Three-dimensional electron microscopy reveals the evolution of glomerular barrier injury

Glomeruli are highly sophisticated filters and glomerular disease is the leading cause of kidney failure. Morphological change in glomerular podocytes and the underlying basement membrane are frequently observed in disease, irrespective of the underlying molecular etiology. Standard electron microscopy techniques have enabled the identification and classification of glomerular diseases based on two-dimensional information, however complex three-dimensional ultrastructural relationships between cells and their extracellular matrix cannot be easily resolved with this approach. We employed serial block face-scanning electron microscopy to investigate Alport syndrome, the commonest monogenic glomerular disease, and compared findings to other genetic mouse models of glomerular disease (Myo1e−/−, Ptpro−/−). These analyses revealed the evolution of basement membrane and cellular defects through the progression of glomerular injury. Specifically we identified sub-podocyte expansions of the basement membrane with both cellular and matrix gene defects and found a corresponding reduction in podocyte foot process number. Furthermore, we discovered novel podocyte protrusions invading into the glomerular basement membrane in disease and these occurred frequently in expanded regions of basement membrane. These findings provide new insights into mechanisms of glomerular barrier dysfunction and suggest that common cell-matrix-adhesion pathways are involved in the progression of disease regardless of the primary insult.

adult glomerulus the GBM is composed of predominately type IV collagen α 3α 4α 5, laminin α 5β 2γ 1 and up to 70 additional structural and regulatory components 10 . Evidence suggests that type IV collagen α 3α 4α 5 is important for the lifelong structural integrity of the GBM 11 , as the type IV collagen α 1α 1α 2 network, produced by both podocytes and glomerular endothelial cells 12 , is unable to compensate for the absence of the α 3α 4α 5 network in Alport Syndrome 13,14 .
Renal biopsy is a valuable diagnostic tool for defining the sub-type of glomerular disease; indeed a diagnostic marker for Alport Syndrome is the pathological basket weave appearance of the GBM, although genetic testing is now superseding biopsy 15 . Other histopathological features of glomerular disease include altered GBM thickness, loss of unique podocyte structures called foot processes (FPs) and a reduction in podocyte number 16 . Interestingly, these features are reported across the spectrum of glomerular disease regardless of the underlying molecular aetiology. Yet the mechanisms underpinning these morphological changes are poorly understood. In particular we have not defined the ultrastructural relationship between podocytes and the GBM during the progression of glomerular disease, however this knowledge could provide insight into disease mechanisms.
To address this knowledge gap we employed serial block face-scanning electron microscopy (SBF-SEM) to examine the relationships of cells and ECM in three-dimensions (3D). This approach was recently used to investigate cell-ECM interactions [17][18][19][20] and the complexities of normal podocyte structure 21 . However, until now, the approach has not been used to investigate changes in podocytes and the GBM during disease progression. We hypothesized that SBF-SEM over a time course would enable the construction of detailed 3D models of the glomerulus in health and disease and would reveal novel pathological features.

Results
3D modelling of glomerular structure in Alport nephropathy. Over a 28-week time course we investigated Col4a3−/− male mice on a C57BL/6J genetic background. We selected this background, which has a slow progression of glomerular disease, to allow a greater window of observation. We analyzed glomeruli using SBF-SEM and created 3D glomerular reconstructions (Supplementary Movie S1). The glomerulus is composed of interwoven capillary loops forming a complex pattern of lumens when viewed in 3D. The localization of the GBM, podocyte cell bodies and foot processes (FPs) can be easily discerned using SBF-SEM (Fig. 1a). From wild type [16][17][18] week (older adult) mouse glomeruli we generated models of podocytes and GBM and these revealed an organized glomerular structure (Fig. 1a,b). We compared these models with glomeruli from 6-week (young adult) Col4a3+ /− and Col4a3−/− animals and found that the majority of podocytes had regular FPs (Fig. 2). This finding was also observed in older adult Col4a3−/− mice although some FPs were flattened or effaced (Fig. 2). In contrast, 28-week (aged) Col4a3−/− mice had striking global loss of podocyte FP organization, GBM thickening and thinning (Fig. 2). Further analysis revealed focal regions of reduced podocyte FP density in all groups of Col4a3−/− mice, even in young adult mice, when compared with age matched wild type and Col4a3+ /− mice (Fig. 3a,b). These areas were often concomitant with thickened, non-uniform sheets of GBM (Fig. 3a,c and Supplementary Movies S2 and S3). Moreover, the density of podocyte FPs decreased with age in Col4a3−/− mice (Fig. 3b) and correlated with increased thickness and variation in thickness of the GBM (Fig. 3c).
Podocytes invade into the GBM in Alport nephropathy. The appearance of cellular material within the GBM has been previously described. Erythrocytes have occasionally been captured traversing the GBM in thin basement membrane nephropathy 22 . Podocyte infolding in the GBM has also been described in a range of glomerular pathologies 23 and more recent investigation of Col4a3−/− mice, suggested that cellular interposition in the GBM is of mesangial cell origin 24 . Mesangial cells are located between adjacent capillary loops and have contact with the GBM at the bases of the capillary loops ( Supplementary Fig. S1a). In contrast, podocytes directly adhere to the GBM of the capillary walls ( Fig. 1a,b). Using SBF-SEM we were able to identify cellular invasions within the GBM that connected to podocyte FPs (Fig. 4a,b and Supplementary Movies S4 and S5). These invasions were rarely observed in young adult Col4a3−/− mice (Fig. 5a) and were never observed in Col4a3+ /− or wild type mice. In contrast, older adult and aged Col4a3−/− mice had frequent podocyte invasions (Figs 4c and 5b-d). The lengths of podocyte invasions were variable, but the mean length increased with age (Fig. 5e). Moreover, they occurred more frequently in thickened and abnormal areas of GBM in Col4a3−/− mice (Figs 4c and 5f). In aged Col4a3−/− mice the entire GBM had irregular thickness and all regions of the GBM contained podocyte invasions (Fig. 5d,f). Since it is known that the genetic background of Alport mice has an effect on the rate of disease progression 25 , we also investigated Col4a3−/− mice on a 129S1/Svlmj background with SBF-SEM. Here, similarly, we found evidence of podocytes invading into the GBM (Supplementary Fig. S2a-e).
Podocyte-GBM invasion is a common feature in glomerular disease. To investigate whether podocyte invasion into the GBM is a unique feature of Alport Syndrome, or shared with other genetic glomerular diseases, we investigated two additional mouse models of human disease. MYO1E mutations in humans cause steroid resistant focal segmental glomerulosclerosis (FSGS) 26 . MYO1E encodes myosin1e, a class I myosin which is expressed by mouse and human podocytes 26,27 . Myo1e−/− mice develop FSGS and renal failure associated with morphological abnormalities in both podocytes and the GBM; we therefore investigated adult Myo1e−/− mouse glomeruli with SBF-SEM. Our analysis revealed that podocyte FPs were flattened, broadened and in many regions completely effaced (Fig. 6a-c). There were expanded areas of GBM that coincided with loss of podocyte FP morphology (Fig. 6d). Overall, there was a loss of podocyte FP density and a thickening of the GBM (Fig. 6f,g). Along with these classical pathological changes in podocyte and GBM ultrastructure, we also discovered prevalent podocyte FP invasion into the GBM (Fig. 7a, Supplementary Fig. S3a and Supplementary Movie S6).
Next we investigated a third mouse model of human glomerular disease. Glomerular epithelial protein 1 (GLEPP1) is a receptor tyrosine phosphatase expressed by podocytes. Mutations in PTPRO the gene that encodes GLEPP1 protein causes childhood-onset nephrotic syndrome in humans 28 . Ptpro−/− mice have impaired renal function and SEM has demonstrated that these mice have shortened podocyte FPs 29 . We investigated adult Ptpro−/− mice with SBF-SEM and found dramatic GBM expansions, overall GBM thickening and focal areas of reduced podocyte FP density ( Fig. 6a-d,f,g). Furthermore we identified podocyte invasions into the GBM Mesangial cells invade the GBM at the mesangial aspect. Using 3D modelling of glomerular structure we also analysed the mesangial cells since they have been reported to invade the GBM in Alport syndrome 24 . Indeed, analysing SBF-SEM from Alport mice we identified mesangial invasions of varying length inside the GBM along the mesangial aspect of the glomerular capillary loop (Supplementary Fig. S1b and Supplementary Movies S9 and S10). These mesangial processes were distinct from podocyte invasions; they were only observed close to the mesangial angle and not in the peripheral capillary loops. We also detected mesangial invasions in both Myo1e−/− and Ptpro−/− mice ( Supplementary Fig. S1c,d and Supplementary Movies S10 and S11). Mesangial invasions were rarer than podocyte invasions and in contrast to podocyte invasions, mesangial invasions were also present in wild type animals (Supplementary Movie S8). However, mesangial invasions were longer in mice with Col4a3, Myo1e and Ptpro mutations compared with wild type mice (Fig. S1E, Movie S9-11).
GBM defects resembling podocyte invasion in human Alport syndrome. We interrogated TEM sections from the renal biopsies of two siblings that we have previously reported 30 . The proband was investigated for persistent microscopic haematuria and mild proteinuria and he had a renal biopsy at 3 years of age. This demonstrated abnormal GBM thickness, which prompted COL4 genetic screening. He was found to be hemizygous for two COL4A5 variants: c.2858G > T; p.(Gly953Val) and c.3097G > C; p.(Gly1033 Arg). His younger brother presented at the age of 6 months with steroid resistant nephrotic syndrome and was also hemizygous for the same COL4A5 variants. In addition he carried two homozygous missense variants in MYO1E: c.352A > G; p.(Lys118Glu) and c.2627C > G; p.(Thr876Arg). In both renal biopsies there was evidence of electron lucent regions within the GBM (Fig. 8). In addition, there was evidence of cellular material within the GBM and these resembled the podocyte invasions that we observed in Col4a3−/−, Myo1e−/− and Ptpro−/− mice. Future SBF-SEM studies of human biopsies are needed to confirm that the cellular material is indeed podocyte protrusions as we have shown in the mouse.

Discussion
In this study we used SBF-SEM to analyze spatial relationships between cellular and extracellular glomerular components in health and disease. We identified podocyte invasions into the GBM as a novel morphological feature of glomerular disease. Irregular expansions of the urinary surface of the GBM were also a dominant feature in all disease models and these irregular expansions were more likely to harbor podocyte invasions. Podocyte invasions occurred in Alport (Col4a3−/−), Myo1e−/− and Ptpro−/− mice and lucent regions of GBM, resembling sites of podocyte invasion, were seen in patients with Alport syndrome. Finally, we demonstrate that GBM thickness and irregularity markedly increases and podocyte FP density decreases with glomerular disease progression.  The GBM is a critical component of the glomerular filtration barrier being exquisitely water permeable yet prohibiting the passage of macromolecules. Super resolution microscopy was recently used to describe the precise molecular topology of the GBM 31 . In this nanoscale investigation of the GBM, type IV collagen was shown to be restricted to the centre of the GBM. This localization indicates that the collagen IV network is too distant from podocytes to interact with integrins at the cell surface and therefore initiate outside-in signaling. This paper also demonstrated that type IV collagen α 1α 1α 2 mislocalizes during Alport syndrome and is found adjacent to podocytes. Type IV collagen α 1α 1α 2 that is proximal to podocytes could conceivably perturb cell-ECM interactions or alter the organization of other components within the GBM. In our investigation, we identified novel podocyte protrusions into the GBM and one explanation for podocyte invasion is that increased GBM degradation and subsequent podocyte migration occur in order to improve podocyte anchorage and to prevent detachment. A second, attractive explanation is that podocytes use invasions to remodel GBM damage that is caused by altered filtration forces. Potentially, podocyte invasions are extensions of plasma membrane used to remodel ECM, similar to fibropositors that regulate the direction of collagen fibril assembly during tendon repair 24 .
Our investigation of human GBM was limited to TEM sections since archival EM blocks from human biopsies are not routinely stained with heavy metals prior to embedding. For SBF-SEM staining is required prior to embedding since sectioning is automated within the microscope. In the human TEM we observed lucent regions in the GBM, resembling the regions in the mouse GBM that were tracked to podocyte protrusions and we therefore propose that the same protrusions occur in human glomerular disease. These observations are likely to overlap with the previously reported podocyte infolding in the GBM, seen across a range of glomerular pathologies 23 .
Defining the molecular composition and nature of podocyte invasions into the GBM would determine whether these could be therapeutically blocked or activated in order to protect glomerular function. There are a number of candidate pathways that could drive podocyte invasion into the GBM. One such pathway is integrin mediated adhesion signalling. Focal adhesion kinase (FAK), a key integrin signalling protein, has been shown to be activated in Alport mouse podocytes by ectopic laminin α 2 32 . This same study revealed that FAK induces matrix metalloproteinase (MMP) release from podocytes. This proposes an explanation for podocyte protrusion into the GBM following podocyte-mediated ECM degradation. Indeed TEM imaging provides evidence of an  electro-lucent halo around FP invasions ( Figure S4), which could be the result of active ECM degradation by podocytes.
In conclusion, SBF-SEM has provided enhanced visualization of the glomerulus and this has led to the identification of previously unseen pathological features of glomerular disease. 3D ultrastructural imaging revealed expected GBM abnormalities in addition to podocyte protrusions, resembling invadosomes, in the GBM. This novel finding provides new insight into disease mechanisms and suggests that a common matrix-adhesion pathway is activated regardless of the primary molecular insult. What remains unknown is whether podocytes actively migrate into the GBM or if these protrusions occur passively. Further study is necessary to understand how and why podocytes protrude into the GBM. Answers to these questions could lead to the development of targeted therapies that would offer new therapeutic avenues for glomerular disease. Serial block face-scanning electron microscopy. Kidney samples were prepared for SBF-SEM as described previously 17 . In brief, mouse kidneys were cut into 1 mm cubes and fixed in situ by using 2% (wt/vol) glutaraldehyde (Agar Scientific, UK) in 0.1 M cacodylate buffer (pH 7.2); stained in 1% (wt/vol) osmium tetroxide, 1.5% (wt/vol) potassium ferrocyanide in 0.1 M cacodylate buffer, followed by 1% (wt/vol) thiocarbohydrazide. After washing, more osmium was added by staining in 1% (wt/vol) osmium tetroxide, and soaked in 1% (wt/vol) uranyl acetate overnight. The final staining step incubation was performed at 60 °C with lead aspartate pH 5.5 for one hour. Samples were dehydrated in ethanol and infiltrated in TAAB 812 hard resin. Tissue was mounted onto an aluminium cryo pin (Electron Microscopy Sciences, cat. no. 70446) using cyanoacrylate glue and all block surfaces trimmed at 90° using a glass knife or diamond trimming tool (Diatome). A gold coating was applied to the block to create a conductive surface. The block was placed in the Quanta 250 FEG (FEI Company) + Gatan 3view system and a 41 μ m × 41 μ m field of view was chosen and imaged by using a 4096 × 4096 scan, which gave an approximate pixel size of 10 nm. The section thickness was set to 50 nm in the Z (cutting) direction.

Methods
Transmission electron microscopy and quantification. Sample fixation and staining was performed as for SBF-SEM. Different blocks from the same sample were used for TEM and SBF-SEM. Tissue blocks were removed from their moulds and sectioned (70-80 nm thickness). Sections were examined using a FEI Tecnai 12G2 Biotwin transmission electron microscope. Magnification varied from 145X to 6800X. n = 6 glomeruli were imaged for each mouse model. Images were saved as original.dm3 files and.tiff files. The dm3 files were opened using Fiji/ImageJ software version 2.0.0-rc-29/1.49r. Collectively 60 randomly selected points of the GBM across 50.dm3 images were measured to assess GBM thickness. The straight-line option was used to draw across the width of the GBM at the chosen point > Analyse > Measure. To quantify podocyte FP number and FP protrusions 50 randomly selected areas of GBM 5 μ m in length were measured as above and FP or protrusion number counted manually. Where protrusions were identified the thickness of the infiltrated GBM was also measured. All measurements were imported into GraphPad Prism software version 5.0 for interpretation and graphical representation.
3D modeling. Typically, Z volumes datasets comprised 1000 images (50 μ m z depth). The IMOD beta version 4.7.11 suite of image analysis software was used to build image stacks, reduce imaging noise, and generate 3D reconstructions 35 . A shell script was run to float images to a common density range, following which they were converted to a size 8 bit smoothed data MRC file as reported previously 17 . A 24″ Wacom tablet with interactive pen tool was used to construct 3D models from the MRC file open in IMOD. The image stack was scrolled through to locate a feature of interest e.g. the GBM. The pen tool was positioned over this feature and a contour drawn around the entire structure. The image stack was moved up or down by one section and a second contour was drawn around the same feature. This process was repeated on every image for the desired number of sections. Statistical analysis. 170 measurements of podocyte FP density and 70 measurements of GBM thickness were taken per group. All identified invasions were quantified and invasion number was normalised to the total volume of glomerulus analysed. Comparisons between individual disease models and wild type were performed in GraphPad Prism using one way ANOVA with a Tukey's post-test, p values less than 0.05 were considered to be significant.