Endothelial cell-specific collagen type IV-α₃ expression does not rescue Alport syndrome in Col4a3^−/− mice

Abstract

The glomerular basement membrane (GBM) is a critical component of the kidney’s blood filtration barrier. Alport syndrome, a hereditary disease leading to kidney failure, is caused by the loss or dysfunction of the GBM’s major collagen type IV (COL4) isoform α₃α₄α₅. The constituent COL4 α-chains assemble into heterotrimers in the endoplasmic reticulum before secretion into the extracellular space. If any one of the α₃-, α₄-, or α₅-chains is lost due to mutation of one of the genes, then the entire heterotrimer is lost. Patients with Alport syndrome typically have mutations in the X-linked COL4A5 gene or uncommonly have the autosomal recessive form of the disease due to COL4A3 or COL4A4 mutations. Treatment for Alport syndrome is currently limited to angiotensin-converting enzyme inhibition or angiotensin receptor blockers. Experimental approaches in Alport mice have demonstrated that induced expression of COL4A3, either widely or specifically in podocytes of Col4a3^−/− mice, can abrogate disease progression even after establishment of the abnormal GBM. While targeting podocytes in vivo for gene therapy is a significant challenge, the more accessible glomerular endothelium could be amenable for mutant gene repair. In the present study, we expressed COL4A3 in Col4a3^−/− Alport mice using an endothelial cell-specific inducible transgenic system, but collagen-α₃α₄α₅(IV) was not detected in the GBM or elsewhere, and the Alport phenotype was not rescued. Our results suggest that endothelial cells do not express the Col4a3/a4/a5 genes and should not be viewed as a target for gene therapy.

INTRODUCTION

Alport syndrome, which is characterized by progressive kidney disease, deafness, and ocular abnormalities (9, 15), is the second most common monogenic cause of renal failure. The disease results from loss of heterotrimeric collagen type IV (COL4) “protomers” composed of COL4 α₃-, α₄-, and α₅-chains . Because COL4 is functional only as a heterotrimer (2), the loss of any one of the α₃-, α₄-, or α₅-chains is sufficient to cause the full Alport phenotype. Many missense mutations and in-frame deletions also cause Alport syndrome, and the phenotype can vary in severity depending on the degree of protomer assembly, secretion, and function (30). It has recently been appreciated that patients with thin basement membrane nephropathy heterozygous for COL4A3 or COL4A4 mutations (16, 30) or women that are heterozygous for X-linked COL4A5 mutations (28) can also develop progressive renal disease. Additionally, many patients diagnosed with focal segmental glomerulosclerosis harbor heterozygous COL4 mutations (21, 26). Other recent reports have demonstrated that clinically silent mutations in other genes (NPHS2/podocin, LAMB2/laminin-β₂, and LAMA5/laminin-α₅) can influence the renal phenotype of Alport syndrome and thin basement membrane nephropathy in patients and in mouse models (6, 35, 36). These data and the estimation that ~1% of the world’s population likely has thin basement membrane nephropathy (32) due to a mutation in COL4A3, COL4A4, or COL4A5 indicate that a preventive or curative therapy for Alport syndrome could alleviate a significant burden of renal disease.

The primary pathology of the renal disease aspect of Alport syndrome is localized to the glomeruli, which are capillary tufts with an outer parietal epithelium and intervening Bowman’s space. The tuft is a network of fenestrated endothelial cell (EC)-lined capillaries anchored by mesangial cells and covered by specialized epithelial cells called podocytes that reside in Bowman’s space (33). A glomerular basement membrane (GBM) constructed by and secreted between the ECs and podocytes is thought to impart size-selective filtration (5) as well as provide attachment sites and signals to ECs, podocytes, and mesangial cells (7). Plasma components pass through the fenestrated endothelium, GBM, and specialized cell-cell junctions between podocytes called slit diaphragms to form the ultrafiltrate.

All basement membranes are built with four major constituents, including laminin, COL4, nidogen, and heparan sulfate proteoglycan (HSPG). Superresolution microscopy of mouse and human GBMs revealed dual layers of laminin-521 and the HSPG agrin at the podocyte and endothelial aspects of the GBM that are separated by a central network of COL4A3/A4/A5 and nidogen (34). A thin layer of COL4A1/A2/A1 localizes to the endothelial side (34), as previously described (17). The Alport phenotype typically begins with hematuria in early to late childhood followed by progressive proteinuria concomitant with glomerulosclerosis and decreasing filtration function (7). The highly variable progression of Alport phenotypes, even between siblings with the same mutation, has long been a mystery that may be partially solved by the recent discovery of genetic modifiers (16, 30).

Like most kidney diseases, the current therapy for Alport syndrome treats secondary effects of the primary molecular problem. The reduction of blood pressure with angiotensin-converting enzyme inhibitor or angiotensin receptor blocker administration is the current standard of care (31). Reduced glomerular pressure and filtration are thought to reduce plasma albumin leak through the glomerular filter; studies in Alport mice have suggested that filtered albumin can injure nephron epithelial cells and promote interstitial disease (13). Inhibition of endothelin A with sitaxentan in Alport mice reduced proteinuria and glomerular and interstitial fibrosis and increased lifespan (3). Inhibition of the transforming growth factor-β target miR-21 also showed significant positive effects on proteinuria and lifespan in Alport mice (8).

Although targeting the secondary effects of COL4 mutations is somewhat effective, this approach does not impact the primary GBM compositional defect. We have previously shown that induced podocyte-specific COL4A3 expression in Col4a3^−/− mice could rescue or limit the Alport phenotype depending on the age at induction (20). This served as a proof-of-principle demonstration that the abnormal GBM could be reconstituted with functional COL4A3/A4/A5 protomers and validates pursuing a gene therapy approach for podocytes. An untested but perhaps more practical gene therapy strategy would be to repair pathogenic COL4 mutations in ECs to promote secretion of COL4A3/A4/A5 protomers into the GBM.

Based on reported COL4 expression profiles, the feasibility of rescuing Alport syndrome via ECs is unclear. Immunogold labeling and in situ hybridization studies have indicated that healthy ECs and mesangial cells typically express α₁ and α₂, whereas podocytes express α₃/α₄/α₅ (1, 12, 18, 24). A kidney chimera study (1) in vivo showed that COL4A3/A4/A5 trimers are expressed by developing and mature podocytes but not by ECs. In the human X-linked Alport kidney, ECs and mesangial cells appear to upregulate α₁ and α₂, but there is no evidence of α₃/α₄/α₅ expression in glomerular ECs (12). Although these published data collectively indicate that glomerular ECs do not express COL4A3, COL4A4, or COL4A5 isoforms in healthy or Alport syndrome contexts, weak expression in the latter case in response to glomerular injury cannot be ruled out.

Forced expression of COL4 isoforms in ECs has never been attempted, and the regulation of COL4 isoform expression is not well understood. Thus, to explore an untested avenue to treat Alport syndrome, we investigated whether ECs could activate COL4A3/A4/A5 expression under circumstances of forced expression of one isoform in the context of disease. We induced EC expression of Col4a3 cDNA in Col4a3^−/− mice to determine whether this would promote the production and secretion of COL4A3/A4/A5 protomers that would polymerize in the GBM and rescue Alport syndrome.

MATERIALS AND METHODS

Mice.

All animal experiments conformed with the National Institutes of Health Guide for the Care and Use of Laboratory Animals and were approved by the Washington University Institutional Animal Care and Use Committee. Generation of the Col4a3-null allele and TetO₇-Col4a3 cDNA transgene was as previously described (20, 23). Tie2-Cre, ROSA26 floxed-Stop rtTA-IRES-EGFP (R26-rtTA), and mTmG (25) transgenic (Tg) mice were purchased from The Jackson Laboratory. The Tie2-Cre transgene was transmitted through male mice due to confounding Cre expression in the female germline. The mice studied were composed of a mixed, primarily C57BL/6J and CBA/J strain background. Pups were treated with doxycycline by providing their mothers with chow containing 0.15% doxycycline beginning at postnatal days 0−7 (P0−P7) and maintained thereafter to induce reverse tetracycline-controlled transactivator (rtTA) activity. Three Col4a3^−/− Tg mice from three different litters paired with Col4a3^+/− control mice also fed doxycycline were euthanized at the indicated ages.

Polymerase chain reaction.

Wild-type Col4a3 and Col4a3-null alleles and R26-rtTA and TetO₇-Col4a3 transgenes were genotyped with primers as previously described (20). The Tie2-Cre transgene was identified with primers that amplify Cre. RNA was isolated from whole kidney lysates preserved in TRIzol reagent and then treated with DNase I and further purified with the RNeasy micro kit (Qiagen). cDNA was generated with Superscript III reverse transcriptase (Invitrogen). Col4a3 cDNA was detected with a sense primer (5′-GTACTCTGGGCAGCTGCCTGC-3′) complementary to a sequence within exon 49 that is deleted in the Col4a3-null allele, and an antisense primer (5′-AATGAAAGAAAAACCTTTCCAGAG-3′) from exon 50 to produce a 290-bp band.

Histology and immunohistochemistry.

Kidneys were harvested at the indicated times and then fixed in 4% neutral buffered formalin overnight for paraffin embedding and sectioning or immediately frozen in OCT medium for frozen sectioning and immunofluorescence staining. For histology, 5-μm-thick sections were stained with periodic acid-Schiff reagent by the Washington University Pulmonary Morphology Core. For immunofluorescence imaging, fresh-frozen tissues were sectioned at 7 μm and stained with rat anti-COL4A3 (clone H31) (11, 29), rabbit anti-COL4A4 (serum 356) (22), rat anti-COL4A1 (clone H11), mouse monoclonal antibody anti-COL4A345 NC1 domain hexamer (clone 26-20) (11), rat anti-nidogen (clone ELM1, Santa Cruz Biotechnology), rabbit anti-agrin (10), and rabbit anti-laminin-α₁ (4). Tissues were imaged on a Nikon Eclipse 800 inverted fluorescence microscope. Images were captured and processed with a DP72 digital camera and Olympus CellSens Standard software (version 1.11).

RESULTS

To test whether endothelium-targeted gene correction is a potential avenue for the treatment of Alport syndrome, we generated Col4a3^−/− global knockout mice carrying three transgenes: Cre recombinase driven by the Tie2 promoter (Tie2-Cre), which is active in ECs (and other hematopoietic lineages) but not podocytes; a floxed-Stop rtTA-IRES-GFP insertion at the Gt(ROSA)26 locus; and TetO₇-driven Col4a3 cDNA (Fig. 1A), referred to hereafter as Tg mice. Our previous podocyte-specific rescue of Alport syndrome with nephrin-rtTA transgene was effective even when doxycycline was administered as late as P21 (20). In the present study, Tg mice and control littermates were given doxycycline beginning at P0–P7 and maintained on it throughout life to induce COL4A3 transgene expression in ECs.

Fig. 1.

Expression of inducible collagen type IV-α₃ (Col4a3) cDNA in endothelial cells does not rescue the Alport phenotype in Col4a3^−/− mice. A: Cre recombinase expressed from the hematopoietic/endothelium-specific Tie2 promoter activates ROSA26-driven reverse tetracycline-controlled transactivator (rtTA) expression by removing a floxed stop sequence. In the presence of doxycycline, endothelium-expressed rtTA activates the TetO₇ promoter to drive Col4a3 cDNA expression. B: PCR of cDNA generated from whole kidney lysates with Col4a3 cDNA primers that cannot amplify the Col4a3-null transcript cDNA produced the expected product in Col4a3^+/− mice (lanes 1, 3, and 5) and in Col4a3^+/− control mice lacking Col4a3 (lanes 5 and 8). PCR from Col4α3^−/− transgenic (Tg) mice (lanes 2, 4, and 6) produced the same band as did Col4a3+/− mice, indicating transcription of the Col4a3 transgene, but Col4a3^−/− mice lacking the transgene did not produce a Col4α3 mRNA PCR product (lane 7). M, marker. C: periodic acid-Schiff staining revealed widespread glomerulosclerosis and tubular protein casts in 78-day-old Col4a3^−/− Tg mice, consistent with progression of the Alport phenotype. Images are representative of 3 control mice and 3 Col4a3−/− Tg mice ranging from 26 to 78 days old. D: Cre recombinase expression from the Tie2 promoter switches mTmG expression from a membrane-bound Tomato fluorophore (left; mTmG) to membrane-bound enhanced green fluorescent protein (EGFP; right; Tie2-Cre; mTmG) via removal of the floxed Tomato expression cassette. Tomato expression was observed in all cells of mTmG mice (left), including the intimal cells of large vessels (arrows). After Cre-mediated excision (right), Tomato was absent, whereas GFP was observed in the glomerular endothelium, in the interstitial endothelium, and in intimal cells of large vessels, which additionally masked the auto-fluorescence (see zoomed images) of the internal elastic lamina compared with mTmG mice (arrowheads). Images are representative of kidneys from two mice analyzed.

To first verify the function and specificity of the Tie2-Cre transgene, we mated mice carrying the mTmG Cre reporter allele with mice expressing Tie2-Cre. Before Cre recombinase activity, the mTmG allele expresses a membrane-tagged Tomato fluorophore, but after Cre activity, Tomato is replaced by membrane-tagged green fluorescent protein. Whereas mTmG mice exhibited exclusively Tomato fluorescence (Fig. 1D, left), Tie2-Cre; mTmG mice exhibited a strong green fluorescent protein signal in ECs, including those in glomeruli (Fig. 1D, right).

To verify TetO7-Col4a3 transgene expression in Tg mice by RT-PCR, we used primers that amplify Col4a3 cDNA derived from the wild-type Col4a3 gene and from the Col4a3 transgene but not from the Col4a3-null allele. PCR of cDNA generated from whole kidney RNA treated with DNase generated the predicted product from Col4a3 heterozygous mice with and without the Col4a3 transgene and from Col4a3^−/− Tg mice that were induced with doxycycline but not from a Col4a3^−/− mouse lacking the Col4a3 transgene (Fig. 1B, top row). RT-PCR products were not generated without mRNA reverse transcription, indicating an absence of genomic DNA contamination (Fig. 1B, bottom row). These data collectively indicate that our approach successfully induced transcription of the Col4a3 transgene, presumably in ECs based on the reporter (Fig. 1D). Despite evidence of induced transgenic Col4α3 expression, periodic acid-Schiff staining in the oldest littermate pair (78 days) revealed protein casts and sclerotic glomeruli (Fig. 1C), indicating that endothelium-derived Col4a3 expression did not rescue the Alport phenotype.

To assay for expression of any COL4A3 protein or COL4A3/A4/A5 protomers, we performed immunofluorescence with antibodies to COL4A3 (Fig. 2), COL4A4 (Fig. 3), and COL4A3/A4/A5 hexamers (Fig. 4) that form upon assembly of a COL4A3/A4/A5 network. Despite PCR-based evidence of transgene expression (Fig. 1), COL4A3, COL4A4, and COL4A3/A4/A5 hexamers were undetectable in Col4a3^−/− Tg mice, indicating that forced expression of the COL4A3 isoform in ECs is insufficient to produce COL4A3/A4/A5 protomers.

Fig. 2.

Collagen type IV-α₃ (COL4A3) was not detected in induced Col4a3^−/− transgenic (Tg) mice. Kidneys from Col4a3^−/− Tg mice and Col4a3^+/− control mice (78 days old shown) were examined by immunofluorescence for COL4A3. Control mice exhibited robust COL4A3 (green) staining that colocalized with the glomerular basement membrane constituent agrin (red). Induced Col4a3^−/− Tg mice exhibited no COL4A3 staining. Images are from individual Col4a3^−/− Tg mice aged as indicated.

Fig. 3.

Collagen type IV-α₄ (COL4A4) was not detected in induced Col4a3^−/− transgenic (Tg) mice. Kidneys from Col4a3^−/− Tg mice and Col4a3^+/− control mice (78 days old shown) were examined by immunofluorescence for COL4A4. Whereas control mice exhibited robust COL4A4 (red) staining in glomeruli and tubules, induced Col4a3^−/− Tg mice lacked any COL4A4 staining. Images are from individual Col4a3^−/− Tg mice aged as indicated.

Fig. 4.

Collagen type IV (COL4)-α₃α₄α₅ NC1 hexamers were not detected in induced Col4a3^−/− transgenic (Tg) glomerular basement membranes. Kidneys from induced Col4a3^−/− Tg mice and Col4a3^+/− control mice (78 days old shown) were examined by immunofluorescence staining for NC1 domain hexamers that would indicate an assembled COL4A3/A4/A5 network. Consistent with staining for COL4A3 and COL4A4, anti-hexameric COL4A345 NC1 domain staining exhibited robust hexamer signals that colocalized with the basement membrane constituent nidogen in tubules and glomeruli of Col4a3^+/− kidneys. No signals were observed in induced Col4a3^−/− Tg kidneys, indicating that no COL4A3/A4/A5 protomers were assembled. Images are from individual Col4a3^−/− Tg mice aged as indicated.

Alport GBMs typically exhibit molecular pathology associated with compensatory mechanisms, including enhanced deposition of COL4A1/A2/A1 and of laminins not found in the healthy adult GBM, including laminin-α₁, laminin-α₂, and laminin-β₁. Our previous podocyte-specific rescue of COL4A3 expression eliminated these indicators of Alport progression (20). Here, in contrast, after 26 days of induction with doxycycline, immunofluorescent staining of Col4a3^−/− Tg kidneys revealed ectopic laminin-α₁ in the GBM that was not observed in control kidneys (Fig. 5). Furthermore, COL4A1/A2/A1, which is increased in the Alport GBM and is thought to incompletely compensate for the loss of COL4A3/A4/A5, was markedly enhanced and continuous in the GBM of all three Tg mice that were examined (Fig. 6). These data support the Col4 expression data above (Figs. 2–4). The data herein collectively indicate that expression of COL4A3 in ECs is not capable of driving restoration of the COL4A3/A4/A5 protomers missing from the GBMs of Col4a3^−/− mice.

Fig. 5.

Ectopic laminin-α₁ staining in the glomerular basement membrane (GBM) of induced collagen type IV-α₃ (Col4a3)^−/− transgenic (Tg) kidneys mimics the Alport syndrome molecular phenotype. Kidneys from induced Col4a3^−/− Tg mice and Col4a3^+/− control mice (78 days old shown) were examined by immunofluorescence staining for laminin-α₁. As expected, control mice exhibited localization of laminin-α₁ (red) in the mesangial matrix and proximal tubule basement membranes but no staining in the GBM. The induced 26-day-old Col4a3^−/− Tg mouse did not exhibit laminin-α₁ in the GBM, but induced 64- and 78-day-old Col4a3^−/− Tg mice exhibited laminin-α₁ deposition throughout the GBM. Images are from individual Col4a3^−/− Tg mice aged as indicated.

Fig. 6.

Enhanced collagen type IV-α₁ (COL4A1) in the glomerular basement membrane (GBM) of induced Col4a3^−/− transgenic (Tg) kidneys reflects Alport syndrome compensatory mechanisms. Kidneys from Col4a3^−/− Tg mice and Col4a3^+/− control mice (78 days old shown) were examined by immunofluorescence staining for COL4A1, which is expressed at low levels by healthy glomerular endothelial cells but is enhanced in the Alport GBM. Compared with control littermates, COL4A1 deposition was dramatically enhanced in the GBMs of induced 64- and 78-day-old Col4α3^−/− Tg mice (arrows). Images are from individual Col4a3^−/− Tg mice aged as indicated.

DISCUSSION

Our previous success with doxycycline-induced podocyte-specific expression of Col4a3 cDNA in Col4a3^−/− Alport mice proved that absent COL4A3/A4/A5 protomers can be replaced and Alport syndrome can be circumvented or attenuated, depending on the timing of transgene induction (20). While targeting podocytes remains a significant challenge for gene therapy strategies, the accessibility of ECs suggests a more promising potential for gene therapy. Although numerous reports have indicated that only podocytes express COL4A3/A4/A5 in mouse and human glomeruli of healthy and Alport tissues (1, 12, 18, 24), transgenic expression of COL4 from the endothelium has not been previously attempted. In the present study, we induced expression of Col4a3 transgene in ECs of Col4a3^−/− Alport mice through Tie2-Cre-mediated activation of ROSA26-floxed-STOP rtTA to drive TetO₇-Col4a3 cDNA transgene. Successful expression of the transgene was demonstrated by RT-PCR. However, Col4a3^−/− Tg mice exhibited no COL4A3 (Fig. 2), COL4A4 (Fig. 3), or assembled COL4A3/A4/A5 hexamer staining (Fig. 4), and they exhibited histochemical and molecular pathology consistent with Alport syndrome in mice. This study provides experimental proof in vivo to support the concept that all three chains must be expressed in the same cell to generate functional COL4 heterotrimers.

The requirement for intracellular assembly of all three COL4 chains for COL4A3/A4/A5 protomer secretion was previously suggested by the absence of the nonmutated isoforms from Alport GBM, with the exception being the presence of ectopic COL4A5/A6/A5 protomers in the GBM of Col4a3^−/− mice on the C57BL/6J background (14). While it is generally accepted that COL4 isoforms lacking assembly partners are degraded rapidly before secretion (27), COL4 expression and stability have not been thoroughly examined in ECs. We speculated that negative regulation may limit expression of the α₃-, α₄-, and α₅-chains in healthy ECs, but this repression may be altered in pathogenic contexts. However, the lack of COL4A3, COL4A4, and hexameric COL4A345 NC1 domain staining in Col4a3^−/− Tg mice agrees with previously published COL4 expression studies and demonstrates that gene therapy-mediated repair of mutant Col4 alleles in the glomerular endothelium will be insufficient to produce COL4A3/A4/A5 protomers and rescue the Alport phenotype.

These data indicate that an endothelium-targeted gene therapy approach for the rescue of Alport syndrome is unlikely to succeed. We additionally speculate that ectopic expression of all three isoforms (if possible) by ECs is not guaranteed to produce posttranslationally equivalent COL4A345 protomers, although they could very well be functional. Furthermore, the subendothelial localization of injected, full-length human laminin-521 trimers into Lamb2^−/− Pierson syndrome mice (19) suggests that size exclusion imparted by the COL4A1/A2/A1 network or by other GBM components could be a significant barrier to proper localization of endothelium-derived COL4A3/A4/A5 protomers. Nevertheless, gene therapy remains an attractive approach to treating Alport syndrome. Recent modification of CRISPR/Cas9 systems has led to sophisticated tools that could be used to ectopically activate multiple promoters in vivo, thereby circumventing limited COL4 expression profiles and allowing testing of the effects of COL4A3/A4/A5 protomer production by glomerular ECs. Although coordinating so many factors would be technically challenging when even experimental gene therapy is not trivial, these and future advancements could someday provide a therapy for this important disease.

GRANTS

This work was funded by National Institute of Diabetes and Digestive and Kidney Diseases Grants R01-DK-078314 (to J. H. Miner) and T32-D-K007126 (to S. D. Funk).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

S.D.F. and J.H.M. conceived and designed research; S.D.F. and R.H.B. performed experiments; S.D.F., R.H.B., and J.H.M. analyzed data; S.D.F. and J.H.M. interpreted results of experiments; S.D.F. and R.H.B. prepared figures; S.D.F. and J.H.M. drafted manuscript; S.D.F., R.H.B., and J.H.M. edited and revised manuscript; S.D.F., R.H.B., and J.H.M. approved final version of manuscript.