Taken together, we hope the methods and data reported here will serve to expedite the much-needed research required to address this unprecedented pandemic.
We would like to thank Natasha K. All authors have read and agreed to the published version of the manuscript. This research was funded by NIH grants to C.
National Center for Biotechnology Information , U. Journal List Viruses v. Published online Jun 6. Alexander S. Author information Article notes Copyright and License information Disclaimer. Received May 13; Accepted Jun 4. This article has been cited by other articles in PMC. Materials and Methods 2. Electron Microscopy Resuspended purified beta-propiolactone treated SARS-CoV-2 virus preps were adsorbed onto mesh formvar-carbon coated nickel grids for 10 min, washed with 0.
Results 3. Open in a separate window. Figure 1. Figure 2. Figure 3. TRIzol TRIzol is a well-known and widely used reagent for the isolation from cells of nucleic acids and in some cases protein. Beta-Propiolactone Beta-propiolactone BPL is a commonly used reagent for the inactivation of viruses for use in vaccine preparations [ 11 , 12 , 13 , 14 ] and it has recently been used in the development of an inactivated SARS-CoV-2 vaccine preparation [ 15 ].
Figure 4. Figure 5. Discussion The ability to grow and accurately quantify infectious virus is critical for virological studies. Acknowledgments We would like to thank Natasha K. Author Contributions A. Conflicts of Interest The authors declare no conflict of interest.
References 1. Bedford J. World Health Organization. Blow J. Virus inactivation by nucleic acid extraction reagents. Darnell M. Haddock E. Effective Chemical Inactivation of Ebola Virus. Rabenau H. Stability and inactivation of SARS coronavirus. Lei S. Gas chromatography-mass spectrometry method for determination of beta-propiolactone in human inactivated rabies vaccine and its hydrolysis analysis.
Lane 3. Left panel presents purified live influenza virus fraction 3. Right panel, red solid curve, presents results of fraction 3. Clearly two populations are present indicating the splitting of the virus was effective. After sterile filtration of this fraction 5. The size of the different intermediate products was measured by DLS; results are presented in Table 6.
The radius of the particles was measured before and after splitting if applicable and before and after SF. The results of the BPL inactivated fractions are given in the columns to the right, while the results of the formaldehyde inactivated fractions are presented in the columns to the left.
DLS analysis of whole virus fractions 3. The mean size before and after splitting did not significantly change, however the PDI did. The increase of size distribution and the two populations in the DLS graphs Fig 4 , right panel indicate that the splitting of the virus with detergent did change its morphology. By performing 0. Due to the fact that large particles contribute more strongly to the signal, the removal of large particles seemingly results in a shift of the size to particles smaller than nm.
However, these small particles are also present in the material before filtration but are not detected. The shift may also be due to disintegration of large particles caused by shear force during sterile filtration. EM pictures of the H3N2 influenza virus, before splitting Fig 5 , top panels show the presence of the spike proteins HA and NA and the particulate nature of the virus.
The EM pictures of the inactivated and split H3N2 influenza Fig 5 , bottom panels clearly show the partially disrupted particular and more heterologous structures, supporting the increase of PDI mentioned above. Enlargement pictures to the left Top row whole virus Fig 1 , fraction 3. Pictures at top: HA and NA spikes are clearly visible on the outside of the particles. The pictures at the bottom show disrupted, heterologous structures.
The left graph presents the six vaccine bulks over a period of 5 months. The right graph presents data from influenza vaccine batches prepared at Intravacc inactivation with BPL and splitting with Triton : stability over a period of twelve months for four WIV products and one Triton split product. The product stabilities are in the same range as in the left panel except for product 5. Lowest stability was found for the ether split, formaldehyde inactivated product 5. These batches are prepared similarly to product 5.
The stability data of an investigational batch, BPL inactivated Triton split product, similar to at Cantacuzino produced 5. The stability study of trivalent subunit influenza vaccine batches by Coenen e. In this study six downstream processes for influenza vaccine manufacturing were executed Fig 1 and compared.
Mostly equipment for industrial scale was used to facilitate later technical transfer of a revised DSP manufacturing process. Two WIV bulks and four split influenza vaccine bulks were manufactured. The bulks were analyzed using a panel of assays Table 4 , as defined by existing specifications for influenza vaccine product release. Additional testing of intermediate products was performed to support better control over product and downstream process.
For the release of influenza vaccine the prime focus is on the presence of antigenic HA, as determined with SRID, and the absence of specific contaminants, such as ovalbumin. For example Cox e.
The effect of detergents or organic solvents used for the disruption of influenza viruses depends on the quality of the input virus. Product properties such as the presence of detergent resistant membrane structures [ 30 ] and the presence of one ribonucleoprotein structure in either small spheroid or large filamentous virus [ 31 ] will affect the quality of the intermediate vaccine product.
The used agent for inactivation can as well influence product quality [ 32 ] because formaldehyde performs cross-linking between molecular structures [ 33 ] and BPL causes acylation and alkylation of molecules [ 34 ]. It is surprising to observe an apparent greater decease in M1, the protein that connects with all viral components and the membrane [ 30 , 35 ], while HA and NP are retained.
No specific explanation for this can be given yet, though detergents will not affect the HA and NA associated in lipid rafts but do interfere with other binding sites given the observed effects of detergent on split vaccine product morphology [ 36 ]. M1 has a strong hydrophobic core and it may form dimers when in solution. In addition M1 dimers may stack up to a ribbon, all with their positively charged area on the same side of the ribbon [ 37 ]. If this causes larges entities, these are removed during the downstream process, especially during sterile filtration.
With respect to the presence of NP which was identified in pandemic vaccine formulations related to narcolepsy, as published recently by Vaarala [ 35 ], the manufacturing process using formaldehyde inactivation and Triton splitting may possibly result in a safer product.
The main reason is the finding that the HA concentration was too low. This can be explained by the fact that the virus and the split virus particles are relative large entities compared to the pore size of the membrane used for SF. After SF still particles are present with size of nm radius of nm and larger, possibly indicating an ongoing association process or equilibrium.
Using these unit operations the protein content decreased even more than the antigenic HA content except for process of ether split, formaldehyde inactivation, where the order of the unit operations is different. Based on these limited HA recovery data, on average processes with BPL inactivation gave higher recoveries. Commonly it is hypothesized that the presence of a certain amount of detergent improves the product stability; Triton X is intentionally added [ 38 ] and present in most of the commercial products.
The majority of the products investigated here are exceptionally stable, despite the almost complete detergent removal. Recovery and composition of the bulk product produced at Cantacuzino Institute Romania during process transfer, with the inactivation using BPL and splitting using Triton, is compared with the average of 6 batches produced in The Netherlands at Intravacc for clinical studies. In Table 7 the main result of the comparison is presented.
SDS-PAGE results of bulks presented in Fig 2 confirm the principal protein composition resemblance of products produced at either location. Therefore as proof-of-concept the process transfer can be considered successful for this one batch.
The product prepared in Romania is more concentrated products produced in The Netherlands were diluted during SF, while Romania product was not , contained more residual ovalbumin and had higher HA recovery.
The product prepared at Intravacc has a lower residual ovalbumin content; eggs are selected for similar size and the allantoic fluid is harvested in a very precise mode, preventing contamination with e.
All influenza vaccine bulk products produced at Cantacuzino met the preset quality criteria WHO, EP , with respect to sterility, presence of NA and maximum ovalbumin content.
This study shows the initial feasibility of process transfer of different influenza DSP processes, as a potential first step towards implementation of a revised manufacturing process, using additional product characterization to support better control over product and process.
Largest losses occurred during the sterile filtration unit operation. The shape of the virus longer than wide may contribute to partial blockage of the sterile filter membrane.
On average BPL inactivated virus product show higher recovery than the formaldehyde treated products. The overall recoveries of the Triton split products, whether BPL inactivated or formaldehyde inactivated, seem to be similar.
WIV products and formaldehyde inactivated Triton split product show better stability in this study based on limited data. This investigation confirmed the influence of choices made for the downstream process on the final product quality, recovery and stability. Lanes numbered above left gel: 1. In the present study, we examined the biological and biochemical characteristics of human H1N1 and H3N2 viruses treated with BPL, and developed an inactivation method for BPL-sensitive viruses.
A significant decrease in HA titer was detected in the H3N2 viruses examined. The decrease in the pH of the virus fluid was not associated with the decreased HA titer, indicating that the decrease in HA titer for the H3N2 virus is the result of the direct effect of BPL. This resulted in six processes: two commonly performed split processes, two hybrid split processes and two WIV processes, as shown in the overview in Fig 1.
In the boxes the unit operations are presented. Fraction identification number is written below the unit operation box. Fraction 1. Six different influenza vaccine batches of bulk vaccine product were produced starting from one batch of clarified allantoic fluid. The main characteristics of the performed production processes are summarized in Table 1 , whereas the respective accompanying process flowcharts are presented in Fig 1.
Good Manufacturing Practices GMP compliant facilities of Cantacuzino were used to produce the influenza vaccine batches. The allantoic fluid was harvested and clarified by centrifugation. The clarified harvest Fig 1 , fraction 1. The two phases were separated by centrifugation CS 50 Centrifugal extractor, CINC, Germany and ether in the top phase was removed by pumping, while the removal of ether was completed by subsequent evaporation Fig 1 , fractions 3.
Fractions 3. Detergents were removed from the fractions Fig 1 , fraction 3. Removal was monitored by UV until UV-absorption did not decrease further. The release tests for influenza vaccine for human use, as specified by the World Health Organization WHO and European Pharmacopeia EP were performed on the bulk products and on several intermediate product fractions. Table 2 is listing the assays including the requirements for the vaccine bulk product. More detail on the test methods is available in S1 File.
Particle size distribution by intensity more relevant for bigger particles and by mass more relevant for smaller particles were evaluated, as well as the polydispersity index PDI which is a measure for size distribution; a value below 0. The HA preservation is expressed as percentage of the HA concentration in the bulk immediately after production.
Based on standard tests the six vaccine bulk products were analyzed. The concentration of total protein and ovalbumin in the starting material, the clarified harvest before ZUC fraction 1. The haemagglutinin HA concentration was not measured below detection limit of the test , The results for HA, total protein and ovalbumin in the fraction after ZUC in phosphate buffer and after ZUC in citrate buffer are given in Table 3 as well. For total protein and HA results are similar; ovalbumin concentration in fraction 2.
In the 5. Haemagglutin HA and total protein concentration are similar; ovalbumin concentration after ZUC in citrate buffer is lower than after ZUC in phosphate buffer.
If applicable the requirements are listed. All products comply with the requirements, except product 5. As shown in Table 4 , almost all products meet the target specification for HA content. The formaldehyde inactivated, ether split product 5. The 5. In addition, because the total protein concentration of the 5. In contrast to the ether split and subsequently formalin inactivated 5. Comparably, in their investigations on influenza split vaccine products prepared by ether splitting and formaldehyde inactivation, Johannsen e.
In our here described study, the DLS results did not indicate significant more aggregation in the 5. The apparent low recovery of the formaldehyde ether split product may be related to other specific chemical or physical changes of the HA protein. From Table 4 , row 6, it can be deduced that all produced bulks comply with this requirement; the NA concentrations in the final fractions Fig 1 , 5. As shown in Table 4 , row 7, the ovalbumin concentration in the produced bulks was in the range of 2.
However: after purification by ZUC Fig 1 , fractions 2. The DLS measurements Table 4 , row 15 of the bulk products showed that whole inactivated virus products contain larger particles than the split products. Whereas WIV vaccines 5. In the intermediate product fractions collected from the process steps before ZUC, no reliable HA concentration could be measured since the concentrations were below the detection limit of the SRID test. The recoveries based on HA quantity are therefore calculated relative to the amount present Table 3 in the fractions after ZUC Fig 1 , fractions 2.
As a consequence the ratio total protein to HA-protein concentration does not change much after this unit operation, in other words the recovery of total protein is indicative for the recovery of HA-protein.
Lanes M were loaded with marker proteins, with the corresponding molecular weight presented to the left. The fraction sample identity Fig 1 is noted above the lane. Left gel: 1. After de-glycosylation the HA bands are more distinct and migration distance has increased right gel, bulks 5. NP and M1 protein bands have not changed position due to the applied de-glycosylation. In the lanes to the right of the right gel, for comparison products prepared at Intravacc site were applied: 5.
The recoveries per unit operation based on antigenic HA, and based on total protein are presented in Table 5. Since the unit operations after zonal centrifugation 2. This is indeed the case except for ether split virus, were a relative large loss of antigenicity was measured after sterile filtration fraction 5. The opposite was observed for Triton split virus: HA purity increased after sterile filtration fraction 5.
This difference in protein composition is confirmed with gel electrophoresis Fig 2 , showing that 5. The most pronounced differences in the recoveries occur after splitting and after sterile filtration SF. In order to identify discrepancies we have used a conservative procedure applied to the log-ratios of the proportional decrease in HA to the proportional decrease in protein [ S2 File ]. This procedure indicates that the discrepancy found for product 5.
The method used controls the false positive rate FDR by means of the Benjamini-Hochberg procedure [ 25 ] in order to prevent the detection of spurious discrepancies. On the other hand, because the method is somewhat conservative, some less obvious discrepancies may have escaped us using this method.
For example the BPL inactivated products seem more similar in HA and total protein recovery over SF than the formaldehyde inactivated products. On average the HA recovery of the three BPL inactivated products is higher than for the three formaldehyde inactivated vaccine products.
In Fig 2 the results of SDS-PAGE under reducing conditions without de-glycosylation left gel and with de-glycosylation right gel of the protein samples are presented. As expected the de-glycosylated HA protein bands are more distinct and at increased migration distance right gel compared to glycosylated HA bands in the left gel.
0コメント