- Email: [email protected]
Translational modifications to improve vaccine efficacy in an oral influenza vaccine
Methods 49 (2009) 322–327
Contents lists available at ScienceDirect
Methods journal homepage: www.elsevier.com/locate/ymeth
Translational modifications to improve vaccine efficacy in an oral influenza vaccine Ewan Bennett *, Alexander B. Mullen, Valerie A. Ferro University of Strathclyde, Strathclyde Institute of Pharmacy and Biomedical Science, 27 Taylor Street, Glasgow G4 0NR, UK
a r t i c l e
i n f o
Article history: Accepted 23 April 2009 Available online 3 May 2009 Keywords: Oral vaccination Influenza IgG IgA Mucosal
a b s t r a c t Oral vaccination using protein antigens is complicated by the degradative effects of the inhospitable conditions in the gastrointestinal tract, such as low pH and digestive enzymes, nescessitating protection and effective delivery of the antigen. The bilosome is a lipid-based, vesicle delivery system incorporating bile salts, which is believed to protect the antigen from degradation, and has been shown to induce significant antibody responses when delivered orally with various vaccines. In translational research, from laboratory bench to industrial scale-up, it is necessary to optimise the manufacturing process in order to improve efficiency and simplify production, giving a more economical end-product. In this study we tested two simplified production methods (3-step and 1-step) along with two different storage methods (lyophilised and non-lyophilised), as well as looking at the effect of buffer pH. The formulations were assessed in a murine system for immunogenicity, alongside characterisation in terms of size and antigen entrapment, with the stability of these aspects assessed with respect to time. Both lyophilised and non-lyophilised 3-step formulations induced significant IgG1, IgG2a and IgA titres, with the lyophilised version showing stable size and antigen entrapment up to 9 months. Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction The influenza virus is an RNA virus, which infects millions of people around the world in seasonal epidemics, and has also killed tens of millions in pandemics that occur when a new human strain of the virus appears [1]. The average excess annual mortality attributed to influenza in the UK is approximately 14,000 deaths, with up to 90,000 deaths occurring in pandemic years [2]. Influenza epidemics also have a significant socio-economic impact, with direct costs such as hospitalisation and treatment, and indirect costs such as loss of productivity and absenteeism. During 1995–1996 the direct cost of treating influenza in Germany was estimated to be US$2.8 million, with indirect costs estimated at US$800 million [3]. At present most influenza vaccination programmes involve subdermal or intramuscular injection of the antigen, a procedure which is invasive, painful, and requires trained professionals. This type of administration only provides systemic protection via IgG antibodies, leaving the principal vector for infection, the mucosal membranes of the respiratory tract, undefended [4,5]. Mucous membranes cover an area of approximately 400 m2 of the body, most of which is vulnerable monolayered, epithelium [6], with protection provided mainly through antibody IgA [5]. Therefore a pain-free, simple, and non-invasive means of administering an
* Corresponding author. E-mail address: [email protected] (E. Bennett). 1046-2023/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ymeth.2009.04.015
influenza vaccine, that induces IgA as well as systemic IgG in humans, would simplify and improve any vaccination programme. Adjuvant formulations have previously been used to enhance the immune response elicited by an antigen [7], and particulate adjuvants are especially useful because of their similarity in size to the pathogens and their ability to present antigen in a multimeric format, causing uptake by antigen-presenting cells (APC) [8]. One such formulation uses the alternating hydrophobic and hydrophilic nature of non-ionic surfactant vesicles (NISV), which allows antigen to be entrapped. NISV are formed by bilayers of hydrated phospholipids, a structure which allows a wide range of chemical and physical modifications of the vesicle to be performed, and compounds with a variety of solubilities to be entrapped. The incorporation of bile salts into these NISV stabilises them against the effects of bile acids [9] and the resulting vesicles are termed bilosomes [10]. The bilosome has been shown to be an effective delivery system for a number of vaccine antigens, including A/Panama [11], tetanus toxoid [12] and hepatitis B [13]. These studies have shown significant antibody titres in a murine experimental model, and an infection challenge study in ferrets vaccinated with influenza haemagglutinin entrapped in bilosomes has shown higher antibody production, lower temperatures and reduced symptoms compared with IM injection [14]. Therefore, the next stage in developing a viable clinical vaccine is to optimise the bilosome formulation into a simple and efficient method whilst retaining immunogenicity. The aim of this paper is to describe how we streamlined the manufacturing process for bilosomes described by Mann et al .
E. Bennett et al. / Methods 49 (2009) 322–327
(2004) [11] for industrial scale-up. The published method was simplified from the original Mann method into a 3-step and a 1-step manufacturing process, with freeze-dried versions of each prepared to examine the suitability of lyophilisation for improving storage stability. In addition, the effect of buffer pH (9.7 versus 7.6), which can restrict certain antigens from use with this system due to precipitation, was also assessed. Comparison between the groups was made of particle size and protein entrapment efficiency, followed by evaluation of vaccine efficacy after oral immunisation of influenza antigen (haemagglutinin, HA from New Caledonian strain) in BALB/c mice using a standardised regime.
2. Materials and methods 2.1. Vesicle preparation 2.1.1. Original Mann et al. (2004) method [11] 1-Monopalmitoyl glycerol (150 lmol, Larodan AG, Sweden), cholesterol (Sigma–Aldrich, UK) and diacetyl phosphate (Sigma– Aldrich, UK) (5:4:1 M ratio) were melted in an oil bath at 140 °C for 2 min, then cooled to 60 °C and hydrated with 3.32 ml of 0.025 M carbonate buffer, pH 9.7. The lipid mixture was homogenised for 5 min at 8000 rpm (Silverson Machines Ltd., UK), followed by addition of 100 mg/ml deoxycholic acid (Sigma– Aldrich, UK) in 0.025 M carbonate buffer, pH 9.7 (preheated to 30 °C) and a further 2 min homogenisation. This mixture was incubated at 30 °C for 2 h in a water bath, then 75 lg of antigen added and the final volume made up to 5 ml with carbonate buffer before being subjected to five freeze–thaw cycles in liquid nitrogen. Unentrapped antigen was removed by centrifugation at 200,000 g for 2 h using a Beckman XL-90 ultracentrifuge (Beckman RIIC, UK), with the vesicles resuspended in 2 ml of 0.025 M carbonate buffer, pH 9.7.
2.1.2. Modified method Briefly, a 5:4:1 M ratio of lipids (all quantities are for a final volume of 10 ml): 248 mg 1-monopalmitoyl glycerol, 234 mg cholesterol and 82 mg dicetyl phosphate were melted in a 25 ml round bottomed flask using an oil bath at 120 °C, then hydrated as follows. (a) 3-Step method: Addition of 3.78 ml of 0.025 M carbonate buffer, pH 9.7 or 7.6 (preheated to 60 °C), then homogenisation for 2 min, followed by addition of 1 ml bile salt (final concentration 10 mM, preheated to 60 °C) and homogenised for 8 min at 8000 rpm. The resultant mixture was then cooled to 30 °C over 2 h in a water bath, followed by the addition of 5.22 ml carbonate buffer (0.025 M, pH 9.7) containing 1.2 mg antigen (preheated to 30 °C) and homogenised for 1 min at 8000 rpm. When examining the effect of pH 0.025 M carbonate buffer was also used at pH 7.6. (b) 1-Step method: One millilitre of 100 mM bile salt dissolved in 0.025 M carbonate buffer, pH 9.7, and 9 ml antigen solution (1.2 mg antigen in 9 ml of 0.025 M carbonate buffer, pH 9.7, preheated to 60 °C), followed by homogenisation for 13 min at 8000 rpm, then cooled to 30 °C over 2 h in a water bath with occasional agitation.
2.2. Lyophilisation of vesicles The vesicles were usually stored as an emulsion at 4 °C, but in order to improve stability, ease of storage and shipping, the vesicles were lyophilised using an Edwards Modulyo freeze drier at
323
45 °C for 24 h, then stored sealed at 4 °C. The effect of lyophilisation on immunisation efficacy was examined. 2.3. Vesicle sizing/zeta potential 2.3.1. By dynamic light scattering Particle sizing and zeta potential measurements were made on a Malvern Instruments Zetasizer Nano ZS at 25 °C using triplicate readings and a 2 min equilibration time. 2.3.2. By freeze fracture electron microscopy (FFEM) Samples were sandwiched between two copper support plates (Bal-tec, Lichtenstein) and frozen in a cryogenic mixture of propane/isopentane (3:1 v/v) at 190 °C. These were then stored at 150 °C. Fracturing was carried out at 100 °C at 106 torr, with the fracture face replicated with platinum/carbon (Pt/C, Agar Scientific, UK) at 45 °C, and carbon coated at 90o to the surface. Acetone was used to clean the replicates and they were collected onto a mesh (100 grid bars/in.), dried and viewed on an LEO 912 electron microscope at 80 kV. 2.4. Protein entrapment Entrapment of protein was quantified using a modified ninhydrin assay, a method reported to be unaffected by lipid interference and previously described by Brewer et al. (1995) [14]. In order to separate entrapped antigen from free antigen a 0.11 ml sample of vesicles, diluted in 4 ml of 0.025 M carbonate buffer, pH 9.7, was spun in a Beckman tube in a Beckman XL-90 ultracentrifuge (Beckman RIIC, UK) at 35,000 rpm for 2 h. The supernatant was then discarded and the pellet resuspended in 0.11 ml 0.025 M carbonate buffer, pH 9.7, then transferred to 1.5 ml microfuge tubes (Elkay, UK). The samples, along with standards prepared with 0, 0.25, 0.5, 0.75, 1, 2.5, 5, 7.5, 10, 15 and 20 ll of antigen, all in 1.5 ml microfuge tubes, were placed in an oven at 90 °C overnight. One hundred and fifty microlitres of 13.5 M NaOH was then added to each tube, with a pinhole made in the lid before autoclaving at 121 °C/131 kPa for 20 min. After removal from the autoclave, the holes were sealed with autoclave tape and the NaOH neutralised with 250 ll glacial acetic acid, followed by vortexing for 5 s. 500 ll of ninhydrin reagent (Sigma–Aldrich, UK) was then added to each tube, each sample vortexed and placed in a water bath at 90 °C for 20 min. Two hundred and fifty microlitres of the resultant mixture was transferred to a fresh tube containing 750 ll 50% (v/v) propan-2-ol and vortexed, and 200 ll of each sample transferred to a flat-bottomed 96-well plate (Iwaki, Japan). Absorbance was read at 540 nm on a SpectraMax 190 plate reader (Molecular Devices, USA), with test sample levels determined by linear regression from the standard calibration curve. The quantity in the supernatant was subtracted from the total protein added to calculate the entrapment efficiency. Experimental conditions were the same for standards and samples and a correlation coefficient of >0.99 was obtained. 2.5. Immunisation and sampling In-house bred male BALB/c mice, 8–10 weeks old, housed in a fully climatised room were randomised and placed into 10 groups (n = 5). All mice were starved, but allowed access to water, for 2 h pre-immunisation, with food and water available ad libitum between immunisations. New Caledonian haemaglutinin (kindly supplied by Solvay Pharmaceutical, Netherlands) was entrapped into: (a) Study 1: bilosomes using the 3-step and 1-step methods and comparison made with the equivalent freeze dried vesicles. (b) Study 2: 3-step methods with either pH 9.7 or 7.6 carbonate buffer. Each dose consisted of 0.4 ml bilosome, containing approximately
324
E. Bennett et al. / Methods 49 (2009) 322–327
50 lg N/Cal HA), administered by intragastric gavage on days 1, 4, 14, and 17. Freeze-dried bilosomes were reconstituted prior to immunisation with 0.4 ml carbonate buffer. 0.4 ml has been found to be a suitable volume to induce a good immune response [12], and retention of the total volume in the stomach could be inferred from the absence of any reflux or nasal discharge. The immunisation regime was also examined by intramuscular dosing fortnightly, on days 1, 14, 14, 17, administered in two sites in the hind leg (0.05 ml per site). The groups consisted of: (a) Study 1: Group 1 (experimental, 3-step bilosomes with antigen), Group 2 (experimental, 3-step, freeze-dried bilosomes with antigen), Group 3 (experimental, 1-step bilosomes with antigen), Group 4 (experimental, 1-step freeze-dried bilosomes with antigen), Group 5 (control, empty 1-step bilosomes), Group 6 (control, intramuscular injection). (b) Study 2: Group 1 (experimental, 3-step bilosomes with antigen, pH 9.7 buffer), Group 2 (experimental, 3-step bilosomes with antigen, pH 7.6 buffer), Group 3 (control, empty bilosomes, pH 9.7 buffer). Tail bleeds were collected in heparinised capillary tubes on days 7, 20, and 33 post-immunisation (Groups 1–5) and days 20, 33, 47 (Group 6), and centrifuged at 1000g in 1.5 ml microfuge tubes for 20 min. Plasma was transferred into fresh 0.5 ml micro-centrifuge tubes (Fischer, UK), and stored at 70 °C until IgG1/IgG2a levels were determined by enzyme linked immunosorbent assay (ELISA). The study was terminated on day 54. IgA levels was determined by ELISA of lung lavages obtained by perfusing the lungs with 0.5 ml PBS. 2.6. Enzyme linked immunosorbent assay (ELISA) Unless otherwise stated, all incubations were carried out for 1 h at 37 °C and washes three times with PBS, pH 7.4, containing 0.01% (v/v) Tween 20 (PBS–Tween), respectively. Ninety-six well tissue culture plates (Iwaki, Japan) were coated with 60 ll/well containing 0.2 lg N/Cal and incubated, then washed three times and blocked with 200 ll/well of 3% (w/v) MarvelTM (Premier Foods UK Ltd.) in PBS. The plates were washed three times and 150 ll/well of a 1:100 dilution of sera added to the first column, then 100 ll/ well of PBS added to the rest of the plate, with a 1:3 serial dilution created by transferring 50 ll from the first column to the next across the plate. The plates were incubated and washed, followed by addition of 100 ll/well of a 1:3000 dilution of IgG1, IgG2 or IgA goat anti-mouse (Southern Biotech, UK), followed by incubation for 45 min. The plates were washed and incubated with 150 ll/well of TMB substrate (250 ll of stock 3,30 ,5,50 -teramethyl benzidine 6 lg/ml dissolved in dimethylsulphoxide (DMSO), added to 25 ml of 0.1 M sodium acetate citrate buffer, pH 5.5, with 4 ll of 30% (v/v) H2O2). After 15 min the reaction was stopped with 50 ll/ well of 10% (v/v) H2SO4, and the absorbance at 450 nm read on a SpectraMax 190 plate reader.
5.0 nm, and empty = 211.8 ± 45.0 nm. These differences were not found to be significant. Lyophilisation of the bilosomes caused an increase in size on rehydration, giving a mean size ± SD of 793.9 ± 39.8 nm for 1-step and 976.7 ± 94.5 nm for 3-step. There was no significant between these groups, with both having p < 0.05 versus all non-lyophilised formulations. The analysis of the 3-step and 1-step vesicles using a LEO912 energy filtering microscope at 80 kV showed spherical structures without lipid sheets or crystalline bodies, with two populations of approximately 150–250 nm and 800–1000 nm, confirming the dynamic light scattering results (Fig. 1). In all cases the zeta potential of the bilosomes was around 100 mV, with no significant differences between the groups (Table 1). 3.1.2. Antigen entrapment For both the 1-step and 3-step protocols, antigen entrapment efficiency was roughly the same, at approximately 30 ± 5%. This was enhanced by lyophilisation, which gave 50 ± 5% entrapment. 3.1.3. Size/entrapment – stability over time and effects of homogenisation time Both 3-step and 3-step lyophilised bilosomes were monitored over 9 months to observe any changes in size or entrapment with time, whilst stored at 4 °C. The size of the 3-step bilosomes remained stable up to 4 weeks after formulation, and then increased in size from around 200 nm at 0 weeks to approximately 350 nm during the mid-stages, with a final size of 560 nm at the 9-month point. As size increased for the 3-step bilosomes there was a concurrent decrease in antigen entrapment, falling from 30% initially to 20% after 9 months (Fig. 2). The size and entrapment of the lyophilised 3-step bilosomes (after rehydration) was unchanged at any time points, remaining within ±1 standard deviation of the original (data not shown). Examining the consequences of homogenisation time showed no apparent affect on size (Fig. 3).
2.7. Statistical analysis Statistical analysis was performed using a two-tailed unpaired t-test with Minitab v15. 3. Results 3.1. Bilosome characteristics 3.1.1. Size/zeta potential Size analysis using the Malvern Instruments Zetasizer Nano ZS indicated that the melt homogenisation methods formed bilosomes between 150 and 200 nm with a small population around 1000 nm, with mean sizes ± SD of: 1-step = 165.9 ± 27.3 nm, 3step = 194.7 ± 83.7 nm, pH 9.7 = 233.9 ± 7.0 nm, pH 7.6 = 250.3 ±
Fig. 1. FFEM of bilosomes produced by the 3-step method before lyophilisation. Imaging was carried out using a LEO912 energy filtering electron microscope at 80 kV.
325
E. Bennett et al. / Methods 49 (2009) 322–327
Formulation
Size (nm) ± SD
Zeta potential (mV) ± SD
1-Step 3-Step 1-Step lyophilised 3-Step lyophilised pH 9.7 pH 7.6 Empty
165.9 ± 27.3 194.7 ± 83.7 793.9 ± 39.8 976.7 ± 94.5 233.9 ± 7.0 250.3 ± 5.0 211.8 ± 45.0
102.7 ± 11.9 104.5 ± 13.9 102.5 ± 13.2 113.5 ± 6.4 113.6 ± 11.9 112.8 ± 14.5 90.1 ± 11.5
A
16.00
Ln mean end point titre +/- S.E.
Table 1 Size/zeta potential of bilosomes formulations determined using 20 ll sample in 1 ml filtered 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS.
14.00
* p = 0.004 ~ p = 0.474
* p = 0.033 ~ p = 0.091
Day 7 Day 20 Day 33
12.00 10.00 8.00 6.00 4.00 2.00 0.00
Size (nm)
800
3s
45
3s fd
1s
% entrapment
Size (nm) +/- S.D.
30
500
25 400 20 300
15
200
10
100
5
0 5
10
15
20
25
30
0 40
35
B Ln mean end point titre +/- S.E.
35
600
Antigen entrapment (%) +/- S.D.
40
700
0
Time (weeks)
empty 1s
IM
12 * p = 0.001 ~ p = 0.004
* p = 0.029 ~ p = 0.039
Day 7 Day 20 Day 33
10
8
6
4
2
0
3s
500 450 400 350 300 250 200
3s fd
1s
1s fd
empty 1s
IM
Group
C Ln mean end point titre +/- S.E.
Fig. 2. Size (nm)/% entrapment versus time, 3-step bilosomes. Size was measured using 20 ll of sample in 1 ml 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS, and antigen entrapment was determined by ninhydrin assay.
Size (nm) +/- S.D.
1s fd Group
9 8
* p = 0.012 ~ p = 0.100
* p = 0.003 ~ p = 0.025
7 6 5 4 3 2 1 0 3s
150
3s FD
1s
1s FD
empty 1s
IM
Group
100 50 0 0
2
4
6
8
10
12
14
16
Homogenisation time (mins) Fig. 3. Size (nm) versus homogenisation time (min). Size was measured using 20 ll of sample in 1 ml of 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS.
3.2. In vivo antibody response (a) IgG1 – Significant endpoint titres were observed for both 3step groups, compared with empty bilosomes (Fig. 4) (3-step Ln titre ± SE 13.17 ± 1.81, p = 0.004, 3-step lyophilised Ln titre ± SE 11.86 ± 2.00, p = 0.033), although there was no significant difference between either of the 3-step groups (p > 0.05). Similarly, both 1-step groups showed no significant difference in titres. Both pH 9.7 and 7.6 buffer groups also induced significant titres compared with the empty bilosome group (pH 9.7 Ln titres ± SE 11.42 ± 0.49, p = 0.002; pH 7.6 Ln titres ± SE 9.66 ± 1.25, p = 0.006; empty Ln titre ± SE 5.92 ± 1.80), with pH 9.7 having a greater effect than pH 7.6 (p = 0.031) (Fig. 5).
Fig. 4. Reciprocal endpoint titres are shown in (A) IgG1, (B) IgG2a and (C) IgA, with each value representing the mean natural log of the endpoint titre ± SE. Significantly high levels of antibody were observed on completion of the study (day 33), indicated by the p values: *comparison with empty bilosome group, comparison with IM group. 3s = 3-step bilosomes, 3s FD = 3-step freeze-dried bilosomes, 1s = 1step bilosomes, 1s FD = 1-step freeze-dried bilosomes, Empty 1s = empty 1-step bilosomes, IM = Intramuscular injection.
(b) IgG2a – Both 3-step groups again induced significant titres compared with the empty bilosome group (3-step Ln titre ± SE 9.00 ± 0.78, p = 0.001; 3-step lyophilised Ln titre ± SE 8.56 ± 1.67, p = 0.029) with no difference between the two groups and no significant titres observed with the 1-step groups. Significant titres were observed with pH 9.7 and 7.6 buffer groups (pH 9.7 Ln titre ± SE 9.00 ± 1.55, p = 0.007; pH 7.6 Ln titre ± SE 9.22 ± 1.81, p = 0.011; empty Ln titre ± SE 5.92 ± 0.92) with no difference between the two pH groups. (c) IgA – Significant titres were observed in both 3-step groups (3-step Ln titre ± SE 6.58 ± 0.92, p = 0.012; 3-step lyophilised Ln titre ± SE 7.24 ± 0.99, p = 0.003), with no difference between the two groups and no significant titres for 1-step groups. Both groups with altered buffer pH failed to induce significant IgA levels.
326
Ln mean end point titre +/- S.E.
A
E. Bennett et al. / Methods 49 (2009) 322–327
14 * p = 0.002
12 * p = 0.006
10
Day 7 Day 20 Day 33
8 6 4 2 0
9.7
7.6
Clortno
Group
Ln mean end point titre +/- S.E.
B
12 * p = 0.011
* p = 0.007
10
Day 7 Day 20 Day 33
8 6 4 2 0
9.7
7.6
Control
Group
Ln mean end point titre +/- S.E.
C
7 6 5 4 3 2 1 0 9.7
7.6
Control
Group
Fig. 5. Reciprocal endpoint titres are shown in (A) IgG1, (B) IgG2a and (C) IgA, with each value representing the mean natural log of the endpoint titre ± SE. Significantly high levels of antibody were observed on completion of the study (day 33) indicated by the p values: *comparison with empty bilosome group, comparison with IM group. 3s = 3-step bilosomes, 3s FD = 3-step freeze-dried bilosomes, 1s = 1step bilosomes, 1s FD = 1-step freeze-dried bilosomes, Empty 1s = empty 1-step bilosomes, IM = Intramuscular injection.
4. Discussion Development of oral formulations for vaccine delivery presents many challenges, and the aim of these studies was to address some of these issues, such as storage, ease of formulation and examining the effect of buffer pH to prevent precipitation of some synthetic peptides. This was performed by characterising, in terms of size, zeta potential and antigen entrapment, and immune response to two manufacturing protocols (3-step and 1-step) as well as the respective lyophilised formulations. With all the methods of formulation, the size range determined by dynamic light scattering was between 150 and 220 nm, although a small population (2– 4%) was observed at around 1000 nm, a result that was confirmed
by FFEM, with no lipid sheets or crystalline structures observed. Lipid vesicles have highly variable sizes, dependant mainly on formulation method and lipid constituents, and can range from 50 nm to several microns in size [15–17]. However, the size range obtained with our modifications is consistent with bilosomes formed by our earlier studies [11]. Lyophilisation of liposomes has previously been shown to produce anywhere between a 2and 7-fold increase in size on rehydration [18], therefore the 4-fold increase observed here is within the expected range. It is possible to limit this expansion using cryoprotectants [19]; however, there appears to be no deleterious effect on the immune response, and indeed IgA titres appear higher in these groups, although not significantly so, making this an unnecessary requirement. The longterm stability of the lyophilised 3-step bilosomes is encouraging; as there is no significant change in size or entrapment, indicating a stable formulation, although the change in size and entrapment, observed with the non-lyophilised variant is to be considered, and it will be desirable to test immunogenicity versus storage time in future work. Antigen entrapment in the 3-step and 3-step freeze-dried formulations of N/Cal was found to be approximately 30% and 50%, respectively, a result which is comparable with previous work with the A/Panama flu vaccine (50%, [11]), although entrapment has been shown to be highly variable and dependent on the antigen used [20]. Zeta potential is mainly dependent on the lipid components, with both cationic [21,22] and anionic [23] liposomes available using different lipids, with both types of liposome also possible with the same components in different ratios [24]. The zeta potential in each of the formulations we tested was around 100 mV, which is similar to other work with liposomes containing both cholesterol and diacetyl phosphate [23]. The dosing regime was well tolerated by the animals, with no overt toxicity observed 24 h post-immunisation. No significant responses were observed for either of the 1-step methods. Both of the 3-step formulations induced similar immune responses to HA, indicated by the significantly raised IgG1 titres compared with the empty bilosome group, with a modest IgG2a response. IgA levels in these two groups were also significantly higher than those treated with empty bilosomes, although only the lyophilised group were significantly greater versus IM injection. In the buffer pH study both pH 9.7 and pH 7.6 induced good IgG1 and IgG2a responses, although responses were higher for IgG1 with pH 9.7. In this study however, no significant IgA titres were observed, an unusual result given our previous data [11]. This may reflect the change in method, from measurement of IgA in faecal matter [11] to measurement in lung lavage fluid. However, the lavage method provides a direct measure of lung IgA levels, whilst analysis of faecal matter is an indirect measure, and may not give an accurate representation of actual levels in the lungs. These results are similar to those of previous work on bilosomes manufactured according to the Mann et al. (2004) [11] method with entrapped A/Panama antigen, although no IgG2a response was observed in this case. The skewing of the response towards IgG1 may be explained by the observations that vesicles of diameter less than 250 nm, administered parenterally, induce a TH2 biased response [25], whilst those greater than 250 nm skew towards TH1 [14]. With the majority of the bilosome population around 200 nm a TH2 response would therefore be expected, however, the small population at around 1000 nm appears to be inducing an additional TH1 response. With around 90% of infections involving the mucosal surfaces of the body [26], it is apparent that induction of mucosal immunity (mediated mainly by IgA) would provide a great advantage in combating these infections. In elderly patients this is particularly relevant, as there is a reduction in immune response with age [27,28],
E. Bennett et al. / Methods 49 (2009) 322–327
and it is estimated that only between 10% and 30% of over 65 s are afforded protection from influenza vaccination [29]. Therefore, they are placed at a disadvantage by enabling the virus to gain a foothold on mucosal membranes before any protective response occurs. Induction of an IgA response has other advantages; it can neutralise the influenza virus itself, in the absence of serum IgG [30–32], and it has been shown that, whilst conventional parenteral administration of vaccine mediated by IgG provides adequate protection against homologous viral infections, it has little efficacy against emerging variant strains [33]. In this situation, where there may be an absence of a direct vaccine for these variants, or the patient may not be immunised against them, protection can still be achieved through induction of an IgA response [31,33,34], an issue which is particularly relevant for high risk demographics. Despite the clear benefits of a mucosal IgA response, it remains necessary for any vaccine to induce virus neutralising antibodies in serum in order to attain regulatory approval [35], something which has not been achieved in clinical studies [4,36–38]. Here we have demonstrated the capability of the bilosome delivery system to induce a mucosal response, with significant IgA titres observed for both 3-step formulations, a result previously observed with this system with A/Panama [11]. The bilosome provides a proven delivery system that offers the potential to preclude the need for IM injection of influenza vaccines, improving patient compliance with the benefits of mucosal as well as systemic protection. Since lyophilised versions offer similar levels of IgG1 and IgA, this removes the need for cold-chain issues, thus improving storage and distribution and allowing vaccine programmes to operate in areas of the world lacking in medical infrastructure. Whilst this may be less relevant in terms of influenza, it would provide a great benefit with other antigens which can be used with bilosomes, such as those derived from cholera toxin [17], hepatitis B [13] and tetanus toxoid [12]. Acknowledgments E.B. is funded through an EPSRC-Strathclyde University Doctoral Training Centre studentship (Bioengineering Department). The authors thank Mr. George Walker, Mr. Gordon Watt for their technical assistance. References [1] N. Johnson, J. Mueller, Bulletin of the History of Medicine 76 (1) (2002) 105– 115. [2] S.L. Johnston, Virus Research 82 (1–2) (2002) 147–152. [3] T. Szucs, Journal of Antimicrobial Chemotherapy 44 (1999) 11–15. [4] Z. Moldoveanu, M.L. Clements, S.J. Prince, B.R. Murphy, J. Mestecky, Vaccine 13 (11) (1995) 1006–1012.
327
[5] P.L. Ogra, H. Faden, R.C. Welliver, Clinical Microbiology Reviews 14 (3) (2001) 641. [6] P. Brandtzaeg, Vaccine 25 (30) (2007) 5467–5484. [7] N. Petrovsky, J.C. Aguilar, Immunology and Cell Biology 82 (5) (2004) 488–496. [8] M. Copland, T. Rades, N.M. Davies, M.A. Baird, Immunology and Cell Biology 83 (2005) 97–105. [9] R. Schubert, H. Jaroni, J. Schoelmerich, K.H. Schmidt, Digestion 28 (1983) 181– 190. [10] M. Conacher, J. Alexander, J.M. Brewer, Vaccine 19 (20–22) (2001) 2965–2974. [11] J.F.S. Mann, V.A. Ferro, A.B. Mullen, L. Tetley, M. Mullen, K.C. Carter, J. Alexander, W.H. Stimson, Vaccine 22 (2004) 2425–2429. [12] J.F.S. Mann, H.E. Scales, E. Shakir, J. Alexander, K.C. Carter, A.B. Mullen, V.A. Ferro, Methods 38 (2005) 90–95. [13] A. Shukla, K. Khatri, G. Gupta, A. Goyal, A. Mehta, S. Vyas, Journal of Pharmacy and Pharmaceutical Sciences 11 (1) (2008) 58–66. [14] J.M. Brewer, C.W. Roberts, W.H. Stimson, J. Alexander, Vaccine 13 (15) (1995) 1441–1444. [15] J.M. Brewer, K.G.J. Pollock, L. Tetley, D.G. Russell, Journal of Immunology 173 (10) (2004) 6143–6150. [16] G.F.A Kersten, D.J.A. Crommelin, Vaccine 21 (2003) 915–920. [17] P. Singh, D. Prabakaran, S. Jain, V. Mishra, K.S. Jaganathan, S. Vyas, International Journal of Pharmaceutics 278 (2004) 379–390. [18] P.A. Bridges, K.M.G. Taylor, Journal of Pharmacy and Pharmacology 53 (3) (2001) 393–398. [19] L.J. Zhang, L. Liu, Y. Qian, Y. Chen, European Journal of Pharmaceutics and Biopharmaceutics 69 (2) (2008) 750–759. [20] G. Gregoriadis, B. McCormack, M. Obrenovic, R. Saffie, B. Zadi, Y. Perrie, Methods 19 (1999) 156–162. [21] J. duPlessis, C. Ramachandran, N. Weiner, D.G. Muller, International Journal of Pharmaceutics 127 (2) (1996) 273–278. [22] Y. Perrie, P.M. Frederik, G. Gregoriadis, Vaccine 19 (23–24) (2001) 3301–3310. [23] D. Gopinath, D. Ravi, B.R. Rao, S.S. Apte, D. Renuka, D. Rambhau, International Journal of Pharmaceutics 271 (1–2) (2004) 95–113. [24] J.C. Birchall, I.W. Kellaway, S.N. Mills, International Journal of Pharmaceutics 183 (2) (1999) 195–207. [25] J.M. Brewer, L. Tetley, J. Richmond, F.Y. Liew, J. Alexander, The Journal of Immunology 161 (8) (1998) 4000–4007. [26] P. Brandtzaeg, International Journal of Medical Microbiology 293 (1) (2003) 3– 15. [27] K. Leder, P.F. Weller, M.E. Wilson, Clinical Infectious Diseases 33 (9) (2001) 1553–1566. [28] G. Wick, B. Grubeck-Loebenstein, Experimental Gerontology 32 (4) (1997) 401–413. [29] D.C. Powers, E.D. Kilbourne, B.E. Johansson, Clinical and Diagnostic Laboratory Immunology 3 (5) (1996) 511–516. [30] R.B. Belshe, W.C. Ruber, P.M. Mendelman, et al., Journal of Infectious Diseases 181 (3) (2000) 1133–1137. [31] F.Y. Liew, S.M. Russell, G. Appleyard, C.M. Brand, J. Beale, European Journal of Immunology 14 (4) (1984) 350–356. [32] K.B. Renegar, P.A. Small, Journal of Virology 65 (4) (1991) 2146–2148. [33] R. Ito, Y.A. Ozaki, T. Yoshikawa, et al., Vaccine 21 (19–20) (2003) 2362– 2371. [34] T.M. Tumpey, M. Renshaw, J.D. Clements, J.M. Katz, Journal of Virology 75 (11) (2001) 5141–5150. [35] A. Huckriede, L. Bungener, T. Stegmann, T. Daemen, J. Medema, A.M. Palache, J. Wilschut, Vaccine 23 (2005) S26–S38. [36] S.S. Avtushenko, E.M. Sorokin, N.Y. Zoschenkova, N.G. Zacharova, A.N. Naichin, Journal of Biotechnology 44 (1–3) (1996) 21–28. [37] K.C. Bergmann, R.H. Waldman, H. Tischner, W.D. Pohl, International Archives of Allergy and Applied Immunology 80 (1) (1986) 107–109. [38] V. Lazzell, R.H. Waldman, C. Rose, R. Khakoo, et al., Journal of Biological Standardization 12 (3) (1984) 315–321.
Contents lists available at ScienceDirect
Methods journal homepage: www.elsevier.com/locate/ymeth
Translational modifications to improve vaccine efficacy in an oral influenza vaccine Ewan Bennett *, Alexander B. Mullen, Valerie A. Ferro University of Strathclyde, Strathclyde Institute of Pharmacy and Biomedical Science, 27 Taylor Street, Glasgow G4 0NR, UK
a r t i c l e
i n f o
Article history: Accepted 23 April 2009 Available online 3 May 2009 Keywords: Oral vaccination Influenza IgG IgA Mucosal
a b s t r a c t Oral vaccination using protein antigens is complicated by the degradative effects of the inhospitable conditions in the gastrointestinal tract, such as low pH and digestive enzymes, nescessitating protection and effective delivery of the antigen. The bilosome is a lipid-based, vesicle delivery system incorporating bile salts, which is believed to protect the antigen from degradation, and has been shown to induce significant antibody responses when delivered orally with various vaccines. In translational research, from laboratory bench to industrial scale-up, it is necessary to optimise the manufacturing process in order to improve efficiency and simplify production, giving a more economical end-product. In this study we tested two simplified production methods (3-step and 1-step) along with two different storage methods (lyophilised and non-lyophilised), as well as looking at the effect of buffer pH. The formulations were assessed in a murine system for immunogenicity, alongside characterisation in terms of size and antigen entrapment, with the stability of these aspects assessed with respect to time. Both lyophilised and non-lyophilised 3-step formulations induced significant IgG1, IgG2a and IgA titres, with the lyophilised version showing stable size and antigen entrapment up to 9 months. Ó 2009 Elsevier Inc. All rights reserved.
1. Introduction The influenza virus is an RNA virus, which infects millions of people around the world in seasonal epidemics, and has also killed tens of millions in pandemics that occur when a new human strain of the virus appears [1]. The average excess annual mortality attributed to influenza in the UK is approximately 14,000 deaths, with up to 90,000 deaths occurring in pandemic years [2]. Influenza epidemics also have a significant socio-economic impact, with direct costs such as hospitalisation and treatment, and indirect costs such as loss of productivity and absenteeism. During 1995–1996 the direct cost of treating influenza in Germany was estimated to be US$2.8 million, with indirect costs estimated at US$800 million [3]. At present most influenza vaccination programmes involve subdermal or intramuscular injection of the antigen, a procedure which is invasive, painful, and requires trained professionals. This type of administration only provides systemic protection via IgG antibodies, leaving the principal vector for infection, the mucosal membranes of the respiratory tract, undefended [4,5]. Mucous membranes cover an area of approximately 400 m2 of the body, most of which is vulnerable monolayered, epithelium [6], with protection provided mainly through antibody IgA [5]. Therefore a pain-free, simple, and non-invasive means of administering an
* Corresponding author. E-mail address: [email protected] (E. Bennett). 1046-2023/$ - see front matter Ó 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.ymeth.2009.04.015
influenza vaccine, that induces IgA as well as systemic IgG in humans, would simplify and improve any vaccination programme. Adjuvant formulations have previously been used to enhance the immune response elicited by an antigen [7], and particulate adjuvants are especially useful because of their similarity in size to the pathogens and their ability to present antigen in a multimeric format, causing uptake by antigen-presenting cells (APC) [8]. One such formulation uses the alternating hydrophobic and hydrophilic nature of non-ionic surfactant vesicles (NISV), which allows antigen to be entrapped. NISV are formed by bilayers of hydrated phospholipids, a structure which allows a wide range of chemical and physical modifications of the vesicle to be performed, and compounds with a variety of solubilities to be entrapped. The incorporation of bile salts into these NISV stabilises them against the effects of bile acids [9] and the resulting vesicles are termed bilosomes [10]. The bilosome has been shown to be an effective delivery system for a number of vaccine antigens, including A/Panama [11], tetanus toxoid [12] and hepatitis B [13]. These studies have shown significant antibody titres in a murine experimental model, and an infection challenge study in ferrets vaccinated with influenza haemagglutinin entrapped in bilosomes has shown higher antibody production, lower temperatures and reduced symptoms compared with IM injection [14]. Therefore, the next stage in developing a viable clinical vaccine is to optimise the bilosome formulation into a simple and efficient method whilst retaining immunogenicity. The aim of this paper is to describe how we streamlined the manufacturing process for bilosomes described by Mann et al .
E. Bennett et al. / Methods 49 (2009) 322–327
(2004) [11] for industrial scale-up. The published method was simplified from the original Mann method into a 3-step and a 1-step manufacturing process, with freeze-dried versions of each prepared to examine the suitability of lyophilisation for improving storage stability. In addition, the effect of buffer pH (9.7 versus 7.6), which can restrict certain antigens from use with this system due to precipitation, was also assessed. Comparison between the groups was made of particle size and protein entrapment efficiency, followed by evaluation of vaccine efficacy after oral immunisation of influenza antigen (haemagglutinin, HA from New Caledonian strain) in BALB/c mice using a standardised regime.
2. Materials and methods 2.1. Vesicle preparation 2.1.1. Original Mann et al. (2004) method [11] 1-Monopalmitoyl glycerol (150 lmol, Larodan AG, Sweden), cholesterol (Sigma–Aldrich, UK) and diacetyl phosphate (Sigma– Aldrich, UK) (5:4:1 M ratio) were melted in an oil bath at 140 °C for 2 min, then cooled to 60 °C and hydrated with 3.32 ml of 0.025 M carbonate buffer, pH 9.7. The lipid mixture was homogenised for 5 min at 8000 rpm (Silverson Machines Ltd., UK), followed by addition of 100 mg/ml deoxycholic acid (Sigma– Aldrich, UK) in 0.025 M carbonate buffer, pH 9.7 (preheated to 30 °C) and a further 2 min homogenisation. This mixture was incubated at 30 °C for 2 h in a water bath, then 75 lg of antigen added and the final volume made up to 5 ml with carbonate buffer before being subjected to five freeze–thaw cycles in liquid nitrogen. Unentrapped antigen was removed by centrifugation at 200,000 g for 2 h using a Beckman XL-90 ultracentrifuge (Beckman RIIC, UK), with the vesicles resuspended in 2 ml of 0.025 M carbonate buffer, pH 9.7.
2.1.2. Modified method Briefly, a 5:4:1 M ratio of lipids (all quantities are for a final volume of 10 ml): 248 mg 1-monopalmitoyl glycerol, 234 mg cholesterol and 82 mg dicetyl phosphate were melted in a 25 ml round bottomed flask using an oil bath at 120 °C, then hydrated as follows. (a) 3-Step method: Addition of 3.78 ml of 0.025 M carbonate buffer, pH 9.7 or 7.6 (preheated to 60 °C), then homogenisation for 2 min, followed by addition of 1 ml bile salt (final concentration 10 mM, preheated to 60 °C) and homogenised for 8 min at 8000 rpm. The resultant mixture was then cooled to 30 °C over 2 h in a water bath, followed by the addition of 5.22 ml carbonate buffer (0.025 M, pH 9.7) containing 1.2 mg antigen (preheated to 30 °C) and homogenised for 1 min at 8000 rpm. When examining the effect of pH 0.025 M carbonate buffer was also used at pH 7.6. (b) 1-Step method: One millilitre of 100 mM bile salt dissolved in 0.025 M carbonate buffer, pH 9.7, and 9 ml antigen solution (1.2 mg antigen in 9 ml of 0.025 M carbonate buffer, pH 9.7, preheated to 60 °C), followed by homogenisation for 13 min at 8000 rpm, then cooled to 30 °C over 2 h in a water bath with occasional agitation.
2.2. Lyophilisation of vesicles The vesicles were usually stored as an emulsion at 4 °C, but in order to improve stability, ease of storage and shipping, the vesicles were lyophilised using an Edwards Modulyo freeze drier at
323
45 °C for 24 h, then stored sealed at 4 °C. The effect of lyophilisation on immunisation efficacy was examined. 2.3. Vesicle sizing/zeta potential 2.3.1. By dynamic light scattering Particle sizing and zeta potential measurements were made on a Malvern Instruments Zetasizer Nano ZS at 25 °C using triplicate readings and a 2 min equilibration time. 2.3.2. By freeze fracture electron microscopy (FFEM) Samples were sandwiched between two copper support plates (Bal-tec, Lichtenstein) and frozen in a cryogenic mixture of propane/isopentane (3:1 v/v) at 190 °C. These were then stored at 150 °C. Fracturing was carried out at 100 °C at 106 torr, with the fracture face replicated with platinum/carbon (Pt/C, Agar Scientific, UK) at 45 °C, and carbon coated at 90o to the surface. Acetone was used to clean the replicates and they were collected onto a mesh (100 grid bars/in.), dried and viewed on an LEO 912 electron microscope at 80 kV. 2.4. Protein entrapment Entrapment of protein was quantified using a modified ninhydrin assay, a method reported to be unaffected by lipid interference and previously described by Brewer et al. (1995) [14]. In order to separate entrapped antigen from free antigen a 0.11 ml sample of vesicles, diluted in 4 ml of 0.025 M carbonate buffer, pH 9.7, was spun in a Beckman tube in a Beckman XL-90 ultracentrifuge (Beckman RIIC, UK) at 35,000 rpm for 2 h. The supernatant was then discarded and the pellet resuspended in 0.11 ml 0.025 M carbonate buffer, pH 9.7, then transferred to 1.5 ml microfuge tubes (Elkay, UK). The samples, along with standards prepared with 0, 0.25, 0.5, 0.75, 1, 2.5, 5, 7.5, 10, 15 and 20 ll of antigen, all in 1.5 ml microfuge tubes, were placed in an oven at 90 °C overnight. One hundred and fifty microlitres of 13.5 M NaOH was then added to each tube, with a pinhole made in the lid before autoclaving at 121 °C/131 kPa for 20 min. After removal from the autoclave, the holes were sealed with autoclave tape and the NaOH neutralised with 250 ll glacial acetic acid, followed by vortexing for 5 s. 500 ll of ninhydrin reagent (Sigma–Aldrich, UK) was then added to each tube, each sample vortexed and placed in a water bath at 90 °C for 20 min. Two hundred and fifty microlitres of the resultant mixture was transferred to a fresh tube containing 750 ll 50% (v/v) propan-2-ol and vortexed, and 200 ll of each sample transferred to a flat-bottomed 96-well plate (Iwaki, Japan). Absorbance was read at 540 nm on a SpectraMax 190 plate reader (Molecular Devices, USA), with test sample levels determined by linear regression from the standard calibration curve. The quantity in the supernatant was subtracted from the total protein added to calculate the entrapment efficiency. Experimental conditions were the same for standards and samples and a correlation coefficient of >0.99 was obtained. 2.5. Immunisation and sampling In-house bred male BALB/c mice, 8–10 weeks old, housed in a fully climatised room were randomised and placed into 10 groups (n = 5). All mice were starved, but allowed access to water, for 2 h pre-immunisation, with food and water available ad libitum between immunisations. New Caledonian haemaglutinin (kindly supplied by Solvay Pharmaceutical, Netherlands) was entrapped into: (a) Study 1: bilosomes using the 3-step and 1-step methods and comparison made with the equivalent freeze dried vesicles. (b) Study 2: 3-step methods with either pH 9.7 or 7.6 carbonate buffer. Each dose consisted of 0.4 ml bilosome, containing approximately
324
E. Bennett et al. / Methods 49 (2009) 322–327
50 lg N/Cal HA), administered by intragastric gavage on days 1, 4, 14, and 17. Freeze-dried bilosomes were reconstituted prior to immunisation with 0.4 ml carbonate buffer. 0.4 ml has been found to be a suitable volume to induce a good immune response [12], and retention of the total volume in the stomach could be inferred from the absence of any reflux or nasal discharge. The immunisation regime was also examined by intramuscular dosing fortnightly, on days 1, 14, 14, 17, administered in two sites in the hind leg (0.05 ml per site). The groups consisted of: (a) Study 1: Group 1 (experimental, 3-step bilosomes with antigen), Group 2 (experimental, 3-step, freeze-dried bilosomes with antigen), Group 3 (experimental, 1-step bilosomes with antigen), Group 4 (experimental, 1-step freeze-dried bilosomes with antigen), Group 5 (control, empty 1-step bilosomes), Group 6 (control, intramuscular injection). (b) Study 2: Group 1 (experimental, 3-step bilosomes with antigen, pH 9.7 buffer), Group 2 (experimental, 3-step bilosomes with antigen, pH 7.6 buffer), Group 3 (control, empty bilosomes, pH 9.7 buffer). Tail bleeds were collected in heparinised capillary tubes on days 7, 20, and 33 post-immunisation (Groups 1–5) and days 20, 33, 47 (Group 6), and centrifuged at 1000g in 1.5 ml microfuge tubes for 20 min. Plasma was transferred into fresh 0.5 ml micro-centrifuge tubes (Fischer, UK), and stored at 70 °C until IgG1/IgG2a levels were determined by enzyme linked immunosorbent assay (ELISA). The study was terminated on day 54. IgA levels was determined by ELISA of lung lavages obtained by perfusing the lungs with 0.5 ml PBS. 2.6. Enzyme linked immunosorbent assay (ELISA) Unless otherwise stated, all incubations were carried out for 1 h at 37 °C and washes three times with PBS, pH 7.4, containing 0.01% (v/v) Tween 20 (PBS–Tween), respectively. Ninety-six well tissue culture plates (Iwaki, Japan) were coated with 60 ll/well containing 0.2 lg N/Cal and incubated, then washed three times and blocked with 200 ll/well of 3% (w/v) MarvelTM (Premier Foods UK Ltd.) in PBS. The plates were washed three times and 150 ll/well of a 1:100 dilution of sera added to the first column, then 100 ll/ well of PBS added to the rest of the plate, with a 1:3 serial dilution created by transferring 50 ll from the first column to the next across the plate. The plates were incubated and washed, followed by addition of 100 ll/well of a 1:3000 dilution of IgG1, IgG2 or IgA goat anti-mouse (Southern Biotech, UK), followed by incubation for 45 min. The plates were washed and incubated with 150 ll/well of TMB substrate (250 ll of stock 3,30 ,5,50 -teramethyl benzidine 6 lg/ml dissolved in dimethylsulphoxide (DMSO), added to 25 ml of 0.1 M sodium acetate citrate buffer, pH 5.5, with 4 ll of 30% (v/v) H2O2). After 15 min the reaction was stopped with 50 ll/ well of 10% (v/v) H2SO4, and the absorbance at 450 nm read on a SpectraMax 190 plate reader.
5.0 nm, and empty = 211.8 ± 45.0 nm. These differences were not found to be significant. Lyophilisation of the bilosomes caused an increase in size on rehydration, giving a mean size ± SD of 793.9 ± 39.8 nm for 1-step and 976.7 ± 94.5 nm for 3-step. There was no significant between these groups, with both having p < 0.05 versus all non-lyophilised formulations. The analysis of the 3-step and 1-step vesicles using a LEO912 energy filtering microscope at 80 kV showed spherical structures without lipid sheets or crystalline bodies, with two populations of approximately 150–250 nm and 800–1000 nm, confirming the dynamic light scattering results (Fig. 1). In all cases the zeta potential of the bilosomes was around 100 mV, with no significant differences between the groups (Table 1). 3.1.2. Antigen entrapment For both the 1-step and 3-step protocols, antigen entrapment efficiency was roughly the same, at approximately 30 ± 5%. This was enhanced by lyophilisation, which gave 50 ± 5% entrapment. 3.1.3. Size/entrapment – stability over time and effects of homogenisation time Both 3-step and 3-step lyophilised bilosomes were monitored over 9 months to observe any changes in size or entrapment with time, whilst stored at 4 °C. The size of the 3-step bilosomes remained stable up to 4 weeks after formulation, and then increased in size from around 200 nm at 0 weeks to approximately 350 nm during the mid-stages, with a final size of 560 nm at the 9-month point. As size increased for the 3-step bilosomes there was a concurrent decrease in antigen entrapment, falling from 30% initially to 20% after 9 months (Fig. 2). The size and entrapment of the lyophilised 3-step bilosomes (after rehydration) was unchanged at any time points, remaining within ±1 standard deviation of the original (data not shown). Examining the consequences of homogenisation time showed no apparent affect on size (Fig. 3).
2.7. Statistical analysis Statistical analysis was performed using a two-tailed unpaired t-test with Minitab v15. 3. Results 3.1. Bilosome characteristics 3.1.1. Size/zeta potential Size analysis using the Malvern Instruments Zetasizer Nano ZS indicated that the melt homogenisation methods formed bilosomes between 150 and 200 nm with a small population around 1000 nm, with mean sizes ± SD of: 1-step = 165.9 ± 27.3 nm, 3step = 194.7 ± 83.7 nm, pH 9.7 = 233.9 ± 7.0 nm, pH 7.6 = 250.3 ±
Fig. 1. FFEM of bilosomes produced by the 3-step method before lyophilisation. Imaging was carried out using a LEO912 energy filtering electron microscope at 80 kV.
325
E. Bennett et al. / Methods 49 (2009) 322–327
Formulation
Size (nm) ± SD
Zeta potential (mV) ± SD
1-Step 3-Step 1-Step lyophilised 3-Step lyophilised pH 9.7 pH 7.6 Empty
165.9 ± 27.3 194.7 ± 83.7 793.9 ± 39.8 976.7 ± 94.5 233.9 ± 7.0 250.3 ± 5.0 211.8 ± 45.0
102.7 ± 11.9 104.5 ± 13.9 102.5 ± 13.2 113.5 ± 6.4 113.6 ± 11.9 112.8 ± 14.5 90.1 ± 11.5
A
16.00
Ln mean end point titre +/- S.E.
Table 1 Size/zeta potential of bilosomes formulations determined using 20 ll sample in 1 ml filtered 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS.
14.00
* p = 0.004 ~ p = 0.474
* p = 0.033 ~ p = 0.091
Day 7 Day 20 Day 33
12.00 10.00 8.00 6.00 4.00 2.00 0.00
Size (nm)
800
3s
45
3s fd
1s
% entrapment
Size (nm) +/- S.D.
30
500
25 400 20 300
15
200
10
100
5
0 5
10
15
20
25
30
0 40
35
B Ln mean end point titre +/- S.E.
35
600
Antigen entrapment (%) +/- S.D.
40
700
0
Time (weeks)
empty 1s
IM
12 * p = 0.001 ~ p = 0.004
* p = 0.029 ~ p = 0.039
Day 7 Day 20 Day 33
10
8
6
4
2
0
3s
500 450 400 350 300 250 200
3s fd
1s
1s fd
empty 1s
IM
Group
C Ln mean end point titre +/- S.E.
Fig. 2. Size (nm)/% entrapment versus time, 3-step bilosomes. Size was measured using 20 ll of sample in 1 ml 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS, and antigen entrapment was determined by ninhydrin assay.
Size (nm) +/- S.D.
1s fd Group
9 8
* p = 0.012 ~ p = 0.100
* p = 0.003 ~ p = 0.025
7 6 5 4 3 2 1 0 3s
150
3s FD
1s
1s FD
empty 1s
IM
Group
100 50 0 0
2
4
6
8
10
12
14
16
Homogenisation time (mins) Fig. 3. Size (nm) versus homogenisation time (min). Size was measured using 20 ll of sample in 1 ml of 0.025 M, pH 9.7, carbonate buffer in a Malvern Zetasizer Nano ZS.
3.2. In vivo antibody response (a) IgG1 – Significant endpoint titres were observed for both 3step groups, compared with empty bilosomes (Fig. 4) (3-step Ln titre ± SE 13.17 ± 1.81, p = 0.004, 3-step lyophilised Ln titre ± SE 11.86 ± 2.00, p = 0.033), although there was no significant difference between either of the 3-step groups (p > 0.05). Similarly, both 1-step groups showed no significant difference in titres. Both pH 9.7 and 7.6 buffer groups also induced significant titres compared with the empty bilosome group (pH 9.7 Ln titres ± SE 11.42 ± 0.49, p = 0.002; pH 7.6 Ln titres ± SE 9.66 ± 1.25, p = 0.006; empty Ln titre ± SE 5.92 ± 1.80), with pH 9.7 having a greater effect than pH 7.6 (p = 0.031) (Fig. 5).
Fig. 4. Reciprocal endpoint titres are shown in (A) IgG1, (B) IgG2a and (C) IgA, with each value representing the mean natural log of the endpoint titre ± SE. Significantly high levels of antibody were observed on completion of the study (day 33), indicated by the p values: *comparison with empty bilosome group, comparison with IM group. 3s = 3-step bilosomes, 3s FD = 3-step freeze-dried bilosomes, 1s = 1step bilosomes, 1s FD = 1-step freeze-dried bilosomes, Empty 1s = empty 1-step bilosomes, IM = Intramuscular injection.
(b) IgG2a – Both 3-step groups again induced significant titres compared with the empty bilosome group (3-step Ln titre ± SE 9.00 ± 0.78, p = 0.001; 3-step lyophilised Ln titre ± SE 8.56 ± 1.67, p = 0.029) with no difference between the two groups and no significant titres observed with the 1-step groups. Significant titres were observed with pH 9.7 and 7.6 buffer groups (pH 9.7 Ln titre ± SE 9.00 ± 1.55, p = 0.007; pH 7.6 Ln titre ± SE 9.22 ± 1.81, p = 0.011; empty Ln titre ± SE 5.92 ± 0.92) with no difference between the two pH groups. (c) IgA – Significant titres were observed in both 3-step groups (3-step Ln titre ± SE 6.58 ± 0.92, p = 0.012; 3-step lyophilised Ln titre ± SE 7.24 ± 0.99, p = 0.003), with no difference between the two groups and no significant titres for 1-step groups. Both groups with altered buffer pH failed to induce significant IgA levels.
326
Ln mean end point titre +/- S.E.
A
E. Bennett et al. / Methods 49 (2009) 322–327
14 * p = 0.002
12 * p = 0.006
10
Day 7 Day 20 Day 33
8 6 4 2 0
9.7
7.6
Clortno
Group
Ln mean end point titre +/- S.E.
B
12 * p = 0.011
* p = 0.007
10
Day 7 Day 20 Day 33
8 6 4 2 0
9.7
7.6
Control
Group
Ln mean end point titre +/- S.E.
C
7 6 5 4 3 2 1 0 9.7
7.6
Control
Group
Fig. 5. Reciprocal endpoint titres are shown in (A) IgG1, (B) IgG2a and (C) IgA, with each value representing the mean natural log of the endpoint titre ± SE. Significantly high levels of antibody were observed on completion of the study (day 33) indicated by the p values: *comparison with empty bilosome group, comparison with IM group. 3s = 3-step bilosomes, 3s FD = 3-step freeze-dried bilosomes, 1s = 1step bilosomes, 1s FD = 1-step freeze-dried bilosomes, Empty 1s = empty 1-step bilosomes, IM = Intramuscular injection.
4. Discussion Development of oral formulations for vaccine delivery presents many challenges, and the aim of these studies was to address some of these issues, such as storage, ease of formulation and examining the effect of buffer pH to prevent precipitation of some synthetic peptides. This was performed by characterising, in terms of size, zeta potential and antigen entrapment, and immune response to two manufacturing protocols (3-step and 1-step) as well as the respective lyophilised formulations. With all the methods of formulation, the size range determined by dynamic light scattering was between 150 and 220 nm, although a small population (2– 4%) was observed at around 1000 nm, a result that was confirmed
by FFEM, with no lipid sheets or crystalline structures observed. Lipid vesicles have highly variable sizes, dependant mainly on formulation method and lipid constituents, and can range from 50 nm to several microns in size [15–17]. However, the size range obtained with our modifications is consistent with bilosomes formed by our earlier studies [11]. Lyophilisation of liposomes has previously been shown to produce anywhere between a 2and 7-fold increase in size on rehydration [18], therefore the 4-fold increase observed here is within the expected range. It is possible to limit this expansion using cryoprotectants [19]; however, there appears to be no deleterious effect on the immune response, and indeed IgA titres appear higher in these groups, although not significantly so, making this an unnecessary requirement. The longterm stability of the lyophilised 3-step bilosomes is encouraging; as there is no significant change in size or entrapment, indicating a stable formulation, although the change in size and entrapment, observed with the non-lyophilised variant is to be considered, and it will be desirable to test immunogenicity versus storage time in future work. Antigen entrapment in the 3-step and 3-step freeze-dried formulations of N/Cal was found to be approximately 30% and 50%, respectively, a result which is comparable with previous work with the A/Panama flu vaccine (50%, [11]), although entrapment has been shown to be highly variable and dependent on the antigen used [20]. Zeta potential is mainly dependent on the lipid components, with both cationic [21,22] and anionic [23] liposomes available using different lipids, with both types of liposome also possible with the same components in different ratios [24]. The zeta potential in each of the formulations we tested was around 100 mV, which is similar to other work with liposomes containing both cholesterol and diacetyl phosphate [23]. The dosing regime was well tolerated by the animals, with no overt toxicity observed 24 h post-immunisation. No significant responses were observed for either of the 1-step methods. Both of the 3-step formulations induced similar immune responses to HA, indicated by the significantly raised IgG1 titres compared with the empty bilosome group, with a modest IgG2a response. IgA levels in these two groups were also significantly higher than those treated with empty bilosomes, although only the lyophilised group were significantly greater versus IM injection. In the buffer pH study both pH 9.7 and pH 7.6 induced good IgG1 and IgG2a responses, although responses were higher for IgG1 with pH 9.7. In this study however, no significant IgA titres were observed, an unusual result given our previous data [11]. This may reflect the change in method, from measurement of IgA in faecal matter [11] to measurement in lung lavage fluid. However, the lavage method provides a direct measure of lung IgA levels, whilst analysis of faecal matter is an indirect measure, and may not give an accurate representation of actual levels in the lungs. These results are similar to those of previous work on bilosomes manufactured according to the Mann et al. (2004) [11] method with entrapped A/Panama antigen, although no IgG2a response was observed in this case. The skewing of the response towards IgG1 may be explained by the observations that vesicles of diameter less than 250 nm, administered parenterally, induce a TH2 biased response [25], whilst those greater than 250 nm skew towards TH1 [14]. With the majority of the bilosome population around 200 nm a TH2 response would therefore be expected, however, the small population at around 1000 nm appears to be inducing an additional TH1 response. With around 90% of infections involving the mucosal surfaces of the body [26], it is apparent that induction of mucosal immunity (mediated mainly by IgA) would provide a great advantage in combating these infections. In elderly patients this is particularly relevant, as there is a reduction in immune response with age [27,28],
E. Bennett et al. / Methods 49 (2009) 322–327
and it is estimated that only between 10% and 30% of over 65 s are afforded protection from influenza vaccination [29]. Therefore, they are placed at a disadvantage by enabling the virus to gain a foothold on mucosal membranes before any protective response occurs. Induction of an IgA response has other advantages; it can neutralise the influenza virus itself, in the absence of serum IgG [30–32], and it has been shown that, whilst conventional parenteral administration of vaccine mediated by IgG provides adequate protection against homologous viral infections, it has little efficacy against emerging variant strains [33]. In this situation, where there may be an absence of a direct vaccine for these variants, or the patient may not be immunised against them, protection can still be achieved through induction of an IgA response [31,33,34], an issue which is particularly relevant for high risk demographics. Despite the clear benefits of a mucosal IgA response, it remains necessary for any vaccine to induce virus neutralising antibodies in serum in order to attain regulatory approval [35], something which has not been achieved in clinical studies [4,36–38]. Here we have demonstrated the capability of the bilosome delivery system to induce a mucosal response, with significant IgA titres observed for both 3-step formulations, a result previously observed with this system with A/Panama [11]. The bilosome provides a proven delivery system that offers the potential to preclude the need for IM injection of influenza vaccines, improving patient compliance with the benefits of mucosal as well as systemic protection. Since lyophilised versions offer similar levels of IgG1 and IgA, this removes the need for cold-chain issues, thus improving storage and distribution and allowing vaccine programmes to operate in areas of the world lacking in medical infrastructure. Whilst this may be less relevant in terms of influenza, it would provide a great benefit with other antigens which can be used with bilosomes, such as those derived from cholera toxin [17], hepatitis B [13] and tetanus toxoid [12]. Acknowledgments E.B. is funded through an EPSRC-Strathclyde University Doctoral Training Centre studentship (Bioengineering Department). The authors thank Mr. George Walker, Mr. Gordon Watt for their technical assistance. References [1] N. Johnson, J. Mueller, Bulletin of the History of Medicine 76 (1) (2002) 105– 115. [2] S.L. Johnston, Virus Research 82 (1–2) (2002) 147–152. [3] T. Szucs, Journal of Antimicrobial Chemotherapy 44 (1999) 11–15. [4] Z. Moldoveanu, M.L. Clements, S.J. Prince, B.R. Murphy, J. Mestecky, Vaccine 13 (11) (1995) 1006–1012.
327
[5] P.L. Ogra, H. Faden, R.C. Welliver, Clinical Microbiology Reviews 14 (3) (2001) 641. [6] P. Brandtzaeg, Vaccine 25 (30) (2007) 5467–5484. [7] N. Petrovsky, J.C. Aguilar, Immunology and Cell Biology 82 (5) (2004) 488–496. [8] M. Copland, T. Rades, N.M. Davies, M.A. Baird, Immunology and Cell Biology 83 (2005) 97–105. [9] R. Schubert, H. Jaroni, J. Schoelmerich, K.H. Schmidt, Digestion 28 (1983) 181– 190. [10] M. Conacher, J. Alexander, J.M. Brewer, Vaccine 19 (20–22) (2001) 2965–2974. [11] J.F.S. Mann, V.A. Ferro, A.B. Mullen, L. Tetley, M. Mullen, K.C. Carter, J. Alexander, W.H. Stimson, Vaccine 22 (2004) 2425–2429. [12] J.F.S. Mann, H.E. Scales, E. Shakir, J. Alexander, K.C. Carter, A.B. Mullen, V.A. Ferro, Methods 38 (2005) 90–95. [13] A. Shukla, K. Khatri, G. Gupta, A. Goyal, A. Mehta, S. Vyas, Journal of Pharmacy and Pharmaceutical Sciences 11 (1) (2008) 58–66. [14] J.M. Brewer, C.W. Roberts, W.H. Stimson, J. Alexander, Vaccine 13 (15) (1995) 1441–1444. [15] J.M. Brewer, K.G.J. Pollock, L. Tetley, D.G. Russell, Journal of Immunology 173 (10) (2004) 6143–6150. [16] G.F.A Kersten, D.J.A. Crommelin, Vaccine 21 (2003) 915–920. [17] P. Singh, D. Prabakaran, S. Jain, V. Mishra, K.S. Jaganathan, S. Vyas, International Journal of Pharmaceutics 278 (2004) 379–390. [18] P.A. Bridges, K.M.G. Taylor, Journal of Pharmacy and Pharmacology 53 (3) (2001) 393–398. [19] L.J. Zhang, L. Liu, Y. Qian, Y. Chen, European Journal of Pharmaceutics and Biopharmaceutics 69 (2) (2008) 750–759. [20] G. Gregoriadis, B. McCormack, M. Obrenovic, R. Saffie, B. Zadi, Y. Perrie, Methods 19 (1999) 156–162. [21] J. duPlessis, C. Ramachandran, N. Weiner, D.G. Muller, International Journal of Pharmaceutics 127 (2) (1996) 273–278. [22] Y. Perrie, P.M. Frederik, G. Gregoriadis, Vaccine 19 (23–24) (2001) 3301–3310. [23] D. Gopinath, D. Ravi, B.R. Rao, S.S. Apte, D. Renuka, D. Rambhau, International Journal of Pharmaceutics 271 (1–2) (2004) 95–113. [24] J.C. Birchall, I.W. Kellaway, S.N. Mills, International Journal of Pharmaceutics 183 (2) (1999) 195–207. [25] J.M. Brewer, L. Tetley, J. Richmond, F.Y. Liew, J. Alexander, The Journal of Immunology 161 (8) (1998) 4000–4007. [26] P. Brandtzaeg, International Journal of Medical Microbiology 293 (1) (2003) 3– 15. [27] K. Leder, P.F. Weller, M.E. Wilson, Clinical Infectious Diseases 33 (9) (2001) 1553–1566. [28] G. Wick, B. Grubeck-Loebenstein, Experimental Gerontology 32 (4) (1997) 401–413. [29] D.C. Powers, E.D. Kilbourne, B.E. Johansson, Clinical and Diagnostic Laboratory Immunology 3 (5) (1996) 511–516. [30] R.B. Belshe, W.C. Ruber, P.M. Mendelman, et al., Journal of Infectious Diseases 181 (3) (2000) 1133–1137. [31] F.Y. Liew, S.M. Russell, G. Appleyard, C.M. Brand, J. Beale, European Journal of Immunology 14 (4) (1984) 350–356. [32] K.B. Renegar, P.A. Small, Journal of Virology 65 (4) (1991) 2146–2148. [33] R. Ito, Y.A. Ozaki, T. Yoshikawa, et al., Vaccine 21 (19–20) (2003) 2362– 2371. [34] T.M. Tumpey, M. Renshaw, J.D. Clements, J.M. Katz, Journal of Virology 75 (11) (2001) 5141–5150. [35] A. Huckriede, L. Bungener, T. Stegmann, T. Daemen, J. Medema, A.M. Palache, J. Wilschut, Vaccine 23 (2005) S26–S38. [36] S.S. Avtushenko, E.M. Sorokin, N.Y. Zoschenkova, N.G. Zacharova, A.N. Naichin, Journal of Biotechnology 44 (1–3) (1996) 21–28. [37] K.C. Bergmann, R.H. Waldman, H. Tischner, W.D. Pohl, International Archives of Allergy and Applied Immunology 80 (1) (1986) 107–109. [38] V. Lazzell, R.H. Waldman, C. Rose, R. Khakoo, et al., Journal of Biological Standardization 12 (3) (1984) 315–321.