pdx1 Knockout Leads to a Diabetic Nephropathy– Like Phenotype in Zebrafish and Identifies Phosphatidylethanolamine as Metabolite Promoting Early Diabetic Kidney Damage

The complex multifactorial pathophysiology of diabetic kidney disease (DKD) involves multiple changes in the renal hemodynamics, hypoxic processes, altered glucose metabolism associated with reactive oxygen species production, inflammation, and renin–angiotensin-aldosterone system upregulation (1). Abnormal angiogenesis is recognized as another major pathway involved in the development of DKD (2). Due to this complexity, singular preclinical animal models of DKD can typically not represent all criteria of DKD, and a multitude of models is used according to primary scientific focus (3).

The zebrafish is a vertebrate animal model with a strong forte for the evaluation of morphological characteristics in vivo and in biochemical parameters, combined with a powerful tool set to modify genetic conditions. Rapid organogenesis is undergoing in the early zebrafish embryo, showcasing a functional renal system with ultrafiltration as early as 40 h postfertilization (hpf). The early zebrafish pronephros already possesses typical features of higher vertebrate kidneys like polarized tubular epithelial cells, interdigitating podocyte foot processes (4), and a fenestrated glomerular endothelium (5,6).

Pancreatic and duodenal homeobox 1 (Pdx1) is guiding β-cell differentiation and physiological insulin gene transcription as one of the earliest active pancreatic transcription factors. pdx1 knockout in zebrafish led to the establishment of a novel model for diabetic retinopathy, in which hyperglycemia leads to the activation of retinal angiogenesis in both larval and adult stages (7,8). In this work, we investigate the pdx1 mutant as a monogenetic zebrafish model for DKD and study the influence of pdx1 deficiency on the larval pronephros and the matured zebrafish kidney.

Embryos of Tg(wt1b:EGFP) (9) and Tg(fli1:EGFP) (10) lines were raised, kept, and genotyped for pdx1 as previously published (7). For experiments, both sexes were included.

Pronephros dimensions were quantified in Tg(wt1b:EGFP) embryos as previously described (11).

SB-pdx1-Mo (5′-GATAGTAATGCTCTTCCCGATTCAT-3′) and control-Mo (5′-CCTCTTACCTCAGTTACAATTTATA-3′) at a dosage of 6 ng/embryo were used as previously described (7).

Pronephros function was determined using a 70-kDa dextran conjugated to a Texas Red fluorochrome as previously described (11).

Tg(wt1b:EGFP) embryos were selectively bred for genotype, screened via fluorescence microscopy at 48 hpf, and raised. Embryos were exposed in toto, while adult zebrafish kidneys were dissected in ice-cold PBS directly after sacrifice and exposed to a primary fixative of 2.5–3.0% glutaraldehyde in 100 mmol/L cacodylate, pH 7.2, overnight at 4°C, followed by secondary fixation with 1% osmium solution and either Spurr or epon embedding. Image acquisition used a JEOL JEM1400 microscope equipped with a TVIPS F416 digital camera at 80 kV and ×3,000–8,000 magnification. Images were analyzed using EM-Measure and ImageJ.

Tg(fli1:EGFP) renal tissue was dissected in ice-cold PBS and fixed in 4% paraformaldehyde for 6 h at 4°C. The 1,024 × 1,024-pixel confocal z-stacks were acquired on a DM6000 B confocal microscope with a Leica TCS SP5 DS scanner using the 63× oil immersion objective. Images were analyzed using ImageJ and Imaris 9.5.

Formalin-fixed paraffin-embedded sections were prepared, and hematoxylin-eosin staining as well as periodic acid Schiff (PAS) reactions were performed. All sections were digitalized using an Aperio AT2 scanner (Leica Microsystems, Wetzlar, Germany). May-Grünwald staining was performed on peripheral blood smears, and images were acquired using an Axio Imager Z1 (Zeiss, Oberkochen, Germany). QuPath v0.2.0 and ImageJ were used for quantification (Supplementary Methods 1).

Zebrafish kidneys were dissected in ice-cold PBS, stored at −20°C in RNAlater, and homogenized simultaneously using ceramic beads and 1% 2-mercaptoethanol. RNA isolation was performed via RNeasy Kit. Library preparation and paired-end sequencing using a TruSeq mRNA Library Kit were performed by ATLAS Biolabs GmbH (Berlin, Germany). The analysis of the RNA-sequencing (RNA-seq) data was executed by the Next-Generation Sequencing Core Facility of the Medical Faculty Mannheim as previously described (12).

Kidney tissue was dissected in ice-cold PBS and immediately frozen in liquid nitrogen. Determination of amino acid levels was done via ultra-performance liquid chromatography fluorescence as described in Weger et al. (13). Fatty acids and primary metabolites were determined by semitargeted gas chromatography-mass spectrometry (GC/MS) analysis. Detailed information is reported in Supplementary Methods 2.

Following fertilization, Tg(wt1b:EGFP) zebrafish eggs were either incubated in egg water with different concentrations of DMSO, phosphoethanolamine (PE), N-(p-amylcinnamoyl)anthranilic acid, and meclizine or injected into the yolk sack with 3 nL corn oil with and without 50 mg/mL phosphatidylethanolamine (PtdE) supplementation. Embryos were dechorionated at 24 hpf, raised until 48 hpf, and analyzed for pronephros alterations.

Results show mean values ± SD. Statistical significance between different groups was analyzed using t test, one-way ANOVA, two-way ANOVA, and appropriate post hoc tests using GraphPad Prism. P values <0.05 were considered significant: *P < 0.05; **P < 0.01, ***P < 0.001. ****P < 0.0001.

All experimental procedures on animals were approved by the local government authority, Regierungspräsidium Karlsruhe (license no. G-160/14), and by Medical Faculty Mannheim (license no. I-19/02) and carried out in accordance with the approved guidelines.

Raw RNA-seq data were uploaded to the Gene Expression Omnibus (National Center for Biotechnology Information) under accession number GSE179104. Other data sets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

The pdx1 mutation was induced in the Tg(wt1b:EGFP) line showing specific EGFP expression in the early pronephros (9). Both pdx1^+/− and pdx1^−/− mutants exhibit significantly shortened pronephric neck length and enlarged glomeruli in comparison with pdx1^+/+ wild-type siblings after 48 hpf (Fig. 1A, C, and D), with indications of a gene dose effect similar to prior observations regarding pancreatic development (7). At 48 h postinjection (hpi), pdx1^−/− larvae show an increased loss of a fluorescent dextran with a molecular weight of 70 kDa mimicking albumin (Fig. 1B and E), indicating functional impairments in the affected glomerular filtration barrier (14).

Zebrafish larvae at 96 hpf show mammalian-like glomerular basement membrane (GBM) morphology accompanied by fenestrated endothelium and evenly spaced podocyte foot processes (Fig. 2C), which are seen fully matured in the electron microscopy (EM) of adult wild types (Supplementary Fig. 1). pdx1^uh10 zebrafish mutants exhibit increased glucose concentrations in the larval state and postprandial hyperglycemic spikes as adults (7). The phenotypical pdx1^−/− pronephros shows significant GBM thickening after just 4 days under diabetic conditions (Fig. 2A and C). Prolonged hyperglycemic exposure in adult pdx1^−/− mutants shows a progressive thickening of the GBM with increased age (Fig. 2B and C). In larval and adult pdx1^−/− mutants, qualitative assessment hinted at focal disorganized and clumped podocyte foot processes.

Homozygous pdx1 knockout leads to a strong survival adversity to reach adulthood in zebrafish, seen in both pdx1^sa280 mutants (15) and our pdx1^uh10 mutants (7). Heterozygous pdx1 mutants with normal survival previously demonstrated susceptibility to develop impaired glucose tolerance and retinal alterations during late adulthood (7). Glomerular morphology was screened via PAS reaction in 12-month-old pdx1^+/− mutants and revealed a significant increase in glomerular nuclear count and glomerular area (Fig. 2D–F). Diabetic glomerulosclerosis was not detectable at this stage via PAS reaction.

Gene set enrichment analysis after RNA-seq of heterozygous pdx1^+/− renal tissue showed significant depletion of multiple biological processes involved in translation and transcription. An underrepresentation of vascular endothelial growth factor (VEGF) signaling, the VEGF–VEGF receptor 2 pathway, and RhoGTPase signaling were found, indicative of angiogenic dysregulation in the diabetic zebrafish kidney (Fig. 3A). However, analysis of the glomerular microvasculature via confocal microscopy in the Tg(fli1:EGFP) zebrafish line revealed there were no alterations in vascular coverage, endothelial nuclear count, or diameter of the glomerular arterioles, and no other obvious gross morphological changes could be identified (Fig. 3B–E). Renal hematopoietic stem cells in zebrafish could constitute a confounding factor; however, pdx1^+/− mutants show no difference in their peripheral blood constitution to wild types (Supplementary Fig. 2).

Kyoto Encyclopedia of Genes and Genomes pathway analysis showed the expected significant downregulation of insulin signaling in adult pdx1^+/− renal tissue and further significant regulations in several other metabolic pathways (Supplementary Fig. 3). In accordance with these, a reduction in free amino acids and increased free fatty acid levels were uncovered by metabolite measurements in pdx1^+/− kidneys. Significant main effects without interaction were found for both groups via two-way ANOVA (Supplementary Figs. 4 and 5). Post hoc analysis uncovered that threonine was significantly decreased, and palmitic acid (C16:0) was significantly increased. Further, hexoses, like d-glucose, were significantly decreased in pdx1^+/− renal tissue (Fig. 4A), with corresponding downregulation seen for the tricarboxylic acid cycle in the pathway analysis.

Increased PE concentrations showed the strongest significance of all metabolite measurements in adult pdx1^+/− mutants (Fig. 4B and Supplementary Fig. 6), while relevance to diabetic nephropathy remains unclear. In the downregulated glycerophospholipid metabolism pathway (Supplementary Fig. 3), multiple mechanisms related to PtdE lead to increasing PE concentrations. Upregulation of zebrafish phospholipase A2 (pla2g4aa) and retinoic acid receptor responder 3-like (rarres3l), which exhibits phospholipase A2–like molecular activity, break down PtdE, leading to PE increase. Novel synthesis of PtdE exhausting PE via ethanolamine-phosphate cytidylyltransferase (pcyt2) was downregulated. Concentrations of up to 50 μmol/L PE showed no effects on the glomerular morphology in wild-type embryos at 48 hpf (Supplementary Fig. 7). However, PtdE injections resulted in a significantly shortened pronephric neck length, partially mimicking the phenotype mediated by pdx1 knockout (Fig. 4C and F). Pharmacological inhibition of Pla2g4aa in wild-type zebrafish similarly led to the induction of the partial pdx1 pronephros phenotype (Fig. 4D). Inhibition of Pcyt2 via meclizine, and therefore inhibition of PtdE synthesis, partially rescued the pronephros phenotype in pdx1 morpholino (pdx1Mo)–injected zebrafish (Fig. 4E and Supplementary Fig. 8). Exposure of wild-type embryos to PtdE via injection led to significant GBM thickening, similar to the observed phenotype in pdx1^−/− larvae (Fig. 4F and G).

This study shows for the first time that pdx1 knockout leads to the development of early DKD features in zebrafish and highlights the involvement of PtdE in the onset of diabetic kidney damage.

Following the classical DKD disease progression, the zebrafish pronephros shows glomerular hypertrophy followed by impairments in the filtration barrier mimicking microalbuminuria, similar to early renal hypertrophy (16,17) and protein loss (18,19) in type 1 diabetes. Simultaneously, significant GBM thickening develops in the pdx1^−/− larvae as a hallmark of DKD. Focal qualitative hints at podocyte foot process effacement are seen in pdx1^−/− larvae and could explain the functional impairments (20).

Glomerular hypertrophy with increased cellularity is also seen in the fully developed heterozygous pdx1 mutant. The pdx1^+/− mutant showed renal underrepresentation of Rho and VEGF signaling matching a dysfunctional podocyte phenotype, as Rho signaling is involved in the control of the actin cytoskeleton (20) and podocytes are the primary glomerular source of VEGF-A (21).

In DKD, initially VEGF upregulation is seen, while in late DKD, downregulation of VEGF due to podocyte damage and incremental uncoupling of endothelial NO production from VEGF signaling sets in (22). Endothelial dysfunction due to a relative lack of paracrine VEGF signaling is probable in the pdx1 mutant, although no morphological microvascular phenotype was found yet. This pathophysiological response could represent an advantage of the fish model (23).

This study further revealed novel insight into PtdE’s relation to diabetic kidney damage. High concentrations of PtdE can induce pronephros alterations and GBM thickening, mimicking the pdx1 phenotype partially in healthy zebrafish without the involvement of hyperglycemia. Via nonenzymatic Amadori-glycation, PtdE can lead to lipid peroxidation, representing a potential mechanism of DKD development (24), which is especially important in the presence of hyperglycemia. Changes in gene regulation suppressing reactive PtdE levels in pdx1^+/− mutants could highlight adaptive responses to reduce kidney damage in the presence of such a diabetic metabolism.

Acknowledgments. The authors thank Hiltraud Hosser (Institute of Neuroanatomy, Medical Faculty Mannheim, Heidelberg University) and Andrea Döbler and Ulrike Ganserer (Electron Microscopy Lab, Institute of Pathology, Heidelberg University Hospital) for technical support regarding the transmission EM of the larval samples. The authors acknowledge the support of the Core Facility Live Cell Imaging Mannheim (DFG INST 91027/10-1FUGG) and the Zebrafish Core Facility of Medical Faculty Mannheim. Tissue samples were processed by the Diabetes-Specific Biobank of the Collaborative Research Center (CRC 1118). The Diabetes-Specific Biobank is affiliated with the Tissue Bank of the National Center for Tumor Diseases Heidelberg, Germany (25). The authors also thank Gernot Poschet and Ruediger Hell from the Metabolomics Core Technology Platform of the Excellence cluster “CellNetworks” (University of Heidelberg) and the Deutsche Forschungsgemeinschaft (grant ZUK 40/2010-3009262) for support with ultra-performance liquid chromatography and GC/MS-based metabolite quantification; Elke Deckert and Uwe Schwahn (Sanofi Deutschland GmbH, Frankfurt, Germany) for technical and analytical support regarding the RNA-seq; and the contributions of Sanofi Deutschland GmbH within the collaboration IRTG 1874/2 DIAMICOM.

Funding. The study was supported by grants from Deutsche Forschungsgemeinschaft (CRC 1118 and IRTG 1874/2 DIAMICOM).

Duality of Interest. P.W. is an employee of Sanofi Aventis Deutschland GmbH. Sanofi Aventis Deutschland GmbH provided the salary for P.W. and enabled access to research materials and in sharing data but did not have any additional role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. No other potential conflicts of interest relevant to this article were reported.

Author Contributions. L.M.W. and L.M. performed experiments, analyzed data, and wrote the manuscript. K.B., S.S., M.B., C.T.T., I.H., and N.V. performed experiments and analyzed data. E.H. and M.B. performed metabolome experiments and analyzed data. C.S. and P.W. performed RNA-seq experiments and analyzed data. H.-P.H. gave conceptual and technological advice and supported drafting of the manuscript. J.K. conceived and designed the study and wrote the manuscript. J.K. is the guarantor of this work and, as such, had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Prior Presentation. Parts of this study were presented at the 53rd Annual DDG Diabetes Congress (German Diabetes Association), Berlin, Germany, 9–12 May 2018, and the 54th Annual Meeting of the European Association for the Study of Diabetes, Berlin, Germany, 1–5 October 2018.