A Novel Y-Specific Long Non-Coding RNA Associated with Cellular Lipid Accumulation in HepG2 cells and Atherosclerosis-related Genes

There is an increasing appreciation for the role of the human Y chromosome in phenotypic differences between the sexes in health and disease. Previous studies have shown that genetic variation within the Y chromosome is associated with cholesterol levels, which is an established risk factor for atherosclerosis, the underlying cause of coronary artery disease (CAD), a major cause of morbidity and mortality worldwide. However, the exact mechanism and potential genes implicated are still unidentified. To date, Y chromosome-linked long non-coding RNAs (lncRNAs) are poorly characterized and the potential link between these new regulatory RNA molecules and hepatic function in men has not been investigated. Advanced technologies of lncRNA subcellular localization and silencing were used to identify a novel intergenic Y-linked lncRNA, named lnc-KDM5D-4, and investigate its role in fatty liver-associated atherosclerosis. We found that lnc-KDM5D-4 is retained within the nucleus in hepatocytes. Its knockdown leads to changes in genes leading to increased lipid droplets formation in hepatocytes resulting in a downstream effect contributing to the chronic inflammatory process that underpin CAD. Our findings provide the first evidence for the implication of lnc-KDM5D-4 in key processes related to fatty liver and cellular inflammation associated with atherosclerosis and CAD in men.


Materials and Methods
Primer design. The human lincRNA catalogues 15,16 were used to select Y-specific lincRNA genes based on their expression in tissues relevant to the liver and CAD, such as heart, liver and white blood cells (WBCs) (Supplementary Materials and Methods). Primers for SYBR ® Green-based real-time polymerase chain reaction (PCR) gene expression analysis were designed using the free online software Primer-BLAST (Basic Local Alignment Search Tool) following the recommendations for probe melting temperature (Tm) values, GC content, and amplicon length provided by Thornton & Basu 17 . All the primers for lincRNAs were designed from the DNA template sequence annotated on LNCipedia, the database for human lncRNA genes and transcripts 18 . Primer sequences are available in Supplementary Table S1.

(Supplementary Materials and Methods).
Human tissue RNA panel. The Human Tissue RNA Panel was obtained from Ambion ® FirstChoice ® Human.
Total RNA Survey Panel which consists of 10 µg pools of total RNA (stored in 1 mmol/L of sodium citrate) from 20 different normal, human tissues including adipose, bladder, brain, cervix, colon, esophagus, heart, kidney, liver, lung, ovary, placenta, prostate, skeletal muscle, small intestine, spleen, testes, thymus, thyroid, and trachea. Each pool is comprised of RNA from 3 tissue donors (with at least 1 male donor per tissue except for the cervix, placenta, and ovary RNA sample pools) and underwent a stringent DNase treatment. After storage at −80 °C on arrival, RNA samples were thawed at 37 °C then placed on ice prior use as recommended by the manufacturer's instructions.

Blood collection.
To study the expression of lincRNAs expressed in leucocytes in men, five healthy male volunteers (positive control group) and two healthy females (negative control group), between 18-50 years old were recruited at the Federation University Australia (Mount Helen campus). Participants were required to meet the following inclusion criteria: current non-smokers, free of recent surgery, ambulatory aids, and acute joint injury, free of underlying medical conditions such as heart disease, un-medicated hypertension and respiratory disorders in order to avoid skewing the results. Participants' blood was withdrawn in Ethylenediaminetetraacetic acid (EDTA) blood collection tubes (tubes commonly used in routine haematology and furnished by the phlebotomist) by a qualified phlebotomist at Mount Helen campus. Then, blood samples were immediately stored on ice and total RNA was purified within an hour of collection to preserve the RNA integrity. This study was approved by the Federation University Ethics Committee at Mount Helen campus and carried out in accordance with Federation University relevant guideline and regulations. were resuspended in 50 µL RNase-free water followed by a column DNase treatment (1 µL per 1000 ng of RNA) using the Recombinant DNase I (Ambion ® , Applied Biosystem ® ) according to manufacturer's instructions. RNA yield and purity were measured by absorbance using a NanoDrop ™ 2000 Spectrophotometer (Thermo Fisher Scientific ® , Australia). The ratio (A 260 nm /A 280 nm ) of ~2.0 was accepted as "pure" for RNA. Finally, the RNA samples were stored at −80 °C until use.
Then, the knockdown was confirmed by real-time PCR.
Reverse transcription and Atherosclerosis RT 2 Profiler ™ PCR Array. The Human Atherosclerosis RT 2 Profiler PCR Array (PAHS-038ZE) from SA Biosciences (Qiagen ® ; Catalogue No. 330231) was performed following the manufacturer's recommendations (Supplementary Materials and Methods). Then, the human tissue-specific network webserver GIANT 22 was used to generate the potential tissue-specific functional interactions between the atherosclerosis-relevant genes.
Statistical analysis. Statistical analysis of data was performed using Prism (GraphPad Software). Data were analysed using a Student's unpaired t-test to test significance of data comparisons between cell treatments, with p-values < 0.05 considered as statistically significant.
The effects of fatty acid palmitate on the transcript expression of Y chromosome-linked lincR-NAs in a steatosis HepG2 cell line. Positive correlations between CAD and high levels of free fatty acids (FFAs) in steatosis hepatocytes have been previously demonstrated in vitro using HepG2 cells [11][12][13][14] . To create a FFA-induced steatosis liver cell line, HepG2 cells were stimulated with 0.3 mmol/L of palmitate for 24 hours to induce lipid accumulation within the cells 14 . Two methods were used to assess the results: quantification of the lipid accumulation by absorbance measurement (Fig. 2A,B) and examination of the intracellular distribution of lipid droplets by Oil Red O (ORO) staining using microscopy (Fig. 2C). To confirm the ability of our assay to generate the required cell phenotypes, we compared the magnitude of lipid accumulation within the target cells with that generated by the positive control for inducing steatosis, the chloroquine (based on manufacturer's recommendations). Results with chloroquine showed a similar significant increase in lipid accumulation within the treated cells when compared to the control cells ( Fig. 2A). The absorbance measurement of the lipid quantification (Fig. 2B) suggested that HepG2 cells had acquired the palmitate-induced steatosis phenotype. These results were further validated by microscopy demonstrating a clear increase of ORO-stained lipid droplets (red) in the peri-nuclear region of the cells in comparison to the control cells (Fig. 2C). Real-time PCR results showed that 24 h-treatment with palmitate triggered a significant 2.16-fold increase in the expression of lnc-KDM5D-4:1 (p-value = 0.00216) in HepG2 cells. No significant increase in the expression of the transcripts lnc-ZFY-1:1, lnc-ZFY-2:1, lnc-RBMY1B-1:1, lnc-RBMY1B-1:4, lnc-USP9Y-1:4, lnc-HSFY2-3:6, and HULC was observed (p-value > 0.05) (Fig. 2D).

The effects of insulin on the lncRNA transcript lnc-KDM5D-4:1 expression in an insulin-resistant
HepG2 cell line. To investigate the possible correlation between steatosis and insulin-resistance, the potential effect of insulin resistance on the expression of lnc-KDM5D-4 was studied. To create an insulin-resistant cell line, HepG2 cells were stimulated with 10 nM of insulin solution over 0, 0.25, 0.5, 2, 4, 8, 24, and 48 h replacing the media and insulin every 8 h. Total cellular protein was extracted using RIPA buffer and 20 μg of total cellular protein was subjected to SDS-PAGE, and subsequently blotted onto Immobilon-PVDF membrane. The relative levels of phosphorylated protein kinase B (pAkt) or phosphorylated insulin-receptor (pIR) were detected relative to Total Akt and Total IR respectively (Fig. 2E). Western blot analysis showed that following 30 min of insulin treatment, there was an increase in the levels of pAkt that were further increased at 2 and 4 h of insulin treatment, as similarly for pIR. The level of phosphorylation then rapidly decreased at 8 h and became almost undetectable at 24 and 48 h. The level of phosphorylation of the IR decreased from 24 h. Overall, there was no change in Total Akt and IR proteins observed, suggesting that the cells had obtained the insulin-resultant phenotype (Fig. 2E).
To determine if lnc-KDM5D-4 responded to insulin-resistance, its expression was analysed by real-time PCR. Results showed no significant change in the expression of lnc-KDM5D-4 in cells stimulated with insulin for 24 h compared to the control cells (Fig. 2F).
A survey of the expression of the lncRNA transcript lnc-KDM5D-4:1 in 21 different normal human tissues. To gain a better understanding about the biological role of lnc-KDM5D-4, we examined its expression across 21 normal human tissue types including adipose, bladder, brain, cervix, colon, esophagus, heart, kidney, liver, lung, ovary, placenta, prostate, skeletal muscle, small intestine, spleen, testes, thymus, thyroid, trachea, male and female leucocytes from WBCs. The real-time PCR analysis revealed that lnc-KDM5D-4 shows ubiquitous expression with the highest level of expression in the spleen and the heart. As expected, no expression was found in female tissues such as cervix, ovary, and placenta as well as in female leucocytes (Fig. 3).

Subcellular localization of lnc-KDM5D-4:1 transcripts in HepG2 cells. To gain fundamental insights
into the biology and potential cellular role of lnc-KDM5D-4 in the regulation of cellular functions 24 , we examined its subcellular localization using RNA fluorescence in situ hybridization (RNA FISH) with single-molecule sensitivity. To achieve a controlled system as background image, the assay was performed without addition of glacial acetic acid (AA) to the fixation solution (Fig. 4A,C,E). This way, although significant cytoplasmic signals were observed for the actin beta (ACTB) housekeeping gene probe (green) (Fig. 4C), any signal obtained was the result of the lnc-KDM5D-4:1 probe (red) (Fig. 4E). These results suggested that lnc-KDM5D-4:1 transcripts were localised only in the nucleus. By adding 1% of AA into the fixation solution, the fluorescence of single-RNA molecule of lnc-KDM5D-4:1 (red) displays a strictly nuclear distribution (Fig. 4F,G and Supplementary Video S1). The ACTB RNA FISH results (Fig. 4C,D) confirmed the efficiency of reagents and methods used for this assay. As negative controls, wells with only secondary probes were tested and no signal displays were seen, as expected, (Fig. 4A,B) confirming that signals obtained for ACTB and lnc-KDM5D-4:1 are not due to the mere expression of the secondary probes. RNA FISH results were then confirmed by real-time PCR using nuclear and cytoplasmic RNAs, separately isolated from HepG2 cells, as shown in Fig. 4H.
Previous studies on transcriptional regulation models showed that lncRNAs, notably lincRNAs, may operate in cis by regulating their immediate neighbouring protein-coding genes leading to either an increase or reduction of their expression 16,25,26 . After confirmation of the lnc-KDM5D-4 knockdown in HepG2 cells, we examined the potential transcriptional activity of the nearest neighbouring protein-coding genes by lnc-KDM5D-4 such as KDM5D (lysine demethylase 5D), EIF1AY (eukaryotic translation initiation factor 1A, Y-linked), and RPS4Y2 (ribosomal protein S4, Y-linked 2) located within 1.1 MB window centred on lnc-KDMM5D-4. Real-time PCR analysis showed that there were no significant changes (p-value > 0.05) in the expression of these neighbouring Y chromosome protein-coding genes in cells transfected with the lnc-KDM5D-4_GapmeR_1 compared to the control cells (Scramble) (Fig. 6).

Discussion
Genetic variation within the Y chromosome has been linked to human lipid levels and cardiovascular disease [2][3][4]6,27,28 . Here we provide novel data suggestive of a potential role for one of the Y chromosome lin-cRNA, lnc-KDM5D-4, in the hepatic metabolism of lipids and atherosclerosis. Our findings demonstrate that lnc-KDM5D-4 is a nuclear-retained lincRNA that has a regulatory effect on the mRNA expression of genes such as Perilipin 2 (PLIN2), that in turn, lead to increased lipid droplets formation in hepatocytes. We propose that this link may, in part, contribute to the association between certain Y chromosome haplogroups and atherosclerosis.
There is currently no information regarding Y chromosome-linked lncRNAs in human traits and disease. This is due to the fact that the Y chromosome is routinely excluded from genome analysis studies into the identification and the functions of lncRNAs in mammals 29,30 . Indeed, due to the haploid nature of the Y chromosome, the usual methods of analysis (such as genome-wide association studies (GWAS)) cannot be employed to investigate variations. This study provides the first evidence of the expression of Y-linked lincRNA transcripts such as lnc-KDM5D-4:1, lnc-ZFY-1:1, lnc-ZFY-2:1, lnc-RBMY1B-1:1, lnc-RBMY1B-1:4, lnc-USP9Y-1:4, and lnc-HSFY2-3:6 in human hepatocellular carcinoma (HCC). Further research focusing on these lincRNAs between primary human hepatocytes and HepG2 cells should be done as these lincRNAs may also play a role in HCC, and could be potential biomarkers for the diagnosis of this cancer in men. Our data also provides the first evidence for the up-regulation of the Y-linked lincRNA, known as lnc-KDM5D-4, in response to palmitate treatment in HepG2 cells. However, lnc-KDM5D-4 was not differentially expressed or affected in insulin-resistance HepG2 cells, demonstrating that the results found with the steatosis cell model were independent of those found in the insulin-resistance cells. These results confirmed that the significant changes in expression of lnc-KDM5D-4 after the FFA-palmitate treatment was triggered by the steatosis phenotype, and not by the insulin-resistance phenotype. It was suggested that lnc-KDM5D-4 was then implicated in hepatic steatosis which can occur independently of insulin resistance in the liver. Given the role of FFAs in the pathogenesis of CAD 31-33 , these results suggest that lnc-KDM5D-4 may play a role in the hepatic metabolism of lipids -a process of well-known relevance to atherosclerosis. Whether aberrant expression of this lincRNA plays an insignificant role in this context, or if it is a mere consequence of disease pathology remains an open question. Further research on lnc-KDM5D-4 should be done to study the role of this Y-linked lincRNA in steatosis-associated atherosclerosis.
Lnc-KDM5D-4:1 transcripts were found to be expressed in adipose, bladder, brain, colon, esophagus, heart, kidney, liver, lung, prostate, skeletal muscle, small intestine, spleen, testes, thymus, thyroid, trachea, and leucocytes suggesting that this lincRNA is widely expressed in male tissues. These findings provide a novel expression profile for lnc-KDM5D-4 across human healthy tissues. Based on these data, we believe that this lincRNA may have other molecular and physiological roles in men. Atherosclerosis is driven by a chronic inflammatory process. Lipid disturbances and other risk factors are thought to cause endothelial injury resulting in monocyte adhesion and migration to the intima, as well as the release of cytokines and growth factors. Low-density lipoprotein (LDL) particles travelling in the blood and carrying cholesterol and triglycerides from the liver to other body tissues get through the endothelium layer due to their size and density, and become oxidized. After migration to the sub-endothelial space, monocytes differentiate into macrophages, which are then able to ingest oxidized-LDL, forming specialized foam cells. Macrophages are not able to process the oxidized-LDL and ultimately grow and rupture depositing a greater amount of oxidized cholesterol into the artery wall. This triggers the recruitment of more monocytes, thus increasing the inflammation and continuing the cycle. This inflammation leads to subendothelial accumulation of fatty substances called atheromatous plaques. In the hepatocytes, the underexpression of the Y chromosome-linked lincRNA lnc-KDM5D-4 results in an overexpression of the gene perilipin 2 (PLIN2) involved in lipid droplet formation within the cells. This increase of expression of PLIN2 may consequently initiates the 'fatty liver' or hepatic steatosis promoting atherosclerosis in the coronary arteries of men.
SCIeNTIFIC RepoRts | 7: 16710 | DOI:10.1038/s41598-017-17165-9 The main known function of lncRNAs to date is regulation of gene expression 34 . Lnc-KDM5D-4:1 RNA FISH results suggested that lnc-KDM5D-4:1 is a lincRNA retained within the nucleus of hepatocytes. Concomitantly, these findings were in agreement with the previous observations showing that lncRNA transcripts were enriched in the cell nucleus in HepG2 cells 16 . This clearly showed that lnc-KDM5D-4 plays a potential role in biological functions taking place within the nucleus 24,35 such as the establishment and maintenance of nuclear domains 36 , shaping of the 3D organization 35,37 , or acting as enhancer-like RNA by activating its neighbouring genes using a cis-mediated mechanism 38,39 . Indeed, an exclusively nuclear localization would argue against putative lncRNAs encoding short peptide sequences as translation occurs in the cytoplasm. However, given that its silencing does not affect the expression of its surrounding protein-coding genes, this may argue that it acts in trans rather than in cis.
Among the genes that were affected by lnc-KDM5D-4 knockdown in HepG2 cells, PLIN2 is known to be implicated in lipid metabolism. This protein-coding gene is located on chromosome 9 and belongs to the perilipin gene family that regulate intracellular lipid storage droplets and very low-density lipoprotein (VLDL) secretion 40 . Overexpression of PLIN2 has been shown to trigger an increase in lipid droplets formation within hepatocytes via the up-regulation of the peroxisome proliferator-activated receptor gamma isoform (PPAR gamma) 41 . Furthermore, the up-regulation of PLIN2 was previously associated with liver steatosis 42 and atherogenesis 43 . More recently, PLIN2 expression was also associated with atherosclerosis in patients with carotid stenosis 44 . On the other hand, the loss of PLIN2 has resulted in reduction of liver steatosis and inflammation 45 . An overview of the involvement of lnc-KDM5D-4 and its potential interaction with PLIN2 in the context of atherosclerosis and CAD is proposed in Fig. 7.
Lnc-KDM5D-4 seems to be not conserved between human and rodents 18 which could be due to the unique forces that drive the evolution of the Y chromosome. It is therefore not possible to study further this lncRNA in rodents.
In conclusion, this is the first study on lincRNAs on the human Y chromosome and gene expression analysis. We provide evidence for the potential involvement of one of these lncRNAs, lnc-KDM5D-4, in atherosclerosis and CAD, possibly through the lipid metabolism-associated gene PLIN2 in hepatocytes. Overall, our data adds to the evidence that the human Y chromosome plays an important role in cardiovascular disease in a male specific manner and provides novel insight into potential new therapeutic targets for CAD.