Lentiviral vector integration in the human genome induces alternative splicing and generates aberrant transcripts

Abstract

Retroviral vectors integrate in genes and regulatory elements and may cause transcriptional deregulation of gene expression in target cells. Integration into transcribed genes also has the potential to deregulate gene expression at the posttranscriptional level by interfering with splicing and polyadenylation of primary transcripts. To examine the impact of retroviral vector integration on transcript splicing, we transduced primary human cells or cultured cells with HIV-derived vectors carrying a reporter gene or a human β-globin gene under the control of a reduced-size locus-control region (LCR). Cells were randomly cloned and integration sites were determined in individual clones. We identified aberrantly spliced, chimeric transcripts in more than half of the targeted genes in all cell types. Chimeric transcripts were generated through the use of constitutive and cryptic splice sites in the HIV 5ι long terminal repeat and gag gene as well as in the β-globin gene and LCR. Compared with constitutively spliced transcripts, most aberrant transcripts accumulated at a low level, at least in part as a consequence of nonsense-mediated mRNA degradation. A limited set of cryptic splice sites caused the majority of aberrant splicing events, providing a strategy for recoding lentiviral vector backbones and transgenes to reduce their potential posttranscriptional genotoxicity.

Figure 1. Schematic maps of the SIN LVs in their proviral forms.

The promoter-less IRES-GFP vector contains a GFP gene inserted downstream an internal ribosomal entry site (IRES). In the CMV.GFP and K14.GFP vectors, the GFP gene is under the control of an internal, immediate-early CMV promoter or a human K14 enhancer-promoter element (K14). The GLOBE vector contains the human β-globin gene under the control of the β-globin promoter (β-p) and the HS2 and HS3 element of the β-globin LCR, in reverse transcriptional orientation. LTRs are represented by black boxes, where Δ indicates the U3 –18 deletion. RRE, rev-responsive element; SD1, HIV gag major SD site; SA7, HIV gag major SA site; (A)_n, polyadenylation signal; wPRE, woodchuck hepatitis posttranscriptional regulatory element. I, II, and III exons are indicated.

Figure 2. 5′RACE and RT-PCR analysis of aberrantly spliced transcripts in clones of Jurkat and SupT1 T cells, primary T cells, HaCaT keratinocytes, and myeloid HEL cells.

(A) Schematic structure of the integrated provirus. The “expression cassette” represents the IRES-GFP splice trap in Jurkat and SupT1 cells, the CMV-GFP cassette in transduced T cells, the K14-GFP cassette in HaCaT cells, and the β-globin gene in reverse orientation in HEL cells (see Figure 1). E(n) indicates the exon upstream of the lentiviral integration, and E(n+1) indicates the downstream exon. SD1, gag major SD site; SA7, gag major SA site. Arrows indicate the vector-specific primer used to reverse transcribe the mRNAs into cDNAs (Lenti-RT), the upstream exon-specific forward primer (E-for), and the vector-specific reverse primer (Lenti-rev) used in the RT-PCR reaction. (B) 5′ RACE and RT-PCR products in representative SupT1/Jurkat, HaCaT, T lymphocytes, and HEL clones stained by ethidium bromide on 1% agarose gels. Transcripts were amplified using the Lenti-rev primer and a primer specific for the upstream exon of the gene identified on top of each lane. Molecular weight markers (sizes in bp) are indicated on the left of each gel.

Figure 3. Analysis of splicing variant types and mapping of cryptic HIV SA sites.

(A) Schematic view of the families of chimeric transcripts generated by alternative splicing to HIV constitutive (type 1) or cryptic (type 2) SA sites or to cryptic SA sites located in the upstream intron (type 3), as identified from sequencing of the PCR products shown in Figure 2B. Exons are indicated by continuous lines, spliced sequences are indicated by dotted lines. (B) Mapping of the cryptic HIV SA sites (red triangles) identified (A) in the intron upstream of the vector integration site and (B–D) in the HIV U5 region of the LTR, (E and F) primer binding site, (G) packaging signal (Ψ), and (H and I) gag gene (SA7). The SA7 site is the gag constitutive, REV-sensitive acceptor site. The dinucleotides at the end of a spliced sequence are indicated in red. The GT dinucleotide at the beginning of the constitutive gag intron is indicated in blue. The U3, R, and U5 region of the 5′LTR are underlined. The frequency of SA site usage in all the sequenced transcripts is reported in Table 2.

Figure 4. Semiquantitative PCR analysis of wild-type and aberrantly spliced transcripts from the HCFC2, PLEKHA7, and DOCK5 genes in HaCaT clones transduced with the K14-GFP vector.

(A) cDNAs were prepared using random hexamer primers from Poly(A)⁺ RNA. Wild-type transcripts were amplified using the E-for and E-rev primers (arrows) annealing to the exons upstream and downstream of the provirus. Fusion transcripts were amplified using the E-for and Lenti-rev primers. (B) PCR reactions were arrested at 24, 28, and 33 cycles and run on 1% agarose gels in the following order (from left): wild-type transcript amplified from the HaCaT clone; wild-type transcript amplified from a HaCaT bulk culture; fusion transcript(s) in the HaCaT clone; and GAPDH transcript in the HaCaT clone, used for signal normalization. The GAPDH transcript in the HaCaT bulk culture was run on each gel but shown only once at the right of all panels. Transcripts were ranked in 4 arbitrary classes of relative abundance, i.e., low, when fusion transcripts were detected 8 PCR cycles later than wild-type transcripts; intermediate, when fusion transcripts were detected 4 PCR cycles later than wild-type transcripts; and high, when chimeric and wild-type transcripts were detected after the same number of PCR cycles. A fusion transcript was classified as rare (data not shown) when it was undetectable after 33 PCR cycles starting from RNA reverse transcribed with random hexamers, although it was detected and sequenced using RNA reverse transcribed with the vector-specific Lenti-RT primer (Figure 2).

Figure 5. Summary of the relative frequency (percentage) of the 4 classes of abundance of aberrantly spliced transcripts recovered in all analyzed cell clones.

(A) Aberrant transcripts generated from LV proviruses integrated in forward orientation in Jurkat/SupT1, primary T cell, HaCaT, and HEL clones, ranked in the high (red), intermediate (yellow), low (blue), rare (gray), and absent (black) abundance classes, as defined by the semiquantitative PCR assay shown in Figure 4. (B) Aberrant transcripts generated from the GLOBE proviruses integrated in reverse orientation in HEL clones. The total number of analyzed proviruses (n) is indicated above each bar.

Figure 6. RT-PCR analysis of aberrantly spliced transcripts in random, unselected clones of HEL cells transduced with the GLOBE vector.

(A) Schematic structure of the GLOBE provirus integrated between exons E(n) and E(n+1) in reverse transcriptional orientation. SD, SD site; SA, SA site. E-for, E-rev, and Globin-rev primers are indicated by arrows. The β-globin HS3 and HS2 LCR elements; promoter; I, II, and III exons; and polyadenylation signal are indicated. (B) Schematic view of the families of chimeric transcripts generated by alternative splicing to the β-globin first intron constitutive SA site (type 4) or to the β-globin promoter (type 5) and HS3 (type 6 and 7) cryptic SA sites, as identified from sequencing of the PCR products. Exons are indicated by continuous lines, spliced sequences are indicated by dotted lines. (C) Mapping of the cryptic SA (red triangles) and SD (blue triangles) sites identified in the HS3 element and the 5′ UTR of the β-globin gene. The dinucleotides at the beginning or end of a spliced sequence are indicated in blue and red, respectively. The frequency of SD and SA site usage in all the sequenced transcripts is reported in Table 3.

Figure 7. Semiquantitative PCR analysis of wild-type and aberrantly spliced transcripts from the EFR3A, QARS, and HMG20A genes in HEL clones transduced with the GLOBE vector.

cDNAs were prepared using random hexamer primers from Poly(A)⁺ RNA. Wild-type transcripts were amplified using the E-for and E-rev primers, annealing to the exons immediately upstream and downstream of the provirus. Fusion transcripts were amplified using the E-for and Globin-rev primers. PCR reactions were stopped at 24, 28, and 33 cycles and run on 1% agarose gels in the following order: wild-type transcript amplified from the HEL clone (first lane from left); wild-type transcript amplified from a HEL bulk culture (second lane); fusion transcript in the HEL clone (third lane); GAPDH transcript in the HEL clone, used for signal normalization (fourth lane); and GAPDH transcript in the HEL bulk culture (fifth lane). Transcripts were ranked in 4 arbitrary classes of relative abundance, i.e., rare, low, intermediate, and high, as described in Figure 4.

Figure 8. Analysis of the strength of the constitutive SA signals in the introns targeted by LV integration.

enoLOGOS plots of the consensus sequences (49 bp) encompassing the SA sites of the exons downstream of the integration sites of proviruses generating rare (top panel) or relatively abundant (middle panel) fusion transcripts. The human SA consensus sequence (23 bp) is shown in the bottom panel. Nucleotide positions are conventionally numbered, starting from the intron/exon boundary (intron, –1 to –46; exon, +1 to +3). For each position, the height of the letter represents the frequency of the corresponding base at that position. A schematic structure of the integrated provirus with upstream and downstream exons is shown in the top panel.

Figure 9. RT-PCR analysis of aberrantly spliced transcripts from the PBX4, PLEKHA7, CNOT6, FLYWCH1, and SYMPK genes in cycloheximide-treated cell clones.

(A) HaCaT and (B) HEL clones were treated with 100 or 50 μg/ml cycloheximide (CHX), respectively, and cDNAs were prepared from poly(A)⁺ RNA using random hexamer primers. Wild-type transcripts were amplified using E-for and E-rev primers (Figure 4), fusion transcripts were amplified using the E-for and Lenti-rev primers (Figure 4) in HaCaT clones and E-for and Globin-rev primers (Figure 6A) in HEL clones. PCR reactions were stopped at 24, 28, and 33 cycles and run on 1% agarose gels in the following order: wild-type transcript amplified from the untreated HaCaT/HEL clone (first lane from left); fusion transcript in the untreated HaCaT/HEL clone (second lane); wild-type transcript amplified from the cycloheximide-treated HaCaT/HEL clone (third lane); fusion transcript in the cycloheximide-treated HaCaT/HEL clone (fourth lane); and GAPDH transcript in the untreated (fifth lane) and cycloheximide-treated (sixth lane) HaCaT/HEL clone, used for signal normalization.