the image plane. We used the range indicator option when collecting images, to be certain that fluorescence in the blood plasma was not saturated. If photodetector saturation was seen, we lowered the laser output in 5–10% increments until there was no saturation of pixels in the plasma image.
The fluorescence signal in Bowman's space is extremely close to background, so to increase its intensity reading (the signal-to-noise ratio), we also used a second approach in five experiments. We collected an image at a low laser output (below saturation of the plasma fluorescence, as above), but then (within a minute) collected a second image at a high laser output (for example, see Fig. 1, A and B). The laser output, i.e., illumination, was changed with an acousto-optic modulator. At the higher laser output, the plasma fluorescence in all capillary loops was completely saturated. At the end of these experiments, we measured fluorescence intensity of a solution of the fluorescently labeled albumin in PBS and of PBS (background) as a function of the laser output at the same settings as were used in the experiment on the animal (see Fig. 1C). If, for example, the fluorescence intensity at a laser output of 20% was 25 and at 60% it was 75, then we would multiply the plasma intensity in the first collected image by 3 and assume that the resulting product was the appropriate plasma intensity for the second image. We found, however, that this approach gave unreliable values for the GSC of albumin.
RSA was obtained from Sigma (St. Louis, MO). Crystalline RSA (cat. no. A4538), ≥99% pure by agarose gel electrophoresis and essentially globulin-free, was further purified by passing it through a Sephacryl 100-HR gel filtration column, using PBS with 2 mM sodium azide as the eluant. Peak albumin fractions were collected. The albumin was labeled with fluorescent Alexa fluor 488 or 568 dyes using kits from Invitrogen Molecular Probes (Eugene, OR). After labeling, the material (0.5 ml solution) was dialyzed in a cold room using 12,000–14,000 molecular weight cut-off dialysis tubing and 2–3 changes of 6 l of distilled water containing 2 mM azide for 1–2 days. We also prepared fluoresceinated crystalline RSA (17), dialyzed it, and then passed the fluoresceinated albumin through the gel filtration column, collecting the peak fractions. In five experiments, we used Fraction V RSA (Sigma, cat. no. A6272), purity 97–99%, labeled it with Alexa fluor dyes, and dialyzed this material, but did not pass it through the gel filtration column. On the morning of the experiment day, albumin solution samples were rinsed by adding isotonic saline in an Amicon Ultra-15 30,000 Da nominal molecular weight limit (NMWL) filter centrifuge tube. Typically, we started with 0.5 ml of labeled albumin solution, added 14.5 ml isotonic saline, ultrafiltered the solution down to a volume of ∼0.5 ml, and repeated this process two to four additional times. Thus, the reduction in azide concentration in the final samples was ∼303- to 305-fold, and the total amount of azide injected into the rat was negligible. This process also removed any residual free dye or contaminants with a molecular weight less than 30,000 Da.
We also studied two molecules, which because of their very large size, should not be filtered at all by kidney glomeruli. Fluoresceinated dextran 2,000,000 (cat. no. D7137) was obtained from Invitrogen Molecular Probes, dialyzed, and repeatedly rinsed and filtered in an Amicon Ultra-15 100,000 Da NMWL centrifuge tube. Bovine thyroglobulin (MW 670,000) was obtained from Sigma (cat. no. T1001), reacted with FITC (17), and exhaustively dialyzed. The fluoresceinated thyroglobulin was then purified by passage through the gel filtration column, and it was repeatedly rinsed and filtered in an Amicon Ultra-15 100,000-Da NMWL centrifuge tube just before use.
Quantitative image analysis.
Images were analyzed using MetaMorph v. 7.1 (Molecular Devices, Downingtown, PA), and calculations were done on a Microsoft Excel spreadsheet. The fluorescence intensity in the plasma was recorded in a capillary loop which typically had the brightest fluorescence in the field. Plasma fluorescence was usually measured at the outer margins of the capillary lumen, a blood cell free zone. We selected an area in the urinary space of Bowman's capsule where there were no blood vessels and where fluorescence intensity was minimal. We recorded the number of pixels (area) and average fluorescence intensities. Calculations of GSCs were based on average intensity values after subtracting background readings for plasma and Bowman's space. Data were analyzed by paired or unpaired t-tests, by an ANOVA after a preliminary test for homogeneity of variances, or by linear regression. A P value <0.05 is considered significant.
Microscope calibration, lower limit of detection, and stability.
To determine the sensitivity of the Zeiss and Olympus microscope systems to fluorescence, we imaged solutions of fluorescein and dextran rhodamine in PBS and measured average pixel intensities. Figure 2 shows a representative calibration for the green channel of the Zeiss microscope using fluorescein solutions and photodetector voltages and laser outputs typical for our in vivo experiments. A perfectly linear relationship between fluorescein concentration and fluorescence intensity was observed, as long as fluorescence intensities were below levels that would saturate the photodetectors. The system was linear even in the range of extremely low intensities (Fig. 2B), intensities in the range of those encountered in Bowman's space in vivo. The Olympus microscope system was also linear, and, with typical settings, the slope of the line (intensity vs. fluorescein concentration) was about eight times higher. The linearity of Fig. 1C (fluorescence intensity of labeled serum albumin vs. laser output) indicates that there was no fluorescence saturation (which is different from photodetector saturation). This means that there was no ground state depletion (30) at the highest laser power intensities that we used.
We also determined the lower limit of detection (3) of fluorescence using the Zeiss and Olympus microscopes and typical photodetector and laser output settings used in vivo. This was done by collecting 4–12 background images per glomerulus from 5–7 glomeruli. The fluorescence intensity was measured in Bowman's space, and means and standard deviations were calculated.
With the Zeiss microscope system's internal photodetectors, Bowman's space background intensity in the green channel averaged 5.9 (SD 0.6), n = 6. The standard deviation of the background intensity averaged 0.06 (SD 0.03), n = 6, and so the lower limit of detection, 4.65 × 0.06, was 0.26. Plasma fluorescence intensity after giving green fluorescent albumin averaged 119 (SD 40), n = 20, in our experiments, so the lower limit of detection of albumin GSC is ∼0.002 (i.e., 0.26/119) in the green channel. For the red channel, background intensity averaged 5.2 (SD 0.5), n = 6. The standard deviation of the background intensity averaged 0.07 (SD 0.01), n = 6, so the lower limit of detection, 4.65 × 0.07, was 0.34. Plasma fluorescence intensity after giving red fluorescent albumin averaged 125 (SD 33), n = 41, in our experiments, so the lower limit of detection of albumin GSC is ∼0.003 (i.e., 0.34/125) in the red channel. With the internal photodetectors on the Olympus microscope system, Bowman's space background intensity in the green channel averaged 48 (SD 2), n = 5. The standard deviation of the background intensity averaged 0.5 (SD 0.2), n = 5, so the lower limit of detection, 4.65 × 0.5, was 2.4. Plasma fluorescence intensity after giving fluoresceinated macromolecules averaged 1,810 (SD 564), n = 5, in our experiments, so the lower limit of detection of GSC is ∼0.001 (i.e., 2.4/1,810) in the green channel. For the red channel, background intensity averaged 48 (SD 2), n = 5. The standard deviation of the background intensity averaged 0.4 (SD 0.2), n = 5, so the lower limit of detection, 4.65 × 0.4, was 2.0. With a plasma fluorescence intensity averaging 1,670 (SD 507), n = 5, in our experiments, these data suggest a lower limit of detection of albumin GSC of 0.001 (i.e., 2.0/1,670) in the red channel.
With the external photodetectors on the Olympus microscope system, Bowman's space background intensity in the green channel averaged 51 (SD 11), n = 7. The standard deviation of the background intensity averaged 0.6 (SD 0.2), n = 7, so the lower limit of detection, 4.65 × 0.6, was 2.8. Plasma fluorescence intensity after giving green fluorescent albumin averaged 1,738 (SD 479), n = 7, in our experiments, so the lower limit of detection of albumin GSC is ∼0.002 (i.e., 2.8/1,738) in the green channel. For the red channel, background intensity averaged 56 (SD 6), n = 5. The standard deviation of the background intensity averaged 1.4 (SD 0.9), n = 5, so the lower limit of detection, 4.65 × 1.4, was 6.3. With a plasma fluorescence intensity averaging 1,883 (SD 402), n = 3, in our experiments, these data indicate a lower limit of detection of albumin GSC of ∼0.003 (i.e., 6.3/1,883) in the red channel.
From the above results, we conclude that the lower limit of detection of GSC on both microscope systems is ∼0.001–0.003. Our values for the GSC of albumin fall extremely close to these limits, which reduces our ability to measure accurately the albumin GSC. We also conclude that there was no clear advantage of one microscope system over the other in determining the GSC of albumin.
The photodetector drift on the Zeiss and Olympus microscope systems was tested over a 2-h period by collecting images with the laser blocked. With the Olympus (external photodetectors), the readings drifted upwards by 5 (green channel) and 4 (red channel) units. Such a change could lead to an error in GSC of ∼0.002. For this reason, when we did measurements in vivo, we always corrected for drift by taking a blank measurement (laser illumination blocked) 1–5 s before each set of images. On the Zeiss system, there was no significant drift (intensity increased 0.05 intensity units over a 2-h period).
In vivo measurements of GSC.
Table 1 summarizes physiological measurements done in our rat experiments. GFR in the imaged kidney and mean arterial blood pressure were normal. As expected, urine protein excretion was much higher in male than in female rats. This sex difference may be explained by excretion of a sex-dependent α-globulin (2, 34), lower rate of tubular reabsorption and catabolism of filtered proteins (4), and higher excretion of albumin in males than in females (2). We did not detect a significantly higher albumin GSC in males than in females. The average plasma protein concentration that we observed (4.35 g/dl) is lower than the average value (5.7 g/dl) reported for euvolemic Munich-Wistar rats (20), and most likely is due to administration of isotonic saline, instead of isoncotic rat plasma, during surgical preparation of our rats and administration of saline (∼0.5 ml) with each injection of fluorescently labeled compounds.
The albumin GSC, determined using the Zeiss two-photon microscope, averaged 0.004 (SD 0.004) in 34 glomeruli (Fig. 3). This mean is significantly different from zero at the P < 0.001 level, but is almost at the lower limit of detection (0.002–0.003). The mean is far below (∼1/8th) the average value for albumin GSC reported by Russo et al.
We sought to determine reasons why higher GSCs were reported in the study of Russo et al. Our first hypothesis was that their use of Fraction V albumin instead of purified, crystalline albumin could explain the discrepancy. In 15 glomeruli we compared the GSC of labeled purified albumin and Fraction V albumin simultaneously. The GSC of the former averaged 0.005 (SD 0.004) and of the latter 0.007 (SD 0.007), but these values were not significantly different (P > 0.1; paired t-test). Therefore, we conclude that impurities in the Fraction V albumin cannot explain the discrepancy.
Our second hypothesis was that the in vitro calibration method used in some of the experiments of Russo et al. was flawed. To increase fluorescence levels in Bowman's space, these authors gave large doses of fluorescent albumin that increased the fluorescence signal in Bowman's space and saturated the plasma fluorescence levels. They then used plasma sampling and measurements of plasma fluorescence in vitro to calculate the plasma fluorescence. We used a different approach. We first measured the albumin GSC at ordinary laser outputs (e.g., 20%). Then, we increased the signal in Bowman's space by increasing the laser output (e.g., 60%) and measured GSC again; the plasma fluorescence (now fully saturated) was calculated from an in vitro calibration curve (Fig. 1C) constructed at the end of each experiment. In 15 glomeruli, the albumin GSC averaged 0.005 (SD 0.004) at low laser outputs and 0.010 (SD 0.011) at high laser outputs; these values were different at the P < 0.05 level (paired t-test). The most likely explanation for this finding is that saturating the plasma fluorescence levels led to increased out-of-focus fluorescence, and, consequently, an erroneously high GSC.
Our third hypothesis was that the high albumin GSCs in the Russo et al. study were due in part to a low GFR. In that study, the animals were often not adequately hydrated and rectal temperature was not rigorously maintained. No tracheostomy was done, arterial blood pressure was not monitored, and there was no measurement of GFR. In two animals, we purposely gave no fluids (other than ∼0.5 ml saline needed to inject the fluorescent albumin) and allowed the rectal temperature of the rat to fall spontaneously to 33–34°C by turning off the warming plate. The albumin GSC in eight glomeruli averaged 0.015 (SD 0.012), almost four times the value observed when physiological conditions were maintained (P < 0.001). High albumin GSCs, averaging 0.017 (SD 0.014), n = 3, were also observed with the Olympus microscope (external photodetectors) in one rat that, for unexplained reasons, had an unacceptably low arterial blood pressure of 72 mmHg. A possible explanation for the greater than normal variability in these GSC measurements is glomerular intermittence (32), i.e., filtration rate in individual glomeruli or from one glomerulus to the next is more variable in stressed animals. Our results suggest that a reduced GFR may have contributed to the high albumin GSCs observed by Russo et al.
Fourth, we suspected that the use of internal vs. external photodetectors may have contributed to our discrepancies. We predicted that the internal photodetectors would detect less out-of-focus fluorescence. With the Zeiss instrument we used only the internal photodetectors, whereas the Russo et al. study had used external photodetectors on a Bio-Rad two-photon microscope. We studied this issue by using both internal and external photodetectors.
To test whether the internal photodetectors on the Zeiss microscope system detect significant out-of-focus fluorescence, we measured the GSC of dextran 2,000,000. In principle, this molecule should be barely filtered because of its very large size, and so its GSC should be zero if the signal in Bowman's space is not contaminated by out-of-focus fluorescence. Indeed, using our usual laser outputs (10–40%) in two rats, we calculated a dextran 2,000,000 GSC of 0.000 (SD 0.001), n = 8 glomeruli.
On the Olympus two-photon microscope system, we used both internal and external photodetectors. With the internal photodetectors, we observed that the GSC of dextran 2,000,000 was 0.000 (SD 0.001), n = 3, and the GSC of thyroglobulin was 0.000 (average for 2 glomeruli; Table 2). These measurements suggest no detectable out-of-focus fluorescence. The GSC of albumin, 0.002 (SD 0.002), n = 5 (Table 2), was almost at the lower limit of detection, 0.001. With the external photodetectors, we observed higher average GSCs for both dextran 2,000,000 and thyroglobulin, which is evidence that out-of-focus fluorescence contributed to the fluorescence ascribed to Bowman's space. For five glomeruli (2 in Table 2 and 3 additional glomeruli), the dextran 2,000,000 GSC averaged 0.005 (SD 0.002; P < 0.01 compared with 0). For five glomeruli (3 in Table 2 and 2 additional glomeruli), the GSC of thyroglobulin averaged 0.023 (SD 0.020; P = 0.057 compared with 0). The albumin GSC averaged 0.013 (SD 0.008), n = 10, significantly higher (P < 0.01) than values observed using internal detectors on this microscope. There was a highly significant correlation (r = +0.948, P < 0.001) between the uncorrected albumin GSC and the thyroglobulin/dextran 2,000,000 GSC (Table 2), which suggests that when out-of-focus fluorescence is low the calculated albumin GSC is low, and when it is high then the calculated albumin GSC is high. If we correct the albumin GSC for the GSCs of dextran 2,000,000 or thyroglobulin, as we can do in those instances in which these measurements were done in the same glomeruli (Table 2), we calculate a corrected albumin GSC of 0.007 (SD 0.004). From these data, it appears that out-of-focus fluorescence, which is more readily detected when external photodetectors are used, is one of several factors that can lead to erroneously high values for the GSC of albumin.
Results from the present study indicate that the GSC of serum albumin is exceedingly low and is almost at the lower limit of detection of current two-photon microscope systems. We found an average albumin GSC of 0.004 using the Zeiss two-photon microscope and 0.002 using internal photodetectors on the Olympus two-photon microscope system. With a total plasma protein concentration of 4.35 g/dl (Table 1) and assuming that albumin constitutes 59% of the plasma proteins in the rat (14), the calculated albumin concentration in the urinary space of Bowman's capsule would be 5–10 mg/dl. These values are higher than most recent micropuncture estimates, which fall in the range 0.7–3.3 mg/dl (13, 15, 22, 28, 31, 43). Micropuncture assessments are known to have problems. For example, it is recognized that binding of albumin to glass micropipettes could lead to falsely low values (22, 25, 29). Furthermore, most estimates of Bowman's space albumin concentrations obtained by micropuncture were calculated by linear extrapolation from proximal tubule fluid collections, and it is not certain that the rates of tubular reabsorption of albumin and water are linear with distance. Direct collection of filtrate by micropuncture is problematic, since insertion of a micropipette tip into Bowman's capsule often produces some visible damage to the glomerulus (personal observation) and can lead to very high and variable Bowman's space protein values (33). The very lowest values for albumin GSC, 0.0006, were reported by Tojo and Endou (43), but they discarded three-fourths of their measurements (the high values), suspecting that these samples were contaminated by extratubular fluids.
Studies on humans with Fanconi syndrome (27), a condition in which there is defective proximal tubule reabsorption of filtered proteins, have suggested an extremely low GSC for albumin (0.00008); this estimate is only valid if tubular reabsorption of albumin is indeed completely absent. Rats show a much higher rate of albuminuria (when factored for body weight or surface area) than humans (2), so it is possible that there are species differences in glomerular permeability to albumin that might explain the higher albumin GSCs reported for the rat. The Munich-Wistar rat, compared with other rat strains, shows an unusually high rate of albumin excretion, and albuminuria is even higher in the Frömter substrain (2). Hence, there might be differences in albumin GSCs in different rat strains or substrains.
Our values for albumin GSC are much lower than the very high values reported by Russo et al. Their measurements suggest a Bowman's space albumin concentration of ∼100 mg/dl. In the present study, we identified several reasons that could explain their high values.
The physiological state of the anesthetized rats in the Russo et al. study is not clear, because they did not report GFR or rectal temperature, do a tracheostomy, or monitor blood pressure. Anesthetized rats become hypothermic if not warmed, and kidney blood flow and GFR fall profoundly when rats are cold (11). In the Russo et al. study, the exteriorized kidney was immersed in a dish of unheated saline; we corrected this problem by using a heating coil for the kidney coverglass dish. It is advisable to do a tracheostomy to maintain a patent airway. Anesthetized, surgically operated rats become blood volume depleted and display a reduced GFR (20). Inadequate volumes of intravenous fluids to support cardiovascular and renal functions were given to many of the animals in the Russo et al. study. Lund et al. (24) provided data in rats demonstrating that the albumin GSC rises as the GFR falls. We believe that a reduced GFR may have contributed to the elevated albumin GSC in the Russo et al. study.
The problem of out-of-focus fluorescence was ignored in the previous study. Light is scattered when it passes through media of different refractive indexes (coverglass, saline, kidney capsule, kidney parenchyma). Although the ballistic, exciting light entering the tissue is concentrated at the focal plane by the microscope objective, fluorescence can also be excited elsewhere due to light scattering, especially when imaging deep in tissue (19, 42, 46). When we evaluated fluorescence levels in Bowman's space, we always subtracted a background reading (taken at the same laser output) collected before the fluorescent albumin was administered. If, however, there is significant out-of-focus fluorescence after administering the fluorescent albumin, then this background is really no longer appropriate, and an overestimate of the albumin GSC will result.
Our measurements with internal photodetectors on the Zeiss and Olympus microscopes suggest that there was no detectable out-of-focus fluorescence contaminating the signal from Bowman's capsule when ordinary laser outputs were used. With the external photodetectors, however, we observed higher GSCs of dextran 2,000,000, thyroglobulin, and albumin (Table 2), likely due to increased detection of out-of-focus fluorescence. External photodetectors collect scattered fluorescent light more efficiently, and are, therefore, recommended for deep tissue imaging (41), but they may not be optimal for measuring the GSC of albumin in superficial glomeruli. With the internal photodetectors, light passes through a pinhole on its way to the photodetector and, consequently, the detection of out-of-focus fluorescent photons is reduced. Ying et al. (46) recommended the use of a confocal pinhole before the photodetector to suppress the effect of light scattering during two-photon microscopy in light-scattering media.
Use of increased fluorescent probe concentrations (38) or higher laser power outputs (present study), both of which result in photodetector saturation of plasma fluorescence, does not appear to be a good strategy to increase the signal in Bowman's space. This approach is problematic because the slopes of in vitro and in vivo calibration lines (fluorescence intensity vs. laser output) will differ due to light scattering in tissue. The in vitro calibration line would be expected to have a steeper slope; this will lead to an overestimate of plasma fluorescence and hence an underestimate of albumin GSC. We observed that this approach actually led to higher calculated GSCs. Hence, increased out-of-focus fluorescence at high laser outputs is an even more serious problem. The same problem would be present at very high fluorescent probe concentrations (38).
In addition to the factors discussed above, there are other pitfalls in the two-photon approach, many of which will tend to increase the albumin GSC. Photodamage, with an ensuing increase in glomerular permeability, is a potential problem, but we believe that this was not present in the Russo et al. study or the current work, because intense light exposure was limited and there was no detectable increase in GSCs when multiple images were collected over time. Fluorescently labeled low molecular weight (i.e., filterable) impurities in the albumin preparation, free dye, or metabolism of the fluorescent albumin in the animal would lead to erroneously high values for albumin GSC. Saturation of plasma pixel intensity is a potential source of error, depending on the number of pixels saturated. Inaccuracies in plasma fluorescence measurements, however, have much less of an impact on the albumin GSC than errors in the Bowman's space value. The human element must also be considered. A human operator must select the region of interest in Bowman's space and plasma where intensity measurements are made. Overestimates of fluorescence emanating from Bowman's space can result if, for example, the investigator fails to recognize that a blood vessel is just above or below the focal plane; the blood vessel (with its ×250–500 higher plasma fluorescence intensity) might contribute a subtle, but significant, amount of fluorescence to the Bowman's space voxel. If the kidney preparation moved during the image collection, e.g., due to the pulse or breathing, then this error may not be obvious. On the other hand, if podocyte cell nuclei (seen as dark round circles) are included in the Bowman's space measurement, falsely low GSCs may result. When different investigators analyzed the same images and calculated the GSC of albumin, the differences among investigators were modest, and do not explain the large discrepancy between the present study and the earlier two-photon study (38).
Two-photon microscopy is an exceedingly powerful, relatively noninvasive technique for quantitative analysis of nephron function in vivo (21, 26), and can be used to measure the GSC of albumin. In the present work, we identified and studied several factors that might lead to errors in the determination of albumin GSC with this method. Most of the possible errors with the two-photon approach would produce an overestimate of albumin GSC, and so the normal albumin GSC is not likely to be higher than our values. Our results suggest that the albumin GSC is ∼0.002–0.004, higher than most values determined by kidney micropuncture in ordinary rats, but an order of magnitude less than reported in an earlier two-photon microscopy study (38). It does not appear to be necessary to postulate that the normal kidney filters nephrotic levels of albumin and reabsorbs massive amounts of intact albumin (38).
I thank the National Kidney Foundation of Indiana for grant support. The studies were conducted at the Indiana Center for Biological Microscopy.
I am greatly indebted to Dr. E. Wang for help with the microscopes and many helpful discussions, and to Dr. J. A. Tanner for help with preparing the manuscript and for support and encouragement.
- Copyright © 2009 the American Physiological Society