A spatial signature of the proximal tubules' metabolic switch during chronic kidney disease progression

Glomerulonephritis (GN) is characterized by inflammation of the glomeruli, leading to impaired kidney function and chronic kidney disease (CKD). This process can result in diffuse interstitial fibrosis beyond the site of initial glomerular injury. Yet, the metabolic pathways of kidney tubular cells and their relationship with inflammatory processes promoting fibrosis remain incompletely understood. This study aimed to characterize lipid metabolism changes in a murine model of GN-induced CKD progression and in kidney biopsies from patients with ANCA-vasculitis-associated GN.

CKD was induced in male mice (10–12 weeks of age) with a C57BL6/J background by retroorbital injection of sheep anti-mouse GBM nephrotoxic serum (NTS).  Mice were sacrificed 21 days after NTS injection. Urines were collected on days 0, 3, 5, 7, 14, and 21, and plasma samples were collected at the time of sacrifice. Functional analysis of plasma creatinine, urea, and triglycerides (TG) was performed by uHPLC using the Biochemistry platform at Bichat Hospital (Paris, France). Kidneys were perfused in situ with cold saline buffer before removal. One half of each kidney was snap-frozen in liquid nitrogen until sample preparation for MALDI-IMS, and the other half was fixed in paraformaldehyde and paraffin-embedded for conventional histological studies.

Renal biopsy specimens for MALDI-IMS analysis after the completion of the diagnostic workup were obtained from the Hôpital Européen Georges Pompidou, Assistance Publique-Hôpitaux de Paris, Paris, France. Human tissue was used with approval from and in accordance with the guidelines of the local ethics committee. Codes were used to ensure patient anonymity. The study complies with the 2000 and 2024 Declaration of Helsinki.

Matrix-Assisted Laser Desorption/Ionization Imaging Mass Spectrometry (MALDI IMS) was used to profile lipid alterations in up to eight renal structures. 

Mapping of nephron segments was performed using multiplexed immunostaining. Sections analyzed by MALDI-IMS were fixed with 4 % paraformaldehyde in phosphate-buffered saline for 10 min at room temperature and washed in cold methanol/acetone (1:1) for 5 min to remove the MALDI matrix. After 1h of blocking with PBS 3% BSA, specific renal segments were immunostained with AQP1 (Proximal tubules and thin descending limbs of Henle's loop), AQP2 (collecting ducts), and Uromodulin (thick ascending limb (TAL) of Henle's loop). An anti-collagen I antibody was used to identify fibrotic areas. Afterward, samples were incubated at room temperature for 1 h with Alexa Fluor 488 and CY3-labeled secondary antibodies. All sections were counterstained with 10 μg/ml Hoechst 33342. After mounting, stained sections were scanned with the high-speed slide scanner (VS200, Olympus).

Sixteen lipid families were identified, totaling 167 lipid species. The control and GN kidneys showed up to 8 distinct lipid fingerprints. GN-associated CKD was characterized by increased phosphatidylcholine (PC) and ether/vinyl ether phosphatidylethanolamine (PE-O/P) levels, alongside decreased hexosylceramide (HexCer) and phosphatidylserine (PS) species.

1- We could distinguish two different lipid signatures for proximal tubules (PTI and PTII). Both are enriched in PC, PE, and phosphatidylinositols (PI), but they differ in their relative expression of ether and vinyl ether lipids (PE O/P and PC O/P), phosphatidylglycerol (PG), and PS; PTII has elevated levels of PE O/P and PC O/P, while PG and PS are more expressed in PTI. Additionally, TAL is characterized by the accumulation of PI O/P, PS, and SFTt, while CDT, regardless of whether it is the contiguous segment to TAL in the nephron, has higher levels of PI, LPI, PE O/P, and PC O/P lipid species.

The PCA results showed distinct separation. Additionally, the confusion matrix and performance metrics for each statistical model tested showed that the Logistic Regression method yielded the best results, achieving an AUC of 0.996, precision of 0.944, and recall of 0.92.

2- During CKD, we calculated the PC/PE, PC/PS, PC/PI, and PC/PG ratios in control and NTS-CKD samples and found significant increases in all ratios in diseased mice. Moreover, a significant Pearson correlation between glycerophospholipids (GP) ratios and plasma creatinine and urea levels suggests functional associations within each kidney structure.

PC/PE ratio showed a significant positive correlation with uremia in PTI, PTII, MDT, and CDT segments and with high creatinemia in PTI, PTII, and whole kidney (WK). 

For the PC/PS ratio, a positive correlation was observed only in the PTII with urea and creatinine levels; for the PC/PG ratio, only a significant positive correlation with plasma urea was observed in PTII, MDT, and CD. All the information suggests that the PTII segment is the most affected renal structure in terms of lipid composition changes and the connection between kidney functional parameters during CKD.

Integration with transcriptomics revealed a coordinated dysregulation of lipid biosynthetic and remodeling pathways. The CDP-ethanolamine pathway was suppressed, consistent with the reduction in phosphatidylethanolamines, while expression of the PS biosynthetic enzyme PSS1 was decreased, matching the IMS-detected loss of PS. In contrast, the CDP-choline pathway and its associated enzyme PEMT were upregulated, favoring PC accumulation. Strikingly, we also observed upregulation of the Lands’ cycle enzyme MBOAT1, which preferentially acylates lysophosphatidylethanolamine with monounsaturated fatty acids(28), thereby directly driving the enrichment of MUFA-PE and MUFA-PE O/P species captured by IMS. This molecular reprogramming provides a mechanistic explanation for the elevated PC/PE and PC/PS ratios observed in diseased kidneys.

Human ANCA-GN and murine NTS-CKD share lipidomic alterations in proximal tubules.

MALDI IMS of lipids in ANCA-GN samples reveals similar lipid modifications within the NTS-CKD model. When plotting the 10 most significantly shifted lipids during ANCA-GN, PS 36:1 and PC 34:2 appear to be among the most affected individual lipid species in the progression of nephropathy. Deep analysis of lipid distribution in human renal biopsies highlights a significant increase in PC 34:2 and a decrease in PS 36:1 in proximal tubules. The increased PC/PS ratio appears to be strongly relevant to the development of kidney disease in both murine and human samples.

3-  Importantly, fatty acid analysis revealed a significant reduction in PUFAs such as docosahexaenoic acid (DHA, FA 22:6) in NTS-CKD kidneys, a known endogenous ligand of PPARγ. Likewise, we found PPARγ downregulation in proximal tubules near fibrotic areas, linking lipid alterations to fibrosis progression in CKD. Immunofluorescence confirmed the loss of PPARγ specifically in AQP1-positive proximal tubules, while collecting ducts (AQP2-positive) remained unaffected.

In this study, we provide a comprehensive analysis of metabolic alterations driving chronic kidney disease progression by integrating high-resolution MALDI-IMS lipidomics, transcriptomics, and functional parameters. Our data reveal that proximal tubules undergo profound lipid remodeling.

MALDI-IMS identified distinct lipid fingerprints across nephron segments, with cortex-to-medulla gradients in sphingomyelins, sulfatides, and hexosylceramides that mirror human kidney lipid architecture. 

During NTS-induced CKD, there was a striking accumulation of phosphatidylcholines and ether-linked phosphatidylethanolamines, accompanied by loss of hexosylceramides and phosphatidylserines. These changes were most pronounced in cortical proximal tubules, whereas increases in PC and PE O/P occurred throughout the nephron. This selective vulnerability of proximal tubules underscores their central role in metabolic reprogramming during fibrosis.  

Our data indicate that PPARγ may act as a critical regulator of tubular lipid metabolism and inflammation. PPARγ expression was markedly reduced in proximal tubules of NTS-CKD mice, coinciding with decreased levels of endogenous ligands such as DHA.

The translational relevance of these findings was confirmed in human ANCA-GN biopsies, in which proximal tubules showed increased PC 34:2 and reduced PS 36:1, and decreased PPARγ expression paralleling the murine data. Although some lipid species, such as sphingomyelins, displayed human-specific profiles, the remodeling of PC/PS ratios emerged as a conserved marker of proximal tubular injury. These findings suggest that phospholipid ratios may hold potential not only as histological readouts but also as candidate biomarkers accessible in biofluids, highlighting their clinical value as indicators of CKD progression.

Some of the data were presented at the European Renal Cell Study Group 2024 meeting.