In vivo gene delivery to immune cells

Abstract

Immune cell therapies are an emerging class of living drugs that rely on the delivery of therapeutic transgenes to enhance, modulate, or restore cell function such as those that encode for tumor-targeting receptors or replacement proteins. However, many cellular immunotherapies are autologous treatments that are limited by high manufacturing costs, typical vein-to-vein time of 3 to 4 weeks, and severe immunerelated adverse effects. To address these issues, different classes of gene delivery vehicles are being developed to target specific immune cell subsets in vivo to address the limitations of ex vivo manufacturing, modulate therapeutic responses in situ, and reduce on- and off-target toxicity. The success of in vivo gene delivery to immune cells – which is being tested at the preclinical and clinical stages of development for the treatment of cancer, infectious diseases, and autoimmunity – is paramount for the democratization of cellular immunotherapies.

Graphical Abstract

Introduction

Recent advances in genetic engineering are now making it possible to precisely modify the genome or to modulate transcriptional activity of target cells and tissues through the delivery of synthetic nucleic acids. Recent drug products that have been approved by regulatory agencies include the use of lentiviral vectors (LVV) to transfer a beta-globin gene to autologous hematopoietic stem cells to treat patients with beta-thalassemia (Bluebird Bio, Zynteglo) [1] and adeno-associated virus (AAV) for in vivo subretinal delivery of choroideremia-encoding DNA to treat retinal dystrophy (Spark Therapeutics, Luxturna) [2]. In the wake of these clinical successes, there is increasing interest in gene delivery to immune cells – such as macrophages, dendritic cells, T cells and B cells – due to their important roles in complex human diseases including cancer [3], viral infections [4] and autoimmune diseases [5]. For instance, using LVVs or retroviral vectors (RVV) to genetically modify autologous T cells with tumor-reactive chimeric antigen receptors (so called CAR T cells) has shown unprecedented therapeutic efficacy against hematological malignancies, as exemplified by the current FDA-approved products on the market [6]. Delivering genes to immune cells can be performed either ex vivo using cells harvested from a patient prior to re-infusion of the genetically modified cells back to the patient, or directly in the body (i.e., in vivo) [7]. In this review, we focus on in vivo gene delivery to immune cells, as it has the potential to bypass the complex logistics and high costs associated with an ex vivo manufacturing pipeline and to increase the functionality of the cell product [8,9]. However, in vivo gene delivery is restricted by the inability to traffic gene delivery vehicles to disease sites or lymphoid organs and the high off-target uptake of the vehicle by non-immune cells such as hepatocytes. To address these issues, we discuss recent advances in lipid nanoparticles (LNPs), polymers, and viruses for gene delivery to immune cells. Although engineering nucleic acids to improve the specificity of gene translation in target cells, in vivo stability, and to reduce immunogenicity is also a major area of investigation, we refer readers to other references on the topic [10–12]. Moreover, we describe therapeutic applications of in vivo engineering of cellular immunity and conclude with considerations for clinical translation and a future outlook.

Lipid nanoparticles

LNPs have emerged as a promising delivery method for genetic cargo as exemplified by the recent clinical and commercial success of the coronavirus disease 2019 (COVID-19) vaccines, mRNA-1273 [13] and BNT162b [14]. LNPs are comprised of ionizable lipids, phospholipids, cholesterol, and polyethylene glycol (PEG)-functionalized lipids which facilitate the encapsulation of nucleic acids, cellular endocytosis, and the subsequent endosomal escape of the genetic material [15]. Delivery of LNPs to specific cell types and organs remains challenging due to the adsorption of serum apolipoprotein E (ApoE) on the surface of LNPs, which facilitates hepatocellular uptake through low-density lipoprotein (LDL) receptors [16]. However, the relative composition and ratio of LNP lipids can affect delivery specificity, as studies showed that by tuning the ratio of common LNP lipids, new formulations of LNPs can deliver encapsulated plasmid DNA to the spleen and predominately transfect antigen-presenting cells (APCs), such as macrophages, B-cells, and dendritic cells [17]. The development of novel ionizable lipids, such as 113-O12B [18], CL4HP [19], and piperazine-containing ionizable lipids [20] also enables specific uptake of nucleic acids by APCs. Another strategy to bias the uptake of LNPs by lymphoid organs is through incorporation of negatively charged lipids in the LNP composition. For example, incorporating the negatively charged lipid 1,2-dioleoyl-sn-glycero-3-phosphate (18PA) altered the tissue tropism of LNPs and allowed for selective delivery to the spleen and subsequent uptake by immune-cells, such as B cells, T cells, and macrophages [21]. These strategies demonstrate that chemically modifying the structure of lipids and composition of LNPs can effectively traffic the LNPs to different lymphoid niches.

The selective transfection of subsets of immune cells with high efficiency and specificity remains a challenge. To address this issue, targeting moieties, such as monoclonal antibodies, are incorporated onto the surface of LNPs to help direct the LNP to specific immune cells [22]. For example, LNPs engineered with CD5 targeting antibodies enabled in vivo T cell transfection of mRNA-encoding CAR against a fibroblast activating protein (FAP), which is expressed by activated fibroblasts that contribute to fibrosis [23]. These CAR T cells were able to kill activated fibroblasts and reduced fibrosis in murine models of cardiac fibrosis, extending the utility of CAR T therapy to treat conditions other than cancer. Other pathologies are more specific in the subsets of immune cells that are responsible for disease progression, such as CD4+ T cells which are infected with human immunodeficiency virus (HIV). To deliver nucleic acids to CD4+ T cells, LNPs can be engineered with CD4 targeting antibodies to facilitate cellular uptake only by this subset of T cells. This strategy was leveraged to deliver Cre recombinase-encoding mRNA cargo to T cell enriched tissues, such as the spleen and lymph nodes and provided dose-dependent loxP-mediated genetic recombination in CD4+ T cells [24]. Although these targeting strategies can improve transfection of specific immune cell types, the conjugation of targeting ligands to LNPs requires additional chemical conjugation and purification steps. This increases the overall complexity of the formulations and can lead to manufacturing challenges for scale up production of LNPs.

CD8+ T cells represent another subset of T cells that are appealing targets for gene delivery due to their cytotoxic capabilities that can be redirected to kill diseased cells. However, the T cell receptor (TCR) repertoire is incredibly diverse with estimates of the number of possible gene rearrangements at ~100 million; therefore, selective delivery of genetic payloads to the specific subset of CD8+ T cells that are relevant for a disease remains challenging. To enable antigen-specific T cell delivery, liposomes that display class I peptide major histocompatibility complexes (pMHCI) were developed to activate antigen-specific T cells in vitro [25] and led to a new class of LNPs coined “antigen-presenting nanoparticles (APN)” that use surface-bound pMHCI to facilitate cellular uptake by T cells in an antigen-specific manner [26]. Using a mouse model of PR8 flu infection, multiplexed delivery of APNs with an encapsulated reporter mRNA resulted in simultaneous transfection of distinct, antigen-specific T cell populations in vivo. Other targeting moieties that are not antibodies can be fused to the surface of LNPs to enable cellular uptake based on the conformational state of integrins expressed on the surface of immune cells, which play a vital role in the migration of cells to disease relevant sites. For example, gut-homing leukocytes are characterized by the expression of the integrin α₄β₇ in a high-affinity conformation state. LNPs that were decorated with a recombinant fusion protein that specifically recognizes the α₄β₇ integrin in a high affinity conformation transfected these gut-homing leukocytes with short-interfering RNA to silence interferon-γ in the gut, decreasing intestinal inflammation in murine models of experimental colitis [27].

There are trade-offs, however, with targeted nanoparticles including LNPs. A large majority of targeted nanoparticles in clinical trials have been unsuccessful with little to no improvements in overall accumulation in tumors or extrahepatic delivery compared to their untargeted counterparts, as well as limited improvement to treatment efficacy [28,29]. Moreover, chemistry, manufacturing and controls (CMC) specifications are increased in complexity arising from the additional steps in the manufacturing pipeline to achieve consistency between batches. These steps include expression of protein targeting ligands, proper removal of endotoxins to decrease potential immunogenicity of targeting moiety, and the additional bioconjugation and purification steps. Additionally, monoclonal antibody-based delivery systems carry the theoretical risk of inducing unwanted immune responses; proper measures need to be to be enforced to minimize immunogenicity, such as humanization and prevention of immunogenic post-translational modifications (e.g. glycation, deamidation and oxidation of amino acid side chains) [30]. Thus it is likely that the success of the current and future gene delivery formulations will not just require development of the gene carrier itself, but also the incorporation of control mechanisms to selectively activate the nucleic acid cargo in the desired tissue or cell type [31,32].

Polymeric nanoparticles

Delivery vehicles comprised of biodegradable polymeric NPs are a promising technology for in vivo gene delivery due to their spontaneous hydrolytic degradation into non-toxic byproducts in aqueous environments, which are easily eliminated by the body [33]. To engineer polymeric NPs capable of immune cell specific delivery, three distinct strategies are employed: (1) modification of existing polymer chemistries, (2) discovery of novel polymer chemistries, (3) targeting through conjugation of targeting moieties.

Poly(lactic-co-glycolic acid) (PLGA), an FDA approved polymer for small molecule drug delivery [34] is commonly chemically modified to improve nucleic acid encapsulation and delivery efficiency of polymeric NPs. Notably for immune cell delivery, modifying PLGA with a positively charged poly(L-lysine) group enables efficient encapsulation of micro-RNA (miRNA) and preferential delivery to splenic T cells, which was used to stabilize regulatory T cell mediated self-tolerance in mice and alleviated systemic lupus erythematosus (SLE) progression [35]. Although modifying FDA approved polymers minimizes the safety risk of the delivery vehicle, the potency and biodistribution may be limited by the intrinsic properties of the chemical structures. To address this, new classes of polymeric NPs are developed by synthesizing novel polymer structures and screening large chemical libraries. For example, charge-altering releasable transporters (CART) were first developed for mRNA delivery [36] and screening combinatorial libraries led to the development of a novel carrier with improved potency in lymphocyte transfections [37]. Intratumoral delivery of mRNA-CART encoding co-stimulatory and immune-modulating factors induced anti-tumor immunity from a single localized injection through the transfection of tumor-infiltrating dendritic cells, macrophages, and T cells, and presents a unique strategy for local in vivo gene therapy [38]. A separate class of polymers known as zwitterionic phospholipidated polymers (ZPPs) enables preferential delivery of mRNA to the spleen and lymph nodes in mice following systemic administration, extending the utility of polymer families for mRNA delivery to lymphoid organs in vivo [39].

Contrary to the prior strategies which focus on polymer chemistry, targeting strategies rely on identification of unique or overexpressed receptors on the target cell and conjugation of the targeting moieties (e.g., receptor ligands, antibodies) onto the nanoparticles. For example, macrophages and dendritic cells express a transmembrane glycoprotein called macrophage mannose receptor 1 (MCR1) and conjugating the MCR1 targeting ligand Di-mannose on the surface of nanoparticles presents a strategy to selectively deliver genetic cargo to these cell types. This strategy was leveraged with poly(glycolic acid) (PGA) polymeric NPs to deliver mRNA to tumor-associated macrophages, inducing a phenotype change to an anti-cancer phenotype [40]. The modular nature of this targeting strategy can be applied to target other immune cell types in vivo, such as T cells. For example, poly(beta-amino ester) (PBAE) NPs with surface conjugated antibodies that target T cell surface markers (CD3 or CD8) delivered disease-specific CARs or T-cell receptors (TCRs) mRNAs and generated therapeutic effects in human leukemia, prostate cancer, and hepatitis B-induced hepatocellular carcinoma mouse models upon repeated dosing [41]. It is important to note that despite being one of the most widely studied polymers for gene delivery, PBAE nanoparticles can trigger an immune response based on the degree of degradation [42]. The immunogenicity of the delivery vehicle is an important factor to assess especially while developing new polymers as it can potentially affect the area of applications and feasibility of repeat dosing.

Aside from mRNA delivery, conjugated polymeric nanoparticles have been shown to successfully deliver CAR pDNA to macrophages and T cells in vivo [43,44]. Encapsulation of both mRNA and DNA by polymers relies on charge interactions between the negatively charged mRNA or DNA and the positively charged polymer [45]. Due to the difference in size and structure, the optimal molar mass and charge density of the polymer used for delivery to maximize transfection efficiency are different between mRNA and DNA [46]. However, DNA delivery faces multiple challenges including the requirement of nucleus localization for transcription to occur and potential insertional mutagenesis [47]. In the case of DNA vaccines, this leads to more stringent FDA recommendations in terms of DNA production and integration studies [48]. While choosing between the delivery of mRNA or DNA, one should consider the delivery routes, delivery vehicles, and downstream regulatory barriers.

Viral vectors and virus-like particles

Lentiviral vectors (LVV) have been crucial in the development of gene-modified cell treatments, as demonstrated by the FDA-approved CAR T cell therapies [49]. The capacity of LVVs to stably integrate transgenes into the host cell genome makes them suitable for transducing replicating immune cells. However, in vivo application of LVVs is limited by the safety concerns associated with their broad tropisms, and thus considerable efforts have been made to engineer the surface glycoproteins of LVVs to enhance target cell specificity [50]. The most effective approaches to date combine site-specific mutations in glycoproteins to diminish the native receptor tropism of the LVV and direct addition of surface-targeting ligands to redirect the LVV. This strategy has been leveraged with engineered glycoproteins that recognize lymphocyte surface markers, which succeeded in establishing CAR T cells in vivo [51]. Recent designs of single-chain variable fragments (scFvs) and designed ankyrin repeat proteins (DARPins) enable selective targeting of T cell co-receptors such as CD3, CD4, and CD8 and provide a powerful method to stably integrate transgenes into the genome of circulating T cells [52–54].

Pseudotyping, the process of introducing non-native glycoproteins to the envelope of LVVs, can alter the tropism of the LVV to match the tropism from which the glycoprotein is derived. LVVs that are pseudotyped with mutated glycoproteins can prevent binding to human cells and be redirected to specific cells through the design of a bispecific antibody that recognizes the mutated glycoprotein and a unique cell marker, which was utilized to target circulating T cells and generate CAR T cells in vivo with a single dose [55]. Furthermore, pseudotyping LVVs with pMHC enabled gene delivery to antigen-specific T cell subsets based on TCR-pMHC interactions [56] and pseudotyping with transmembrane domains displaying B-cell antigen epitopes enables antigen-specific B-cell gene delivery [57]. Additionally, pseudotyped LVVs with a barcoded viral genome were developed to directly link immune cell receptors with their cognate antigens [58]. This high-throughput platform utilized single-cell sequencing to screen and match TCR- and B cell receptor (BCR)-interactions with a library of antigen-displaying LVVs to determine novel receptor pairs associated with disease. Although the success of LVVs for stably introducing transgenes is well documented, there remain concerns about the risk of insertional mutagenesis and adverse immune reactions [59]. Even though there is currently no direct evidence suggesting cancer development due to insertional mutagenesis of LVVs, there are reports of aberrant cell growth and clonal expansion after LVV transduction [60,61]. This concern is reflected in the FDA’s guidelines in choosing the input multiplicity of infection (MOI) of all retroviruses [62].

Adeno-associated virus (AAV) vectors can also be used to engineer lymphocytes in vivo. Systemic injection of both standard and exosome-associated preparations of AAV8 vectors mediated in vivo transduction of fluorescent reporter proteins in immune cells, including CD4 T cells, CD8 T cells, B cells, macrophages, and dendritic cells, establishing the feasibility of AAVs to engineer lymphocytes in vivo [63]. A similar approach was leveraged for in vivo delivery of AAV-DJ vectors encapsulating anti-CD4 CAR DNA, which led to significant tumor regression in murine models of human adult T cell leukemia after a single dose [64]. Additionally, AAV-DJ vectors have been used to genetically engineer B cells in vivo. After intravenous delivery of two AAV-DJ vectors, one coding for Staphylococcus aureus Cas9 and the other an anti-HIV broadly neutralizing antibody, B cells were stably edited in vivo and secreted high titers of neutralizing antibodies [65]. Future efforts to modify the capsid of the AAV vectors, such as with DARPins targeting mouse CD8 [66], may enhance the efficiency and specificity of the cell types transduced, but further in vivo testing is necessary to see the on/off-target effects of these targeted AAV vectors.

Virus-like-particles (VLPs) have been explored as alternative strategies to deliver gene-editing agents, since they can exploit the cell targeting advantage of viral delivery while avoiding the risks associated with viral genome integration and prolonged expression of the editing agent [67]. VLPs are multimeric protein complexes composed of viral structural proteins, which are structurally analogous to native viruses but lack the viral genome. Recent studies used VLPs to deliver immunostimulatory oligodeoxynucleotides in vivo to modify tumor-associated-macrophages [68] as well as deliver Cas9 ribonuclear protein (RNP)/single-guide RNA (sgRNA) complexes and base editors ex vivo to program lymphocytes [69,70]. Similar to viruses, VLPs can be pseudotyped to alter the tropism of the particle to immune cells, such as T and B cells, through native viral glycoproteins that enable lymphocyte targeting ex vivo, such as HIV-1, baboon endogenous virus (BaEV), and vesicular stomatitis virus (VSV) glycoproteins. VLPs ability to efficiently package RNPs outcompetes LNPs, which offers unique advantages compared to mRNA delivery, such as minimal exposure to gene editing agents due to the shorter lifetime of RNPs in the target cell, thus lowering the chances of off-target editing [71]. However, future work is still needed to eliminate off-site delivery at the tissue and cell level to enable efficient in vivo delivery of therapeutic gene editing agents to immune cells.

Conclusion

In vivo gene delivery to immune cells is a promising avenue for the future success of immune cell-based therapy. Advances in engineered gene delivery vehicles such as LNPs, polymeric NPs, viruses, and VLPs hold great clinical promise for in situ modulation of immune cells and reducing the cost and time required for ex vivo immune cell manufacturing. Their modularity enables a wide variety of applications dependent on the genetic payload encapsulated (e.g. mRNA, plasmid DNA, CRISPR/Cas9, siRNA) and the cell type targeted (e.g. T cells, dendritic cells, macrophages). Looking ahead, as the field of synthetic biology further advances, we anticipate a future shift to focus not just on the delivery of single protein systems (e.g. cytokines, receptors) but also to the delivery of synthetic biocircuits with increasing complexity such as sense-and-response genetic components to allow autonomous or remote control of immune cell activity [72–75]. Recent work in mRNA circuits which enables the regulation of mRNA translation based on cell type specific microRNA expression profile or exogenous siRNA holds potential in increased specificity in immune cell engineering [31,32]. These advances could also lead to the in situ production of immune cells as living diagnostics that can home, target and query sites of future disease by releasing synthetic biomarkers [76]. Recent demonstrations using adoptively transferred cell sensors such as engineered macrophages or probiotics showed that such approaches using synthetic biomarkers can lead to significantly improved detection sensitivity and specificity compared to naturally occurring biomarkers shed by diseased cells and tissues [77,78]. A significant hurdle for in vivo gene delivery continues to be the need for increased specificity and localization of genetic carriers to lymphoid organs and immune cells while minimizing off-target activity and systemic toxicity. These challenges will be compounded for chronic diseases such as autoimmune disorders since repeated dosing will likely be required for disease management. Solving these challenges, however, would have an enormous impact on accessibility of cellular immunotherapies and accelerate their adoption for numerous indications beyond oncology.

Figure 1: Uptake and intracellular release mechanisms of lipid nanoparticles (LNPs), polymeric nanoparticles (NPs), virus-like nanoparticles (VLPs), and lentiviruses.

Depending on the genetic cargo delivered, LNPs, polymeric NPs and VLPs can achieve both transient and stable effects. Non-viral nanoparticles rely on charge interactions to disrupt the endosome and release the genetic material. Genetic components delivered by lentivirus integrate into the genome to achieve prolonged effect. Lentivirus contains reverse transcriptase in the delivered viral genome to facilitate the conversion of cargo RNAs into DNAs before integrating into the host’s genome.

Figure 2: Targeting specific immune cells in vivo leads to diverse clinical applications.

A. Tumor-associated macrophages (TAMs) can be polarized from a cancer-promoting phenotype (M2) to an anti-tumor phenotype (M1) through the targeted delivery of mRNA encoding for interferon regulator factor 5 (IRF5) and the kinase IKKβ [40], siRNA to target and silence the STAT3 and HIF-1α genes [19], and CpG nucleotides [68]. B. Antigen presenting cells are targets for improved vaccination strategies. Targeted delivery of mRNA encoding for tumor-antigen peptides to dendritic cells enables antigen processing and presentation on major histocompatibility complexes I and II (MHC class I and MHC class II) to bolster immune response [18]. C. Chimeric antigen receptor (CAR) T cells engineered in vivo kill activated fibroblasts that are responsible for promoting cardiac fibrosis [23]. D. Targeting T cells in vivo and delivering micro-RNA (miRNA) that stabilizes T_reg mediated-self-tolerance restores effector/regulatory T cells balance for autoimmune diseases treatment [35]. E. CAR T cells are engineered in vivo through targeted delivery of mRNA [41] or viral RNA [53,55,79] for transient or durable CAR expression respectively.

Table 1:

Comparison of genetic cargo and delivery vehicles used to engineer immune cells in vivo and their associated applications.

Delivery Vehicle	Advantages and caveats	Cell Type	Genetic Cargo	In Vivo Disease Models
Lipid Nanoparticles (LNPs)	LNPs (+) Proven scale up capability (+) Proven safety profile in gene delivery (+) Tissue specificity (−) Lack of cell type specificity (−) Requires low temperature storage	Tumor associated macrophages	siRNA	Cancer (OS-RC-2 renal cell carcinoma tumor model) [19]
		Antigen presenting cells	Plasmid DNA/mRNA	Vaccination [17]
		CD8+ T cells	mRNA	Cardiac fibrosis (fibroblast activation protein (FAP) CAR) [23]
		Antigen specific CD8+ T cells	mRNA	Influenza [26]
		High-affinity α4β7 expressed by Gut-homing leukocytes	siRNA	Colitis [27]
Polymeric Nanoparticles (NPs)	Polymer NPs (+) FDA approved for non-gene delivery applications (+) Single component formulation (+) Simple formulation process (+) Less safety concerns compared to virus (−) Higher immunogenicity than LNPs (−) Requires lyophilization for storage	Splenic T cells	miRNA	SLE [35]
		Intratumoral immune cells	mRNA	Cancer (A20 B cell lymphoma model and CT26 colon carcinoma model) [38]
	Targeted NPs (+) Cell type specificity (−) Higher production cost compared to nontargeted (−) Potential tissue level off-target delivery (−)Requires lyophilizationfor storage	CD3+/CD8+ T cells	mRNA	Cancer (1928z CAR, HBcore18–27 TCR and ROR1 CAR) [41]
		Tumor associated macrophages	mRNA	Cancer (vascular epithelial growth factor (VEGF)-expressing ID8 ovarian tumor, B16F10 lung metastatic tumor and PDGFβ-driven glioma) [40]
Virus	(+) Efficient and stable gene transfer (+) Transgene expression can be controlled by virus (transient or persistent) (+) Cell type specificity (dividing and non-dividing cells) (−) Difficult to manufacture (−) Higher immunogenicity (−) Low packaging capacity (−) safety concerns associated with insertion mutagenesis (−) High cost	CD4+ T cells	RNA	Cancer (CD19 CAR) [52]
		CD8+ T cells	RNA	Cancer (CD19 CAR) [53]
		CD3+ T cells	RNA	Cancer (CD19 CAR) [54,55]
Viral-like Particles	(+) Better loading capacity compared to viral vectors (+) non-infectious (+) deliver CRISPR and base editor (−) off-target delivery to tissue	Tumor associated macrophages	CpG Oligonucleotides	Cancer (CT26 murine colon cancer model) [68]

Highlights:

• Immune cell therapies are a new class of treatments for complex human diseases.

• Engineering immune cells in vivo can improve manufacturing, efficacy, and safety.

• Gene delivery vehicles can selectively target immune cells in the body.

• In vivo gene therapy can be used to treat cancer, infection, and autoimmunity.