LINC01013 Is a Determinant of Fibroblast Activation and Encodes a Novel Fibroblast-Activating Micropeptide

Myocardial fibrosis confers an almost threefold mortality risk in heart disease. There are no prognostic therapies and novel therapeutic targets are needed. Many thousands of unannotated small open reading frames (smORFs) have been identified across the genome with potential to produce micropeptides (< 100 amino acids). We sought to investigate the role of smORFs in myocardial fibroblast activation. Analysis of human cardiac atrial fibroblasts (HCFs) stimulated with profibrotic TGFβ1 using RNA sequencing (RNA-Seq) and ribosome profiling (Ribo-Seq) identified long intergenic non-coding RNA LINC01013 as TGFβ1 responsive and containing an actively translated smORF. Knockdown of LINC01013 using siRNA reduced expression of profibrotic markers at baseline and blunted their response to TGFβ1. In contrast, overexpression of a codon-optimised smORF invoked a profibrotic response comparable to that seen with TGFβ1 treatment, whilst FLAG-tagged peptide associated with the mitochondria. Together, these data support a novel LINC01013 smORF micropeptide-mediated mechanism of fibroblast activation. Graphical Abstract TGFβ1 stimulation of atrial fibroblasts induces expression of LINC01013, whose knockdown reduces fibroblast activation. Overexpression of a smORF contained within LINC01013 localises to mitochondria and activates fibroblasts Supplementary Information The online version contains supplementary material available at 10.1007/s12265-022-10288-z.


Introduction
Myocardial fibrosis is a common final pathology of almost all forms of heart disease. It leads to arrhythmia and haemodynamic failure, and is an independent predictor of mortality [1][2][3]. No current treatment for myocardial fibrosis has been shown to offer prognostic benefit [4]. Although some major determinants of myocardial fibrosis are recognised, inhibition of upstream molecules, such as transforming growth factor beta-1 (TGFβ1) has detrimental pleiotropic effects, limiting their viability as therapeutic targets [5,6]. Therefore, identification of novel therapeutic targets, downstream of TGFβ1, is urgently needed.
Cardiac fibroblasts comprise a heterogeneous mesenchymal cell population that produce and remodel extracellular matrix, and whose activation is a precursor to the fibrotic response [7]. The final consequence of fibroblast activation is their transdifferentiation into extracellular matrix (ECM) producing and contractile myofibroblasts, the persistence of which turns the fibrotic response from physiological to maladaptive, with the production of pathological quantities of extracellular matrix. Markers of fibroblast activation include interleukin-11 (IL11), as a key determinant of fibroblast activation [8], snail (SNAI1), which is necessary for myofibroblast transdifferentiation and fibrosis [9][10][11][12][13], and periostin (POSTN) and alpha-smooth muscle actin (α-SMA, ACTA2) which are both established markers of myofibroblast transdifferentiation [7,14].
Small open reading frames (smORFs) are increasingly recognised to play a significant role in regulating cellular processes [15,16]. These short regions with coding potential can be identified both within the upstream and downstream regions of messenger RNAs traditionally labelled as untranslated (UTRs), and within long non-coding (LNC) RNAs-an abundant class of RNAs with largely undescribed functional roles [17]. In order to visualise smORFs which are actively translated, ribosome profiling (Ribo-Seq) leverages the property of ribosomes to protect RNA from endonuclease digestion. Sequencing and mapping of ribosome-protected fragments (RPFs) of RNAs generate a snapshot of regions of active translation [18][19][20]. From these studies, it is now clear that many regions, previously annotated as non-coding, in fact contain smORFs which are not only actively translated but can be demonstrated to produce biologically active micropeptides with a diverse range of actions, including in the cardiovascular system [15,16,[21][22][23][24][25].
LINC01013 is a largely uncharacterised intergenic lncRNA, previously considered to be non-coding, which has been seen to have a role in chondrogenesis [26], cancer progression [27,28] and regulation of DNA repair in endothelial cells [29]. A recent epigenomics study found it to be co-ordinately expressed with a key fibrotic gene CCN2/ CTGF [30] in the context of human umbilical cord vascular endothelial (HUVEC) cells, suggesting a potential role in fibrosis. To date, no study has addressed the possibility of LINC01013 encoding a bioactive peptide.
Here, we demonstrate a direct role of LINC01013 in the response of cardiac fibroblasts to TGFβ1. Moreover, we use ribosome profiling to identify a small open reading frame (LINC01013ORF) with strong evidence of translation. LINC01013ORF is independently able to activate cardiac fibroblasts and encodes a small peptide that localises to the mitochondria. Taken together, these data show that LINC01013 and its encoded micropeptide are likely to play a key role in fibroblast activation and represent novel targets for antifibrotic therapy.

Cell Culture of HeLa Cells
Human HeLa cells (ATCC) were cultivated in a humidified incubator at 37 °C with 5% CO 2 using Dulbecco's modified eagle medium (DMEM) with high glucose (4.5 g/l), 10% foetal bovine serum (FBS), 2 mM l-glutamine and 1 mM sodium pyruvate. Medium was changed after 48 h and cells were split at confluency rates of 80-90% using standard trypsinization techniques.

RNA-Seq and Ribo-Seq
mitochondrial RNA, ribosomal RNA and transfer RNA using Bowtie [33] were discarded as previously described in [31]. The remaining reads were aligned to the human genome (hg38) using STAR [34] and read coverage on known coding genes was calculated using featureCounts from the Subread package [35]. Differential expression analysis was carried out using DESeq2. Ribotaper [36] was used to detect actively translated small open reading frames (smORFs) based on Ribo-Seq and RNA-Seq reads.

LINC01013ORF3 Overexpression
A codon optimised sequence with identical coding potential but altered RNA sequence to LINC01013ORF was cloned into expression vector pcDNA3.1 (Addgene) and expressed under control of a CMV promoter/enhancer. Control vector contained eGFP also expressed from CMV promoter/ enhancer in the pMaxGFP vector (Lonza).

Immunofluorescence
Immunofluorescence experiments were performed as described previously [25] with slight adjustments. HeLa cells were grown on 15 mm glass slides for 24 h and transfected with 3xFLAG tagged LINC01013ORF (LIN-C01013ORF-3xFLAG) in a pcDNA3.1 vector, using Lipofectamine 2000 reagent. Twenty-four-hour post transfection, cells were fixed with 4% paraformaldehyde for 10 min at room temperature and washed three times with ice-cold phosphate-buffered saline (PBS). Cells were permeabilized and blocked for 1 h at room temperature using 2.5% bovine albumin serum, 10% anti-goat serum and 0.1% Triton X or 30 mg/mL digitonin followed by 3 washing steps. Whilst Triton X permeabilizes the outer and inner mitochondrial membrane and therefore allows the staining of mitochondrial matrix proteins, digitonin leaves the inner mitochondrial membrane intact and prevents staining of mitochondrial matrix proteins. Overexpressed LINC01013ORF-3xFLAG was stained for 1 h at room temperature using an anti-FLAG mouse monoclonal antibody (1:500, F1804, Sigma Aldrich) and co-stained with the mitochondrial matrix protein ATPIF (1:1000, rabbit ATPIF1 #13,268, Cell Signaling Technology; Danvers, MA, USA) or the mitochondrial outer membrane protein TOM20 (1:200, rabbit TOM20 #42,406, Cell Signaling Technology; Danvers, MA, USA). Slides were then washed and incubated with fluorescently-labelled secondary antibodies (1:500, Alexa Fluor 488 anti-rabbit & Alexa Fluor 594 anti-mouse; Invitrogen, Carlsbad, CA, USA) for 30 min at room temperature. Cells were washed again, stained with 4-6-diamidino-2-phenylindole (NucBlue Fixed Cell ReadyProbes Reagent, R37606, Thermo Fisher) for 5 min at room temperature and mounted onto glass slides using ProLongTM Gold antifade reagent (Molecular Probes; InvitrogenTM). Images were visualised using a LEICA SP8 confocal microscope using a × 63 objective. Image analysis was performed using the Leica confocal software Las X (v3.5.2) and ImageJ (v1.52a).

In Vitro Translation of LINC01013 RNA
The full LINC01013 RNA sequence (including 5′UTR, ORF region and 3′UTR) was synthesized and inserted into an expression plasmid by Genewiz Europe (Leipzig, Germany; construct available upon request). Linearized plasmid DNA (0.5 µg) was transcribed and translated in vitro as described previously [25] using the TnT Coupled Wheat Germ Extract system (Promega, Mannheim, Germany) in the presence of 10 mCi/mL [35S]-methionine (Hartmann Analytic, Braunschweig, Germany) according to manufacturer's instructions. To visualise the translation products, 5µL lysate was denatured for 2 min at 85 °C in 9.6 µL Novex Tricine SDS Sample Buffer (2X) (Thermo Fisher Scientific) and 1.4 µL DTT (500 mM). Proteins were separated on 16% Tricine gels (Invitrogen) for 1 h at 50 V followed by 3.5 h at 100 V and blotted on PVDF-membranes (Immobilon-PSQ Membrane, Merck Millipore).

Statistics
Continuous data were expressed as mean ± SEM. For two groups with normal distribution, data were compared using Student's t-test with multiple test correction if required. For normally distributed data with more than two groups and one dependent variable, one-way ANOVA was used, with Tukey's multiple test correction. For normally distributed data with more than two groups and two dependent variables, two-way ANOVA was used with Tukey's multiple test correction. Nonparametric data of more than two groups was analysed using Friedmann's test. Statistical analysis was performed with Prism 8.4.3.

LINC01013 Is Expressed in Activated Human Cardiac Fibroblasts and Contains an Actively Translated smORF
In order to establish the profile of LINC01013 RNA expression in cardiac fibroblasts, we first evaluated transcripts using data previously derived from primary human cardiac fibroblasts (HCFs) activated with TGFβ1 and analysed by RNA-Seq [8]. This identified two alternate transcripts: a major form corresponding to NR_038981.1 and a shorter, minor form corresponding to NR_146223.1 (Fig. 1a). We interrogated transcripts for potentially translated regions using ribosome profiling (Ribo-Seq) data likewise obtained from TGFβ1-stimulated HCFs [31]. This identified four smORFs: three of these had only limited evidence of translation but one, located in exon 4 of and within the major NR_038981.1 transcript, had a notably high density of ribosome footprints with consistent trinucleotide periodicity, highly indicative of active translation (Fig. 1b-c). This smORF spans 168 bp, which has an ATG in a moderate Kozak context (AAG ATG A), with coding potential for a 56AA micropeptide of a predicted mass of 6.3KDa. The NR_038981.1 transcript (hereon termed LINC01013) was therefore prioritised for further study together with the smORF contained therein (hereon named LINC01013ORF).
To evaluate LINC01013 expression in vivo, we interrogated single cell nucleus RNA sequencing (Nuc-Seq) data previously compiled from 14 healthy human hearts [38]. LINC01013 was detected almost exclusively in fibroblasts, myeloid cells and neuronal cells (Fig. 1e). Within the fibroblast population, LINC01013 was associated with an extracellular matrix-producing cluster of fibroblasts (described as 'FB4' in Litviňuková et al. [38]), which express a range of markers of TGFβ1 stimulation and fibroblast activation (Fig. 1f). Conversely, LINC01013 was not associated with the quiescent 'FB1' fibroblast subcluster, which has the lowest expression of TGFβ1-and fibroblast activation-associated genes. LINC01013 was expressed in fibroblasts across all regions of the healthy human heart (Supplementary Fig. 1; information on fibroblast subtypes can be found in Supplementary Table 1).
Upregulation of the LINC01013 transcript under the experimental conditions used here was confirmed using cultured primary HCFs analysed by qPCR. LINC01013 expression was readily detectable and consistently and robustly upregulated after treatment with 5 ng/ml TGFβ1 for 24 h (Fig. 1d).

siRNA-Mediated Knockdown of LINC01013 Reduces the Fibrotic Phenotype at Baseline and After TGFβ1 Stimulation
To evaluate the effect of knocking down LINC01013 RNA on the fibrotic phenotype, we used a model of fibroblast activation: HCFs were serum-starved for 16 h, then treated with 5 ng/ml of TGFβ1 for 24 h, after which RNA markers of fibroblast activation were quantified: α-smooth muscle actin (ACTA2), interleukin 11 (IL11), periostin (POSTN), fibronectin (FN1) and collagen 1α1 (COL1A1) and were robustly increased as determined by qPCR, whilst matrix metalloproteinase (MMP2) remained unchanged (Supplementary Fig. 2).

at 24 h.
To evaluate the effect of LINC01013 knockdown on TGFβ1-mediated fibroblast activation, HCFs were serum starved for 16 h before treatment with either anti-LINC01013 siRNA or nontargeting control siRNA for 24 h and further cultured in either standard media or standard media + 5 ng/ml TGFβ1. LINC01013 knockdown blunted the response to TGFβ1, reducing ACTA2 and IL11 (Fig. 2b,  c), but not POSTN (Fig. 2d) expression.

Overexpression of the LINC01013ORF Leads to Fibroblast Activation
We first used in vitro translation to validate production of the LINC01013ORF protein product ( Supplementary  Fig. 4). To evaluate its functional role and to distinguish peptide driven effects, we used a codon-optimised form of LINC01013ORF, cloned into a pcDNA3.1 vector (pcDNA3.1-LINC01013ORF), driven by a constitutive CMV promoter. Control HCFs were transfected with an eGFP expression construct and transfection validated by visualisation of eGFP ( Supplementary Fig. 5). In control cells, TGFβ1 treatment increased myofibroblast transdifferentiation markers: ACTA2 and POSTN, profibrotic transcription factor SNAI1 and its target ECM protein, FN1. Overexpression of LINC01013ORF increased these same markers (Fig. 3a-d) to the same level as seen with TGFβ1.

LINC01013ORF Encodes a Micropeptide that Localises to the Mitochondrial Matrix
To investigate the subcellular localisation of LIN-C01013ORF peptide, we overexpressed a LIN-C01013ORF-3 × FLAG fusion peptide in HeLa cells, and co-stained with DAPI (nuclei), anti-FLAG (LIN-C01013ORF-FLAG) and anti-ATPIF (mitochondrial matrix). When cells were permeabilised with triton-X, the peptide encoded by LINC01013ORF (LINC01013ORFpep) clearly co-localised with ATPIF (Fig. 4) indicating its mitochondrial localisation. Additionally, we permeabilised cells with digitonin which does not penetrate the  , c). High ribosome drop-off at the LINC01013ORF stop codon further supports active translation of this ORF (data not shown). Arrowed lines: intron; grey boxes: exon, green: translated region. d LINC01013 expression in HCFs is increased by TGFβ1 stimulation in vitro, using primers amplifying LINC01013ORF. N = 6. *** = p < 0.0001. e Nuc-Seq data showing that LINC01013 is expressed in fibroblasts, myeloid and neuronal cells. CM: cardiomyocyte, EC: endothelial cell, FB: fibroblast, PC: pericyte, SMC: smooth muscle cell. f LINC01013 expression is associated with markers of TGFβ1 stimulation and fibroblast activation, and ECM-producing fibroblast subtype 'FB4' mitochondrial inner membrane and thus prevents antibody staining of mitochondrial matrix proteins, but permits that of those in the intermembrane space or outer mitochondrial membrane (OMM). Dropout of both ATPIF and FLAG signals, but not TOMM20 (OMM protein) with digitonin permeabilisation (Supplementary Fig. 6), indicates that LINC01013ORF micropeptide, like ATPIF, localises to the mitochondrial matrix.

Discussion
We have demonstrated that LINC01013 is upregulated in the response of cardiac fibroblasts to TGFβ1, and that it harbours a previously undocumented smORF that is translated and encodes a biologically active peptide. Knockdown of LINC01013 reduces markers of fibroblast activation at baseline and blunts the response to TGFβ1, whilst overexpression of LINC01013ORF induces markers of activation to the same level as seen with TGFβ1 stimulation. To our knowledge, LINC01013ORF is the only known translated smORF described to date to be associated with TGFβ1 signalling.
To date LINC01013 has remained largely uncharacterised and the presence of a smORF encoded peptide has not been previously reported. Our data show that LINC01013 is directly implicated in TGFβ1-mediated fibroblast activation and that this effect is mediated, at least in part, due to the presence of a smORF encoded peptide. Whilst this study was under review, others have reported similar activation of LINC01013 by TGFβ1 in aortic valve fibroblasts [39].
TGFβ1-mediated fibroblast activation can occur both via canonical (SMAD-mediated) and noncanonical (SMADindependent) pathways [40]. We propose that the effects of LINC01013ORF peptide may act primarily through the canonical pathway: siRNA-LINC01013 mediated knockdown reduced expression of SMAD-dependent genes ACTA2 and IL11, but did not reduce POSTN (Fig. 2) the expression of which is upregulated by noncanonical p38 MAPK [41]. LINC01013ORF overexpression activated the canonical target genes ACTA2 and SNAI1 (Fig. 3). It also resulted in increased FN1 and POSTN expression which is likely due to direct SNAI1-mediated transcription [42], or activation of the Wnt/β-catenin, pathway which is known to act on a broad range of genes involved in fibroblast activation [43]. Our data therefore align with previous findings of a LINC01013-SNAIL1-FN1 pathway [27], and suggest the translated ORF within LINC01013 may be the active player involved.
Further supportive of its role in TGFβ1 signalling, we show that in the human heart, LINC01013 is positively associated with the previously described [38] subpopulation of fibroblasts that express markers of TGFβ1 activation and ECM production, and negatively associated with a subpopulation of quiescent fibroblasts.
The mechanisms by which LINC01013ORF encoded peptide acts upon TGFβ1 signalling remain uncertain. When overexpressed, LINC01013ORF peptide localises to the mitochondria. Mitochondria-mediated metabolic drivers that broadly activate fibroblasts are well described, acting via reactive oxygen species (ROS) activating p38 and ERK1/2 pathways [44]. It is therefore possible that the profibrotic effect of LINC01013ORF peptide may be through overall metabolic stress. Of note, overexpression can influence a protein localisation and future work should resolve the localisation of the endogenous protein to substantiate the connection to mitochondria and fibroblast activation. FLAG-tag fusion can also potentially influence subcellular localisation, though has not typically induced mitochondrial localisation in the study of other smORFs [25].
More broadly, our data demonstrate how Ribo-Seq can be leveraged in the discovery of novel translated targets. Many lncRNAs have been implicated in cardiac disease [45], including myocyte dysfunction [46] and as potential biomarkers post infarct [47], and our work highlights the importance of considering an active role of biologically active translated smORFs within these: there is a growing body of evidence for functional smORF peptides [22,23,48]. lncRNAs harbouring smORFs may thus potentially act via coding and non-coding mechanisms, with synergistic or independent functions.
In conclusion, we propose that LINC01013 functions downstream of TGFβ1 and may represent a potential therapeutic target in limiting fibroblast activation. We have demonstrated it to be both necessary and sufficient for fibroblast activation at the RNA level, acting potentially via the canonical TGFβ1 pathway. Fibrosis is a determinant of clinical outcome in the heart, liver, lung and kidneys, and so large cohorts of patients with established fibrosis, or who are at risk of its development, can potentially benefit from antifibrotic therapy. To our knowledge, LINC01013ORF is the first smORF to be directly implicated in fibroblast activation, and our work highlights the importance to consider biologically active smORFs within putatively non-coding regions. Finally, whilst our observations have been made in cardiac fibroblasts, due to mechanistic homology, they are potentially translatable to fibrosis in other tissues.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.