The hepatitis C virus internal ribosome entry site facilitates efficient protein synthesis in blood vessel endothelium during tumour angiogenesis

Abstract

The development of gene delivery systems for therapeutic use involves vectors (often retrovirus or adenovirus) which typically encode one target protein, but the use of internal ribosome entry sites (IRES) can confer the ability to express more than one protein from bi- or polycistronic mRNAs. IRES elements can display tissue-specific expression, so it is necessary to determine suitable IRES for specific clinical applicability. Blood vessel endothelial cells are important clinically since many different conditions involve neo-vascularisation (angiogenesis). We have demonstrated that the viral hepatitis C IRES element is a powerful mediator of protein synthesis in angiogenesis, such as found in solid tumours. Homologous recombination was used to introduce IRES-lacZ sequences into the Lmo2 gene, which is expressed in endothelial cells. β-Galactosidase expression was determined during vascular remodelling in mouse embryos and in sprouting endothelium during growth of solid tumours, and showed that the hepatitis C IRES is used efficiently for protein synthesis in endothelial cells. This IRES element can provide the means to express two or more therapeutic genes in blood vessel endothelium in clinical conditions, such as cancer, which depend on angiogenesis.

INTRODUCTION

Specific expression of genes following delivery to target cells is a key objective for molecular-based therapies, and blood vessel endothelial cells are important targets in a variety of clinical indications. The remodelling of established vasculature (angiogenesis) occurs at specific times in normal conditions, such as wound healing or menstruation, and in a number of conditions, such as cancer, ischaemia, diabetic retinopathy and inflammatory diseases (1). In addition, neovascularisation, establishment of collateral circulation in ischaemic diseases and formation of granulation tissue in inflammatory diseases are a hallmark of these pathological conditions (1,2). In cancer, angiogenesis is important to support in situ growth of primary solid tumours and also in metastasis where tumour cells traverse the infiltrating blood vessels (3), enter the blood stream and eventually form metastatic deposits in remote locations. The process of angiogenesis has been advocated as an important point of therapeutic intervention in cancer (4), as depleting growing tumours of blood supply can inhibit their growth. Many molecular targets will be confined to intracellular locations which require access of molecular therapeutics to the inside of the cell. This can be achieved with the use of viral vectors such as adenovirus (5) or lipid formulations which carry DNA vectors across the plasma membrane (6). These impose restrictions in that they are usually designed to express a single gene within the target cell. The use of internal ribosome entry sites (IRES) in bicistronic mRNAs has been a common strategy for dual gene expression, but these structural elements in mRNA display developmental and cell type specificity (7,8).

In view of the importance of gene expression in endothelial cells, we have assessed the potential of IRES sequences to achieve high level expression in sprouting endothelial cells in vivo. Using IRES-lacZ cassettes introduced into the Lmo2 gene [which is expressed in embryonic and tumour endothelial cells (9,10)], we have examined β-galactosidase expression during vascular remodelling in embryos and during the growth of solid tumours. Our results show that the hepatitis C (HC) virus IRES, but not the encephalomyocarditis (EMC) IRES, is used efficiently for protein synthesis in endothelial cells, proving the means to co-express proteins in blood vessel endothelium in clinical conditions which depend on angiogenesis.

MATERIALS AND METHODS

Plasmid preparation and gene targeting

The plasmids for homologous recombination in the Lmo2 gene were based on pKO5tk, which has a unique BamHI restriction site mutated in exon 2 (11). The HC-IRES-lacZ Lmo2 knock-in targeting clone was prepared by inserting an HC-IRES-LacZ-MC1neopA cassette into the BamHI site of pKO5tk. A 400 bp BamHI fragment including the HC-IRES (12) was first cloned into the BglII–BamHI sites of a modified pBSpt vector (pBspt-BGB4) to generate the precursor pBSpt-HC-IRES with a unique BamHI site into which was cloned the lacZ gene and pMC1-neo-pA (13). The EMC-IRES Lmo2 knock-in targeting clone was prepared by inserting the lacZ gene fragment into pEMC IRES and addition of pMC1neo-pA, and this cassette was cloned into pKO5tk. The in-frame fusion of lacZ with exon 2 of the Lmo2 gene has been described previously, as have the generation and characterisation of the germline mouse carriers of the targeted allele (14).

Generation and analysis of gene targeted mice

Embryonic stem (ES) cells (CCB) were transfected and selected for G418 resistance and gancyclovir sensitivity as described (11) and targeted clones characterised by Southern filter hybridisation using two external probes, A and B (Fig. 1A). Targeted clones were injected into C57Bl6 blastocysts, and chimaeric mice were generated, from which germline transmission was obtained by breeding male chimeras with C57Bl6 females. Timed matings were set up between heterozygous mice carrying one of the three Lmo2 knock-in alleles and wild-type C57Bl6 mice. At the appropriate times, the pregnant females were euthanased, embryos removed and whole mount stained with X-gal to detect β-galactosidase as described (10). Post-fixed embryos (10% formalin) were sectioned after wax embedding. Sections (4 µM) were mounted on microscope slides and counter-stained with haematoxylin and eosin. Detection of the endothelial marker PECAM (CD31) was carried out using MEC13.3 anti-CD31 antibody (Pharmingen) by the avidin– biotin-conjugated peroxidase method as described (15).

Figure 1.

Activity of IRES elements in mouse embryo vascular endothelium. (A) Constructs for homologous recombination. Top line: the partial restriction map of the mouse Lmo2 gene shows the location of Lmo2 exons 2 and 3, together with the two probes (A and B) used to detect homologous recombination (10). Middle line: a map of the targeting vector pKO5tk (10) which has a BamHI restriction site introduced within exon 2 to facilitate cloning of exogenous elements into Lmo2. Bottom line: the maps of the lacZ gene insertions cloned in the exon 2 BamHI site for Lmo2-lacZ (in-frame lacZ fusion with the 5′ end of Lmo2) (14), HC-IRES and EMC-IRES. (B) Whole-mount X-gal staining of mouse embryos at embryonic stages E9.5, E10.5 and E12.5 showing expression of β-galactosidase from the Lmo2 gene in de novo capillary formation (vasculogenesis) and endothelial remodelling (angiogenesis) during mouse embryo development. Wt = wild-type C57Bl6

Tumour endothelial cell analysis

Lewis lung carcinoma cells were injected into both flanks of mice from each of the Lmo2 knock-in mouse lines or C57Bl6 controls (about 10⁶ cells per site). When primary site solid tumours reached ∼1 cm size, the recipient mice were euthanased, tumours resected and whole-mount X-gal staining carried out as for the embryos. After post-fixation in 10% formalin, sections were prepared from wax-embedded specimens, and 4 µM sections were mounted and counter-stained with haematoxylin and eosin.

RESULTS

Efficiency of HC-IRES in vascular endothelium during embryogenesis

A range of IRES elements potentially could be used to express proteins in endothelial cells (16). We compared the ability of the HC virus IRES and EMC virus IRES to facilitate protein synthesis in blood vessel endothelial cells during embryonic development. The Lmo2 gene is expressed in and is necessary for sprouting endothelium in embryogenesis (9) and tumour growth (10). We chose this gene as a test situation for expressing bicistronic mRNA species in endothelial cells in vivo since the mouse Lmo2 gene is amenable to gene targeting in ES cells (11). We have created two lines of mice in which the expression of lacZ is controlled from an IRES element in the mRNA, namely HC-IRES and EMC-IRES lines, respectively (Fig. 1A). In addition, we compared the Lmo2-lacZ mouse line in which an in-frame fusion has been made between the lacZ gene and Lmo2 (9). Timed matings were established for the three lines and embryos were whole mount stained with X-gal to detect β-galactosidase activity at embryonic day E9.5, 10.5 and 12.5 (Fig. 1B). As previously reported (9), the developing vasculature of the Lmo2-lacZ embryos expresses the Lmo2 gene which can be detected readily via the β-galactosidase reporter. No β-galactosidase activity was detected in wild-type embryo littermates (Fig. 1B). In the developing Lmo2-lacZ embryos, β-galactosidase is widely expressed in blood vessels, being found in whole body developing vasculature which coincides with expression of the pan-endothelial marker PECAM/CD31, detected with anti-CD31 antibodies in histological sections of embryos at E10.5 (Fig. 2, top panels).

Figure 2.

Histology of Lmo2-lacZ knock-in E10.5 embryos shows co-expression of β-galactosidase and the pan-endothelial marker CD31. E10.5 embryo specimens were whole mount stained with X-gal (Fig. 2), sectioned (4 µM), and counter-stained with haematoxylin and eosin. CD31 protein expression was detected in serial sections using anti-CD31 antibody and peroxidase. The montage shows embryo sections from each indicated Lmo2 knock-in mouse line, or wild-type (wt) controls stained only with X-gal (left) or co-stained with X-gal and anti-CD31 (right). Arrowheads indicate endothelial cells lining the blood vessel walls.

The levels of β-galactosidase reporter expression in the knock-in mouse lines with the Lmo2-HC-IRES-lacZ gene were less than in the direct lacZ gene knock-in Lmo2 (Fig. 1B), but the detectable β-galactosidase in the blood vessel endothelial cells in the EMC-lacZ mice was very low and indeed virtually undetectable at embryonic day E10.5 (Figs 1B and 2). This suggests that the EMC virus IRES is unsuitable for endothelial expression in vivo. The HC-IRES, on the other hand, yielded readily detectable levels of β-galactosidase activity. By embryonic day E12.5, profound levels of endothelial expression had occurred, indicating that the HC-IRES was used efficiently by the protein synthesis machinery of endothelial cells of mouse embryos.

The HC-IRES mediates endothelial protein synthesis in tumour angiogenesis

Angiogenesis is a target of cancer therapy (4,17,18), requiring targeting of anti-endothelial reagents to these specific cells. The efficacy of the HC-IRES in tumour blood vessels was tested using the lacZ knock-in mouse lines to support growth of tumour grafts which become vascularised by sprouting of existing blood vessels from the host. Lewis lung carcinoma cells were injected subcutaneously into the Lmo2-lacZ and HC-IRES mice (and C57Bl6 wild-type controls), and solid tumours were allowed to develop in situ at the site of injection. As the vascularisation of these tumours is contributed by the recipient mouse, the blood vessel endothelium would be expected to express the Lmo2-based lacZ reporter (Fig. 3A). This was analysed by staining isolated tumours with X-gal and histological sectioning to examine endothelial expression. Figure 3B shows a comparison of sections made from X-gal-stained tumours of the three sources, showing that the Lmo2-lacZ and HC-IRES-transplanted tumours had comparable levels of β-galactosidase activity in this situation. The HC-IRES therefore has significant activity in the developing vasculature of tumours.

Figure 3.

Expression of β-galactosidase from HC IRES in Lewis lung solid tumours. Lewis lung carcinoma cells were implanted subcutaneously in the Lmo2 HC-IRES knock-in mouse line and Lmo2-lacZ or wild-type (wt) controls. (A) The vasculature of the tumours growing in the recipient mice comes from the latter, and therefore the endothelial cells will be expressing the Lmo2 reporter of the recipient. In the case of Lmo2-lacZ and HC-IRES mouse lines, the expression of β-galactosidase is detected using X-gal substrate. (B) After tumour growth, solid tumours were whole mount stained with X-gal, and 4 µM sections made for examination of tumour vascular endothelium which forms by sprouting of the existing endothelium from recipient mice. Arrowheads indicate endothelial cells lining the blood vessel walls.

DISCUSSION

IRES have evolved in viruses and allow viruses to express more than one gene per mRNA, and the cell types in which this activity occurs depends on the virus. IRES elements bind to cellular protein factors, and these can be cell type specific, lending an internal degree of specificity to the system. The corollary is that, as a therapeutic or molecular biology methodology, use of IRES elements in the design of vectors for in vivo genes must take cell specificity into account. Thus, if the objective is targeting protein synthesis within endothelial cells, a suitably efficient IRES must be employed. Our data show that for blood vessel endothelial expression, the HC-IRES is used efficiently.

There are a number of important clinical indications where angiogenesis is an important consequence. Neovascularisation occurs around malignant tumours in order to supply enough oxygen and CO₂ exchange for rapidly dividing cells (18). In chronic inflammatory diseases such as rheumatoid arthritis, sustained inflammation results in the formation of vascular-rich granulation tissues in the synovial membrane (1). Thus, in these circumstances, preventing blood vessel remodelling and neovascularisation is a potential therapeutic approach (2,4, 17). In circumstances where gene delivery is envisaged as a means of introducing proteins into target endothelial cells for therapy, the HC-IRES element could prove invaluable. In anti-angiogenesis therapies, a virus or other expression vector could encode therapeutic proteins [such as intracellular antibody fragments (19)] to two distinct intracellular targets, adding efficacy to the desired therapeutic effect. Alternatively, in solid tumour therapy, intracellular protein targets of angiogenesis, such as LMO2 (10), could be tackled by introduction of vectors encoding two blocking reagents aimed at prohibiting the function of the target protein in distinct ways [e.g. using an intracellular antibody fragment and a peptide aptamer (20)]. Methods for delivery of vectors to specific cells in vivo are becoming more effective, and specific ways of putting vectors into endothelial cells have been reported (21). Combining these delivery methods with the ability to express efficiently two or more proteins which can combat the function of specific targets is a possible approach to anti-angiogenesis therapies.