One Million Exomes Identify FNIP1, a Metabolic Brake

Abstract

Exome sequencing of over one million people recently uncovered dozens of genes linked to a lipid-based biomarker of body energy state. Rare loss-of-function variants in FNIP1, a suppressor of mitochondrial energy expenditure, were associated with markedly lower cardiometabolic disease risk, nominating selective FNIP1 inhibition as a potential therapeutic strategy.

Genome-wide association studies (GWAS) have transformed our understanding of the genetic architecture of human diseases, uncovering hundreds of genomic regions associated with circulating lipid levels and cardiometabolic disease risk [1]. However, conventional microarray-based GWAS are inherently limited in their ability to assess rare variation. These studies directly genotype hundreds of thousands of predominantly common variants and statistically impute tens of millions of additional variants using reference panels. Although highly effective for detecting common associations, this framework provides incomplete coverage of rare and ultra-rare variants. This limitation is consequential because rare protein-coding variants can exert large phenotypic effects and more directly implicate causal genes and the biological consequences of their perturbation. Large-scale exome sequencing addresses this gap by directly interrogating protein-coding variation across the allele-frequency spectrum. In a study of more than one million exomes, Hindy et al. identify FNIP1 as a key regulator of energy metabolism [2].

Over the past decade, advances in DNA sequencing have enabled large-scale exome sequencing among participants in national biobanks and health systems. These datasets support gene-based burden tests that aggregate rare coding variants within genes to boost statistical power [3,4], as well as single-variant association analyses of even ultra-rare alleles [5]. Such studies have nominated potential therapeutic targets for cardiometabolic diseases. Predicted loss-of-function (pLoF) variants in GPR75 are associated with lower obesity risk [3], whereas a pLoF variant in RORC is associated with lower low-density lipoprotein cholesterol levels and coronary artery disease risk [5]. Protective pLoF associations such as these are rare but exceptionally valuable. They provide naturally occurring, lifelong experiments in target inhibition, offering human genetic evidence that pharmacologically mimicking the same loss-of-function may be effective and well tolerated.

Hindy et al. pushed this approach to an unprecedented scale, reporting an exome-sequencing analysis of 1,032,116 individuals from 11 cohorts spanning three continents [2]. The authors focus on the ratio of triglycerides to high-density lipoprotein cholesterol (TG:HDL). TG:HDL is a simple, widely available biomarker that tracks closely with visceral adiposity, liver fat, and insulin resistance, as well as the incidence of cardiometabolic diseases across five ancestry groups. These epidemiological findings position TG:HDL not merely as a lipid ratio but as an integrative proxy for whole-body metabolic state, supporting its use as a phenotype for gene discovery.

Gene-based burden analysis of rare coding variants identified 59 genes independently associated with TG:HDL, 44 of which had not been previously implicated. This markedly expands on the largest prior rare-variant study of TG:HDL, which was conducted in approximately 170,000 exomes [6]. The identified genes were enriched for those expressed in the liver and adipose tissue, as well as for approved or clinical-stage drug targets, indicating that TG:HDL captures a biologically coherent and druggable network.

Among the new findings, FNIP1 stands out for its combination of rarity, effect size, and mechanistic tractability. Ultra-rare protein-truncating variants in FNIP1, carried by roughly 1 in 7,000 sequenced individuals, were associated with a markedly lower TG:HDL, reduced liver fat, lower glycated hemoglobin (HbA1c), more favorable fat distribution, and approximately 60% lower odds of cardiometabolic disease outcomes. FNIP1 encodes a binding partner of folliculin (FLCN) that normally restrains mitochondrial biogenesis and energy expenditure by suppressing the transcription factors TFEB and TFE3. Heterozygous loss of FNIP1 may release this metabolic brake, promoting mitochondrial and lysosomal gene expression and lipid catabolism. Notably, complete biallelic loss of FNIP1 causes a severe immunodeficiency syndrome, yet heterozygous carriers in this study showed no significant association with a composite phenotype capturing these manifestations. This gene-dosage distinction highlights a critical consideration for any future FNIP1-targeted therapies.

The authors strengthen the genetic findings through mechanistic experiments. Silencing FNIP1 in primary human hepatocytes induced lysosomal and lipid-catabolism genes, supporting the proposed mechanism. In mice, liver-specific inhibition of Flcn, or combined inhibition of Fnip1 with its paralog Fnip2, protected against diet-induced weight gain, reduced hepatic triglycerides, and improved insulin sensitivity. Because hepatocyte-selective siRNA therapeutics are already approved for other targets, the authors propose that liver-specific FNIP1 silencing could replicate the metabolic benefits seen in heterozygous carriers while avoiding the potential side effects of systemic loss-of-function, including the immunodeficiency caused by homozygous FNIP1 deficiency [7] and Birt–Hogg–Dubé syndrome caused by FLCN haploinsufficiency [8].

Several questions remain. Selective hepatic inhibition of Fnip1 alone did not reproduce the phenotype in mice, and combined inhibition of Fnip1 and Fnip2 was required. This discrepancy may reflect species differences in functional redundancy between FNIP1 and FNIP2, underscoring the limitations of extrapolating rodent biology to humans. It also remains unclear whether the liver alone, or a combination of metabolically active tissues, should ultimately be targeted. Long-term liver safety also warrants scrutiny. A recent report found that hepatic Flcn loss can predispose mice to liver injury and cholangiocarcinoma [9]. The present study’s shorter-term experiments and existing human carrier data cannot definitively exclude this risk.

Nonetheless, this study compellingly demonstrates how million-scale exome sequencing, paired with mechanistic follow-up, can identify a new therapeutic target for cardiometabolic diseases that remain leading causes of death worldwide (Figure).

Figure. Human genetic evidence combined with cellular and animal experiments identifies a novel drug target for metabolic disease.

Exome sequencing of more than one million individuals identified ultra-rare loss-of-function variants in FNIP11, carried by roughly 1 in 7,000 people, that are protectively associated with a lower triglyceride-to-HDL-cholesterol ratio (TG:HDL), a biomarker of whole-body metabolic state. Inhibition of FNIP1 in cellular models induced lipid breakdown, and inhibition of Fnip1 and Fnip2, or Flcn in mice reduced diet-induced obesity, recapitulating the favorable metabolic phenotype observed in human variant carriers. Together, this convergence of human genetic, cellular and animal evidence identifies FNIP1 as a promising novel therapeutic target for metabolic disease. Created in BioRender. Enzan, N. (2026) https://BioRender.com/lrrd9f7.

The implications extend beyond FNIP1. This study linked a widely available biomarker to rare coding variants across a cohort large enough to catch even ultra-rare, large-effect alleles. This approach is agnostic to the disease in question. Exome and genome cohorts are growing and are increasingly linked to a widening array of phenotypes beyond lipids. As this phenotypic net widens, similar strategies may uncover additional protective coding variants and genetically validated therapeutic targets across a broad range of diseases. This study thus extends the reach of human genetics-guided drug discovery well beyond the story told here.