Lipid droplet-associated lncRNA LIPTER preserves cardiac lipid metabolism

Lipid droplets (LDs) are cellular organelles critical for lipid homeostasis, with intramyocyte LD accumulation implicated in metabolic disorder-associated heart diseases. Here we identify a human long non-coding RNA, Lipid-Droplet Transporter (LIPTER), essential for LD transport in human cardiomyocytes. LIPTER binds phosphatidic acid and phosphatidylinositol 4-phosphate on LD surface membranes and the MYH10 protein, connecting LDs to the MYH10-ACTIN cytoskeleton and facilitating LD transport. LIPTER and MYH10 deficiencies impair LD trafficking, mitochondrial function and survival of human induced pluripotent stem cell-derived cardiomyocytes. Conditional Myh10 deletion in mouse cardiomyocytes leads to LD accumulation, reduced fatty acid oxidation and compromised cardiac function. We identify NKX2.5 as the primary regulator of cardiomyocyte-specific LIPTER transcription. Notably, LIPTER transgenic expression mitigates cardiac lipotoxicity, preserves cardiac function and alleviates cardiomyopathies in high-fat-diet-fed and Leprdb/db mice. Our findings unveil a molecular connector role of LIPTER in intramyocyte LD transport, crucial for lipid metabolism of the human heart, and hold significant clinical implications for treating metabolic syndrome-associated heart diseases. Han et al. identify the long non-coding RNA LIPTER as a key mediator of lipid droplet transport and metabolism in human cardiomyocytes. LIPTER overexpression mitigates cardiomyopathy and preserves cardiac function in obese and diabetic mouse models.

Lipid droplets (LDs) are highly dynamic cellular organelles ubiquitously existing in prokaryotes and eukaryotic cells. Synthesized in the endoplasmic reticulum (ER), LDs consist of a neutral lipid core, primarily containing triacylglycerols (TAGs) and sterol esters, surrounded by a phospholipid monolayer with associated proteins 1 . Although initially considered storage reservoir for excess lipids, LDs have been increasingly recognized for their critical roles in intracellular lipid trafficking, lipid homeostasis and membrane synthesis 1,2 . Under hyperlipidaemic conditions, LDs accumulate in the myocardium (that is, cardiac steatosis) and cardiomyocytes (CMs) 2 . While increased LD levels afford transient cardiac protection by storing surplus fatty acids (FAs) in CMs 3,4 , long-term lipid intermediary accumulation results in detrimental outcomes, such as apoptosis, tissue injury and cardiac dysfunction (that is, lipotoxicity) 2,5,6 . Intramyocyte LD accumulation in humans is associated with cardiomyopathies and heart failure in hyperlipidaemia-related metabolic disorders, including obesity and diabetes mellitus 6,7 . Animal model studies have also revealed that an expanded pool of LDs can lead to early impairment of cardiac contractility and subsequent heart dysfunction [8][9][10] . Despite these findings, the CM intrinsic mechanisms determining LD accumulation and mobilization, as well as the aetiology of intramyocyte LD accumulation in obesity and diabetes mellitus, remain poorly understood.
LDs play a vital role in cellular metabolism, particularly under nutrient-deprived conditions. Under starvation conditions, LDs are mobilized and transported to mitochondria for β-oxidation to generate energy for the cell 11 . TAG within LD core is digested to FAs via Article https://doi.org/10.1038/s41556-023-01162-4 type 2 diabetes (T2DM) (3 individuals per group) were compared. We identified the top four CM-enriched lncRNAs, LINC00881, TTN-AS1, SLC8A1-AS1 and NAV2-AS2, which were downregulated in T2DM hearts compared with non-T2DM hearts (Fig. 1a,b and Extended Data Fig. 1a). However, RT-qPCR validation confirmed that only LINC00881 expression was significantly reduced to less than 50% (P < 0.05) in all collected T2DM hearts compared with non-T2DM hearts ( Fig. 1c and Extended Data Fig. 1b). Multiple tissue-specific gene expression databases demonstrated that LINC00881 expression is highly enriched in the human hearts ( Fig. 1d and Extended Data Fig. 1c-e). Additionally, the analysis of single-cell RNA sequencing (scRNA-seq) data from a 6.5-7 week human embryonic heart 25 revealed that LINC00881 expression was specifically enriched in the NKX2.5 + and cardiac Troponin T (cTnT) + CM clusters (Fig. 1e), with an expression level comparable to the key cardiac transcription factor (TF) NKX2. 5 (ref. 26) (Fig. 1f). LINC00881 is conserved in human and non-human primates but not across other species (Fig. 1g). Although three putative open reading frames (ORFs) were predicted in LINC00881 by ORF-FINDER 27 (Extended Data Fig. 2a), the precited peptides were not detected using western blotting or immunostaining (Extended Data Fig. 2b,c), albeit increased RNA level of LINC00881 detected in transfected 293T cells (Extended Data Fig. 2d). Altogether, these results demonstrate that LINC00881 is highly and specifically enriched in human CMs and significantly downregulated in human T2DM hearts. In this study, LINC00881 was renamed LIPTER due to its role in LD transport within human CMs.

LIPTER deficiency impairs LD metabolism and transport
During CM differentiation from human induced pluripotent stem cells (hiPSCs) by forming embryoid bodies 23 (EBs; Fig. 1h), LIPTER expression rapidly increased by over 10,000-fold (CMs versus hiPSCs), similar to NKX2.5 (Fig. 1i). Next, two LIPTER knockout (LIPTER KO ) hiPSC clones were established by CRISPR/Cas-9 (ref. 28) (Fig. 1j,k and Extended Data Fig. 3a). LIPTER KO did not affect the ratios of beating EBs and cTnT + CMs or the expression levels of CM markers, cTnT and MYH6, at day 20 of CM differentiation (Extended Data Fig. 3b-e), indicating that LIPTER is not required for CM differentiation. However, since LIPTER expression declines in T2DM hearts (Fig. 1c), we cultured hiPSC-EBs with high glucose (22.75 mM) for an additional 20 days. At day 40, significantly reduced ratios of cTnT + CMs were observed in LIPTER KO compared with wild-type (WT) hiPSC-EBs ( Fig. 1l and Extended Data Fig. 3f), suggesting that LIPTER is required for CM survival under high-glucose condition and implicating a potential role of LIPTER in CM metabolism. To investigate global changes in metabolites upon LIPTER deficiency, untargeted metabolomics ( Fig. 2a and Supplementary Table 1) were performed on enriched day 40 hiPSC-CMs. The top increased metabolites in LIPTER KO compared with WT hiPSC-CMs were lipid and lipid-like molecules, including phosphatidylinositol (PI), PAs and TAGs. Notably, TAG is the main core lipid of LDs 1,29 , while PI and PA are phospholipids on LD membranes 29 . These data suggest increased lipolysis and lipophagy processes on LDs 1,12 . The extensive interactions of LDs with various organelles, such as the ER, endosomes and mitochondria, facilitate intracellular lipid trafficking and channelling 13 . LD-associated proteins mediate LD-organelle membrane contacts. For example, FATP1 and DGAT2 at the LD-ER interface regulate LD expansion 14 , while LD-mitochondria tethering in skeletal and heart muscle is mediated by PLIN5 for channelling FAs from LDs to mitochondria 15 . Although LD biogenesis, composition and turnover have been extensively studied, the molecular mechanism underlying LD transport from the ER to other organelles remains elusive. Previous studies have implicated myosin motors, such as Myosin I and V, in the transport of ER-derived vesicles 16,17 . Additionally, Myh9 depletion has been shown to impair LD dissociation from the ER 18 . Mammalian CMs predominantly express non-muscle myosin IIB (Myh10) (ref. 19), which can interact with membrane lipids 20 and link cytoskeleton with the ER 21 . While the key roles of protein factors in intracellular LD trafficking have been extensively studied, the direct involvement of RNA molecules in this process remains unknown. Gaining insight into this aspect is crucial for understanding intramyocyte lipid trafficking and channelling, which are essential for maintaining lipid homeostasis in the human heart.
Although RNA is primarily known for carrying and regulating genetic information, recent evidence indicates that RNA interacts with various molecules, such as nucleic acids, ions, proteins and lipids 22 , to perform diverse functions. RNA-lipid interactions have revealed new regulatory mechanisms in essential cellular processes, implying previously undefined functions of RNA. In this study, we identified a human long non-coding RNA (lncRNA), named Lipid-Droplet Transporter (LIPTER), which facilitates LD transport within human CMs. LIPTER deficiency impairs LD transport and metabolism, mitochondrial function and CM viability. Mechanistically, LIPTER directly binds two phospholipids on LD membrane, phosphatidic acid (PA) and phosphatidylinositol 4-phosphate (PI4P), as well as the MYH10 motor protein, connecting LDs with cytoskeleton for intramyocyte transport. Importantly, LIPTER overexpression mitigates cardiomyopathies and preserves cardiac functions in mouse models of obesity and diabetes, highlighting LIPTER's potential clinical relevance in treating human metabolic syndrome-associated heart disease and failure. Our findings reveal that RNA can directly participate in intracellular vesicle transport, expanding RNA's functional dimensions and suggesting a potential ancient role of RNA-lipid interaction in forming the first living system in the primordial 'RNA world'.

LINC00881 (LIPTER) is specifically expressed in human CMs
Using our previously established method 23 , human pluripotent stem cells (hPSCs) were differentiated into cardiovascular cells. Transcriptomic profiles of hPSCs, hPSC-derived multipotential cardiovascular progenitors (MCPs), CMs, smooth muscle cells (SMCs), endothelial cells (ECs) 24 and left ventricle tissues from individuals with and without (n = 718 cells) and MYH7 (n = 582 cells) expression in CMs by analysing scRNA-seq data from e. Whiskers show the minimum to maximum values, and bounds of boxes represent first and third quantiles and the centre line indicates the median. g, Interspecies conservation analysis of LINC00881 sequence. h, A scheme of CM differentiation from hiPSCs by forming EBs. i, Expression dynamics of LIPTER and NKX2.5 during CM differentiation. Dots are presented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison (n = 3 independent experiments). j, Dual gRNAs were designed to completely knock out LIPTER in hiPSCs using CRISPR/ Cas-9. k, LIPTER expression in WT and LIPTER KO hiPSC-derived EBs at day 20 (n = 4 independent experiments). l, Ratios of cTnT + CMs in day 40 hiPSC-EBs (n = 4 independent experiments). In b, c, k and l, bars are represented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison. Source numerical data are available in source data.   Intramyocyte LD accumulation could be due to increased uptake of free FAs via CD36, a membrane receptor to import extracellular FAs 30 . However, both CD36 expression (Extended Data Fig. 4a) and lipid uptake capability were reduced in LIPTER KO hiPSC-CMs compared with WT hiPSC-CMs (Fig. 2d, experiments). f,g, Representative fluorescence images of LD accumulation/ distribution in WT and LIPTER KO hiPSC-CMs treated with 200 μM palmitate for 6 h (f), followed by palmitate depletion for an additional 12 h (g). h, Quantification of relative LD densities in whole CMs, and in 1/2 cytosolic and 1/2 nucleus surrounding areas in CMs (n = 4 independent experiments). i, Measurement of Rhodamine B fluorescence levels in mitochondria isolated from hiPSC-CMs post treatment with Rhodamine B-palmitic acid for 2 h (n = 4 independent experiments). In c, e, h and i, bars are presented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison. Source numerical data are available in source data. Article https://doi.org/10.1038/s41556-023-01162-4 result of enhanced FA uptake. We detected no significant changes in genes for TAG synthesis and lipolysis on LDs, except for ~30% reductions of GPAM and ATGL1 (refs. 1,31) (Extended Data Fig. 4a-c), upon LIPTER KO , suggesting that accumulated LDs in LIPTER KO CMs might not be primarily caused by altered LD formation or lipolysis process. Interestingly, LIPTER KO prominently reduced expression and induced accumulation of PLN5 (Extended Data Fig. 4a-d), which was reported to mediate LD-mitochondria tethering in CMs 15 . These results implied that LIPTER KO could impair LD transport to and/or interaction with mitochondria.
To trace LD transport, hiPSC-CMs were cultured with palmitate (200 μM) for 6 h to induce LD formation (Fig. 2d). In WT hiPSC-CMs, LDs surrounded nucleus and were broadly distributed in the cytoplasm (Fig. 2f, top, arrowheads), while cytosolic transport and distribution of LDs were retarded in LIPTER KO hiPSC-CMs (Fig. 2f,bottom). We then cultured CMs without palmitate for additional 12 h to mobilize the LDs (Fig. 2d). While LDs in WT hiPSC-CMs were fully mobilized, large LDs accumulated in LIPTER KO hiPSC-CMs (Fig. 2g, arrowheads). Statistical results show that, from 6 h to 18 h, LD densities in WT hiPSC-CMs rapidly declined in whole CMs, cytoplasm and the surrounding region of nucleus, whereas high amounts of LDs remained in LIPTER KO hiPSC-CMs (Fig. 2h). Since mobilized LDs could transfer stored lipids to mitochondria for β-oxidation 1,2 , we conducted live cell imaging to trace LD-mitochondria interaction. Rhodamine-B-labelled palmitic acid was added to CM culture for 6 h and enriched in LDs. In WT hiPSC-CMs, approximately 28% of LDs Rh-B fused with mitochondria and then disappeared (Extended Data Fig. 4e, top, yellow arrows; Extended Data Fig. 4f), whereas only about 8% of LDs Rh-B fused with mitochondria in LIPTER KO hiPSC-CMs (Extended Data Fig. 4e, middle, red arrows; Extended Data Fig. 4f). Finally, mitochondria were isolated from hiPSC-CMs after 2 h of Rhodamine-B-labelled palmitic acid treatment. Significantly reduced fluorescence levels were detected in LIPTER KO versus WT mitochondria (Fig. 2i), indicating reduced Rhodamine-B-labelled palmitic acid transfer to mitochondria upon LIPTER ablation. Collectively, these results demonstrate that LIPTER deficiency disrupts LD transport and mobilization in human CMs, leading to extensive LD accumulation.

LIPTER deficiency induces mitochondrial dysfunction and apoptosis
Next, we performed whole messenger RNA sequencing (mRNA-seq) (Supplementary Table 2) and discovered that LIPTER KO altered the transcriptome of hiPSC-CMs (Fig. 3a). The genes with significant expression changes were enriched into Gene Ontology (GO) pathways, including elevated apoptotic signalling, aberrant mitochondrial functions and lipid storage (Fig. 3b), as well as toxicity signalling pathways, such as increased heart failure and cardiac dysfunction (Fig. 3c). Consistently, transmission electron microscopy (TEM) revealed compact and rod-shaped mitochondria in WT hiPSC-CMs (Fig. 3d, yellow arrowheads), while ~60% of mitochondria in LIPTER KO hiPSC-CMs displayed giant and/or swollen morphology (Fig. 3d, red arrows, Fig. 3e), indicating mitochondrial dysfunction. Mass spectrometry (MS) of purified mitochondria revealed that LIPTER KO altered the expressions of mitochondrial respiratory proteins, such as reduced levels of NDUFB2, NDUFB11 and COX7B (Supplementary Table 3). Compared with WT hiPSC-CMs, mitochondrial maximal respiration capacity, spare respiratory capacity (Fig. 3f,g) and long-chain fatty acid oxidation (FAO) capacities of LIPTER KO hiPSC-CMs ( Fig. 3h and Extended Data Fig. 4g) were significantly reduced. Furthermore, increased apoptosis was detected in LIPTER KO versus WT hiPSC-CMs (Fig. 3i,j and Extended Data Fig. 4h,i). To investigate whether these CM abnormalities were LIPTER dependent, LIPTER expression was rescued in LIPTER KO hiPSC-CMs, which restored the swollen and dysfunctional mitochondria and suppressed apoptosis of LIPTER KO CMs (Fig. 3i,j, LIPTER KO/OE versus LIPTER KO ). Consequently, rescued LIPTER enhanced the ratios of cTnT + CMs at day 50 of EB differentiation compared with LIPTER KO (Fig. 3k,l).
Altogether, these results reveal the crucial roles of LIPTER in mitochondrial function and viability of human CMs.

NKX2.5 controls CM-specific LIPTER transcription
Since LIPTER expression is enriched in human CMs and downregulated in T2DM hearts (Figs. 1b,c and 4a) and extensive LDs were observed in both LIPTER KO and T2DM human CMs (Figs. 2b,c and 4b,c), we posited that LIPTER transcription was under the control of a CM-specific regulatory mechanism that could respond to hyperglycaemia. The LIPTER promoter (~2.5 kb upstream of transcriptional start site) was predicted to contain multiple putative binding sites for NKX2.5 RXRA and CEBPB, which are associated with CM-specific gene regulation and lipid metabolism (Fig. 4d). NKX2.5 was the most effective in increasing LIPTER promotor activity, and its co-transfection with RXRA or CEBPB further enhanced the activity (Fig. 4e). Knockdown of NKX2.5 (NKX2.5 KD ) by NKX2.5-shRNAs led to reduced expressions of NKX2. 5 and LIPTER (Fig. 4f), but not CM markers cTnT and MYH10 (Extended Data Fig. 5a). NKX2.5 KD also prominently increased LD accumulation and apoptosis of WT hiPSC-CMs, phenocopying LIPTER KO hiPSC-CMs (Fig. 4g,h and Extended Data Fig. 5b-e, shNKX2.5 versus shControl in WT CMs). All these results indicate that NKX2.5 can control LIPTER transcription in human CMs. Next, we found exposure to high glucose levels (11.0 and 22.0 mM) for 14 days significantly reduced LIPTER expression ( Fig. 4i), and the mRNA and nuclear protein levels of NKX2.5, RXRA and CEBPB in WT hiPSC-CMs (Fig. 4j,k and Extended Data Fig. 6a,b). Similar reductions in these TFs were observed in T2DM hearts compared with non-T2DM human hearts (Fig. 4l,m and Extended Data Fig. 6c,d). These results are consistent with previous reports that high glucose suppressed NKX2. 5 (ref. 32) and RXRA 33 expressions in vivo. Finally, LIPTER overexpression was found to repress NKX2.5 KD -induced LD accumulation and apoptosis (Extended Data Fig. 5b-e, shNKX2.5 versus shControl in LIPTER KO/OE CMs), suggesting a therapeutic potential of LIPTER for treating diabetes-associated dilated cardiomyopathy (DCM) 34 . Altogether, these results demonstrate that NKX2.5 is the primarily regulator of CM-specific LIPTER transcription and that hyperglycaemia can repress the NKX2.5-LIPTER axis expression in human CMs.

LIPTER selectively binds phospholipids and MYH10 protein
We sought to elucidate the molecular mechanisms by which LIPTER regulates LD transport. RT-qPCR detected cytosolic LIPTER expression with GAPDH, but not in the nucleus of hiPSC-CMs (Extended Data Fig. 7a). RNA fluorescence in situ hybridization (RNA FISH) revealed that ~74% of LDs co-localized with LIPTER ( Fig. 5a,b), suggesting potential binding between LIPTER and LDs. We then isolated total lipids from WT hiPSC-CMs, which contain lipid-binding RNAs, followed with RT-qPCR. Compared with the house-keeping gene ACTB, LIPTER was highly enriched in isolated lipids (>150-fold, Fig. 5c), suggesting a possible direct interaction between LIPTER and lipids. Next, using the RNA-lipid overlay assay 35 , we identified that LIPTER selectively bound PA and PI4P, whereas antisense-LIPTER (AS-LIPTER) could not bind any lipid (Fig. 5d). Given that PA and PI4P are found on the LD membranes, we asked whether LIPTER could bind PA and PI4P in a membrane context. Giant unilamellar vesicles (GUVs) generated with TopFluor-labelled PI4P/PA were incubated with Alexa-594-labelled LIPTER/AS-LIPTER, respectively. LIPTER bound both PA-and PI4P-GUVs, whereas AS-LIPTER did not (Fig. 5e). Furthermore, neither LIPTER nor AS-LIPTER bound GUVs generated by PI (Extended Data Fig. 7b). The binding affinities of LIPTER-PA and LIPTER-PI4P were assessed using microscale thermophoresis (MST) assay 36 , with K d values of 706 and 70 in the liquid phase ( Fig. 5f,g), indicating strong and specific LIPTER-PA and LIPTER-PI4P interactions, respectively. Collectively, these data reveal the selective binding of LIPTER with PA and PI4P on LD membranes.
Next, we employed the MS2-BioTRAP system to identify LIPTER-interactive proteins and trace LIPTER in live hiPSC-CMs as previously described 37 (Fig. 5h). As shown in Fig. 5i lower panels, control    independent experiments). f, Mitochondrial OCR measurement in hiPSC-CMs. g, Analyses of maximum respiratory capacity and spare respiratory capacity in OCR, normalized to total protein content of each well (n = 3 independent experiments). h, Quantification of FAO rates. i, Ratios of TUNEL + CMs in day 40 hiPSC-EBs (n = 5 in the first 2 groups and n = 4 in the last group of independent experiments). j, Ratios of cleaved caspase-3 + CMs in day 40 hiPSC-EBs (n = 4 in the first 2 groups and n = 3 in the last group of independent experiments). k, FACS analysis of cTnT + CM ratios in day 50 hiPSC-EBs, with statistical results in l (n = 3 independent experiments). In e, g-j and l, bars are presented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison. Source numerical data are available in source data. hiPSC-CMs expressed MS2-tag and MS2 binding protein MS2 YFP , which was ubiquitously distributed in CMs. In comparison, MS2 YFP bound to MS2-tagged-LIPTER, forming green LIPTER-MS2 YFP particles observed only in the cytoplasm and co-localized with LDs ( Fig. 5i, top right, yellow arrows). Since cytosolic LIPTER-LD co-localization was observed by both MS2-BioTRAP and RNA FISH (Fig. 5a,i), these results indicate that the MS2-tag did not affect cellular location and interaction of LIPTER.
LIPTER-interacting proteins were then pulled down in 293T cells and hiPSC-CMs using antibody against MS2 FLAG or MS2 YFP , respectively, followed by MS analysis and western blotting validation. LIPTER specifically pulled down non-muscle Myosin IIB (NM-IIB, or MYH10) protein, as compared with AS-LIPTER ( Fig. 5j and Extended Data Fig. 7c Fig. 7f), as myosin motors move along Actin filaments 16,17    To identify LIPTER domains that specifically interact with PA, PI4P and MYH10, we generated truncated LIPTER fragments (Fig. 5l) for RNA-lipid overlay and GUVs assays. Exon 3 of LIPTER was found to contain binding domains with PA and PI4P ( Fig. 5m-o). Further analysis revealed the PI4P binding domain on exon 3-1, and PA binding domains on both exon 3-2 and exon 3-3 regions (Fig. 5m). A MYH10-binding domain was identified on exon 3-4 ( Fig. 5p). LD-LIPTER-MYH10 interactions were observed in hiPSC-CM using confocal fluorescence microscopy (Fig. 5q). These results demonstrate that LIPTER selectively binds PA, PI4P and MYH10 via distinct RNA domains. Overall, our data demonstrate that LIPTER functions as a molecular linker, connecting LDs with the cytoskeleton for intramyocyte LD transport (Fig. 5r).

MYH10 is indispensable for LIPTER function in CMs
MYH10 was knocked out (MYH10 KO ) in hiPSCs using CRISPR/Cas-9 ( Fig. 6a,b and Extended Data Fig. 8a). Compared with WT hiPSC-CMs, MYH10 KO hiPSC-CMs displayed increased LD accumulation (Fig. 6c,d), ~50% swollen mitochondria (Fig. 6e,f), reduced mitochondrial spare respiratory and maximal respiration capacities (Fig. 6g), and decreased FAO capacity ( Fed with high-fat diet (HFD, 45 kcal% fat) for 3 months, Myh10 CKO male mice exhibited severe cardiac lipid accumulation, including enhanced LD deposition (Fig. 6q, arrowhead) and elevated concentrations of TAG and FAs, compared with Myh10 f/f littermates (Fig. 6r,s). Notably, Myh10 deficiency significantly reduced the FAO rates of whole Myh10 CKO hearts compared with Myh10 f/f hearts (Fig. 6t). Lastly, left ventricular ejection fraction (EF) and fractional shortening (FS) values significantly declined in Myh10 CKO mice compared with Myh10 f/f mice (Fig. 6u,v). Collectively, all these in vitro and in vivo results demonstrate a conserved and critical role of Myh10 in LD metabolism of mammalian heart muscle cells.

Gain-of-LIPTER mitigates lipotoxicity of hiPSC-CMs
Palmitate overload has been shown to induce cardiac lipotoxicity in vivo, which compromised mouse heart function and increased CM death 39 . In vitro, palmitate overload can incite oxidative and ER stress of CMs 2 . We found that palmitic acid (400 μM) treatment for 4 days significantly induced apoptosis in hiPSC-CMs (Fig. 7a,b and Extended Data Fig. 9a, Vector-Ctrl versus Vector-PA), which was prominently reduced by LIPTER OE (Fig. 7b,c, LIPTER OE -PA versus Vector-PA; Extended  In b, f, d-i, l-n and r-v, bars are presented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison. Source numerical data are available in source data. Article https://doi.org/10.1038/s41556-023-01162-4 Data Fig. 9a,b). These data reveal an in vitro protective role of LIPTER against lipotoxicity.

In vivo LIPTER transgene displays a cardiac protective role
Intramyocyte LD accumulation occurs in human obese and diabetic hearts when cardiac function is still normal 5,39 , suggesting that LD metabolic disturbance precedes the onset of cardiac dysfunction.
We hypothesized that gain-of-LIPTER function could alleviate obesityand diabetes-associated cardiomyopathies and cardiac dysfunctions. To test this, we generated a LIPTER transgenic (LIPTER(Tg)) mouse line by knocking in LIPTER into the Rosa26 locus, resulting in high LIPTER expression in the heart (Extended Data Fig. 9c,d). Six-week-old male LIPTER(Tg) and WT mice were fed a HFD (45 kcal% fat) to induce obesity, insulin resistance and cardiac abnormalities as previously reported    Fig. 7d) 40,41 . After 7 months, HFD-fed LIPTER(Tg) mice showed no significant change in heart weight/tibia length ratios compared with WT mice (Extended Data Fig. 9e). However, LIPTER(Tg) significantly reduced LD levels, and FA and TAG concentrations compared with WT hearts (Fig. 7e-g). Notably, LIPTER(Tg) mice exhibited significantly enhanced FAO rates in whole hearts (Fig. 7h) and globally altered transcriptome (Supplementary Table 4). Metabolic process, energy homeostasis, cardiac muscle development and growth, and sarcoplasmic reticulum Ca 2+ transport biological processes were significantly over-represented in the LIPTER(Tg) upregulated genes (Fig. 7i). We also observed significantly reduced fibrotic areas (Fig. 7j,k) and apoptosis (Extended Data Fig. 9f,g) in HFD-fed LIPTER(Tg) hearts compared with similarly treated WT hearts. Importantly, LIPTER(Tg) preserved cardiac function in HFD-fed mice. Significantly reduced EF and FS values were observed in WT mice fed with HFD for 10 months compared with WT mice fed with normal chew (NC) (Fig. 7l,m, WT-HDF versus WT-NC). However, EF/FS values of HFD-fed LIPTER(Tg) mice were significantly higher than those of HFD-fed WT mice and were preserved at levels similar to NC-fed WT mice (Fig. 7l,m, LIPTER(Tg)-HFD versus WT-HFD). These results demonstrate that gain-of-LIPTER can mitigate HFD-induced cardiomyopathy and preserve cardiac function. Furthermore, to test whether the transgenic LIPTER could interact with mouse Myh10 protein, we performed RNA immunoprecipitation sequencing (RIP-seq) to pull down and sequence all mouse Myh10-binding RNAs from LIPTER(Tg) and WT mouse hearts. Although no Myh10 binding signals were detected on control Actb and Gapdh mRNAs, mouse Myh10 protein bound LIPTER in LIPTER(Tg) heart (Extended Data Fig. 9h). Interestingly, mouse Myh10 binding signals were enriched on exon 3 of transgenic LIPTER (Fig. 7n, red arrow), which was found to interact with human MYH10 (Fig. 5p). These results suggest that LIPTER-MYH10 interaction might play a conserved role in LD transport of mammalian CMs, although the mouse homologue of LIPTER remains to be identified.

Myh10 expression level in hearts
Finally, we examined the impact of CM-specific LIPTER transgene on cardiomyopathy of leptin-receptor-deficient Lepr db/db mouse (Fig. 7o), a model of T2DM and obesity that develops cardiac hypertrophy and dysfunction from 10 weeks of age 42,43 . We employed an AAV9 virus carrying a chicken cTnT promoter to selectively deliver LIPTER into CMs 44 . Retro-orbital injection of AAV9-cTnT-GFP virus (2 × 10 10 vg/g) into Lepr db/db mice led to robust GFP expression in mouse CMs (Extended Data Fig. 10a,b). Subsequently, AAV9-LIPTER, control AAV9-GFP and AAV9-AS-LIPTER viruses were administered to 6-week-old male B6.BKS(D)-Lepr db/db mice using the same approach. Age-matched WT C57BL/6J male mice without AAV-9 injection also served as controls. Six weeks post AAV-9 injection, transgenic LIPTER, AS-LIPTER and GFP expressions were detected in the mouse hearts ( Fig. 7p and Extended Data Fig. 10c). AAV9-LIPTER administration resulted in reduced intramyocyte LD accumulation (Fig. 7q), increased whole heart FAO rates (Fig. 7r), reduced CM size ( Fig. 7s and Extended Data Fig. 10d), and preserved EF and FS values in Lepr db/db mice ( Fig. 7t-u) compared with control groups. Since AAV9-LIPTER transgene in CMs did not impact blood glucose levels of Lepr db/db mice (Extended Data Fig. 10e), these findings indicate that CM-restricted LIPTER transgene can mitigate cardiomyopathy and preserve cardiac function in Lepr db/db mice.

Discussion
Our data demonstrate that LIPTER functions as an RNA linker, facilitating LD transport in human CMs by connecting LDs via binding PA/PI4P and the cytoskeleton via binding MYH10. Despite the importance of lncRNAs in regulating cellular processes, the roles of lncRNA-lipid interactions remain largely unexplored. Lin et al. reported that LINK-A bound PI(3,4,5)P3 to hyperactivate AKT in cancer cells 35 , and SNHG9 was found to bind PA to facilitate LATS1 liquid-liquid phase separation, promoting oncogenic YAP signalling 45 . Our study reveals that LIPTER binds both PA and PI4P (Fig. 5d,e), which play crucial roles in membrane-membrane interactions that facilitate LD transfer. Specifically, PA on the ER membrane contributes to the formation of newly formed LDs 46 , while PI4P on the Golgi and plasma membrane can recruit and bind lipid transport carrier proteins on LDs 47 . Our MS analysis of proteins pulled down by LIPTER (Supplementary Table 5) did not detect typical LD-associated proteins, except TFG (Trafficking From ER To Golgi Regulator). However, we did not detect any interaction between TFG and LIPTER (Extended Data Fig. 10f). These findings, in conjunction with the GUV results (Fig. 5e), indicate that LIPTER can directly bind PA and PI4P on LDs without additional protein co-factors. Currently, factors determining the selective interactions of RNA with different lipids are not well understood. A few studies have suggested that RNA-lipid interaction may depend on RNA length, base pairing and nucleotide content 48,49 . Tomasz et al. also reported that guanine and G-quadruplex formation are critical for RNA-lipid interactions 49 .
Our study demonstrates that MYH10 depletions disrupted LD metabolism and transport in hiPSC-CMs, and compromised cardiac function in Myh10 CKO mice, highlighting the crucial and conserved role of cytoskeleton in LD trafficking within mammalian CMs. Additionally, we observed enlarged LDs in LIPTER KO hiPSC-CMs compared with WT hiPSC-CMs (Fig. 2g). Myh9 depletion in U2OS cells was found to reduce LD dissociation from the ER and increase LD size 18 . Since MYH9 is not expressed in human CMs, our data suggest that disconnection of MYH10 from LDs due to LIPTER deficiency may impair the dissociation of newly formed LDs from ER, leading to increased LD size.
Diabetic cardiomyopathy is recognized as a non-ischaemic form of DCM, characterized by progressive muscle loss, global systolic dysfunction and heart failure. It has been reported that approximately 75% of patients with idiopathic DCM are diabetic 34 . An 85-fold increase in apoptosis and 4-fold increase in necrosis of myocytes have been observed in diabetic hearts compared with non-diseased hearts 50 . Our data reveal that hyperglycaemia suppresses the NKX2.5-LIPTER axis in hiPSC-CMs, and NKX2.5/LIPTER deficiencies induce LD deposition and CM death Red O and haematoxylin staining of mouse hearts after 7 month HFD feeding. f,g, FA (f, n = 5 mice each group) and TAG (g, n = 7 mice each group) concentrations were next measured. h, FAO rates of whole WT and LIPTER(Tg) mouse hearts at 3 months of age (n = 5 mice each group). i, GO biological processes significantly enriched in the upregulated genes in LIPTER (Tg) versus WT mouse hearts. j, Picrosirius Red and Fast Green co-staining of mouse hearts after 10 month HFD feeding. k, Quantification of relative red fibrotic areas in whole hearts (n = 3 mice each group). l,m, EF (l) and FS (m) measurements of WT and LIPTER(Tg) mice fed on HFD or NC for 10 months. n = 5 mice (WT + NC), n = 6 mice (WT + HFD), n = 7 mice (Tg + HFD). n, RIP-seq reads plot showing LIPTER pulldown by anti-Myh10 antibody in LIPTER (Tg) mouse heart. Red arrow indicates the reads enrichment on LIPTER exon 3. The experiment was carried out in two mice, each genotype with similar outcomes. o, A scheme for delivering LIPTER and controls (GFP, AS-LIPTER) into Lepr db/db mouse CMs using AAV9 virus. p, RT-qPCR detecting LIPTER expressions in Lepr db/db mouse hearts 6 weeks post AAV9 injection (n = 5 mice per group, except n = 4 mice in the GFP group). q, Oil Red O staining of Lepr db/db mouse hearts 6 weeks post AAV9 injection. r, FAO rates of Lepr db/db mouse hearts 6 weeks post AAV9 injection (n = 5 mice per group, except n = 4 mice in the No AAV group). s, CM size analysis using WGA staining 6 weeks post AAV9 injection (n = 5 mice per group, except n = 4 mice in the GFP group). t,u, EF (t) and FS (u) measurements in WT and Lepr db/db mice 6 weeks post AAV9 injection (n = 5 mice per group, except n = 4 mice in the GFP group). In b, c, f-h, k-m, p and r-u, bars are presented as mean ± s.e.m. Unpaired two-tailed t-test is used for comparison. Source numerical data are available in source data. Article https://doi.org/10.1038/s41556-023-01162-4 (Extended Data Fig. 5). Although the mechanism of NKX2.5 downregulation by hyperglycaemia remains to be investigated, our data suggest that a declined NKX2.5-LIPTER axis could possibility contribute to CM loss in diabetic hearts. We hypothesize that hyperglycaemia could differentially repress NKX2.5-LIPTER expression levels in individual CMs, and CMs with critically low LIPTER expression levels could gradually die, with this slow progress of muscle loss eventually leading to DCM of diabetic hearts. Since intramyocyte LD accumulation precedes  onset of cardiac dysfunctions in individuals with obesity or diabetes 5,39 , our in vivo LIPTER transgene data strongly suggest that targeted LIPTER delivery into heart muscle might be an effective strategy to prevent cardiac dysfunction and heart failure in metabolic syndromes. Compared with DNA, RNA exhibits a broader range of interacts with various molecules, suggesting more diverse and complex functional dimensions. LDs are ancient organelles that exist in prokaryotes and eukaryotic cells. Our findings reveal a molecular linker role of RNA in mediating the transport of lipid-covered vesicle through RNA-lipid/ protein interactions. This may reflect ancient RNA functions in the primordial 'RNA world', where RNA-lipid/protein interactions could have play critical roles in forming the first living system, which encompassed lipid membranes and a mixture of RNA, protein and other molecules.
In summary, this study uncovers an lncRNA-mediated LD transport system in human CMs and reveals the crucial role of LIPTER and MYH10 in lipid metabolism of mammalian hearts. The mouse homologue of LIPTER has not been identified, limiting the in vivo loss-of-function analysis of LIPTER in rodents. Moreover, considering that a single lncRNA can interact with multiple proteins 28,51,52 , it is not yet clear whether LIPTER could regulate LD metabolism through other mechanisms, such as epigenetic regulation of genes in TAG synthesis and lipolysis on LDs (Extended Data Fig. 4a). Lastly, FAO rate measurements were conducted at a 6:1 palmitate-to-albumin ratio, much higher than the physiological ratio (1:1-3:1) (ref. 53), raising a concern of lipid overload during measurement. Additionally, whole heart FAO rate measurement using heart tissue homogenate were not performed under the same oxygen concentration as in vivo conditions, resulting in the majority of radiolabelled palmitate being converted into ASMs rather than CO 2 . To address these technical challenges, FAO rates will be further evaluated in live animals as previously described 54 .

Online content
Any methods, additional references, Nature Portfolio reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/s41556-023-01162-4.

Animals and diets.
All animal experiments were performed following the approval and guidelines of the Institutional Animal Care and Use Committee of Indiana University. Animals are maintained in accordance with the applicable portions of the Animal Welfare act and the DHHS Guide for the Care and Use of Laboratory Animals. LIPTER(Tg) C57BL/6J mice were generated using CRISPR/Cas9 at the Genome Editing, Transgenic, and Virus Core at UPMC Magee-Women's Research Institute. Six-week-old male homozygous Lepr db/db mice were obtained from the Jackson Laboratory (strain number 000697). Myh10 floxed mice with C57BL/6J background were obtained from Dr Robert S. Adelstein lab at Laboratory of Molecular Cardiology, National Heart, Lung and Blood Institute 57 . C57BL/6J Tnnt2 Cre mice were obtained from Dr Chenleng Cai lab at Indiana University. Conditional Myh10 cKO in mouse CMs was generated by crossing Myh10 f/f mice with Tnnt2 MerCreMer mice. Six-week-old male Myh10 f/f and Myh10 f/f /Tnnt2 MerCreMer mice were injected with tamoxifen (0.1 mg g −1 body weight) for days 1, 3 and 5 by intraperitoneal injection and then fed with a HFD (45 kcal% fat, Research Diets Inc., D12451i) for 3 months. Six-week-old male LIPTER(Tg) mice and age-matched WT C57BL/6J mice were fed with a HFD (45 kcal% fat, Research Diets Inc., D12451i) or NC (18 protein, Inotiv, 2018sx) for 10 months.

Method details
Human heart tissue acquisition. Human left ventricle tissues were obtained from the Duke Human Heart Repository, which is a Duke University Health System institutional review board-approved tissue repository. Samples were procured by the Duke Human Heart Repository in accordance with an approved Duke University Health System institutional review board protocol with written informed consent. All patient information was de-identified. Left ventricle tissues were immediately flash frozen in liquid nitrogen and stored in a −80 °C freezer.

LIPTER coding potential prediction and verification
Protein coding potential was predicted using Open Reading Frame Finder (https://www.ncbi.nlm.nih.gov/orffinder/). FLAG sequences (GAC TAC AAA GAC GAT GAC GAC AAG) were inserted after each predicted ORF before the stop codon and cloned into a pQCXIP vector. Three LIPTER ORFs with FLAG plasmids were transfected into HEK293T cells (CRL-3216, ATCC) respectively using the X-tremeGENE 9 DNA Transfection Reagent (Roche, 6365787001). A non-relevant protein-coding gene with a FLAG-tag was transfected as a positive control. RNA expression levels were detected using RT-qPCR. FLAG was detected using anti-FLAG antibody (1:1,000, Cell Signaling, 14793S) by immune fluorescent staining and western blotting.

Luciferase assay
Luciferase assay was performed using the Dual-Glo Luciferase Assay System (Promega, E2920) following the manufacturer's protocol. Briefly, HEK293T cells were co-transfected with the pGL3-LIPTER-promoter and NKX2.5, CEBPB or RXRa plasmids, or a vector control using the X-tremeGENE 9 transfection reagent (Roche, 6365787001). Cells were collected 48 h post-transfection. Firefly and Renilla luciferase activities in cell lysates were measured using the GLOMAX Explorer microplate reader system (Promega).

Lipid uptake assay
Rhodamine-B-labelled palmitic acid (Avanti, 810104) was dissolved in ethanol and then in 10% bovine serum albumin (BSA). The final concentration of PA was 5 mM. HiPSC-CMs were dissociated into single cells and seeded at a density of 1 × 10 6 cells per well in a 12-well plate. Cells were treated with 200 μM Rhodamine-B-labelled palmitic acid for 1 h and lysed with 100 μl RIPA lysis buffer. For the mitochondrial lipid uptake assay, hiPSC-CMs were treated with 200 μM Rhodamine-labelled palmitic acid for 2 h. Subsequently, mitochondria were isolated using the Mitochondria Isolation Kit for Cultured Cells (ThermoFisher, 89874) and lysed with 100 μl RIPA lysis buffer. Fluorescence values were detected at an excitation of 520 nm with a 580-640 nm filter on a GLOMAX Explorer microplate reader system (Promega).

Mitochondrial proteomics
WT and LIPTER KO hiPSC-CMs were selected and dissociated into single cells. Mitochondria were isolated from 5 × 10 6 CMs using Mitochondria Isolation Kit for Cultured Cells (ThermoFisher, 89874). MS and data analyses were performed at the Proteomics Core Facility at the Indiana University School of Medicine.

Flow cytometry
Flow cytometry was performed to detect cTnT positive cell ratios during EB differentiation, as previously described 28 . Briefly, day 20 or Nature Cell Biology Article https://doi.org/10.1038/s41556-023-01162-4 40 EBs were collected and dissociated with 1 mg ml −1 Collagenase B for 30 min, followed by 0.25% trypsin-EDTA for 5 min at 37 °C. Dissociated single cells were fixed in 4% paraformaldehyde (PFA) for 15 minutes at room temperature and washed three times with phosphate-buffered saline (PBS). Cells were incubated in a blocking buffer containing 1× PBS, 10% goat serum and 0.1% saponin. Cells were then incubated with primary antibody diluted in 1× PBS with 2% BSA and 0.1% saponin for 1 h at 37 °C, followed by incubation with Alexa Fluor 488-labelled goat anti-mouse secondary antibody (1:200, Thermo Fisher Scientific, A-11001) for 1 h at 37 °C. Flow cytometry analysis was conducted using an Attune NxT flow cytometer (Invitrogen). Data were analysed using FlowJo (Treestar). Supplementary Fig. 1 shows the gate setting.

Nuclear and cytoplasmic RNA detection
A total of 2 × 10 6 hiPSC-CMs were separated into nuclear and cytoplasmic fractions using NE-PER Nuclear and Cytoplasmic Extraction Reagents (ThermoFisher, 78833). RNAs from both nuclear and cytoplasmic fractions were extracted using the RNeasy Mini Kit (Qiagen, 74106). LIPTER, GAPDH (cytosolic marker) and U6 (nuclear marker) expression levels were detected using RT-qPCR on a QuantStudio 6 Flex system (Applied Biosystems).

Extraction of lipid-associating RNAs
A total of 1 × 10 6 hiPSC-CMs were treated with 100 μM palmitate acid for 48 h, followed by lipid extraction using the Lipid Extraction Kit (Cell Biolabs, STA-612) according to the manufacturer's protocol. The isolated total lipids were dried in a vacuum concentrator at 4 °C and resuspended in 700 μl of TRIzol. RNA was then extracted by using the RNeasy mini kit (Qiagen, 74106).

TEM
hiPSC-derived EBs at day 40 of differentiation were fixed in 2.5% glutaraldehyde in 1× PBS for 1 h. Then EBs were washed with PBS and post-fixed with 1% osmium tetroxide and 1% potassium ferricyanide for 1 h at 4 °C, dehydrated with a graded series of EtOH (30%, 50%, 70%, 90% and 100%). Subsequently, EBs were embedded and sectioned. TEM images were taken using the JEM-1400 Flash Electron Microscope at the Indiana Center for Biological Microscopy.

Echocardiography
Mice were anaesthetized with 1-2% isoflurane, and chest hairs were shaved. The anaesthetized mice were placed on a heated platform in the prone position. Pre-heated ultrasound coupling gel was applied to the chest area, and a linear array transducer (18)(19)(20)(21)(22)(23) was positioned to obtain one-dimensional M-mode images or two-dimensional B-mode parasternal long-and short-axis views on a VisualSonics Vevo 2100 Imaging System. Left ventricle FS percentages and EF percentages were calculated from the M-mode measurements using Vevo LAB ultrasound analysis software at the IU Small Animal Ultrasound Core.

NKX2.5 knockdown
AAV9-shNKX2.5 were utilized to knock down NKX2.5 in hiPSC-derived CMs. Briefly, AAV9-shRNA-ctrl (Addgene, 85741) was used as a control scramble shRNA. Two shRNAs targeting human NKX2.5 were designed and inserted into the AAV9 plasmid to replace the shRNA-ctrl sequence. The pAdDeltaF6 and AAV2/9 plasmids were used for packaging viruses. Viruses with a concentration of 1 × 10 7 GC ml −1 were used to infect hiPSC-derived EBs. Media containing AAV9 virus were changed every three days and maintained for 9 days.

Cryosection and CM area measurement
HiPSC-derived EBs were fixed with 4% PFA for 30 min, washed with PBS and dehydrated in sucrose solutions of increasing concentration (10%, 20% and 30%) for 1 h each. Dehydrated EBs were embedded in OCT (Fisher Scientific, 23730571) and stored at −80 °C. Frozen EBs were sectioned at 7 μm on a Leica CM3050 cryostat and stored in a −80 °C freezer. For human heart LV tissues and mouse heart sectioning, the tissues were fixed with 4% PFA overnight, and dehydrated in sucrose solutions of increasing concentration (10%, 20% and 30%) for 3 h each. Fixed tissues were embedded in OCT and stored at −80 °C freezer. To measure myocyte cross-sectional area, the sections were stained with Alexa Fluor 488-conjugated wheat germ agglutinin (Invitrogen, W11261) for membrane and with DAPI for nuclei. A minimum of 50 CMs per section were manually outlined and single-cell cross-sectional area was measured using ImageJ software (National Institutes of Health (NIH)).

Fast Green and Picoro-Sirius Red staining
Mouse heart sections were fixed with Bouins' solution at 55 °C for 1 h and washed with tap water for 45 min. Until yellow colour disappeared, sections were stained with 0.1% Fast Green for 30 min, followed by 1% acetic acid for 15 s and tap water for 2 min. Sections were then stained with 0.1% Sirius Red for 30 min, dehydrated with EtOH (70-100%), cleared with xylene and mounted with Cytoseal 60 (Fisher Scientific, 23-244257).

RNA FISH
Cryosections of hiPSC-EBs and human heart tissues were prepared by following the sample preparation protocol for fixed frozen tissues from ACD Inc. Briefly, the tissue sections were washed with PBS, baked at 60 °C for 30 min, post-fixed with 4% PFA for 15 min at 4 °C, and then dehydrated with 50%, 70% and 100% EtOH for 1 × 5 min at room temperature. Tissues were baked in target retrieval solution at 98-102 °C for target retrieval, and then treated with Protease IV for 30 min at 40 °C. The LIPTER specific probes labelled with C2 and negative control probes were synthesized at ACD Inc. The prepared tissue and EB sections were hybridized with LIPTER or control probes following the RNAscope Fluorescent Multiplex Assay protocol. Briefly, the sections were incubated with Amp 1-FL for 30  with Wash Buffer twice at room temperature between each inculcation. For FISH and lipid double stainings, after FISH, the sections were stained with 1:1,000 LipidTOX Deep Red (ThermoFisher, H34477) in PBS for 30 min at room temperature. After three washes with PBS, samples were mounted with DAPI Fluoromount-G (SouthernBiotech, 0100-20) for imaging.

Immunofluorescent microscopy
EBs or tissue sections were washed with PBS to remove OCT and blocked with 5% goat serum in PBST for 1 h at room temperature. CMs were fixed with cold 4% PFA for 15 min at room temperature. The blocked sample sections and fixed CMs were incubated with anti-cTnT antibody (

TUNEL staining
TUNEL staining was performed following the manufacturer's protocol of In Situ Cell Death Detection Kit (TMR red) (Roche, 12156792910). Briefly, after immunofluorescent staining with anti-cTnT antibody (1:200, Thermo Fisher Scientific, MS-295-P), heart tissue sections or cell samples were permeabilized with permeabilization solution (0.1% Triton X-100 in 0.1% sodium citrate) for 2 min on ice and washed with PBS twice. Samples were then stained with TUNEL reaction mixture for 1 h at 37 °C. After washing with PBS three times for 5 min each, the samples were mounted with DAPI Fluoromount-G medium (Southern-Biotech, 0100-20) for imaging.

RNA isolation, cDNA synthesis and RT-qPCR
Total RNAs were extracted from cells or tissues using TRIzol Reagent (Invitrogen, 15596018) and purified with the RNeasy mini kit (Qiagen, 74106). DNase treatment was performed using RNase-free DNase (Qiagen) to remove any contaminating genomic DNA. Total RNAs were reverse transcribed into complementary DNA using the High-Capacity cDNA Reverse Transcription Kit (Applied Biosystems, 4374966). RT-qPCR was performed on a QuantStudio 6 Flex system (Applied Biosystems) using SYBR Green Master Mix (Applied Biosystems, 4385614). Results were analysed using the 2 −ΔΔCt method and normalized to the expression of ACTB gene. RT-PCR was performed with Thermo Fisher Scientific DreamTaq Green PCR Master Mix. Primer sequences are described in Supplementary Table 6.

RNA-lipid overlay assay
The RNA-lipid overlay assay was performed according to the previously published method 58 . Briefly, Biotinylated LIPTER RNA, truncated LIPTER RNA fragments and control AS-LIPTER were transcribed using the TranscriptAid T7 High Yield Transcription Kit (ThermoFisher, K0441), and labelled with RNA 3′ End Biotinylation Kit (ThermoFisher, 20160) following the manufacturer's protocols. Lipid strips membrane (Echelon Biosciences, P-6002) was blocked with RNA-lipid binding buffer and then incubated with biotinylated RNA (1 μg ml −1 ) in RNA-lipid binding buffer supplemented with 50 U ml −1 RNase inhibitor. The membrane was then washed three times for 10 min with RNA-lipid binding buffer supplemented with 0.05% IGEPAL CA-630 (Sigma, I8896), and signal was detected using the Chemiluminescent Nucleic Acid Detection Module (ThermoFisher, 89880) of a ChemiDoc Imaging System (Bio-Rad).

Giant lipid vesicles assay
To generate giant lipid vesicles 59 , fluorescent-labelled phospholipids including TopFluor TMR PA (Avanti Polar Lipids, 810240), Top-Fluor PI4P (Avanti Polar Lipids, 810185) and TopFluor PI (Avanti Polar Lipids, 810187) were dissolved in CHCl 3 . A total of 250 nmol of each fluorescent-labelled PA, PI4P or PI in 500 μl CHCl 3 was transferred to a small glass tube (ThermoFisher, 14-958) and dried using nitrogen air in a chamber. A thin lipid layer was formed on the bottom of the glass bottle. RNA-lipid binding buffer was slowly added to the bottle, and after 48 h of incubation at room temperature in the dark, giant lipid vesicles were collected from the upper layer of the solution. LIPTER, LIPTER-antisense and LIPTER exon 1-2 and exon 3 fragments were labelled with the Alexa Fluor 594 Nucleic Acid Labeling Kit (ThermoFisher, U21654) according to the manufacturer's protocol. The lipid vesicles and the Alexa594-labelled RNAs were gently mixed at a volume ratio of 5:1, incubated at room temperature for 15 min in the dark, and dropped on glass coverslips. Images were captured using a DMi8 inverted fluorescent microscope (Leica).

MST assay
MST assays were performed on a NanoTemper Monolith NT.115 with blue/red filters (NanoTemper Technologies) at the Physical Biochemistry Instrumentation Facility at Indiana University, Bloomington. TopFluor TMR PA (Avanti Polar Lipids, 810240) and TopFluor PI4P (Avanti Polar Lipids, 810185) were dissolved in RNA-lipid binding buffer (100 nM), and a final concentration of 25 nM of each was used. LIPTER and LIPTER-antisense RNA were synthesized in vitro and serially diluted (1-9,600 nM) in the RNA-lipid binding buffer into 16 vials with 10 μl per vial. Fluorescent-labelled PA or PI4P (10 μl) was added and mixed well in each vial. Next, 10 μl of the mixed solution from each vial was removed and loaded into a capillary. Fluorescence intensities of the 16 capillaries were measured at 40% MST power and 90% LED power. The K d value was calculated using MO Affinity Analysis software.

Live cell imaging
To trace LIPTER RNA in live CMs, LIPTER-24xMS2SL and LIPTER-antisense-24xMS2SL were separately inserted into a pHAGE lentiviral vector with puromycin resistance. The pHR-tdMCP-YFP (MS2-coat binding protein tagged with YFP) lentiviral plasmid was obtained d from Addgene (plasmid number 99151). A total of 1 × 10 6 hiPSCs were infected with packaged pHAGE-LIPTER-24xMS2SL or pHAGE-LIPTER-antisense-24xMS2SL lentivirus for 72 h. After infection, cells were selected using 1 μg ml −1 puromycin for 4 days. The remaining cells were then infected with pHR-tdMCP-YFP lentivirus. YFP + cells were sorted by a BD FACSAria II Cell Sorter, expanded and differentiated into CMs. The differentiated CMs were seeded into poly-d-lysine-coated 35 mm glass-bottom dishes (MatTek, P35GC-0-10-C). Live cell video was taken on a Nikon live cell imaging system with an Apo 60× Oil lens at Indiana Center for Biological Microscopy. Consecutive images were taken during 4 h with 3 min intervals, and a total of 81 images were taken per field. The video was exported with 16 frames per second using the NIS Elements software.

Quantifications of FA and TAG in mouse hearts
The concentration of free FA was quantified by using the Free Fatty Acid Assay kit (Abcam, ab65341), and TAG concentrations were quantified using a Triglyceride Quantification Assay Kit (Abcam, ab65336). Approximately 25 mg of mouse heart tissue per heart was collected and homogenized in 500 μl lipid extraction buffer (Abcam, ab211044) using a Beadbug Homogenizer. After centrifugation, the supernatant was removed to a new tube. FA or TAG concentration was measured according to the manufacturer's protocols. The FA and TAG concentrations in WT and genetically modified mouse hearts were normalized to the collected heart tissue weights.

Whole mouse heart FAO rate assay
Freshly collected mouse heart was minced into small pieces and thoroughly homogenized in ice-cold STE buffer (0.25 M sucrose, 10 mM Tris-HCl and 1 mM EDTA, at pH of 7.4) using a Dounce homogenizer. After homogenization, the specimen was centrifuged at 420g at 4 °C for 10 min. Supernatant containing crude mitochondria was transferred into chilled Eppendorf tubes for protein quantification and FAO assay as previously described [61][62][63] . In brief, 30 μl of tissue homogenate was added into Eppendorf tubes in triplicate (including STE blank control for background reading), followed by the immediate addition of 370 μl oxidation reaction mixture (100 mM sucrose, 10 mM Tris-HCl, 5 mM KH 2 PO 4 , 0.2 mM EDTA, 80 mM KCl, 1 mM MgCl 2 , 2 mM l-carnitine, 0.1 mM malate, 0.05 mM Coenzyme A, 2 mM ATP, 1 mM dithiothreitol, 0.5 mM palmitate, 0.7% BSA and 0.4μCi 14 C-palmitate), and incubated at 37 °C for 1 h. After incubation, the entire reaction mixture was transferred into tubes containing 200 μl of 1 M perchloric acid and the CO 2 trap disc, and incubated for 1 h at room temperature. Then the trap disc was transferred to scintillation vials for complete oxidation measurement, whereas the supernatant obtained by centrifugation of the remaining acid solution was used to measure acid-soluble metabolites (ASMs) produced by incomplete oxidation. In addition, the nonincubated reaction mixture was counted in triplicate to get the measurement of the amount of radioactivity input into each reaction.
The specific activity of the reaction mixture was determined by dividing the disintegrations per minute (DPM) of the input by the number of nanomoles of palmitate (cold + hot) per reaction. The specific activity can be used to convert DPM to nanomoles for the other samples.
Specific activity = DPM of input Total nmol of palmitate To determine the rate of conversion of 14 C-palmitate to 14 CO 2 , we divide the nanomoles of 14 CO 2 by the amount of time in the reaction and the amount of protein per action. A similar calculation can be done to determine the rate of conversion of 14 C-palmitate to ASMs.

Untargeted metabolomics
Untargeted metabolomics was performed, and data were analysed by Creative Proteomics. Briefly, 10 million enriched hiPSC-CMs were collected at day 40 and washed with PBS. Metabolites were extracted with 800 μl of methanol, 10 μl of DL-o-chlorophenylalanine (2.8 mg ml −1 ) and 10 μl of LPC (12:0), followed by ultrasonication for 30 min. The samples were analysed using the Ultimate 3000 HPLC and UHPLC combined with Q Exactive MS systems (ThermoFisher) and screened with electrospray ionization (ESI)-MS. Raw data were acquired and aligned using the Compound Discover (3.0, ThermoFisher) based on the m/z value and the retention time of the ion signals. Ions from both ESI − and ESI + were merged and imported into the SIMCA-P program (version 14.1) for multivariate analysis.

Seahorse mitochondrial function assays
Seahorse XFp Extracellular Flux Analyzer was used to measure oxygen consumption rate (OCR) and extracellular acidification rate of hiPSC-CMs according to the manufacturer's protocol. Briefly, 5 × 10 4 CMs (based on FCCP titration assay) were seeded into Matrigel-coated eight-strip utility plate for 48 h before Mito Stress Test assay. We prepared the pH 7.4 assay medium (with 4 mM l-glutamine and 25 mM glucose),and hydrated the cartridge in calibration medium at 37 °C without CO 2 before running the assay. Then 1.5 μM oligomycin was loaded into port A, 0.25 μM FCCP into port B and 0.5 μM Rot/AA into port C on the cartridge. The Mito Stress Test was run according to the standard process of analyzer. After the assay, the protein concentration of each well was measured using BCA protein assay kit (ThermoFisher, 23225). Raw data were processed using Seahorse Wave software.
Long-chain FAO was measured according to the protocol of Palmitate Oxidation Stress Test Kit on a Seahorse XF96 Analyzer. Briefly, 5 × 10 4 CMs were seeded into one well of Matrigel-coated 96-well utility microplate, cultured in normal culture medium for 48 h and then switched to the substrate-limited growth medium (DMEM plus 0.5 mM glucose, 1.0 mM GlutaMAX, 1% FBS and 0.5 mM l-carnitine) overnight. The FAO assay buffer was prepared according to the protocol and replaced the substrate-limited growth medium 45 min before measurement. FAO buffer or 40 μM etomoxir was added into each well 15 min before measurement. BSA control or palmitate:BSA was added into well before the Mito Stress test was started. FAO level was quantified on the basis of the differences in the values of the maximal oxygen consumption in palmitate:BSA without or with presence of etomoxir.

Illumina RNA sequencing and data analysis
Transcriptional profiles of human ESC-derived CMs, SMCs and ECs were obtained by Illumina mRNA deep sequencing using a service from LC Sciences as we previously reported 24 . Three non-failure hearts, three non-failure hearts with T2DM, three hearts with DCM, and three hearts with DCM and T2DM were subject to mRNA-seq. Briefly, total RNAs were extracted from human left ventricle tissues by using the RNeasy kit (Qiagen, 74106), and sequenced by using paired-end deep-transcriptome sequencing with the Illumina platform at IU Genomic Core. All lncR-NAs were profiled according to their fragments per kilobase of transcript per million mapped reads values. Additionally, mRNAs were extracted from enriched day 40 WT and LIPTER KO hiPSC-CMs, as well as hearts of WT and LIPTER(Tg) mice fed with HFD for 7 months and sequenced by using paired-end deep-transcriptome sequencing with the Illumina platform. The RNA-seq data were collected and analysed Article https://doi.org/10.1038/s41556-023-01162-4 as we previously described 24,64 . In general, the sequencing reads were mapped to the reference genome (either hg38 or mm10) by STAR (version 2.7.2a). Gene expression levels were evaluated by the fea-tureCounts on uniquely mapped reads. Following gene expression normalization based on trimmed mean of M (TMM) values, edgeR (refs. 65,66) was employed to perform differential analysis given the comparison between LIPTER KO and WT for human samples or LIPTER(Tg) and WT for mouse heart samples. The genes with false discovery rate (FDR)-adjusted P values <0.05 after multiple test correction were determined as differentially expressed genes.

Functional enrichment analysis
GO analysis was conducted on differentially expressed genes using DAVID [67][68][69] . The gene set enrichment analysis (GSEA) 70 was performed (http://www.gsea-msigdb.org/gsea/index.jsp, V4.1.0) based on the fold change (log 2 ) of gene expression between either KO versus WT CMs or Tg versus WT hearts by focusing on C5 biological process (bp) from GO database or specific toxicity gene sets retrieved from Ingenuity Pathway Analysis 71 . The significantly enriched GO terms or toxicity-related gene sets were selected for presentation in figures.

RIP-seq for mouse hearts
RNA immunoprecipitation was performed with mouse hearts as described above. Hearts were collected from WT and LIPTER (Tg) mice. One-hundred milligrams of left ventricle tissue per heart was collected and lysed with RIP lysis buffer according to the manufacturer's protocol (Sigma, 17701). Anti-Myh10 antibody (1:400, Abcam, ab230823) was used to pull down Myh10-binding RNAs. Normal rabbit IgG was used as negative control. The pulldown RNAs were extracted by using the RNeasy mini kit (Qiagen, 74106). Total RNA sequencing was performed at the IU Genomic Core. The sequence reads were mapped to the mouse genome mm10 using STAR (v2.7.2a) with the the following parameter: '-outSAMmapqUnique 60'. Then unmapped reads were mapped to the human genome hg38 by STAR with the same parameter. Peak calling on mapped reads was performed by MACS2 compared with IgG signals with a cut-off of FDR-adjusted P value <0.01.

ScRNA-seq data analysis
The gene count matrix of scRNA-seq data from a foetal human heart was retrieved from a previously published report 25 . Normalization, dimensionality reduction and clustering of scRNA-seq data were performed using Seurat (v3) (ref. 72) according to the parameters previously utilized 25 . The t-distributed stochastic neighbour embedding was used to visualize single-cell clusters, and gene expression in a reduced 2D space. Annotation of cell clusters was mapped from the annotation of the clusters presented in this report 25 .

Statistics and reproducibility
Statistical results were analysed using Prism 8 (GraphPad Software). Unpaired Student's t-tests (two groups) and two-way analysis of variance were used in data analyses unless specified in the legend. Column bar plots show mean ± standard error of the mean (s.e.m.) from at least three independent experiments. Statistical parameters, including the exact value of n, statistical test method and statistical significance, are reported in the figures, figure legends and related data resources. Data were judged to be significant when P < 0.05. No statistical method was used to pre-determine sample size. No animal or data were excluded from the analyses. The experiments were not randomized. The Investigators were not blinded to allocation during experiments and outcome assessment.

Reporting summary
Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data availability
RNA-seq data of WT and LIPTER KO hiPSC-CMs, and WT and LIPTER (Tg) mouse hearts were deposited in NCBI GEO database (GSE175370). The UCSC Genome Browser view of RNAseq results from T2DM and NF human hearts can be found at https://genome.ucsc.Edu/s/samal/hg38_Yan-gLei_ILMN550. The metabolomics data can be accessed in Supplementary  Table 1. The datasets used to determine LINC00881 expression in human hearts can be accessed at NCBI Accession GSE64283 as shown in Extended Data Fig. 1c 73  showing predicted peptide sequence of each ORF. A FLAG tag was inserted right after each ORF. b, Western blotting detection of peptide expressions in HEK293T cells transfected with positive control, vector, and FLAG tagged ORF1, ORF2 or ORF3. Positive control is an irrelevant FLAG tagged protein coding gene. The experiment was carried out 2 times with similar outcomes. c, Immunofluorescent staining to detect expressions of FLAG tagged peptides in HEK293T cells transfected with positive control, vector, and FLAG tagged ORF1, ORF2 or ORF3. d, RT-qPCR detecting LINC00881 RNA expression levels in HEK293T cells transfected with positive control, vector, and FLAG tagged ORF1, ORF2 or ORF3. Bars are presented as mean ± s.e.m. Unpaired twotailed t-test is used for comparison. **p < 0.01, ***p < 0.001, (n = 3 independent experiments). Source numerical data and unprocessed blots are available in source data. Fig. 9 | LIPTER transgene displays cardiac protective effects in HFD-fed mice. a, Representative FACS results showing cleaved-Caspase-3/ cTnT double positive cell ratios in control and LIPTER OE hiPSC-CMs treated with or without 400 μM palmitic acid for 4 days. The experiment was carried out 3 times with similar outcomes. b, Representative Immunofluorescent images of TUNEL and cTnT co-staining in control and LIPTER OE hiPSC-EBs treated with 400 μM palmitic acid for 4 days. The experiment was carried out 3 times with similar outcomes. c, Schematic of LIPTER(Tg) mouse generation using CRISPR/Cas-9, with the Rosa26 locus targeted by a single gRNA. d, RT-qPCR examining LIPTER expression levels in WT and LIPTER(Tg) mouse hearts. (n = 5 mice per group). e, Representative images of WT and LIPTER (Tg) mouse hearts after 7 months of HFD feeding. The heart weight/tibia length ratio was calculated. (n = 5 mice in WT, and n = 6 in Tg). f, Representative immunofluorescent images of TUNEL and cTnT co-staining in mouse heart sections after 7 months of HFD feeding. The experiment was carried out 4 times with similar outcomes. g, Quantification of TUNEL + CM ratios in (f). (n = 4 mice per group). h, RIP-seq results showing LIPTER pulldown in LIPTER (Tg) mouse heart using anti-Myh10 antibody, with the red arrow indicating enriched signals on exon 3 of transgenic LIPTER. Control Actb and Gapdh RNAs were not pulled down by anti-Myh10 antibody. The experiment was carried out in 2 mice each genotype with similar outcomes. In d, e, g, bars represent mean values ± s.e.m. Unpaired two-tailed t-test is used for comparison. ***p < 0.001, *p < 0.05, (n.s), no significance. Source numerical data are available in source data.