Tubule-Specific Compensatory Responses to Cpt1a Deletion in Aged Mice

Abstract

Key Points

• Aging plus high fat diet suppresses expression of genes related to metabolism and fatty acid oxidation in the proximal tubules.

• Proximal tubules lacking Cpt1a have significant transcriptional increases in Hmgcs2, peroxisomal fatty acid oxidation genes, and omega-oxidation genes.

• Distal convoluted tubules lacking Cpt1a have fewer changes in metabolic genes, but significantly downregulated expression of differentiation markers.

Background

Fatty acid oxidation (FAO) is the preferred energy pathway in the proximal tubule (PT), and carnitine palmitoyltransferase 1A (Cpt1a) is the rate-limiting enzyme of mitochondrial FAO. Cpt1a expression and FAO decrease after renal injury. Our recent work demonstrated that genetic deletion of tubular Cpt1a did not significantly worsen the response to injury or aging and did not completely block FAO, suggesting compensatory metabolic pathways. In addition, Cpt1a was most highly expressed in distal convoluted tubule (DCT), a segment not known for FAO. Therefore, we used single-nuclear RNA sequencing to explore a cell-specific responses to aging with high fat diet (HFD aging), to define compensatory metabolic pathways in PT segments lacking Cpt1a, and to determine the role of Cpt1a in the DCT.

Methods

Cpt1a floxed (Cpt1a^fl/fl) and tubule-specific conditional Cpt1a knockout (Cpt1a^CKO) mice were aged for 2 years with HFD. Single-nuclear RNA-sequencing was performed on these HFD-aged mice and young controls.

Results

HFD-aged mice had increased fibrosis, inflammation, and more injured PT cells than young mice. Whereas PT segments from HFD-aged mice had significant transcriptional changes in metabolism-related pathways, the DCT had more changes in inflammation-related pathways. Compared with floxed mice, HFD-aged Cpt1a^CKO mice had increased lipid deposition and increased inflammation, but no significant differences in fibrosis or renal function. PT segments from HFD-aged Cpt1a^CKO mice had significantly upregulated Hmgcs2, a promoter of ketogenesis and FAO, and upregulated genes in peroxisomal FAO and omega-FAO (CYP4A family) pathways. DCT from HFD-aged Cpt1a^CKO mice had decreased expression of DCT-specific markers of cell differentiation.

Conclusions

The upregulated Hmgcs2, peroxisomal FAO genes, and CYP4A genes may compensate for impaired mitochondrial metabolism of long chain fatty acids in PT cells lacking Cpt1a. Our data suggest that Cpt1a may be important in maintenance of cell differentiation for DCT.

Visual Abstract

Introduction

The kidney, particularly the proximal tubule (PT), has high metabolic demands to support the transporters necessary to reclaim 99% of filtered solutes.¹ To generate the high amount of required ATP, the PTs preferentially use fatty acid oxidation (FAO).² Long chain fatty acids (LCFAs) require carnitine palmitoyltransferase 1 (CPT1) for entry into the mitochondria. The kidney exclusively expresses the isoform carnitine palmitoyltransferase 1A (Cpt1a), the rate-limiting enzyme for FAO.³ Decreased Cpt1a gene expression was observed in human chronic kidney injury,⁴ and overexpression of Cpt1a in murine tubules improved the response to chronic kidney injury, suggesting that impaired FAO worsens response to injury.⁵ We generated mice with selective Cpt1a deletion in kidney tubules (Cpt1a^CKO) and either aged them for 2 years or induced chronic injury.¹ Surprisingly, the Cpt1a^CKO mice had only subtle differences from floxed controls after aging or injury, despite robust recombination.¹

These prior published data¹ raise questions regarding the role of Cpt1a and kidney metabolism. We demonstrated that Cpt1a deletion did not completely block LCFA oxidation. If Cpt1a is required for LCFA import into the mitochondria, what compensatory mechanisms are responsible for LCFA metabolism and maintaining the metabolic health of the PT on Cpt1a deletion? In addition, Cpt1a protein and gene expression were found to be most highly expressed in the distal convoluted tubule (DCT), not in the PT.¹ This finding challenges the dogma that the distal tubule is primarily glycolytic and raises the question of whether Cpt1a may be mediating nonmetabolic functions in this tubule segment. Single-nuclear RNA sequencing (snRNA-seq) techniques are ideal for investigating these tubule segment-specific effects of Cpt1a deletion.

Kidney function declines with aging, and the pathophysiology is thought to include inflammation, fibrosis, senescence, and dysfunctional mitochondria,^6–10 but the exact mechanisms underlying age-associated injury are poorly understood. Altered metabolism has been implicated in aging-associated kidney dysfunction, and mice lacking peroxisome proliferator-activated receptorα (PPARα), a master regulator of FAO oxidation genes,¹¹ sustained greater aging-associated kidney injury. Most studies with aging use regular chow, but this does not reflect the average Western diet, which has a much higher fat content. In this study, we aged Cpt1a^CKO mice and floxed controls for 24 months while on a high fat diet (HFD-aging) for 16 months and performed snRNA-seq to address the following questions: (1) How does Cpt1a deletion alter the tubular response to HFD-aging? (2) What compensatory mechanisms maintain proximal tubular metabolic health in aged mice lacking tubular Cpt1a? (3) What role is Cpt1a playing in the DCT, which has the highest Cpt1a expression? To the best of our knowledge, this work represents the first single cell/nuclear transcriptomic dataset on aged kidneys with HFD. We define potentially important compensatory metabolic pathways upregulated in PT segments from HFD-aged Cpt1a^CKO mice, as well as nonmetabolic effects of Cpt1a deletion in the DCT.

Methods

Animals

All procedures were approved by the Institutional Animal Care and Use Committee of Vanderbilt University Medical Center and conducted according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals. Male Cpt1a floxed (Cpt1a^fl/fl) mice¹² on a mixed background crossed with mice containing Pax8-rtTA; tetO-Cre allele¹³ were fed doxycycline-containing chow at 5–8 weeks to induce recombination, generating conditional tubule-specific knock-down (Cpt1a^CKO) or control (Cpt1a^fl/fl) mice. One cohort of Cpt1a^CKO and Cpt1a^fl/fl mice were euthanized at 3 months, whereas another cohort was maintained on a HFD (Bio-Serv Mouse Diet; 60% kcal from fat) from 8 months of age until euthanasia at 24 months.

Renal Function Measurement

Plasma was collected at euthanasia, and creatinine was measured by the University of Alabama at Birmingham's O'Brien Center by high performance liquid chromatography. BUN was measured using Quantichrom Urea Assay Kit (BioAssays, DIUR-100).

Single Nucleus Isolation, Library Construction, and Informatics

Please see Supplemental Material.

Western Blotting

Kidney tissue was minced in lysis buffer (150 mM NaCl, 50 mM Tris HCl pH 7.4, 1 mM EDTA, 2% SDS), sonicated, and centrifuged. Supernatants were diluted with NuPage sample buffer (Thermo), electrophoresed on 10% SDS-PAGE gels, and transferred to PVDF membranes. They were blocked in 5% milk and then incubated overnight with the following antibodies: mouse anti-CPT1A (1:1000, Abcam ab128568 [8F6AE9]), rabbit anti-3-hydroxyl-3-methylglutaryl CoA synthetase 2 (Hmgcs2; Abcam, ab173043), or mouse anti-β-actin (1:5000, cell signaling 3700) before 1-hour incubation with the species appropriate HRP-conjugated secondary antibody. Membranes were developed with Clarity or Clarity MAX ECL substrate kits (Bio-Rad) on a Bio-Rad ChemiDoc MP imager.

Immunohistochemistry, Immunofluorescence, Quantitative PCR, Staining, and Statistics

Please see Supplemental Material.

Results

Tubular Cpt1a Deletion Did Not Worsen Kidney Function or Fibrosis after Aging and HFD

Mice with Cpt1a^fl/fl and Cpt1a^fl/fl; Pax8-rtTA; tetO-Cre (Cpt1a^CKO) were given doxycycline-containing diet and euthanized at 3 months or fed a HFD and euthanized at 24 months (Figure 1A).¹³ Robust recombination was confirmed in HFD-aged kidneys by immunoblots (Figure 1B). There were no differences in total body or kidney weights between Cpt1a^CKO and Cpt1a^fl/fl HFD-aged mice (Supplemental Figure 1, A and B). Compared with young mice, HFD-aged mice had mildly increased tubular injury by H&E and kidney injury molecule-1 (KIM-1) gene expression, increased interstitial fibrosis by Sirius Red, and increased medullary macrophage infiltration, but no significant differences were noted in renal function (Figure 1, C–J and Supplemental Figure 1, C–E). HFD-aged Cpt1a^CKO mice had significantly increased lipid droplet accumulation and macrophage infiltration by Oil Red O and F4/80 staining, respectively (Figure 1, H–L), but they did not have significantly altered KIM-1 expression, kidney function, or fibrosis compared with aged controls (Figure 1, D–G and Supplemental Figure 1, C and D). Although HFD-aged mice had increased KIM-1 expression compared with aged mice on regular chow (Supplemental Figure 1E), the differences between HFD-aged Cpt1a^CKO and Cpt1a^fl/fl mice were very similar to the differences between these genotypes in aging without HFD, as we recently published.¹

Figure 1.

Aging with HFD did not exacerbate injury in Cpt1a conditional knockout mice. (A) Male Cpt1a^fl/fl and Cpt1a^fl/fl; Pax8-rTtA; tetO-Cre (Cpt1a^CKO) mice were placed on Dox containing diet at 5 weeks of age for 3 weeks, and schematic for generating 3 month or HFD-aged mice is shown. (B) Western blotting of cortical tissue lysates from HFD-aged mice assessing CPT1A expression with β-actin as loading control. (C) H&E staining is shown in young and HFD-aged mice. KIM-1 (Havcr1) gene expression showing no significant differences between genotypes in either aged (24 months) or aged+HFD mice (D). Representative picrosirius red staining of young and aged+HFD (24 months) kidneys (E) with quantification of tubulointerstitial (F) and glomerular (G) area that stains positive (i.e., red). F4/80 staining shown of the renal cortex and medulla with quantification (H–J). Oil red O staining (K) and quantitation (L) revealed significant lipid accumulation in aged Cpt1a^CKO mice versus aged Cpt1a^fl/fl mice. Data presented as means±SD. n=3–7. *P < 0.05, **P < 0.01, ***P < 0.001 by one-way ANOVA with Fischer's LSD test (E, G, and H) or Mann–Whitney U unpaired t test (J). Scale bars=100 μm. Cpt1a, carnitine palmitoyltransferase 1A; Cpt1a^fl/fl, control Cpt1a^fl/fl mouse; Cpt1a^CKO, mouse with conditional tubular deletion of Cpt1a; Dox, doxycycline; H&E, hematoxylin and eosin; HFD, high fat diet; LSD, least significant difference; mo, month.

Kidney Cell-Specific Effects of Aging Plus HFD

To better investigate both the effects of HFD-aging on the kidney, as well as tubule-specific changes on Cpt1a deletion, we performed snRNA-seq on young and HFD-aged Cpt1a^CKO and Cpt1a^fl/fl mice. Clustering of sequenced single nuclei with the Seurat pipeline identified 16 clusters (Figure 2A) after quality control filters and batch correction with the Harmony package (Figure 2B and Supplemental Figure 2, A–C), with average transcript and gene counts >2500/cell and >2000/cell, respectively (Supplemental Figure 2B). Canonical cell type-specific gene markers^14–22 exhibited enriched expression in 75%–100% of their respective cell types, with very low expression in other clusters (Figure 2C).

Figure 2.

snRNA-seq shows increased inflammatory cells in HFD-aged mice. (A) Unsupervised clustering of barcoded and quality-controlled nuclei from kidneys of young Cpt1a^fl/fl (n=2) and Cpt1a^CKO (n=3) mice and HFD-aged Cpt1a^fl/fl and Cpt1a^CKO mice (n=3 each) in Seurat produced 16 renal and immune cell clusters. (B) Harmony-mediated batch correction integrated both age and genotype into cell type-specific clusters. (C) Canonical renal and hematopoietic cell marker genes were utilized to identify cell types. (D–H) Multiple immune cell clusters identified in macrophage (E and F) and lymphocyte (G and H) subclusters were predominantly derived from aged mice. Act. B cells, activated B cells; B&Tcells, B and T lymphocytes; CD-PC, collecting duct principal cells; CNT, connecting segment; DC, dendritic cells; DCT, distal convoluted tubule; DLH, descending Loop of Henle; EnC, endothelial cells; ICA, type A intercalated cells; ICB, type B intercalated cells; Inj. PT, injured proximal tubule; MC, mesangial cell; Mcphg, macrophage; NK, natural killer; PCT, proximal convoluted tubule; Podo, podocyte; PT, proximal tubule; snRNA-seq, singlenuclear RNA sequencing; TAL, thick ascending limb; T-regs, T regulatory cells; UMAP, uniform manifold approximation and projection.

To determine the changes induced by HFD aging, we compared data from 3-month-old and HFD-aged mice with Cpt1a intact. Consistent with F4/80 staining (Figure 1, F–H), the macrophage population came predominantly from the HFD-aged mice, as did the lymphocyte subcluster (B and T cells; Figure 2D). The macrophage population was subclustered to demonstrate that most of the increased macrophages in aged mice were resident macrophages (Figure 2, E and F and Supplemental Figure 2E). Similarly, most subsets of B and T cells also came from the HFD-aged mice (Figure 2, G and H and Supplemental Figure 2F). Thus, inflammation is broadly upregulated in HFD-aged compared with young mice.

A subpopulation of PT cells has been described as injured based on enriched expression of Vcam1, Havcr1, and Icam1 (Supplemental Figure 3). There were virtually no injured PT cells in the young mice, but HFD-aged mice had a significant subpopulation of injured PT cells identified as expressing markers of S1 and S3 (Figure 3, A and B and Supplemental Figure 3). To further explore how HFD aging differentially affects specific tubule segments, we analyzed upregulated inflammation-specific Kyoto Encyclopedia of Genes and Genomes ontology pathways (Figure 3C). Some inflammatory pathways were upregulated in both proximal and distal tubules (Hippo, advanced glycation end-products, autophagy, senescence), but the DCT had more inflammatory-related pathways upregulated than PT segments (Figure 3C). Most of these DCT-specific upregulated inflammatory pathways related to T cell activity or growth factor/proliferation signaling (e.g., TGF-β, ErbB, MAPK, and Ras signaling).

Figure 3.

HFD plus aging differentially affects proximal and distal tubules. Analysis of aged versus young Cpt1a^fl/fl clusters revealed enrichment of injured PT S1 and S3 segment cells in aged mice (A and B). Inflammation (C) and metabolism (D) KEGG pathway ontologies generated from statistically significant, upregulated DEGs of aged versus young Cpt1a^fl/fl mice. The different DEGs within the oxidative phosphorylation pathway that are differentially upregulated by PT and DCT segments (E). (F) KEGG pathway ontologies driven by down-regulated DEGs show a reduction in fatty acid metabolism in the S1 and S3 proximal segments. Ontologies and −log10 P values were derived in EnrichR from adjusted P values using default settings. DEG, differentially expressed gene(s); KEGG, Kyoto Encyclopedia of Genes and Genomes.

In contrast to the higher number of inflammatory pathways in the DCT, PT segments, particularly S1, upregulated more metabolism ontologies than the DCT (Figure 3D). Metabolic pathways affecting all substrates (fatty acid, glucose/pyruvate, and amino acids) were increased in PT segments. Although the oxidative phosphorylation pathway was upregulated in both proximal and distal tubules, the differentially expressed genes (DEGs) within this pathway were different between PT segments and the DCT (Figure 3E). Most of the downregulated pathways in PT segments related to metabolism, whereas the DCT had reduced expression of ontologies related to morphogenesis (Figure 3F). Within the HFD-aged PT segments, S1 and S2 downregulated pathways primarily involved in fatty acid metabolism. Taken together, these data indicate that HFD aging induces significant changes in the metabolic transcriptome of PT segments, whereas the DCT has changes related to inflammation and morphogenesis.

Effect of Tubular Cpt1a Deletion on HFD Aging

Deleting tubular Cpt1a increased macrophage infiltration (Figure 1, F–H). Analysis of HFD-aged macrophage subclusters revealed that most resident and infiltrating macrophages came from Cpt1a^CKO mice (Figure 4, A and B). Resident macrophages from HFD-aged Cpt1a^CKO mice exhibited significantly higher expression of M1 markers Cd86 (Figure 4C) and Il1β (Supplemental Figure 4A) relative to HFD-aged Cpt1a^fl/fl mice, whereas M2 markers such as Mrc1 (Figure 4C) were not significantly different.²³ This suggests that HFD-aged Cpt1a^CKO mice have more resident and infiltrating macrophages with a higher number of proinflammatory, M1 resident macrophages compared with HFD-aged Cpt1a^fl/fl mice.

Figure 4.

Characterization of macrophages and tubule-specific DEGs in aged Cpt1a^CKO mice. Macrophage subclusters in aged Cpt1a^CKO and Cpt1a^fl/fl mice were analyzed (A and B) with violin plots for M1 marker Cd86 and M2 marker Mrc1 (C). (D–G) The top 15 up-regulated DEGs in Cpt1a^CKO versus Cpt1a^fl/fl proximal and distal tubule segments. *P < 0.1×10⁻³. P values are Bonferroni-corrected adjusted P values derived from the Wilcoxon rank-sum test. Pct, percentage of cells.

To examine how deleting Cpt1a differentially affects proximal and distal tubules in HFD aging, we examined the top 15 upregulated genes in Cpt1a^CKO relative to Cpt1a^fl/fl mice. The most upregulated gene in both S1 and S2 PT segments was Hmgcs2, which had 3–4× increased expression (Figure 4, D and E). Hmgcs2 encodes for Hmgcs2, which is the rate-limiting enzyme in ketogenesis, but is also implicated in FAO.^24,25 Consistent with our previous data from young mice,¹ peroxisomal β-oxidation genes, including the rate-limiting Acox1 and Ehhadh, were among the most highly upregulated in HFD-aged Cpt1a^CKO PT S1 and S2 cells. In the DCT, the most upregulated gene was Pdk4, encoding pyruvate dehydrogenase kinase 4 (Figure 4G). Pyruvate dehydrogenase kinase 4 inhibits pyruvate dehydrogenase through phosphorylation and promotes pyruvate metabolism to lactate (i.e., anaerobic glycolysis) rather than oxidation in mitochondria.

Metabolic Adaptations to Cpt1a Deletion in PTs

To address the question of how PT segments compensate metabolically for the absence of Cpt1a, we focused on the metabolism-related DEGs significantly upregulated in these tubule segments (Figure 5A). Ppara, encoding PPARα, is a potent stimulator of mitochondrial and peroxisomal FAO and was upregulated in PT segments. As mentioned above, expression of peroxisomal FAO genes Acox1 and Ehhadh was upregulated in PT segments of Cpt1a^CKO mice. In addition, Abcd3, which encodes the transporter that imports fatty acids into the peroxisome, was increased, as was Pxmp4, which encodes for peroxisomal membrane protein 4, in HFD-aged PT segments of Cpt1a^CKO mice (Figure 5, A–F). Peroxisomal FAO has been suggested to be upregulated when mitochondrial import of LCFA (i.e., CPT1/2) is blocked.²⁶

Figure 5.

Metabolic changes in PT segments from aged Cpt1a^CKO mice. (A) Upregulation of metabolic DEGs in PT and DCT tubule segments from aged Cpt1a^CKO compared with Cpt1a^fl/fl mice. (B–F) Violin plots of upregulated genes associated with peroxisomal FAO in PT segments and (G) the glycolysis-regulating Pdk4 in the DCT. Violin plots (H) and immunoblots (I) show increased Hmgcs2 transcript and Hmgcs2 protein levels, respectively, in aged kidney cortices of Cpt1a^CKO and Cpt1a^fl/fl mice. Liver was used as a positive control for Hmgcs2. (J) IF staining demonstrated increased Hmgcs2 expression in LTL+ tubules of aged Cpt1a^CKO mice compared with aged Cpt1a^fl/fl mice. *P < 0.01×10⁻²⁰. Scale bars=100 μm. FAO, fatty acid oxidation; Hmgcs2, 3-hydroxyl-3-methylglutaryl CoA synthetase 2; IF, immunofluorescence; LTL, lotus tetragonolobus lectin.

In addition to genes related to peroxisomal FAO, PT segments also had transcriptional increases in fives genes of the cytochrome P450 CYP4A family (Figure 5A). The CYP4A family members can promote omega-oxidation of fatty acids,²⁷ which occurs in the endoplasmic reticulum and microsomes. Omega-oxidation involves adding a carboxyl group to fatty acids, starting with a hydroxylation step that can be catalyzed by CYP4A proteins. The resulting dicarboxylic acids can then be further oxidized by peroxisomal FAO.^27,28 This omega-oxidation pathway has been implicated as an adaptive metabolic pathway that may compensate when mitochondrial β-oxidation is blocked.^29,30 Thus, the upregulated expression of genes related to peroxisomal FAO and omega-oxidation in HFD-aged Cpt1a^CKO mice suggests that these pathways may be compensating for impaired mitochondrial import of LCFA.

Increases in Ppara and Hmgcs2, the rate-limiting enzyme for ketogenesis, also suggest that ketogenesis may be upregulated in PT segments from Cpt1a^CKO mice, although the role of ketogenesis in the kidney remains controversial.²⁴ These metabolic changes in peroxisomal FAO, ketogenesis, and CYP4A proteins were specific to PT segments and not present in other cell types in HFD-aged Cpt1a^CKO mice (Figure 5A and Supplemental Figure 4, B and C). In addition, these metabolic genes were less enriched in the injured PT cluster (Supplemental Figure 4B), suggesting that lack of metabolic adaptation to Cpt1a loss could cause injury. Violin plots also show upregulation of peroxisomal FAO-related genes (Figure 5, B–F) and CYP4A family genes (Supplemental Figure 4C). By contrast, Pdk4 was most highly enriched in DCT and thick ascending limb epithelia lacking Cpt1a (Figure 5A and Supplemental Figure 4B). Thus, deleting Cpt1a induced FAO-related metabolic changes in PT segments and more glycolytic pathways in distal segments.

We chose to validate Hmgcs2 as this was the most highly upregulated gene transcript in HFD-aged Cpt1a^CKO PT S1 and S2 segments (Figures 4, D and E and 5H). Protein expression of Hmgcs2 was significantly increased in cortical lysates of HFD-aged Cpt1a^CKO compared with Cpt1a^fl/fl mice (Figure 5I). Immunofluorescence showed that Hmgcs2 expression was predominantly in lotus tetragonolobus lectin+ (i.e., PT segments) and that upregulated protein expression required both conditional knockout of Cpt1a and HFD aging (Figure 5J). In addition to roles in ketogenesis and FAO, Hmgcs2-derived ketones can be used nonoxidatively for cholesterol or fatty acid synthesis.²⁴ Acat2 and Aacs, which promote the incorporation of ketones into fatty acids or cholesterol, respectively, were significantly upregulated in PT segments from Cpt1a^CKO mice (Supplemental Figure 4, D–G).^24,31,32 Urinary ketones did not show a difference in Cpt1a^CKO compared with Cpt1a^fl/fl mice (Supplemental Figure 5). There was a nonsignificant trend toward a reduction in urinary ketones in urine from HFD aged compared with young mice (Supplemental Figure 5). Thus, deleting Cpt1a led to upregulated Hmgcs2 gene and protein expression in HFD-aged PT segments, but its functional role needs further investigation.

Role of Cpt1a in HFD-Aged DCT

The DCT had the greatest protein and gene expression of Cpt1a¹ despite this segment being characterized as glycolytic, so we hypothesized that it might be playing a nonoxidative role in the DCT. Ontologic assessment of genes in HFD-aged Cpt1a^CKO DCT versus HFD-aged Cpt1a^fl/fl DCT revealed that four of the eight most significantly downregulated pathways are involved with kidney development (Figure 6, A and B). The DCT canonical cell markers Calb1 and Slc8a1 were among the most highly downregulated genes in aged Cpt1a^CKO DCT epithelia, as well as distal markers Klotho (Kl) and Umod, and the kidney development transcription factors Pax8 and Mtss1 were also downregulated (Figure 6, B–D). We verified that Calb1, encoding calbindin 1, was decreased at the level of protein expression in HFD-aged Cpt1a^CKO DCT (Figure 6, E and F). These data suggest that Cpt1a may be playing a role in maintenance of DCT cell differentiation.

Figure 6.

Distal tubules from aged Cpt1a^CKO mice have suppressed expression of differentiation markers. (A) Top GO biologic process ontologies derived from down-regulated DEGs of Cpt1a^CKO versus Cpt1a^fl/fl DCT. (B) A network plot of DEGs related to kidney and nephron development ontologies. (C and D) Dot plot and violin plots show downregulation of distal tubule marker genes and differentiation-related transcription factors in aged Cpt1a^CKO versus aged Cpt1a^fl/fl DCT. (E and F) IF images taken from serially sectioned tissues demonstrated localization of calbindin protein with tubules expressing the DCT marker NCC and reduced calbindin signal in Cpt1a^CKO mice versus Cpt1a^fl/fl mice. Data in (F) presented as mean±SD of N=3–4. *P < 0.01; **P < 0.01×10⁻²⁰. Scale bars=100 μm. GO, gene ontology; KO, Cpt1a^CKO; NCC, sodium chloride cotransporter; WT, Cpt1a^fl/fl.

Discussion

This work builds on prior data showing that deleting tubular Cpt1a had a minimal effect on aging and response to chronic injury.¹ Given the presence of increased lipid accumulation in Cpt1a^CKO mice, we aged these mice on a HFD to provide further metabolic stress. Then we assessed by snRNA-seq how the HFD-aging model affects the kidney and the tubule-specific effects of deleting Cpt1a. Although the HFD-aging model had greater tubular injury, assessed by KIM-1, than aging alone, the addition of HFD did not produce more marked changes between genotypes than what we previously published with aging on normal chow diet.¹ In both aging and HFD aging, the Cpt1a^CKO mice had increased lipid and macrophage accumulation, but no changes in fibrosis or kidney function compared with age-matched Cpt1a^fl/fl mice. The number of mice in these aging studies was small, and it is possible that with larger numbers, differences might emerge.

To the best of our knowledge, this is the first study to perform snRNA-seq on kidneys after aging plus HFD. The usual mouse chow has about 12%–14% of calories from fat, which is significantly less than the average Western diet, which is close to 40%. Although the HFD at 60% exceeds the fat intake for a Western diet, a noted limitation, aging on the regular chow diet may miss transcriptional differences relevant to humans. Our HFD-aging model showed increased inflammation, KIM-1 expression, and injured PT segments (Figures 1J and 2, D–H and Supplemental Figure 1E) compared with young mice, consistent with other studies examining kidney aging.^33,34 Some of the increased DEGs in HFD-aged tubules were cytokines and growth factor signaling pathways (e.g., advanced glycation end-products signaling, Hippo signaling, and EGF) that have been described in either kidney aging (human and mice) or murine kidney injury, suggesting that aging may induce similar transcriptional effects in kidney tubules, as does injury.^33,35 PT segments of HFD-aged mice showed significant downregulation of metabolism-related ontologies. Although some of these changes could be enriched by the HFD, this is consistent with a previous study using transcriptomics and proteomics to show downregulation of fatty acid metabolic processes in the aged kidney.³⁶

Single-cell/nuclear transcriptomics provide an ideal tool to address the question of how Cpt1a null PT cells maintain metabolic health. Our data strongly suggest that both peroxisomal FAO and omega-oxidation pathways are upregulated in the absence of Cpt1a. Our current data (Figure 5A) and prior bulk RNA-seq data¹ suggest that Cpt1a^CKO mice have increased PPARα activity/gene expression. PPARα is an established transcriptional promoter of peroxisomal FAO and CYP4A proteins that mediate omega-oxidation,²⁸ so it is possible that PPARα is driving this metabolic compensation in Cpt1a^CKO mice. Our prior data also identified upregulated peroxisomal FAO genes in young Cpt1a^CKO mice.¹ Peroxisomes usually prefer very LCFAs rather than the LCFA typically imported through Cpt1a into mitochondria for beta-oxidation. However, others have shown that, in the absence of a functional carnitine shuttle (i.e., CPT1/2), peroxisomes may compensate by taking up LCFA and oxidizing them to medium chain fatty acids, which can enter mitochondria without the need for Cpt1a.^26,37 Consistent with our prior data showing increased peroxisomal FAO gene expression in PT segments of young Cpt1a^CKO mice (i.e., Acox1, Ehhadh, Abcd3), this study confirms these findings while adding Ppara and Pxmp4 as peroxisome-related genes upregulated in HFD-aged Cpt1a^CKO mice. We previously showed that this increase was associated with lower kidney concentrations of very LCFAs,¹ exclusively metabolized in peroxisomes, an indirect measure of peroxisomal FAO. Further knockout studies are needed to confirm functional compensation by peroxisomal FAO.

Another pathway of metabolic compensation in Cpt1a^CKO mice is the CYP4A family. Several DEGs in the CYP4A family (Figure 5A and Supplemental Figure 4C) were significantly upregulated in HFD-aged PT segments from Cpt1a^CKO mice. The CYP4A family members are known promoters of fatty acid omega-oxidation, a relatively understudied process. This process is believed to occur primarily in the liver and kidney and has been described in conditions in which there is a mitochondrial FAO defect.^28,30 In such cases, the resulting dicarboxylic acids undergo fatty acid β-oxidation by the peroxisomes.^38,39 Thus, peroxisomal FAO and omega-oxidation may be working together to overcome the metabolic defect of Cpt1a deletion. Future studies are necessary to determine the functional roles of these pathways.

The most significantly upregulated gene in PT (S1 and S2) from HFD-aged Cpt1a^CKO mice was Hmgcs2, and increased PT-specific protein expression was confirmed (Figure 5, I and J). Although Hmgcs2 is the rate-limiting enzyme for ketogenesis, the liver is the primary organ responsible for ketone production. It is controversial whether the kidney is capable of ketogenesis.²⁴ Fasting is a potent stimulus for ketogenesis, and others have shown that fasting induces Hmgcs2 expression in renal PT segments.⁴⁰ However, organ-specific knockout models of Hmgcs2 revealed that the kidney does not contribute to systemic ketone production, implying a ketogenesis-independent role of Hmgcs2 in the PT segments.⁴⁰ It is also possible that ketogenesis does occur in the PT, but the resulting ketones are either metabolized within the kidney or used to generate cholesterol or lipids.^24,31,32 Ppara and Hmgcs2, both upregulated in PT segments of Cpt1a^CKO mice, can induce expression of each other, as well as stimulate FAO.^25,41,42 It is possible that the increased lipid accumulation in Cpt1a^CKO mice stimulates Hmgcs2, potentially through PPARα, to reduce the lipid burden through ketone production and oxidation or cholesterol synthesis, but further studies are necessary to confirm. We have identified Hmgcs2 as being increased in HFD-aged PT segments of Cpt1a^CKO mice, but the exact role of this protein requires additional investigation.

In contrast to the PT segments in which Cpt1a deletion led to significant alterations in multiple metabolism-related DEGs, the DCT had far fewer transcriptional changes in metabolism-related genes (Figures 4, D–G and 5A). In addition, the PT segments from HFD-aged Cpt1a^CKO mice primarily upregulated FAO-related genes (peroxisomal FAO, omega oxidation, and ketogenesis), whereas the DCT's top upregulated DEG, Pdk4, relates to glycolysis (Figure 4G). We previously identified Pdk4 as an upregulated target in bulk RNA-seq on aged Cpt1a^CKO mice and confirmed this at the protein level.¹ Our snRNA-seq data defines the DCT as the cellular source for Pdk4 upregulation (Figure 5A).

In the DCT, the tubular segment with the highest CPT1A expression,¹ genes related to kidney development and cellular differentiation were significantly downregulated in DCT of HFD-aged Cpt1a^CKO mice. Within these ontologies, the expression of several genes (e.g., Calb1, Slc8a1, and Pax8) related to cell differentiation were downregulated, suggesting that Cpt1a is needed to maintain healthy, differentiated DCTs. The functional effects of reduced expression of DCT markers in HFD-aged Cpt1a^CKO mice should be investigated in future studies.

In summary, we provide the first snRNA-seq data of kidneys that have been aged 24 months while on HFD. Cpt1a^CKO PT segments had significantly upregulated expression of genes involved in peroxisomal FAO and omega-oxidation (CYP4A family), as well as increased gene and protein expression of Hmgcs2, changes that may compensate for impaired mitochondrial long chain FAO. Cpt1a deletion in the DCT did not induce as many metabolic changes as in the PT, but did impair maintenance of its differentiated state, suggesting nonoxidative roles for Cpt1a in the distal tubule.

Disclosures

Disclosure forms, as provided by each author, are available with the online version of the article at http://links.lww.com/KN9/A960.

Funding

L.S. Gewin: National Institute of Diabetes and Digestive and Kidney Diseases (R01DK108968 and 1U54DK137332), US Department of Veterans Affairs (BX03425), and Longer Life Foundation. J.R. Koenitzer: National Institute of Diabetes and Digestive and Kidney Diseases (5K08HL159418-03). B.D. Humphreys: National Institute of Diabetes and Digestive and Kidney Diseases (UC2DK126024 and 1U54DK137332), Janssen Pharmaceuticals, Pfizer, and Chinook Therapeutics. R. Zent: National Institute of Diabetes and Digestive and Kidney Diseases (DK069921, DK088327, and DK127589) and US Department of Veterans Affairs (BX002196). S.C. Huen: National Institute of Diabetes and Digestive and Kidney Diseases (R35GM137984, R56DK134582, and R01DK135555).

Author Contributions

Conceptualization: Leslie S. Gewin.

Data curation: Steven D. Funk, Leslie S. Gewin, Justin T. Kern, Jeffrey R. Koenitzer, Elizabeth Sulvaran-Guel.

Formal analysis: Steven D. Funk, Jeffrey R. Koenitzer.

Funding acquisition: Leslie S. Gewin.

Investigation: Steven D. Funk, Leslie S. Gewin.

Methodology: Jacob S. Blum, Kyle C. Feola, Steven D. Funk, Leslie S. Gewin, Sarah C. Huen, Benjamin D. Humphreys.

Project administration: Leslie S. Gewin, Justin T. Kern.

Resources: Sarah C. Huen, Benjamin D. Humphreys, Olga M. Viquez, Roy Zent.

Software: Jacob S. Blum, Jeffrey R. Koenitzer, Elizabeth Sulvaran-Guel.

Supervision: Leslie S. Gewin.

Writing – original draft: Steven D. Funk.

Writing – review & editing: Leslie S. Gewin, Sarah C. Huen, Benjamin D. Humphreys, Justin T. Kern, Olga M. Viquez, Roy Zent.

Data Sharing Statement

Data related to transcriptomic, proteomic, or metabolomic data. Raw Data/Source Data. Gene Expression Omnibus. FASTQ files from snRNA-seq studies are available at the Gene Expression Omnibus repository, accession #GSE277335.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/KN9/A961.

Supplemental Methods

Supplemental Figure 1. Renal function was not significantly affected by aging or selective deletion of Cpt1a.

Supplemental Figure 2. Robust quality control in single-nucleus RNA sequenced samples.

Supplemental Figure 3. Injured PT cells were comprised predominately of segments S1 and S3.

Supplemental Figure 4. Inflammatory markers and tubular metabolic changes in Cpt1a^CKO kidneys.

Supplemental Figure 5. Urinary β-hydroxybutyrate.