Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation

G Model

ARTICLE IN PRESS

PBA-11999; No. of Pages 9

Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

Contents lists available at ScienceDirect

Journal of Pharmaceutical and Biomedical Analysis journal homepage: www.elsevier.com/locate/jpba

Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation Sasmit S. Deshmukh a,b,1 , Federico Webster Magcalas a,c,1 , Kristen N. Kalbfleisch a,c , Bruce W. Carpick a , Marina D. Kirkitadze a,∗ a

Sanofi Pasteur Ltd., 1755 Steeles Avenue West, Toronto, ON, M2R 3T4, Canada SGS Canada, Biopharmaceutical Services, 6490 Vipond Drive, Mississauga, ON, L5T 1W8, Canada c Biotechnology Advanced Program, Seneca College, 70 The Pond Road, Toronto, ON, M3J 3M6, Canada b

a r t i c l e

i n f o

Article history: Received 6 December 2017 Received in revised form 28 May 2018 Accepted 30 May 2018 Available online xxx Keywords: ® IC31 adjuvant H4-IC31 vaccine candidate Structural changes PALS Raman spectroscopy

a b s t r a c t Tuberculosis (TB) is one of the leading causes of death worldwide, making the development of effective TB vaccines a global priority. A TB vaccine consisting of a recombinant fusion protein, H4, combined with ® a novel synthetic cationic adjuvant, IC31 , is currently being developed. The H4 fusion protein consists ® of two immunogenic mycobacterial antigens, Ag85 B and TB10.4, and the IC31 adjuvant is a mixture of KLK, a leucine-rich peptide (KLKL5KLK), and the oligodeoxynucleotide ODN1a, a TLR9 ligand. However, efficient and robust methods for assessing these formulated components are lacking. Here, we developed and optimized phase analysis light scattering (PALS), electrical sensing zone (ESZ), and Raman, FTIR, and CD spectroscopy methods to characterize the H4-IC31 vaccine formulation. PALS-measured conductiv® ity and zeta potential values could differentiate between the similarly sized particles of IC31 adjuvant and the H4-IC31 vaccine candidate and could thereby serve as a control during vaccine formulation. In addition, zeta potential is indicative of the adjuvant to antigen ratio which is the key in the immunomod® ulatory response of the vaccine. ESZ was used as an orthogonal method to measure IC31 and H4-IC31 ® particle sizes. Raman, FTIR, and CD spectroscopy revealed structural changes in H4 protein and IC31 adjuvant, inducing an increase in both the ␤-sheet and random coil content as a result of adsorption. Furthermore, nanoDSF showed changes in the tertiary structure of H4 protein as a result of adjuvanta® tion to IC31 . Our findings demonstrate the applicability of biophysical methods to characterize vaccine components in the final H4-IC31 drug product without the requirement for desorption. © 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

1. Introduction Tuberculosis (TB) remains one of the leading causes of death worldwide with 10.4 million new cases reported in 2015. In the same year, 1.8 million people died from TB even with the availability of a vaccine [1]. The Bacillus Calmette-Guerin (BCG) vaccine is the only vaccine available against TB, despite its variable efficacy and its failure to control spread of the disease [1]. Although the BCG vaccine provides protection against manifestations of TB in children, it is not able to provide reliable protection against adult pulmonary TB [2]. Hence, there is an urgent need to develop a vaccine that can surpass the efficacy of the current BCG vaccine. One of the strategies in TB vaccine development is to modify or boost

∗ Corresponding author. E-mail address: Marina.Kirkitadze@sanofi.com (M.D. Kirkitadze). 1 Authors contributed equally

the immune response of the BCG vaccine [3–5]. Since BCG still confers protective immunity against TB in children, it is considered unethical to develop a vaccine strategy that does not include BCG [2]. One of the candidate TB vaccine formulations, H4-IC31 [6], is being developed by Sanofi Pasteur in collaboration with Statens Serum Institut (Copenhagen, Denmark) and Valneva (Lyon, France). ® The candidate vaccine’s novel adjuvant IC31 , which acts as an immunostimulant, is composed of two biopolymers: an antimicrobial leucine-rich peptide (KLK or KLKL5KLK or KLKLLLLLKLK) and an oligodeoxynucleotide of repeating deoxyinosine and deoxycytosine dinucleotides (polyI:C) with a natural phosphodiester backbone (ODN1a) in a molar ratio of 25:1. The KLK peptide acts as a delivery vehicle by which it forms a delivery port for antigens at the injection site, making it a very interesting adjuvant concept [7]. KLK also facilitates the uptake and delivery of ODN1a into TLR9-positive intracellular vesicular compartments [8]. ODN1a co-localizes with

https://doi.org/10.1016/j.jpba.2018.05.048 0731-7085/© 2018 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model PBA-11999; No. of Pages 9 2

ARTICLE IN PRESS S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

TLR9 positive compartments, which are endosomes and endoplas® mic reticular structures [8]. Due to these properties of IC31 , the adjuvant was shown to have a profound effect on immune response triggered by TLR9-agonists [9]. The combined effects of KLK and ODN1a stimulate antigen specific Th1 and/or Th17 adaptive mem® ory through a TLR9 signaling pathway [10]. In this study, IC31 components were characterized using spectroscopic techniques. The antigen protein component of the vaccine, H4, is a 41.3 kDa recombinant fusion protein consisting of Ag85 B and TB10.4 antigens from Mycobacterium tuberculosis [6,11]. The positive total ® surface charge of IC31 promotes the association with H4 protein through electrostatic interactions. The H4 protein and the individ® ual components of IC31 adjuvant are water soluble. Once KLK and ® ODN1a are combined, micron-sized insoluble particles of IC31 are formed and remain insoluble upon formulation with H4. Conse® quently, the physical appearance of the IC31 adjuvant and H4-IC31 vaccine candidate are very similar. Due to the presence of these particles, many widely used analytical techniques that require soluble analytes cannot be applied. Moreover, biophysical methods that can be applied to samples containing particles may not be able ® to distinguish between IC31 adjuvant alone and H4-IC31 vaccine formulation. Bioformulations such as vaccines are developed and assessed through a complex, multi-stage process. Key elements include: knowledge of underlying health and disease factors; knowledge of functional mechanisms of relevant bioformulations; development and production of candidate formulations; quality control involving product quality, safety, potency and stability; and delivery system for clinical trials and subsequent commercialization. One of the key steps in this process is to select and characterize both the individual components of these bioformulations as well as the final vaccine product. Characterization typically includes assessment of vaccine antigen structure throughout the manufacturing process. Stability assessment and quality control of antigen proteins require knowledge of antigen structure and of interactions with the surrounding matrix. Characterization of vaccine components at different manufacturing stages is critical as part of a quality control strategy [12]. Formulation of the vaccine product by adsorption of the antigen protein to the adjuvant is a critical manufacturing process step. Accordingly, a Phase Analysis Light Scattering (PALS) method was ® developed to characterize IC31 adjuvant and H4-IC31 vaccine formulation using zeta potential, conductivity, and particle size as reportable values. A particular advantage of this technology was its ability to distinguish between adjuvant alone and the adjuvanted vaccine. Protein conformational changes may result from physical adsorption to an adjuvant, and biophysical assays that are straightforward to apply to proteins in solution may be challenging in complex matrices such as adjuvants. Desorption of the antigen from the adjuvant prior to analysis may require buffer conditions which alter the protein structure and make subsequent interpretation of results problematic. In this study we report structural changes asso® ciated with H4 protein upon adsorption to IC31 adjuvant using CD, Raman, and FTIR spectroscopy, along with nano differential scanning fluorimetry (nanoDSF). The focus of this study was on the structural changes of the fully adsorbed H4 protein antigen, ® and not the adjuvant in excess, IC31 . Where the entire population of H4 undergoes structural changes as a result of adsorption, ® a majority of the IC31 population remains unaffected. Moreover, ® the constituents of IC31 function as immunostimulants, but do ® not exhibit any secondary structural elements on their own. IC31 is a much larger molecule (∼3 ␮m) compared to H4 (12 nm) with a pseudo ␤-sheet structure, which is non-protein molecule containing elements that result in spectral features as ␤-sheet containing

protein. Hence, any changes in pseudo ␤-sheet structure occurring ® in IC31 do not affect its function as an immunostimulant. Raman spectroscopy was used to characterize the individual ® IC31 components, KLK and ODN1a. In summary, we have developed a panel of biophysical methods that can monitor the TB H4-IC31 vaccine adjuvantation process, and provide characterization information on formulated vaccine components without the need for desorption. 2. Materials and methods All samples were obtained from different stages of the manufacturing process of the candidate vaccine product. The following samples were used without further modification unless otherwise ® stated: H4 antigen, IC31 adjuvant, and H4-IC31 [6,11]. Recombinant fusion H4 protein was purified as described in [13]. Its predicted pI value is 5.19, based on the amino acid sequence of individual AG85 B and TB10.4 proteins, as per the ExPASy portal. Adjuvant raw materials, KLK and ODN1a, were purchased from Bachem AG (Bubendorf, Switzerland) and Agilent Technologies Inc. (Boulder CO, USA) respectively. The buffer used for H4 and H4IC31 was 10 mM Tris at pH 7.78. The molar ratio of H4:IC31 is 1:9, ® whereas IC31 consists of ODN1a:KLK in a 1:25 molar ratio. H4 was 1.2–1.4 mg/ml in purified samples, whereas after formulation H4 concentration decreases to 0.03 mg/ml. H4 is fully adsorbed (100%) ® to the IC31 adjuvant. A formulation containing higher concentration of H4 (0.1 mg/ml) had the same molar ratio of 1:9 and was fully ® adsorbed to the IC31 adjuvant. All measurements were performed in triplicate. More than three measurements were performed by PALS and ESZ. 2.1. Phase analysis light scattering (PALS) Zeta potential is an essential parameter because depending on the magnitude of net charge, the stability of colloidal suspension can be assessed. When external potential is applied, the charged particle moves in the solution with altered velocity known as electrophoretic mobility. It depends on factors such as solvent viscosity and dielectric constant. The electrical potential at the boundary of diffuse layer is called zeta potential. The magnitude of this potential determines whether the repulsive forces exceed the attractive forces of the particles in colloidal solution, and this can ultimately be interpreted as stability of the colloidal solution. An optical probe focuses a laser beam through the colloidal sample. The light scatters back to the probe from the particles. Zeta potential can be calculated using the Smoluchowski’s equation. ␨ = (␩␮)/␧ Where, ␨: Zeta potential ␩: solvent viscosity ␮: electrophoretic mobility ␧: dielectric constant of solvent This method also measures the diameter of the particle based on the principles of dynamic light scattering (DLS), where the speed at which the particles are diffusing through solvent due to Brownian motion is measured by recording the fluctuations in the intensity of the scattered light. Using Stokes-Einstein equation one can estimate hydrodynamic radius or diameter. The particle size measurement is an intensity averaged distribution. d(H) = kT/3␲␩D where, d(H): hydrodynamic radius of the particle k: Boltzmann’s constant

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model PBA-11999; No. of Pages 9

ARTICLE IN PRESS S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

T: absolute temperature ␩: solvent viscosity D: diffusion coefficient (determined from Brownian motion of the particle). The values for the zeta potential and conductivity are displayed with the particle size distribution graph by the Microtrac FLEX software. PALS analyses for particle size and zeta potential was performed using the Zetatrac instrument (Microtrac Inc., Montgomeryville, PA, USA) with the Microtrac FLEX 10.6.1 software for data collection and analysis. Particle size, surface conductivity, and ® zeta potential values were obtained using this method. The IC31 and H4-IC31 samples were measured following 70 and 60 fold dilu® tions with Milli-Q water. The viscosity of IC31 and H4-IC31 were 3.8 and 3.75 cPs at 20 ◦ C. However since the samples were diluted prior to measurement, the viscosity of water (1 cPs at 20 ◦ C) was used for the measurements. The refractive index for the particle analysis was set to 1.57 and the total run time was 180 s for each sample. Several measurements were done using a fresh dilution of the sample. The range of the measurements was from 2.2 ␮m to 4.4 ␮m for both IC31 and H4-IC31. 2.2. Electrical sensing zone (ESZ) The MultisizerTM 4 Coulter Counter employs a non-optical method called Electrical Sensing Zone to measure the volume displaced by the particle and correlating it with the particle diameter. The samples under study were suspended in a weak electrolyte solution which was drawn at a steady rate through an aperture separated by two electrodes. When the electric current flows through the electrodes, a sensing zone is created. When a particle passes through this sensing zone, it displaces the volume equivalent to its particle size causing an increase in impedance in the aperture. These minute changes in impedance are converted into voltage pulses which are directly proportional to the volume of particles. The volume of particle is converted into equivalent spherical diameter and the size distribution is acquired by scaling the pulse heights. The instrument provided the particle size distribution graphs and the corresponding mean size distribution of the samples under study. The resulting graphs for H4 and H4-IC31 were re-plotted using SigmaPlot from the raw data without further processing. ESZ was performed using the Beckman Coulter Multisizer 4 instrument (Beckman Coulter, Miami, FL, USA) equipped with the Multisizer 4 software. An aperture of 70 ␮m was installed and calibrated using a 10 ␮m polystyrene beads standard (Beckman Coulter). For each measurement, a sample volume of 200 ␮l was ® ® added to an Accuvette ST cell with 20 ml of Isoton II (0.9% w/v NaCl solution; Beckman Coulter).

3

any user intervention or prior information aside from selection of the lambda factor which can be adjusted to manipulate the auto fitted background so as to achieve the desired background removal result. The instrument reports the intensity as a function of Raman shift which is displayed in a graph. Before measurement, the background vibrations were corrected by acquiring a dark scan in the absence of the laser. This was then automatically subtracted from each measurement. Arithmetic difference between the spectra was performed to obtain the desired spectrum. These spectra were then re-plotted using SigmaPlot. 2.4. Fourier transform infrared spectroscopy (FTIR) The FTIR spectroscopy was performed using a Vertex 70 FTIR Spectrometer (Bruker Optics, Bremen, Germany), equipped with a cryogenically-cooled MCT (mercury-cadmium-telluride) detector and BioATRII sampling accessory. A sample volume of 20 ␮l was loaded onto the sample cell of the BioATRII. The spectra were collected at a resolution of 0.4 cm−1 at 25 ◦ C with a wavenumber accuracy of 0.01 cm−1 at 2000 cm−1 . The samples were allowed 1 min to stabilize on the ATR crystal. Buffer (Milli-Q water) and sample measurements were conducted with each measurement averaging 200 scans. Data acquisition and analysis were performed using OPUS 6.5 software (Bruker Optics, Bremen, Germany). OPUS automatically subtracts the background (buffer) signal from the sample to produce the spectrum for the analyte. All measurements were carried out at 25 ◦ C using the Haake DC30/K20 temperature controller (Karlsruhe, Germany). After acquiring the FTIR spectra, the baseline was corrected by removing the scattering signal by using the OPUS software. The derivative spectrum was generated using the Savitzky-Golay algorithm which allowed simultaneous smoothing of the spectrum. Arithmetic subtractions and re-plotting were performed using SigmaPlot. 2.5. Circular dichroism (CD) spectroscopy CD spectroscopy was performed on Jasco Spectropolarimeter J-810 (Jasco Inc., Maryland, USA). The spectropolarimeter was constantly purged with nitrogen gas to protect optical elements of instrument while operating in the far-UV region. The spectra were recorded from 190 to 280 nm with temperature regulation using Peltier temperature controller at 25 ◦ C. All measurements were carried out using a 1 mm path length quartz cuvette (Hellma GmbH & Co, Müllheim, Germany). In order to obtain the spectra of H4 in H4-IC31, The spectra of IC31 was corrected and subtracted from H4-IC31. The Jasco Secondary Structure Estimation program based on Yang’s model was used to estimate secondary structural content [14].

2.3. Raman spectroscopy ®

Raman spectroscopy was performed using i-Raman Plus portable Raman system (B&W Tek Inc., Newark, DE) equipped with a 785 nm laser. The spectra were collected from 0 to 3200 cm−1 . The exposure time and laser output were optimized such that the spectral intensity is maximized for identification of all spectral features. This was important to trace minor contributions to the overall spec® trum and also for consistency of analysis. All liquid samples (IC31 and H4-IC31) were dried using a SpeedVac concentrator to improve the signal to noise ratio and extract additional spectral information. The samples were tested using quartz windows supported by aluminum foil, and the laser was directly focused on the samples. Data acquisition and analysis were performed using BWSpec4 software, whereby baseline correction was applied for all measured spectra with an empirically selected lambda factor of 355173. The software uses a baseline correction algorithm which does not require

2.6. Nano differential scanning fluorimetry (nanoDSF) The nanoDSF method was performed on a Prometheus NT.48 system (Nano Temper Technologies, Munich, Germany). nanoDSF uses intrinsic fluorescence, which is a dye-free method to evaluate changes in aromatic residues (fluorophores) within proteins in response to the changes in their local environment. The shift and intensity change in fluorescence emission is monitored, with a change in the intrinsic fluorescence indicating that the protein has unfolded. Thermal stability of protein is characterized using the melting temperature (Tm ), which indicates the point at which half the protein is unfolded. In the nanoDSF method, this is determined by using the ratio of fluorescence recorded at 330 nm and 350 nm; this ratio has shown to be more sensitive in detecting Tm as compared to the use of a single wavelength. Samples were filled in capillary tubes without any further preparation and excited at

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model

ARTICLE IN PRESS

PBA-11999; No. of Pages 9

S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

4

Fig. 1. Representative particle size distribution curves by phase analysis light ® scattering The H4-IC31 (blue trace) and IC31 (red trace) have mean particle sizes of 3.5 ± 0.72 ␮m and 2.6 ± 0.83 ␮m respectively. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Table 1 ® PALS analysis of H4-IC31 and IC31 for zeta potential and conductivity values. Sample

Zeta potential (mV)

Conductivity (␮S/cm)

H4-IC31 (60 fold dilution)

32.05 34.57 25.32 27.98 65.41 74.40 68.50 63.33

304 299 300 295 314 315 313 309

®

IC31 (70 fold dilution)

285 nm with 20% power output. The thermal profiles were recorded from 20 to 95 ◦ C with 2 ◦ C/min scan rate. 3. Results

Fig. 2. Representative particle size distribution curves by ESZ. The mean size ® distribution of H4-IC31 (blue trace) and IC31 (red trace) were in the range of 2.73.2 ␮m. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.) Table 2 Peak positions and tentative assignments of Raman bands for KLK and ODN1a. Peaks (cm−1 )

Assignment

Sample

1453 1344 1175 1123 1046 958 874 788 1485, 1328, 720 1417, 998, 596 1236 1095 780

Side chain, CH3 CH2 bend C CH bend Backbone CH2 twist C C stretch CH3 rock CH2 wag N C–C bend dA Backbone dT PO2 − − − symmetric stretch O-P-O stretch, dC

KLK KLK KLK KLK KLK KLK KLK KLK ODN1a ODN1a ODN1a ODN1a ODN1a

dA, dT: deoxynucleotides.

3.1. Particle characterization PALS technology (Zetatrac instrument) was used to measure particle size distribution, zeta potential values and conductivity ® for IC31 adjuvant and H4-IC31 adjuvanted vaccine. The overall ® particle size distribution profiles are similar for both IC31 and H4® IC31 (Fig. 1). The predominant particle sizes of IC31 and H4-IC31 were 2.6 ± 0.83 ␮m and 3.5 ± 0.72 ␮m, respectively. Theoretically the particle size of H4 should be around 3 nm but purified H4 protein contains monomers and oligomers as reported by SEC-MALS (data not shown) and DLS (Fig. S1). Therefore, a 0.9 ␮m increase in size in case of H4-IC31 is likely due to multiple layer adsorption. The polydispersity index was in the range of 0.194 − 0.217. The slightly larger H4-IC31 particles also showed a narrower size distribution ® ® versus IC31 alone (Fig. 1). IC31 had higher zeta potential and conductivity values than H4-IC31 (Table 1). Conductivity and zeta potential values changed as storage conditions such as incubation time and temperature changed. As ® incubation temperature of IC31 increased from 4 ◦ C to 25 ◦ C, the conductivity increased by approximately 40 ␮S/cm, whereas zeta potential decreased by approximately 25 mV. These differences in zeta potential and conductivity values attest to the ability of PALS to assess the stability of the adjuvant suspension in differentiating ® IC31 and H4-IC31. Hence, PALS can be used as a characterization test for both the adjuvant and the formulated vaccine product to examine the consistency between different manufacturing batches and also in comparability studies to support process changes. This is particularly important since other biophysical methods (i.e. MFI, ® DLS) were not able to differentiate between IC31 and H4-IC31.

The next step would be to further assess PALS as a potential product quality attribute, and as a potential stability indicating assay ® for IC31 and/or H4-IC31. These activities will be tied to ongoing efforts to develop in vitro potency readout(s) for the vaccine. Electrical Sensing Zone (ESZ) was applied as an orthogonal and ® more sensitive method than PALS to measure particle size of IC31 and the H4-IC31vaccine candidate. The procedure was similar to the one reported previously for BCG Immunotherapeutic [15]. Both ® H4-IC31 and IC31 were in the range of 2.7-3.2 ␮m in size (Fig. 2). The coefficient of variation values were 2.54% and 2.51% for H4® IC31 and IC31 , respectively. Consistent with results obtained with PALS, the average particle size by ESZ for H4-IC31 was slightly ® higher than IC31 alone. 3.2. Characterization of vaccine components ®

The two components of IC31 , KLK and ODN1a, were characterized by Raman spectroscopy. Raman spectra of KLK and ODN1a showed distinct spectral features associated with the backbone and other vibrational modes (Fig. 3a and 3b). The peptide component showed Raman peaks for amide I (1666 cm−1 ) and amide III (1244 cm−1 ) along with other assigned vibrational modes (Table 2). In the case of ODN1a, O-P-O stretch, dA (deoxynucleotide) and PO2 − − − symmetric stretch were observed at 780 cm−1 , 720 cm−1 and 1095 cm−1 , respectively. These peaks are in close agreement with other similar studies involving polypeptides and oligodeoxynucleotides [16,17]. As mentioned earlier, these spectral features can be used as fingerprints to identify raw materials

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model

ARTICLE IN PRESS

PBA-11999; No. of Pages 9

S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

®

Fig. 3. Raman spectra of IC31 components, a. KLK and b. ODN1a.

and potentially use this information as a quality measure. This will also help to distinguish between the native structures of KLK and ODN1a, and their non-native states that might be induced by the presence of contaminants, or by changes to the storage factors conditions. 3.3. Structural changes of H4 protein upon adjuvantation ®

Since H4 protein is adsorbed on the surface of IC31 particles, the protein was hypothesized to undergo conformational changes

5

resulting from the altered matrix. These structural changes within H4 were identified by three different spectroscopic methods: CD, FTIR, and Raman spectroscopy. ® Due to the large particle sizes of H4-IC31 and IC31 and their increased scattering in the UV region, approximate secondary structural content was probed using CD spectroscopy and compared with FTIR and Raman spectroscopy data. The spectra of H4 and H4-IC31 are presented in Fig. S2. The H4 protein showed minima at 208 nm and 222 nm indicating ␣-helical content whereas upon adjuvantation only a broad minimum was observed at 218 nm for ␤–sheet content. FTIR spectroscopy also revealed changes in the secondary struc® ture of H4 as a consequence of adsorption to IC31 adjuvant (Fig. 4). ® The FTIR spectra of IC31 and H4-IC31 appear to be similar, while the difference is only apparent via the second derivative (Fig. 4). Buffer contribution was removed from all the spectra by normalizing them to the 1065 cm-1 peak where Tris buffer has an intense C O stretch peak from alcohol (dashed traces in Fig. 4). This region was selected for normalization because it doesn’t overlap with the amide region. In order to further probe these changes, second derivative spectra were calculated from the original spectra (Fig. 4) and buffer contribution was removed in a similar way (dashed traces in Fig. 4). The final second derivative spectrum of adjuvanted H4 shown in Fig. S3b was calculated from Fig. 4 by subtracting ® IC31 from H4-IC31. The peak shifts observed within amide I and ® amide II regions when H4 was combined with IC31 indicate that the secondary structure of H4 is altered upon adsorption. As mentioned earlier, structural changes to H4 protein upon adjuvantation ® are noticeable. On the other hand, structural changes to IC31 in the presence of H4 protein are marginal due to the large popula® tion of unchanged IC31 particles (Fig. S3a). Only two new peaks ® at 1096 and 1491 cm−1 were observed in IC31 in the presence of H4 as a result of changes in the local environment of the PO2− ® group and KLK. Both H4 and IC31 showed structural changes once formulated into a complex. Nevertheless, H4 appears to undergo greater changes on adjuvantation as shown by an increase of ␤sheet content in a formulation containing a higher concentration of H4 antigen. This allowed the detection by FTIR secondary structure

®

Fig. 4. (a) FTIR spectra of H4 (green), H4-IC31 (blue), IC31 (red) and Tris-HCl buffer (black). These spectra are shown with vertical offset for clarity. Tris-HCl buffer was subtracted from each sample (represented by dashed traces with similar color codes) to eliminate the buffer contribution in the sample spectra. (b) Calculated second derivatives of the corresponding spectra from panel a. Contributions from Tris-HCl buffer were subtracted from all second derivative spectra (represented by dashed traces with similar color codes). All spectra were normalized to 1065 cm-1 (solid vertical line) before subtraction where Tris-HCl buffer has large absorption and does not screen information about secondary structural changes. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model PBA-11999; No. of Pages 9 6

ARTICLE IN PRESS S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

®

Fig. 5. Raman spectra of H4-IC31 (blue trace), IC31 (red trace), and free H4 (green). Traces are vertically shifted by offset for clarity. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

analysis software Quant 2, which showed that H4-IC31 contained approximately 20% ␣-helical content, and 65% ␤-sheet content (Fig. S4). Individually, free H4 had approximately 20% ␣-helical con® tent and 45% ␤-sheet content, and IC31 had approximately 30% ␣-helical content and 60% ␤-sheet content (Fig. S4). The expected ␤-sheet content for the complex based on molar ratios is approximately double the ␣-helical content, however greater than 3 times the content is seen. The changes observed can be attributed to the ® ␤-sheet content of H4 and in the pseudo ␤-sheet content of IC31 . Raman spectroscopy was also done to identify structural changes of H4 upon adjuvantation (Fig. 5). In H4, ␣-helical content was observed by Raman peaks at 529 cm−1 and 890 cm−1 while a weak signal for ␤-sheet was present at 1670 cm−1 . In case of adsorbed H4 the peak at 1670 cm−1 has much higher intensity than the corresponding peak in free H4. Vibrations of aromatic side chains were also observed at 1618 cm−1 in the free form. The Raman peak at 640 cm−1 from tyrosine is quite apparent in H4-IC31 indicating formation of more hydrogen bonds in H4 after adjuvantation. ® Spectra for H4, H4-IC31 and IC31 were background corrected, while the difference spectrum was generated from H4-IC31 and ® IC31 spectra to yield spectral information of the adjuvanted form of the protein within the formulation matrix (Fig. S5a). Similarly, to ® probe structural changes of IC31 in the presence of H4, the Raman spectrum of H4 is subtracted from H4-IC31 (Fig. S5b). Changes in ® IC31 are marginal, confirming results found by FTIR (Fig. S3). The spectral differences observed in Fig. S5a are due to the changes in H4 structure as a result of adjuvantation. The tertiary structure was monitored by nanoDSF, in order to examine the effect of adjuvant adsorption on the H4 (Fig. 6). For the non-adjuvanted H4 protein (Fig. 6, green trace), a thermal transition was detected at 66 ± 1 ◦ C (first derivative in Fig. 6), whereas no distinct thermal transition was observed in the adjuvanted sample (Fig. 6, blue trace). This finding was also supported by theDSC results that showed no distinct transition for H4-IC31 complex (Fig. S6a). A difference of 10 arbitrary units in the 350/330 nm ratio was observed between the two curves is due to increased light scattering in H4-IC31 (Fig. S7c). The scattering also caused decrease in fluorescence intensity in the emission spectrum (Fig. S7a and b). ® This was also demonstrated by titrating IC31 into H4 (Fig. S8). The fluorescence signal was significantly lower in H4-IC31 as compared to H4, and it was blue shifted by 4 nm, but it was still detectable (Fig. S7a and b, Fig. S8a and b). Furthermore the effect of high temperature on H4 and H4-IC31 was verified, in part to support nanoDSF data, by recording FTIR spectra of these samples in native and temperature stressed state (Fig. S9). Stress temperature was selected based on DSC (Fig. S6b) and nanoDSF (Fig. 6) profiles. H4 protein has peaks at 1655 and 1545 cm−1 for ␣-helices in the native state while

Fig. 6. (a) Thermal profile of intrinsic fluorescence emission ratio (350 nm/330 nm) of H4-IC31 (blue trace) and free H4 (green trace). (b) First derivative of thermal profiles from panel a. Traces were smoothed using negative exponential function. The vertical dashed line represents melting temperature (Tm ) of the protein. The Tm of H4 is 66 ± 1 ◦ C while H4-IC31 did not show any thermal transition. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

temperature stressed sample showed more ␤-sheet content as the peak at 1638 cm−1 became apparent. In case of H4-IC31peak for ␤sheet at 1625 shifted to 1626 cm−1 and peak for ␣-helices altered slightly at 1654 cm−1 , while in the amide II region peak position for ␤-sheets did not shift from 1538 cm−1 . These differences are noticeable from second derivative spectra. Hence, H4 showed prominent changes in the secondary structure as compared to H4-IC31. 4. Discussion In order to develop and control vaccine adjuvant formulation processes, it is important to understand how the different components interact. In the case of H4-IC31, the individual components ® are the H4 recombinant fusion protein and the IC31 adjuvant. This study focusses on the characterization of those components alone and in the final vaccine product using PALS, ESZ, CD, nanoDSF, Raman and FTIR spectroscopy. Previously, no biophysical techniques were available to characterize and distinguish H4-IC31 ® from IC31 formulations that did not require a desorption step, due to the low concentration of H4 antigen and the similarity of ® H4-IC31 and IC31 in terms of particle size and morphology. It has ® been reported that a high IC31 to H4 ratio is necessary for optimal vaccine immunomodulation [18], which would further complicate ® the ability of any method to differentiate IC31 from H4-IC31. The ® focus of this study was the biophysical characterization of the IC31 adjuvant and its components, as well as characterization of the H4 antigen protein to understand its conformational changes upon adsorption to the adjuvant. The results from particle size distribution suggest that the overall particle size does not increase dramatically upon formulation ® of IC31 adjuvant with H4; however, the nature of the particles

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model PBA-11999; No. of Pages 9

ARTICLE IN PRESS S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

may be different. The results from PALS are in agreement with ESZ measurements for particle size distribution (Figs. 1 and 2). Zeta potential, also termed electrokinetic potential, is an important property of colloidal suspensions and can be defined as the average electrostatic potential at the interface of the fluid and the ® suspended particle [19]. IC31 is positively charged due to the presence of the KLK peptide, while H4 is predicted to have a net negative charge, and based on the amino acid sequence, the theoretical pI value is 5.19 as per ExPASy portal. Therefore, ionic forces likely ® contribute to the physical association between IC31 and H4, and this in turn may impact immunogenicity of the vaccine formulation. Furthermore, zeta potential is an important indicator of the stability of the colloidal suspension; in general, the higher the zeta potential, the more stable the suspension is [20]. Zeta potential of 30 mV and greater is indicative of stable suspension. Hence, PALS ® data suggest that both IC31 and H4-IC31 are stable suspensions; ® however IC31 is more stable than H4-IC31 as indicated by the higher zeta potential values (Table 1). An additional goal of this study was to probe the structural ® ® changes of H4 and IC31 as a result H4 of adsorption on IC31 , different spectroscopic techniques were used. CD analyses (Fig. S2) revealed an increase in the ␤-sheet content from 37.2 ± 0.21% to 57.3 ± 0.55% while ␣-helical content decreased from 16.3 ± 0.38% to 5.6 ± 0.37% upon adjuvantation. The turns disappeared in the adjuvanted H4 protein from the initial contribution of 17.4 ± 0.49% in the free form. A variation was also observed in random coil content, which increased from 29.1 ± 0.32% in the free H4 protein to 37.1 ± 1.54% in the adjuvanted form, most likely indicative of protein stretch on the surface of the adjuvant as a result of electrostatic ® interactions with and potential structural changes in IC31 . In all of the H4-IC31 samples studied, the adjuvant was at least 9-fold in excess with respect to the antigen [18]. Due to such a low final concentration of H4 protein in the H4-IC31 formulation and the fact that it was in the adsorbed form, it becomes challenging to study this molecule using most biophysical assays. In FTIR spectroscopy, the ground state spectrum does not reveal any apparent change in the spectral features. Therefore, second derivative spectra were calculated to highlight small changes associated with the protein antigen. Although the second derivative spectrum yielded a low signal-to-noise ratio stemming from the marginal differences between H4 and H4-IC31, it still elucidates structural changes of H4 protein antigen as a result of adjuvantation. Moreover, these findings are in agreement with the results from CD and Raman spectroscopy as discussed in this section. In the FTIR spectra the peaks observed at 1689 cm−1 and 1588 cm−1 for H4-IC31 indicate the presence of ␤-sheet structure in adjuvanted H4, and are not observed in free H4 (Fig. 4 and S3b). This suggests that ␤-sheet content was increased upon adjuvantation. A shift of 8 cm−1 was also observed in the random coil region at 1288 cm−1 as a result of adjuvantation. The peak at 1666 cm−1 in the free H4 spectrum, corresponding to the turn region, disappeared in the adjuvanted form (brown trace in Fig. S3b). These findings are consistent with those observed in the CD spectroscopy experiments discussed above. The peaks at 1360 cm−1 and 1338 cm−1 indicate CH2 bending and wagging vibrations of the amino acid side chains. The difference in the peak position would indicate changes to the side chains as H4 ® protein adsorbs to IC31 . Raman spectroscopy results also support the observation of increased ␤-sheet structure as intensity of 1670 cm−1 peak increased in the adsorbed H4 protein (Fig. 5 and S5). As illustrated in the 3D structures of Ag85 B and TB10.4 (Fig. S11, [21,22]), the components of H4, the predominant secondary structure is ␣-helix. The structures of Ag85 B and TB10.4 are prepared from PDB codes 1F0N and 2KG7, respectively. The PDB source suggests 37% ␣-helical and 20% ␤-sheet contents for Ag85 B whereas only 59.8% ␣-helical content for TB10.4. The results also demonstrate high ␣-helical content in H4 protein prior to adjuvan-

®

7

tation. Once adsorbed to IC31 , ␤-sheet becomes the predominant secondary structure in H4. The CD results in this study demonstrated ␤ −sheet and ␣-helical content in the fused Ag85B-TB10.4 (H4) protein, suggesting that Ag85 B imparts more contribution to the CD spectrum. This is further proved in the results of Shi et al. [23], in which the CD spectrum of Ag85B-TB10.4 is similar to the CD spectrum obtained for Ag85B. All of the results discussed above indicate that as a result of adsorption to the adjuvant, H4 undergoes secondary structural changes that primarily involve formation of ␤-sheets. Different proteins are known to undergo a conformational switch under different environmental conditions. Conformational changes from ␣-helices to ␤-sheets and vice versa have been documented in several instances such as in photosystem II reaction center when exposed to bright illumination [24], amyloid plaques from Alzheimer’s disease upon proteolytic cleavage [25,26]. A universal deformation mechanism, proposed earlier for proteins with more than 26 amino acid residues feature ␣-␤ transition under stretching conditions [27], is the most plausible explanation for the ® changes observed in H4 upon adjuvantation to IC31 . The decrease in ␣-helicity of some adsorbed proteins has also been documented. Bovine serum albumin (BSA) and hen’s egg lysozyme both show a decrease in ␣-helical content upon adsorption to silica particles. This transformation can also be dependent on the ratio of coverage of the protein to the surface of the particles. The BSA study showed that increasing the coverage of the surface by the protein diminishes the decrease in ␣-helical content of the adsorbed protein. For hen’s egg lysozyme, it was observed that there is a direct correlation between the increase of structural rearrangements and the increase in charge contrast between the protein and the sorbent [28]. This would mean that once H4 protein adsorbs ® to the surface of IC31 it would experience a similar stretch and adapt to a more favorable conformational state (i.e. conversion of ␣-helices to ␤-sheets). Alterations in secondary structure of the H4 ® protein upon adsorption to IC31 detected by Raman Spectroscopy also resulted in changes to its tertiary structure as shown by nanoDSF (Fig. 6). The absence of a detected thermal transition in the adjuvanted H4 (Fig. 6, blue trace) may indicate that the thermal transition of H4 is beyond the range of the instrument or it lacks ® ordered hydrophobic core. To probe this further, the IC31 concentration was decreased by 12 folds compared to actual H4-IC31 ® product. This decrease of IC31 concentration below 1:1 ratio with H4 resulted in the appearance of free H4 signal in the nanoDSF ® profile (Fig. S10). The Tm of H4 with low IC31 concentration was ◦ 66.5 C, which is deemed the same as of free H4 (Fig. 6). This experiment showed the co-existence of two populations of H4: bound and free, where the former has altered tertiary structure that showed no thermal transition, whereas free form showed thermal transition with lower intensity. The DSC thermogram (Fig. S6a) showed only an exothermic transition for H4-IC31 which is most likely due to a release of heat by the aggregating particles. As a result, ® a potential H4 endothermic transition is masked by IC31 exothermic contribution. The other possibility is that the protein undergoes conformational change upon adjuvantation (for example, due to complete unfolding), but this seems unlikely given that the Raman, CD and FTIR spectroscopy results indicate that H4 retains defined, albeit altered, higher order structure, and the PALS and ESZ results show that H4 contributes to the particle size and distribution. This ® suggests that the protein is stabilized by IC31 , and that this lack of thermal transition is most likely due to an increase in ␤-sheet content of H4 protein. FTIR data for the temperature stressed samples also indicate prominent changes in H4 secondary structure, whereas H4-IC31 showed small changes to the secondary structure and remained largely as ␤-sheet (Fig. S9). As a result of adjuvantation, the H4 protein undergoes a conformational switch to form

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model PBA-11999; No. of Pages 9

ARTICLE IN PRESS S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

8

more ␤-sheets and likely a more compact structure on the sur® face of IC31 , resulting in changes in tertiary structure of H4 in the H4-IC31 formulation that is not detectable within the normal temperature range of the DSF instrument (Fig. 6). Raman spectroscopy data also supports this as hydrogen bonding patterns change upon adjuvantation resulting in the ring deformation noted at 640 cm−1 (Fig. 5 and S5) [29]. This study highlights various biophysical techniques to characterize the TB H4-IC31 adjuvanted vaccine formulation. In addition to the particle size and stability of the adjuvant and antigenadjuvant suspension by PALS, Raman spectroscopy was applied to ® the characterization of the raw material components of the IC31 adjuvant, along with CD, FTIR, nanoDSF, and Raman spectroscopy to study the conformational changes in H4 upon adsorption to the adjuvant. Previously, only immunological methods were applied directly to the H4-IC31 formulated vaccine [6,11,13,18]. PALS analysis, in addition to providing a measurement of particle size, was ® capable in distinguishing between IC31 and H4-IC31, both in terms of size and zeta potential. Zeta potential results further showed that both samples were relatively stable as colloidal sus® pensions, with IC31 being more stable of the two. ESZ confirmed the PALS results for particle size and also detected a size difference ® between IC31 and H4-IC31. The particle size distribution results are also in agreement with the data obtained by Micro Flow Imaging (MFI) (data not shown). Raman spectroscopy, in addition to detecting and identifying the ODN1a and KLK raw material components ® of IC31 , was able to detect conformational changes in H4 upon formulation with the adjuvant. Although Raman Spectroscopy was ® performed on the dried samples of H4, IC31 , and H4-IC31, and may not predict that in solution condition, the results were supported by FTIR and nanoDSF studies. Together with Raman, CD, FTIR, and nanoDSF experiments showed that H4 undergoes changes in secondary structure in the environment of the adjuvant matrix, specifically an increase in ␤-sheet and random coil, together with a decrease in ␣ −helix and turn content. The absence of a detectable thermal tran® sition by nanoDSF in H4 upon adsorption with IC31 adjuvant may reflect an increase in thermal stability.

4.1. Application This work describes an innovative approach whereby multiple biophysical methods were applied to a complex vaccine adjuvant ® formulation. A 9-fold molar ratio of IC31 to H4 is required for H4 TB immune modulation. The ratio of the antigen to adjuvant affects the zeta potential values of H4 and H4-IC31. The zeta potential of IC31 switches from a negative value to a positive value once the molar ratio of adjuvant to antigen exceeds 9 folds [18]. As demonstrated in mice, the adjuvant to antigen ratio is the key to induce strong protection by selectively increasing the number of polyfunctional T cells [30]. It was also shown that the H4-IC31 vaccine candidate is safe and induces a persistent polyfunctional CD4 T-cell response in adults, whereas the response decreases at high dose of H4 [5] due to changes in the adjuvant to antigen ratio. These studies indicate the importance of the adjuvant to antigen ratio to the immunomodulatory activity of the vaccine. Because this ratio is reflected by the surface charge or zeta potential of H4-IC31 formulations, this readout may correlate with vaccine immunogenicity and is thus a potential product quality attribute. FTIR analysis showed differences in the secondary structure content of different formulations of the vaccine. In one formulation in which 20–40% thermal degradation of H4 protein was noted, the FTIR results showed a decrease in ␣-helix from 42% to 32%, and ␤-sheets remained the same. Peak shifts in the Amide I and Amine II regions were observed. In the clinical representative lot, the peak from 1625 cm−1 shifted to 1629 cm−1 and another peak from 1546 cm−1 to 1542 cm−1 in

the 20–40% degraded protein sample (data not shown). Thus, FTIR can be used to verify the integrity of H4 in vaccine formulation and report stability of H4-IC31. All of the developed methods can be performed directly on the formulation without desorption or other harsh prior manipulation of the sample. Such a capability represents an important advantage for vaccine product development, since traditionally vaccine quality has been assessed by means of animal immunogenicity models, which have ethical disadvantages and other limitations, or by immunological readouts which may be technically complex and rely on specialized reagents such as antibodies. Product attributes such as particle size, stability, and protein conformation may have an impact on immunogenicity and thus should be routinely monitored, particularly in early stages of vaccine development when biological and immunological readouts for potency may not yet be available. While biophysical assays such as these may not be subsequently identified as critical quality attributes for the vaccine product (and thus required to be performed for routine product release with defined acceptance criteria), they are nonetheless useful in demonstrating consistency of the manufacturing process. In addition, they can be assessed as stability indicating assays, and applied to comparability assessments to support manufacturing scale-up and other changes, throughout the product life cycle.

5. Conclusion The novel approach presented in this study allows direct testing of the final H4-IC31 drug product formulation without desorption or extensive sample processing. The components and final product of H4-IC31 vaccine were successfully characterized by a panel of biophysical techniques. Changes in secondary and tertiary structure of H4 protein were observed as a result of adjuvantation. The content of ␤-sheet and random coil of H4 protein increased in the adsorbed form. The thermal stability of H4-IC31 was likely enhanced compared to H4 protein alone. The individual methods developed are time efficient, cost effective, and reliable. This general approach should be considered as part of a characterization strategy for protein-based vaccine antigens formulated with novel adjuvants.

Conflict of interest The authors are employees of Sanofi Pasteur Ltd. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. Thus includes employment, consultancies, stock ownership or options, or royalties. No writing assistance was utilized in the production of this manuscript.

Author contributions Sasmit Deshmukh, Federico Webster Magcalas, Kristen Kalbfleisch, Bruce Carpick, and Marina Kirkitadze have contributed towards data analysis and interpretation, critical revisions and final approval of the manuscript. Sasmit Deshmukh and Federico Webster Magcalas designed the study, acquired the data, and drafted the manuscript. Marina Kirkitadze has also contributed to the concept of the study and drafting of the manuscript. Kristen Kalbfleisch investigated the thermal stability of the samples.

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048

G Model

ARTICLE IN PRESS

PBA-11999; No. of Pages 9

S.S. Deshmukh et al. / Journal of Pharmaceutical and Biomedical Analysis xxx (2018) xxx–xxx

Acknowledgements This work was supported by Sanofi Pasteur Ltd. The authors would like to thank Roger Brookes, Emily Xiao, and Jin Su, Sanofi Pasteur Ltd., for generously providing samples for this study, and Jasminder Singh for the FTIR experiments on degraded H4-IC31 samples, and to Jian Hu for DSC experiments on H4 and H4-IC31. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at https://doi.org/10.1016/j.jpba.2018.05.048. References ¨ [1] World Health Organization, “Global Tuberculosis Report,2016. [2] P. Andersen, T. Doherty, The success and failure of BCG − implications for a novel tuberculosis vaccine, Nature Rev. Microbiol. 3 (2005) 656–662. [3] S. Bertholet, G. Ireton, D. Ordway, H. Windish, S. Pine, M. Kahn, T. Phan, I. Orme, T. Vedvick, S. Baldwin, R. Coler, S. Reed, A defined tuberculosis vaccine candidate boosts BCG and protects against multidrug-resistant Mycobacterium tuberculosis, Science Transl. Med. 2 (2010) 53–74. [4] P.L. Lin, J. Dietrich, E. Tan, R.M. Abalos, J. Burgos, C. Bigbee, M. Bigbee, L. Milk, H.P. Gideon, M. Rodgers, C. Cochran, K.M. Guinn, D.R. Sherman, E. Klein, C. Jansen, J.L. Flynn, P. Anderson, The multistage vaccine H56 boosts the effects of BCG to protect cynomolgus macaques against active tuberculosis and reactivation of latent Mycobacterium tuberculosis infection, J. Clin. Invest. 122 (2012) 303–314. [5] H. Geldenhuys, H. Mearns, D. Miles, M. Tameris, D. Hokey, Z. Shi, S. Bennett, P. Andersen, I. Kromann, S. Hoff, W. Hanekom, H. Mahomed, M. Hatherill, T. Scriba, M. von Rooyen, J. McClain, R. Ryall, G. Bryun, H4:IC31 Trial Study Group, The tuberculosis vaccine H4:IC31 is safe and induces a persistent polyfunctional CD4 T cell response in South African adults: a randomized controlled trial, Vaccine 33 (2015) 3592–3599. [6] J. Dietrich, C., Aagaard, P. Andersen, Vaccines comprising TB10.4. United States of America Patent US008557258B2, 15 October 2013. [7] J. Fritz, S. Brunner, M. Birnstiel, M. Buschle, A. Gabain, F. Mattner, W. Zauner, The artificial antimicrobial peptide KLKLLLLLKLK induces predominantly a TH2-type immune response to co-injected antigens, Vaccine 22 (2004) 3274–3284. [8] M. Aichinger, M. Ginzler, J. Weghuber, L. Zimmermann, K. Riedl, G. Schütz, E. Nagy, A. von Gabain, R. Schweyen, T. Henics, Adjuvating the adjuvant: facilitated delivery of an immunomodulatory oligonucleotide to TLR9 by a cationic antimicrobial peptide in dendritic cells, Vaccine 29 (2010) 426–436. [9] K. Lingau, K. Riedl, A. von Gabain, IC31 and IC30, novel types of vaccine adjuvant based on peptide delivery systems, Expert Rev, Vaccines 6 (2007) 741–746. [10] A. Szabo, P. Gogolak, K. Pazmadi, K. Kis-Toth, K. Riedl, B. Wizel, K. Lingnau, A. ® Bacsi, B. Rethi, E. Rajnavolgyi, The two-component adjuvant IC31 boosts type I interferon production of human monocyte-Derived dendritic cells via ligation of endosomal TLRs, PLoS One 8 (2013) e55264. [11] R. Billeskov, T. Elvang, P. Anderson, J. Dietrich, The HyVac4 subunit vaccine efficiently boosts BCG-primed antimycobacterial protective immunity, PLoS One 7 (2012) e39909. [12] S. Deshmukh, K. Bhandal, B. Carpick, M. Kirkitadze, Identification of individual components from the manufacturing chain of a final vaccine product by Raman spectroscopy, Am. Pharm. Rev. 19 (2016) 5, www. americanpharmaceuticalreview.com/1429-AuthorProfile/6011-SasmitDeshmukh/ (Accessed 25 May 2017).

9

[13] J. Dietrich, C. Aagaard, R. Leah, A. Olsen, A. Styhn, T. Doherty, P. Andersen, Exchanging ESAT6 with TB10.4 in an Ag85 B fusion molecule-based tuberculosis subunit vaccine: efficient protection and ESAT6-based sensitive monitoring of vaccine efficacy, J. Immun. 174 (2005) 6332–6339. [14] J. Yang, C. Wu, H. Martinez, Calculation of protein conformation from circular dichroism, Meth. Enzymol. 130 (1986) 208–269. [15] M. Kirkitadze, E. Remi, K. Bhandal, B. Carpick, Visible and subvisible particles ® in the BCG immunotherapeutic product Immucyst , Comput. Struct. Biotech. J. 14 (2016) 154–160. [16] B. Hernández, Y.-M. Coïc, G. Gouyette, M. Ghoumi, Probing the interactions of oligodeoxynucleotides with a cationic peptide by raman scattering application of raman spectroscopy to biology, Netherlands, IOS Press BV 5 (2011) 58–71. [17] G. Guiffo-Soh, B. Hernández, Y.-M. Coïc, F.-Z. Boukhalfa-Heniche, G. Fadda, M. Ghoumi, Vibrational analysis of amino acids and short peptides in hydrated media. 3. successive KL repeats induce highly stable ␤-Strands capable of forming non-H-Bonded aggregates, J. Phys. Chem. B 112 (2008) 1282–1289. [18] S. Aboutorabian, J. Hakimi, F. Boudet, S. Montano, A. Dookie, C. Roque, S.F. ® Ausar, N. Rahman, R.H. Brookes, A high ratio of IC31( ) adjuvant to antigen is necessary for H4 TB vaccine immunomodulation, Hum. Vac. Immunother. 11 (2015) 1449–1455. [19] IUPAC, in: A.D. McNaught, A. Wilkinson (Eds.), Compendium of Chemical Terminology, 2nd ed., Blackwell Scientific Publications, Oxford, 2006, https:// goldbook.iupac.org/. (Accessed 25 May 2017). [20] B. Derjaguin, L. Landau, Theory of the stability of strongly charged lyophobic solution and of the adhesion of strongly charged particles in solutions of electrolytes, Acta. Phys. Chemica USSR 14 (1941) 663. [21] D.H. Anderson, G. Harth, M.A. Horwitz, D. Eisenberg, An interfacial mechanism and a class of inhibitors inferred from two crystal structures of the Mycobacterium tuberculosis 30 kDa major secretory protein (Antigen 85B), a mycolyl transferase, J. Mol. Biol. 307 (2001) 671–681. [22] D. Ilghari, K.L. Lightbody, V. Veverka, L.C. Waters, F.W. Muskett, P.S. Renshaw, M.D. Carr, Solution structure of the Mycobacterium tuberculosis EsxG·EsxH complex: functional implications and comparisons with other M. tuberculosis Esx family complexes, J. Biol. Chem. 286 (2011) 29993–30002. [23] S. Shi, L. Yu, D. Sun, J. Liu, A. Hickey, Rational design of multiple TB antigens TB10.4 and TB10.4-Ag85 B as subunit vaccine candidates against mycobacterium tuberculosis, Pharm. Res. 27 (2009) 224–234. [24] W.-Z. He, W. Newell, P. Harris, R. Chapman, J. Barber, Protein secondary structure of the isolated photosystem II reaction center and conformational changes studies by Fourier-transform infrared spectroscopy, Biochemistry 30 (1991) 4552–4559. [25] P. Fraser, J. Nguyen, W. Surewicz, D. Kirschner, pH-dependent structural transitions of Alzheimer amyloid beta protein fibrillogenesis, Biophys. J. 60 (1991) 1190–1201. [26] M. Kirkitadze, M. Condron, D. Teplow, Identification and characterization of key intermediates in amyloid beta protein fibrillogenesis, J. Mol. Biol. 312 (2001) 1103–1119. [27] Z. Qin, M. Buehler, Molecular dynamics simulation of the alpha-helix to beta-sheet transition in coiled protein filaments: evidence for a critical filament length scale, Phys. Rev. Lett. 104 (2010) 198304. [28] W. Norde, J. Favier, Structure of adsorbed and desorbed proteins, Colloids Surf. 64 (1992) 87–93. [29] Z. Arp, D. Autrey, J. Laane, S.A. Overman, G.J. Thomas Jr., Tyrosine Raman signatures of the filamentous virus Ff are diagnostic of non-hydrogen-bonded phenoxyls: demonstrationby Raman and Infrared spectroscopy of p-cresol vapor, Biochem 40 (2001) 2522–2529. [30] C. Aagaard, Hoang T.T.K.T, A. Izzo, R. Billeskov, J.L. Troudt, K. Arnett, A. Keyser, T. Elvang, P. Andersen, J. Dietrich, Protection and polyfunctional t cells ® induced by Ag85B-TB10.4/IC31 against mycobacterium tuberculosis is highly dependent on the antigen dose, PLoS One 4 (2009) e5930.

Please cite this article in press as: S.S. Deshmukh, et al., Tuberculosis vaccine candidate: Characterization of H4-IC31 formulation and H4 antigen conformation, J. Pharm. Biomed. Anal. (2018), https://doi.org/10.1016/j.jpba.2018.05.048