To date, investigations of orthopaedic gene therapy have primarily focused on the delivery of cDNA-encoding bioactive proteins that are secreted from the modified cells. Several distinct advantages favor this approach. First, the cDNA of signaling molecules tend to be small and can be inserted fairly easily in the limited space available in most viral vectors. Second, a relatively small population of genetically modified cells can release transgene products into the extracellular fluids to affect regional cell populations in a paracrine manner. Further, the soluble gene products present in conditioned media, biologic fluids, and tissue homogenates can be quantified by enzyme-linked immunosorbent assay, allowing compilation of dose response and pharmacokinetic profiles of both vector and gene product to define the functional parameters of the procedure.⁴

Alternatively, numerous gene products with therapeutic potential function intracellularly. Because there is no common mechanism by which exogenous proteins can be taken in by a cell and retain function, gene transfer is the only method by which the activities of certain types of molecules, for example, nuclear receptors, transcription factors etc., can be exploited for clinical use. Because
the effects of the gene product are limited specifically to the population of modified cells, the chance of adverse response from unintended diffusion to a nontarget tissue is eliminated. On the downside, overexpression of various transcription factors essential for chondrogenic differentiation and bone formation, such as RUNX2 and SOX9, has been shown to induce adverse skeletal phenotypes⁵ and tumorigenic activation.⁶ From a technical standpoint, for procedures involving direct in vivo gene delivery, for an intracellular gene product to induce a meaningful response at the tissue level or organ level, a large proportion of the resident cell population must be genetically modified by the vector. Regardless of vector type, functional in vivo gene delivery of this magnitude is an extremely challenging undertaking, even with viral systems in small laboratory animals. Further, because efficacy is a function of the number and locations of the cell populations modified by the vector (rather than total protein expression), exhaustive marker studies are required to assess and optimize dosing relative to the density and distribution of transduced cell populations in the target environment.

A virus is a biologic entity composed of genetic material (RNA or DNA) packaged in a protective shell (capsid). Through evolution, viruses have fine-tuned both their genome and capsid components for peak transduction efficiency making them attractive as vectors. The challenges lie with engineering a recombinant vector that is technically manipulable, maintains efficient transduction, and can be manufactured at high titer while at the same time eliminating its ability to reproduce in the host and cause pathology.⁹

In general, a viral vector is created by removing the coding sequences from its genome while leaving in place the noncoding, cis-acting sequence elements required for efficient replication and packaging into the viral capsid. Because the genome length of the wild type virus approximates the maximum amount of genetic material that can be efficiently inserted into the viral capsid, the removal of viral genes creates room for insertion of an exogenous expression cassette. Transfer and expression of the viral coding sequences required for replication in a complementing cell line allows for selective replication and
packaging of the vector DNA containing the cis-acting recognition sequences. The resulting recombinant viral particles can infect and transduce target cells but can only replicate in the complementing cell line. Removal of viral coding sequences from the vector genome precludes their expression in transduced cell populations, which reduces their immunogenicity and prolongs transgene expression (discussed later). As each viral vector is unique, with different genomes, host-cell range (tropism), and molecular mechanisms for transduction, the key to a successful platform involves tailoring the vector and expression cassette to the therapeutic needs of the target disease. The following sections describe the salient features of the most common viral vector systems in clinical gene therapy: adenovirus, adeno-associated virus (AAV), and retrovirus (gamma retrovirus [γ-retrovirus] and lentivirus) (Table 1).

Vectors derived from adenovirus have been widely used in orthopaedic research because of their broad tropism, high-level infectivity, and ease of propagation. Wild-type adenovirus is common in nature and generally associated with self-limiting respiratory tract infections. The viral capsid is a nonenveloped icosahedron approximately 80 to 100 nm in diameter that encases a linear, double-stranded DNA genome approximately 35 kb in length. The genome is flanked on either end by inverted terminal repeat (ITR) sequences, with a short psi sequence positioned immediately downstream from the 5’ (left hand) ITR that marks the DNA for packaging into the adenovirus capsid. The wild type genome encodes approximately 35 proteins, which are sequentially expressed in the early (E) and late phases of viral infection and replication.⁹

Early-generation adenovirus vectors were created by removing the immediate E1A and E1B genes, and later the E3 gene, creating room for insertion of an expression cassette of approximately 7.5 kb. Because expression of the E1A gene is required for transcription of the other viral genes, removal of the E1 locus from the adenovirus genome renders the vector replication deficient in normal cells. The vector is propagated by infection of the widely used 293 cells, derived from a human embryonic kidney cell line. A total of 293 cells were engineered to stably harbor the left hand in 11% of the wild-type adenovirus genome, and constitutively express the E1A and E1B proteins required for activation of the remaining viral genes that reside on the vector genome.¹¹ Infection of 293 cells with recombinant adenovirus provides the E1 proteins in trans-, complementing the defective vector genome to allow its replication. The ease of propagating recombinant adenovirus vectors in this way is very convenient but requires vigilant quality control because serial propagation selects for virions that have jettisoned their transgene cassette baggage to replicate more efficiently.

Adenovirus vectors offer several advantages for gene transfer applications: (1) highly efficient transduction of dividing and nondividing cells from a wide range of species and tissues, (2) a nonintegrating genome maintained as an extrachromosomal (episomal) element in nondividing cells; (3) high-level transgene expression with rapid onset; (4) a relatively large packaging capacity; and (5) technically straightforward methods of vector propagation. A distinct limitation of adenovirus vectors is the tendency to provoke immune responses against transduced cell populations in vivo. Despite removal of the E1 locus, leaky readthrough transcription allows low-level expression of the residual adenovirus coding sequences in transduced cells leading to neutralizing immune reactions.

To reduce the immunogenicity of vector-modified cells, third-generation vectors were developed in which all viral coding sequences were removed from the vector genome, leaving only the flanking ITRs and the psi packaging sequence. These gutted or high-capacity vectors accommodate up to 36 kb of exogenous DNA but require coinfection with a helper adenovirus for replication. Although removal of the viral coding sequences reduces the immunogenicity of the transduced cells and prolongs transgene expression, innate immune responses to the capsid protein and viral infection cause the vector to be inflammatory;¹² elimination of the helper virus from vector preparations presents an additional challenge.¹³

Because of the prevalence of wild-type adenovirus in nature, much of the human population is seropositive for neutralizing antibodies (NAbs) against one or more human variants, which renders common adenoviral vectors ineffective. To address this, vector systems have been developed from nonhuman variants common in other species. Currently, adenoviral vectors account for approximately 50% of the clinical trials worldwide, primarily for vaccination and anticancer protocols. Notably, two different adenoviral vectors, both developed from variants found in chimpanzees, are used to deliver and express the coding sequence for the SARS-CoV-2 spike protein and are in widespread use as vaccines against COVID-19.¹⁴

Recombinant adeno-associated virus (rAAV) is another nonintegrating viral vector with the ability to transduce both dividing and nondividing cells.¹⁵ Compared with other prominent viral vector systems, rAAV is accepted as the least toxic and, relative to adenovirus vectors, induces relatively low innate and adaptive immunity against transduced cells in vivo. Though its genome remains episomal, rAAV is capable of achieving long-term (>10 years) transgene expression in quiescent cell populations in vivo.¹⁶ The first human trial of AAV-mediated gene delivery was reported in 1998 for cystic fibrosis. Because of its favorable safety profile in this and more than 200 subsequent clinical studies, rAAV is currently the
preferred vector system for human protocols involving in vivo gene delivery.

Wild-type AAV is a small (<50 nm diameter), nonenveloped, single-stranded DNA virus that is nonpathogenic in humans. Naturally replication defective, wild-type AAV requires coinfection with a second virus (eg, adenovirus or herpes simplex virus) to provide helper functions necessary for replication.¹⁷ The 4.7 kb genome harbors four open reading frames that contain the rep and cap genes and coding sequences for two accessory proteins. ITR sequences flank the viral genome and are the only required sequence elements for vector replication and packaging; thus, the vector genome is remarkably simple and comprises an expression cassette with an ITR sequence on either end. The packaging limit of approximately 5 kb precludes the use of large cDNA and promoter sequences and complex multigenic cassettes. The viral vector is propagated by cotransfection of 293 cells with the vector plasmid and plasmid(s) harboring the AAV rep and cap genes and the adenoviral helper functions.¹⁵

More than 1,000 naturally occurring wild-type AAV variants have been identified, along with at least 12 natural serotypes, which preferentially bind to various cell surface glycans and secondary receptors.¹⁸ Following entry into the cell, the virus is trafficked through late endosomal and lysosomal compartments before being shuttled into the nucleus where the genome is unencapsidated. The single-stranded DNA genome of both wildtype AAV and the conventional rAAV vector requires second-strand synthesis before it can be recognized by the transcriptional apparatus of the infected cell.¹⁵ The ITR sequences facilitate intermolecular and intramolecular recombination to form concatenated double-stranded DNA circles, which provide stability and enable the vector genomes to be maintained as episomal elements. In mesenchymal tissues, where the resident cell populations are largely quiescent, the requirement for second-strand DNA synthesis can limit transgene expression. However, double-stranded self-complementary AAV genomes have been engineered that are fully functional at the time of infection.¹⁵,¹⁹ The transduction efficiency of self-complementary AAV vectors is substantially higher than that of single-stranded AAV vectors in mesenchymal cells and tissues, with rapid onset of expression—often within 24 hours of delivery.²⁰ However, the self-complementary modification reduces the vector genome by one-half (approximately 2.5 kb),¹⁹ which further restricts the size and composition of the expression cassette.

Wild-type AAV is prevalent in nature and, though nonpathogenic, childhood infections are common and typically induce a potent humoral immune response and life-long production of NAb specific to the capsid of the infecting serotype. Depending on location, anywhere from 30% to 70% of the human population have circulating NAb to one or multiple AAV variants. The ability to cross-package (or pseudotype) the vector genome in different capsids alters its tropism, providing the opportunity to increase the efficiency of gene transfer in specific tissues, as well as the potential to evade recognition by preexisting capsid-specific NAb. Engineering the AAV capsid to obtain designer vectors with enhanced properties for specific applications is currently an area of intense investigation.¹⁵

Vectors derived from members of the retrovirus family, γ-retrovirus and lentivirus, are spherical enveloped viruses approximately 100 nm in diameter whose genomes are composed of two copies of single-stranded RNA of the sense (+) strand.²¹ Nonenveloped viruses, for example, adenovirus and AAV, reproduce by lytic infection, releasing thousands of viral progeny in a burst that kills the host cell. Retroviruses, in contrast, are enveloped and reproduce by continuous budding from the surface of the infected cell. Though the genome is encased in a protective capsid, the outer envelope comprises lipid bilayer pilfered from the plasma membrane of the host cell as the virus is released from the surface.

Retroviruses can be classified as either simple or complex based on the composition and organization of their genome. γ-retroviruses, which harbor only three genes, are considered to be simple, whereas lentiviruses with nine overlapping coding regions are complex.⁹ The genome of the retrovirus family is flanked on each end by complex long terminal repeat (LTR) sequences containing strong promoter/enhancer elements that drive expression of the entire viral genome. Other cis-acting elements positioned downstream of the 5′ LTR include sequences for replication priming and a psi-sequence for encapsidation. All retroviral genomes contain three core genes: gag, pol, and env. For both types of retroviral vectors, the endogenous env gene, which codes for the surface glycoprotein used for cell-surface receptor binding, is often replaced with the coding sequence for the vesicular stomatitis virus glycoprotein.²² Pseudotyping the virus in this manner dramatically expands the tropism of the vector, enhances infectivity, and increases the stability of the vector particle during production. The pol gene codes for reverse transcriptase and integrase proteins that are carried within the viral envelope.²² The viral reverse transcriptase converts the RNA genome to double-stranded DNA, which the integrase inserts into the host genome as a provirus, preferentially targeting regions that are transcriptionally active.²³ Vector integration provides the potential for stable expression of the transgene, which can be amplified with subsequent cell divisions in vitro and in vivo. Retroviral vectors are highly infectious, elicit relatively weak
immune responses, and, depending on the biology of the target cell, can support long-term expression of therapeutic gene products.

The vector derived from the genome of Moloney murine leukemia virus was among the first γ-retroviral vector systems developed and the first used successfully in a clinical trial.²⁴,²⁵ As γ-retroviruses are unable to penetrate the nuclear envelope of an infected cell, they can only gain access to the host genome during mitosis, when the nuclear envelope is disassembled.²⁴ This limits their host range to cells that are actively dividing and confines their use to ex vivo applications. In early vectors, the gag, pol, and env genes were replaced by the cDNA of interest, and expression of the transgene was driven by the endogenous promoter/enhancer of the 5′ LTR.²⁵ The viral vector was propagated in complementing cell lines where the gag, pol, and env genes are stably expressed and supplied in trans. In later self-inactivating vector generations, a more conventional expression cassette is used.

γ-retroviral vectors preferentially integrate near the transcription start sites of active genes, with a particular affinity for proto-oncogenes, which raises the potential for insertional mutagenesis and oncogenic activation of the infected cell.²³ In one of the earliest gene therapy clinical trials, a γ-retroviral vector was used to modify hematopoietic stem cells to treat X-linked severe combined immunodeficiency. Though the treatment was successful and stably reversed disease in 9 of 10 male infants, within a few years of treatment, four of the boys went on to develop T-cell acute lymphocytic leukemia. DNA sequence analysis of the leukemic cells revealed insertional mutagenesis of the vector genome leading to aberrant activation of proto-oncogenes LMO2 and BMI1.²⁶,²⁷ Based on this and similar adverse events in other trials using γ-retroviral vectors, modifications have been made to the genome, including deletion/inactivation of the LTR enhancers to create self-inactivating vectors incapable of generating replication competent retrovirus with reduced risk of vector-induced oncogenesis.²⁸ As one of the earliest viral vector systems, γ-retroviral vectors have been used in numerous clinical trials, including gene therapy for rheumatoid arthritis,³ some of which are currently ongoing.

Distinct from γ-retrovirus, lentiviral vectors are more complex and use active transport mechanisms that enable the viral assembly to traverse the pores of the nuclear envelope to access the host chromosomes. This endows lentiviral vectors with the ability to transduce both dividing and nondividing cells with similarly high efficiency and expands the utility of lentiviral vectors to both ex vivo and in vivo applications.²⁹ Although lentiviruses also preferentially integrate in genes that are transcriptionally active, they target the gene body rather than the start site, which reduces the likelihood of insertional oncogenesis.²³ Because of their increased versatility and reduced genotoxicity, recombinant lentivirus has emerged as the preferred retroviral system for gene delivery, especially for clinical ex vivo applications.

Historically, the motivation for pursuing nonviral gene transfer has been based on several theoretical advantages over viral-based systems³⁰ specifically (1) ease of manipulation, (2) lower production costs, (3) lack of immune response, (4) increased safety, and (5) large payload capacity. With advancements in viral vector technology over the past several years, the extent to which these benefits still exist is highly questionable. Although exogenous nucleic acids do not provoke adaptive immune responses, they are potent inducers of innate immune pathways³¹ and are characteristically inflammatory following delivery in vivo. Regarding safety, available data from over 200 clinical trials involving more than 3,000 patients indicate that AAV gene therapy when administered at moderate doses is safe, well-tolerated, and efficacious: a profile compatible with use in common, non-life-threatening disorders.³² Finally, although it is possible to create plasmid DNA constructs that are large and complex, transfection efficiency is inversely proportional to size of the DNA, such that plasmid uptake and expression drops precipitously with constructs over 3 kb in length and is most efficient with minicircles of 650 bp or less.³³,³⁴ Although nonviral vectors are perhaps easier to manipulate and somewhat less costly to produce than recombinant virus, these advantages are irrelevant considering the functional limitations of nonviral gene transfer in vivo. Central among these is the inefficient delivery and brief duration of transgene expression. Plasmid DNA must be taken into the cell either through endocytosis or pinocytosis then passively find its way to the nucleus and traverse the nuclear envelope before it can be transcribed. Much like γ-retrovirus, transfection efficiency is far higher in mitotic cells, which strongly favors transfection in vitro. Technologies such as electroporation, hydrodynamic injection, and ultrasonography, which perturb or temporarily disrupt the cellular membranes, can enhance uptake and gene expression in vivo. In larger tissues, these technologies are difficult to administer effectively and, in some cases, provoke significant tissue damage. Regardless of the transfection reagent or method of delivery, gene expression in vivo is modest and transient. These limitations effectively restrict nonviral applications to vaccine development and ex vivo procedures.

In 1999, the concept of a gene-activated matrix was introduced, whereby plasmid DNA containing the cDNA for human parathyroid hormone 1 to 34 (teriparatide)
was linked to a scaffold and subsequently implanted into critical-size bone defects in a canine model.³⁵ It was theorized that local progenitor cells would infiltrate the matrix, acquire the plasmid, and express the transgene to stimulate local osteogenesis and enhance repair of critical size defects. Unfortunately, the early reports of efficacy have not been widely reproducible. Attaching or incorporating exogenous genetic material into a scaffold reduces its bioavailability, such that the already inefficient process of DNA uptake becomes even more so. The problem with nonviral gene transfer is not a matter of controlled release but inefficient uptake. Newer iterations of the gene-activated matrix concept include viral vectors.