Writing and erasing ceramides to alter liver disease

Abstract

M⁶A RNA modifications mediate RNA processing and stability. Ceramides are lipid metabolites containing an amino acid-based backbone, which promote metabolic dysfunction. Wang et al. describe a novel m⁶A-dependent regulatory node that tunes ceramide-generating enzymes.

The accumulation of ectopic lipid metabolites in tissues not suited for fat storage drives many of the cellular dysfunctions that underly diabetes and non-alcoholic fatty liver disease (NAFLD). Of the numerous types of lipids that accumulate, sphingolipids such as ceramides are amongst the most deleterious, as they modulate signalling and metabolic pathways to drive insulin resistance, apoptosis and fibrosis¹. Owing to a strong association of serum ceramides with insulin resistance and major adverse cardiac events, the use of serum ceramides as a clinical biomarker has increased². Yet, the regulatory mechanisms governing changes in ceramide biosynthesis remain minimally described. In this issue of Nature Metabolism, Wang et al. elucidate a novel mechanism whereby the postnatal liver fine-tunes sphingolipid metabolism via methylation of mRNA transcripts encoding ceramide synthesis enzymes³. In turn, altered balance of m⁶A methylation tips the scale towards ceramide overaccumulation, mitochondrial damage and endoplasmic reticulum stress (Fig. 1).

Fig. 1 |. By altering the stability of mRNA transcripts, Mettl3–m⁶A alters sphingolipid metabolism.

Schematic of ceramide synthesis via de novo, salvage and sphingomyelinase pathways (left). Enzyme colour denotes the degree of expression change in Mettl3^ΔHep livers. Visual representation of Mettl3^ΔHep model (right), whereby constitutive liver-specific knockout of Mettl3 reduces methylation of sphingolipid-related mRNA such as Smpd3 mRNA, diminishing transcript decay and leading to an upregulation of ceramide-generating enzymes, ultimately contributing to liver injury.

N⁶-methyladenosine (M⁶A) RNA modifications are a recently discovered regulator of RNA processing and stability. They are deposited by a methyltransferase enzyme complex, which includes methyltransferase-like 3 and 14 (METTL3 and METTL14) in association with Wilms’ tumour 1-associating protein (WTAP), which serves as an accessory factor. Often referred to as ‘writers’, these transferases catalyse methylation of the nitrogen atom at position 6 of the RNA adenine nucleotides of select mRNA transcripts⁴. By contrast, demethylase enzymes capable of reversing or ‘erasing’ this modification include alkB homologue 5 (ALKBH5) and fat-mass and obesity-associated protein, which has long been noted for strong genetic correlations with obesity and metabolic disease⁵. As functional sensors of M⁶A modifications, the YTH domain family of proteins promote translation, decay and translocation of M⁶A-modified mRNA transcripts.

Wang et al. use models of Mettl3 deletion to elucidate a mechanism whereby the postnatal liver keeps sphingolipid metabolism in check to allow for normal maturation through a fine-tuned M⁶A-driven mechanism³. Through the constitutive or inducible ablation of Mettl3 selectively in mouse liver, Wang and colleagues identify that ceramide synthetic pathways are coordinately regulated via M⁶A methylation. In particular, they highlight a novel mettl3–m⁶A–neutral sphingomyelinase 3 (Smpd3) axis that allows for the proper postnatal liver maturation via controlling Smpd3 expression levels and, thereby, ceramide levels during a critical developmental stage.

Despite being less abundant than glycerolipids, sphingolipids play crucial roles in cell structure, signalling and cell cycle regulation, and are essential for life at all developmental stages. As opposed to commonly considered glycerolipids, sphingolipids lack a glycerol moiety and instead contain a sphingoid backbone derived from the condensation of an amino acid with a long chain fatty acyl CoA — canonically an addition of palmitoyl-CoA onto serine. This committed step in the de novo synthesis of ceramides is catalysed by a serine palmitoyl-transferase complex consisting of SPTLC1 and SPTLC2. Ceramides can additionally be formed by a salvage pathway (which re-acylates the sphingosine backbone) or by hydrolytic action of sphingomyelinase enzymes (such as Smpd3), which removes the choline headgroup from sphingomyelin, to regenerate ceramide⁶.

Although ceramides are the key building block for all complex sphingolipids, their overaccumulation may serve as an evolutionarily conserved sensor of nutrient excess. They commonly accumulate during overnutrition or cellular stress. In humans, liver ceramides strongly correlate with insulin resistance and steatosis in NAFLD patients who do not carry the common PNPLA3^I148M variant that contributes to steatosis. In preclinical models, liver-specific degradation of ceramides or interruption of ceramide synthesis prevents hepatic steatosis.

Mechanistically, hepatic ceramides promote fatty acid uptake via translocation of CD36, induce triglyceride synthesis via enhanced Srebf1 expression and upregulate gluconeogenesis by impeding AKT phosphorylation⁷. C16:0 ceramides, which contain a 16-carbon acyl chain attached to the sphingoid backbone, have proved particularly deleterious to liver function by promoting mitochondrial fission and decreasing mitochondrial efficiency at the level of the electron transport chain^8,9.

Wang et al. analysed a publicly available transcriptional dataset of C57BL6/J mouse livers and noticed a temporal expression of methyltransferase complex enzymes during the prenatal to postnatal shift. Methyltransferase differential expression coincides with a critical window of liver growth and maturation. METTL3 and METTL14 enzymes show increased expression in embryonic and neonatal murine hepatocytes and decreased expression, along with liver m⁶A levels, at 3–4 weeks of age. The downregulation of these genes is concomitant with the upregulation of hepatic metabolic genes. This curious observation propelled Wang et al. to probe further into the role of the catalytically active Mettl3 in postnatal liver development.

The researchers used a constitutive murine genetic knockout model of Mettl3 driven by the Albumin Cre promoter (hereafter referred to as Mettl3^ΔHep), effectively reducing m⁶A levels in the Mettl3^ΔHep mouse livers. They discovered that depleting this methyltransferase enzyme in the prenatal period led to stunted growth, hepatocyte hypertrophy and liver injury. Interestingly, inducible deletion of Mettl3 1 week after birth using adeno-associated virus serotype 8 (AAV8) carrying a thyroid-binding globulin promoter (hereafter AAV8-Cre) did not recapitulate the phenotype seen with prenatal deletion. Neither was liver injury seen with tamoxifen-induced deletion of Mettl3 4 weeks after birth using Albumin Cre^ERT2 described by Xu et al.¹⁰. Wang et al. deciphered the contradiction between the phenotype achieved by the constitutive and inducible models by examining hepatic m⁶A levels in their AAV8-Cre mice livers and comparing them to that of the Mettl3^ΔHep mice. They found that AAV-driven deletion of Mettl3 was incomplete 2 weeks after AAV injection (3 weeks of age), illustrated by significant m⁶A levels in the AAV8-Cre livers compared to the Mettl3^ΔHep livers. The differences between the models further supported Wang et al.’s hypothesis that m⁶A levels play a crucial role in supporting normal hepatic maturation in this developmental stage.

Transcriptional analysis revealed that this liver damage phenotype also coincided with the upregulation of sphingolipid-related enzymes, including Smpd3. The m⁶A modification is generally deposited on the consensus GGAC motif and primarily targets areas around stop codons, a highly conserved phenomenon in humans². Wang et al. illustrated that decreased levels of m⁶A in the locus of Smpd3 around its stop codon led to reduced decay of the RNA transcript in the postnatal period in their Mettl3^ΔHep mice, leading to upregulated Smpd3 expression and aberrant accumulation of toxic ceramides.

The ceramides that appeared to accumulate in Mettl3^ΔHep mice included the C16:0 chain length, which is associated with disease and mitochondrial dysfunction^8,9,11. Interestingly, mitochondria from primary hepatocytes originating from the Mettl3^ΔHep mice resembled the fragmented mitochondria observed by Hammerschmidt et al. in their Hepatic CerS6 overexpression model, which is the ceramide synthase enzyme responsible for synthesizing the C16:0 ceramides through the addition of a 16-carbon acyl chain to the sphingoid backbone. In addition, the Mettl3^ΔHep hepatocytes also had signs of mitochondrial and endoplasmic reticulum unfolded protein response, apoptosis and elevated reactive oxygen species.

The researchers also performed rescue experiments with pharmacological inhibition of Smpd3, Smpd3 knockdown via siRNA or overexpression of sphingomyelin synthase 1, which catalyses the reverse reaction to Smpd3. These experiments decreased levels of ceramides and attenuated liver damage, suggesting that Smpd3 upregulation and subsequent aberrant accumulation of ceramides led to liver damage due to reduced m⁶A deposition and decreased Smpd3 transcript decay.

According to recent studies, the effects of Mettl3 deletion are controversial. For example, Li et al. reported no developmental defects in their liver-specific Mettl3 knockout model driven by the Albumin Cre promoter¹². Studies like this suggested that there may be significant Mettl3-independent m⁶A deposition. Wang et al. suggested that the discrepancy in their findings with Li et al. may be due to the different exons targeted by the genetic manipulation between the models. While their study had Mettl3 exon 4 floxed, Li et al. had exon 2 floxed. Indeed, a recent study by Poh et al. compared the two models and found that the protein product of exon 2-deleted cells retained some catalytic activity and corresponded to a higher level of m⁶A modification, unlike the exon 4-deleted cells, which had a non-catalytically active protein product. This finding confirms that Mettl3 is the primary enzyme responsible for most m⁶A transcriptomic modifications.

The work by Wang and colleagues raises exciting unanswered questions regarding the other sphingolipid enzymes upregulated in the Mettl3^ΔHep model, including serine palmitoyl transferase 1 and 2 (SPTLC1/2) and ceramide synthase 5 and 6 (CerS5/6). Though strong upregulation of these de novo ceramide synthesis genes was also observed in Mettl3^ΔHep mice, the contribution of de novo ceramide synthesis to hepatic dysfunction was not formally interrogated. While the knockdown of Smpd3 did alleviate the liver damage phenotype, it didn’t entirely rescue it. The authors speculate that upregulation of de novo ceramide synthesis occurred with Mettl3 deletion leading to liver damage, which Smpd3 worsened. Further research into the role of m⁶A-dependent and independent mechanisms leading to the upregulation of these enzymes would provide additional evidence for epitranscriptomic sphingolipid regulation.

In summary, Wang et al. elucidate a novel role of m⁶A modification in curating physiological homeostasis in the postnatal developmental stage to allow for proper hepatic maturation. These findings reveal a previously unknown mechanism of sphingolipid metabolism regulation. Effective therapeutic strategies to erase ceramide-producing enzymes may benefit the treatment of cardiometabolic diseases, while tuning a local increase of ceramides within tumours could provide new pro-apoptotic therapies.