MicroRNA perturbation of the MRN complex buffers DNA damage response from VEGF signaling

MicroRNAs contribute to biological robustness by buffering cellular processes from external perturbations. Here we report an unexpected link between DNA damage response and angiogenic signaling that is buffered by two distinct microRNAs. We demonstrate that genotoxic stress-induced miR-494 and miR-99b inhibit the DNA repair machinery by targeting the MRE11a-RAD50-NBN (MRN) complex. Functionally, miR-494 and miR-99b affect telomerase activity, activate p21 and Rb pathways and diminish angiogenic sprouting in vitro and in vivo. Genetic and pharmacological disruption of VEGFR-2 signaling and the MRN complex reveal a surprising co-dependency of these pathways in regulating endothelial senescence and proliferation. miR-99b diminishes VEGF signaling, transcriptional responses and proliferation. Disruption of the MRN complex induces CD44, a known driver of senescence and regulator of VEGF signaling. Our work identifies a putative miR-facilitated mechanism by which endothelial cells can be insulated against VEGF signaling to facilitate the onset of senescence.


Introduction
Accumulation of DNA damage can overwhelm the repair machinery and lead to senescence (Bautista-Nino, Portilla-Fernandez et al. 2016). Endothelial senescence leads to progressive damage and deterioration of cellular structure and function over time (Regina, Panatta et al. 2016) (Gates, Strain et al. 2009). Two major pathways of senescence in endothelial cells are replicative senescence and stress induced senescence. Replicative senescence is one of the hallmarks of aging and is associated to telomere shortening. Stress-induced premature senescence (SIPS) is triggered by external stimuli, including oxidizing agents and radiation, both of which can induce DNA damage and cell cycle arrest.
Recent studies indicate that DNA repair proteins in ECs, such as ATM kinase and histone H2AX, have an inherently pro-angiogenic role (Economopoulou, Langer et al. 2009, Okuno, Nakamura-Ishizu et al. 2012. For example, ATM deficiency kinase decreases tumor angiogenesis while enhancing the anti-angiogenic action of VEGF blockade, suggesting that pathological neoangiogenesis requires ATM. Similarly, global or EC specific deletion of the histone H2AX in mice results in a substantial decrease of pathological angiogenesis in proliferative retinopathy, hindlimb ischemia and tumor angiogenesis. These findings suggest that key regulators of DNA damage repair (DDR) modulate pathological angiogenesis. Recently, we identified a seven-miR signature of DNA damage in ECs induced in common by genotoxic stress, oxidative stress and radiation (Wilson, Espinosa-Diez et al. 2016). We demonstrate that two DNA damage induced miRs from this signature, miR-494 and miR-99b, regulate endothelial cell cycle, senescence, survival and sprouting angiogenesis in vitro and in vivo. Using gene expression arrays, we identified the MRE11a-RAD50-NBN (MRN) complex, a critical DNA damage response checkpoint (Uziel, Lerenthal et al. 2003, Williams and, as a primary target of these miRs.

MRN acts as a sensor of DNA double strand breaks (DSB), initiating Homologous
Recombination (HR) or Non-Homologous-End-Joining (NHEJ) pathway. Once MRN detects DSBs, it tethers the ends, activates and recruits DNA damage response proteins such as ATM ) (Lafrance-Vanasse, Williams et al. 2015).
Interestingly, MRN is also associated with telomere maintenance, playing a role in the formation and disassociation of the t-loops (Dimitrova andde Lange 2009, Porro, Feuerhahn et al. 2014). The MRN complex relationship with aging and cell senescence has been described (Ju, Lee et al. 2006, Gao, Singh et al. 2015, however its role in endothelial cell and vascular pathology is unclear. Similarly, while several miRs have been shown to be involved in EC senescence (Boon and Dimmeler 2015, Regina, Panatta et al. 2016, Santulli 2016, proliferation, viability and migration (Fish and Srivastava 2009) (Shi, Li et al. 2010) (Landskroner-Eiger, Moneke et al. 2013), our work identifies a specific link between DDR driven induction of miR-494, miR-99b, the MRN complex, EC senescence and angiogenic signaling.
Our results indicate that DNA damage induces a miR mediated regulatory process that can inhibit repair, diminish telomerase activity and therefore, enhance cellular senescence.
These phenomena are also exacerbated by the ability of these miRs to interfere with VEGF signaling and transcription, thereby amplifying the anti-angiogenic effect of these miRs. We propose that miR-494 and miR-99b serve to insulate replicative senescence from growth factor driven proliferation by virtue of their effect on the MRN complex and DDR.

DNA damage induced miRs drive EC senescence and inhibit angiogenesis
We previously reported a seven microRNA signature (miR-103, miR-494, miR-99b, miR-21, miR-224, miR-92a and let-7a) specifically upregulated in ECs after radiation, hydrogen peroxide and cisplatin treatment. Among these, we found that miR-99b and miR-494 are both transcribed rapidly in human umbilical vein endothelial cells (HUVECs) in response to γ -radiation with maximal induction occurring at a lower (2 Gy) dose of radiation. ( Figure   1A-B).
To understand the role of these two miRs, we transfected HUVECs with the corresponding miR mimics and observed a significant increase in senescence associated β-galactosidase ( Figure 1C). This phenotype was also observed in human microvascular endothelial cells (HMVECs) (Supplementary Figure 1). Phalloidin staining in HUVECs after 48 hours of miR-494 treatment revealed flattened and multinucleated cells ( Figure 1D), a morphological phenotype associated with stress dependent senescence. Consistent with these findings, we observed an increase in the cell cycle regulator p21 as well as a decrease in Rb hyperphosphorylation ( Figure 1E). Importantly, we also observed a decrease in telomerase activity ( Figure 1F), confirming that the expression of miRs-494 and 99b induced senescence in ECs.
Next we investigated whether this senescence induction in ECs had functional consequences on angiogenesis. Transfection of either miR-99b or miR-494 mimic inhibited sprouting angiogenesis in a 3D-fibrin bead assay ( Figure 1G). To determine if this inhibition of angiogenesis occurred in vivo, we implanted bFGF containing Matrigel plugs in mice and treated the mice with miR-494 mimic in EC targeted 7C1 nanoparticles (Dahlman, Kauffman et al. 2014, Wilson, Espinosa-Diez et al. 2016. We found more than a 50% decrease in CD31 staining in miR-494 treated plugs compared to the control miR mimic treated group ( Figure 1H). These experiments suggest that miRs -494 and -99b not only induce EC senescence but also cause a reduction in angiogenesis.

miR-494 and miR-99b target the MRN DNA repair complex
To analyze the possible targets of these miRs responsible for this phenotype, we ectopically expressed these two miRs in HUVECs and analyzed gene expression using a DNA damage array (Supplementary Figure 2). Interestingly, we identified three common targets: MRE11a, RAD50 and NBN. RNA hybrid modeling suggested putative binding sites for both miRs on all three target 3'UTR regions ( Supplementary Figures 3-4). Both miR-494 and -99b directly bound MRE11A mRNA, and to a lesser extent, RAD50 and NBN mRNAs, as measured by a miRTrap assay (Clontech) (Figure 2A). Consistent with this finding, transfection of miR mimics decreased MRN RNA levels by qRT-PCR ( Figure 2B) and protein levels as measured by a western blot ( Figure 2C). We corroborated the interaction of these miRs with their targets by cloning the 3´-UTRs in luciferase reporter vectors, and co-transfecting them in HEK-293T cells. Either of the miRs transfected alone was able to decrease the luciferase activity downstream of the MRN 3´-UTR reporter ( Figure 2D). Finally, miR-494 transfected HUVECs showed a decrease in MRE11a nuclear foci ( Figure 2E). miRs typically regulate several target mRNA transcripts making their functional attribution challenging (Bartel 2009), (Li, Cassidy et al. 2009, Inui, Martello et al. 2010, Osella, Bosia et al. 2011, Ebert and Sharp 2012. To address this, we used a target protector, a locked nucleic acid (LNA) stabilized oligonucleotide, that binds completely to a specific miRbinding site on a single target mRNA, to rescue it from the negative regulation of the miR.
We confirmed that in the presence of the MRE11A target protector, miR-494 did not decrease MRE11A RNA levels (Supplementary Figure 5A). Functionally, the MRE11A target protector significantly restored the telomerase activity and to a lesser extent the senescence associated β -gal levels in miR-494 transfected cells (Supplementary Figure   5B-C).

Disruption of the MRN complex phenocopies miR-494 and miR-99b effects
We then sought to determine if disruption of the MRN complex using siRNAs ( Supplementary Fig 6A-B) or small molecule inhibitors resulted in a similar phenotype.
Indeed, silencing each of the components of the MRN complex induced senescence and impaired telomerase activity ( Figure 3A-B). Similar to our miR effects, siRNAs targeting MRE11a and NBN inhibited sprouting angiogenesis significantly ( Figure 3C). Furthermore, Mirin-1, a small molecule inhibitor specific to MRE11a (Garner, Pletnev et al. 2009), induced senescence in HUVECs ( Figure 3D). Consistent with the effects in HUVECs, siRNAs targeting the MRN complex and Mirin-1 all induced senescence in HMVECs, Fig 7A-B). Taken together our data indicate that miR -494 and -99b induce senescence, impair telomerase activity and affect angiogenesis, likely by their effects on the MRN pathway.

(Supplementary
To understand the shared molecular pathways that may drive senescence downstream of miR-494, miR-99b and the MRN complex, we utilized a human senescence gene signature array. We found 6 genes with similar changes across both the miR and siRNA treatment groups (Supplementary Fig 8A). Interestingly, CD44, a known negative regulator of VEGF signaling (Ponta, Sherman et al. 2003, Tremmel, Matzke et al. 2009), was the most significantly upregulated gene in the four groups. This increase in gene expression was also reflected by increased cell surface expression of CD44 (Supplementary Figure   8B).

MRN-VEGF codependency
While disruption of DNA damage response (DDR) in ECs can modulate both pathological and developmental angiogenesis via VEGF (Economopoulou, Langer et al. 2009, Okuno, Nakamura-Ishizu et al. 2012, it is not clear if this phenomenon is unique to histone H2AX and/or ATM or if targeting MRN would also influence VEGF signaling. To address this, we silenced VEGF-receptor 2 (VEGFR2) and tested the ability of miR-99b, miR-494 or the specific siRNAs against the MRN complex to induce senescence. Surprisingly, knockdown of VEGFR2 diminished the senescence promoting activities of miR-494, miR-99b, siRNAs targeting the MRN and Mirin-1 (Fig 4A-C). Similarly, combinatorial disruption of both VEGFR2 and MRE11a either with siRNA ( Fig 4D) or with Mirin-1 and Vandetanib, a small molecule inhibitor for VEGFR2 (Fig 4E), led to a decrease in proliferation of HUVECs. We suspected that the disruption of the MRN complex destabilized the activity of VEGFR2, either at the signaling level or at the transcriptional level. Interestingly, we found that miR-99b expression blocked specifically VEGF early gene transcription ( Fig 4F). miR-99b also decreased VEGF-dependent phosphorylation of ERK ( Figure 4G). This dampening effect of VEGF induced ERK activation was also recapitulated by Mirin-1 ( Fig 4H).
These results demonstrate that the DNA repair pathway and the MRN complex have an essential functional role in angiogenesis and we propose that we can exploit this DNA repair-VEGF pathway codependency for future antiangiogenic therapies. Genotoxic stress agents induce DNA damage in ECs; however, they also increase expression of miR-99b and miR-494. These two miRs lead to senescence by targeting the DNA repair complex MRN. Moreover, disruption of MRN decreases proliferation and transcription of early VEGF genes likely by targeting ERK phosphorylation ( Fig 5) and amplifying the induction of senescence and inhibition of angiogenesis.

Discussion
We have postulated that a network of stress-dependent miRs target DNA repair pathways leading to impaired angiogenesis (Wilson, Espinosa-Diez et al. 2016). Among this group of stress-induced miRs, miR-99b and miR-494 lead to senescence, likely through MRN disruption. miR-494 has been shown to be highly upregulated in retinoblastoma (Zhao, Yang et al. 2009), cardiovascular pathologies such as cardiac injury (Wang, Zhang et al. 2010) and in atherosclerotic lesion development (Wezel, Welten et al. 2015). miR-494 has also been reported as an angiogenesis inhibitor, supporting our data here (Asuthkar, Velpula et al. 2014, Welten, Bastiaansen et al. 2014, Chen, Zhao et al. 2015. Recently, Esser et al demonstrated that miR-494 is downregulated in ECs treated with the proangiogenic factor BMP4, in opposition to an increase in the pro-angiogenic miR-126 (Esser, Saretzki et al. 2017). miR-494 has been implicated in tumor senescence and the development of resistance to radiation and chemotherapy (Comegna, Succoio et al. 2014) (Weng, Yu et al. 2016) (Ohdaira, Sekiguchi et al. 2012) (Liu, Li et al. 2015).
In EC-progenitors, miR-99b affects the expression of PECAM-1 and VE-cadherin (Kane, Howard et al. 2012). miR-99b also modulates β-catenin in endothelial junctions (Zhuang, Peng et al. 2016). Interestingly, miR-99b expression has been proposed as a prognostic marker to assess the response to VEGFR2 Tyrosine Kinase Inhibitors (TKI). Lukamowicz-Rajska et al described higher expression of miR-99b-5p in clear cell renal cell carcinoma (CCRC) patients with complete response to TKI treatments (Lukamowicz-Rajska, Mittmann et al. 2016) compared to partial or non-responders. Another example is dermal wound healing, where TK pathways have decreased signaling. Prior studies proposed that miR-99b targets several members of the IGFR1/AKT/mTOR signaling pathway. They demonstrated that miR-99b was downregulated in the early phases of wound healing likely facilitating cell proliferation (Jin, Tymen et al. 2013). We propose here that miR-99b expression is leading to senescence through MRN disruption, but that its effect on the VEGFR2/ERK signaling pathway enhances the anti-angiogenic phenotype. Similarly, our observations of miR-99b diminishing a specific VEGF-dependent early-gene signature (Schweighofer, Testori et al. 2009), also suggests that the induction of miR-99b in complete responders can enhance the effect of VEGFR2 TKIs.
We find that miR-494 and miR-99b target MRE11a, NBN and RAD50 to affect EC senescence and sprouting angiogenesis. EC senescence leads to dysfunction in cardiovascular diseases (Minamino, Miyauchi et al. 2002, Kurz, Decary et al. 2004). Loss of function mutations in the MRN genes cause inherited genetic disorders that are characterized by elevated sensitivity of patients' cells to radiation damage (O'Malley, Li et al. 2003, Dzikiewicz-Krawczyk 2008. Interestingly, there is some evidence that NBN disruption has an anti-angiogenic effect in vivo (Araki, Yamashita et al. 2010). Our observations also imply that the senescence phenotype maybe mediated by the increase in CD44 expression on the cell surface (Mun and Boo 2010). CD44 is a known regulator of endothelial interactions with leukocytes in addition to its other critical roles in EC junction integrity (Flynn, Michaud et al. 2013). Induction of CD44 in response to DNA damage by miR disruption of the MRN complex may increase adhesiveness of the endothelium (Orian-Rousseau, Chen et al. 2002, DeLisser 2009).
Although the role of MRN in DNA damage repair and likely effects on senescence are well studied, emerging studies have implicated deficient telomere maintenance as leading to senescence when MRN function is compromised. For example, decreased MRE11a has been shown to increase T-cell aging in arthritis due to compromised telomere maintenance and heterochromatin unraveling (Li, Shen et al. 2016). Recent structural studies also shed light on how the phosphorylation of NBS influences its interaction with TRF2 and dictates the repair of telomeres (Rai, Hu et al. 2017). RAD50 downregulates the association of TRF1 from telomeres and also contribute to maintain telomere length (Wu, Xiao et al. 2007). We saw a decrease in telomerase activity across miRs and siRNAs experiments targeting the MRN complex. Moreover, the MRE11a target protector largely restored this function, implying this was the miR-494 target responsible for the compromised telomerase activity. Therefore, we suspect that the telomere maintenance function of the MRN complex drives the EC senescence downstream of miRs-494 and miR-99b induced by DNA damage.
Our observations indicate that the function of these miRs in regulating telomerase activity and EC senescence may depend on the expression of VEGFR2. This is surprising, since expression of VEGFR2 may steer cells into a robust replication program rather than a cell cycle exit characteristic of senescence. However, it is possible that replication stress and DNA damage are enhanced in the presence of these miRs. Similarly, our findings demonstrate a decrease in MRN activity either by miR-99b induction or Mirin-1 affects the VEGFR signaling at the level of P-ERK and a subset of the VEGF early response genes.
These data argue that the MRN pathway and the VEGFR2 signaling cross-talk functions as a threshold to determine the level of DNA damage in ECs. In the presence of significant DNA damage the MRN pathway is disrupted and the VEGF signaling is diminished thereby decreasing angiogenesis.
It is possible that during physiological aging, increase in miR-494 and miR-99b levels render the ECs less responsive to VEGF levels and therefore, decrease angiogenic responses. We envision that this senescence associated decrease in VEGF sensitivity functions in a feedforward loop and contributes to cardiovascular aging.

Cell Culture and Reagents
HUVECs (Lonza) were cultured in EGM-2 media (Lonza) supplemented with 10% Fetal Calf Serum (Hyclone). ASMCs (Lonza) were cultured in Medium 231, supplemented with Smooth muscle growth serum (GIBCO). Cells were tested and found negative for mycoplasma contamination before use in the assays described. Mirin-1 and Vandetanib were purchased from Cayman Chemical and Selleckchem respectively. VEGF was purchased from PeproTech, Inc.

miRs/anti-miRs/siRNAs.
miR mimics, inhibitors and respective controls were purchased from Life Technologies and Exiqon. For in vivo studies, high-performance liquid chromatography-purified miRs were purchased from Life Technologies in bulk quantities. siRNAs against MRE11a, RAD50 and NBN were purchased from Life Technologies. VEGFR2 Gapmer and MRE11a target protector were purchased from Exiqon.

Vectors/Plasmids.
MRE11a Luciferase-3-UTR plasmid was purchased from SwitchGear Genomics. RAD50 and NBN luciferase constructs were generated by cloning the entire 3'UTR regions into pmiR-REPORT vector (Ambion). Luciferase assay reagents were purchased from SwitchGear Genomics and Promega.

Transfection
Cell were transfected at 50-60% confluence using standard forward transfection protocols using RNAimax reagent (Life Technologies) for miRs, siRNAs or gapmers and Lipofectamine 2000 for plasmid or plasmid RNA dual transfections. Typically 50 nM RNA and 1-2 μ g plasmid DNA were used for transfections. Target protectors were transfected at a concentration of 50 nM or equivalent to the miR amounts.

Radiation of cells
Cells were irradiated on a Shepherd 137 cesium irradiator at a rate of B166 1.34 cGy min.
β-Gal Senescence assay HUVEC were transfected for 48 hours with the correspondent microRNA or siRNA. After this time cells were washed with cold PBS and then stained for β-galactosidase activity following manufacturer's protocol (Senescence Cells Histochemical Staining Kit, Sigma).

Telomerase activity assay
Cells were transfected with microRNAs or siRNAs for 24 hours. Cells were lysed and processed according to manufacturer´s instructions (Quantitative Telomerase Detection Kit, Allied Biotech). The telomerase activity level in the cell extract was determined through its ability to synthesize telomeric repeats onto an oligonucleotide substrate. The resultant extended product was subsequently amplified by polymerase chain reaction (PCR).

RNA extraction, RT-PCR, miR Profiling
Total RNA and microRNA were isolated using a miRVana microRNA isolation kit (Ambion).
Reverse transcription was performed using TaqMan™ Advanced cDNA Synthesis Kit (Life Tech) according to the manufacturer's instructions. RT-PCR was performed using multiplexed TaqMan primers (Applied Biosystems). The relative quantification of gene expression was determined using the 2 -ΔΔCt method (35). Using this method, we obtained the fold changes in gene expression normalized to an internal control gene, GAPDH or U6 snRNA, respectively. For target analysis a 92 gene DNA damage array (Life tech, 4418773) was used and for senescence phenotype profiling the Human Cellular Senescence array was utilized (SA Biosciences).

miR-TRAP/RISC TRAP assay.
293 T cells were co-transfected with a plasmid coding for a flag-tagged dominant negative GW418 mutant (Clontech kit #632016) along with a control mimic, miR-99b or miR-494 mimic according to kit instructions. Twenty-four hours later the RNA protein complexes were crosslinked and the RISC complex was immunoprecipitated using an anti-FLAG antibody and RNA was isolated for quantitative real-time PCR of target genes. The fold enrichment was calculated using pre and post IP controls as well as normalization to the control mimic pull-downs.

3-D Angiogenic Sprouting Assay
Early passage HUVECs were coated on cytodex-3 beads (GE Healthcare) at a density of 10 million cells/40 μl beads and incubated in suspension for 3-4 hours with gentle mixing every hour. They were plated on TC treated 6 well dishes overnight and resuspended in a 2mg/ml fibrin gel with 200,000 human smooth muscle cells. The gel was allowed to polymerize and complete EGM-2 media was added. Sprouts were visualized from days 3-4 via confocal imaging after overnight incubation with FITC labeled Ulex europaeus lectin (Vector labs). Immunofluorescence imaging was performed on a Yokogawa CSU-W1 spinning disk confocal microscope with 20 0.45 Plan Fluor objective (Nikon).

Immunofluorescence and microscopy
In some experiments, CD31a and MRE11a were visualized using immunofluorescence staining from OCT sections. Slides were fixed with 4% PFA and stained overnight for Fluor objective (Nikon). All images were taken with a channel series. Images were analyzed with Image J software for quantitation.

In vivo assays
All animal work was approved by the OHSU Institutional Animal Use and Care Committee.
Immune-compromised 8-10 week old female nu/nu mice were purchased from Jackson Labs. Growth factor reduced Matrigel (BD) with 400 ng/ml recombinant human bFGF (Millipore) was injected subcutaneously in nu/nu mice. Mice were injected i.v. with 7C1nanoparticles containing miR-494 or control miR (~1mg/kg, i.v) 3 or 4 days after plugs were implanted. At day 7 mouse tissues were harvested and processed to obtained RNA or frozen in OCT for tissue staining.

Statistics.
All statistical analysis was performed using Excel (Microsoft) or Prism (GraphPad). Twotailed Student's t-test was used to calculate statistical significance. Variance was similar between treatment groups.   HUVECs were transfected for a total of 48h. At 24h cells were starved overnight and then treated with VEGF (50ng/ml) for 3h. Graphs depict mean mRNA levels + SEM. *p≤ 0.05.

G.
Representative P-ERK Elisa assay. HUVECs were transfected for a total of 48h. At 24h cells were starved overnight and then treated 10 min with VEGF (50ng/ml). P-ERK levels were assayed using an ELISA quantified via colorimetric detection at 450 nm. Graphs depict mean p-ERK levels + SEM. *p≤ 0.05. H. HUVECs were starved overnight before VEGF treatment and treated with Mirin-1. Then cells were treated 10 min with VEGF (50ng/ml) and p-ERK levels were measured as in G. Graphs depict mean p-ERK levels + SEM. *p≤ 0.05.

Figure 5. Schematic summary -DNA damage repair and VEGF signaling cross talk
Genotoxic stress in ECs induces DNA damage that leads to increase expression of miR-99b and miR-494. These microRNAs target the MRN complex inducing EC senescence. In ECs when VEGF/VEGFR2 signaling pathway is activated ERK1/2 is phosphorylated, leading to increased proliferation. However when MRN complex is decreased by miRs or siRNAs or Mirin-1, ERK pathway is also decreased. On the contrary, when VEGF signaling is inhibited increased DNA damaged does not lead to senescence. Therefore we hypothesize that MRN and VEGF are codependent and that crosstalk between the two pathways dictate endothelial senescence.