|Home | About | Journals | Submit | Contact Us | Français|
Artemis is an endonucleolytic enzyme involved in nonhomologous double-strand break repair and V(D)J recombination. Deficiency of Artemis results in a B− T− radiosensitive severe combined immunodeficiency, which may potentially be treatable by Artemis gene transfer into hematopoietic stem cells. However, we recently found that overexpression of Artemis after lentiviral transduction resulted in global DNA damage and increased apoptosis. These results imply the necessity of effecting natural levels of Artemis expression, so we isolated a 1 kilobase DNA sequence upstream of the human Artemis gene to recover and characterize the Artemis promoter (APro). The sequence includes numerous potential transcription factor-binding sites, and several transcriptional start sites were mapped by 5′ rapid amplification of cDNA ends. APro and deletion constructs conferred significant reporter gene expression in vitro that was markedly reduced in comparison to expression regulated by the human elongation factor 1-α promoter. Ex vivo lentiviral transduction of an APro-regulated green fluorescent protein (GFP) construct in mouse marrow supported GFP expression throughout hematopoeitic lineages in primary transplant recipients and was sustained in secondary recipients. The human Artemis promoter thus provides sustained and moderate levels of gene expression that will be of significant utility for therapeutic gene transfer into hematopoeitic stem cells.
Genome integrity is maintained by way of essential DNA double-strand break (DSB) repair mechanisms. Multicellular eukaryotic organisms resolve DSBs primarily through the canonical nonhomologous end-joining (NHEJ) pathway, which repairs genomic insults generated by both external-damaging agents and internal cellular processes. The NHEJ cascade begins when the Ku70/Ku80 heterodimer recognizes and binds a DNA DSB. The DNA-dependent protein kinase catalytic subunit (DNA-PKcs) then complexes with the DNA bound Ku to form the DNA-PK holoenzyme (Yaneva et al., 1997; West et al., 1998). The DNA-PK holoenzyme subsequently recruits and binds Artemis, which acquires endonucleolytic activity recognizing 5′ and 3′ overhangs (Ma et al., 2002). After Artemis digests the DNA overhangs, a heteromultimer of XLF, XRCC4, and DNA Ligase IV is recruited to the synaptic complex to repair and seal the break (Grawunder et al., 1997, 1998).
NHEJ also plays a fundamental role in the rearrangement of immunoglobulin (Ig) genes and T cell receptor (TCR) genes (van Gent et al., 1996; Zhu et al., 1996). This site-specific rearrangement process begins when the RAG-1 and RAG-2 complex is recruited to recombination signal sequences (RSS) flanking each V (variable), D (diversity), or J (joining) coding segment (Oettinger et al., 1990). The RAG complex introduces a nick adjacent to each RSS; the resulting 3′ hydroxyl group nucleophilically attacks the antiparallel DNA strand to form a hairpin structure at the coding ends (Roth et al., 1992; McBlane et al., 1995). The Artemis:DNA-PK complex endonucleolytically cleaves the coding end hairpin, and then the DSB is processed and repaired through the NHEJ pathway (Ma et al., 2002).
Deficiency of Artemis disrupts both DNA DSB repair and V(D)J recombination, manifesting in humans as a radiation sensitive form of severe combined immunodeficiency characterized by the inability to rearrange Ig and TCR genes, ultimately resulting in the absence of both B and T lymphocytes. Artemis deficiency is a primary immunodeficiency (PI) designated SCID-A due to a founder mutation in Athabascan-speaking Native Americans and significant incidence of the disease in this population. In general, PIs are treatable by allogeneic hematopoeitic stem cell transplantation, with an attendant risk of infection, graft rejection, and graft-vs.-host disease associated with about 20% mortality worldwide. Accordingly, there is great interest in the development of more effective therapeutic approaches for this disease. Recent results from clinical trials have demonstrated the effectiveness of transplantation using autologous HSC after ex vivo genetic correction by retroviral transduction for two PIs, one caused by adenosine deaminase deficiency and the other caused by absence of the common gamma chain for several cytokine receptors (Aiuti et al., 2009; Hacein-Bey-Abina et al., 2010). Both of these studies reported long-term engraftment of corrected stem cells in the majority of patients, ultimately resulting in reconstitution of cellular and humoral immunity (Aiuti et al., 2009; Hacein-Bey-Abina et al., 2010). These results demonstrate the effectiveness of gene transfer for the treatment of PIs in general, potentially including SCID-A. However, we recently reported that overexpression of Artemis after lentiviral transduction is associated with cytotoxicity, inducing a halt in cell cycle progression, fragmentation of genomic DNA, and, ultimately, apoptosis (Multhaup et al., 2010). These results highlight the importance of providing Artemis expression at a level that is nontoxic and yet sufficient to correct the T−B− phenotype in preclinical studies and in clinical application to human SCID-A.
To explore the potential of innate regulation using the endogenous human Artemis promoter (APro) region, a 1-kilobase (kb) sequence directly upstream of the human Artemis translational start site (TSS) was isolated by polymerase chain reaction (PCR) amplification, characterized, and shown to support green fluorescent protein (GFP) and luciferase reporter gene expression in vitro. 5′ rapid amplification of cDNA ends (RACE) revealed evidence for multiple transcriptional start points, and several deletion constructs were found to promote reporter gene expression in vitro. Additionally, ex vivo transduction of mouse bone marrow with an APro-regulated GFP lentiviral vector resulted in GFP expression at a significantly reduced level in comparison with control mice transplanted with EF1α-GFP transduced marrow. The human Artemis promoter supported GFP expression in all hematopoeitic lineages that persisted in secondary transplant recipients. These results demonstrate that the human Artemis promoter provides moderate and yet reliable levels of expression in hematopoeitic lineages, essential for regulated expression of potentially toxic gene products (such as Artemis) and reduced risk of genotoxicity associated with integrative gene transfer.
A 1kb sequence directly upstream of the human Artemis TSS was amplified from human genomic DNA (Roche) by PCR using forward primer 5′ ACCGGTCAGAGAGCCGAAATCACGCC 3′ and reverse primer 5′ ACCGGTAGCGCCGCCGATCCCAGAGT 3′ (flanked by AgeI restriction sites, underlined). Two rounds of PCR amplification were necessary to achieve a 1-kb product (5min at 95°C; 35 cycles of 1min at 95°C, 1min at 69°C, and 1.5min at 72°C, with a final 10min extension at 72°C). The product was gel extracted (Qiagen) and cloned into TOPO vector PCR2.1 to generate pCR2.1/APro. The resulting clone was sequenced and found to be identical to the published human chromosome 10 sequence 1kb directly upstream of the human Artemis TSS (PubMed Reference: NT_077569).
The full-length APro sequence was digested with AgeI and ligated into the AgeI site of pGL3Basic (Promega) directly upstream of the firefly luciferase gene to generate pGL3/APro. Several APro deletion mutants were generated from the full-length APro sequence by PCR amplification. Primers utilized to amplify truncated APro sequences are illustrated in Table 1. PCR products containing various portions of the APro sequence were digested and cloned into the AgeI site of pGL3Basic.
A lentiviral vector plasmid designed for regulation of GFP by the full-length human APro was generated from pCSIIEG (Agarwal et al., 2006). The CMV promoter was excised from pCSIIEG by AgeI digestion, and then the APro sequence was ligated into the AgeI site directly upstream of the GFP gene to generate pCSII/APro-GFP.
HEK 293T and murine NIH 3T3 tk− cells were cultured in Dulbecco's modified Eagle medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 1% antibiotic antimycotic at 37°C and 5% CO2. HEK 293T cells were transfected with 2μg of pGL3/APro or derivative deletion constructs using the DNA calcium phosphate coprecipitation technique as described (Zufferey et al., 1997). Briefly, 10μL of 2.5M CaCl2 was added to 2μg of luciferase expression plasmid DNA to a final volume of 100μL for each well in a 6-well plate to be transfected. An equal volume of 2X BBS [0.5M BES, 150mM NaH2PO4, 2.8M NaCl; pH 7.05] was added to the DNA-calcium mixture in a drop-wise fashion and thoroughly mixed. The CaCl2-DNA-BBS solution was added to each well containing 5×104 HEK 293T cells and incubated at 37°C with 3% CO2. Each culture was cotransfected with 0.25μg of phRL-CMV (Promega) as an internal control for transfection efficiency; assay results are expressed as firefly RLU/Renilla RLU. pKT2/Cal, a firefly luciferase construct regulated by the strong CAGS promoter (Wilber et al., 2007), was utilized as a luciferase positive control, and the promoter-less pGL3Basic (Promega) was used as a luciferase background control.
BJAB, BLIN-1 K+, and BLIN-1 K− cells were cultured in RPMI 1640 medium (Invitrogen) supplemented with 10% FBS and 1% antibiotic antimycotic at 37°C with 5% CO2 and transfected in an Amaxa Nucleofector (Lonza). Cells were harvested, washed, and resuspended in Amaxa supplemented nucleofection buffer C. About 5×105 cells were mixed with 5μg of pGL3Basic/APro or derivative deletion construct plus 0.5μg of phRL-CMV and then electroporated at program setting X-001. Immediately after nucleofection, cells were transferred to prewarmed culture medium and incubated at 37°C with 3% CO2.
Forty-eight hours post-transfection, cells were collected and analyzed for firefly and renilla luciferase expression via Promega's Dual Luciferase Reporter Assay System. Briefly, cells were harvested by trypsinization, washed with 1X phosphate-buffered saline (PBS), and resuspended in passive lysis buffer (provided in the Dual Luciferase Reporter Assay System). After a 5-min incubation at room temperature, the lysate was vortexed vigorously and then cleared by centrifugation for 2min at 13,200 g, collecting the supernatant. A 20μL aliquot of lysate supernatant was incubated with Promega firefly luciferase substrate, and then, subsequently, incubated with Stop and Glo Renilla luciferase substrate, assaying for luminescence in a Lumat LB 9507 luminometer. Student's t-test was utilized to statistically evaluate differences in luciferase expression levels between transfected cell populations.
5′ RACE was carried out using RNA extracted with Trizol reagent (Invitrogen) and the 5′ RACE System for Rapid Amplification of cDNA Ends (Invitrogen), Version 2.0 kit and reagents, according to the manufacturer's instructions. Primers utilized for the 5′ RACE protocol are listed in Table 1. RACE products were isolated by gel electrophoresis, extracted, cloned into TOPO vector PCR2.1, and sequenced. Vector NTI Software (Invitrogen) was used to align 5′ RACE product sequences against the published human Chromosome 10 sequence (PubMed Reference: NT_077569).
Vesicular stomatitis virus G (VSV-G) pseudotyped lentiviral vectors were generated as described (Zufferey et al., 1997; Gori et al., 2007). Briefly, 24h pretransfection, 1.4×107 HEK 293T cells were seeded into poly-L-lysine coated 15-cm2 plates and cultured in DMEM supplemented with 1% penicillin streptomycin and 8% FBS at 37°C with 5% CO2. Lentiviral vector plasmid constructs were cotransfected with pΔNRF to provide lentiviral structural and enzymatic proteins and pMD.G to provide VSV-G envelope protein. Twelve hours post-transfection, the medium was replaced using DMEM supplemented with 4% FBS. Viral supernatants were collected 24, 36, and 48h post-transfection, pooled, and then concentrated 100-fold by centrifugation at 23,000 g in a Sorvall RC5B centrifuge. Vector was resuspended in Iscove's modified DMEM, aliquoted, and stored at −80°C for future use. For quantification of vector titers, NIH 3T3 tk− cells were transduced with varying amounts of vector in the presence of 8μg/mL polybrene. Forty-eight hours post-transfection, cells were harvested for flow cytometric analysis of GFP expression to determine the percentage of cells transduced. DNA was also extracted from the transduced cells and subjected to quantitative PCR as previously described, using a probe specific for the integrated lentiviral strong stop sequence or a probe for the GFP sequence as a measure of lentiviral vector titer (Gori et al., 2007).
All procedures were reviewed and approved by the University of Minnesota Institutional Animal Care and Use Committee. CD45.2 and CD45.1 C57Bl/6 mice were obtained from the National Cancer Institute (Frederick, MD) and provided food and water ad libitum. Bone marrow was flushed from the long bones of the hind limbs of donor mice into DMEM supplemented with 10U/mL heparin, 10% FBS, and 1% PenStrep antibiotic/antimycotic. Red blood cells were lysed using ammonium chloride hemolysis buffer [0.8% NH4Cl with 0.1mM EDTA] (StemCell Technologies). The remaining nucleated cells were washed with 1X PBS and then triturated into a single-cell suspension in transduction medium (complete StemPro-34 SFM media with supplement, 2mM L-glutamine, 1% PenStrep [all from Invitrogen], 100ng/mL murine IL-3, 100ng/mL murine IL-6, 100ng/mL murine TPO, 100ng/mL murine SCF [all cytokines from R&D Systems], and 8μg/mL polybrene [Sigma-Aldrich]). The marrow was transduced twice, once immediately after marrow harvest and a second time 20h after the first exposure. Transduced cells were harvested, washed, counted, and prepared as a single-cell suspension in Iscove's modified DMEM. Equal numbers of bone marrow cells were injected via lateral tail vein into lethally irradiated (800 rads, cesium source) congenic recipients. For secondary transplants, marrow samples were collected individually from primary recipients as just described, isolating the nucleated fraction and transplanting 5×106 cells into each of three secondary irradiated (800 rads) recipients.
Blood was collected via cheek puncture and assayed for GFP expression in specific hematopoietic compartments by flow cytometry. Whole blood was treated with ammonium chloride hemolysis buffer (0.8% NH4Cl with 0.1mM EDTA; StemCell Technologies), washed, and then pelleted leukocytes were resuspended in staining buffer [1× PBS plus 1% FBS, 0.002% sodium azide and fluorochrome-conjugated monoclonal antibodies for identification of hematopoeitic cell type]. All antibodies utilized for flow cytometric analysis were purchased from eBiosciences. GFP expression as well as immunophenotype analysis was conducted using monoclonal mouse antibodies against CD45.1, CD45.2, B220 (B lymphocytes), CD3ε (T lymphocytes), NK1.1 (Natural Killer cells), and Gr-1 (myeloid lineages) and assayed on an LSRII instrument. Data were collected using CellQuest Pro (BD Biosciences) and analyzed using FlowJo (Tree Star, Inc.) software. Isotype staining provided an internal control and was used to determine appropriate gating to distinguish between GFP-positive and GFP-negative cells. Mean fluorescence intensity (MFI) was determined using the FlowJo software statistics tool. Significance of differences between groups was determined via Student's t-test.
To identify the promoter of the human Artemis gene (APro), we examined the sequence of human chromosome 10p13 directly upstream of the human Artemis translational start site (TSS). A total of fourteen potential regulatory sites were identified within this region using the search engine TFSEARCH: Searching Transcription Factor Binding Sites (ver1.3) (Heinemeyer et al., 1998) including several known hematopoeitic transcription factor binding sites such as GATA-1, 2, 3 (−967, −880, −849, −175, and −16), AML1A (−32) and Lyf1 (−356) (Fig. 1). Additionally, a CAAT box at −811 and a TATA box at −595 were identified (Fig. 1): These motifs plus a potential 5′ splice site (AAG|GTTAG) at base pair −501 and several downstream 3′ splice sites predicted the possibility of an intron in the 5′ untranslated region of messages initiated at around −500bp or further upstream.
Considering the observed sequence characteristics and potential regulatory elements, we recovered the 1kb segment of human chromosome 10 genomic DNA located directly upstream of the human Artemis TSS. The PCR product was cloned into TOPO vector PCR2.1, sequenced, and found to be identical to the published human genomic sequence (PubMed Reference: NT_077569).
A reporter construct was assembled in which the 1-kb APro sequence was inserted immediately upstream of the firefly luciferase gene (pGL3Basic/APro). Due to its essential role in NHEJ, Artemis expression is not cell type specific. We therefore introduced pGL3Basic/APro into readily transfectable HEK 293T cells, where the APro sequence supported luciferase expression that was about 105-fold lower than that of pKT2/Cal, a strong CAGS regulated firefly luciferase construct (Fig. 2A). Luciferase expression driven by a downstream 500bp fragment (pGL3Basic/−500_TSS) was equal to or greater than that mediated by the full-length 1kb APro sequence. In contrast, luciferase expression driven by an upstream 500bp sequence (pGL3Basic/−1000−500_TSS) was indistinguishable from background (Fig. 2A). These results suggest that the 500nt region immediately upstream of the TSS contains the human Artemis promoter.
To map those regions within the downstream 500nt of the APro sequence that are important for gene expression, deletion constructs were generated based on the clustered location of TFSEARCH-identified transcription factor binding sites. The region from −374 to the Artemis TSS, including USF and Lyf1 binding sites (pGL3Basic/−374_TSS), was found to mediate the same level of luciferase expression as compared with the full-length construct (p=0.41) but only half the amount of luciferase expression conferred by the −500_TSS construct (p<0.01) (Fig. 2A). However, the region from −184 to the Artemis TSS (pGL3Basic/−184_TSS), including a GATA-1 binding site, supported luciferase expression at a level similar to that of the full-length APro (p=0.08) and −500_TSS sequences (p=1.0) (Fig. 2A). Finally, a fragment extending from nt −100 to the human Artemis TSS, containing two major transcription factor binding site clusters (AML1A, GATA-1, STAT-1, Elk-1, and NRF-2), and the upstream APro construct −1000 to −500 supported minimal luciferase activities at levels slightly above background (p<0.008 and p<0.03, respectively) (Fig. 2A). These results define a functional promoter region in HEK 293 cells extending as little as 184bp upstream of the TSS of the human Artemis gene.
Artemis plays a key role in lymphocyte development during the receptor V(D)J recombination stage. To assess the expression profile of APro deletion constructs in lymphoid cell lines, Jurkat cells (an immortalized human T cell line) and BJAB cells (a human EBV-negative Burkitt-like B cell lymphoma) were electroporated and luciferase expression was assayed 2 days later. Surprisingly, Jurkat cells expressed a high level of luciferase after transfection with pT2/Cal but yielded only background levels of luciferase expression in all deletion construct transfection reactions (data not shown), suggesting that Artemis expression in mature T cells may be very moderate to null. On transfection into BJAB cells, the full-length APro sequence and deletion constructs −374 and −184 displayed luciferase expression patterns similar to those seen in HEK 293T cells, that is, the observed expression level was substantially less than that mediated by the CAGGS promoter, yet significantly above background (Fig. 2B). Interestingly, the upstream APro construct −1000 to −500, containing the TFSearch-identified TATA box, generated the highest level of luciferase expression in BJAB cells (Fig. 2B), in contrast to the background levels observed in 293T cells transfected with this construct (p=0.05) (Fig. 2A). This upstream region mediated luciferase expression at levels indistinguishable from the full-length APro construct (p=0.3) as well as the −500 and −184 to TSS regions (p=0.08 and p=0.1, respectively). These results may not be surprising, considering the numerous lymphoid specific transcription factor-binding sites located in this upstream region, suggesting the potential for two or more regions of the APro sequence that regulate Artemis gene expression.
To determine the position of transcript initiation, 5′ rapid amplification of cDNA ends was carried out as described in Materials and Methods using human liver RNA and total cellular RNA extracted from BLIN-1 K+, BLIN-1 K−, BJAB, and Jurkat cell lines. BLIN-1 is a cell line derived from a human B cell lymphoma; BLIN-1 K− is characterized as retaining the ability to continuously undergo V(D)J recombination, whereas BLIN-1 K+ is characterized as kappa light chain positive, therefore having completed V(D)J recombination. These cell lines are of interest not only due to their lymphocytic lineage but also because of the anticipated requirement of Artemis for V(D)J recombination in BLIN-1 K−. BLIN-1 K+ serves as an internal control for the cessation of V(D)J recombination.
Several 5′ RACE products were identified that grouped into four clusters (Fig. 3). The most 3′ cluster from −18 to −36 (Group 1) was the only group amplified from BLIN-1 K− RNA, but was also amplified from BLIN-1 K+ and liver RNA. Group 2 consisted of isolated 5′ RACE products dispersed between −46 and −79, generated only from BLIN-1 K+ and liver RNA. A single start site was identified in the BJAB lymphoma line, located at base pair −108 (Group 3), which was not observed in any of the other samples. Finally, a far upstream group of 5′ RACE products from −376 to −421 (Group 4) was amplified only from liver RNA. No 5′ RACE products were identified using RNA extracted from Jurkat cells.
The transcriptional start sites identified by 5′ RACE are consistent with the luciferase expression profile observed on APro deletion analysis. The full-length pGL3Basic/APro and deletion constructs pGL3Basic/APro-500_TSS and pGL3Basic/APro-184_TSS that exhibited the highest levels of luciferase activity in 293 cells (Fig. 2) correspond with transcription start points identified in liver RNA located at base pairs −420 and −80, respectively (Fig. 3). Moreover, 5′ RACE products were not recovered using RNA extracted from Jurkat cells, consistent with the lack of expression observed on transfection of APro regulated constructs. The absence of Groups 1 and 2 5′ RACE products from BJAB cells is consistent with the higher level of luciferase expression observed after transfection of the upstream −1000 to −500 APro deletion construct, even though we did not observe 5′ RACE products from this region containing a TFSearch-identified TATA box.
Our motivation in characterizing the human Artemis promoter is to incorporate it into vectors to achieve moderate expression levels in vivo. We, thus, generated a lentiviral vector with the GFP coding sequence positioned immediately downstream of the full-length APro sequence (CSII/AProGFP; Fig. 4). Considering the sequence motifs contained in this region and expression profiles generated from deletion analysis, we determined that the full 1kb APro sequence would be most appropriate to include in our vector design. Since CSII has a self-inactivating vector design, GFP expression in transduced cells is entirely dependent on expression driven by the APro sequence. Vectors were generated, and vector titer was quantified by flow cytometry as described in Materials and Methods. Flow cytometric analysis of NIH 3T3 cells transduced with CSII/AProGFP demonstrated that the CSII/AProGFP vector achieved a similar gene transfer frequency as the control CSIIEG vector, in which GFP is regulated by the strong EF1α promoter (Fig. 4). However, the CSII/AProGFP transduced population exhibited an MFI of 21.5, 10-fold lower than populations transduced with CSIIEG, which demonstrated an MFI of 216 (Fig. 4). The APro sequence thus confers a low level of expression compared with stronger promoters such as CAGGS or EF1α after transfection (Fig. 2A) or after lentiviral transduction (Fig. 4), respectively.
To test for APro mediated gene expression in vivo, whole bone marrow was harvested from C57BL/6 CD45.1 animals and transduced overnight with either CSII/AProGFP or the control vector CSIIEG at a multiplicity of infection of 10. The transduced marrow was transplanted into lethally irradiated CD45.2 recipient animals, which were subsequently assayed monthly for donor engraftment and GFP expression by flow cytometry. Animals from both groups exhibited high levels of engraftment (~90%) with GFP expression persisting up to 4 months. Animals transplanted with CSII/AProGFP-transduced marrow expressed GFP in about 33% of the donor leukocyte compartment, whereas animals transplanted with CSIIEG-transduced marrow maintained GFP expression in about 11% of the leukocyte compartment 4 weeks post-transplant (Fig. 5). Interestingly, at 16 weeks post-transplantation we found that CSII/AProGFP transduced cells in the peripheral blood exhibited a diminished level of GFP expression (average MFI=24.5) in comparison with the CSIIEG control (average MFI=1458) (p<0.001) (Fig. 5), similar to the shift in MFI observed in transduced 3T3 cells.
Secondary transplantation was carried out by harvesting marrow from each primary recipient 4 months post-transplant and infusing each marrow sample into three recipient animals preconditioned with 800 rads, cesium source. GFP expression was sustained in the secondary transplant recipients receiving either CSII/AProGFP or CSIIEG transduced marrow, demonstrating the capability of these lentiviral vectors to maintain gene expression after transduction into primitive hematopoietic stem cells engrafted in the primary recipients (Fig. 5). Since our intention is ultimately to use the human Artemis promoter to regulate gene expression for correction of SCID-A, we assayed for APro mediated expression in lymphocyte populations engrafted in secondary recipients. GFP expression was observed in both lymphoid and myeloid compartments, including B cells (B220+NK1.1−), T cells (CD3e+), and myeloid cells (CD11b) in animals that received marrow transduced with CSII/AProGFP (Fig. 5). These results verify that the Artemis promoter provides a moderate level of expression in differentiated lymphoid cells after lentiviral transduction into long-term repopulating hematopoeitic stem cells.
We isolated a 1-kb region of human genomic DNA directly upstream of the Artemis TSS by PCR amplification (APro). An in silico transcription factor binding site analysis of this sequence uncovered several transcription factor binding sites, leading us to further characterize the genomic region for promoter activity. 5′ RACE analysis indicated multiple transcriptional start points in RNA extracted from liver and lymphoid cell lines, and deletion analysis revealed divergent expression patterns between 293T cells and the BJAB Burkett lymphoma-like cell line. Transfection studies in vitro revealed the potential for this sequence to regulate gene expression in multiple cell types; additionally, APro mediated expression at a level that was substantially lower than that mediated by the strong EF1α promoter. Further, GFP expression regulated by APro was observed in vivo in mice transplanted with marrow that had been transduced with an APro-regulated lentiviral vector. Finally, expression was sustained in secondary transplant recipients in both myeloid and lymphoid lineages, establishing the effectiveness of APro to serve as a proficient promoter for gene expression within the hematopoeitic system.
The human APro region identified through this study exhibits several characteristics similar to those observed for promoters of other NHEJ proteins. The downstream promoter region extending from −184 to the human Artemis TSS contains neither a TATA box nor a CCAAT box, motifs also absent in the promoters for DNA-PKcs (Connelly et al., 1998), Ku70 (Takiguchi et al., 1996) and other housekeeping genes. Lack of a distinct TATA box within this region of the human APro sequence is associated with considerable variability in the location of transcriptional start sites, as determined by 5′ RACE, analogous to variability in transcriptional initiation sites observed for other NHEJ messages such as DNA-PKcs, in which at least 6 transcriptional start sites were observed (Connelly et al., 1998). We also tested full-length APro and deletion constructs transfected into 293T cells for the effect of irradiation or exposure to the radiomimetic chemicals bleomycin and etoposide, but induced expression upon DNA damage was not observed (data not shown). Similarly, neither transcription nor translation of DNA-PKcs are induced upon DNA damage in either human or mouse cells (Lee et al., 1997; Connelly et al., 1998). These observations indicate the similarity of human Artemis transcriptional regulation with that of other NHEJ components.
Our 5′ RACE results and the in vitro deletion mapping presented in this study suggest the potential for multiple regulatory regions that comprise the endogenous Artemis promoter. This is consistent with Artemis' involvement in several distinct cellular functions such as NHEJ (Ma et al., 2002), V(D)J recombination (Ma et al., 2002), and apoptosis (Britton et al., 2009). It is tempting to speculate that Artemis expression may be spatially and/or temporally regulated per required function, similar to several other proteins demonstrated to be differentially regulated through transcript initiation at multiple promoter sites. For example, the murine α-amylase gene exhibits tissue specific expression regulated by two separate promoters (Schibler et al., 1983). S1 nuclease mapping revealed that liver tissues yielded one minor α-amylase transcript, whereas the parotid gland yielded two α-amylase transcripts including the minor transcript plus one additional major transcript; additionally, synthesis of each transcript was found to initiate directly downstream of TATA boxes identified by sequence analysis (Schibler et al., 1983). Our data show two potential promoter regions regulating human Artemis expression: one containing a well-defined TATA box with the ability to regulate expression in B-lymphoid cells and the downstream promoter region being less defined in that it does not contain a TATA box, yet regulates expression in several other cell types. It will be of interest to study the relationship between Artemis function and tissue type expression and address the possibility that one region of the APro sequence is necessary for expression during V(D)J recombination, whereas the other region confers constitutive expression in support of NHEJ.
Gene transfer is emerging as a promising approach for treatment of genetic disorders. However, recently observed adverse events have called to attention the importance of regulating expression of therapeutic genes. Specifically, two independent studies have reported correction of X-linked SCID by ex vivo transduction of CD34+ hematopoeitic stem cells using a retroviral vector expressing the common cytokine-receptor gamma chain (common γ chain) (Gaspar et al., 2004; Hacein-Bey-Abina et al., 2010). Long-term engraftment of corrected stem cells was observed in the majority of patients, ultimately resulting in reconstitution of a functional lymphocyte compartment. To date, however, five out of twenty patients treated for X-linked SCID by gene transfer have developed T cell outgrowth resulting in leukemia, from which one child has died (Howe et al., 2008; Hacein-Bey-Abina et al., 2008). Although insertional activation of the LMO2 oncogene was reported in four of the leukemic cases, overexpression of the common γ chain induces cellular proliferation and thus may have contributed to the T lymphocyte clonal outgrowth (Howe et al., 2008; Amorosi et al., 2009). More tightly regulated expression of the common γ chain may thus reduce the risk of oncogenesis resulting from common γ chain overexpression.
Achieving transgenic expression of human Artemis for the correction of SCID-A will be challenging considering previous results regarding Artemis overexpression. Recently, two independent groups reported correction of a murine model of SCID-A by lentiviral vector-mediated gene transfer (Mostoslavsky et al., 2006; Benjelloun et al., 2008). In both studies, SCID-A animals receiving HSC transduced with a lentiviral vector encoding a human Artemis cDNA regulated by the PGK promoter exhibited repopulation of both B and T lymphocyte compartments. However, Mostoslavsky et al. reported the inability of either CMV or EF1α regulated human Artemis lentiviral vectors to restore B and T cells in RAG-1 deficient animals receiving SCID-A HSCs transduced with either vector (Mostoslavsky et al., 2006; Benjelloun et al., 2008). Considering the endonucleolytic nature of Artemis, these results lead us to consider whether Artemis overexpression might be inherently toxic. We subsequently characterized the effect of Artemis overexpression and found it to be associated with cytotoxicity, ultimately resulting in a halt in cell cycle progression, fragmentation of genomic DNA, and apoptosis (Multhaup et al., 2010). These results emphasize the importance of providing Artemis expression at a level that is nontoxic and yet sufficient to correct the T−B− phenotype in preclinical studies and in clinical application to human SCID-A.
The endogenous human Artemis promoter provides expression both in vitro and in vivo at levels significantly lower than those of well-characterized strong promoters such as CMV and EF1α. Further, when studied in vivo, the Artemis promoter regulated expression in several lymphoid cell populations, demonstrating the ability of Artemis promoter to mediate expression in important hematopoeitic compartments after gene transfer into hematopoietic stem cells. The APro sequence thus has great potential as a regulator of therapeutic gene expression, including expression of the human Artemis gene for in vivo correction of human SCID-A. With this in mind, we have recently generated lentiviral vectors employing innate regulation of human Artemis cDNA using its own endogenous promoter sequence and have achieved successful ex vivo gene transfer resulting in functional lymphocyte repopulation of a murine model of SCID-A (M. Multhaup et al., manuscript in preparation). Overall, these results demonstrate that providing innate Artemis expression via ex vivo lentiviral transduction into hematopoietic stem cells may serve as a clinically relevant and feasible treatment of human SCID-A. As an element that supports moderate but reliable levels of expression, the human APro also has the potential for reduced genotoxicity associated with integration of other therapeutic genes.
We acknowledge the assistance of the Flow Cytometry Core Facility of the Masonic Cancer Center, University of Minnesota, a comprehensive cancer center designated by the National Cancer Institute, supported in part by P30 CA77598. This work was supported by Public Health Service grant ROI A1063340 from the National Institute of Allergy and Infectious Disease.
The authors declare that no competing financial interests exist.