In vivo gene delivery and expression by bacteriophage lambda vectors

Abstract

Aims: Bacteriophage vectors have potential as gene transfer and vaccine delivery vectors because of their low cost, safety and physical stability. However, little is known concerning phage-mediated gene transfer in mammalian hosts. We therefore performed experiments to examine phage-mediated gene transfer in vivo.

Methods and results: Mice were inoculated with recombinant lambda phage containing a mammalian expression cassette encoding firefly luciferase (luc). Efficient, dose-dependent in vivo luc expression was detected, which peaked within 24 h of delivery and declined to undetectable levels within a week. Display of an integrin-binding peptide increased cellular internalization of phage in vitro and enhanced phage-mediated gene transfer in vivo. Finally, in vivo depletion of phagocytic cells using clodronate liposomes had only a minor effect on the efficiency of phage-mediated gene transfer.

Conclusions: Unmodified lambda phage particles are capable of transducing mammalian cells in vivo, and may be taken up -- at least in part -- by nonphagocytic mechanisms. Surface modifications that enhance phage uptake result in more efficient in vivo gene transfer.

Significance and impact of the study: These experiments shed light on the mechanisms involved in phage-mediated gene transfer in vivo, and suggest new approaches that may enhance the efficiency of this process.

Figure 1

In vivo luciferase (luc) expression in BALB/c mice is dose dependent. Mice (three per group) were injected ID at the tail base with 1 × 10¹⁰, 5 × 10¹⁰, 1 × 10¹¹ or 5 × 10¹¹ PFU of gpD (luc) phage or 1 × 10¹¹ PFU of gpD (no luc) phage. Twenty-four hours later, luc expression was imaged at the tail base site of injection. The graph is representative of the average photon flux (photons sec⁻¹ cm⁻² sr⁻¹) and SD for each group.

Figure 2

In vivo luciferase (luc) expression in BALB/c mice injected with wild-type luc-encoding phage persists over a time frame very similar to that of mice injected with purified lambda luc DNA. Mice (eight per group) were injected ID at the tail base with either 1 × 10¹¹ PFU of wild-type gpD (luc) phage (panels a, c) or 5 μg of purified lambda luc DNA (panels b, d); note that 5 μg DNA corresponds to the expected amount of genomic lambda DNA contained in 1 × 10¹¹ PFU of lambda phage. Luc expression was measured at the tail base site of injection, 1, 3, 5, 7, and 10 days following injection. Results shown in panels a and b represent mean luc expression values ± SDs. In vivo luc expression in mice injected with gpD (luc) is similar to expression levels obtained from mice injected with 5 μg purified lambda luc DNA. Panels c and d represent an analysis of luc expression in each of the individual eight animals from each experimental group that is represented in the summary graphs (panels a and b). The dashed horizontal line drawn across each of the panels denotes the cut-off of the assay (1 × 10⁵ photons sec⁻¹ cm⁻² sr⁻¹).

Figure 3

Luciferase (luc)-encoding phage is resistant to DNase I treatment. Mice (eight per group) were injected ID at the tail base with 1 × 10¹¹ PFU of gpD (luc) phage that were either treated with 10 U of DNase I for 30 min at 37°C prior to injection, or not exposed to DNase I. As a control, mice (four per group) were injected with 1 × 10¹¹ PFU of matching phage that lacked the luc expression cassette (no luc), which were physically mixed with 5 μg of lambda luc DNA and then either treated with 10 U of DNase I for 30 min at 37°C prior to injection, or not exposed to DNase I. Twenty-four hours later, luc expression was measured at the tail base site of injection. DNase I treatment had no detectable effect on luc expression levels in mice that were injected with gpD (luc) phage (P > 0·05, Student's two-tailed t-test). However, the amount of DNase I added was sufficient to degrade 5 μg of exogenous DNA (compare luc expression levels in the two no luc groups).

Figure 4

In vivo luciferase (luc) expression in BALB/c mice injected with gpD (luc) phage is increased when mice are pre-immunized with bacteriophage lambda. Mice were immunized IM with either 1 × 10¹¹ PFU of gpD (no luc) bacteriophage lambda (lambda) in 50 μl of suspension media or 50 μl of suspension media alone (SM only). Two weeks postimmunization, all mice were injected ID at the tail base with 1 × 10¹¹ PFU of gpD (luc) phage, and luc expression was then measured 24 h later at the tail base site of injection. (a) Data shown represent mean luc expression values ± SDs; the data shown were combined from two separate experiments that used a total of seven mice. There was a statistically significant difference in in vivo luc gene expression between mice that were pre-immunized with bacteriophage lambda vs mice that were pre-immunized with suspension media alone (P < 0·05, Student's two-tailed t-test). (b) Sera were collected from mice at 14 days following the initial phage immunization and analysed for lambda-specific IgG antibodies by ELISA. Antibodies specific for bacteriophage lambda were detected in mice pre-immunized with bacteriophage lambda, but not in control mice.

Figure 5

3JCLI4 (luc) phage targets αvβ3 in vitro and increases in vivo luc expression. (a) K562-αvβ3 cells were incubated with gpD (luc) (▪) or 3JCLI4 (luc) () phage (containing a modified gpD protein bearing the 3JCLI4 integrin-binding peptide) at an MOI of 10⁵, in the presence or absence of increasing concentrations of soluble 3JCLI4 protein. Two hours later, cells were washed and lysates were prepared in order to quantify internalized phage, which were then titrated on LE392 Escherichia coli cells. Data shown represent mean phage titres ± SDs (calculated from three independently analysed wells of a cell culture dish, each of which was titrated in triplicate). There was a statistically significant increase in phage internalization between the 3JCLI4 (luc) phage particles and the gpD (luc) phage particles (*P < 0·05, two-way anova, Tukey's post-test). In addition, internalization of the 3JCLI4 (luc) phage was reduced, by the addition of soluble 3JCLI4 protein, in a dose-dependent fashion (this reduction was statistically significant at all concentrations of soluble 3JCLI4 that were added; P < 0·001 in all cases, two-way anova, Tukey's post-test). (b) gpD (luc) (▪) or 3JCLI4 (luc) () phage was added to K562-αvβ3 or wild-type K562 cells in 96-well plates, at an MOI of 10⁵. The plates were subjected to centrifugation at 900 g for 15 min, in order to enhance the efficiency of phage binding to the target cells (O'Doherty et al. 2000; Scanlan et al. 2005; Harui et al. 2006a). After this, the cultures were returned to a 37°C incubator for 1 h and 45 min, and the cells were then washed to remove unbound phage. The cultures were again returned to a 37°C incubator, and 48 h later, cell lysates were prepared. After normalization of the protein content of the cell lysates, luc expression was measured. Data shown represent mean luc expression values ± SDs (calculated from three independent experiments, each of which analysed triplicate wells of a cell culture dish). There was a statistically significant increase in luc expression between the 3JCLI4 (luc) phage particles and the gpD (luc) phage particles, when tested in K562-αvβ3 cells but not when tested in wild-type K562 cells (*P < 0·05, two-way anova, Tukey's post-test). Thus, 3JCLI4 (luc) can be targeted to cells expressing αvβ3. (c, d) To examine the ability of 3JCLI4 (luc) to increase in vivo gene delivery, mice (eight per group) were injected ID at the tail base with 1 × 10¹¹ PFU of either gpD (luc) phage or 3JCLI4 (luc) phage (containing a modified gpD protein bearing the 3JCLI4 integrin-binding peptide). As a negative control, four mice were injected with 1 × 10¹¹ PFU of gpD (no luc). (c) Mice were killed at either 1 or 3 days following phage injection, and the tail base site of injection was excised using a tissue punch. The tissue sample was then homogenized in luc sample buffer, and luc activity was measured using a chemiluminescent assay; results are expressed as relative light units (RLU) per μg of tissue extract. Data shown represent mean luc expression values ± SDs (four mice per group). There was a statistically significant difference in in vivo luc gene expression between mice that received 3JCLI4 (luc) phage particles vs animals that received the gpD (luc) phage particles at the 3-day time point (P < 0·05, Student's two-tailed t-test). The data for the 1-day time point did not achieve statistical significance. In this experiment, the control group (which received ‘no luc’ phage) was analysed at 1 day following phage delivery (only). The analysis of control animals was not repeated at the 72-h time point (N.D.), because it was felt that a single time point was sufficient to establish the background in the luc assay (□, 24 h; ▪, 72 h). (d) Luc expression was analysed at the tail base site of injection in live animals at the day 1 and 3 time points, using the Xenogen IVIS system. Data shown represent mean luc expression values ± SDs (eight mice per group at day 1 and four mice per group at day 3). In vivo luc gene expression was found to be greater in animals injected with 3JCLI4 (luc), although this increase did not achieve statistical significance (P > 0·05, Student's two-tailed t-test). In this experiment, the control group (which received ‘no luc’ phage) was analysed at both 1 and 3 days following phage delivery. The luc signal measured in these control animals fell below the background cut-off of the assay (□, 24 h; ▪, 72 h).

Figure 6

Surface display of an integrin-binding peptide (3JCLI4) does not alter the physical stability of lambda phage particles. The physical stability of various lambda phage particles was assessed using a panel of in vitro assays. For these experiments, the lambdaD1180 (luc) phage genome, which contains an amber termination mutation in the gpD coat protein gene, was packaged into phage particles either using an in vitro complementation system to express wild-type gpD (black bars) or gpD-3JCLI4 fusion protein (grey bars) on the phage surface (Zanghi et al. 2005) or by lytic-phase growth of the phage in Escherichia coli host cells containing an amber suppressor gene (‘lytic phage’, denoted by dotted bars). The physical stability of the resulting phage particles was assessed. Phages were incubated in naïve BALB/c mouse sera (diluted 1:10 in SM) at 37°C for 30 min (a), or in increasing concentrations of EDTA (b) and SDS (c), as described in the Materials and Methods section. Data shown represent mean phage titre values ± SDs. There were no statistically significant differences in the stability of phage bearing the chimeric gpD fusion protein (3JCLI4; grey bars), when compared with otherwise identical phage that displayed only wild-type gpD coat protein (gpD; black bars), except at the highest concentrations of SDS (0·02% and 0·2%). However, these SDS concentrations resulted in almost complete inactivation of phage infectivity (>99%), regardless of the coat protein composition of the phage particles.

Figure 7

In vivo luciferase (luc) expression in BALB/c mice injected with wild-type or 3JCLI4-targeted phage is only modestly affected by clodronate-mediated depletion of phagocytic cells. (a) Mice (four per group) were injected with clodronate liposomes via a combined IP and ID route, to deplete both local and systemic phagocytic cells (black bars) or were left untreated (white bars). Forty-eight hours later, 1 × 10¹¹ PFU of either gpD (luc) phage or 3JCLI4 (luc) phage, or 100 μg of gWIZ (luc) plasmid DNA was injected ID at the tail base site, and luc expression was measured 24 h thereafter at the tail base site of injection. A decrease in in vivo luc gene expression in animals treated with clodronate liposomes prior to phage or DNA injection was observed, although this decrease was not statistically significant (P > 0·05, Student's two-tailed t-test). (b) After imaging, mice were killed and splenocytes were stained for F4/80 (a cell surface marker for macrophages). Mice that received clodronate liposomes had a decrease in F4/80-positive splenocytes, as measured by flow cytometric analysis. A representative staining profile for one control animal (left) and one clodronate-treated animal (right) is shown in the lower part of panel b, and mean data from all animals are shown in graphical form in the upper part of panel B (bars denote SDs).