Contributions of Gene Marking to Cell and Gene Therapies

Abstract

The first human genetic modification studies used replication-incompetent integrating vector vectors to introduce marker genes into T lymphocytes and subsequently into hematopoietic stem cells. Such studies have provided numerous insights into the biology of hematopoiesis and immune reconstitution and contributed to clinical development of gene and cell therapies. Tracking of hematopoietic reconstitution and analysis of the origin of residual malignant disease after hematopoietic transplantation has been possible via gene marking. Introduction of selectable marker genes has enabled preselection of specific T-cell populations for tumor and viral immunotherapy and reduced the threat of graft-versus-host disease, improving the survival of patients after allogeneic marrow transplantation. Marking studies in humans, murine xenografts, and large animals have helped optimize conditions for gene transfer into CD34⁺ hematopoietic progenitors, contributing to the achievement of gene transfer efficiencies sufficient for clinical benefit in several serious genetic diseases such as X-linked severe combined immunodeficiency and adrenoleukodystropy. When adverse events linked to insertional mutagenesis arose in clinical gene therapy trials for inherited immunodeficiencies, additional animal studies using gene-marking vectors have greatly increased our understanding of genotoxicity. The knowledge gained from these studies is being translated into new vector designs and clinical protocols, which we hope will continue to improve the efficiency, effectiveness and safety of these promising therapeutic approaches.

Preclinical studies using gene marking have enabled researchers to optimize ex vivo manipulation and transduction of hematopoietic stem cells. In this review, Drs. Barese and Dunbar focus on the contributions of gene-marking approaches to the fields of hematopoietic stem-cell transplantation, T-cell immunotherapy, and gene therapy. They also summarize recent insights into the risks of vector integration and genotoxicity.

Introduction

One of the greatest success stories of gene transfer is the inclusion of genes or truncated gene sequences in hematopoietic stem cells or mature blood lineages to “mark” a target cell population and track their contributions to hematopoiesis or immune function over time. Such genes or gene sequences, most frequently delivered via retroviral vectors, are not inherently therapeutic by themselves, but rather integrate into the cells' genome without disturbing normal cellular functions. The persistence of these gene sequences in the progeny of transduced cells has wide-ranging applications. Since the 1990s, gene marking has been used in both experimental and clinical situations. Marking of T cells or hematopoietic progenitor cells has provided invaluable information regarding cell fate and therapeutic potential. Analyses of marker genes in tumor cells revealed new information regarding the origin of residual disease after autologous transplantation. Preclinical studies using gene marking enabled researchers to optimize ex vivo manipulation and transduction of hematopoietic stem cells (HSCs), increasing the efficiency of gene transfer and eventually resulting in clinically efficacious gene therapy in patients with genetic immunodeficiencies and storage disorders. This review will focus on how these gene-marking approaches have contributed to the fields of hematopoietic stem cell transplantation (HSCT), T-cell immunotherapy, and gene therapy. Finally, we will summarize recent insights into the risks of vector integration genotoxicity, using animal models and marking vectors. These results have informed the search for safer methods of gene delivery to HSCs and have important implications for future gene therapy applications.

Available Marker Genes

Because marker genes are designed to facilitate the identification and tracking of gene-marked cells in vitro and in vivo, ideally they should be safe, nontoxic, nonimmunogenic, and easily monitored and they should facilitate positive immunoselection if purification of marked cells is desirable. Genetically marked cells can be followed either by analysis for expression of a transgene product or via marker detection of vector DNA sequences. Depending on the experimental or clinical situation, there are a number of factors to consider in selecting the optimal gene-marking approach. Table 1 summarizes the major approaches studied to date.

Table 1.

Summary of Genetic Marking Approaches

Gene marker	Advantages	Disadvantages
NEO	No need of high technology methods for detection (Dick et al., 1985)	Selection by G418 resistance is time consuming and time in culture may change cell characteristics, which can have a negative effect should selection by this method be used.
	Most data indicates no impact on cell function (Wu et al., 1998)
EGFP (and FPs)	Rapid detection and ability to select by fluorescence-activated cell and/or fluorescent microscopy (Hawley et al., 2001; Persons et al., 1997, 1998)	Possible immunogenicity (Morris et al., 2004)High level of some FPs toxic (Hanazono et al., 1997)
	Transgenic animals available	Cell death by free radicals release after prolonged in vitro excitation
		DNA methylation effects (Liu et al., 1999)
ΔLNGFR	Human protein, nonimmunogenic.	AML in murine studies using ΔNGFR- (Li et al., 2002)
	Detectable by flow cytometry.
	Endogenous NGFR not present on hematopoietic cells
	Commercially available monoclonal antibodies for ex vivo immunoselection and detection
	Reported safe in human clinical trials targeted to T cells (Bonini et al., 2003)
ΔCD19	Human protein	Limited clinical experience
	Not immunogenic
	Detectable by flow cytometry.
	Commercially available monoclonal antibodies for ex vivo immunoselection and detection
tCD34	Human protein	Impact on homing and differentiation of transduced cells (Fehse et al., 2000)
	Detectable by flow cytometry
	Commercially available monoclonal antibodies for ex vivo immunoselection and detection
Drug-selectable FPs	Fluorescent-mediated detection and drug-mediated selection (Weber et al., 2010)	Limited experience

EGFP, enhanced green fluorescent protein; FPs, fluorescent proteins; AML, acute myeloid leukemia; NGFR, nerve growth factor receptor.

Antibiotic resistance genes such as neo were used extensively for the first decade of gene marking and gene therapy studies. The neo marker gene had practical advantages for vector production, served as an unequivocal DNA tag for transduced cells, and appeared minimally toxic and immunogenic. However, fluorescent proteins such as green fluorescent protein (GFP) rapidly supplanted neo for most marking applications following their discovery, given the ease of analysis of GFP-marked cell populations via flow cytometry, microscopy, or even whole-animal imaging. Cell surface molecules, such as truncated forms of the nerve growth factor receptor (ΔNGFR) or CD34, have also been developed, and allow efficient and straightforward immunoselection of transduced cells prior to administration. Recently, a series of marking lentiviral vectors containing drug-selectable fluorescent proteins (LeGo) became available and have great potential for multicolor marking and tracking studies (Weber et al., 2010) Transduction of cells with marking vectors containing independent DNA “bar codes” is another intriguing approach for sophisticated clonal tracking of transduced cells (Gerrits et al., 2010).

Marking T Cells

The high incidence of resistance to conventional cancer chemotherapeutics stimulated the development of T-cell immunotherapies. Interest in knowing whether tumor-reactive T cells homed in on and persisted in tumors stimulated the early application of gene-marking techniques in immunotherapy protocols. A summary of the pioneering studies using gene-marked T cells and the lessons learned from these clinical trials follows.

In 1989, Rosenberg et al. conducted the first study using genetically modified cells in a human clinical trial, using a replication-incompetent integrating retrovirus carrying the neo marker gene to transduce autologous tumor-infiltrating lymphocytes (TIL) purified from melanoma patients (Rosenberg et al., 1990). The marker gene neo was introduced into TIL from 10 patients with metastatic melanoma, and selection of transduced cells was performed by continuous culture in G418 following transduction, prior to re-infusion of the TIL into the patients. Peripheral blood and tumor biopsy specimens analyzed for neo gene sequences by PCR revealed the presence of transduced TIL in tumor tissues for up to 64 days post-infusion, with less consistent detection in the circulation, indicating that at least some of the gene-modified TIL homed to the tumor and persisted long term. These early studies demonstrated the potential for adoptive cell therapies and suggested that infusion of genetically modified T cells was safe and feasible. Lessons learned from these early marked TIL studies have informed all subsequent studies from Rosenberg and colleagues, culminating in more encouraging clinical results when immunoablation was given prior to infusion of lymphocytes genetically engineered to target tumor cells (Morgan et al., 2006; Rosenberg and Dudley, 2009).

Donor-derived T cells have been used extensively in allogeneic stem cell transplantation to boost antitumor and antiviral immunity. Genetic marking of these cells has provided valuable information about their trafficking, homing, and efficacy. For example, Epstein Barr virus (EBV)-associated post transplant lymphoproliferative disorder (PTLD) occurs in up to 30% of patients undergoing T-cell–depleted stem cell transplantation. Infusion of donor-derived EBV-specific cytotoxic lymphocytes (CTLs) can successfully treat this complication, but the life-span and distribution of infused CTLs in vivo were unknown (Rooney et al., 1995; Heslop and Rooney, 1997). In a series of studies from Heslop, Brenner, and Rooney's groups, EBV-specific donor CTLs were transduced with a neo marking vector and infused either preemptively for patients at very high risk of PTLD or as treatment in patients with established PTLD or Hodgkin disease linked to EBV (Heslop and Rooney, 1997; Bollard et al., 2001). The most recent update reported 100% effectiveness of donor EBV-specific CTL infusions in preventing PTLD (101 patients), as well as a remission rate of higher than 80% when EBV-CTLs were used as therapy in 13 patients that had already developed EBV-related tumors following either hematopoietic or solid organ transplantation (Heslop et al., 2010). Twenty-six patients reported in the long-term follow-up analysis received donor EBV-CTLs cells marked with the neo transgene, and these patients had similar clinical outcomes to those receiving unmarked CTLs, suggesting full functionality of the marked cells, despite undergoing transduction procedures and having integrated vector within their genome. Marked CTLs could be detected for up to 9 years in the transplant recipients, without evidence for clonal outgrowth or other adverse effects resulting from the presence of vector-containing cells. This series of careful clinical trials were a real tour de force that used gene marking to give further insights into the behavior of EBV-specific CTLs in vivo and provided invaluable information for further development of EBV-CTLs as a therapeutic option for the treatment or prevention of life-threatening EBV-related lymphoproliferative disease in transplant patients.

The efficacy of T-cell therapies for the treatment of cancer has been demonstrated by the complete and sustainable clinical tumor responses achieved in patients undergoing allogeneic HSCT (allo-HSCT) followed by infusions of donor T cells. Although donor T cells have been shown to be effective as both graft-versus-leukemia (GvL) and graft-versus-infection (GvI) therapies, the high risk of developing graft-versus-host disease (GvHD) still represents a major obstacle to the utilization of donor T-cell infusions after allo-HSCT.

To overcome this life-threatening complication, Bordignon et al. (1995) transduced donor T cells with a “suicide gene” to render them specifically sensitive to drug-induced killing, should deleterious severe GvHD occur following T-cell administration (Bonini et al., 1997). The retroviral vector cleverly encoded both a suicide gene and a selectable cell surface marker gene, allowing in vitro selection of transduced cells before infusion into the patients. The marker was a truncated form of ΔNGFR. Endogenous NGFR is not present on hematopoietic cells, and the ΔNGFR form does not transduce any intracellular signals. It has several advantages, including the opportunity for positive selection via commercial immunoselection methods, selection without toxicity, in contrast to cells selected by growth in G418, and most importantly, transgene expression does not interfere with T-cell function (Bonini et al., 2007). After in vitro selection for ΔNGFR-positive cells, the pure population of transduced donor lymphocytes can be infused into patients during or after allo-HSCT. Bordignon's group used the Herpes simplex virus thymidine kinase gene (HSVtk) as a suicide gene, allowing in vivo specific killing of vector-containing cells via administration the relatively nontoxic drug ganciclovir. Importantly, this strategy has proved to be effective in treating or to ameliorate GvHD in several Phase I/II clinical trials (Bonini et al., 1997; Tiberghien et al., 1997; Ciceri et al., 2007; Qasim et al., 2007).

Concerns regarding the ΔNGFR marker gene arose when Li et al. (2002) reported that a mouse developed acute myeloid leukemia (AML) after receiving bone marrow transduced with a retroviral vector containing ΔNGFR, which correlated with an integration of the provirus into the murine Evi1 proto-oncogene. Some contribution of the ΔNGFR marker itself to the process, due to some residual potential signaling activity, was hypothesized. However, with time and further investigation it became apparent that this was a rare outcome because no malignant transformations were observed in other studies (Bonini et al., 2003) that used retro- or lentivirus vectors expressing ΔNGFR in transduction of human lymphocytes or murine HSCs. Conversely it became apparent that any integrating gammaretroviral vector, regardless of the transgene, could activate Evi1 and result in adverse outcomes, as described by Metais and Dunbar (2008). Several clinical trials in aggregate including 110 patients, targeting either T lymphocytes or HSCs, with vectors containing ΔNGFR, have been conducted to date, with no reported major adverse effects or development of malignancies (Recchia et al., 2006). Other groups investigated alternative positive selectable marker gene–suicide gene combinations, including a truncated CD34 protein cloned as a fusion with the HSVtk gene in either retro- (Fehse et al., 2002) or lentiviral vectors (Qasim, 2007), allowing rapid positive selection of transduced human T lymphocytes via standard CD34 cell selection devices. In a murine model of GvHD, the fusion protein tCD34/HSVtk was suitable for selection of transduced cells and also for monitoring the gene-marking level in peripheral blood (Rettig et al., 2004). Although the use of a fusion gene comprising the selectable marker and the gene of interest offers the advantage of equimolar production of a protein responsible for both functions, unexpected consequences can occur. In a recent paper Bennour et al. (2008) reported an unusual loss of functionality of the fusion protein tCD34/cHSVtk arose from deletions in HSVtk that yielded ganciclovir-resistant transduced cells but left intact expression of the tCD34⁺ surface marker.

A truncated, nonsignaling CD19 (ΔCD19) surface marker was also tested recently for in vitro selection and enrichment of donor-positive T cells transduced with the human inducible caspase 9 (iCasp-9) suicide gene (Tey et al., 2007). The CD19 marker allows rapid in vivo tracing of positive T cells after infusion; Di Stasia et al. (2010) recently presented the preliminary results of a Phase I clinical trial (CASPALLO). The authors reported that two patients in the trial developed grade I/II GvHD, and that administration of the specific chemical inducer of dimerization (AP1903) that activates iCasp9 and results in apoptosis was effective in eliminating more than 90% of alloreactive T cells after allo-HSCT, completely resolving the skin and liver manifestations of acute GvHD within the first 24 hr. This approach has a number of potential advantages compared to the HSVtk suicide gene system. The iCasp-9 suicide gene product is unlikely to be immunogenic, since it consists almost wholly of an endogenous human gene product, in contrast to HSVtk, a completely foreign viral protein. Inclusion of HSVtk also precludes use of ganciclovir to treat CMV post-transplantation, without ablation of donor T cells, and thus undesirable impairment of antitumor and anti-infectious immunity, in the absence of GvHD.

Marking HSC/Progenitors Grafts

Source of relapse post autologous transplantation

In some of the initial human clinical trials, gene-marking vectors were used in an attempt to investigate the source of relapse in patients undergoing high-dose chemotherapy and autologous stem cell rescue for a number of malignancies. Brenner and co-workers carried out a series of studies using retroviral vectors carrying the neo gene to transduce autologous bone marrow from patients with AML and neuroblastoma (Rill et al., 1992; Brenner et al., 1993a). Tracking the neo transgene in malignant cells by PCR or colony forming unit (CFU) selection with G418, they found that some pediatric patients with AML relapsed with gene-marked blasts (Brenner et al., 1993b). In similar gene-marking studies, adults with chronic myeloid leukemia (Deisseroth et al., 1994) who relapsed after receiving autologous HSCT tested positive for the marker gene in bcr-abl positive marrow CFU. These pioneering studies provided absolutely definitive evidence, perhaps not surprisingly, that at least one source of relapse following autologous transplantation can be the graft itself.

However, other trials involving cancer patients did not show evidence of marked cells after transplantation with autologous neo-transduced grafts. A considerable number of adult patients with multiple myeloma, breast cancer (Alici et al., 2007; Dunbar et al., 1995; Emmons et al., 1997), follicular lymphoma (Bachier et al., 1999), AML (Cornetta et al., 1992), and also pediatric patients with AML (Heslop et al., 1996) whose underlying malignancy relapsed did not demonstrate marked tumor at the time of relapse. However, lack of detectable marked relapse is much less informative than the converse, since explanations for relapse without detectable marked cells includes lack of transduction of tumor cells ex vivo, due to specific features of the tumor cells, versus lack of tumor cells contaminating the graft, and overall poor transduction and engraftment due to suboptimal conditions for ex vivo manipulation of the cells.

Genetic marking/tracking studies

As previously reviewed, retroviral transduction conditions worked out using small animal and in vitro models in the early 1990s did not translate into successful stable engraftment with genetically modified cells in early human clinical trials (Donahue and Dunbar, 2001; Hu and Dunbar, 2002). Preliminary in vivo rodent studies showed potentially clinically relevant levels of engraftment (Ferrari et al., 1991) and evidence of an intrinsic advantage on survival and growth of the transduced cells, which aided engraftment into the host without the need of myeloablation (Ferrari et al., 1992). So, in the early-to-mid 1990s, several groups of European and American investigators performed the first gene therapy clinical trials on humans with adenosine deaminase deficient severe combined immunodeficiency (ADA-SCID), using T cells (Blaese et al., 1995), bone marrow (Bordignon et al., 1995; Hoogerbrugge et al., 1996), or cord blood CD34⁺ cells as targets of transduction (Kohn et al., 1995). Although one of the studies demonstrated a selective survival advantage of genetically modified T cells after gene transfer, detectable gene-marked lymphocytes in peripheral blood were 0.1–0.01% (Kohn et al., 1995), far below clinical benefit. Thus, the patients continued polyethylene glycol-modified adenosine deaminase (PEG-ADA) enzyme replacement treatment. Similar results were obtained in initial attempts to correct chronic granulomatous disease and Gaucher disease, in which the therapeutic gene does not confer a selective advantage for either survival or growth to the gene-modified cells. In a Phase I clinical trial for X-linked chronic granulomatous disease (X-CGD), the investigators incorporated flt3-ligand and granulocyte-macrophage colony-stimulating factor and utilized fibronectin in a 4-day in vitro transduction protocol. The patients underwent multiple cycles of mobilization and infusion of transduced cells without preconditioning. Nine months after transplantation, the levels of gene-corrected neutrophils were 0.06%–0.2%, which is below the desired 5%–10% required for therapeutic effects (Malech et al., 1997). In patients with Gaucher disease, there was low gene marking, little or no expression, and no disease correction following transplantation of retrovirally transduced autologous HPCs (Dunbar et al., 1998).

The gene-marking trials carried out in patients undergoing autologous transplantation for cancer had dual purposes, first to investigate the source of relapse, as detailed earlier, and second to explore gene transfer efficiency and persistence in humans. Even with full myeloablation, patients had levels of marked circulating blood cells significantly less than 1% long-term, if marking could be detected at all (Cornetta et al., 1992; Brenner et al., 1993a; Dunbar et al., 1995; Emmons et al., 1997). However, these studies did unequivocally demonstrate the autologous mobilized peripheral blood CD34⁺ cells contained long-term repopulating stem cell activity, since very long-term marking from these transduced cells was documented.

Despite the discouraging results of these first attempts, they highlighted the need to improve methods for more effective retroviral gene transfer into true long term repopulating HSCs. These early trials also underscored that some sort of conditioning was very likely necessary to improve engraftment of gene-modified cells. A number of preclinical studies in the late 1990s and early 2000s, using neo or GFP marker genes, provided invaluable insights on culture conditions necessary for improving gene transfer into repopulating cells of primates and humans. The use of optimized cytokine cocktails, such as stem cell factor, Flt-3 ligand, and thrombopoietin, together with shortening the ex vivo culturing time, enabled increased transduction of more primitive HSCs endowed with long-term repopulating ability (Dunbar et al., 1996; Tisdale et al., 1998). Coating culture flasks with a specific fragment of human recombinant fibronectin to co-localize retroviral particles and CD34⁺ cells produced demonstrable improvements in the efficiency of transduction into rhesus and human HSCs. Gene-marking levels with potential clinical utility were observed in studies using these optimized conditions (Hanenberg et al., 1996; Dao et al., 1998; Kiem et al., 1998). Additionally, gene marker studies in large-animal models helped address the effectiveness of submyeloablative conditioning (320–500 cGy radiation) prior to HSCT. Persistence of as much as 15% of gene-marked leukocytes was observed in some animals for as long as 33 weeks after transplantation, suggesting that the gene correction may be achievable while reducing transplant-related toxicities (Huhn et al., 1999; Rosenzweig et al., 1999). These and other preclinical studies set the stage for the second generation of human gene therapy trials for inherited immunodeficiencies (Cavazzana-Calvo et al., 2000; Aiuti et al., 2002, 2009; Gaspar et al., 2004; Thrasher et al., 2005; Gaspar et al., 2006; Ott et al., 2006; Boztug et al., 2010), which have been discussed extensively in the literature.

Gene-marking studies in large animals have also been of great interest to study the properties of hematopoietic stem and progenitor cells in vivo, offering insights into clinically relevant processes that are difficult to explore directly in any other way. For instance, clonal analysis of granulocyte insertions has resulted in estimates for the frequency and cycling characteristics of primate long-term repopulating cells of the bone marrow. These estimates match those obtained from measurement of telomere attrition (Shepherd et al., 2007). They have also provided direct evidence for short- and long-term repopulating clones, analogous to the distinct populations identified in murine studies (Schmidt et al., 2002). Furthermore, the behavior of individual stem and progenitor cell clones could be tracked following nonmyeloablative conditioning or chronic cytokine therapy (Kuramoto et al., 2004; Laukkanen et al., 2005).

Drug resistance genes for in vivo selection

Inclusion of drug resistance genes as in vivo selectable markers can increase engraftment of genetically modified HSCs following exposure to drugs that are toxic to unmodified cells in the graft. Such in vivo selectable markers include the genes for multi-drug resistance (MDR-1), dihydrofolate reductase (DHFR), and O6-methylguanine-DNA methyltransferase (MGMT). Preclinical studies in mice, and then more importantly in dogs and nonhuman primates that underwent HSCT with MGMTP140K gene-modified CD34⁺ cells following submyeloablative conditioning have shown impressive enhancement of in vivo gene-marking levels following treatment with O6-benzylguanine and 1,3-bis-(2-chloroethyl)-1-nitrosourea or tremozolomide (Gerull et al., 2007; Beard et al., 2010), along with eventual protection from the myelosuppressive effects of repeated cycles of these chemotherapeutic agents. However, two other groups did not find this approach efficient or nontoxic in an alternative primate model (Larochelle et al., 2009). Based in the somewhat promising preclinical findings, phase I/II clinical trials in glioblastoma are ongoing, with the goal of allowing nontoxic 1,3-bis-(2-chloroethyl)-1-nitrosourea dose escalation.

Insights into genotoxicity from gene-marking trials

Studies performed prior to the availability of the human and murine genome sequences and without high-throughput methodologies suggested that retroviral integration was random, and that thus insertional mutagenesis was likely to be an exceedingly rare event in the absence of replication-competent vectors and required activation of multiple genes (Baum et al., 2003). However, the unexpected occurrence of insertional oncogenesis in human clinical trials for primary immunodeficiencies (Hacein-Bey-Abina et al., 2003; Ott et al., 2006; Howe et al., 2008) and in animal models (Li et al., 2002; Kustikova et al., 2005; Modlich et al., 2005) firmly indicated the need to re-examine and modify gene transfer vectors to avoid this serious adverse consequence of gene therapy using integrating vectors. While HIV vectors integrate within genes, MLV integrates within 5 kb of transcriptional start sites, in particular in actively transcribed genes. The integration preferences of gene transfer vectors have been discussed in several recent review papers (Nienhuis et al., 2006; Baum, 2007). The targeted coding regions likely depend on the genes that are highly expressed in the transduced cell and murine serial transplantation models using retrovirus encoding marker genes showed viral insertions in or near genes committed to self-renewal or survival of stem cells (Kustikova et al., 2005). In addition, in the rhesus macaques transplantation studies using MLV vectors containing solely marker genes, one animal demonstrated clonal dominance with integrations found in an anti-apoptotic gene (BCLA2) shortly after transplantation. Importantly, the clone remained quiescent for 5 years, then re-emerged, with progression to an overt myeloid leukemia (Seggewiss et al., 2006). These studies indicate the importance of constantly monitoring viral integration profiles and the value of in vitro assays (Modlich et al., 2006) as well as in vivo tumor-prone animal models (Montini et al., 2006), which may be of considerable utility in designing approaches to prevent oncogenesis in future clinical studies.

To date, no adverse effects have been reported when T cells were the targets for gene transfer. Interestingly, the evaluation of vector insertion sites in vitro and after in vivo infusion of herpes tk-transduced T cells in the clinical allogeneic transplantation trials demonstrated that integrations mostly landed close to genes that were prone to down-regulate T-cell functions. However, this did not lead to abnormal immune responses in the transplanted patients (Recchia et al., 2006). The lack of malignant transformation was also demonstrated in a murine study in which T cells and HSCs were genetically modified with gammaretrovirus vectors that expressed T-cell oncogenes. While the enforced expression of LMO2, TCL1, or TrkA into HSCs resulted in T-cell leukemia/lymphoma in the transplanted animals, the infusion of similar genetically modified T cells was well tolerated (Newrzela et al., 2008).

Further gene therapy studies in which the occurrence of insertional oncogenesis as adverse effects was not observed include the ADA-SCID trials (Cassani et al., 2009). In contrast to the SCID-X1 trial, in all the ADA-SCID patients treated with gene transfer, the long-term analysis of integration sites revealed a polyclonal pattern along with a lack of in vivo skewing for risky insertions. The fact that the therapeutic gene confers only selective survival to the transduced HSCs and does not participate in proliferation signaling within the cell can account for a lower risk of participating in oncogene activation (Kohn, 2008; Aiuti et al., 2009).

Consequences of Genotoxicity Results

The combination of the adverse events in X-SCID and then the X-CGD clinical trials targeting HSCs, in addition to the clear evidence for nonrandom integration patterns for both retroviruses and lentiviruses in cell lines and primary cells stimulated intense further investigation into integration profiles for various vector systems and new vector designs predicted to reduce the likelihood of insertional activation of proto-oncogenes. Mitchell et al. (2004) showed that avian sarcoma leukosis virus (ASLV) has little or no bias toward integration into active genes and does not favor integration near transcription start sites. Hu et al. (2007) demonstrated the feasibility of obtaining efficient long-term engraftment in rhesus macaques CD34⁺ cells transduced with an ASLV-derived RCAS (Replication Competent ALV long-terminal repeat [LTR] with a Splice Acceptor) vector and GFP as a gene marker. Furthermore, the authors analyzed approximately 300 insertions and showed that the vector integrations did not favor gene-rich regions, transcription start sites, or CpG islands (Hu et al., 2008), demonstrating that ASLV may be a safer candidate for gene transfer into human HSCs compared with MLV. The ASLV viral enhancer is only active in avian cells, making constitutive activation of adjacent proto-oncogenes less likely following insertion of ASLV vectors into mammalian genomes. Self-inactivating (SIN) vectors have the strong viral promoter/enhancer deleted from the proviral LTR, and instead transgene expression is driven from an internal tissue-specific or constitutive promoter. This design has been employed almost universally with lentiviral vectors, and more recently there has been a resurgence of interest in SIN MLV vectors. SIN vector design has been shown to greatly reduce the risk of insertional leukemogenesis (Gabriel et al., 2009) in the murine in vivo model. MLV SIN gammaretrovirus vector with neo and cellular phosphoglycerate kinase as an internal promoter was evaluated in an in vitro immortalization assay using murine HSCs and demonstrated lower level of myeloid immortalization than a non-SIN vector (Bosticardo et al., 2009). Certainly, MLV SIN vectors are not completely devoid of insertional oncogenesis risk, and SIN vectors completely lacking enhancer activity in the LTR or internally may be required.

Lentiviral vectors derived from the human immunodeficiency virus (HIV) or simian immunodeficiency virus (SIV) are also attractive tools for improving safety in gene therapy. Their lack of potent enhancer sequences in the modified LTRs reduces the risk of activating neighboring genes. Recent studies demonstrate that vectors derived from lentiviruses will be useful for the correction of genetic blood diseases (Hayakawa et al., 2009). The inclusion of physiological promoters in SIN vectors, such as enlongation factor-1α, also significantly reduces the risk of insertional transformation when evaluated in recently established serial replating assays that detect insertional transformation (Zychlinski et al., 2008). Hanawa et al. (2004) recently reported high gene-marking levels and long-term polyclonal reconstitution of hematopoiesis in rhesus macaques after transplantation of mobilized CD34⁺ cells transduced with SIV vectors and GFP as a gene marker. Follow-up of these animals now for almost 5 years has shown no common integration sites in known proto-oncogenes, highly polyclonal hematopoiesis, and very stable levels of vector-containing cells and transgene expression (Kim et al., 2009). Several HSC-targeted therapy trials using HIV-derived lentiviral vectors were recently reported or are in progress. Two patients with adrenoleukodystrophy who lacked a histocompatible bone marrow donor received infusions of HSCs transduced with the therapeutic ALD gene following myeloablative conditioning. Importantly, at 36 months post-transplantation both patients had highly polyclonal reconstitution of hematopoiesis and normal levels of ALD protein, which appeared to retard the progressive cerebral demyelization process (Cartier et al., 2009). Since the ALD protein is very unlikely to impact the behavior of HSCs and progenitor cells, giving neither a positive nor a negative growth advantage, these patients, in addition to the nonhuman primates discussed earlier, will provide very valuable information on the dynamics of in vivo hematopoiesis over time, besides the very encouraging clinical outcomes regarding the underlying disease. The recent report of a clonal expansion from integration of lentivirus vector in the DNA-binding protein HMGA2 (high mobility group AT-hook 2) gene in a patient treated with gene therapy for thalassemia imposing, once more, further caution (Cavazzana-Calvo et al.).

Taken together, these data reflect a mixed message: gammaretrovirus vectors in which the transgene is not a growth/survival-promoting factor may have a relatively low but still appreciable risk for inducing clonal dominance and eventual leukemia in patients treated with HSC-directed gene therapy. As soon as the first two children with X-SCID developed leukemia linked to insertional mutagenesis, human gene-marking trials became unacceptable almost overnight, with the realization that risk was not minimal and could not be justified, given that there was no chance for clinical benefit. The further development and validation of predictive animal models including human–murine xenografts, dogs, and nonhuman primates obviated the need to carry out additional nontherapeutic marking studies in patients. Instead, carefully chosen clinical trials are being designed to further test putatively safer vectors in situations with some hope for clinical benefit.

Nonviral Methods of Gene Transfer

Physical-chemical methods have been used with great success to enable, enhance, or localize gene delivery. These nonviral methods of gene transfer into primary human T cells (Tahvanainen et al., 2006; Magg et al., 2009) and HSCs (Sumiyoshi et al., 2009) are currently being developed and tested in a growing number of in vitro studies using gene markers. Some of the methods have shown good transfection efficiency and expression levels. However, most of these methods result in significant loss of viability of primary cells, particularly if functional assays are performed, such as proliferation in vitro or engraftment in vivo. Using optimized electroporation for delivering a dual-expression vector into the nucleus Magg et al. (2009) achieved successful stable transfection of primary human T cells expressing the gene of interest and ΔNGFR as a gene marker. This enabled rapid and efficient positive selection by cell sorting. Another group (Hackett et al., 2010) has recently combined electroporation with the Sleeping Beauty (SB) transposase/transposon DNA plasmid system to introduce a CD19-specific chimeric antigen receptor to redirect the specificity of human T cells. By improving the efficiency of electroporation and integration by iterative adaptations of the SB system, this potentially safer nonviral approach is moving towards clinical trials. Two recent reports document the potential for the SB system even in delivering genes to CD34⁺ HSCs (Sumiyoshi et al., 2009; Xue et al., 2009). Although the efficiency of gene transfer is below that achieved with viral vectors, the potential for these approaches warrant further studies to optimize and evaluate the safety of this novel strategy in the future.

Conclusions

Gene-marking studies constitute the earliest application of gene transfer to HSCs and T cells. Since the 1990s, the contribution of gene-marking studies to experimental and clinical hematology has been remarkable, providing invaluable translational data to the fields of HSCT, T-cell immunotherapy, and gene therapy. Pioneering marking studies in large animals greatly aided in the optimization of conditions for gene transfer to hematopoietic stem and progenitor cells, furthered the expression of therapeutic genes to clinically relevant levels, and provided the foundation for the second generation of human gene therapy trials. More recently, the knowledge gained from in vitro and in vivo marking studies designed to assess the risks of the genotoxicity phenomena associated to gene transfer have been translated to the development of safer viral vectors for gene delivery into HSCs. In addition, nonviral methods of gene delivery are currently being evaluated in studies using marker genes. Taken together, the contribution of gene marking to the continued development of these cell and gene therapies provides considerable hope for improving the effectiveness and safety of these applications for malignant and nonmalignant hematologic diseases in the near future.