mRNA-based in vivo induction of ETV2 to restore blood flow in a preclinical mouse hindlimb ischemia model

Abstract

Regenerative medicine aims to repair tissue defects and functional abnormalities by enhancing endogenous cellular processes. Additional to cell-based therapy, viral and nonviral gene therapies have raised substantial interest as promising approaches that allow the direct modulation of key signaling pathways for tissue regeneration in the body. E26 transformation-specific (ETS) variant transcription factor 2 (ETV2) is a well-validated transcription factor which enables vascular regeneration. The therapeutic efficacy of this factor was demonstrated using lentiviral-based in vivo transduction of ETV2 in a murine model of hindlimb ischemia (HLI). However, given the theoretical safety risks inherent to the clinical use of lentivirus, alternative approaches are considered desirable. Here, we investigated the use of a messenger RNA (mRNA)-based approach to in vivo induction of ETV2. Consistent with previous reports using ETV2 lentivirus, we show that ETV2 translated from synthetic mRNA promotes the upregulation of key endothelial genes in vitro. Furthermore, we demonstrate that the intramuscular injection of ETV2 mRNA encapsulated in lipid nanoparticle (LNP) induces local expression of ETV2 protein in skeletal muscle stromal cells and accelerates blood flow recovery in a murine HLI model. These results provide the first demonstration of ETV2 mRNA with LNP as a potential therapeutic tool for targeting peripheral artery disease.

Graphical abstract

Toyonaga and colleagues investigate an mRNA/LNP-based approach to deliver a pro-angiogenic transcription factor ETV2 for promoting blood flow recovery in a preclinical mouse hindlimb ischemia model.

Introduction

Regenerative medicine aims to repair tissue defects and functional abnormalities in patients by enhancing endogenous cellular processes. In addition to cell-based therapy, viral and nonviral gene therapies are attractive approaches offering the direct modulation of key signaling pathways for tissue regeneration via the expression of therapeutic proteins in the body. Recent advances in mRNA technology have driven the development of non-viral in vivo gene transfer methods, and hold promise for transformative approaches in regenerative medicine.

E26 transformation-specific (ETS) variant transcription factor 2 (ETV2; also known as ER71) is a well-studied transcription factor that is associated with endothelial and hematopoietic development. ETV2 is transiently expressed during embryogenesis and plays a critical role in early vascular development, including endothelial cell differentiation from multipotent stem cells^1,2 and proliferation of endothelial cell progenitors² and endothelial cells³ through increased expression of vascular endothelial genes such as Cdh5 (vascular endothelial [VE]-cadherin), Kdr (vascular endothelial growth factor receptor [VEGFR]-2), and Pecam1 (CD31)^2,4,5 as well as other ETS transcription factors such as Fli1, Erg, Ets1, and Ets2.⁶ Along with these features, growing evidence shows that ETV2 converts somatic cells such as fibroblasts into functional endothelial cells in vitro.^7,8,9 Given these ETV2 modes of action together, ETV2 is an essential transcription factor that can directly upregulate a diverse set of genes necessary for vascular development, and can be expected to be applied to ischemic disease treatment.

Angiogenic effects of in vivo ETV2 transduction have been reported in several previous studies with a lentiviral-based gene therapy approach to animal models of peripheral artery disease (PAD).^3,10 PAD is a type of atherosclerosis of arteries in the legs. The disease condition is associated with the occlusion of arteries by a cholesterol-rich material called plaque, resulting in irreversible damage to the legs due to inadequate blood flow supply. For advanced PAD, endovascular therapy by balloons or stents, or surgical bypass may be considered to improve blood flow within the ischemic legs. More than 10% of patients with PAD are estimated to develop critical limb ischemia (CLI).¹¹ In the most severe cases there may be no treatment option, raising the risk of serious complications such as major leg amputation and death.^12,13 An effective treatment for PAD/CLI therefore remains a high unmet medical need.

However, theoretical safety risks are inherent to the clinical use of lentivirus, namely the generation of novel replication-competent lentiviruses and insertional oncogenesis.¹⁴ The Food and Drug Administration (FDA) released guidance for long-term follow-up that included a tightening of requirements for clinical studies of cell and gene therapies that use viral vectors.¹⁵ This places a further burden on patients. Therefore, alternative methods with clinical compatibility remain necessary. Recent advances in mRNA therapeutics have overcome the instability of the mRNA molecule itself and the potential inflammatory response it induces, expanding its application.¹⁶ Together with the added benefit of the absence of concerns about genomic integration,¹⁶ the mRNA-based approach is expected to become an effective therapeutic modality for the delivery of angiogenesis-related transcription factors to ischemic tissue.

Here, we demonstrate the use of an mRNA-based approach for in vivo expression of ETV2. We show that synthetic mRNA encoding ETV2 translated the functional ETV2 protein and resulted in the upregulation of key endothelial genes (i.e., CDH5, KDR, and PECAM1) in vitro. Our selected lipid nanoparticle (LNP) enabled the highest expression of protein upon intramuscular injection among the materials used and resulted in local expression in stromal cells of skeletal muscle tissue. Furthermore, we demonstrate that ETV2 mRNA encapsulated in this LNP accelerated blood flow recovery in a mouse HLI model. These observations provide the first demonstration of an approach to utilizing ETV2 mRNA as a potential therapeutic tool for targeting PAD.

Results

Synthetic mRNA encoding ETV2 is efficiently translated into functional protein in vitro

ETV2 mRNA transcripts were synthesized with the replacement of uridine to N1-methylpseudouridine (N1mΨ) and reverse-phase high-performance liquid chromatography (RP-HPLC)-based purification to reduce theoretical innate immune response caused by in vitro transcript components, as previously reported.¹⁷ To determine whether the ETV2 mRNA transcript was translated into a functional protein, we transfected the ETV2 mRNA into human adipose-derived mesenchymal stem cells (hADSC) using a transfection reagent, and confirmed the expression of ETV2 protein by western blotting (Figure 1A). Furthermore, we demonstrated significantly upregulated expression levels of endothelial markers, including CDH5 (VE-cadherin), KDR (VEGFR-2), and PECAM1 (CD31), in the ETV2 mRNA-transfected cells, compared with enhanced green fluorescent protein (EGFP) mRNA-transfected cells (Figure 1B). Upregulated protein expression of CDH5 (VE-cadherin), which is a direct downstream target of ETV2, was consistently confirmed by western blotting (Figure 1C).

Figure 1.

Synthetic mRNA encoding ETV2 is efficiently translated into functional protein in vitro

(A) EGFP or ETV2 protein expression at 15 h after mRNA transfection with MessengerMAX. Representative blight filed image and fluorescence image. Scale bars, 75 μm. TF, transfection reagent. (B) After mRNA transfection with MessengerMAX, the cells were collected at 24, 48, or 72 h time point. Total RNA extraction, cDNA synthesis, and qPCR analysis by ΔΔCT method were conducted. For each condition, n = 4; mean ± SD. ∗∗p < 0.01 by two-way ANOVA with Bonferroni correction. (C) CDH5 protein expression at 24, 48, and 72 h time points after mRNA transfection were analyzed by western Blotting. The number above each gel lane represent the fold-change in intensity relative to control (TF only, 24 h).

LNP-1 enabled the highest protein expression in vivo among the compared groups

LNPs represent a clinically validated, non-viral approach that enables the in vivo expression of therapeutic protein by delivering mRNA. To identify a potent LNP for local protein expression on intramuscular injection, firefly luciferase (Fluc) was chosen as a reporter gene for a carrier selection study. We compared three LNPs containing a different ionizable lipid, as well as naked mRNA. Each LNP formulation maintained the same lipid nitrogen-to-phosphate ratio (N:P) and molar composition of lipid components (ionizable lipids, cholesterol, phospholipid, and polyethylene glycol (PEG) lipid). Mice were administrated intramuscularly in the thigh muscle with LNPs encapsulating Fluc mRNA. The thigh muscle was harvested 24 and 72 h later and subject to ex vivo assay (Figure 2A). Those time points were selected because higher levels of local protein expression are expected to be achieved by 24 and 72 h for LNPs¹⁸ and naked mRNA,¹⁹ respectively. All mice survived until the conclusion of the study and the body weight changes of the mice were insignificant during the study (Figure 2B). Luciferase activity was measured by adding luciferase substrate to the muscle tissue lysate and standardized with the protein content measured by the bicinchoninic acid assay (BCA) assay. We found that LNP-1 demonstrated higher luciferase activity per mg protein than other LNPs and naked mRNA at the 24 h time point (Figure 2C). The physicochemical properties of LNP-1 are summarized in Figure S1A. With these results, we selected LNP-1 for further studies.

Figure 2.

In vivo LNP screening for the protein expression levels in muscle

(A) Experimental design of the LNP screening study. LNPs (30 μg/limb) were administered intramuscularly on day 0. Muscle tissues were collected at 24 and 72 h after administration. (B) Body weight changes over the study. (C) Luciferase activity was measured in muscle tissue lysate by adding luciferase substrate and protein content was measured in muscle tissue lysate with BCA assay to normalize.

ETV2 mRNA expresses ETV2 protein at the skeletal muscle stromal cells after intramuscular administration

We next investigated the in vivo expression of ETV2 protein following intramuscular administration of ETV2 mRNA/LNP. First, we collected quadriceps muscles, including the injection site, at 4 h after intramuscular administration and examined their tissue homogenates by western blotting. However, no signal was detected by this method (data not shown). Using in situ hybridization by RNAscope and immunohistochemistry (IHC), we next assessed the serial sections of quadriceps injected with ETV2 mRNA/LNP to visualize the injected ETV2 mRNA and expressed protein, respectively. Figure 3A shown the longitudinal section of the whole quadriceps stained with RNAscope. In the high-magnification images of the area with high ETV2 mRNA density, the ETV2 mRNA was observed to localize mainly at intermuscular connective tissue such as endomysium and perimysium, and not at muscle fibers (Figure 3B). Corresponding with this mRNA distribution, the expressed ETV2 protein was observed in the muscle stromal cells (Figure 3C). It is notable that the expressed protein signals were shown in the limited number of cells despite the much larger abundance of mRNA observed in the same area. This indicates that some mRNAs were present without uptake by cells or were taken up by cells but not translated to protein. These findings were also observed when EGFP mRNA was substituted with ETV2 mRNA for the LNP formulation, indicating that it is a typical cellular distribution pattern when this LNP formulation is used (Figures S2A–S2C). For all of the RNAscope assays, positive and negative control probes were used as technical controls to assess sample and RNA quality (Figures S2D and S2E). For IHC assays, a negative control sample was stained with anti-ETV2 antibodies to confirm there was no non-specific binding or endogenous ETV2 expression (Figures S2F and S2G). Also, non-specific binding derived from secondary antibody was not observed (data not shown).

Figure 3.

In vivo biodistribution of the injected ETV2 mRNA with LNP and protein expression in skeletal muscle

(A) ETV2 mRNA visualization with RNAscope. Quadriceps were collected 5 h after LNP (0.5 μg/limb) administration intramuscularly. Longitudinal section of the whole quadriceps was analyzed. Scare bars, 3 mm (low-magnification). (B and C) Representative images of (B) ETV2 mRNA visualization with RNAscope and (C) ETV2 protein visualization with IHC. Scale bars, 100 μm (high-magnification).

Profiling the cellular functional distribution of intramuscularly administrated mRNA/LNP using an Ai6 ZsGreen reporter mouse model

In skeletal muscle, the connective tissue surrounding muscle fibers is composed of several types of muscle resident cells²⁰—i.e., immune cells, endothelial cells, nerve cells, and fibro-adipogenic progenitors (FAPs)—which represent a multipotent mesenchymal cell population within skeletal muscle. Next, we examined what cell types were functionally delivered with intramuscularly administrated mRNA/LNP. For this, we utilized the transgenic Ai6 mouse model, which contains a LoxP-flanked stop cassette that prevents the transcription of fluorescent protein ZsGreen. Cre-Lox recombination upon Cre mRNA delivery and translation in a cell induces stop cassette excision and the cell then constitutively expresses ZsGreen (Figure 4A). For Cre mRNA functionality, an in vitro cell system was used for validation (Figure S3). Cre mRNA encapsulated in LNP was formulated and administrated intramuscularly. Skeletal muscle harvested at 72 h after administration was dissociated to single cells (Figure 4B) and analyzed by flow cytometry with a gating strategy for leukocytes, endothelial cells, and FAPs (Figure 4C). Compared to the PBS control group (i.e., group without Cre mRNA administration), a significant increase in ZsGreen signals was observed in leukocytes and FAPs cells. On the other hand, the change in ZsGreen signal in endothelial cells was slight and no significant difference was observed (Figures 4D and 4E).

Figure 4.

Profiling the cellular functional distribution of intramuscularly administrated mRNA/LNP

(A) Schematic of the Ai6 transgenic mouse LoxP-flanked stop cassette preventing the transcription of ZsGreen. Upon delivery of Cre recombinase via Cre mRNA, the stop cassette is excised, and the cell expresses ZsGreen. (B) Experimental design of the functional delivery assay in vivo. (C) Gating strategy for each cell population: leukocytes, endothelial cells (ECs), and fibro/adipogenic progenitors (FAPs). (D) Flow cytometry scatter plots of ZsGreen expression in each cell population in PBS or Cre mRNA injected Ai6 mice. (E) ZsGreen expression in each cell population (n = 2 mice). Data are represented as mean ± SD. ∗∗p < 0.01.

ETV2 mRNA/LNP promotes blood flow recovery in a murine HLI model

We next examined the effect of ETV2 mRNA administration in vivo using a murine HLI model generated by femoral artery excision. Athymic nude mice that lacks T cells but retain other immune cells intact (i.e., B cells, DCs, monocytes, macrophages, NKs, etc.), are commonly used to induce HLI and also selected in this study because of their ischemia-susceptible feature compared with immunocompetent mice. Accumulating evidence^21,22 indicates that T cells contribute to vascularization and are potent sources of a number of angiogenic factors, including vascular endothelial growth factor (VEGF). Just after hind-limb surgery to remove a femoral artery, mice were treated with intramuscular administration of ETV2 lentiviral particle or ETV2 mRNA/LNP into the quadriceps and adductor muscles close to the site of HLI surgery (Figure 5A). To monitor blood perfusion levels after the surgery and treatment, laser Doppler perfusion imaging (LDI) for healthy leg and ischemic leg were conducted over 5 weeks. Blood flow ratio was calculated by the following formula: blood flow ratio (%) = blood flow in the ischemic right limb/blood flow in the normal left limb × 100.

Figure 5.

LNP encapsulating ETV2 mRNA promotes blood flow supply in a hindlimb ischemia model using BALB/c athymic mice

(A) Experimental design of HLI study. On day 0 post the HLI surgery, PBS or LNP (3 μg/limb) was administered intramuscularly once a week up to 4 weeks. ETV2 lentivirus was administered intramuscularly on day 0 post HLI surgery. Blood flow was analyzed by LDI by day 35. (B) Representative LDI images on day 0 pre/post HLI surgery and day 14 post HLI surgery. (C) The ratio of blood flow (left injured limb/right normal limb) at all time points for all groups was shown. For each group, n = 6–15; mean ± SE. (D) The ratio of blood flow (left injured limb/right normal limb) at day 14 and day 28 was shown. ∗p < 0.5 and ∗∗p < 0.01 by two-way ANOVA with Bonferroni correction.

Although our mouse HLI model naturally recovered over time, as observed in the placebo (PBS) group, the ETV2 mRNA/LNP group showed superior blood flow supply from day 14 post-HLI surgery (Figures 5B and 5C). Furthermore, ETV2 mRNA/LNP showed significantly higher blood flow supply on days 14 and 21 compared with the control Fluc mRNA encapsulated in the same LNP (Figure 5D). This result confirms that the mRNA-based approach to induce ETV2 expression in vivo produces a beneficial effect in this HLI model. An additional point to note is that the Fluc mRNA/LNP caused a weaker but certain level of blood flow recovery, especially after day 28, with a significant difference (Figures 5C and 5D). We also tested ETV2 lentivirus in the same HLI model. Although a recovery trend was observed in the ETV2 lentivirus group over time (Figure 5C), the level of blood flow supply was lower than that in the ETV2 mRNA/LNP group at the earlier time points of days 14, 21, and 28 (Figure 5D).

Discussion

PAD is a chronic progressive disease that leads to peripheral vascular occlusion by plaque accumulation. Some patients with PAD progress to arterial ulceration, claudication, resting limb ischemia, and limb amputation. One theoretical therapeutic strategy for patients with a severe disease state is to induce the restoration of blood flow through neovascularization of ischemic areas using therapeutic agents, such as gene therapy. A variety of studies has revealed that in vivo ETV2 induction is a promising strategy to promote angiogenesis and arteriogenesis for treating ischemia.^3,10,23,24 This effect is attributable to the generation of endothelial cells through the directed differentiation of stem cells^1,2 or direct conversions of somatic cells,^7,8,9 and the enhancement of endothelial cell function such as vascular integrity.⁷ For instance, ETV2 transduction using lentiviral vectors demonstrated improved blood flow via ETV2-mediated angiogenesis and arteriogenesis in pre-clinical HLI models.^3,23 Further, the feasibility of in vivo ETV2 gene therapy using adeno-associated virus (AAV) was demonstrated in a pre-clinical myocardial infraction model.^23,24

Here, we demonstrate that the intramuscular administration of ETV2 mRNA encapsulated in LNP accelerated blood flow recovery in a mouse HLI model (Figures 5C and 5D). We also showed that the intramuscular administration of the LNP enabled the functional delivery of mRNA to skeletal muscle stromal cells around the injection site, such as the leukocytes and FAPs that represent a multipotent mesenchymal cell population within skeletal muscle (Figure 4E). It is suggested that our ETV2 mRNA delivered by LNP modulates ETV2 signaling pathways at these skeletal muscle-resident stromal cells, including multipotent mesenchymal stem cells in vivo. Our in vitro experiments using ADSC showed that ETV2 mRNA resulted in upregulated expression of endothelial markers, including CDH5 (VE-cadherin), KDR (VEGFR2), and PECAM1 (CD31). Furthermore, several studies reported that lentiviral-based ETV2 transduction in hADSC facilitated differentiation to functional endothelial cells. These results indicate that mRNA-based ETV2 induction can serve as an effective therapeutic approach to PAD/CLI therapy by facilitating vascularization through ETV2-mediated endothelial cell generation around injection sites, as with other viral-based approaches.

Consistent with previous studies,^3,10 our study also showed that intramuscular administration of ETV2 lentiviral particles promoted blood flow recovery. However, the observed recovery was delayed compared to that by ETV2 mRNA encapsulated in LNP in our study. In part, this might be explained by differences in protein expression kinetics or target cell types. Generally, high protein expression is observed in mRNA-based transfection as early as several hours¹⁸ post-treatment, versus a few to several days in lentivirus-based transduction.²⁵ This is because mRNA can be translated to protein shortly after uptake into cytoplasm whereas lentivirus initiates protein expression after genomic integration of its DNAs to begin mRNA transcription and protein translation. In fact, our in vivo experiment confirmed that mRNA administration led to peak translation at 4 h post-administration (data not shown), which is consistent with a previous report.¹⁸ In our HLI model study, mRNA or lentivirus was administered just after HLI surgery, so that the timing of therapeutic protein expression might have led to the difference in efficacy. With regard to target cell type, we showed in the biodistribution assay that ETV2 mRNA encapsulated in LNP was delivered to stromal cells but that delivery to endothelial cells was low (Figure 4E). A consistent observation regarding the local cellular distribution of intramuscularly injected mRNA/LNP was recently reported in a mouse study.²⁶ On the other hand, this finding contrasts with a previous report showing that lentiviral particles cause protein expression in endothelial cells.¹⁰ ETV2 has several potential modes of action as follows: (1) endothelial cell differentiation from multipotent stem cells^1,2; (2) proliferation of endothelial cell progenitors² and endothelial cells³; and (3) direct conversion of somatic cells into endothelial cells.^7,8,9 The inducible mode of action might differ depending on the cell type expressing ETV2, causing a difference in efficacy. Because our LNP shows the favorable delivery to stromal cells such as FAPs and leukocytes, not to endothelial cells (Figures 4D and 4E), our mRNA/LNP-based approach could rely on endothelial cell induction from mesenchymal stem cells and/or somatic cells in part. An in-depth study to investigate the optimal cell types for ETV2 induction could support selection of an ETV2 induction method that further improves efficacy. Practically, investigation utilizing in an inducible in vivo expression system²⁷ of ETV2 in specific cell types and tracking technology of ETV2 expressed cells could assist to understand the impact of each mode of action in HLI model. Furthermore, for instance, if ETV2-mediated endothelial cell differentiation from muscle resident-mesenchymal stem cells such as FAPs is the predominant mode of action in showing higher efficacy, optimizing mRNA delivery to that cell type would be worthwhile. In addition, it is also noteworthy that our mRNA/LNP showed limited numbers of functionally delivered cells in IHC assessment despite the confirmation of abundant mRNA in the RNAscope (Figures 3B and 3C). This low efficiency of mRNA delivery could motivate efforts to identify a more effective delivery strategy.

We also found that the control Fluc mRNA encapsulated in LNP resulted in a certain degree of blood flow recovery on and after day 21 post-HLI surgery compared with the placebo (PBS) group (Figures 5C and 5D). This might indicate that some factor in mRNA and/or LNP affects blood flow via an ETV2-independent mechanism, such as inflammatory response. In fact, mRNA/LNPs or even empty LNPs are known to activate immune signaling and enhance the efficacy of vaccine through their adjuvant-like effect.²⁸ Furthermore, intramuscularly injected mRNA/LNPs or even empty LNPs induces transient local inflammation, which recruits immune cells, including monocytes.^29,30,31 Infiltrating monocytes at ischemic sites contribute to angiogenesis and arteriogenesis.^32,33,34,35 Although nude mice that lack T cells were employed in our HLI model, the remaining immune system could be activated by mRNA and/or LNP. Further investigation using empty LNPs would be beneficial to understand responsible components that lead to blood flow recovery at the later time point. This effort may contribute to determining the appropriate composition of mRNA therapeutics for PAD/CLI treatment.

Several limitations of our study warrant mention. First, we did not assess the differences in ETV2 protein expression levels after single and multiple dosing of ETV2 mRNA/LNP in vivo because quantitative method of ETV2 protein detection was limited. With mRNA COVID-19 vaccines, the anti-PEG antibodies production after multiple dosing was demonstrated, which would promote accelerated clearance of LNP containing PEG lipid and ultimately lead to decreased protein expression.^36,37 Understanding pharmacokinetics of ETV2 mRNA/LNP formulation and induced ETV2 protein would assist proper design of dosing regimen, coupled with further optimization of LNP formulation for multiple dosing. Second, we did not investigate whether the significant blood flow recovery induced by intramuscular administration of ETV2 mRNA/LNP in the in vivo mouse model improves ambulatory ability. In the clinic, measurement of walking ability is usually incorporated into the evaluation of all treatment for PAD patients.³⁸ While patients commonly present with chronic ischemia, current pre-clinical HLI models, including the model that we used in this study, are designed to produce acute ischemia, resulting in natural recovery over a period of approximately 4 weeks.^39,40 This leads a limitation in functional assessment using the current HLI model. Therefore, investigation in clinically relevant disease models that assess functional recovery are required. Finally, we did not conduct vascularization assessment, which would support functional recovery in part. The practical methodology for vascularization assessment is histological assessment by staining the ischemic tissue with CD31 or alpha smooth muscle actin (α-SMA) to investigate capillary density or collateral artery density, respectively. The results of these comprehensive assessments of our ETV2 mRNA-based treatment approach would be of much interest.

In summary, this study provides evidence that in vivo induction of pro-angiogenic transcription factor ETV2 using synthetic mRNA enables blood flow recovery in an HLI model, suggesting that mRNA-based therapeutics may be a new approach for PAD/CLI treatment. Lentiviral-based gene therapy has drawbacks, such as genomic integration risk, which can theoretically disrupt the normal regulation of cell development and proliferation, leading to oncogenesis. These characteristics of lentiviral vectors limit their clinical practice. mRNA therapeutics have the advantages of speed, transience, and the safe expression of therapeutic genes, as well as of time-effectiveness in manufacturing, as proven by COVID-19 mRNA vaccine development.⁴¹ Therefore, an mRNA-based approach likely has great potential for the therapeutic delivery of ETV2.

Materials and methods

mRNA

CleanCap EGFP mRNA, CleanCap Fluc mRNA, CleanCap Cre mRNA, and CleanCap Codon-optimized ETV2 mRNA (N1mΨ, RP-HPLC purified) were synthesized at TriLink Biotechnologies (San Diego, CA). Each mRNA sequence incorporates the 5′ and 3′ untranslated regions and a poly-A tail. Open reading frame of ETV2 mRNA was listed in the Table S1.

In vitro mRNA transfection

For hADSCs (Lonza, Lexington, MA), mRNA was forward transfected at 0.8 μg/20,000 cells/well (12-well plate) or 2 μg/50,000 cells/well (6-well plate) using Lipofectamine MessengerMAX mRNA Transfection Reagent (Thermo Fisher Scientific, Waltham, MA). At 15, 24, 48, or 72 h, the cells were washed with PBS and lysed with RLT RNeasy Plus lysis buffer (a component of RNeasy Mini Kit, QIAGEN, Germantown, MD) containing 2-mercaptoethanol or radioimmunoprecipitation assay (RIPA) buffer containing phosphatase/protease inhibitors. Total RNA was extracted with an RNeasy Min kit and reverse transcribed to cDNA with a Maxima First Strand cDNA Synthesis Kit (Thermo Fisher Scientific), followed by RT-qPCR using TaqMan Probes and TaqMan fast Advanced Master Mix (Applied Biosystems, Waltham, MA) and CFX Opus 384 Real-Time PCR system (Bio-Rad, Hercules, CA) to assess mRNA expression levels of CDH5, KDR, PECAM1, and GAPDH (TaqMan Probes ID nos. Hs00901465_m1, Hs00911700_m1, Hs01065279_m1, and Hs02786624_g1). Relative expression was determined by the ΔΔCt method. Total protein was extracted in RIPA buffer. Two micrograms of the extracted total protein was separated by SDS-PAGE and immunoblotted by antibodies of EGFP (CST, Danvers. MA, cat no. 2956S), ETV2 (Abcam, Waltham, MA, cat no. ab181847), CDH5 (CST, cat no. 2500S), or ACTINB (CST, cat no. 3700) followed by IRDye 800CW or 680CW Infrared Dye conjugated secondary antibodies (LI-COR, Lincoln, NE). Near-infrared fluorescence was imaged in an Odyssey DLx Imager (LI-COR).

LNP formulation

mRNAs were encapsulated in LNP by NanoAssemblr Benchtop (Precision NanoSystems, British Columbia, Canada). Briefly, the ionizable lipids (Table S2), the helper lipid 1,2-distearoyl-sn-glycero-3-phosphorylcholine (DSPC) (NOF America, White Plains, NY, cat no. MC-8080), cholesterol (Nacalai USA, San Diego, CA, cat no. NS460303), and PEG lipid DMG-PEG2000 (NOF America, cat no. GM-020) were dissolved in absolute ethanol with a molar ratio of 50:10:38.5:1.5. mRNA was dissolved in 6.25 mM acetate buffer (pH 5.0) at 0.15 mg/mL. The lipid mixture and mRNA were mixed at an N/P ratio of 5.67 using a NanoAssemblr cartridge at a total flow rate of 12 mL/min and a flow rate ratio of 3:1 (aqueous phase: organic phase). After formulation, the LNPs were subjected to buffer exchange with 1× Dulbecco's Phosphate Buffered Saline (DPBS) (pH 7.4) and concentrated using an Amicon Ultra-15 Centrifugal Filter Unit (100 kDa NMWL, Millipore Sigma, Burlington, MA), followed by filtration through a 0.22 μM filter (PVDF, Millipore Sigma).

LNP characterization

Average size, size distribution, and zeta-potential of the LNPs were assessed by Zetasizer Pro (Malvern Panalytical, Worcestershire, UK). mRNA encapsulation efficiency and concentration were determined with a Quant-it RiboGreen RNA Assay Kit (Thermo Fisher Scientific). Briefly, LNPs were diluted with 1× TE working solution in the absence or presence of 1% Triton X-100. mRNA standard curves were generated within 0.3–5.0 μg/mL with 1× TE working solution in the presence of 1% Triton X-100. Then, the mixture was incubated at 37°C for 20 min followed by addition of the same volume of 1× Quant-it RiboGreen RNA Reagent to each sample. Fluorescence was measured by SpectraMax (Molecular Devices, San Jose, CA) at 485 nm excitation and 525 nm emission. mRNA encapsulation efficiency was calculated by the following formula: encapsulation efficiency (%) = [(fluorescence)total mRNA–(fluorescence)outside mRNA]/(fluorescence)total mRNA × 100.

Animal studies

In vivo studies using wild-type C57BL/6 mice were performed at the CROs Biomodels LLC (Waltham, MA) or Charles River Accelerator and Development Lab (CRADL, Cambridge, MA). The in vivo study using Ai6 (B6.Cg-Gt(ROSA)26Sortm6(CAG-ZsGreen1)Hze/J) mice was performed at the Astellas Institute for Regenerative Medicine. The in vivo HLI study using BALB/c nude mice (BALB/c-CAnN.Cg-Foxn1nu/Crl) was performed at the CRO Pharmaseed Ltd. (Ness Ziona, Israel). All animal study protocols were approved by the Institutional Animal Care and Use Committee of each institute.

Tissue protein expression analysis

Wild-type C57BL/6 male mice (10–12 weeks old) were injected in the right thigh muscle with 100 μL (30 μg) of LNPs encapsulating Fluc mRNA. At 24 or 72 h post administration, the mice were euthanized and the right thigh muscles were collected and snap-frozen. The frozen muscles were pulverized using a mortar and pestle on dry ice and ∼30 mg of pulverized muscle pieces was transferred to Bead Ruptor Pre-Filled Bead Tubes (OMNI International, Kennesaw, GA, cat no. SKU 19–628). For luciferase assay, a Luciferase Assay System (Promega, Madison, WI, cat no. 1500) was used. Briefly, the pulverized muscle pieces in the OMNI tubes were mixed with 1× Luciferase Cell Culture Lysis Reagent (Promega, cat no. E153A) and processed with Bead Ruptor Elite at 4.5 m/s for six cycles of 20 s, with 30-s dwells on ice. Then, the OMNI tubes were centrifuged at 3000 × g for 1 min at room temperature (RT). The supernatant of tissue lysate was transferred to a new tube and mixed with Luciferase Assay Substrate (Promega, cat nos. E151A and E152A) according to the manufacturer’s protocol. Luminescence was measured by EnVision (PerkinElmer, Waltham, MA). To determine protein concentration, a BCA Protein Assay Kit (Thermo Fisher Scientific, cat no. 23227) was used. The tissue lysate was diluted with PBS and mixed with 1× working regent according to the manufacturer’s protocol. Albumin standard (BSA) curves were generated within 25–1,000 μg/mL, followed by the addition of 1× working reagent. The mixture was then incubated at 37°C for 30 min. Absorbance at 562 nm was measured by SpectraMax. Protein concentration was used to normalize luciferase activity.

Tissue RNAscope (ISH) analysis

Wild-type C57BL/6 male mice (10–12 weeks old) were injected intramuscularly with 0.5 μg of EGFP mRNA-LNP formulation or ETV2 mRNA mRNA-LNP formulation to the quadriceps. At 5 h post administration, the mice were euthanized and quadriceps were collected, followed by fixation in 4% paraformaldehyde for 24 h. The quadriceps were paraffine-embedded and sectioned at U-Mass Morphology Core Facility. In situ hybridization (ISH) was performed using an RNAscope 2.5 LS Reagent Kit-BROWN (Advanced Cell Diagnostics [ACD], Hayward, CA, cat no. 322100) for use with BOND RX Stainer (Leica Biosystems, Buffalo Grove, IL) according to the manufacturer’s instructions. Target probes with proprietary sequences were designed by ACD to target EGFP mRNA and ETV2 mRNA. Control probes to the housekeeping gene Mus musculus ubiquitin C (Mm-Ubc) (ACD, cat no. 310778) were used as positive control, and the bacterial gene DapB (ACD, cat no. 312038) as negative control. Images were acquired using PhenoImager HT (Akoya Biosciences, Marlborough, MA).

Tissue IHC analysis

IHC was performed using the BOND RX Stainer. Sections adjacent to those used for RNAscope analysis were baked and deparaffinized, followed by epitope retrieval for 20 min at 95°C using Leica Epitope Retrieval Buffer 2 (Leica Biosystems, cat no. AR9640). The sections were blocked with 2% horse serum on the slides for 20 min at RT. Anti-EGFP antibody (Abcam, Cambridge, UK, cat no. ab183734) and anti-ETV2 antibody (LifeSpan Biosciences, Seattle, WO, cat no. LS-C809808) were used at 1:100 or 1:400 dilution, respectively. Secondary antibody and detection were performed using a Bond Polymer Refine Detection Kit (Leica Biosystems, cat no. DS9800). Images were acquired using PhenoImager HT.

Single cell analysis

Ai6 (B6.Cg-Gt(ROSA)26Sortm6(CAG-ZsGreen1)Hze/J, The Jackson Laboratory, strain no. 007906) mice were injected intramuscularly with 0.5 μg of Cre mRNA-LNP formulation to the quadriceps. At 72 h post administration, the mice were euthanized and quadriceps were collected, followed by the procedure for single cell suspension preparation using a Skeletal Muscle Dissociation Kit (Miltenyi, Santa Barbara, CA, cat no. 130-098-304) according to the manufacturer’s protocol. Cells were pre-incubated with Fc Blocker (BD Bioscience, Franklin Lakes, NJ, cat no. 553142) in flow cytometry buffer (PBS containing 1% BSA and 2mM EDTA) and then stained with anti-mouse antibody cocktails CD45 (BD Bioscience, cat no. 559864, clone 30-F11), CD31 (BD Bioscience, cat no. 561410, clone 390), or Sca1 (BD Bioscience, cat no. 560653). After staining for 20 min on ice, the cells were washed with flow cytometry buffer 3 times and resuspended in flow cytometry buffer containing 1× 7-AAD (BD Bioscience, cat no. 559864). Flow cytometry data collection was performed with MACSQuant X (Miltenyi, Bergisch Gladbach, Germany). Data were analyzed with FlowJo software (Ashland, OR).

Lentiviral particles

Lentiviral particles for ETV2 transduction were synthesized at GeneCopoeia (Rockville, MD, cat no. ULP-H1877-Lv166-EF1a).

Hindlimb ischemia model

BALB/c nude mice (BALB/c-CAnN.Cg-Foxn1nu/Crl) were purchased from Charles River (Wilmington, MA) for development of an HLI mouse model at Pharmaseed. Briefly, under anesthesia and analgesia, the mice were placed supine. On the day of surgery (day 0), one hour before surgery, Meloxicam was administered (2 mg/kg SC) and an incision was made in the skin in the inguinal area of the right hindlimb. The femoral artery just below the iliac artery and below the saphenous artery was ligated twice with 6-0 silk thread and excised between the two ligatures. The wound was closed with 5-0 Vicryl absorbable thread and the mouse was allowed to recover. The mice were treated with Baytril 0.2 mg/mL in drinking water for one week after surgery to prevent post-surgery infection. A high-protein diet was given to enhance recovery. Blood flow in both legs for each mouse was measured with a non-contact Perimed LASER Doppler (Peri Scan PIM II System) before surgery as baseline and after surgery for inclusion criteria (only animals in which blood flow was reduced by at least 30% compared to the uninjured leg were included) and then used for animal group assignment. Treatment with lentiviral particles and mRNA-LNP formulation began shortly after group allocation. Each mouse was injected in the right thigh muscle close to the HLI surgery area, with 1 × 10⁶ TU/head or 3 μg/head of mRNA-LNP formulation. The mRNA-LNP formulation was injected once a week up to day 28. Further blood flow measurements were performed on days 7, 14, 21, 28 and 35 (i.e., 1, 2, 3, 4, and 5 weeks post-operation). Blood flow ratio was calculated by the following formula: blood flow ratio (%) = blood flow in the ischemic right limb/blood flow in the normal left limb × 100.

Statistical analyses

Two-way ANOVA with Bonferroni correction was performed. All statistical analyses were performed using GraphPad Prism 8 software (La Jolla, CA).

Data and code availability

All data from this study are available from the authors upon request.

Acknowledgments

We thank current and former employees at Astellas Pharma Inc., Astellas Innovation Management LLC., and Astellas Institute of Regenerative Medicine for conceptualizing, coordinating, and conducting the studies and Pharmaseed ltd, Biomodels LLC, and CRADL for their in vivo study support.

Author contributions

H.T., K.A., L.C., and H.T. conceptualized the research and designed the experiments. H.T., K.A., and M.Z. coordinated and conducted the experiments. H.T. drafted the manuscript, and L.C. and H.T. reviewed it. This research was funded by Astellas Pharma.

Declaration of interests

The authors declare no competing interests.