One of the most striking results that is seen with the introduction of labelled albumin into the circulation is that the labelled material in the urine is mainly associated with small peptides (Figure 1) (Osicka et al., 1997; Osicka et al., 2000; Russo et al., 2002; Comper et al., 2008; Vuchkova and Comper, 2015). This is the result of filtered albumin being degraded by the proximal tubular cell (PTC). The extent of this degradation can be modulated by the disease process. For example, in diabetic patients the excreted labelled albumin can be predominantly intact and is accompanied by a smaller quantity of labelled peptides (Osicka et al., 2000) for the same amount of labelled material being excreted as in healthy controls. This inhibition of degradation is a characteristic of the disease process. The degradation pathway is quite non-specific; all labelled proteins studied undergo degradation.

The first and only demonstration so far of circulating plasma albumin being filtered and then returned to the renal vein (retrieval) was published in 1999 (Eppel et al., 1999). Retrieval was measured through the introduction of a small pulse of radioactive [³H]albumin into the rat renal artery in vivo, followed by examination of the radioactive profile from the renal vein effluent. It is viewed that the filtered portion of [³H]albumin in the pulse will be taken up by the PTC retrieval pathway, which will return the albumin to the blood supply undegraded (Figure 1). This recycled albumin appeared as a second peak(s) in renal vein effluent after the initial bolus had passed. Quantitation of the glomerular flux of albumin from the size of the second peak relative to the control (controls included either [³H]horseradish peroxidase, inulin labeled with radioactive carbon-14, or [³H]albumin in a nonfiltering kidney) accurately estimated the predicted glomerular flux of albumin governed by size selectivity alone; the calculated GSC was 0.04. This meant that this pathway was relatively high capacity as compared to the degradation pathway. Similar results were obtained using ex situ kidney perfusion.

The existence of the PTC retrieval pathway was confirmed 14 years later using a transgenic mouse with podocyte-specific expression of doxycycline-inducible tagged murine albumin (Tenten et al., 2013).

Variation in albuminuria, without hypoalbuminemia, is common in diabetic kidney disease and in other diseased states. Figure 2C demonstrates that most of the data lies in the range where the degradation pathway would operate (below the DD line for protein clearance) indicating that albuminuria/proteinuria in diabetic states is predominantly a metabolic event associated with the integrity of the albumin being excreted (Osicka et al., 1997; Osicka et al., 2000). This is often confused for a change in GFB permeability for albumin and IgG but in actual fact it is a change in the balance of intact material versus degraded material being excreted; urine assays used for the detection of protein generally only measure the intact material, not the degraded material.

In contrast, nephrotic albuminuria associated with hypoalbuminemia is accompanied by protein fractional clearances that approach the fractional clearances of Ficoll. Partial nephrosis, as manifested as intermediate fractional clearances between the dd-line and n-line are generally not seen for non-diabetic nephrotic states in humans (Figure 2B). The same is observed for nephrotic states in rats and mice (Russo et al., 2002; Comper et al., 2008; Comper et al., 2016). Therefore, the evolution of nephrotic proteinuria/albuminuria is seen as a ‘saltation effect’ rather than the gradual change seen in proteinuria/albuminuria in normo-proteinemic states.

The interesting question arises as to whether there is inhibition of the degradation pathway in nephrotic states. In relative terms there is; there is a marked predominance of intact material excreted over that of degraded material (Greive et al., 2001). Yet, if you consider the degradation in terms of the capacities of the Park and Maack binding sites then there is no inhibition within the error analysis. Greive et al. (Greive et al., 2001) found essentially the same excretion of degradation product from filtered albumin, or transferrin or IgG as measured in healthy rats as compared to nephrotic rats with either anti-GBM disease or puromycin aminonucleoside (PAN) nephrosis.

Not many studies have characterized this. Russo et al. have demonstrated no change in size selectivity, and inhibition of the retrieval pathway occurs in CD2AP KO mice (Russo et al., 2013). Relative inhibition of the degradation pathway is apparent but the excretion rate of degradation products again is not too dissimilar to that observed in healthy mice. These characterizations overlap significantly with those in nephrotic acquired kidney disease.

The enthusiasm for generating a charge selectivity concept was derived from the original micropuncture studies (see below) that suggested that albumin was essentially impermeable to the GFB.

When confounding factors such as cell uptake, metabolic degradation, binding to plasma proteins and other components are eliminated it becomes clear that charged inert transport probes of equal or greater negatively charged valence to albumin are not affected by charge selectivity. This has been demonstrated for dextran sulfates (Comper et al., 1994; Vyas et al., 1996), carboxymethyl Ficolls (Greive et al., 2001; Guimarães et al., 2003) and negatively charged dextran and hydroxy ethyl starch (Schaeffer et al., 2002). These experiments demonstrate unequivocally that glomerular charge selectivity or glomerular electrochemical phenonema do not exist.

Early studies by Brenner’s group (Chang et al., 1975b; Bohrer et al., 1979) established that for dextrans and Ficolls, used in size selectivity fractional clearance studies (radius range ∼20–40 Å), there was essentially no tubular reabsorption of filtered material as measured by micropuncture studies; that changes in fractional clearance could be directly related to changes in transglomerular transport. Interestingly, there has been no published work that has seen any change in transport of these probes in proteinuric/albuminuric states ¹ .

Perhaps the single most influential experiment that has formed the basis of thinking about glomerular ultrafiltration is from the original studies on micropuncture (Wearn and Richards, 1924), performed almost 100 hundred years ago. Wearnand Richards demonstrated that the filtrate was essentially protein free, not just albumin free. There started the dogma in Nephrology that the glomerular filter was essentially impermeable to albumin and other plasma proteins. It also became apparent that the expected low concentrations of albumin in the primary filtrate influenced data selection in later micropuncture and 2-photon studies (see below).

Confidence in the micropuncture technology was enhanced by Brenner’s group (Chang et al., 1975a; Chang et al., 1975b; Bohrer et al., 1979) who demonstrated that the GSC of inert probes like dextran and Ficoll were the same as their fractional clearance demonstrating that their reabsorption by the PTC was negligible. Then something strange happened. Efforts at measuring the GSC for albumin, through micropuncture, were met with difficulty as highly variable results were obtained. The expectation was that the GSC would be low, so when high values were obtained they were discarded on the basis that it must be some kind of contamination from extra-tubular sources (Oken and Flamebaum, 1971) (note that no such contamination or data elimination was ever described for studies on dextran and Ficoll). In spite of all the micropuncture studies performed by Brenner’s group they never reported the results of albumin GSC by micropuncture. Furthermore, Oken and Flamebaum did not find contamination of albumin GSC in nephrotic states (Oken and Flamebaum, 1971).

The micropuncture studies of Tojo and Endou (Tojo and Endou, 1992) exemplify the problem. Basically flow was stopped in a nephron and fluid fractions of proximal tubule filtrate fluid near the GFB were serially collected to yield 4 fractions. The first fraction collected would have given a GSC for albumin of 0.06 (not dissimilar from Ficoll fractional clearance in controls and nephrotics and also albumin fractional clearance in nephrotics, Figure 2). They thought this high value was due to contamination. Tojo and Endou took another 3 fractions of the fluid over a period of time and eventually came to a constant albumin concentration yielding a GSC of 0.0006 which they felt was correct. So the questions arise; where did the initial contaminating albumin ultimately go as it was not seen in the final fraction collected? If it was taken up by proximal tubule cells, does it mean that if they had waited they may not have seen the contaminating albumin? Or were these results simply normal leakage of albumin due to lack of GFB charge and retrieval of albumin over time?

Our interpretation of these data is that the earliest sample taken by Tojo and Endou probably reflected the true GSC of albumin and that over time albumin was being continually removed from the stationary filtrate by both the retrieval pathway and the degradation pathway in the PTC.

Rippe and coworkers (Lund et al., 2003; Rippe et al., 2006; Rippe et al., 2007) used a glomerular uptake technique to determine the GSC of albumin of 0.0006. They used an 8 min infusion into plasma of ¹²⁵I-labelled protein. After infusion, a whole body vascular washout was started with heparinized horse serum via the jugular vein at a rate of 20 ml min⁻¹ for approximately 8 min. The trichloroacetic acid (TCA) precipitable urine and cortical tissue (not the inner medulla) were included in radioactivity measurements. The overall problems with the glomerular uptake technique is that it does not include the non-TCA-precipitable radioactivity in the urine derived from renal degradation of filtered material, does not include the material that is bound to the surface of the PTCs that could be exchanged with unlabelled albumin during the washout, fails to prove the assumption that there was no washout of retrieved albumin in the PTCs during the 8 min washout period and is unable to account for retrieved albumin that had already been returned to the blood supply during the 8 min infusion.

Once it was realized that charge selectivity was not a factor in governing the transglomerular transport of albumin there was a compelling need to measure the albumin GSC directly. We expected the GSC to be high yet when the first two photon measurements of the albumin GSC of 0.03 were published (Russo et al., 2007) there was an outcry in terms of their veracity. It became clear that many issues can affect the measured the two photon GSC; these have been investigated extensively and thoroughly by Sandoval, Molitoris and others (Sandoval et al., 2012; Sandoval et al., 2014). Depending on the settings of the microscope, investigators can get any value of GSC. As it turns out some two photon investigators have chosen to publish the very low values of albumin GSC (Peti-Peterdi, 2009; Schießl et al., 2015; Butler et al., 2019) (no doubt influenced by the micropuncture/charge selectivity dogma in the kidney field) without any qualifying data, particularly the use of transport calibrants of known GSC. The criticism that can then be levelled at all these studies is ‘how do they know they have measured the correct GSC?’ It is absolutely necessary for two photon studies that adequate transport calibrants of known GSC be used to ensure the correct settings on the microscope. The only studies that have done this is by Russo et al. (Russo et al., 2009) and Sandoval et al. (Sandoval et al., 2012) and accordingly they have obtained the relatively high GSCs for albumin. These GSC values have been confirmed by a body of work over the years (Wagner et al., 2016; Sandoval et al., 2019; Molitoris et al., 2022).

These studies do provide what was originally expected when cell-mediated effects are not involved and knowing that charge selectivity is not a factor; the GSC of 36 Å radius Ficoll in healthy control humans = GSC of 36 Å radius Ficoll in nephrotic humans = the GSC of albumin in nephrotic humans = GSC of albumin in healthy controls (rats).

There is an interesting set of studies that induces reversible nephrotic-like albuminuria; that is overload proteinuria as studied in rats (Davies et al., 1978; Bliss and Brewer, 1985; Weening et al., 1987; Lemley, 1993; Koltun and Comper, 2004; Wagner et al., 2016). In overload proteinuria, an artificial small increase in plasma protein (∼20%) induced by iv injection of albumin solution in healthy rats can generate profound and disproportionate increases in the urinary excretion of protein (maximal 7-fold increase in total protein excretion, 200-fold increase in intact albumin excretion (see also (Wagner et al., 2016)) and an increase in the excretion of other plasma proteins such as transferrin (Koltun and Comper, 2004)) without any significant change in size selectivity (Koltun and Comper, 2004) or GSC for albumin (Wagner et al., 2016). ‘Nephrotic-like’ albuminuria and proteinuria can occur by simply increasing the plasma albumin concentration by a very small amount. The rat removes this excess albumin/protein in less than 3 days. These results have been interpreted in the following way; these levels of excretion would never occur megalin/cubilin mediated degradation pathway but rather through a saturable receptor (retrieval pathway) that is unable to fully accommodate the extra filtered albumin due to its handling of endogenous filtered albumin; the albumin not taken up by the receptor is excreted into the urine. These types of studies reveal the finely tuned balance between albumin filtration rate, proximal tubule uptake and the concentration of filtered material; in healthy controls it is basically operating at maximal capacity and any further perturbation tips the system into an overflow situation where the albumin is readily excreted. In the Koltun and Comper (Koltun and Comper, 2004) study, an average increment in plasma protein content of ∼10 mg ml⁻¹ (day 3), due to albumin overload, yielded an increment of ∼360 mg protein day⁻¹ in excretion rate which corresponds to a protein fractional clearance of ∼0.02. In a system where glomerular permeability is essentially unaltered and with endocytotic mechanism by PTCs being nearly saturated but with normal fragmentation still occurring, the results imply that the GSC is at least an average of 0.02 for all glomerular-filtered protein, the majority of which is likely to be albumin. Again, this is in agreement with the two photon studies described above. It is notable that overload proteinuria would never occur if the only tubular uptake was the low capacity degradation pathway associated with a very low GSC.

The immediate albumin filtrate concentration in the rat will be ∼1 mg ml⁻¹ yet the results in Figure 1 would suggest that at this concentration the high capacity receptor (albeit for rabbit tubules) is nowhere near saturation. The reason is that the real protein concentration in vivo is going to be considerably higher due to the presence of non-albumin plasma proteins that are also filtered and retrieved. If we assume that albumin is approximately 50% of the total protein present then we should be looking at P_ALB near 2 mg ml⁻¹ (assuming albumin is just like other plasma proteins here) where the receptor is far closer to saturation. The transition from normal filtration in vivo to overload/saturability is very tight and much tighter than exemplified by the line of best fit of the upper inflexion of the uptake curve shown in Figure 1.

The functional interplay of the retrieval and degradation pathways has still to be worked out. We have noted, however, that the tertiary structure of albumin, beyond simple size and charge, is a critical determinant for albumin processing by the retrieval pathway by the kidney (Clavant and Comper, 2003). It suggests that a specific albumin-recognition event by the kidneys is critical to normal renal handling of albumin otherwise the molecule is not retrieved but is disposed of by the degradation pathway.

Table 3 summarizes the albumin GSC data as determined in rats that we judge as valid. All the techniques give similar values in the range of 0.02–0.06. These are the values expected when size selectivity alone is the primary biophysical factor governing transglomerular transport and where charge selectivity is non-existent. The relatively high GSC values are entirely consistent with the original observations by Ryan and Karnovsky (Ryan and Karnovsky, 1976) that albumin can transverse the GFB relatively quickly.

When Park and Maack (Park and Maack, 1984) perfused tubules at the low albumin concentration of 0.03 mg ml⁻¹ they determined that the majority of the degradation products were observed in the bathing solution. Many investigators subsequently interpreted these studies that degradation products produced in vivo would be expected to be trancytosed in the PTC and be destined for the circulation. However, we have never seen this in vivo (Osicka et al., 1997; Osicka et al., 2000; Russo et al., 2002; Comper et al., 2008; Vuchkova and Comper, 2015); the degradation products are exclusively recycled back to the tubular lumen and excreted. There is the likelihood that the vectorial fate of the degradation products is dependent on the albumin perfusate concentration; when the perfusate concentration was relatively high at 2.2 mg ml⁻¹ we did not see those products on the baso-lateral side of the perfused tubule (Zhou et al., 2013).

The results in Figure 2 demonstrate that the fractional clearance of proteins as measured in healthy human subjects increases 10,000–100,000-fold when studied in nephrotic patients. It is remarkable then that genetic perturbation of the GFB can have such a biophysically profound effect (as commonly proposed in the literature) when there is absolutely no clear explanation as to what those biophysical changes are (Kestilä et al., 1998; Tryggvason, 1999; Funk et al., 2020; Benzing and Salant, 2021). The effect of genetic perturbation on glomerular filtration was initially elaborated by Tryggvason et al. (Kestilä et al., 1998; Tryggvason, 1999) when they identified nephrin mutations in the slit diaphragm as being responsible for congenital nephrotic syndrome. There have been many studies since (Tryggvason, 1999; Benzing and Salant, 2021) that have identified various genetic perturbations of the GFB giving rise to altered permeability and even nephrotic proteinuria; how these occur has never been seriously addressed. Tryggvason (Tryggvason, 1999) invoke the ‘importance of size, shape and charge selective properties’ of the GFB in determining the passage of plasma macromolecules whereas Benzing et al. (Benzing and Salant, 2021) invokes additional glomerular centric biomechanical effects on the GFB. As we discussed earlier, none of these speculations quantitatively account for differences in the fractional clearance of plasma proteins from healthy controls to nephrotics as shown in Figure 2. There is no doubt that the etiology of the proteinuria can directly related to the occurrence of the perturbation associated with the podocyte or GFB but there is no evidence that the perturbation itself is directly responsible for altered transglomerular transport properties giving rise to nephrotic proteinuria.

In speculating the initial events leading to nephrotic proteinuria, we could easily envisage a factor(s) being produced by stressed podocytes ³ (through genetic perturbation or other means) that inhibit the retrieval pathway downstream. The nature of the factors is not known.

Alternatively we could consider an overload proteinuria effect. The overload effect we described above does not explain nephrotic proteinuria. We could turn the thinking around, however, and say that if there is a very small change in the exclusion properties of the GFB and that the output of albumin filtration across the GFB is increased by 10–20% then this would mimic overload proteinuria (Russo et al., 2013; Comper and Vuchkova, 2021). The changes could conceivably be a result of podocyte stress or deletions in proteins of the GFB altering the transport properties of the extracellular matrix milieu. They may be too small to accurately measure directly in terms of GSC or fractional clearance. Yet these small changes would be amplified in terms of protein excretion in a synonymous way as in overload proteinuria. This mechanism sets the basis for understanding how a ‘podocyte related event’ can have a disproportionate influence on albumin/protein excretion determined by the PTCs. There is also an important biomarker that establishes the common phenomenology of overload proteinuria and nephrotic proteinuria; the lack of inhibition of the degradation pathway. This is exactly what occurs in overload proteinuria where there is a saturation of the retrieval receptor but the degradation pathway is still operative (Koltun and Comper, 2004).

It is interesting to speculate how a small increase in GFB permeability can ultimately generate hypoalbuminemia ⁴ . In both healthy controls and nephrotics the plasma level of albumin will be the net result of albumin synthesis, albumin plasma elimination, kidney retrieval and uptake by proximal tubules. It is assumed that optimal residence in the circulation would occur at the K_m value for the retrieval pathway in Figure 1. Any concentration above that corresponding to the K_m value may result in some albumin loss into the urine ⁵ . Therefore, an increase in GFB permeability will have to find a new steady state when the different amount of hypoalbuminemic albumin is presented to the proximal tubule receptors. In this case, further albumin loss will occur and ultimately result in inhibition of the retrieval pathway and maximal hypoalbuminemia.

We view the initial etiology of the nephrotic state and particularly in nephrotic podocytopathies is a small aberration in the permeability of the GFB. This will result in an overload effect on the PTC with subsequent loss of filtered protein including albumin into the urine. Hypoproteinemia will be set in play. This evolution of nephrosis is probably never measured because, as noted earlier, partial nephrosis is rarely if ever observed. Rather what is seen is the end point of these overload effects where the fractional clearances of proteins particularly albumin and IgG, approach those of their Ficoll counterparts of the same radius and hypoproteinemia has reduced the plasma protein concentration by two-thirds ensuring that overload effects on PTC uptake will not occur (Figure 1). The reason for the release of nephrotic levels of protein into the urine, at this stage, is that the retrieval pathway now appears to be completely inhibited which is consistent with the equivalence of the plasma elimination of proteins with their Ficoll counterparts. The reason for the inhibition is not known although several mechanisms can be envisaged; podocyte factors may be released and operate downstream on the proximal tubular cell to turn off the retrieval pathway or it may be that the retrieval pathway is inhibited by virtue of the hypoproteinemic state. Future delineation of these various mechanisms will determine the primary vascular/glomerular/tubular event that ultimately determines the nephrotic condition.