Genetic engineering of th
surface nee ral vectors is time-consuming, sinc loning. I
by means of molecular c p m
, a more versatile targeting
ace and easily attachable counterpart
ia and provides a one-step method to A interaction fullfills th
surface.
attach ligands on the viral
protein was inco us su
en aying baculovirus, Baavi, by fusing avidin to
g, entry and endosomal
ajor e 4 is essential for viral buddin
re ma
scap 996) modifications we
expressed under ba to the only gp64 cop
olh promoter. It has been reported t et al
ptide inse ease the virus efficiency (Erns
system was selected, since the af
-15 , 1975), di vidin is
(Kd~10 ) (Green
d and avidin is essential for k l display
modification of the virus s blems in
and W
p ential for several steps of bacu
eting on the ard
1 d gp64 are comp y p
production or oligo Mangor et al., 2001
of the gp64-fusion protein may l particle (vp) to plaque fo (
a rs and vira
a
the viral envelope may vary from dimeric to tetrameric forms or the fusion protein may form native gp64 (Boublik et al., 1995), visualized in Figure 10.
hetero-oligomeric structures with
igure 10. Possible models for assembly of trimeric membrane glycoprotein fused together with a tetrameric
of dimeric or Laitinen et al., 2001) or single/dual chain avidins
nd et al., 2004; N fusion se gp64 oligo ation.
y, this kind o t in formin i roduct.
It has been reported tha p64 a elope, while
monomeric forms are de inutes after synthesis (Oomens et al., 1995). In the
velope, the trim zed by interm Vo an and
h, 1984) and 10 e needed to r (Mar al.,
e factors fav gp h hown
that slightest modification will result in a decrease in the amount and lower titers (Ernst et al.,
Optical biosensor analysis (I/ fig 4), flu spectroscopy (FCS)
fig 3) and tra n microscopy er incub with
iotinylated gold particles confirmed that the avidin-gp64 fusion protein had retained the biological to 20 biotinylated ligands. In icroscopy images, the amount of avidin molecules / virus is around tens/ slice, which is in agreement with the (FCS) results (I/ fig 3a) of minim els/
virus. As with other baculovirus display reports (Mottershead et al., 2000; Tami et al., 2000;
Toivola et al., 2002; Ojala et al., 2004; Matilainen et al., 2006; Makela et al., 2006; Ernst et al., 2006), it seem e fusion partner affect the number of gp64-fusion proteins in the viral surface, possibly by the random acquirement of the fusion protein from viral budding. Display of GFP-gp64 fusion protein resulted in an average of 3.2 fusion proteins per virus (Toivola et al., 2002). It is suggested that larger fusion partners interfere with the viral incorporation of the fusion p
As com arrott et al., 2003) the am
is more than after adenovirus fibre metabolical biotinylation, shown to be sufficient for targeted viral transduction although direct comparison of different viruses is difficult. Fusion protein ability to bind biotin indicated that the protein was able to fold correctly, further con by optical
F protein.
The use monomeric avidins (
(Nordlu ordlund et al., 2005) as partners might ea meriz
Consequentl f approach may assis t only trimeric forms of g
graded within 30-45 m
g more homo-oligomerized fus re included into the viral env
on p
viral en ers are stabili olecular disulfide bonds ( lkm
Goldsmit of those trimers ar esult in fusogenic activity kovic et 1998). Thes or the formation of all native 64 trimers and several studies
gp64 ave s
2006).
orescence correlation
assays (I/ nsmission electro imaging (I/ fig 2) aft ation
b
activity and FCS suggested that one viral particle could bind to 10 addition, according to the electron m
um 10-20 lab s that size and composition of th
rotein (Matilainen et al., 2006).
pared to adenovirus biotinylation (P ount in this work firmed
biosensor affinity results (I/ fig 4). Altogether these data suggested that the fusion protein is incorporated into the viral envelope in a functional form.
Modification of a viral envelope could lead to increased cytotoxicity or immuno response. The MTT-assay suggested that only low cytotoxity could be associated with the virus even with moi 1000 (I/table 1), correlating with later in vivo results (III/table 2).
5.1.2 Titering of baculovi
It is essential d which determinates the viral concentration; the titering. The standard meth tious titer and moi to determinate the gene delivery efficiency.
The infectious point dilution method (O'Reilly et al., 19
ability of the baculovirus to infect Sf9-cells (production cell line). The resul titer is
indicated as pl liliter (pfu/ml). Other methods are to use r
of viral partic ive), determined by optical density and assay (similar to end-point dilution). Recently other methods, such as flow
ansducing titer, TU (Chan et al., 2006a), quantitative real-time PCR (Chan et al., 2006b) and cell diameter (Janakiraman et al., 2006) based methods have been published.
Regardless of the method used, the viral surface display might change the binding properties of the virus and result in altered viral efficiency in infecting the insect cells used for the titer determination. If the modification results in improved insect cell infection, while not changing the properties for gene delivery for mammalian cells, actual pfu titer would be lower than in reality, leading to decreased mammalian cell transduction efficiency. Respectively, decreased pfu would lead to increased transduction efficiency. In a recent study, infectious titers were determinated to be partly related to, but not reliably reflect, the capability to transduce mammalian cells (Chan et al., 2006a).
Although titers can be determined with different methods, the total viral particles to pfu ratio should remain constant within the same production facility and therefore reflect established methods. In addition, as baculoviruses are degraded to some extent at +4 °C storage (Jorio et al., 2006), titering should be performed at intervals to remain certain of the reproducibility of the obtained results. However, comparison of total viral particles to pfu ratio can be difficult:
according to the literature the ratio of total viral particle / pfu of baculovirus has been estimated at 300 (Knudson and Tinsley, 1974), 128 (Volkman et al., 1976), 4-6 (Dee and Shuler, 1996) or an average of 3.7 (Shen et al., 2002) as compared to 7 (or generally <10) of adenovirus (Ugai et al., 2005). However, these contradicting reports may reflect improvement in the viral production between different laboratories, making it difficult to compare experiments between groups if ratio is not measured.
5.1.3 Baavi resulted in enhanced transduction in vitro
It has been shown that biotin, when covalently attached to cell surface membrane proteins, enables efficient entry of avidin bioconjugates into nucleated cells (Wojda et al., 1999) providing a possible way to transduce cells without a specific receptor for a virus. This approach offers a possible method to increase local transduction efficiency without having to increase the viral dosage and create cytotoxicity problems. As regarding possible clinical use, cell biotinylation experiments in vivo have resulted in successful targeting of molecules via biotin-avidin interaction (Hoya et al., 2001; Yolcu et al., 2002; Rybak et al., 2005), without major toxicity. In addition, the surface biotinylation followed by avidin-particles has been demonstrated to be a feasible method to achieve 85% transduction efficiency in rabbit renal arteries in vivo (Hoya et al., 2001). Interestingly, it has been reported that cell surface biotinylation in vitro has been maintained for weeks (Yolcu et al., 2002), possibly widening the temporal window for avidin-biotin targeting in vivo.
rus to examine the metho od is to use the infec
titer is assayed by end- 94), based on the
ting virus aque forming units per mil
les (infective + non-infect the total numbe
plaque formation cytometry based tr
In order to examine the avidin-binding of Baavi, BT4C and RaaSMC cells were avi or wild type virus. Baavi resulted in 100 to 270 % increase in
ount of avidin per virus to be adequate to achieve attachment to cells. It was also observed th ro the transduction volume of th virus affects the transduction efficiency, larger volumes favouring the avidin-biotin interaction, possibly decreasing the effect of viral sedimentation (Dee and Shuler, 1996). In accordance, hig cell confluency reduced the baculovirus transduction efficiency, possibly preventing the access t the basolateral side, as suggested by Bilello et al., 2003.
We also found that as compared to w baculoviruses the gene transfer efficiency of Baavi per se was significantly higher in BT4C a C cell lines (I/ Fig 5). The effect was more evident at lower mois, but diminished with higher mois used. This could indicat the saturation of the transduction pathway with hig resulting diminishment of th differences caused by avidin display. Similar results we detected with VSV-GED virus (IV fig 3). The enhanced transduction efficiency may be due to the high pI ratio of avidin which is reported to enhance cellular uptake (Pardridge and Boado, 1991) or possibly binding to hepari (Kett et al., 2003; Kett et al., 2005). Avidin could enable negatively charged cell surface because avidin pI 10.5 presents a high positive net charge at physiological pH and baculovirus entry has been suggested to be affected by electrostatic interactions with the negative heparin sulphate proteoglycans (Duisit et al., 1999). While the hypothesi e net charge is reasonable, on can not overrule the unspecific binding to cells by avidin ins (Marttila et al., 2000).
5.1.4 Targeting the Baavi
he display of avidin on the viral surface should enable coating of the virus with targeting molecules. We therefore examined the binding of biotin-conjugated EGF to EGF receptor overexpressing SKOV-3 cells and observed enhanced binding with both Baavi and EGF-coated Baavi as compared to the wild-type virus (I /fig 6). Similar reports of increased binding have been reported by using gp64-ZZ domain displaying baculovirus (Ojala et al., 2001) and truncated VSV-G-ZZ displaying baculoviruses (Ojala et al., 2004), both coated with cell-specific antibodies.
However, no increased transduction efficiency to permissive cells could be observed either with these approaches or by ours. However, by including a RGD- motif peptides from the foot-and-mouth disease virus protein VP1 (Ernst et al., 2006) or coxsackie virus A9 or VP1 protein from human parechovirus 1 to gp64 (Matilainen et al., 2006), improved binding and transduction efficiencies are reported. The difference might be explained by the different route after the binding to the cell surface, enhanced binding with antibodies or ligands not triggering further internalization and transduction. Even though integrin specific motifs have shown promising results, the native copy of gp64 still retains its binding properties overruling weaker interactions, such as with baculovirus displaying integrin α2 specific motif (Riikonen et al., 2005).
In the future it might be essential to remove the native gp64 and provide the necessary functions for infection and endosomal escape by using other glycoproteins with reduced binding.
With this method, targeting might be possible by using a pathway leading to triggered viral internalization and subsequent endosomal maturation.
Yet, with cells defined as non-permissible (for example EaHY), the restrictions in the later endocytic pathway as suggested by a previous study (Kukkonen et al., 2003), might restrict the positive effect of increased or targeted binding. If true, this would require novel methods for capsid targeting or unblocking the obstacles in nuclear traffic to take advantage of enhanced binding to cells.
biotinylated and incubated with Ba
transduction efficiency which suggests the am
e
uction in vitro
he equation describing the magnetic force affecting the particles in a magnetic field is presented in
itations and applications in the magnetic targeting, since a magn
5.1.5 Magnetically targeted transd T
equation 1, where B is the magnetic flux density (field strength), ∇B magnetic field gradient, χ2 is the volume magnetic susceptibility of the magnetic particle, χ1 is the volume magnetic susceptibility of the surrounding medium and μ0 is the magnetic permeability of free space (Dobson, 2006). This equation is the basis of understanding the lim
etic field gradient is always needed to cause magnetic force into the particle and large volume particles and high magnetic susceptibility result more magnetic force being affected.
) 1 (
) (
0 1
2 x V B B
mag = x − ∇
μ Equation 1
As an approximation, it can be stated that the smaller the particle, the higher the agnetic field gradient is needed to affect the particle and compensate for the Brownian motion.
According to the literature review (Dobson, 2006) a 0,7 mT field strength requires a constant field F
While increasing the size of the particle might be feasible in vitro, the physiology of the capillaries sets boundaries to the size of the iron particles in vivo as the smallest capillaries are 3-4 µm (Young and Heath, 2000). Larger particles are likely to cause emboli within the capillary bed of the lungs.
m
gradient of at least 100T/m to capture most magnetic particles in vivo. Permanent NdFeB magnets provide >1T magnetic fields, thus enabling magnetic field strengths of > 10T/m up to 3mm with in vitro use (Schopf et al., 2005).
In order to examine magnetic targeting of Baavi in vitro we used biotinylated Spherotech 1,1 µm small paramagnetic iron oxide particles (SPIO) and 1T NdFeB magnets with plated monolayer BT4C cells. When Baavi was coated with biotinylated SPIO, clear targeting was seen in the areas where the magnet was beneath the wells (I/ fig 7). Similar in vitro transduction patterns matching the magnet area have been described by others using biotinylated retro- and adenoviruses (Pandori et al., 2002a; Hughes et al., 2001). When avidin binding was blocked with excess biotin, no targeting effect was seen with the magnets, providing proof that the binding was due to the specific bond between avidin and biotin. We also analyzed the effect of the SPIO material on the targeting (Table 11) and found that while silica coated SiMAG-particles were able
clear medi
to um from avidin-displaying viruses efficiently (a), the same experiment with wild type virus suggested unspecific binding with SiMAG (b), which was later also recognized by the literature (Bagwe et al., 2006). The experiment also proved it was possible to concentrate Baavi from the dilute mediums (unpublished data).
Table 11. Constant volume of a) Baavi or b) wild-type virus was mixed with equal amount of SiMAG-biotin (Chemicell, silica based) or Spherotech-biotin magnetic particles (polystyrene-coating) in similar volumes to te the binding efficiency. Transduction of pelleted particles was performed with NdFeB magnets in BT4C cells and presented in scale of – (no transduction) to +++ (extensive transduction).
st
a) Baavi Transduction at 24 h
SiMAG +++
Spherotech ++
b) Wild-type virus
SiMAG +
Spherotech -
With a weaker ferrite magnet (0.1 T) the targeting effect was more subtle and increased sporadic transduction was seen throughout the well, indicating that the increase in the magnetic field strength augments the particle sedimentation and dimishes the Brownian mo (Huth et al., 2004). In addition, an increase in the external magnetic field strength can be used force the magnetic particles through the cellular membrane, possibly bypassing the normal cellular uptake (Mondalek et al., 2006).
tion to
pproaches in vitro. The use of paramagnetic particles enabled the physical concentration of irus with magnetic force (Nesbeth et al., 2006; Hughes et al., 2001), resulted in increased
The use of magnetism together with non-viral or viral systems has been utilized in several a
v
transduction efficiencies in vitro (Chan et al., 2005) and targeting (Plank et al., 2003; Rudge et al., 2001). However, while in vitro the high magnetic field gradient of a few millimeters is sufficient to draw the magnetic particles to the monolayer cells, in vivo such depth would be inadequate.
Although the idea of highly specific 3D magnetic targeting in vivo would be tempting, due to the properties of magnetic fields (Figure 11) lateral targeting (Li et al., 2005), localization to organs near the surface (Goodwin et al., 1999; Arbab et al., 2004) and retention to extremities (Alexiou et al., 2003) have been most feasible. Nevertheless, magnetic materials have proven to be promising in several clinical studies including hyperthermia and magnetic drug targeting (Lubbe et al., 2001). In some experiments the depth of magnetic field of 0.1 T has been increased to 10 cm resulting to first pass targeting and retainment for six days (Goodwin et al., 1999).
Figure 11. Magnetic field lines of a bar magnet, S and N denoting poles of the magnet. The magnetic field gradient is largest where the magnetic lines are sparse, while the magnetic field is linear only in a small portion of the magnet.
5.1.6 Additional in vivo data
To analyze the feasibility of magnetic targeting in vivo we selected a model allowing us to study the retainm ne iron-coated Baavi by using high magnetic field gradients (unpublished). We injected 15 µl of the solution into the tailvein of BALB/c nude mice and placed a bar shaped 1 T NdFeB magnet with the gradient orthogonally to the beginning of the mouse tail for 15 minutes. Although the results are preliminary, some retainment could be seen as compared to the control without the magnet (Figure 12). The microscopical analysis showed aggregated clusters of iron particles, mainly seen on the blood vessel structures. Possibly due to the complement, no transduction could be seen in the area after beta-galactosidase staining (data not shown).
t of
Figure 12. Hematoxylin stained crossections from whole mount muscle tissue of nude BALB/c mice after injection of SPIO-coated Baavi a) with 1 T NdFeB magnet held for 15 min (gradient facing from right to left), b) without any magnet c) extract of a and d) extract of b. Images a and b original magnification of 100x. Arrows indicate iron clusters.
Due to the promising results of a previous studies (Lehtolainen et al., 2002; Kukkonen et al., 2003), we speculated on the possibility of imaging the biodistribution after intracerebroventricular injection of either 1.1 µm SPIO or Gd3+-chelate coated Baavi in BDIX rats with MRI. The preliminary experiment with SPIO however resulted in very limited transduction and biodistribution (data not shown), possibly due to the hindered motion of the large sized viral complex in the ventricles. Gadolinium chelate did not reach the detection threshold and further improvements were needed (data not shown). We concluded that smaller iron particles with a nanometer scale would be required for in vivo use to maintain the viral B
as possible. rownian motion as native
Based on the properties of MRI, discussed in 2.4.1.1, there are two possible classes of contrast gents to be selected from: T1 and T2. While the signal increases with the increase of T1 agents, the opposite effect takes place with T2 agents and signal loss occurs. When considering the use in a e quite high, a
increase occurring during viral entry, there might be some decrease in the of 10
MTT assay to analyze the toxicity of the USPIO (Ultrasmall paramagnetic iron oxide particles) particles together with viruses, without any signs of toxicity (Table 12). Since ultrasmall paramagnetic iron oxide particles had proven to be feasible for use in vivo with MRI, we biotinylated 50 nm USPIO (bUSPIO) and coated Baavi with them.