RGD peptide-based lipids for targeted mRNA delivery and gene editing applications

mRNA therapeutics are promising platforms for protein replacement therapies and gene editing technologies. When delivered via non-viral gene delivery systems, such as lipid nanoparticles (LNPs), mRNA therapeutics are easy to produce and show low toxicity and immunogenicity. However, LNPs show limited delivery efficiency and tissue specificity in certain applications. To overcome this, we designed RGD peptide (Arg-Gly-Asp) based ionizable lipids, which can be formulated into LNPs for integrin binding on cells and targeted mRNA delivery. RGD-LNPs were formulated using microfluidic devices and screened in vitro for size, mRNA encapsulation efficiency, transfection efficiency, and cell viability. A lead candidate, 1A RGD-based hybrid LNP, showed effective mRNA encapsulation and transfection, and was selected for further testing, including the co-delivery of Cas9 mRNA and sgRNA for gene editing applications. In vitro, 1A RGD-based hybrid LNP outperformed a non-targeted control LNP and showed GFP knockout efficiencies up to 90%. Further, the improved cellular uptake was reversed in the presence of soluble RGD, supporting the hypothesis that this improved uptake is RGD-dependent. In vivo, 1A RGD-based hybrid LNPs showed comparable mRNA delivery to the liver and spleen, when compared to a non-targeted control, and had increased expression in the whole body. Overall, this RGD-based hybrid LNP system is a promising platform for targeted mRNA delivery, which may allow for mRNA-based protein replacement and gene editing in a more efficient and specific manner with reduced off-target effects.


Introduction
In recent years, messenger RNA (mRNA), a transient intermediate between genes and proteins, has emerged as a promising new approach for therapeutic applications including protein replacement therapies, vaccines, and gene editing applications. [1][2][3] However, mRNA rapidly degrades and cannot easily cross the cell membrane due to its large size and negative charge. 4 Therefore, the usage of mRNA therapeutics requires safe, effective, and stable delivery systems to protect from degradation and allow cellular uptake and functionality. 1,[5][6][7][8] In clinical trials, both viral and non-viral vectors are used for systemic delivery of mRNA. 9 Viral vectors have high transfection efficiency and higher specicity for cell targeting, but have several intrinsic limitations including delayed cellular immune response, the possibility of carcinogenesis, and limited opportunity for repeated administration due to acute inammatory response. [10][11][12][13] In CRISPR-based genome editing specically, non-viral systems for CRISPR-Cas9 delivery are preferred due to their improved biosafety proles and the possibility for repeat dosages. 14,15 However, low genome editing efficiency and shorter-lived gene expression using these non-viral systems remains a concern. 16 Lipid nanoparticles (LNPs) represent a broad class of materials that are being investigated to deliver these mRNA therapeutics, including the co-delivery of mRNA Cas9 and single guide RNA (sgRNA). 17,18 LNPs have been shown to provide potent, intracellular nucleic acid delivery across a variety of cell lines with minimal toxicity and immunogenicity, and are also easier to synthesize than viral vectors. [19][20][21][22] LNPs are typically composed of four components: (i) ionizable amino lipids, which acquire charge at low pH, allowing for complexation with nucleic acids and eventual endosomal escape into the cytoplasm; (ii) phospholipids, which fortify the LNP bilayer structure and aid in endosomal escape; (iii) cholesterol, which enhances LNP stability and promotes membrane fusion; and (iv) a lipid-polyethylene glycol (PEG) conjugate, which inserts into the LNP bilayer and provides a PEG coating that reduces LNP aggregation and nonspecic endocytosis by immune cells. [23][24][25][26] These four components are easily modied to inuence the physicochemical properties of LNPs to affect their delivery and uptake into cells, and their variation may encompass 10 10 or more distinct LNP formulations. 27,28 Targeted delivery is of great interest in LNP studies due to its ability to enhance transfection, potentially overcoming current limitations of low genome editing efficiency and allowing for more specicity in delivery to organ and cell types of interest. Through targeting, LNPs can be developed to reach new cells and tissues, reduce toxicity and off target effects, and improve efficiency in difficult to transfect targets. In the literature, amino acids and peptides have been explored as a method to introduce targeting to the ionizable lipid component of LNPs. 29,30 This method has shown successful, selective, and potent delivery of nucleic acid cargo. 31,32 Here, we specically examine Arg-Gly-Asp (RGD), the bronectin tripeptide binding domain, as a potential peptide targeting moiety (Fig. 1). RGD is recognized by a v b 3 and a 5 b 1 integrins that can be overexpressed on several solid tumors, and integrins in general can mediate uptake into cells. 33 In this work, we developed a library of 20 RGD-based ionizable lipids. These lipids, combined with DOPE, cholesterol, and lipid-PEG, were used to formulate 20 unique LNPs which were characterized and screened for luciferase mRNA delivery in vitro. Top performing RGD-lipids were mixed with C12-200 in various ratios to form 7 additional LNP formulations, which were screened similarly. The top LNP formulation, 1A RGDbased hybrid LNP, was found to effectively encapsulate and deliver mRNA in vitro and in vivo, with improved cell viability and whole-body delivery than the untargeted C12-200 control LNP. Further, this formulation was used to co-deliver Cas9 mRNA and sgRNA and was found to knockout green uorescent protein (GFP) expression of HepG2 cells with efficiencies up to 90%. Overall, this RGD-lipid LNP formulation could provide a platform for protein replacement and gene editing applications with improved mRNA uptake.

RGD peptide-based lipid synthesis
Twenty RGD peptide-based ionizable lipids as well as C12-200 were prepared via Michael addition or nucleophilic addition/ SN 2 reactions (Fig. 2). Note that the reaction between amine cores (1-5) with alkyl tails (C and E) in this library is regarded as a Michael addition reaction, while the reactions with epoxide (alkyl tails: A, B, and D) is considered a nucleophilic addition/ SN 2 reaction. Briey, RGD peptides (GenScript, Piscataway, NJ) with reactive amino groups were dissolved in ethanol and combined with excess lipid tails (Sigma-Aldrich, St. Louis, MO) in a 4 mL glass vial under gentle stirring with a magnetic stir bar for 3 days at 70 C. The reaction mixture was dried using a Rotovap R-300 (Buchi, New Castle, DE) and used for LNP formulation.

mRNA synthesis
The rey luciferase gene sequence was codon optimized, synthesized, and cloned into a proprietary mRNA production plasmid. Transcription was carried out using MegaScript T7 RNA polymerase (Invitrogen, Waltham, MA), and mRNA was precipitated using lithium chloride, puried by cellulose chromatography, and stored frozen at À80 C for future use.

LNP formulation
Ionizable lipid (RGD peptide-based lipid or C12-200) was combined in an ethanol phase with cholesterol (Sigma-Aldrich), 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (DOPE, Avanti, Alabaster, AL), and 1,2-dimyristoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000] (ammonium salt) (C14-PEG2000, Avanti) to a total volume of 100 mL. 25 mg of luciferase mRNA was prepared in 10 mM citrate buffer (pH ¼ 3) to a total volume of 300 mL. A 10 : 1 weight ratio of ionizable lipid to luciferase mRNA was used to ensure encapsulation via electrostatic interactions. A syringe pump was used to combine the ethanol and citrate phases in a microuidic device designed with herringbone features to formulate mRNA-containing LNPs via chaotic mixing. 34 Then, LNPs were dialyzed against 1Â PBS in cassettes with a molecular weight cutoff of 20 kDa for 2 h, ltered using a 0.22 mm lter, and stored at 4 C for future use. All materials were prepared and handled ribonuclease-free throughout synthesis, formulation, and characterization steps. For gene editing experiments, Cas9 mRNA (TriLink BioTechnologies, San Diego, CA) and single-guide RNA (sgRNA) targeting EGFP (Axolabs, Kulmbach, Germany) were combined at ratios of 4 : 1 or 2 : 1 in 10 mM citrate buffer for LNP formulation.

Dynamic light scattering
Dynamic light scattering (DLS) was used to evaluate LNP diameter and polydispersity using a ZetaSizer Nano Series (Nano ZS, Malvern Instruments, UK) at a scattering angle of 173 . 6 mL of each LNP solution was diluted 100Â in 1Â PBS in 4 mL disposable cuvettes (Malvern Panalytical, Malvern, UK). Three measurements each with at least 10 runs were recorded for each sample at room temperature.

LNP pK a measurements
Surface ionization measurements to calculate the pK a of selected LNP formulations were performed as previously described. 35 Briey, buffered solutions containing 150 mM sodium chloride, 20 mM sodium phosphate, 20 mM ammonium acetate, and 25 mM ammonium citrate were prepared with pH measurements of 2 to 12 in increments of 0.5.5 mL of LNP and 125 mL of each pH-adjusted solution were added in black 96-well plates in triplicate. 2-(p-Toluidinyl)naphthalene-6sulfonic acid (TNS) was then added to each well to a nal TNS concentration of 6 mM. An Innite 200 Pro plate reader (Tecan, Morrisville, NC) at an excitation wavelength of 322 nm and an emission wavelength of 431 nm was used to test the uorescence intensity. Using least squares regression, the pK a was taken as the pH corresponding to half-maximum uorescence intensity, i.e., 50% protonation.

LNP encapsulation efficiency
mRNA encapsulation efficiency of each LNP formulation was calculated using the Quant-iTRiboGreen (Thermo Fisher Scientic, Waltham, MA) assay. 36 LNP samples were diluted to approximately 2 ng mL À1 in microcentrifuge tubes containing either 1Â TE buffer or 0.1% (v/v) Triton X-100 (Sigma-Aldrich). LNPs in TE buffer or Triton-X as well as mRNA standards were plated in triplicate in black 96-well plates following which uorescent RiboGreen reagent was then added in each well. Fluorescence intensity was read on an Innite 200 Pro plate reader (Tecan) at an excitation wavelength of 490 nm and an emission wavelength of 530 nm. Least squares linear regression (LSLR) was used to estimate a standard curve to quantify the RNA content. Encapsulation efficiency was calculated as the comparison of RNA content in TE buffer and in TX buffer.
In vitro LNP-mediated luciferase mRNA delivery and cell viability in HepG2 cells LNP-mediated in vitro transfection was evaluated in HepG2 cells seeded in 96-well plates at a density of 5000 cells per well in 100 mL of Dulbecco's Modied Eagle Medium (DMEM) media. Cells were seeded and cultured for 24 h at 37 C before treatment in triplicate with a volume of LNPs corresponding to 10 ng of luciferase mRNA, including wells treated with C12-200 LNPs as a positive control. For the RGD binding studies, HepG2 cells were treated with excess RGD peptide (0.1 mg mL À1 ) for 1 h prior to the addition of RGD-lipid LNPs. 37 24 h following treatment with luciferase mRNA-containing LNPs, the medium was aspirated, cells were lysed with 20 mL per well of 1Â lysis buffer (Promega, Madison, WI), and luciferase assay substrate (Promega) was added to each well. The luminescence intensity corresponding to luciferase protein expression was measured using a Innite 200 Pro plate reader. Similarly, for cell viability experiments, 24 h following treatment with LNPs, cells were treated with CellTiter-Glo (Promega) and luminescence intensity corresponding to ATP production was measured using a Innity 200 Pro plate reader. Luminescence measurements were normalized to untreated cells and a one-way ANOVA with the Dunnett correction for multiple comparisons was used to compare means between RGD LNP groups and C12-200 LNP.

Luciferase imaging and quantication
To assess LNP biodistribution and in vivo transfection, eightweek old C57BL/6 mice were treated with either 0.1 mg kg À1 luciferase mRNA-containing RGD-lipid or C12-200 LNPs (n ¼ 3) via tail vein injection. 6 h later, the mice received D-luciferin and potassium salt (Biotium, Fremont, CA) at 150 mg kg À1 via intraperitoneal injection. 10 min later, mice were anesthetized using isourane and imaging was performed using an in vivo imaging system (IVIS, PerkinElmer, Waltham, MA) to evaluate whole-body luciferase expression. Next, mice were euthanized using CO 2 , organs were immediately collected, and bioluminescence imaging was performed using IVIS with 60 s exposure times.
Image analysis was conducted using the Living Image soware (PerkinElmer). To quantify luminescence ux, a rectangular region of interest (ROI) was placed around the whole body or organ as well as in an area without any luminescence signal in the same image. Normalized ux was calculated by dividing the total ux from the whole body or organ ROI by the total ux from the background ROI. An unpaired, two-tailed t test was used to compare total ux between RGD and C12-200 only LNPs.
In vitro gene editing in GFP + -HepG2 cells GFP-expressing HepG2 cells were seeded in 6-well plates at a density of 100 000 cells per well and 24 h later, sgRNA/Cas9 mRNA-containing LNPs were used to treat cells. Aer 7 days, media was exchanged for fresh cell culture medium and GFP expression was evaluated using ow cytometry. GFP expression was normalized to untreated cells to determine gene editing efficiency and lipofectamine MessengerMAX (Thermo Fisher Scientic) was used as a positive control for the study. A one-way ANOVA with the Dunnett correction for multiple comparisons was used to compare means between the treatment groups and the GFP positive control.

In vivo cytotoxicity
For cytotoxic evaluation of Luc-mRNA containing LNPs, plasma samples collected from mice 12 h aer LNP or PBS injection were used to assess AST and ALT liver enzyme levels via colorimetric assay kits (Cayman Chemical, Ann Arbor, MI, USA). Mice AST/ALT data were normalized to the protein concentration in the sample as determined by the microBCA assay (Thermo Fisher, Waltham, MA).

Animal studies
C57BL/6J mice were purchased from Jackson Laboratory. All animal procedures were performed in accordance with the Guidelines for Care and Use of Laboratory Animals of University of Pennsylvania and approved by the Animal Ethics Committee of University of Pennsylvania (protocol number 806540).

Design, characterization, and in vitro evaluation of LNP library
In this report, we designed RGD peptide-based ionizable lipids for their potential to improve potency and targeting for LNPmediated intracellular mRNA delivery (Fig. 1). A library of 20 RGD peptide-based ionizable lipids was synthesized by conjugating RGD peptide heads 1-4 to alkyl lipid tails (A, B, C, D, and E) (Fig. 2). 38 These RGD peptide-based ionizable lipids were then used to formulate LNPs and combined in an ethanol phase with three other excipients: (i) cholesterol, to enhance LNP stability and promote membrane fusion, (ii) DOPE, a helper lipid to fortify the bilayer structure of the LNP and promote endosomal escape, and (iii) lipid-anchored polyethylene glycol (C14-PEG2000), to reduce aggregation and nonspecic endocytosis (Fig. 1A). 39,40 This ethanol phase was then combined with an aqueous phase containing mRNA using a microuidic device to induce chaotic mixing and LNP formulation. The resulting LNP library was characterized using dynamic light scattering (DLS); the diameter of LNPs ranged from 50 to 200 nm and all LNPs had a polydispersity index (PDI) less than 0.3 (Table S1 †). However, the mRNA encapsulation efficiency of these RGD peptide-based LNPs was lower than typically seen, with less than 50% encapsulation for all LNPs in the library.
These LNPs were formulated with luciferase mRNA as a model cargo to evaluate in vitro mRNA delivery via bioluminescence. Upon intracellular uptake and endosomal escape, luciferase mRNA is translated into luciferase protein which reacts with luciferin reagent to produce bioluminescence as an indicator of functional mRNA delivery. 38 HepG2 cells were selected to evaluate in vitro mRNA delivery as they are a commonly used immortalized cell line for drug metabolism and hepatotoxicity studies. Seven LNPs from the 20 LNP library -1A, 1B, 1C, 1D, 1E, 2A, and 2Bwere used to treat HepG2 cells and luciferase expression was evaluated 24 h later. The results demonstrate little to no transfection for the seven RGD peptidebased LNPs compared to C12-200 as a positive control, when the RGD peptide-based lipids are the only ionizable lipid in the LNP formulation (Fig. 3a).
Following these results, we formulated a small library of LNPs to evaluate the effect of partially substituting the C12-200 ionizable lipid with our 2A RGD peptide-based lipid at various RGD-lipid to C12-200 ratios (10 : 90, 20 : 80, 30 : 70, 40 : 60, and 50 : 50). Upon characterization of these LNPs, we found improved encapsulation efficiency for all formulations compared to their counterparts without C12-200 (Table S2 †). Additionally, when screened in vitro, the 20 : 80 ratio of 2A RGDlipid to C12-200 had more comparable delivery to C12-200 only LNPs than its counterpart LNP without C12-200 (Fig. 3b). As a result, we proceeded with a 20 : 80 ratio of RGD-lipid to C12-200 for further exploration due to its highest encapsulation efficiency and most comparable in vitro delivery to C12-200 only LNPs.
Therefore, we next formulated seven RGD peptide-based LNPs with a 20 : 80 ratio of RGD-lipid to C12-200. The characterization results demonstrate similar diameter and PDI measurements to C12-200 only LNPs with encapsulation efficiencies greater than 75% (Table S3 †). For example, the diameter of LNPs formulated with RGD peptide-based lipid 1A had a diameter of 106.7 nm (Fig. S1 †), a pK a value of 6.41 (Fig. S2 †), and an mRNA encapsulation efficiency of 89.49%, all of which are similar to our positive control C12-200 LNPs. These results conrm the successful formulation of LNPs with RGD peptidebased lipids with desirable physicochemical properties and efficient mRNA encapsulation for further investigation in different biological applications.
These seven RGD peptide-based LNPs -1A, 1B, 1C, 1D, 1E, 2A, and 2Bformulated with a 20 : 80 ratio of RGD-lipid to C12-200, all showed some level of mRNA delivery and resulting luminescent signal. LNPs formulated with a 20% substitution of RGD peptide-based lipid 1A demonstrated signicantly higher (****p < 0.0001) luminescence signal than the C12-200 only control (Fig. 3c). Ligand-receptor interactions play an important role in many biological processes. RGD containing peptides are known to mimic the binding domain of the extracellular matrix protein bronectin and selectively bind to a subset of integrin receptors. Integrins are a class of transmembrane receptors expressed in various cell types that are involved in tumor progression. 33 Here, we used HepG2 cancer cells to evaluate cell uptake. According to confocal images ( Fig. S3 †), the top performing RGD-based hybrid LNP 1A exhibited greater uptake by HepG2 cells compared to C12-200 LNPs, supporting the delivery results shown here. In terms of cell viability, six of the seven RGD LNPs had signicantly improved (**p < 0.01, ***p < 0.001, ****p < 0.0001) cell viability compared to the C12-200 only control (Fig. 3e). Taken together, these results prompted us to select the C12-RGD (1A) LNP as our lead formulation due to its increased luciferase mRNA delivery and decreased toxicity compared to the C12-200 control.
Finally, to demonstrate whether the improved delivery efficacy with this RGD-LNP formulation is RGD-dependent, HepG2 cells were pre-treated with excess soluble RGD 1 h before the addition of 1A RGD-based hybrid LNPs. RGD will bind with the integrin receptor, which may block RGD-based lipid binding with integrins and the resulting RGD-LNP cell uptake. Again, the luciferase expression for our RGD LNPs without pretreatment was signicantly higher (***p < 0.001) than the C12-200 only control. Additionally, luciferase expression was reduced in the group pre-treated with the RGD peptide (Fig. 3d). There were no changes in cell viability upon pre-treatment with soluble RGD peptide (Fig. 3f). These results suggests integrin- RGD binding plays a key role in the intracellular uptake and luciferase expression of RGD-LNP. 37

In vivo evaluation of RGD peptide-based LNPs
To further assess the transfection efficiency of our novel RGD peptide-based LNPs in vivo, we evaluated LNP-mediated luciferase mRNA delivery for our lead 1A RGD lipid LNP compared to C12-200 only LNPs in mice. LNPs were administered to C57BL/6 mice via tail vein injection at a dose of 0.1 mg kg À1 of luciferase mRNA. Six hours aer injection, D-luciferin was injected intraperitoneally and whole-body imaging was subsequently performed using an in vivo imaging system (IVIS) to quantify luciferase expression (Fig. 4a). Using ROI quantication of the whole-body images, there was signicantly higher (**p < 0.01) luciferase expression between our RGD lipid LNPs and the C12-200 control (Fig. 4c). Next, the mice were euthanized and organs including the heart, liver, spleen, lung, and kidneys were isolated as evaluated for luciferase expression via IVIS (Fig. 4b).
The images demonstrate LNP-mediated mRNA delivery to the liver and spleen in both groups. However, there was no significant difference in luciferase expression between the RGD lipid LNPs and C12-200 LNPs for either the liver or spleen (Fig. 4c).
Due to the high luciferase expression in the liver, we sought to explore LNP-mediated hepatotoxicity by quantifying the liver enzymes alanine aminotransferase (ALT) and aspartate aminotransferase (AST) in mouse plasma 12 h following LNP administration. There was no signicant difference in ALT or AST levels for either the RGD lipid LNP or C12-200 LNP groups compared to the PBS control (Fig. S4 †). This indicates that LNPs formulated with RGD peptide-based lipids enable efficient mRNA delivery to the liver without inducing enzyme releaseassociated toxicity. Overall, our RGD-lipid LNPs have low toxicity and higher cell viability compared to C12-200 LNPs.

In vitro gene editing in GFP expressing HepG2 cells
To evaluate an additional biological application of our lead RGD peptide-based lipid, LNPs were formulated with either 20% of 1A RGD lipid or C12-200 alone to deliver Cas9 mRNA and sgRNA for in vitro gene editing in GFP expressing HepG2 cells. 24 h aer seeding, cells were treated with LNPs at three different doses (0.4, 0.8, and 1.6 mg mL À1 ) and two different Cas9 mRNA to sgRNA ratios (4 : 1 and 2 : 1). Lipofectamine MessengerMAX was used as an industry standard transfection reagent to serve as a positive control. In typical mRNA delivery applications, LNPs only need to release mRNA into cytoplasm for its translation into functional proteins. However, since the functional protein Cas9, and the resulting Cas9/sgRNA complex targets DNA in the nucleus, gene editing applications require not only cytoplasm delivery, but trafficking to the nucleus. Further, changes to the genome do not immediately result in changes to the protein composition within the cell. In this work, editing cannot be measured using ow cytometry until a cycle of protein degradation and new protein translation has occurred. Therefore, we did not detect any obvious editing proles at day  one and day three. Seven days post-treatment, in vitro gene editing was assessed using ow cytometry (Fig. S5 †). The results demonstrate that GFP knockout is dependent on total RNA dose (Fig. 5a) and that treatment with RGD-lipid LNPs reduces GFP mean uorescence intensity (MFI) beyond the C12-200 and lipofectamine groups (Fig. 5b). Of note, RGD-lipid LNP at a dose of 0.8 mg mL À1 and a 4 : 1 ratio of Cas9 mRNA to sgRNA induced up to 90% knockout of GFP expression compared to the untreated control (Fig. 5c). Together, these results suggest that our RGD peptide-based lipid may expand the therapeutic potential of LNPs for mRNA therapeutics and CRISPR/Cas9 gene editing.

Conclusions
The RGD peptide-based lipids introduced in this work are a potential novel approach in designing LNPs for targeted nucleic acid delivery. This multicomponent delivery system shows efficient mRNA encapsulation, signicant transfection efficiency of nucleic acids, and an excellent safety prole. The introduction of this RGD-peptide lipid into a C12-200 LNP formulation demonstrated improved cellular uptake and lower toxicity when compared to LNPs formulated with C12-200 alone. Additionally, RGD lipid LNPs were selected to co-deliver Cas9 mRNA and sgRNA for in vitro gene editing, successfully knocking out green uorescent protein (GFP) expression in up to 90% of HepG2 cells. This RGD-LNP system shows signicant potential both in vitro and in vivo, not only improving RNA delivery efficacy, but also potentially opening a new avenue for targeted delivery of other nucleic acids into specic cells and tissues of interest. This technology may be applicable for potential in vivo gene editing applications or for mRNA therapeutic delivery in tumor-bearing mouse models.

Author contributions
J. Q. and L. X. were responsible for conceptualization of this work and initial manuscript writing. J. Q., L. X., N. G., and H. Z. were responsible for data collection and initial analysis. J. Q, S. J. S., and K. L. S. were responsible for data analysis and the production of graphs. S. J. S., R. M. H., and K. L. S. were responsible for manuscript writing, review, and editing.

Conflicts of interest
There are no conicts to declare.