In the current study we have determined that podocyte density and apparent glomerular podocyte number are substantially reduced in rats and mice injected with STZ, a model of Type 1 diabetes mellitus, quite early after initiation of diabetes. Reduction in apparent podocyte number and podocyte density occurred as early as 2 weeks after STZ injection, and appeared to worsen somewhat by 6 and even further by 8 weeks after STZ administration. Importantly, STZ did not appear to have an acute toxic effect on podocytes, at least in the murine model, since 3 days after completion of the five day STZ injection protocol apparent podocyte number and density was completely normal in STZ treated mice. Also significantly, insulin treatment had only a modest effect in maintaining podocyte number, despite inducing a substantial improvement in fasting blood glucose levels by 2 weeks after initiation of diabetes. In contrast, administration of the antioxidant, α-lipoic acid was more effective in maintaining podocyte number and density. These latter data suggest that near normalization of glucose levels, even as early as 2 weeks after initiation of STZ-diabetes, cannot fully prevent the reduction in podocyte number, possibly because the early effects of hyperglycemia are dominant in this model. In addition, the protection with α-lipoic acid suggests that oxidant-mediated injury plays a role in the early diminution in podocyte density consistent with other effects of oxidants in mediating or enhancing diabetic complications [e.g., [12
Our methods of podocyte counting relied on accurately identifying podocyte nuclei (WT-1) or cytoplasm (GLEPP1). The WT-1 staining tends to be more precise since it is difficult to distinguish the cytoplasmic extent of any single podocyte using these methods. As noted by these investigators, a potential disadvantage of this method is that podocyte nuclei are not uniformly WT-1 positive under some pathologic circumstances [10
]. As noted by these investigators, a potentially disadvantage of this method is that podocyte nuclei are not uniformly WT-1 positive under some pathologic circumstances ([10
]). However, as they also noted, there are no data to indicate that podocytes in DN are WT-1 negative. In addition, WT-1 is an excellent marker of the normal mature differentiated podocyte and even podocytes with effaced foot processes remain WT-1 positive. Hence, if a podocyte is WT-1 negative, then it is likely to be severely and perhaps terminally damaged. Thus, we believe that our data are consistent with true and quite early reductions in podocyte density in experimental diabetic nephropathy and at the least reflect severe podocyte damage.
Although the morphometric methods utilized in this study are somewhat different from those developed by Weibel and Gomez and Cavalieri, they clearly assess very similar parameters and have been well validated in the previous publications [5
]. Nonetheless, since we did not survey glomerular serial sections in these studies, nor investigate both thick and thin sections of the same glomerulus, we cannot claim that we have demonstrated an absolute reduction in numbers of podocytes/glomerulus. The increase in glomerular size in diabetes may have contributed to the reduction in podocytes/glomerular section. However, it is clear that glomerular density was substantially reduced and the glomerular volume subsumed by each podocyte was increased in these Type 1 diabetic models. Although there appears to be a great deal of precision using these methods, as noted by the small variances in each study, there was a substantially greater reduction in apparent podocyte number after 8 weeks than after 6 weeks of STZ-diabetes (34 vs. 18 %). Whether this represents true continued podocyte loss over the intervening time period is uncertain, and would need to be validated by a larger single trial in which all animals were made diabetic at the same time followed by assessment of podocyte loss in samples at regular intervals.
Previously, several studies in Type 1 and Type 2 diabetic patients have suggested that diabetes leads to a reduction in glomerular podocyte number as a relatively early marker of diabetic nephropathy [2
]. Moreover, Meyer and colleagues showed that the number of podocytes per glomerulus at the start of the study was predictive of the degree of albuminuria after a four year follow-up, suggesting that podocyte loss precedes and predicts progressive diabetic nephropathy [3
]. More recently, in a study of Italian patients with Type 2 diabetes mellitus, substantial abnormalities in podocyte structure and density were found to occur during the early stages of diabetic nephropathy [13
]. Specifically, glomerular podocyte density declined early in diabetes and correlated inversely with albumin excretion rates, although the number of podocytes per glomerulus did not correlate with progression of disease. These findings prompted the authors to conclude that the density, not the absolute number, of podocytes may be important for the progression of nephropathy in Type 2 patients. In a report by Steffes et al., for the International Diabetic Nephropathy Study Group, Type 1 diabetic patients were found to have a significantly lower number of podocytes than age-matched control subjects [6
]. A more recent study has also shown that podocyte loss accompanies Type 1 diabetes but that this podocyte loss did not appear to occur as early as in Type 2 patients [7
]. Although it reported that podocyte number was lower in diabetic patients than in nondiabetic controls and that the surface covered by each podocyte increased gradually with time, the absolute number of podocytes did not predict the degree of albuminuria. Nonetheless, in the longitudinal component of the study, there was a significant correlation between the decrease in podocyte number and the degree of albuminuria over the period from the baseline observation to follow-up three years later [7
]. From all of these human studies, it remains impossible to conclude whether decreased podocyte number or density actually results in increased albuminuria and progressive nephropathy or whether the podocyte changes and albuminuria are simply associated findings.
Since the human studies cannot ascertain whether the decrease in podocyte number causes or simply correlates with advancing albuminuria and nephropathy, it would be beneficial to identify animal models that recapitulate similar changes. Such models would aid the investigation of the mechanisms by which diabetes results in podocyte loss and allow determination of whether podocyte loss directly results in albuminuria. The early podocyte loss that occurs in humans with diabetic nephropathy appears to occur quite early in both rat and mouse STZ diabetes. While the changes are more profound in the rat models, similar although less substantial reductions in podocyte density occur in mice after only 2 weeks of STZ diabetes. Although the current study is the first to show that experimental murine and rat models of diabetes undergo very early podocyte loss, several previous studies using experimental models of diabetes have demonstrated podocyte loss or abnormalities after longer periods of diabetes. Gross and colleagues have shown that similar reduction in podocytes occurs after 6 months of streptozotocin (STZ) diabetes in rats [14
] and that loss of podocytes in this model was prevented by treatment with the angiotensin converting enzyme inhibitor, trandolopril. In a separate study, these investigators also showed substantial podocyte loss after 6 months of diabetes in the SHR/N-cp rat, a model of type II diabetes that spontaneously develops pronounced abnormalities in renal histology. In comparison to STZ-diabetic rats, which develop relatively modest glomerular changes at 6 months of diabetes, glomeruli from the SHR/Ncp rats contained fewer and larger podocytes, smaller mesangial cells and a more expanded mesangial matrix [15
Mifsud and colleagues did not examine podocyte number but documented that the number of slit pores per unit length of glomerular basement membrane in rat glomeruli was decreased 24 weeks after STZ injection, consistent with podocyte foot process broadening [9
]. These changes, as well as the increased albumin excretion seen in this model, were also ameliorated by treatment with either an angiotensin receptor blocker or an angiotensin converting enzyme inhibitor [9
]. Gassler and colleagues studied nephron degeneration in Zucker fa/fa male rats, a model of Type 2 diabetes, at 10 months of age and concluded that degeneration began with damage to podocytes [16
]. They demonstrated sclerosis in approximately 25% of the diabetic glomeruli and found evidence of more extensive, "pre-sclerotic" podocyte injury including foot process effacement, pseudocyst formation, and accumulation of lysosomal granules and lipid droplets in podocyte cytoplasm. These podocyte changes appeared to play a significant role in the progression of the segmental glomerular injury to global sclerosis as well as to the degeneration of the corresponding tubule. Hoshi and colleagues also found electron microscopic evidence of podocyte degeneration and the development of tuft adhesions that they concluded were responsible for the glomerular sclerosis in the same model [8
In addition, several studies have found that diabetes was associated with a significant reduction in expression of the podocyte slit diaphragm protein, nephrin, in both human [17
] and animal models [18
] of diabetes. However, in the studies in STZ diabetes in rats there was no change in nephrin expression after one week of diabetes [17
]. Since our data indicate that podocyte changes occur quite early in STZ diabetes and is statistically significant as early as 2 weeks after induction of diabetes, it seems likely that the change in nephrin expression was either a response to podocyte injury or stress, or developed independently of the early podocyte changes found in this model.
The biochemical and metabolic signals that result in diabetic glomerulopathy have been the subject of investigation over the past several decades. A unifying factor in promoting most if not all of the abnormalities found in the diabetic glomerulus appears to be the increase in mitochondrial oxidative stress generated by enhanced glucose metabolic flux [19
]. For these studies, we used a potent inhibitor of mitochondrial superoxide generation, α-lipoic acid, which virtually eliminated the effects of STZ-diabetes on reduction in podocyte density and apparent number, supporting the notion that mitochondrial reactive oxygen species are critical in these early changes of diabetic glomerulopathy.
Since STZ also has nephrotoxic effects, especially when given at high doses, it is conceivable that STZ had a direct toxic effect on podocytes and that some of the effect of α-lipoic acid was to protect against such toxicity. While our studies do not absolutely rule this out, we feel this is unlikely for several reasons: 1) other models of diabetic nephropathy that arise spontaneously show similar decreases in podocyte number and density [e.g., [15
]]; 2) relatively low dose STZ as used in these experiments has not been associated with proximal tubule toxicity and the proximal tubule cell should be especially sensitive to STZ since, like the pancreatic β cell, it expresses high levels of the facilitative glucose transporter, GLUT2 which transports STZ into cells [21
], 3) finally and most importantly, we demonstrated that no changes in podocyte density or apparent number occurred within 3 days after completion of a 5-day low dose STZ protocol in mice, which strongly suggests that STZ has no acute podocyte toxicity.
Although the early apparent loss of podocytes in the STZ diabetic rat and mouse models implicates podocyte damage in the pathogenesis of diabetic nephropathy, it is also striking that such changes precede albuminuria and glomerulosclerosis by at least several months. Moreover, the STZ rat and mouse models, as is true for most animal models of diabetic complications, fail to develop classic changes of human diabetic nephropathy. Thus it seems likely that the early podocyte changes, though important, may not be sufficient to lead to progressive diabetic nephropathy and renal failure. It may be that a certain threshold of podocyte damage must be achieved before sclerosis occurs, as suggested by Kim et al. [5
]. Conversely, nephrosclerosis may be dependent on mesangial and other factors that are not directly affected by podocyte changes, as suggested by many studies over the past 20 years (for an overview, see reference 22
). Nonetheless, the sensitivity of the podocyte to hyperglycemic damage is now evident. These early changes in podocytes may lead to long-term glomerular responses that are critical for the development of diabetic nephropathy.