LncRNA PVT1 Mediated by ZFP36L2 Regulates Myocardial Ischemia/Reperfusion Injury and Attenuates Mitochondrial Fusion and Fission via Activating miR-21-5p/MARCH5 Axis

Weifeng Huang Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Qin Tan Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Yong Guo Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Yongmei Cao Department of Critical Care Medicine. Shanghai Jiaotong University A liated Sixth People's Hospital Jiawei Shang Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Fang Wu Departmen of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Feng Ping Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Wei Wang Department of Critical Care Medicine, Shanghai Jiaotong University A liated Sixth People's Hospital Yingchuan Li (  yingchuan.li@sjtu.edu.cn ) Shanghai Jiaotong University A liated Sixth People's Hospital


Abstract Background
Among several leading cardiovascular disorders, ischemia-reperfusion (I/R) injury causes severe manifestations including acute heart failure, in ammation, and systemic dysfunction. Recently, there has been increasing evidence suggesting that alterations in mitochondrial morphology play a role in the prognoses of cardiac disorders. Long non-coding RNAs (lncRNAs) form major regulatory networks to modify gene transcription and translation. While several roles of lncRNAs have been explored in cancer and tumor biology, their implications on mitochondrial morphology and functions remain to be elucidated.

Methods
The functional roles of ZFP36L2 and lncRNA PVT1 were determined by a series of cardiomyocyte hypoxia/ reoxygenation (H/R) in vitro and myocardial I/R injury in vivo experiments. Quantitative Reverse transcription-polymerase chain reaction (qRT-PCR) and western blot analysis were used to detect the mRNA levels of ZFP36L2 and mitochondrial ssion and fusion markers in the myocardial tissues and cardiomyocyte. Cardiac function was determined by immunohistochemistry, H&E, Masson's staining and echocardiogram. Ultrastructural analysis of mitochondrial ssion was performed using transmission electron microscopy (TEM). The mechanistic model of PVT1 with ZFP36L2 and miR-21-5p with MARCH5 was detected by subcellular fraction, RNA pull down, FISH, and luciferase reporter assays.

Results
In this study, we report a novel regulatory axis involving lncRNA PVT1, microRNA miR-21-5p, and E3 ubiquitin ligase MARCH5, which alters mitochondrial morphology during myocardial I/R injury. Using an in vivo I/R injury mouse model and in vitro cardiomyocyte H/R model, we observed that zinc nger protein ZFP36L2 directly associated with PVT1 and altered mitochondrial ssion and fusion. PVT1 also interacted with miR-21-5p and suppressed its expression and activity. Furthermore, we identi ed MARCH5 as a modi er of miR-21-5p, and expression of MARCH5 and its effect on mitochondrial ssion and fusion were directly proportional to PVT1 expression during H/R injury.

Conclusions
Our ndings demonstrated that manipulation of PVT1-miR-21-5p-MARCH5-mediated mitochondrial ssion and fusion via ZFP36L2 may be a novel therapeutic approach to regulate myocardial I/R injury.

Background
Hypoperfusion of the heart for short periods of time in response to sepsis, transplantation, or other syndromes is known as ischemia. Subsequent reperfusion or restoration of blood ow in some conditions causes injuries to ischemic tissues, known as ischemia-reperfusion (I/R) injury. Prolonged I/R injury can lead to myocardial infarction and acute coronary syndrome [1]. Prolonged hypoxic conditions lead to anaerobic respiration and dysregulation of the electron transport chain in mitochondria. These results in lower levels of ATP, release of stress factors, and generation of antioxidative reactive oxygen species [2], could trigger cellular dysfunction, DNA damage, and apoptosis [3]. In the process of apoptosis, mitochondria are fragmented (mitoptosis), which is regulated by several proteins including dynamin related protein 1 (Drp1) and ssion 1 (Fis1) [4].
Plasmacytoma variant translocation 1 (Pvt1) is a form of long non-coding RNA (lncRNA) that are classically differentiated from other non-coding microRNAs (miRNAs) based on its size, where lncRNAs are typically greater than 200 bp in length whereas miRNAs are approximately 22 nucleotides in length. Pvt1 has been identi ed in Burkett's lymphoma and has been implicated for decades in its oncogenic functions [4][5][6][7]. Aberrant expression of Pvt1 and its ability to regulate several miRNAs are hallmarks of cancer invasion and progression [8,9]. In the context of cellular degradation, lncRNAs have been extensively characterized for its role in autophagy through interaction with with miRNAs. For example, miR-30a inhibition promoted autophagy protecting neuronal bers from I/R injury [10]. However, the role of lncRNAs in speci cally regulating mitochondrial ssion/ fusion process has not been explored so far.
In a study by Cho et al, microarray analysis in single cell muscle bers identi ed Pvt1 as a regulator of mitochondrial respiration, ssion/fusion, and mito/autophagy [11].
Zinc nger RNA binding protein ZFP36L2 is part of a family of proteins that contain tandem zinc nger domains that can associate with adenine-uridine rich elements present most often in the 3'-untranslated region (UTR) of mRNAs thereby interfering with posttranscriptional modi cations and as a result affecting protein translation [12]. It causes cell cycle arrest, especially during embryonic development and has recently been shown to alter immune functions in T cells [13][14][15]. ZFP36L2 can also positively and negatively regulate adipogenesis and development of B cells. So far, regulatory roles of ZFP36L2 in cardiomyocytes and in the context of I/R injury have not been described.
On the other hand, miRNAs and their role in altering mitochondrial morphology or cardiomyocyte functions have been broadly studied. Mitochondrial miRNAs are more speci cally targeted towards regulating pathways related to cell apoptosis, proliferation and differentiation [16]. More speci cally to processes related to mitochondrial morphogenesis, miR-200a-3p has been shown to promote mitochondrial elongation by targeting mitochondrial ssion factor (Mff) [17]. Using an anti-tumor drug, doxorubicin-induced cardiomyopathy can be counteracted using miR-532-3p, which targets mitochondrial ssion and fusion processes in cardiomyocytes [18]. In the context of I/R injury, several miRNAs including miR-140 and miR-15 have been shown to alter mitochondrial ssion/fusion and apoptosis [19][20][21][22][23].
In this study, we investigated the novel role of miR-21-5p in regulating PVT1 expression, and identi ed an E3 ubiquitin ligase MARCH5 that played a role in PVT1-mediated alterations of mitochondrial ssion and fusion processes during I/R injury.

Materials And Methods
Cardiomyocyte culture and treatment Murine cardiomyocytes were harvested from hearts of 2-day-old mice. Suspension cultures were treated with HEPES solution containing 12 mg/mL pancreatin and 0.14 mg/mL collagenase (Worthington, Freehold, NJ, USA). Cells were cultured in vitro in Dulbecco's Modi ed Eagle's Medium supplemented with 5% heat-inactivated horse serum and antimycotic cocktail (Thermo Fisher Scienti c, Waltham, MA, USA) and plated on laminin-coated (10 μg/mL) culture dishes. Cells were transduced with non-targeting shRNA (NC), or shRNA against ZFP36L2, PVT1, or MARCH5 for 48 h, or transfected with a PVT1 overexpression construct using Lipofectamine 2000 (Invitrogen, Carlsbad, CA, USA) according to the manufacturer's instructions. For the hypoxia/reoxygenation (H/R) based injury model, myocytes were stimulated with palmitate (100 μM) for 2 h followed by induction of hypoxia (1% O 2 ) for 4 h and reoxygenation for 1 h.
Cells were then stained for further mitochondrial analyses or harvested to evaluate mRNA and protein expressions.

Myocardial IR injury model
All in vivo animal experiments were conducted in accordance with the requirements and principles of the Animal Care and Use Committee of Shanghai Jiaotong University A liated Sixth People's Hospital and performed according to established guidelines. Eight-week-old C57BL/6 wild-type mice were anesthetized using iso urane to perform thoracotomy. To assess myocardial I/R injury, mice were subjected to 45 min myocardial ischemia followed by 4 h reperfusion. Sham-operated group underwent the same procedure except that the snare was left untied. After reperfusion, evans blue dye (1 ml of a 2% solution; Sigma-Aldrich) was injected through jugular vein to delineate the ischemic area at risk. The mice were euthanized by cervical dislocation. Then the heart was rapidly excised and sectioned. The heart slices were incubated in 1.0% 2, 3, 5-triphenyltetrazolium chloride (TTC; Sigma-Aldrich) for 15 min at 37°C to differentiate live (red) from dead or infarcted myocardium (white). After washing in ice-cold sterile saline, the slices were xed in 10% formaldehyde, weighed and photographed from both sides. The infarct area (INF) and the risk zone were assessed using computer-assisted planimetry by a histologist blinded to treatment conditions. The INF/LV ratio (%) LVIDd (mm) and the infarct size (de ned as % of risk zone) were calculated.
For lncRNA delivery, sh-PVT1 or sh-NC was administered by intravenous injection at a dose of 30 mg/kg per day for three consecutive days. The mice were then subjected to I/R treatment. sh-MARCH5 (250:1 m.o.i.), or sh-NC (250:1 m.o.i.) were injected with a catheter from the LV apex into the aortic root. The mice were subjected to I/R treatment ve days after injection of adenoviruses.
Quantitative reverse transcription-PCR Total RNA was extracted from heart samples or cells using a RNEase kit (Qiagen, Hilden, Germany) according to the manufacturer's protocol. A total of 500 ng of RNA was used for subsequent cDNA preparation and quantitative PCR using SYBR-Green (Thermo Fisher Scienti c), and the results were normalized to glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The primer sequences used for ampli cation of DRP1, Fis1, Mff, Mfn1, Mfn2, ZFP36L2, PVT1, miR-21-5p, MARCH5, and GAPDH are listed in Table S1.

Histology
Heart sections were analyzed after hematoxylin and eosin (H&E) staining and by immunohistochemistry (IHC). Samples were prepared for histology by infusing 4% paraformaldehyde into the heart. 5-μm microsections were then subsequently stained with H&E or Masson's trichome to assess in ltration and brosis. For IHC, tissue sections were stained using primary antibodies against ZFP36L2 or MARCH5 overnight at 4°C, then imaged using a light microscope.

Subcellular fractionation
Cells were washed with fresh PBS and resuspended in fractionation buffer [20 mmol HEPES (pH 7.5), 10 mmol KCl, 1.5 mmol MgCl 2 , 1 mmol EGTA, 1 mmol EDTA, 1 mmol DTT, and 0.1 mmol phenylmethanesulfonyl uoride], 250 mmol sucrose, and 20mmol protease inhibitor cocktail (Sigma-Aldrich, St. Louis, MO, USA). The suspension was homogenized using a Dounce homogenizer and centrifuged at 750 × g for 5 min. The nuclei were pelleted, leaving the cytosolic and ER fraction in the supernatant. The supernatant was subsequently centrifuged at 10,000 × g for 15 min to pellet the mitochondrial fraction. A third centrifugation step of the supernatant yielded the cytosolic fraction. The fractions were analyzed for proteins using western blotting.
DNA pull-down assay A complementary DNA probe (biotinylated) to CARL was synthesized (Sigma-Aldrich) and incubated with streptavidin-coated magnetic beads in binding buffer. DNA bound beads were then used to pull-down RNA from lysates prepared from cardiomyocytes. RNA bound to the probe was eluted and analyzed using northern blotting.
Luciferase assay PVT1 and MARCH5 wild-type and mutant sequences were expressed using a pGL3 vector (Promega, Madison, WI, USA) encoding the re y luciferase gene. Cardiomyocytes were co-transfected with the luciferase constructs using Lipofectamine 2000 (Invitrogen), and cells were harvested 48 h posttransfection for analysis of the Dual luciferase Reporter Assay kit (Promega) according to the manufacturer's instructions. 30 μL protein samples were analyzed in a luminometer. Fire y luciferase activities were normalized to Renilla luciferase activity.

Northern blot analysis
Samples were subjected to polyacrylamide-urea gel electrophoresis, blotted onto positively charged nylon membranes, and cross-linked using UV irradiation. Membranes were hybridized using 100 pmol 30digoxigenin (DIG)-labeled probes against ZFP36L2, PVT1 or miR-21-5p overnight at 4°C, and detected using a DIG luminescent detection kit (MyLab) according to the manufacturer's instructions.
Immuno uorescence assay Cells were seeded onto poly-L-lysine coated coverslips, xed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100, and blocked using standard 5% bovine serum albumin in PBS blocking solution and incubated with primary antibody against Drp1 (1:1000, Abcam, UK) for 1 h at room temperature. After washing three times using PBS, the coverslips were incubated with Alexa Fluor 488-labeled goat antirabbit secondary antibody (1:1,000, abcam, UK) for 1 h at room temperature. The antibody was removed by washing three times with PBS and the samples were mounted for visualization and imaged using confocal microscopy.

Mitochondrial staining
Cells were seeded onto poly-L-lysine coated coverslips, stained with MitoTracker Red CMXRos (0.02 μM; Molecular Probes, Eugene, OR, USA), and analyzed using a confocal microscope (LSM 510 META; Zeiss, Oberkochen, Germany). Total cells with fragmented mitochondria were represented as a percentage by counting at least 300 cells per treatment group, from six different elds of view.

Electron microscopy
Ultrastructural analysis of mitochondrial ssion was performed using transmission electron microscopy (TEM) as described previously [29]. Tomograms from sections were obtained using a JEM-1230 transmission electron microscope (JEOL, Tokyo, Japan) and analyzed using Image-Pro Plus software (Media Cybernetics, Rockville, MD, USA). More than 1200 mitochondrion were analyzed to determine the mitochondria sizes and structures < 0.6 mm 2 , which were classi ed as ssion mitochondria.

Statistical analysis
All data are expressed as the mean (±SD) of at least three independent experiments, and statistical signi cance was calculated using Student's t-test to compare two groups, and one-way analysis of variance for comparing multiple groups. A value of P < 0.5 was considered statistically signi cant.

ZFP36L2 knockdown inhibited myocardial ischemia/reperfusion (I/R) injury and attenuated mitochondrial ssion in vivo.
Regulatory functions are imposed by RNA binding proteins in response to cellular stress, which alters transcription and translation [30]. In an in vivo mouse model, upon induction of ischemia followed by reperfusion, a 3-fold increase in ZFP36L2 protein level was observed. This increase was attenuated upon treatment with a shRNA targeted against ZFP36L2 (Fig. 1A). Reducing the levels of ZFP36L2 signi cantly lowered infarction size, infarct area/left ventricle (INF/LV), and improved the functioning of the left ventricle as measured by the left ventricular internal diastolic diameter (LVIDd) (Fig. 1B-C). An echocardiogram showed reduced capacity of ejection and fractional shortening upon I/R induction, which was signi cantly improved upon knockdown of ZFP36L2 (Fig. 1D). Analyses of cardiomyocytes observed by H&E and Masson's staining revealed that IR injury caused discontinuous tissue architecture due to extensive apoptosis, but knockdown of ZFP36L2 alleviated this effect (Fig. 1E). Ultrastructural analyses using TEM showed that mitochondrial ssion was signi cantly increased upon IR injury, as indicated by increased expression of Drp1, Fis1, and Mff, and a decrease in Mfn1 and Mfn2 fusion proteins. This effect was dependent on ZFP36L2 ( Fig. 1F-G). Together, these results identi ed ZFP36L2 as a key regulator of cardiomyocyte injury during I/R.
To further extend our ndings in vitro, we used a H/R model with cardiomyocytes treated with 1% O 2 (hypoxia) for 4 h followed by reoxygenation. As previously observed, ZFP36L2 levels were increased upon H/R injury ( Fig. 2A). Mitochondrial ssion, as observed by staining of mitochondria with MitoTracker Red, revealed a 50% increase in ssion upon H/R induction. This effect was signi cantly reduced in the ZFP36L2 knockdown group when compared to the non-targeting control-treated group (Fig. 2B). Similarly, western blotting and immuno uorescence assays showed that proteins mediating ssion or fusion were signi cantly decreased or increased, respectively, upon knockdown of ZFP36L2 ( Fig. 2C-E). Together, these results showed that in vitro regulation of ZFP36L2 levels altered mitochondrial ssion and fusion processes in response to H/R injury.
ZFP36L2 bound to lncRNA PVT1 and regulated its expression.
To identify the mechanism responsible for ZFP36L2-mediated regulation of I/R injury, we performed RNA pull-down and immunoprecipitation experiments. The lncRNA PVT1 was bound to ZFP36L2 and was signi cantly enriched ( Fig. 3A-C). DNA FISH and subcellular fractionation assays showed that a fraction of PVT1 was localized in the cytoplasm, while ZFP36L2 was localized in the nucleus ( Fig. 3D-E). We also observed that knockdown of ZFP36L2 reduced the expression of PVT1 and overexpression increased PVT1 levels (Fig. 3F). However, knockdown and overexpression of PVT1 did not affect ZFP36L2 levels (Fig. 3G). These results indicated that ZFP36L2 bound to PVT1 and directly regulated its expression.
PVT1 knockdown suppressed I/R injury and attenuated mitochondrial ssion.
To verify the functional implication of the association between PVT1 and ZFP36L2, we used an in vivo I/R injury model to determine the effects of PVT1. Upon intraperitoneal injection of shRNA against PVT1 at 24 h post-I/R injury, we observed a reduction in endogenous levels of PVT1 in cardiomyocytes when compared to increased expression upon I/R induction or treatment with control shRNA (NC) (Fig. 4A). This also signi cantly reduced infarction size, INF/LV ratios, and diastolic diameter of the left ventricle (LVIDd) when compared to the NC-treated group ( Fig. 4B-D). Tissue analyses showed improved cell architecture and reduced apoptosis with low levels of PVT1 (Fig. 4E). In addition, TEM and qPCR analyses showed decreased levels of mitochondrial ssion and improved fusion upon knockdown of PVT1 in cardiomyocytes, in both the in vivo and in in vitro H/R injury models ( Fig. 4G and Supplementary  Fig. S1).
PVT1 directly bound to miR-21-5p and suppressed its activity.
PVT1 is known to regulate functions of microRNAs (miRNA) during tumorigenesis [31,32]. We identi ed a region in miR-21-5p having a complementary binding sequence to PVT1, using Starbase 3.0 [33], and also generated a binding-de cient mutant of PVT1 (PVT1-mut) (Fig. 5A). Luciferase assay results showed that treatment with a miR-21-5p mimic signi cantly reduced luciferase activity, indicating an inhibition in wildtype PVT1 (wt-PVT1) expression. This effect was lost in mut-PVT1, and when an inhibitor against miR-21-5p was used (Fig. 5B). A biotin-based pull-down assay with miR-21-5p and a mutant version that did not associate with PVT1 showed an enrichment of PVT1 speci cally in the miR-21-5p wild-type fraction (Fig.   5C). Additionally, we performed DNA pull-down experiments and observed that miR-21-5p was speci cally expressed in the pull-down fraction, using PVT1 (Fig. 5D). Fractionation experiments revealed that miR-21-5p speci cally associated with PVT1 in the cytoplasm and not the nucleus (Fig. 5E-F). To test for the effect of miR-21-5p-mediated regulation of PVT1 expression in mitochondrial ssion upon I/R injury, we treated I/R induced mice with a miR-21-5p mimic alone or in combination with PVT1 overexpression and/or the respective controls. TEM and tissue analyses revealed that treatment with a miR-21-5p mimic signi cantly reduced I/R induced effects on mitochondrial ssion and cell apoptosis, and that this effect was lost when PVT1 was co-expressed ( Fig. 5G-H and supplementary Fig. S2A). In vitro analyses also showed the total number of cells with fragmented mitochondria was below 20% in the control and miR-21-5p mimic-treated cells, and over 40% when H/R was induced and PVT1 was overexpressed ( Fig. 5I and Supplementary Fig. S2B-C). Upon knockdown of PVT1, expression of miR-21-5p was signi cantly higher and overexpression reduced these levels, using both in vivo and in vitro analyses ( Fig. 5J-K). Taken together, these results showed that miR-21-5p and PVT1 were negatively regulated by direct association and altered mitochondrial morphology during I/R injury.
The miR-21-5p associated with and regulated MARCH5 expression.
To further understand how miR-21-5p regulated PVT1 mediated mitochondrial ssion, we looked for known components binding to miR-21-5p using Starbase 3.0, an online omics database for miRNA networks [33], and identi ed MARCH5, an E3 ubiquitin ligase that promotes mitochondrial ssion by Drp1 [34,35]. MARCH5 has a complementary sequence to miR-21-5p in its 3'-UTR region (Fig. 6A). We veri ed binding by treating cells encoding luciferase under the promoter of wild-type MARCH5 (MARCH5-wt) or a mutant version that could not associate with miR-21-5p (MARCH5-mut). Luciferase assays in cardiomyocytes showed reduced expression of MARCH5-wt upon treatment with a miR-21-5p mimic, and this effect was lost in the MARCH5-mut (Fig. 6B). Luciferase expression was increased when an inhibitor against miR-21-5p was used (Fig. 6C). Both in vitro and in vivo, cardiomyocytes responded to treatment with the miR-21-5p mimic by showing increased mRNA levels of miR-21-5p when compared to treatment with the NC mimic, which signi cantly reduced the infarction size 3-fold ( Fig. 6D-E). Treatment with the miR-21-5p mimic also reduced MARCH levels in cardiomyocytes, and treatment with an inhibitor or overexpression of MARCH5 increased protein levels ( Fig. 6F-G). Together, these results identi ed MARCH5 to be negatively regulated by miR-21-5p in cardiomyocytes.
We next tested the in vivo and in vitro importances of MARCH5 expressions in I/R or H/R injury. Intraperitoneal treatment of shRNA MARCH-treated cardiomyocytes 24 h post-I/R injury signi cantly reduced MARCH5 expression compared to shNC-treated mice (Fig. 7A). Infarction size decreased more than 50% in the shMARCH5 treated group compared to the controls (Fig. 7B). A signi cant decrease in LVIDd and infarction areas was also observed upon shMARCH5 treatment (Fig. 7C). Ultrastructural analyses of TEM images revealed reduced mitochondrial ssion, and histology also showed improved tissue architecture and reduced apoptosis in the shMARCH5-treated group (Fig 7D-E). Knockdown of MARCH5 signi cantly reduced Drp1 mRNA expression along with Fis1 and Mff. However, Mfn1 and Mfn2 levels were increased (Fig. 7F). In vitro analyses also showed decreased fragmentation in mitochondria upon treatment with shMARCH5, and reduced Drp1, Fis1, and Mff mRNA and protein expressions along with increased Mfn1, Mfn2 mRNA, and protein levels ( Fig. 7G and Supplementary Fig S3). These results showed that MARCH5 was functionally implicated in mitochondrial dysfunction during I/R injury.
To assess the roles of lncRNA PVT1, miR-21-5p, and MARCH5 in I/R injury, we tested different conditions in vitro using miR-21-5p mimics, PVT1 overexpression, and shRNA-mediated MARCH5 knockdown. Luciferase expression in response to MARCH5 3'-UTR in 293T cells was increased upon PVT1 overexpression, and this effect was reduced when treated with a miR-21-5p mimic (Fig. 8A). In vitro, cardiomyocytes showed decreased MARCH5 expression when PVT1 was targeted by shRNA (sh-PVT1), and these levels increased when treated with a miR-21 inhibitor, whereas overexpression of PVT1 and treatment with miR-21-5p showed an opposite effect ( Fig. 8B-C). Overexpression of PVT1 also increased mitochondrial fragmentation, altered associated genes, mRNA levels, and protein expression levels ( Fig.  8D-G). However, this effect was lost upon co-expression of shMARCH5 because mitochondrial ssion was reduced and Drp1, Fis1, Mff mRNA, and protein levels decreased along with an increase in Mfn1 and Mfn2 (Fig. 8D-G). Taken together, these results showed that MARCH5 played a key role in the interplay between miR-21-5p-mediated regulation of PVT1 expression and its effects on mitochondrial morphology.

Discussion
Mitochondria are enriched in cardiomyocytes and undergo constant ssion and fusion in response to physiological conditions. I/R causes changes in structure and function that determine various cell functions [36,37]. In this study, we showed that lncRNA PVT1 played a critical role in altering mitochondrial ssion and fusion processes. The levels of PVT1 were regulated by the presence of the zinc nger protein, ZFP36L2, which directly associated with PVT1. Furthermore, we also showed that miR-21-5p negatively regulated PVT1 and MARCH5 levels, thereby reducing mitochondrial ssion during I/R injury.
Mitochondrial division during mitoptosis involves pro-apoptotic BAX and BAK proteins along with Drp1, in contrast to the roles of Fis1 and Mdv1 (mitochondrial division 1) proteins during the ssion processes [38]. High levels of Drp1 result in morphologically altered mitochondria and decreased total numbers per cell, along with increased cytochrome c levels and cell death [39]. We observed that Drp1 levels along with Mff and Fis1 were increased upon induction of I/R or H/R. These levels were decreased upon knockdown of ZFP36L2 or PVT1. LncRNAs are classical regulators of gene expression. ZFP36L2 is a nuclear-DNA encoded lncRNA that acts in the mitochondria. Transport of these lncRNAs by RNA binding proteins during I/R injury remains to be elucidated. LncRNA can also form complex three dimensional RNA-RNA hybrid structures to which zinc nger proteins such as PVT1 can bind and regulate their transmodulatory effects [40]. Future studies directed at identifying the binding motif of ZFP36L2, which associates with lncRNA PVT1, may provide the possibility to regulate its effects on mitochondrial ssion/fusion processes.
In other cell types such as skeletal muscles, PVT1 is localized to both nuclear and cytoplasmic fractions [11], whereas in cardiomyocytes it was largely expressed in the cytoplasm. We found that miR-21-5p preferentially associated with PVT1 in the cytoplasm and reduced its expression. This effect was rescued by treatment with a miR-21-5p inhibitor. Although it is known that PVT1 regulates functions of miRNA to promote cell proliferation and invasion in cancer [41,42], it is still not clear how I/R injury, and interference with mitochondrial ssion/fusion processes are directed by PVT1 and miR-21-5p. The miR-21-5p could bind to ssion proteins such as Fis1 or Mff to inhibit its translation as shown previously for miR-484 and miR-761 in cardiomyocytes and speci c to I/R injury respectively [43,44]. Indirectly, it could inhibit Drp1-mediated pathway thereby suppressing p53 mediated triggering of apoptosis.
MiRNAs are capable of interacting with the 3'-UTR regions of mRNAs to alter posttranscriptional modi cations or translation [45], and several miRNAs, such miR-27 and miR-30 have been shown to alter the mitochondrial ssion/fusion process [46,47]. We identi ed MARCH5 as possibly binding to miR-21-5p, and previous studies have shown that it alters Drp1-mediated mitochondrial ssion by redirecting Drp1 from ssion sites [35]. We showed that miR-21-5p bound to the 3'-UTR region of MARCH5, and its knockdown reduced mitochondrial ssion in I/R injury. Knockdown of MARCH5 may prevent ubiquitination and subsequent proteasomal degradation of miR-21-5p, thereby retaining fusion properties of mitochondria [48]. We observed that in the presence of MARCH5 and overexpression of PVT1, mitochondrial ssion was signi cantly induced, and these effects were lost upon MARCH5 knockdown.
While our data suggests that regulation of mitophagy by MARCH5 is via targeting miR-21-5p, MARCH5 is also capable of directly ubiquitinating ssion/ fusion proteins [24,25]. Therefore there is a possibility of additional mechanisms and direct targets for MARCH5 to coordinate mitochondrial morphogenesis during I/R injury in cardiomyocytes.
Hence, we have described a novel mechanism of miR-21-5p/ MARCH5-mediated regulation of mitochondrial morphology by PVT1 during I/R injury. Among increasing data describing a role for miRNAs as a link between I/R injury and mitochondrial dysfunction, our study contributes to a further understanding of these mechanisms, which may eventually translate into improved therapeutic approaches.

Consent for publication
Not applicable. All data presented in this article is non-identi able.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no con icts of interest.  (G) Expression of mitochondrial fusion-related genes Drp1, Fis1, Mff, Mfn1, and Mfn2 after myocardial I/R injury as evaluated by qRT-PCR. Data are represented as the mean ± SD (n = 3, technical replicates). Statistical signi cance was calculated by one-way analysis of variance with multiple comparisons. *P < 0.05; **P < 0.01 and ***P < 0.001. (G) Expression of mitochondrial fusion-related genes Drp1, Fis1, Mff, Mfn1, and Mfn2 after myocardial I/R injury as evaluated by qRT-PCR. Data are represented as the mean ± SD (n = 3, technical replicates).
Statistical signi cance was calculated by one-way analysis of variance with multiple comparisons. *P < 0.05 **P < 0.01 and ***P < 0.001, n.s. = no signi cance. inhibitors and sh-NC/sh-MARCH5 followed by induction of H/R. MARCH5 expression was evaluated by western blots and qRT-PCR. Data are represented as the mean ± SD (n = 3, technical replicates).
Statistical signi cance was calculated by one-way analysis of variance with multiple comparisons. *P < 0.05 **P < 0.01 and ***P < 0.001, n.s. = no signi cance. condition. Data are represented as the mean ± SD (n = 3, technical replicates). Statistical signi cance was calculated by one-way analysis of variance with multiple comparisons. **P < 0.01 and ***P < 0.001.  . Cells with fragmented mitochondria were quanti ed. At least 100 cells were counted per condition. Data are represented as the mean ± SD (n = 3, technical replicates). Statistical signi cance was calculated by one-way analysis of variance with multiple comparisons. **P < 0.01 and ***P < 0.001. This is a list of supplementary les associated with this preprint. Click to download.