A Novel Combination of Chicken Hypersensitive Site-4 Insulator Elements Improves Titers and Restores Full Insulator Activity.

Abstract 3566 Poster Board III-503 Chromatin insulators separate active transcriptional domains and block the spread of heterochromatin in the genome. A prototypic insulator, the 1.2Kb chicken hypersensitive site-4 (cHS4) element utilizes CTCF and USF-1/2 motifs in the proximal 250bp. This smaller “core” cHS4 element provides enhancer blocking activity and reduces position effects. However, the cHS4 core sequences alone do not insulate viral vectors effectively. Two copies of the core, while effective in plasmid based systems, are unstable in viral vectors. In contrast, the full-length cHS4 has excellent insulating properties, but its large size severely compromises vector titers (Urbinati et al, Mol Ther 2009). Therefore, we performed a structure-function analysis of the full-length cHS4 in the context of self inactivating lentivirus-vectors to identify minimal insulator elements required for optimal insulation. Specifically, we analyzed transgene expression in the clonal progeny of primary murine hematopoietic stem cells using the secondary bone marrow transplant assay, and analyzed epigenetic changes in cHS4 and the transgene promoter in vitro in clonal integrants. As expected, the full length cHS4 insulator reduced clonal variegation in transgene expression, reduced position effects and blocked silencing-associated epigenetic modifications over the insulator core and the transgene promoter. However, while either the 5′ cHS4 250bp core or the 3′ cHS4 400bp sequences similarly effected only lower clonal variegation in transgene expression, when they were combined these 650bp sequences recapitulated the activity of the full length 1.2kb insulator, with minimal impact on viral titer. The distal 3′ 400bp fragment contains no consensus sites for USF or CTCF. However, ChIP analysis on proviruses carrying only the 3′ 400bp showed that it binds CTCF. USF-1 binding to the 3′ 400bp, however, only occurred when both the 5′ 250bp core and the 3′ 400bp fragment were present in the proviruses. Indeed, the silencing associated epigenetic marks over the 3′ 400bp region were blocked only when the vector carried both these ends of cHS4 insulator sequences, suggesting that USF-1 bridges the 5′ core and the 3′ 400bp to confer full insulator activity. Furthermore, the 650bp sequences or the full length insulator had the maximal reduction in clonal dominance in the in vitro immortalization assay of lineage negative primary murine hematopoietic cells (Arumugam et al, Mol Ther. In press). Our studies confirm and extend earlier observations on the 5′ 250bp insulator core and identify a new “core-like” insulator activity in the 3′ end of cHS4. The specific elements in the 3′ 400bp sequences that promote interaction with the 5′ 250bp sequences would be important to determine, and may be present in other insulators in and across the genome/s. In the meanwhile, new vector systems flanked by this optimized ‘650bp’ cHS4 sequences, can provide excellent insulation of the transgene without significant loss in viral titers and have important safety and efficacy implications for gene therapy. Our data have important implications in understanding the molecular basis of insulator function and design of gene therapy vectors. Disclosures: No relevant conflicts of interest to declare.

Mechanism of Reduction in Titers from Lentivirus Vectors Carrying Chromatin Insulator Elements in the 3′ LTR

Random integration of viral vectors can result in undesirable activation of surrounding genes by enhancers in the vectors (vector genotoxicity), or result in variable expression due to effects of surrounding chromatin on the vector transgene (chromatin position effects). Vector genotoxicity has become an area of intense study since the occurrence of gene therapy related leukemias in 5 patients in the French X-SCID trial. Additionally, we find consistent and therefore 2–3 fold higher expression from vectors insulated by the chicken b-globin hypersensitive site (cHS4) insulator; and these vectors achieve therapeutic correction in human b0-thalassemia major, where a very high transgene expression is necessary. However, vectors that insulate the correcting transgene from position effects and genotoxicity significantly compromise viral titers by an order of magnitude. In order to define the mechanism by which this occurs and improve the titers of insulated vector systems, we placed the 1.2Kb cHS4 insulator or different regions of cHS4 that may have insulator activity and/or inert DNA spacers in the 3′LTR of self-inactivating lentiviruses carrying a large b-globin transgene and regulatory elements. We also lengthened the β-globin lentivirus vector by an additional 1.2Kb, by insertion of an internal transgene cassette. We found that addition of 1.2Kb transgene internally to a large “globin” vector did not reduce vector infectious titers. However, when cHS4 sequences or inert DNA spacers of increasing size were placed in the 3′LTR, infectious titers decreased proportional to the length of the insert. This effect occurred regardless of the type of sequence inserted in the 3′ LTR. Vectors carrying the1.2Kb cHS4 or l-DNA spacer in the 3′ LTR had the lowest titers. We next examined the stage of the vector life-cycle affected by large LTR inserts in packaging cells, the quantity and quality of the virus particles generated, and post-entry viral steps in target cells. Equal amounts of full-length viral genomic transcripts were produced in the packaging cells with vectors with or without the 1.2Kb insulator insert. All insertion in the 3′ LTR are placed in the U3 enhancer deleted region, proximal to the viral polyadenylation signal. Also, self-inactivating vectors have a U3 enhancer deletion that also deletes enhancers of polyadenylation. However, despite the insertion of cHS4 elements in this region, no increase in viral readthrough transcription (as measured by northern blot analysis and an enzyme-based assay) occurred with vectors carrying the large 1.2Kb insert. Packaging efficiency was also identical with insulated and uninsulated vectors. Similar degree of viral genome encapsidation occurred, as measured by p24 ELISA, virus associated reverse transcriptase and viral RNA analysis, demonstrating that similar amounts of intact viral particles were produced with insulated and uninsulated vector plasmids. However, lentiviruses carrying the 1.2Kb insert in the 3′LTR were inefficiently processed following target-cell entry, with reduced reverse transcription and integration efficiency. This primarily occurred from increased homologous recombination resulting in increased 1-LTR circles of the insulated vector viral DNA; resulting in reduced vector integration and hence lower transduction/infectious titers. Thus, we found that large inserts in the viral 3′ LTR are packaged efficiently, but have inefficient post-entry viral mRNA processing. In a parallel study, we also did a structure-function analysis of cHS4 fragments and identified key elements necessary for optimal insulation. Vectors constructed with this minimized 650bp cHS4 sequences had a minimal reduction in titers, yet retained full insulator activity. These studies have important implications in the design of gamma-retrovirus or lentivirus vectors with insulator, transgenes or enhancer inserts into the 3′ LTR. [FU and PA contributed equally to this work]

Mechanism of Reduction in Titers with Vectors Flanked by Chromatin Insulator Elements

Random integration of viral vectors can result in undesirable activation of surrounding genes by enhancers in the vectors (vector genotoxicity), or result in variable expression due to effects of surrounding chromatin on the vector transgene (position effects). Vector genotoxicity has become an area of intense study since the occurrence of gene therapy related leukemias in patients in the French X-SCID trial. Additionally, variability in expression has been shown to compromize therapeutic efficacy in gene therapy for beta-thalassemia, where a high consistent level of expression is necessary. Vectors flanked by the cHS4 insulator, one of the best characterized insulator element, can reduce vector genotoxicity (Neinhuis et al, 2007) and chromatin position effects in γ-retrovirus (Rivella et al, 2000; Yannaki et al, 2002) and lentivirus vectors (Arumugam et al, 2007). Despite these favorable features, the full-length cHS4 insulator, necessary for optimal insulator activity, is infrequently used because it lowers vector infectious titers by 10–15 fold (Ramezani et al, 2003). This reduction in titers is especially limiting with vectors carrying large inserts. We analyzed mechanisms by which this occurred. Insertion of an additional 1.2kb internal cassette to the large human β-globin-LCR (hβ-LCR) lentiviral (LV) vectors did not reduce vector infectious titers. However, insertion of the 1.2Kb cHS4 element reduced titers by more than an order of magntude; showing that reduction in titers by cHS4 was not secondary to a further lengthening of vector genome but this occurred either due to a large insert in the 3′LTR or specific cHS4 sequences that bind cellular factors and hinder viral genome transcription. We then inserted varying sized fragments of cHS4, tandem repeats of the cHS4 core element, or inert DNA spacers in the 3′LTR of the hβ-LCR LV vector, sBG, resulting in vectors with 3′LTR inserts of 0, 250, 400, 500, 650, 800 and 1200bp. After a threshold length of 650bp, infectious titers fell proportional to the size of the insert into the LTR. The same effect was seen with inert DNA spacer elements, showing that this phenomenon was not sequence-specific. We next examined the stage of the vector life-cycle affected by large LTR inserts: lengthening of the 3′LTR did not increase viral readthrough transcription, as measured by northern blot analysis and an enzyme-based assay for readthrough transcription (Higashimoto et al, 2007). Equal amounts of full-length viral genomic transcripts were produced in the packaging cells with vectors with and without the insulator. Similar degree of viral genome encapsidation occurred, as measured by p24 ELISA, virus associated reverse transcriptase and viral RNA analysis, demonstrating that similar amounts of intact viral particles were produced with insulated and uninsulated vector plasmids. However, the insulated vector was inefficiently processed following target cell entry, resulting in less integrated vector; and thus lowers infectious titers. Of note, a vector carrying tandem repeats of the cHS4 core (two 250bp repeats) resulted in increased rate of recombination, with deletion of the insulator core at a high frequency. Thus, we found that large inserts in the viral 3′LTR are packaged efficiently, but have inefficient post-entry viral mRNA processing. These studies have important implications in the design of γ-retrovirus and LV vectors with insulator and other transgene/promoter/enhancer inserts into the LTR.

Elements of a Chromatin Insulator Necessary for Consistent, High-Level Globin Gene Expression

Vectors flanked by the chicken b-globin locus hypersensitive site-4 (cHS4) insulator provide consistent, improved expression (chromatin barrier activity) and reduced genotoxicity (enhancer-blocking activity); features important for randomly integrating viral vectors. However, full-length 1.2Kb cHS4 reduces vector titers significantly. Initial studies mapped the enhancer blocking activity of cHS4 to the first 250bp ‘core’ in a copy-number dependent manner (Felsenfeld and colleagues). However, elements required for the barrier effect are not well defined. We studied various regions of cHS4 for chromatin barrier effect. Beta-globin-LCR lentivirus vectors were constructed with the core (sBGC), 2 copies of the core (sBG2C), 400bp (sBG400) and 800bp (sBG800) fragments (both including the core) and compared to the un-insulated vector (sBG) and the vector with the full-length cHS4 (sBG-I). Barrier activity (improved probability and consistency of expression from individual integrants) in MEL cells and the clonal progeny of hematopoietic stem cells was analyzed. Single-copy MEL clones showed that sBG-I clones had a significantly higher proportion of hβ+ cells/clone. However, proportion of hβ+ cells from all other vectors carrying various fragments of cHS4 were not significantly different from the un-insulated vector (Table 1). Interestingly, all clones carrying a fragment containing the core showed reduced clonal variegation (coefficient of variation of hβ expression). We next analyzed hβ expression from the sBG, sBGC, sBG2C, sBG400 and sBG-I vectors in primary and secondary thalassemia (Hbb th3/+) mice. In the primary mice and secondary CFU-S analyzed thus far (Table 2), it appears that all vectors with cHS4 insulator fragments containing the core provide some barrier activity, but the full-length insulator is necessary for maximum barrier effect. To corroborate whether elements outside the core contribute to barrier activity, we also tested vectors with inert DNA spacers in addition to the core (sBG400−S, sBG-800−S and sBG1200−S). Single copy MEL clones confirmed that the sBG400−S, sBG800−S and 1200−S vectors had significantly lower proportion of hβ+ cells/clone [41±4.4%, 41±7% and 28±6% respectively, n=36]. These data suggest that regions of 950 bp region of cHS4 outside the core contains elements that cooperate with the core for optimal barrier activity. These studies have important implications for vector design for gene therapy for thalassemia, where variability in expression from uninsulated vectors compromizes efficacy. Table 1. β globin expression in MEL cell clones from vectors carrying portions of the cHS4 insulator sBG sBG c sBG 2c sBG 400 sBG 800 sBG-I Values are expressed as Mean±SEM, *P<0.05, **P<0.01 by ANOVA (Dunnet: comparisons of all vectors to the control vector, sBG) % hβ+ cells 59±5 49±6 51±3 51±3 49±6 84±3** CV 53±4 37±3** 40±2** 42±1.5** 42±3** 36±2** Table 2. β globin expression in primary thalassemia mice and in the clonal progeny of hematopoietic stem cells of secondary mice Primary mice (16 weeks) Mock sBG sBG c sBG 2c sBG 400 sBG-I Values are expressed as Mean ± SEM, *P<0.05, **P<0.01 by ANOVA (Dunnett: comparison of all vectors to the control, sBG) HCT (%) 20±3 32±3 36±2 37±2 37±3 39±2 Reticulocytes (%) 28±2 16±5 11±1** 13±6** 7±2** 7±2** %HbA (mαhβ by HPLC)/vector copy NA 32±19 24±5 29±3 31±3 43±4* % hβ+RBC/ vector copy NA 36±11 57±8* 67±13* 76±17* 104±18** Secondary CFU-S (Single Copy) % hβ+ cells/clone NA 26±3 45±3* 34±2 48±9* 75±3** CV NA 88±9 68±2* 71±2* 72±3 61±2**