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.