Most viral diseases display a variable clinical outcome due to differences in virus strain virulence and/or individual host susceptibility to infection. Understanding the biological mechanisms differentiating a viral infection displaying severe clinical manifestations from its milder forms can provide the intellectual framework toward therapies and early prognostic markers. This is especially true in arbovirus infections, where most clinical cases are present as mild febrile illness. Here, we used a naturally occurring vector-borne viral disease of ruminants, bluetongue, as an experimental system to uncover the fundamental mechanisms of virus-host interactions resulting in distinct clinical outcomes. As with most viral diseases, clinical symptoms in bluetongue can vary dramatically. We reproduced experimentally distinct clinical forms of bluetongue infection in sheep using three bluetongue virus (BTV) strains (BTV-1IT2006, BTV-1IT2013 and BTV-8FRA2017). Infected animals displayed clinical signs varying from clinically unapparent, to mild and severe disease. We collected and integrated clinical, haematological, virological, and histopathological data resulting in the analyses of 332 individual parameters from each infected and uninfected control animal. We subsequently used machine learning to identify the key viral and host processes associated with disease pathogenesis. We identified five different fundamental processes affecting the severity of bluetongue: (i) virus load and replication in target organs, (ii) modulation of the host type-I IFN response, (iii) pro-inflammatory responses, (iv) vascular damage, and (v) immunosuppression. Overall, our study using an agnostic machine learning approach, can be used to prioritise the different pathogenetic mechanisms affecting the disease outcome of an arbovirus infection.### Competing Interest StatementThe authors have declared no competing interest.

Most viral diseases display a variable clinical outcome due to differences in virus strain virulence and/or individual host susceptibility to infection. Understanding the biological mechanisms differentiating a viral infection displaying severe clinical manifestations from its milder forms can provide the intellectual framework toward therapies and early prognostic markers. This is especially true in arbovirus infections, where most clinical cases are present as mild febrile illness. Here, we used a naturally occurring vector-borne viral disease of ruminants, bluetongue, as an experimental system to uncover the fundamental mechanisms of virus-host interactions resulting in distinct clinical outcomes. As with most viral diseases, clinical symptoms in bluetongue can vary dramatically. We reproduced experimentally distinct clinical forms of bluetongue infection in sheep using three bluetongue virus (BTV) strains (BTV-1IT2006, BTV-1IT2013 and BTV-8FRA2017). Infected animals displayed clinical signs varying from clinically unapparent, to mild and severe disease. We collected and integrated clinical, haematological, virological, and histopathological data resulting in the analyses of 332 individual parameters from each infected and uninfected control animal. We subsequently used machine learning to select the key viral and host processes associated with disease pathogenesis. We identified and experimentally validated five different fundamental processes affecting the severity of bluetongue: (i) virus load and replication in target organs, (ii) modulation of the host type-I IFN response, (iii) pro-inflammatory responses, (iv) vascular damage, and (v) immunosuppression. Overall, we showed that an agnostic machine learning approach can be used to prioritise the different pathogenetic mechanisms affecting the disease outcome of an arbovirus infection.

The host immune response usually aims to inhibit virus replication in order to avoid cell damage and disease. In some cases, however, the infected host avoids the deleterious effects of infection despite high levels of viral replication.

The pandemic spread of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the etiological agent of Coronavirus Disease 2019 (COVID-19), represents an ongoing international health crisis. A key symptom of SARS-CoV-2 infection is the onset of fever, with a hyperthermic temperature range of 38 to 41°C. Fever is an evolutionarily conserved host response to microbial infection that can influence the outcome of viral pathogenicity and regulation of host innate and adaptive immune responses. However, it remains to be determined what effect elevated temperature has on SARS-CoV-2 replication. Utilizing a three-dimensional (3D) air-liquid interface (ALI) model that closely mimics the natural tissue physiology of SARS-CoV-2 infection in the respiratory airway, we identify tissue temperature to play an important role in the regulation of SARS-CoV-2 infection. Respiratory tissue incubated at 40°C remained permissive to SARS-CoV-2 entry but refractory to viral transcription, leading to significantly reduced levels of viral RNA replication and apical shedding of infectious virus. We identify tissue temperature to play an important role in the differential regulation of epithelial host responses to SARS-CoV-2 infection that impact upon multiple pathways, including intracellular immune regulation, without disruption to general transcription or epithelium integrity. We present the first evidence that febrile temperatures associated with COVID-19 inhibit SARS-CoV-2 replication in respiratory epithelia. Our data identify an important role for tissue temperature in the epithelial restriction of SARS-CoV-2 independently of canonical interferon (IFN)-mediated antiviral immune defenses.

Background SARS-CoV-2 infection, the causative agent of COVID-19, has resulted in over 2,500,000 deaths to date1. Although vaccines are becoming available, treatment options remain limited. Repurposing of compounds could reduce the time, cost, and risks associated with the development of new drugs and has been the focus of many clinical studies. Here, we summarise available evidence on 29 FDA-approved compounds, from in vitro results to clinical trials, focussing on remdesivir, galidesivir and favipiravir, and test 29 antiviral compounds’ activity in vitro. Methods A comprehensive search strategy was used to retrieve trials and publications related to antiviral compounds with potential efficacy to treat coronaviruses. These data were used to prioritise testing of a panel of antiviral drugs in vitro against patient isolates of SARS-CoV-2. An in vitro screen was carried out to determine the activity of 29 FDA-approved compounds. Results 625 clinical trials investigated 16 repurposed antiviral candidate compounds for the treatment of COVID-19. In vitro studies identified ten drug candidates with demonstrable anti-SARS-CoV-2 activity, including favipiravir, remdesivir, and galidesivir. To validate these findings, a drug screen was conducted using two cell lines and wildtype isolates of SARS-CoV-2 isolated from patients in the UK. While eight drugs with anti-SARS-CoV-2 activity were identified in vitro, activity in clinical trials has, as yet failed to demonstrate a strong effect on mortality. Conclusions So far, no repurposed antiviral has shown a strong effect on mortality in clinical studies. The urgent need for novel antivirals in this pandemic is clear, despite the costs and time associated with their development. Research in Context Evidence before this study Repurposing of existing compounds for the treatment of COVID-19 has been the focus of many in vitro studies and clinical trials, saving time, costs and risks associated with the research and development of new compounds. Added value of this study We reviewed the literature for 29 FDA-approved compounds with previously reported (or suspected) anti-SARS-CoV-2 activity and found 625 clinical trials that have been undertaken on 16 different drugs. We determined if repurposed antivirals are suitable for clinical trials based on previously published data, and conducted an additional in vitro screen using locally circulating strains in the UK (PHE2 and GLA1). We report the difference in IC50 from published data using Wuhan1/Wash1 strains with PHE2 and GLA1, including IC50 values below 100μM for galidesivir in wild-type virus. Given the limited success of repurposed compounds in the treatment of COVID-19, we comment on the urgent need for new antivirals specifically targeting SARS-CoV-2. Implications of all the available evidence Our data show that most prospective compounds for repurposing show no anti-SARS-CoV-2 activity, and antiviral activity in vitro does not always translate to clinical benefit. So far, no repurposed compound has shown a strong effect on mortality in clinical studies. Drugs, including monoclonal antibody therapies, that have been developed to target SARS-CoV-2 virus itself have shown most promise.

The recent emergence of Severe Acute Respiratory Syndrome Coronavirus 2 (SARS-CoV-2), the underlying cause of Coronavirus Disease 2019 (COVID-19), has led to a worldwide pandemic causing substantial morbidity, mortality, and economic devastation. In response, many laboratories have redirected attention to SARS-CoV-2, meaning there is an urgent need for tools that can be used in laboratories unaccustomed to working with coronaviruses. Here we report a range of tools for SARS-CoV-2 research. First, we describe a facile single plasmid SARS-CoV-2 reverse genetics system that is simple to genetically manipulate and can be used to rescue infectious virus through transient transfection (without in vitro transcription or additional expression plasmids). The rescue system is accompanied by our panel of SARS-CoV-2 antibodies (against nearly every viral protein), SARS-CoV-2 clinical isolates, and SARS-CoV-2 permissive cell lines, which are all openly available to the scientific community. Using these tools, we demonstrate here that the controversial ORF10 protein is expressed in infected cells. Furthermore, we show that the promising repurposed antiviral activity of apilimod is dependent on TMPRSS2 expression. Altogether, our SARS-CoV-2 toolkit, which can be directly accessed via our website at https://mrcppu-covid.bio/, constitutes a resource with considerable potential to advance COVID-19 vaccine design, drug testing, and discovery science.

Remdesivir (RDV), a broadly acting nucleoside analogue, is the only FDA approved small molecule antiviral for the treatment of COVID-19 patients. To date, there are no reports identifying SARS-CoV-2 RDV resistance in patients, animal models or in vitro. Here, we selected drug-resistant viral populations by serially passaging SARS-CoV-2 in vitro in the presence of RDV. Using high throughput sequencing, we identified a single mutation in RNA-dependent RNA polymerase (NSP12) at a residue conserved among all coronaviruses in two independently evolved populations displaying decreased RDV sensitivity. Introduction of the NSP12 E802D mutation into our SARS-CoV-2 reverse genetics backbone confirmed its role in decreasing RDV sensitivity in vitro. Substitution of E802 did not affect viral replication or activity of an alternate nucleoside analogue (EIDD2801) but did affect virus fitness in a competition assay. Analysis of the globally circulating SARS-CoV-2 variants (>800,000 sequences) showed no evidence of widespread transmission of RDV-resistant mutants. Surprisingly, we observed an excess of substitutions in Spike at corresponding sites identified in the emerging SARS-CoV-2 variants of concern (i.e., H69, E484, N501, H655) indicating that they can arise in vitro in the absence of immune selection. The identification and characterisation of a drug resistant signature within the SARS-CoV-2 genome has implications for clinical management and virus surveillance.

Virus-virus interactions influence the epidemiology of respiratory infections. However, the impact of viruses causing upper respiratory infections on severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) replication and transmission is currently unknown. Human rhinoviruses cause the common cold and are the most prevalent respiratory viruses of humans. Interactions between rhinoviruses and cocirculating respiratory viruses have been shown to shape virus epidemiology at the individual host and population level. Here, we examined the replication kinetics of SARS-CoV-2 in the human respiratory epithelium in the presence or absence of rhinovirus. We show that human rhinovirus triggers an interferon response that blocks SARS-CoV-2 replication. Mathematical simulations show that this virus-virus interaction is likely to have a population-wide effect as an increasing prevalence of rhinovirus will reduce the number of new coronavirus disease 2019 cases.

Remdesivir (RDV) is used widely for COVID-19 patients despite varying results in recent clinical trials. Here, we show how serially passaging SARS-CoV-2 in vitro in the presence of RDV selected for drug-resistant viral populations. We determined that the E802D mutation in the RNA-dependent RNA polymerase was sufficient to confer decreased RDV sensitivity without affecting viral fitness. Analysis of more than 200,000 sequences of globally circulating SARS-CoV-2 variants show no evidence of widespread transmission of RDV-resistant mutants. Surprisingly, we also observed changes in the Spike (i.e., H69 E484, N501, H655) corresponding to mutations identified in emerging SARS-CoV-2 variants indicating that they can arise in vitro in the absence of immune selection. This study illustrates SARS-CoV-2 genome plasticity and offers new perspectives on surveillance of viral variants. One Sentence Summary SARS-CoV-2 drug resistance & genome plasticity

The pandemic spread of SARS-CoV-2, the etiological agent of COVID-19, represents a significant and ongoing international health crisis. A key symptom of SARS-CoV-2 infection is the onset of fever, with a hyperthermic temperature range of 38 to 41°C. Fever is an evolutionarily conserved host response to microbial infection and inflammation that can influence the outcome of viral pathogenicity and regulation of host innate and adaptive immune responses. However, it remains to be determined what effect elevated temperature has on SARS-CoV-2 tropism and replication. Utilizing a 3D air-liquid interface (ALI) model that closely mimics the natural tissue physiology and cellular tropism of SARS-CoV-2 infection in the respiratory airway, we identify tissue temperature to play an important role in the regulation of SARS-CoV-2 infection. We show that temperature elevation induces wide-spread transcriptome changes that impact upon the regulation of multiple pathways, including epigenetic regulation and lncRNA expression, without disruption of general cellular transcription or the induction of interferon (IFN)-mediated antiviral immune defences. Respiratory tissue incubated at temperatures >37°C remained permissive to SARS-CoV-2 infection but severely restricted the initiation of viral transcription, leading to significantly reduced levels of intraepithelial viral RNA accumulation and apical shedding of infectious virus. To our knowledge, we present the first evidence that febrile temperatures associated with COVID-19 inhibit SARS-CoV-2 replication. Our data identify an important role for temperature elevation in the epithelial restriction of SARS-CoV-2 that occurs independently of the induction of canonical IFN-mediated antiviral immune defences and interferon-stimulated gene (ISG) expression.

Bluetongue is a vector-borne disease of ruminants caused by bluetongue virus (BTV). BTV can infect essentially all domestic and wild ruminants but the clinical outcome of infection differs substantially between host species. Clinical disease induced by BTV, including haemorrhagic fever in severe cases, is normally evident only in sheep. Conversely, cattle are more resilient to BTV infection, as they develop high levels of viremia and can be reservoirs of infection, but rarely show clinical signs. Here, we concentrated on BTV-host cell interactions using primary cells as an experimental system. First, we determined that BTV reaches higher titres in ovine cells, compared to bovine cells although it induces comparable levels of antiviral cytokines in both cell types. Importantly, these differences are abolished by inhibiting the Jak/Stat pathway. In addition, pre-treatment with interferon (IFN) severely hampers BTV replication in bovine, but not in ovine, primary cells. These data suggest that bovine, unlike ovine, IFN-stimulated genes (ISGs) are effective in controlling BTV replication. Using a high-throughput flow cytometry approach, we screened an expression library of over 300 bovine ISGs to identify genes with antiviral properties against BTV. We have identified ∼10 bovine ISGs that negatively impact BTV replication (by at least 50%). Currently, we are assessing the sheep orthologues to the bovine ISGs of interest in order to investigate host-species differences. Our study provides novel insights on how bovine cells restrict BTV replication and could provide an intellectual framework to understand the host determinants involved in disease severity.

African horse sickness virus (AHSV) is the causative agent of African horse sickness (AHS), a highly fatal disease of equids. Currently, live attenuated vaccines are used to control AHS especially in South Africa, but they are in general subject to restrictions. We took advantage of previously published AHSV structural and antigenic data to simplify AHSV vaccine development using an AHSV-4 reverse genetics approach. We systematically substituted the tip and central domains of AHSV-4 outer capsid protein VP2 with the corresponding region from other serotypes to generate a chimeric protein. AHSV S2 chimeric and mono-reassortant viruses for all 9 serotypes were recovered by reverse genetic using AHSV-4 as backbone and used to immunise guinea pigs. Presence of neutralising antibody titres were determined using a fluorescence-based neutralisation assay. The exchange of the tip and central domain of VP2 switched the serotype specificity of the rescued chimeric viruses, however, sera from AHSV-4VP2DTip only neutralised the homologous AHSV-4 reference virus. Mono-reassortants, but not recombinant viruses containing chimeric VP2, induced antibodies with low levels of cross-neutralisation between phylogenetically related serotypes. Interestingly, most of the sera raised were able to neutralise AHSV-4, indicating the presence of other neutralising epitopes within the virus. These results raise the possibly of generating a single virus that affords protection against multiple serotypes. Our research highlights the ability to manipulate the AHSV genome to rapidly generate ‘synthetic’ viruses using a single platform approach for vaccine development.

African horse sickness is a major infectious disease of equids and is caused by African horse sickness virus (AHSV), a dsRNA virus with 10 genome segments encoding for 7 structural and 4/5 non-structural proteins. Here, we focused on the characterisation of the AHSV NS4, the latest protein found to be expressed by this virus. In silico analysis of available sequences confirmed the existence of two phylogenetically distinct AHSV clades: NS4-I and NS4-II. NS4-II is further divided into three subtypes (a, b and g). Confocal microscopy demonstrated that all AHSV NS4 types localised in the cytoplasm of infected cells, unlike the BTV NS4 which, has a strong nucleolar localisation. The replication kinetics of reverse genetics derived AHSV NS4 deletion mutants (AHSVDNS4) were similar to their wild type counterparts in insect (Kc) or interferon incompetent (BSR) cells. However, replication of AHSVDNS4 mutants in primary horse endothelial cells was restricted, in comparison to wild-type viruses. Importantly, primary cells restriction to AHSV replication was dependent on the JAK/STAT pathway. Furthermore, AHSVDNS4 mutants were not able to efficiently suppress the secretion of anti-viral cytokines from primary cells, while the wild-type viruses suppressed this response to varying degrees. Importantly, AHSVDNS4 mutants were less virulent than their wild type counterparts in a murine model of AHSV infection. These results indicate that AHSV NS4 has a role in interferon IFN antagonism and a determinant of viral virulence. We are currently carrying out mass spectrometry analyses to identify the cellular proteins interacting with the AHSV NS4.

Amount and diversity of bird-dispersed seed rain play important roles in determining forest composition, yet neither is easy to quantify. The complex ecological processes that influence seed movement make the best approach -highly context specific. Although recent advances in seed rain theory emphasize quantifying source-specific seed shadows, many ecological questions can be -addressed u sing a less mechanistic approach that requires fewer assumptions. Using seed rain rates from 0.38 m(2) hoop traps sampled twice monthly over the course of a year, we show that number of traps required to identify changes in seed rain varies across seed species and forest type. Detecting a 50% increase in amount of seed rain required from 65 to > 300 traps, while detecting a 200% increase generally required <= 50 traps. Trap size and ecological context dictate the number of seeds found in each trap, but the coefficient of variation (CV) across traps in a given ecological context can help inform future studies about number of traps needed to detect change. To better understand factors influencing variation around estimates of seed rain, we simulated both clustered and evenly distributed patterns of fecal deposition using three different levels of seed aggregation (number of seeds in each fecal deposit). When patterns of fecal -deposition were clustered, rather than evenly dispersed across the study area, they required > 1.5 times the number of traps to identify a 100% increase in seed rain. Similarly, we found that low seed aggregation required > 1.5 times the number of traps to detect a 100% change than when aggregation was medium or high. At low aggregations, fewer seed rain traps contained seeds (low, 33 +/- 5%; medium, 23 +/- 4%; high, 24 +/- 5%), resulting in more variation across traps than medium and high aggregations. We also illustrate the importance of training -observers to discern between morphologically similar seeds from different species and provide resources to help identify bird-dispersed seeds commonly found within midelevation mesic Hawaiian forests.

Bluetongue virus (BTV) is endemic in many parts of the world, often causing severe hemorrhagic disease in livestock. To date, at least 27 different serotypes have been recognized. Vaccination against all serotypes is necessary to protect susceptible animals and to prevent onward spread of the virus by insect vectors. In our previous studies, we generated replication-deficient (disabled infectious single-cycle [ DISC]) virus strains for a number of serotypes and reported preliminary data on their protective efficacy in animals. In this report, to advance the DISC vaccines to the marketplace, we investigated different parameters of these DISC vaccines. First, we demonstrated the genetic stabilities of these vaccine strains and also the complementing cell line. Subsequently, the optimal storage conditions of vaccines, including additives, temperature, and desiccation, were determined and their protective efficacies in animals confirmed. Furthermore, to test if mixtures of different vaccine strains could be tolerated, we tested cocktails of DISC vaccines in combinations of three or six different serotypes in sheep and cattle, the two natural hosts of BTV. Groups of sheep vaccinated with a cocktail of six different vaccines were completely protected from challenge with individual virulent serotypes, both in early challenge and after 5 months of challenge without any clinical disease. There was no interference in protection between the different vaccines. Protection was also achieved in cattle with a mixture of three vaccine strains, albeit at a lesser level than in sheep. Our data support and validate the suitability of these virus strains as the next-generation vaccines for BTV.IMPORTANCE Bluetongue (BT) is a debilitating and in many cases lethal disease that affects ruminants of economic importance. Classical vaccines that afford protection against bluetongue virus, the etiological agent, are not free from secondary and undesirable effects. A surge in new approaches to produce highly attenuated, safer vaccines was evident after the development of the BTV reverse-genetics system that allows the introduction of targeted mutations in the virus genome. We targeted an essential gene to develop disabled virus strains as vaccine candidates. The results presented in this report further substantiate our previous evidence and support the suitability of these virus strains as the next-generation BTV vaccines.

In the late 1990s, there were rumblings that Bluetongue virus (BTV) was on the move. The 2006 summer outbreak changed the way that the European economic and scientific communities viewed its importance. It shifted from being a neglected disease confined to the tropical regions of the world to a potentially important threat to agriculture. Suddenly, BTV was sharing research priorities and the limelight with other important viruses of animals such as foot-and-mouth disease virus and avian influenza virus.

Viruses have often evolved overlapping reading frames in order to maximize their coding capacity. Until recently, the segmented dsRNA genome of viruses of the Orbivirus genus was thought to be monocistronic, but the identification of the bluetongue virus (BTV) NS4 protein changed this assumption. A small ORF in segment 10, overlapping the NS3 ORF in the +1 position, is maintained in more than 300 strains of the 27 different BTV serotypes and in more than 200 strains of the phylogenetically related African horse sickness virus (AHSV). In BTV, this ORF (named S10-ORF2 in this study) encodes a putative protein 50-59 residues in length and appears to be under strong positive selection. HA- or GFP-tagged versions of S10-ORF2 expressed from transfected plasmids localized within the nucleoli of transfected cells, unless a putative nucleolar localization signal was mutated. S10-ORF2 inhibited gene expression, but not RNA translation, in transient transfection reporter assays. In both mammalian and insect cells, BTV S10-ORF2 deletion mutants (BTV8ΔS10-ORF2) displayed similar replication kinetics to wt virus. In vivo, S10-ORF2 deletion mutants were pathogenic in mouse models of disease. Although further evidence is required for S10-ORF2 expression during infection, the data presented provide an initial characterization of this ORF.

Bluetongue virus (BTV) encodes a single capping protein, VP4, which catalyzes all reactions required to generate cap1 structures on nascent viral transcripts. Further, structural analysis by X‐ray crystallography indicated each catalytic reaction is arranged as a discrete domain, including a nucleoside‐2′‐O‐methyltransferase (2′‐O MTase). In this study, we have exploited the structural information to identify the residues that are important for the catalytic activity of 2′‐O MTase of VP4 and their influence on BTV replication. The effect of these mutations on GMP binding, guanylyltransferase (GTase) and methylase activities were analysed by a series ofin vitro biochemical assays using recombinant mutant proteins; subsequently their effects on virus replication were assessed by introducing the same mutations in replicating viral genome using a reverse genetics system. Our data showed that single substitution mutations in the catalytic tetrad K‐D‐K‐E were sufficient to abolish 2′‐O MTase activityin vitro and to completely abrogate BTV replication in cells; although these mutants retained the upstream GMP binding, GTase and guanine‐N7‐methyltransferase activities. Mutations of the surrounding substrate‐binding pocket (predicted to recruit cap0) had variable effects onin vitro VP4 capping activity. Only triple but not single substitution mutations of these residues in genome resulted in reduced virus replication kinetics. This is the first report investigating the importance of 2′‐O MTase function for any member of theReoviridae and highlights the significance of K‐D‐K‐E tetrad and surrounding residues for the efficiency of 2′‐O MTase activity and in turn, for virus fitness.

Baculovirus expression system offers the advantage of expression of several large proteins simultaneously by a single recombinant virus. To date, expression of multiple large (>100 kDa) proteins has been hampered by the need to generate large constructs and repeat use of homologous sequence and promoter. The development of multi-loci baculovirus expression system overcomes these issues by enabling the recombination of large foreign sequences into different regions of the genome. In this paper, we have examined the co-expression of African horse sickness virus (AHSV) VP2 proteins from multiple serotypes in a single recombinant baculovirus. To this end, recombinant baculoviruses expressing multiple AHSV VP2 proteins were generated and it was found that up to six different AHSV serotypes (serotype 1, 3, 4, 5, 7 and 8) VP2 proteins (∼120 kDa) could be expressed simultaneously from different loci of baculovirus genome. The expression of VP2 of one serotype was not significantly hindered by the presence of other serotypes, although there were slight differences in expression level between different serotypes. The expression of VP2 of further serotypes from additional loci resulted in a lesser expression level of VP2 proteins. Based on these findings, three additional recombinant baculoviruses encompassing all nine AHSV serotypes were constructed (serotypes 1, 7, 8 or serotypes 2, 4, 5 or serotypes 3, 6, 9) and each of the triple recombinant viruses exhibited similar expression level of each VP2. This system allows for the expression of a number of large proteins that has the potential to be exploited for multivalent vaccines production.