Allogeneic transplantation: opportunities and limitations
Sibling or matched unrelated allogeneic transplants of adult or cord blood hematopoietic stem cells (HSC) remain the only curative therapy for patients with hereditary disorders including hemoglobinopathies, immune deficiency, and inborn errors of metabolism. However, the toxicity of myeloablative preparative regimens, risks of graft vs. host disease (GVHD), infectious complications of immunosuppression, and limitations in the availability of suitable related or unrelated donors limit the number of patients to whom allogeneic bone marrow (BM) transplants can be offered. Particularly in neonatal models, myeloablative transplantation approaches have been associated not only with higher morbidity and mortality, but also with subsequent abnormal development and growth (1
Approaches to decrease toxicity associated with HSC transplantation include reducing the intensity of preparatory regimens and T cell depletion to decrease GVHD. Partially mis-matched donors such as haploidentical parents and siblings have been used to increase the available donor pool. Indeed, the successful use of haplo-identical donors for the treatment of severe combined immunodeficiency disease, pioneered by Dr Good and other colleagues, was one of the landmark advances in the use of transplantation to treat this disorder. However, immune responses against donor-specific antigens and immune-mediated rejection of transplanted hematopoietic progenitor cells and resultant graft failure remain important obstacles. To date, the primary strategies for avoiding immunological rejection of allogeneic transplants have been to minimize antigenic differences between donor and recipient by matching Human Leukocyte Antigens (HLA; the Major Histocompatibility Complex (MHC) antigens in humans), and to condition the host by suppression/ablation of the host immune system using irradiation or potent chemotherapy.
Limitations for application of allogeneic transplantation: HLA matching
The most important loci for graft survival are the HLA Class I antigens A and B, and the Class II antigen DR (). However, HLA are highly polymorphic with more than 220, 460, 110, and 360 molecularly defined epitopes for HLA-A, B, C, and DR respectively. Mismatching of the serological antigens increases the probability of graft failure (2
). Even with serological cross-matching, small molecular genetic differences may result in graft rejection (5
). Further, graft vs. host disease, associated with HLA antigen mismatching between donor and recipient, results in considerable morbidity and mortality (8
). Despite the growing size of national and international registries there continues to be a lack of suitable donors for both stem cell and organ donation. Procuring suitable HLA-matched donors represents the major limitation to the field of organ and stem cell transplantation. Finally, even with well-matched HLA-A, B, and DR antigens, potent immunosuppressive regimens are required. Thus, for patients there is a fine balance between “too little” immunosuppression and graft rejection, or “too much” immunosuppression resulting in infection and toxicity to the recipient, and in some cases leading to post-transplant malignancies (10
Targeting HLA Class I to reduce graft immunogenicity
HLA down-regulation as a strategy for reducing donor/host immune responses
We have been pursuing a novel strategy for genetically modifying graft cells to achieve long-term suppression of cellular proteins that elicit or potentiate host immune responses against mismatched donor antigens. In principle, modifications to reduce or eliminate expression of specific HLA expression on donor cells would reduce alloreactive immune responses, the need for full HLA matching, and the need to suppress the host immune system. The goal of this approach is to achieve down-regulation of HLA expression sufficient to alleviate graft rejection without deleteriously impacting immune responses against foreign pathogens. Support for the concept that long-term suppression of HLA may enable donor cells to evade host immune responses comes from studies showing that grafts from MHC class I and class II “knockout” mice survive longer than controls (12
). However, antibody-based strategies such as graft pretreatment with haplotype-specific anti-Ia mAb, haplotype-specific immunoconjugates, or mAb directed against graft antigen presenting cells (APC), adhesion or costimulation molecules, were found to have little efficacy in promoting acceptance of allografts in immunocompetent hosts (13
Virus-based gene transfer vectors, such as lentivirus vectors, adeno-associated virus (AAV) vectors, and helper-dependent adenovirus vectors may be used to transduce graft cells with sequences encoding small interfering RNAs (siRNAs), anti-sense mRNA, and ribozymes directed against critical immune recognition molecules or immunostimulatory factors. Engineering the graft using these approaches could decrease donor cell immunogenicity and increase the pool of unrelated bone marrow donors, while reducing the need for toxic myeloablative pre-conditioning or immunosuppressive regimens.
Recent studies have demonstrated that adenoviral-mediated expression of an anti-human MHC I single-chain intrabody achieved ‘phenotypic knockout’ of Class I HLA in human primary keratinocytes and in endothelial (HUVEC) cells (14
). Intrabody-transduced cells were protected from lysis by sensitized allogeneic cytotoxic T lymphocytes (CTLs), while control cells from the same donor were not (15
), establishing proof-of-concept for targeted inhibition of MHC expression in reducing immunogenicity of adult tissue allografts. However, in terms of practical implementation, adenoviral-mediated approaches are unlikely to have long term benefits for graft survival as adenoviruses are a non-integrating vectors that direct short-lived expression and are highly immunogenic due to the leaky expression of viral proteins. Further, the synthetic anti-MHC intrabody protein itself also represents a neo-antigen that may be immunogenic.
Application of siRNA-based methods for knockdown of HLA expression
In contrast to adenoviral vectors, the use of oncoretrovirus- or lentivirus-based gene transfer vectors enables long-term transduction of donor HSC and expression of sequences encoding siRNAs, anti-sense mRNA, or ribozymes directed against immune recognition molecules such as HLA, or co-stimulatory molecules such as CD80. These vectors could also be used for gene delivery of specific ubiquinating enzymes, dominant-negative inhibitory proteins, and transmembrane or secreted immunosuppressive factors. The effect of such manipulations will be to decrease immunogenicity of donor cells and associated host and graft mediated immune responses. A potential limitation of this approach is that complete silencing of HLA may render donor cells susceptible to recognition and attack by non-HLA-restricted effector cells such as natural killer (NK) and lymphokine-activated killer (LAK) cells. It may be necessary to modulate siRNA-mediated silencing and HLA expression to a level that does not allow recognition by alloreactive T cells and yet will not elicit NK-mediated lysis. Alternatively, precise knockdown of specific HLA alleles may be used to nullify potential mis-matches. For example, lentiviral-mediated class- and gene-specific silencing of HLA would broaden the application for transplantation of existing human embryonic stem cell (hESC) pools in future regenerative medicine approaches by reducing the problem of HLA mismatches in recipients of hESC-derived grafts (16
RNA interference (RNAi) is a potent genetic tool for silencing gene expression that triggers post-transcriptional degradation of homologous transcripts through a multi-step mechanism involving double-stranded siRNA (18
). Recently, Cooper, Rossi and colleagues demonstrated that T cells stably transfected with plasmids containing siRNA expression cassettes showed HLA down-regulation that provided protection from cytolysis even when the transfected cells were loaded with a target peptide and challenged with peptide-specific CTLs (20
). However, these stable transformants expressing high copy numbers of siRNA could only be isolated by co-transfection and selection with antibiotic resistance genes in an immortalized Jurkat T cell line. This methodology is not feasible for hematopoietic progenitors and particularly quiescent stem cells. Thus, our focus has been to use lentivirus-based vectors, capable of highly efficient transduction of quiescent cells, including HSC. Permanent integration into the host cell genome, provides the potential for long-term suppression of HLA in adult cell and organ transplantation.
Advantages of Lentiviral-based approaches for HLA-silencing
In contrast toadenovirus, AAV, and non-viral gene transfer approaches, both oncoretroviral vectors and lentiviral vectors have the capacity to integrate into the host genome and direct stable gene expression. Onco-retroviral vectors remain one of the most frequently utilized gene delivery modalities in clinical trials to date. During the retroviral life cycle the reverse transcriptase converts the viral RNA genome to double-stranded DNA, which is then permanently integrated into the chromosomes of the host cell, enabling long-term gene expression. Since native coding sequences have been completely replaced by therapeutic genes of interest there is no possibility that expression of native viral genes might lead to immunological rejection of cells transduced by replication-defective retrovirus vectors. Progressive understanding of the characteristic limitations of MLV-based vectors has spurred numerous improvements in vector production and transduction methods, particularly for ex vivo
applications, and has led to their more judicious application to relevant clinical scenarios. In this regard, arguably one of the leading successes to date has been the amelioration of X-linked severe combined immunodeficiency (SCID); however, enthusiasm regarding this success has been tempered by sobering reminders of the potential for risk, as well as benefit, of viral-mediated gene integration into host chromatin (21
Among the most promising basic advances in gene delivery vector technologies in recent years has been the development of lentivirus-based vector systems. Lentiviruses are complex retroviruses that contain additional regulatory and pathogenicity-enhancing “accessory” genes in addition to the gag, pol and env structural proteins classically expressed by oncoretroviruses (). Although the overall life cycle of lentiviruses is similar to that of oncoretroviruses, there are several major differences between these vectors. Most notably, oncoretroviruses such as MLV can only transduce cells that divide shortly after infection, because the MLV pre-integration complex cannot achieve chromosomal integration in the absence of nuclear envelope breakdown during mitosis.
Advantages of using lentivirus-based vectors for efficient and stable gene-silencing of HLA expression in human cells
In contrast, lentiviruses can infect non-proliferating cells, owing to the karyophilic properties of the lentiviral pre-integration complex, which allows recognition by the cell nuclear import machinery ( and
). Correspondingly, lentiviral vectors can transduce cell lines that are growth-arrested in culture, as well as terminally differentiated primary cells including hematopoietic stem cells, neurons, hepatocytes, cardiomyocytes, endothelium, alveolar pneumocytes, keratinocytes and dendritic cells [18
]. Hence, there is increasing interest in the development of vector systems based on a wide variety of lentiviruses, including human immunodeficiency virus (HIV), simian immunodeficiency virus (SIV), feline immunodeficiency virus (FIV), and equine infectious anemia virus (EIAV). We have focused primarily on HIV-based lentiviral vectors for permanent integration of transgenes to achieve long-term modification of cellular phenotype, as this technology has progressed the most rapidly and is already in clinical trials.
Molecular events associated with retrovirus or lentivirus vector -mediated transduction
Development of lentiviral vectors for silencing of HLA Class I expression
To test two different targets for HLA silencing, candidate siRNA sequences directed against unique sequences in human HLA-A0201 (HLA-A2 allele-specific) and against common sequences conserved among Class I loci (HLA-ABC pan-specific) were designed as short hairpin RNA (shRNA) loop structures and cloned into pLentiLox-DsRed (). This HIV-derived lentiviral vector contains multiple cloning sites for insertion of shRNA constructs to be driven by an upstream U6 promoter, and a downstream CMV promoter-DsRed fluorescent protein (marker gene) cassette flanked by loxP sites. FACS analysis for cell surface HLA expression after transient tranfection of 293T human embryonic kidney cells with plasmid demonstrated highest silencing activity with either an HLA-A,B,C pan-specific shRNA construct targeting the conserved HLA Class I sequence 5′-GCTACTACAACCAGAGCGAG-3′, or with an allele-specific shRNA construct targeting the unique HLA-A0201 sequence 5′-GGATTACATCGCCCTGAAAG-3′ (, indicated by arrows). These vector constructs were selected for virus construction and further testing. Virus production was performed using a standard third-generation lentiviral system (23
), consisting of packaging plasmids (pMD.G encoding VSV-G envelope, pMDLg/p encoding HIV gag-pol, and pRSV-REV encoding HIV rev) co-transfected along with each pLentiLox-DsRed vector plasmid into 293T cells by calcium phosphate precipitation. The resulting viruses deliver both the U6 promoter-driven shRNA cassette and a CMV promoter-driven DsRed fluorescent marker gene cassette upon infection of target cells. Lentiviral vectors expressing only DsRed were also prepared in parallel as negative controls. Virus titers, as determined by FACS analysis for expression of the co-expressed DsRed fluorescent marker protein, were 2 to 10 × 10e8 transducing units (TU) per mL.
Lentiviral vector constructs for short hairpin RNAs directed against HLA
Lentiviral vectors for shRNA targeting of pan-specific and allele-specific HLA sequences were used to transduce naïve 293T cells at increasing multiplicities of infection (MOI; i.e., virus-to-cell ratio). The 293T cell line predominantly expresses HLA Class I antigen HLA-A2 and lower levels of HLA-B7 and -Cw7. FACS analysis of transduced cells showed an increase in mean fluorescence level in the red channel due to co-expression of DsRed. A dose-dependent reduction in HLA levels due to shRNA expression was detected as a reduction in fluorescence in the green channel after staining with HLA-A2-specific or -ABC pan-specific antibodies conjugated with fluorescein isothiocyanate (FITC). Transduction at higher MOI (10
) with allele-specific and pan-specific shRNA vectors reduced cell surface expression of HLA-A and HLA-ABC up to 50% and over 80%, respectively, compared to HLA expression levels in cells transduced with lentivirus expressing DsRed only (, (24
Studies using lentivirus-based vectors for efficient and stable gene-knockdown of HLA expression in human cells
Resistance to alloreactive CTL-mediated killing without LAK cell-mediated killing
The effect of lentiviral mediated HLA silencing on CTL-mediated killing of allogeneic cells was tested in vitro
. Peripheral blood mononuclear cell (PBMC)-derived allo-reactive human effector T lymphocytes (allo-CTL), kindly provided by Dr. Carol Kruse at the Sidney Kimmel Cancer Center, were pre-activated against human stimulator cells expressing the Class I antigens HLA-A2, B44, and C5. Target 293T cells transduced with lentiviral vectors encoding either HLA-A2 allele-specific or HLA-ABC pan-specific shRNA, or with pLentiLox-DsRed control vector were analyzed for sensitivity to cytolysis by incubation with HLA-activated allo-CTL at a ratio of 10:1 (effector: target cell ratio). After 48 hr incubation, the level of interferon-gamma (IFNγ) production by allo-CTLs was measured by ELISA, and the viability of adherent target cells remaining after washing was determined by MTS assay and annexin V staining (24
). Target cells transduced with HLA-ABC shRNA and HLA-A0201 shRNA vectors induced significantly less IFNγ production from allo-CTLs, and showed enhanced resistance to alloCTL-mediated killing by both annexin V and MTS assays (p<0.05) compared to cells transduced with control vector expressing DsRed only (24
A potential limitation of the HLA-silencing approach in the transplant setting is that complete loss of HLA expression may have the undesirable effect of increasing sensitivity to lysis by non-HLA-restricted effectors such as NK and LAK cells. However, we did not observe significant differences in the survival of HLA-ABC shRNA or HLA-A0201 shRNA vector-transduced target cells compared to DsRed only-transduced control cells after incubation with LAK cells derived from the same donor PBMC (24
Reducing histocompatibility barriers by conditioning the graft, not the host
Our studies demonstrate that lentiviral vectors expressing pan-Class I specific shRNA constructs directed against conserved sequences in HLA-A,B,C, and HLA-A2-specific shRNA constructs can achieve dose-dependent knockdown of HLA levels in human cells, associated with induction of resistance to killing by allo-reactive T effector cells, without incurring significant sensitivity to non-HLA-restricted killer cells.
In concurrent and more recent studies, Blasczyk, Horn, and colleagues have also reported that lentiviral vectors expressing shRNA cassettes targeted against HLA-A and beta2-microglobulin (β2
M) could achieve similar allo-protection in HeLa and immortalized primary B cell lines (25
). Surrogate assays showed prevention of HLA-A-specific antibody-mediated, complement-dependent cytotoxicity and reduced CD8+ T cell proliferation and IFNγ secretion (25
). While complete elimination of HLA expression was not achieved, transduced cells showed protection from T cell allo-reactivity by siRNA transduction, and only elicited moderate NK cell reactivity with conditional expression of a lentiviral shRNA vector (26
). Our results with lentiviral-mediated delivery of different HLA-A0201-specific and HLA-ABC pan-Class I specific shRNA constructs showed similar results including effectiveness in conferring resistance to allo-reactive CTL by direct measurement of target cell survival (24
Combining siRNA technology with lentivirus vector-mediated genetic engineering of donor cells or tissues, offers the potential to achieve immunological evasion with less toxicity than currently used non-specific immunosuppressive agents. In this context, it may be advantageous that HLA knockdown with these particular pan-Class I shRNA constructs did not result in complete loss of expression even at high MOI, as this may avoid sensitivity to non-HLA-restricted killer cell activity. Down-regulation of specific HLA alleles by precise siRNA targeting may expand the histocompatibility and utility of existing donor cells and tissues by nullifying certain classes of HLA sequences and thus making it easier to find matches with the remaining HLA sequences. In the future, issues regarding potential NK /LAK mediated lysis of donor cells with knock-down HLA expression will have to be explored in in vivo models of transplantation where the responses of host and donor immune effector cells may be tested.
Global or allele-specific knockdown of HLA in order to effect long-term modulation of the immunogenicity of transplanted cells and tissues, would represent a fundamental shift in the approach to achieving histocompatibility, by engineering the graft rather than immunosuppressing the recipient. While efficient transduction of entire solid organs remains a significant technical hurdle, application of this strategy can be readily envisioned for ex vivo transduction of cellular transplants in which HLA matching is a rate-limiting factor, such as bone marrow transplants, skin grafts and pancreatic islet cells. With regard to HSC transplantation in particular, the primary technical goals now are to improve the efficiency of gene transfer and the engraftment of genetically modified hematopoietic progenitor cells to clinically relevant levels. As noted above, lentiviral vectors are a more efficient system for gene transfer to quiescent HSC than previously tested vectors because they encode proteins that permit active import of the viral genome into the nucleus in the G0/G1 phase of the cell cycle (, ). However, remaining hurdles to the application of gene therapy for hereditary and other disorders include variable transduction of HSC and gene expression, and immune responses to vectors, therapeutic genes, and allo-antigens in the case of allogeneic transplantation.
Approaches to exploring in vivo chemoselection to enhance hematopoietic reconstitution
Clearly, additional complementary strategies for genetic engineering of HSC that would provide a selective advantage for transplanted donor cells and enable successful engraftment with less toxic preparative and immunosuppressive regimens would increase the numbers of individuals to whom this potentially curative therapy could be offered. The rarity and quiescence of HSCs has led to the development of several strategies to amplify these cells either ex vivo
or in vivo
. The ability to increase the numbers of HSCs ex vivo
could increase the efficiency with which stem cells could be modified by expression of shRNA or by corrective therapeutic genes, and would also increase the numbers of cells available for transplantation. Many earlier studies focused on the use of combinations of recombinant cytokines to enhance proliferation of HSCs ex vivo
. However, concerns were raised that extended culture decreased the re-populating and self-renewal activity of cultured cells (27
As an alternative approach, HSC transduction with vectors carrying drug resistance genes has been tested as a method for achieving positive selection of the transduced cells in vivo
. The human M
esistance gene (MDR1) or the dihydrofolate reductase (DHFR) gene were introduced into BM cells by oncoretroviral-mediated transduction after expansion in culture with IL-3, SCF, IL-6 for 12 days (30
). Transduced cells carrying MDR1 were selected in the presence of Taxol, while DHFR-containing cells were selected with trimetrexate. There was a 100-fold in vitro
expansion of cells with either of the selective agents; however, engraftment of transplanted DHFR transduced cells was only transient with in vivo
). In contrast, BM populations transduced with MDR1 and selected in vivo
with taxol showed long-term engraftment, but all mice transplanted with these cells developed a myeloproliferative syndrome (30
). Another approach for HSC expansion has been the introduction of HoxB4 (31
). However, expression of HoxB4 has been associated with the emergence of leukemic clones in a large animal model (32
). Thus, these approaches have been associated with considerable liabilities.
Lentiviral-mediated expression of P140K-MGMT and in vivo chemoselection of HSC
A recent advance in positive selection strategies has been to use lentiviral transduction and in vivo
selection of HSC expressing a modified DNA alkyltransferase, the P140K variant of the MGMT gene (P140K-MGMT) (33
). In vivo
expansion of HSC populations is achieved by administration of benzylguanine (BG) to inhibit endogenous MGMT activity and the nitrosourea BCNU, or other alkylating agents such as temozolomide, to which the P140K-MGMT shows 1000-fold resistance compared with the wild type enzyme (34
). Transplantation of P140K-MGMT transduced congenic HSC in adult animals followed by successive cycles of BG /BCNU in vivo
provides selection at the stem cell level and high levels of donor chimerism, without major toxicity (37
). Chemo-selection after P140K-MGMT-lentiviral transduction and amplification has also been successfully demonstrated with human NOD/SCID repopulating cells (36
) and with allogeneic canine CD34+ cells in a large animal model, with sustained multilineage enrichment of transduced repopulating cells (39
). Similarly, in an adult murine syngeneic transplant model, peripheral blood showed high levels of GFP+ cells in multiple lineages after two cycles of in vivo
Neonatal models for testing genetic engineering approaches to modify donor stem cells
We are combining the MGMT-based positive selection strategy for in vivo amplification of drug-resistant donor cells with transplantation during the neonatal period, when tolerance may be more readily achieved. Our rationale is that the neonatal period is a unique window during immune ontogeny for transplantation of genetically modified allogeneic HSC to achieve long-term engraftment. Neonatal transplantation provides a model system for exploring mechanisms of tolerance induction to neo-antigens, and allogeneic stem cell engraftment. The MGMT-based in vivo chemoselection strategy with successive cycles of BG /BCNU will both expand donor stem cells and potentially deplete allo-reactive cells of donor and host origin, reducing the need for toxic ablative or immunosuppressive treatment.
In collaborative studies with Dr Stanton Gerson’s laboratory, Case Western Reserve University, we have tested the efficacy of a bifunctional P140K-MGMT-GFP (MAG) lentiviral vector in providing a selective advantage to transduced adult BM HSC after neonatal transplantation and in vivo
selection with BCNU and BG (). The functionality of the vector was first tested in in vitro
studies of lentivirally transduced 293T cells. Transduced cells and untransduced were cultured with BG/BCNU for 6 days (). While MAG-transduced 293T cells were resistant to BG/BCNU, control untransduced cells showed greater than 90% cell death. Studies in the Gerson laboratory have demonstrated efficient transduction of human CD34+ at low MOI and in vivo
selection with BCNU (41
). Significantly, myeloablation was not required for efficient in vivo
selection of syngeneic transduced, transplanted cells. The P140K-MGMT mutant is 20-fold more BG-resistant than other variants such as G156A-MGMT (37
In vitro gene delivery of MGMT into HSC followed by transplantation and in vivo chemoselection
In vitro and In vivo chemoselection of syngenic and allogeneic HSC
The feasibility of this in vivo HSC selection strategy was then assessed in our neonatal transplantation model. BALB/c whole BM was transduced with MAG lentivirus and transplanted into day 2 BALB/c neonates after non-ablative conditioning (). Mice were treated with BG/BCNU at 5 and 11 weeks after birth and flow analysis of GFP expression performed before, and 1 month after, each drug cycle. After two cycles, 39.5% of mononuclear cells in peripheral blood were GFP positive showing successful engraftment and in vivo chemoselection of syngeneic GFP+ donor cells without significant toxicity in recipients.
We have also demonstrated the feasibility and efficacy of MAG-lentiviral mediated transduction and in vivo chemoselection of transduced allogeneic BM in the neonatal transplant model. BALB/c whole BM was transduced and transplanted into day 2 C57Bl/6XBALB/c F1 neonates after treatment with busulfan and low dose radiation at birth. Two cycles of BG/BCNU were administered 5 and 10 weeks after transplant. Flow analysis of peripheral blood prior to chemoselection demonstrated 1-3% GFP+ cells. However, after two cycles of BG/BCNU up to 37% GFP+ donor cells were seen in peripheral blood (). Levels of donor chimerism were stable thereafter, and no signs of GVHD or immune rersponses to the GFP neo-antigen were detected in animals observed longitudinally.
This neonatal transplantation model will be a valuable tool for evaluating different and complementary strategies for genetic engineering of HSC. The model will not only be useful for addressing technical hurdles to successful gene therapy and improved safety and applicability of allogeneic tranplantation, but will also provide insights into mechanisms of immune tolerance induction during immune ontogeny. We have shown that transplantation in the neonatal period permits stable high-level engraftment of syngenic or allogeneic HSC with minimal preparative regimens using in vivo chemoselection. Further, enrichment of p140-MGMT-GFP (MAG) transduced HSC after either syngeneic or semi-allogeneic neonatal transplant and BG/BCNU selection was achieved without evidence of immune responses to either neo-antigens (GFP) or allo-antigens. In the future, combining approaches including RNAi mediated knockdown of HLA, modulation of immune responses, and strategies to provide selective advantages for transplanted genetically modified HSC or other donor cells hold great promise for the field of tissue transplantation.