Viral and Nonviral Transfer of Genetic Materials to Adipose Tissues: Toward a Gold Standard Approach

Our understanding of physiology is heightened by our ability to manipulate cells and tissues. Through gene transfer, we can deliver genetic agents for gene silencing, protein overexpression, or genome modification to further our understanding of biological phenomena. To accomplish gene transfer in a tissue-specific manner, one must consider two critical components: the vector and the route of administration. An ideal vector shields its cargo from degradation, targets a specific tissue, facilitates unloading of its cargo into cells, and evades the host immune response. Both viral and nonviral vectors meet these criteria (1–3). Secondly, to ensure the vector reaches its desired destination, one must determine an appropriate route of administration. In general, vectors are administered systemically or locally, depending on the tissue of interest. Gene transfer has largely been aimed at the liver, heart, skeletal muscles, and central nervous system; however, its application to adipose tissue has been met with limited success. Collectively, adipose tissues comprise the largest endocrine organ in the body with important regulatory functions for energy balance and metabolism, thus making it an attractive focal point for understanding obesity and the metabolic syndrome. In this section, we will outline how both vector and route of administration influence transduction efficiency. By careful optimization of these two factors, gene transfer can be accomplished in an adipose-specific fashion.

Adipocytes are located throughout the body mainly in subcutaneous, visceral, and marrow depots as white (WAT), brown (BAT), or more specialized adipose tissues (4–6). At the cellular level, mature adipocytes comprise only 11–40% of the total cell population in adipose tissue, which also contains preadipocytes, stem cells, macrophages, endothelial cells, and vasculature, collectively referred to as stromal vascular cells (7). WAT constitutes the bulk of adipose tissue and is found in subcutaneous and visceral depots. In mice, the subcutaneous WAT includes the interscapular, subscapular, axillary, and cervical depots in the anterior and the dorso-lumbar, inguinal, and gluteal depots in the posterior. Subcutaneous WAT are also closely associated with lymphatic tissues and in females comprise part of the mammary glands. Visceral adipocytes are contained within the peritoneal cavity and include the mesenteric, perirenal, retroperitoneal, and gonadal (epididymal/ovarian) depots. An inability to expand these depots to store excess energy is typically associated with insulin resistance and metabolic dysfunction (5,6). Additionally, there are several smaller depots including the intramuscular, periarticular, paracardial, epicardial, retro-orbital, and dermal WAT. By contrast, BAT is predominantly found in the interscapular depot but also in the periaortic, perirenal, and intercostal depots (8). Brown adipocytes are characterized by a multilocular lipid droplet morphology and are rich in mitochondria (4–6,8). In response to temperatures below thermoneutrality (lower than ∼30°C in mice), brown adipocytes express uncoupling protein 1 (UCP1), a key mechanism for adaptive nonshivering thermogenesis (8). Although much smaller in terms of adipose tissue mass, BAT metabolism is the predominant energy consumer in mice at cold temperatures, making it an attractive therapeutic target for treating obesity (8,9).

Given the diversity and distribution of adipose tissue, gene transfer has remained challenging. In vivo manipulation has been limited to the generation of Cre-Lox and/or drug-inducible transgenic models. Cre-Lox systems enable the expression of Cre recombinase under the control of adipocyte-specific promoters. Several adipocyte-specific promoters including fatty acid binding protein-4 (FABP4) and adiponectin (AdipoQ) have been developed (10). Although both have been used extensively, FABP4-Cre activity has been detected in the brain, endothelial cells, macrophages, adipocyte progenitors, and embryonic tissues (11–13). Thus, the field has shifted to conditional (AdipoQ-Cre) or inducible (AdipoQ-Cre^ERT, AdipoChaser) lines (13–15). While useful, Cre mice must be bred to transgenic mice harboring a floxed gene. In some cases, a new floxed line must also be generated, which is time-consuming and expensive and involves large cohorts of animals. Therefore, more efficient strategies, such as gene transfer, would be beneficial to the field.

Adipose tissue is very heterogenous, and in many cases, vectors transduce stromal vascular cells more efficiently than the adipocytes (7). Thus, a vector must successfully navigate the adipose tissue microenvironment to home to the adipocytes. Important considerations in selection of a vector should include cargo, tropism, expression profile, and immune response (Table 1). In this section, we will discuss advancements using viral and nonviral vectors for transducing adipocytes.

Different classes of viruses, including lentivirus, adenovirus, and adeno-associated virus (AAV), have all been used for gene transfer to adipose tissue. While lentiviruses and adenoviruses efficiently transduce adipocytes in vitro, efforts to mediate gene transfer in vivo have been less successful. For example, lentiviruses target CD4⁺ cells and bind to the human adenovirus C receptor, which is not expressed by adipocytes (16,17). Moreover, lentiviruses integrate into the genome leading to insertional mutagenesis (16). While several studies using adenovirus have been performed, concerns regarding its strong immunologic response and generation of antibodies against its capsid proteins have caused the field to shift to AAV (18–26). AAV is characterized by low immunogenicity, epichromosomal expression, and a multitude of serotypes and capsid variants (27).

Seminal work from the Bosch laboratory demonstrated that AAV8 and AAV9 preferentially transduce white and brown adipocytes with variable transduction efficiencies across depots (28). Following systemic delivery, AAV8 and AAV9 GFP were highly expressed in the epididymal WAT and interscapular BAT but marginally expressed in the inguinal, retroperitoneal, and mesenteric depots. Interestingly, variability in GFP expression was observed across different strains of mice including C57/BL/6J, ICR, ob/ob, and db/db, suggesting that genetic background may influence transduction efficiency (28,29). For improvement of tropism, recombinant AAV (rAAVs) have been engineered by capsid shuffling to increase affinity for adipose tissue. For example, Liu et al. (30) generated rAAVs harboring GFP (Rec2-GFP) and demonstrated increased adipocyte transduction in comparison with the natural AAV serotypes 1, 8, and 9. Rec2 vectors were outfitted with Cre recombinase and injected into the inguinal WAT depots of floxed–insulin receptor mice, which led to ∼53% knockout of the insulin receptor and 50% decrease in inguinal WAT mass. Interestingly, these findings are in contrast to AdipoQ-Cre–mediated knockout mice, which develop lipodystrophy and metabolic dysfunction characterized by insulin resistance and hyperglycemia (31). This might be explained by the fact that Rec2-Cre knockout occurs in the adult mouse, whereas AdipoQ-Cre causes recombination early in adipocyte development.

As an alternative approach, nonviral vectors, including organic and inorganic nanoparticles, have also been explored for gene transfer to adipose tissue. Nonviral nanoscale vectors (<200 nm in diameter) have unique characteristics that make them attractive targets for gene transfer and pharmacologic applications. For example, nonviral vectors can be functionalized with macromolecules to impart tissue specificity (2,3). Efforts to impart adipose specificity have largely focused on conjugating nonviral vectors with the peptide sequence CKGGRAKDC (adipocyte-targeting sequence [ATS]), which binds to the prohibitin receptor on the subcutaneous WAT vasculature (32). Hossen et al. (33) formulated ATS-conjugated nanoparticles that accumulated in the adipose tissue vasculature of mice with diet-induced obesity. In another study, cadmium-based quantum dots functionalized with ATS targeted the WAT of obese Wistar rats (34). While the mechanism by which ATS-based vectors escape the vasculature to transduce adipocytes remains unknown, research has shown that targeting adipose tissue vasculature directly can influence the surrounding tissue. For example, Xue et al. (35) developed polymer-based nanoparticles conjugated to ATS and loaded with rosiglitazone or prostaglandin E2 analog. In treated obese mice, WAT vascular density was increased coupled with the upregulation of the BAT markers Ucp1, Cidea, and Dio2. Thus, modulation of adipose tissues can be achieved by targeting the vasculature.

Modulation of adipose tissue can also be achieved by targeting the resident macrophages to suppress obesity-induced inflammation. In the obese state, endogenous low-molecular-weight hyaluronic acid binds to CD44-expressing cells to trigger inflammation and insulin resistance (36). Rho et al. (36) developed hyaluronic acid nanoparticles (HA-NPs) to block this pathway and reduce inflammation. HA-NP–treated obese mice had significant reductions in body weight, glucose tolerance, and insulin resistance. Additionally, HA-NPs reduced macrophage infiltration and production of inflammatory cytokines in epididymal WAT, likely by outcompeting low-molecular-weight hyaluronic acid for CD44 binding sites. It should be emphasized, however, that HA-NPs accumulated predominantly in liver, lungs, and kidneys, with only marginal transduction observed in the epididymal, subcutaneous, and brown depots. In another study, Ma et al. (37) synthesized dextran conjugates loaded with dexamethasone to reduce inflammation in obese mice. Impressively, a single dose reduced the expression of TNFα, IL-6, and MCP-1 in the gonadal, perirenal, mesenteric, and subcutaneous depots. Although these studies do not involve gene transfer per se, it can be reasoned that genetic cargo can be incorporated into these nonviral vectors. In fact, several nonviral nanoscale vectors have entered clinical trials for the delivery of siRNA to the vasculature (38). Thus, future strategies may involve non-viral-mediated gene transfer to adipocytes, vasculature, or macrophages.

Although both viral and nonviral vectors have been investigated (Fig. 1), there is no consensus method for transducing adipose tissue largely because adipocyte specificity remains a major limitation. Therefore, future success will depend on creating targeted vectors that enhance tissue tropism and impart tissue-specific transcriptional control. For improvement of tissue tropism, viral envelope glycoproteins have been modified with ligands, peptides, and antibodies. Yang et al. (39) created an efficient method that incorporates antibodies onto the viral surface and enables the virus to bind to a specific cell type and deliver its transgene. While this method has not been used to transduce adipocytes directly, one can envision the use of adipocyte-specific antibodies against adipocyte-selective targets such as the amino acid transporter ASC-1 or the brown cell surface markers PAT2 or P2RK5 to improve tropism (40). Adipose-specific promoters such as AdipoQ or UCP1 can limit off-target transgene expression to impart transcriptional control (10,13). It should be noted that the large size of the AdipoQ locus required for adipocyte specificity restricts its use with AAV, although “mini” promoters, such as mini/UCP1, have been implemented (28). Another strategy uses tissue-specific miRNAs targeting posttranscriptional regulatory elements of the transgene. For example, Jimenez et al. (28) incorporated multiple miRNAs to prevent transgene expression in liver and heart but retained transgene expression in WAT and BAT, respectively. Thus, targeted vector design that enhances tropism and imparts adipose-selective transcriptional control can significantly improve transduction efficiency and limit off-target expression. A combination of these strategies will likely emerge as the consensus method for gene transfer to adipose tissue. It should be emphasized that careful consideration should be paid to validating transduction efficiency and selectivity, either by incorporating a fluorescent probe or by separating the adipocytes from stromal vascular cells. We will now turn our attention to the second critical component of delivery: route of administration.

Generally speaking, vector administration can proceed via systemic or local delivery (Fig. 2). Ideally, a vector should transduce every adipocyte across all adipose depots, although this goal has never been fully realized. Therefore, in selection of a route one should carefully consider the experimental aims, anatomical location, degree of vascularization, and invasiveness in order to optimize transduction efficiency.

Intravenous injection capitalizes on the circulatory system’s direct access to all tissues in the body. Despite this, the varying degree of vascularization in each adipose depot coupled with vector accumulation in reticuloendothelial organs (liver, lung, heart, and kidney) limits adipocyte transduction efficiency (1,2,5,6). Successful gene transfer using intravenous injection via the tail vein has only been accomplished with adenovirus or AAV; however, the field has since shifted exclusively to AAV because of its favorable properties. For example, O’Neill et al. (41) intravenously injected AAV2/8 encoding leptin to leptin-deficient mice. A single injection increased plasma leptin levels to 7% of that observed in wild-type mice, which led to decreased body weight, food intake, and fasting insulin levels and improved glucose tolerance. In another study, rAAV8 vectors expressing perilipin A were intravenously injected into mice (42). Treated mice showed elevated expression of perilipin A in visceral and subcutaneous WAT and the liver with a decrease in serum free fatty acids. Interestingly, however, both of these studies reported transgene expression in reticuloendothelial organs despite having transcriptional control features, thus highlighting the off-target issues associated with systemic delivery.

Intraperitoneal injection can be rapidly accomplished in conscious animals with minimal restraint. Vectors injected intraperitoneally are primarily absorbed into the mesenteric vessels that drain into the portal vein and pass through the liver (43). Huang et al. (44) injected mice with Rec2-GFP vectors encoding a liver-specific miRNA to prevent off-target expression. Substantial GFP expression was observed in the epididymal WAT but not in other adipose tissues examined or in reticuloendothelial organs. GFP was exchanged for leptin and injected into leptin-deficient mice, which increased lean mass and circulating leptin levels, improved glucose tolerance, and reversed hepatic steatosis. An important caveat to this study is that vectors transduced only one adipose depot despite being injected via a systemic route.

Enteral delivery through oral gavage is economical, convenient, and relatively safe and requires moderate technical skill (43). In a proof of concept study, Huang et al. (45) examined the effect of oral gavage on the biodistribution of Rec2 vectors. Interestingly, Rec2-GFP expression was only detected in the BAT, inguinal WAT, and liver and not in the stomach, intestine, heart, kidney, or brain. Rec2-Cre vectors were orally administered to floxed-VEGF mice for targeted deletion in BAT. Cold exposure (4°C for 4 h) led to an inability to defend core body temperature, decreased BAT mass, and VEGF protein depletion by ∼80%. VEGF was then overexpressed with use of orally administered Rec2, which led to an increase in BAT mass and VEGF, UCP1, and CD31 protein levels. The potential mechanism by which Rec2 bypasses the gastrointestinal system to preferentially transduce BAT has not been described; however, the authors postulate that the lymphatic system may play a role, as Rec2 was detected in the mesenteric adipose depot, which is rich in lymph nodes.

In contrast to systemic delivery, local delivery is a more targeted approach in which vectors are administered directly into a tissue. Although slightly more invasive, time-consuming, and resource heavy, local delivery via intradepot injection retains vectors at the site of injection, enabling the study of depot-specific function (27,43,46). For adipose tissue, intradepot injection has proceeded through intra-WAT or intra-BAT injection of viral and nonviral vectors. Before attempting intradepot injection, we recommend reviewing references (4–6,27,46) for guidance on how to access these tissues.

Mizukami et al. (47) were the first group to optimize intra-WAT injection by administering AAV1-erythropoietin into the subcutaneous WAT. Treated mice showed elevated plasma erythropoietin concentration compared with controls. The transduced subcutaneous WAT was then surgically removed, causing plasma erythropoietin concentrations to return to baseline, indicating that increased circulating erythropoietin was the result of gene transfer to WAT specifically. Work from the Bosch laboratory explored AAV-mediated gene transfer to overexpress fibroblast growth factor 21 as a therapeutic strategy for treating obesity (48). AAV8 that expressed fibroblast growth factor 21 was administered directly into liver, epididymal WAT, or skeletal muscle of obese mice. Interestingly, a single injection into the epididymal WAT led to a dose-dependent reduction in adipocyte size, improvement in insulin sensitivity, and prevention of hepatic steatosis, results that were comparable with those of both liver- and skeletal muscle–specific injections. Collectively, these studies demonstrate that a single dose of AAV to one adipose depot is sufficient to mediate systemic changes to metabolism.

Lentiviruses have also been used for intra-WAT gene transfer applications. Gnad et al. (49) injected a lentivirus expressing the A_2A receptor into the inguinal WAT of obese mice as an antiobesity therapy. Stimulation of sympathetic drive causes BAT to release adenosine, which binds to the A_2A receptor to increase energy expenditure. It was hypothesized that overexpression of the A_2A receptor in inguinal WAT, which has the capacity for browning, might have therapeutic potential. Interestingly, lentiviral expression of the A_2A receptor significantly increased thermogenic and lipogenic markers and led to a brown adipocyte–like morphology in the inguinal WAT with increased expression of Ucp1, Pgc1α, and Prdm16.

Only one study has attempted intra-WAT injection of a nonviral vector for gene transfer. The Kim laboratory synthesized an ATS-oligopeptide complex to administer shRNA to silence fatty acid binding protein-4 (FABP4) and combat obesity (50). Obese mice were administered the complex via intra-WAT injection on one side of the “abdominal” adipose tissue. Strikingly, FABP4 expression was decreased by 70% compared with the contralateral depot. Treated mice also showed significant improvements in insulin sensitivity and reduced blood glucose levels. Perhaps the most surprising data are that treated mice lost nearly 20% of their body weight following a single administration to one WAT depot. While these data are interesting, they also raise several red flags. Firstly, the mechanism by which the ATS-oligopeptide-shRNA complex binds to prohibitin on adipose tissue vasculature and escapes to transduce the adipocytes has not been explained. Secondly, the authors do not report changes in energy balance, lipolysis, inflammation, apoptosis, or cell death or changes in tissue mass in the injected WAT depot, leading us to question how silencing FABP4 in one adipose tissue equates to a 20% reduction in body weight. This study is also in direct contradiction with previous work that shows FABP4 knockdown or knockout in obese mice increases body weight and fat mass without changes in glucose homeostasis (51,52). Moreover, it does not report on FABP5 expression, which is compensatory to loss of FABP4 (53).

Similar to intra-WAT injection, AAV has remained the most commonly used vector for intra-BAT injection. Two aforementioned studies attempted intra-BAT injection using AAVs. Jimenez et al. (28) showed that intra-BAT injection of AAV9 encoding VEGF164 under the control of the mini/UCP1 promoter increased vessel density and VEGF164 expression specifically in BAT. In the second study, Liu et al. (30) administered Rec2-Cre bilaterally into the BAT of floxed–insulin receptor mice, which decreased BAT size by 50% and expression of insulin receptor protein by ∼60%. Surprisingly, disruption of insulin receptor signaling in BAT induced gene expression changes in the hypothalamus, highlighting that depot-specific gene transfer can be used to study the interaction between adipose tissue and the central nervous system.

Although significant progress has been made using viral and nonviral vectors, the field lacks a gold standard approach. Experimental aims will always dictate the selection of a vector and route of administration; however, to guide the reader we propose the following set of guidelines:

• 1) Transcriptionally controlled AAVs currently show the most promise for transducing adipocytes in adipose tissues. Researchers should prioritize rAAV, AAV8, or AAV9 that incorporates an adipose-specific promoter–driven transgene (mini/AdipoQ or mini/UCP1), an miRNA to limit transgene expression in off-target tissues, and a reporter molecule to validate transduction efficiency.

• 2) To maximize adipocyte transduction efficiency and limit accumulation in off-target tissues, researchers should administer vectors locally via intra-WAT or intra-BAT injection.

• 3) For validation of your experiment, follow the gold standard approach outlined in Fig. 3. For validation of adipocyte transduction, vectors should be outfitted with fluorescent probes or administered to reporter lines such as the mT/mG mouse (54). Successful transduction is characterized by a fluorescent “halo” surrounding the adipocyte. As adipocytes represent only a small fraction of the total cells in adipose tissue, they should be separated from stromal vascular cells by buoyant-density centrifugation, and both fractions should be analyzed for transduction.

While we have focused on transducing adipocytes, it should be noted that targeting the vasculature or resident macrophages is an alternative strategy for manipulating adipose tissues. This is particularly true for nonviral vectors, although continued research is required.

With the emergence of powerful gene-editing technologies like CRISPR/Cas9, researchers can modify virtually any region of the genome. CRISPR/Cas9 is composed of two components: the Cas9 endonuclease and a programmable single-guide RNA (sgRNA) that directs Cas9 where to create double-strand breaks in genomic DNA (55). CRISPR/Cas9 is thus easily amendable to viral or nonviral gene transfer, as its constitutive parts can be genetically encoded. Despite the adoption of CRISPR/Cas9 gene transfer to a variety of tissues (56–59), only one study reports its use in adipose tissue (60). We recommend caution, however, as this study uses the ATS-oligopeptide complex described above for CRISPR-interference silencing of FABP4 (50,60).

Two studies have recently reported on in vitro delivery of CRISPR/Cas9 to cultured adipocyte precursors. The Czech laboratory successfully transduced isolated primary preadipocytes with nonviral vectors loaded with Cas9 protein and an sgRNA targeted to Nrip1 (61). Treated cells showed a mutation frequency of 43.8% in the Nrip1 locus coupled with increased expression of the brown-like genes Ucp1, Cidea, Pgc1α, Prdm16, and Cpt1b. Importantly, no off-target mutations were detected and CRISPR/Cas9 did not affect the capacity to differentiate. In a second study, Kamble et al. (62) electroporated human preadipocytes with a ribonucleoprotein complex of Cas9 and an sgRNA targeted to Pparg or Fkbp5. Impressively, treated cells achieved 90% efficiency and did not require selection or clonal isolation of edited cells. While these studies demonstrate in vitro proof of principle, the field has yet to report successful in vivo gene transfer of CRISPR/Cas9 to adipose tissue.

CRISPR/Cas9 gene transfer to adipose tissue is in its infancy; however, we can gain several insights from its use in other tissues and apply these to the gold standard approach. For example, CRISPR/Cas9 gene transfer can proceed via exogenous expression or a combination of exogenous/endogenous expression of its components. In the exogenous method, both Cas9 and sgRNA are delivered via a vector. In the exogenous/endogenous method, the sgRNA is delivered to a Cas9-expressing mouse line (63,64). Thus, rAAVs can be encoded with one or multiple sgRNAs and administered locally to AdipoQ-Cre–dependent Cas9 mice to enable spatial control (63). Cas9 expression can also be temporally regulated using a doxycycline-inducible line developed by Dow et al. (64). Ultimately, CRISPR/Cas9 has immense potential for elucidating the underlying mechanisms to a myriad of metabolic diseases. For instance, future application of CRISPR/Cas9 might entail homology-directed repair to correct single nucleotide polymorphisms associated with aberrant metabolism (65). Particular focus should be paid to integrating CRISPR/Cas9 with well-established gene transfer techniques such as AAV and Cre-Lox mouse models (63,64). Moreover, we also recommend a strong adherence to the gold standard approach to validate transduction efficiency. With these powerful technologies in hand, the field of adipose tissue biology is poised to enter a new era of research.

Acknowledgments. The authors thank members of the MacDougald laboratory for their helpful suggestions and review of the manuscript.

Funding. This review was supported by grants from the National Institutes of Health to O.A.M. (R24 DK092759, R01 DK121759, and R01 DK125513) and S.M.R. (T32 GM835326 and F31 DK12272301).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.M.R. researched and wrote the manuscript. O.A.M. provided scientific guidance and editorial assistance.