Alternative Methods of Making Influenza Vaccines

Author: Alan Shaw

The keys to survival of a pandemic will be surveillance, public health planning, and a publicly agreed strategy for behavior.

When talking about the development of vaccines to avert an influenza pandemic, we must keep in mind a sobering fact. A very well run vaccine development program, from early development to finished product launch and including all regulatory filings and product requirements, currently takes about 13 years to complete. A good example is the human papilloma virus (HPV) vaccine. The program started in 1993 and was licensed in the summer of 2006 in the United States. This was a full-throttle, fully funded project that did not run into major problems and did not have to cut any corners. It is important to keep in mind that 13 years or more is the norm.

Before reviewing some of the ways of making an influenza vaccine, it is important to consider the structure of the influenza virus and to review the primary targets for the vaccine. Figure 1(see PDF version for figures) is a schematic drawing of the influenza virus showing the primary target the vaccine industry has focused on for the last 60 years, the hemagglutinin (HA) protein (spikes). HA is the major ingredient assayed in classic egg-based vaccine production, and it is the component upon which a dose is based. The neuraminidase (NA) protein on the virus surface (pegs) is another target. HA and NA denote different subtypes of influenza viruses (e.g., H5N1). Finally, there is the M2 ion channel, which is discussed in detail below.

Current methods of making vaccines for influenza involve growing the virus in a bioreactor, either a fermenter containing VERO, Madin-Darby canine kidney (MDCK) cells (WHO, 2005), PerC.6 cells, or embryonated eggs. Once the virus is grown, it can be inactivated and taken apart; the parts are then used to make the vaccine.

Figure 2 shows a ribbon model of the HA protein, the structural feature of the influenza virus targeted by most influenza vaccines. HA is a trimer of a glycoprotein with a long skinny stalk under a globular head. The transmembrane region at the bottom of the stalk interacts with the matrix protein that holds the virus together.

Recombinant DNA-Based Approaches
HA protein can be synthesized by a variety of recombinant DNA (rDNA)-based approaches: insect cells; plants; yeast; bacteria; chemically synthesized peptides; naked or enhanced DNA; and vectored gene-delivery systems. Most of these approaches are in various stages of development, but it is not clear whether any of them will make it to market.

Every rDNA HA production system has a promoter section, an HA coding sequence, and a terminator section; this rDNA HA protein “cassette” can be inserted into a plasmid or some other type of genetic element and placed inside any number of systems (e.g., brine shrimp, bacteria, yeast, or plants). The biggest variables are the size and nature of the bioreactor and the process for extracting the HA from the producing organism.

Insect Cells
Several companies (Novavax and Protein Sciences are two of the main players) are working on using insect cells as vectors for making HA protein. Insect cell-culture systems grow at relatively low temperatures (~22?C), and incubators must be equipped with both cooling and heating capacity to maintain a constant temperature. The devices in which insect cells are grown are very similar to the equipment used in blood banks. In fact, insect cells can be incubated in large blood bag type containers (wave reactors) that constantly rock back and forth in rocker trays.

One of the most common methods of inserting the rDNA HA into insect cells is via a baculovirus (BV) expression system (Kost et al., 2005). The natural BV virus makes an abundant protein, polyhedrin, and the virus contains a very strong promoter for this polyhedrin gene. The gene of interest, in this case rDNA HA protein, is inserted into the BV insect virus, and the promoter for the viral polyhedrin drives the product synthesis (rDNA HA) in cells infected with recombinant BV.

There are two alternative ways to use this system. One is to make the HA by itself (the Protein-Sciences approach); the other is to co-express HA with NA and the matrix protein that underlies the membrane and make a virus-like particle (the Novavax approach). The virus-like particles apparently have a high propensity for being taken up by the immune system; an effective dose of a virus-like particle vaccine in a mouse model is about one-tenth the dose necessary for HA itself. The potential advantage here is the reduction in the amount of vaccine necessary to elicit an effective immune response.

Work on rDNA protein production in plants has been done at the Boyce Thompson Institute (BTI) for Plant Research, Cornell University, and at Arizona State University. One of BTI’s first forays into vaccine production was a hepatitis B surface antigen expressed in potatoes. Previously established immunity could be boosted after the ingestion of the raw potato with its expressed hepatitis B surface antigen. One of the difficulties with this approach was that the potatoes had to be eaten raw because cooking or heating them denatures the protein. BTI as well as Monash University in Melbourne, Australia, are now working on tomatoes and other plants that are eaten raw and plants from which the vaccine can be extracted. However, because of societal resistance to genetically modified organisms and foods, this type of vaccine production may meet with opposition.

Yeast is one of the workhorses of the vaccine industry. Vaccines currently made in yeast include the recombinant proteins hepatitis B surface antigen and the newly approved HPV vaccine. Yeast has the capability of secreting and assembling large proteins, which are then glycosylated, but not exactly the way human cells operate. GlycoFi Inc. reported early this year that they have engineered the human enzymology for gylcosylation side chain addition into yeast, so a more human-looking sugar additive is produced as proteins are formed (Li et al., 2006). Yeast should be a good system for making HA.

Avant Immunotherapeutics Inc. is making innovative bacterial-vectored vaccines from bacteria that normally invade the lining of the human gut (e.g., salmonella, shigella, cholera). Because the genetics of these bacteria are now understood, their metabolic processes can be used to advantage. First, knock-out mutations are made in various metabolic pathways so the bacteria cannot survive outside of an artificial environment. After ingestion, they travel to the gut, invade the mucosa, and stop growing because they are missing a critical element for growth.

The plasmids in the bacteria have been genetically engineered to express the protein-coding gene fragment that has been loaded into them (i.e., the HA antigen). The plasmid is a double-stranded, typically circular DNA molecule, not part of the cell nucleus with its chromosomal DNA, but capable of autonomous DNA replication. Plasmids in genetic engineering are called vectors. Once released into the mucosa, the protein is picked up and processed by the immune system of the person, elicits a response, and establishes immunity. Immunogenicity of the HA surface antigen is currently being evaluated in mouse and rabbit models.

Chemically Synthesized Peptides
Another attractive possibility for vaccine production is fabrication of peptides-reductive immunology taken to its end state. The process for making selected peptide-containing antigenic epitopes by chemical synthesis is outlined in Figure 3). Figure 3a shows a solid support resin to which amino acids are added one by one by synthetic chemistry in a controlled sequence. In the most preferred application for vaccines, an array is fabricated with the appropriate density of these epitopes to elicit an immune response. For example, Figure 3b shows how multi-armed lysine trees can be made and the peptides of choice attached in multiple copies in a very dense array.

Many companies are working with synthetic chemistry peptide technology. The challenge is to pick the right conserved epitopes out of the 600 amino acid sequence of the HA protein.

Naked or Enhanced DNA
The principle of using naked DNA is to select a plasmid vector into which the gene of interest can be inserted with the right controlling sequences around it. The plasmid is then propagated in E. coli, which can be engineered to make profuse amounts of the DNA containing the gene of interest. The plasmids are then purified and injected directly into muscle tissue or are coated onto gold beads that can be injected under the skin with a gene gun. The DNA is taken up by the human host cell, carried to the nucleus, and expressed, thus allowing the host to make the active antigen.

The limitation so far in this approach is the inefficiency of DNA transfer. The human body contains many nucleases and other enzymes that destroy “free” DNA, simply because DNA is not supposed to be free. Thus, a large dose of free DNA (about 5 mg) is necessary in humans because most of it is broken down before it reaches the nucleus. Although very little DNA makes it to the nucleus, the DNA that does get there is highly effective in expressing protein and eliciting an immune response. A variety of vehicles for enhancing the uptake of DNA have been applied to this technology, largely by Vical of San Diego, the originator of this technology.

Adjuvants are agents or drugs that have few or no antigenic effects or properties but may increase the efficacy or potency of vaccines or other drugs when given at the same time. Some of the most common adjuvants are aluminum salts and monophosphoryl lipid (MPL)/QS21 cocktails (being developed by Glaxo Smith Kline). Adjuvants and enhancements make it possible for larger fragments of naked DNA to reach the intended target. Although poorly understood scientifically, the successful development of adjuvants is important because they can greatly increase the efficacy of existing vaccine supplies. This technology is currently being used for vaccines against malaria and HIV, as well as for influenza.

Virus-Vectored Gene Delivery
Adenoviruses, which infect both humans and animals, can be rendered nonpathogenic by taking out part of their genomes, so that the virus cannot replicate on its own unless the host cell provides the missing portion of the genome (usually the E1 genes); this vector is then taken up by the host cells (Figure 4). Adenoviruses have a propensity for infecting dendritic cells, which are the antigen-presenting cells that can carry antigens to the lymph nodes. This is advantageous because proteins expressed by an adenovirus are fairly immunogenic; in fact, the leading candidate for an HIV vaccine is a viral-vectored system.

Most humans have antibodies against adenoviruses; in fact the virus was first isolated from the human adenoids (tonsils), from which the name is derived. Thus, the effectiveness of an adenovirus-vector system could be limited because of pre-existing immunity to the vector. Nevertheless, these systems are being used in the development of vaccines.

Alpha Viruses
The alphavirus Venezuelan Equine Encephalitis virus, which is normally pathogenic for horses, has an unusual property. Its genome is expressed in two parts (i.e., separate systems for expressing early genes and late genes). The RNA for the late genes can be commandeered to express the protein of interest; Alphavax is the leader in alphavirus-vector technology (Schultz-Cherry et al., 2000). Alphaviruses are not normally pathogens for humans, so immunity against most alphaviruses does not exist in the human population.

Alternative Delivery of Antigens
Vaccines can be delivered in ways other than by direct intramuscular (IM) injection. Any method can be used to introduce the vaccine directly to dendritic cells (the cells that carry antigen to the lymph nodes) located in the intradermal compartment of the skin, whence they will be effectively taken into the immune system. Thus, vaccines can be delivered through the skin using patches, micro-abrasion, or micro-tines coated with antigen. They can also be delivered through the intranasal system; Flu-Mist? administers a live, attenuated vaccine by spray into the intranasal compartment to create local mucosal immunity.

Alternative Influenza Antigens
The influenza virus (Figure 1) has ten gene products: HA and NA on the surface; the M2 ion channel and a variety of intraviral proteins; the M1 matrix proteins; and various RNA packaging proteins that make up the viron. HA and NA, which are on the surface of the virus, are available for attack by the immune system, but they are subject to immune pressures and antigenic drift (mutations) in their genetic material over time.

Antigens other than HA might be good candidates for making a flu vaccine. The neuraminidase (NA) protein is one possible candidate, although it is susceptible to antigenic drift and immune pressure, so it may not have any advantages over HA other than as another surface target. NA is not a strong neutralizing antigen; rather it is an enzyme that cleaves progeny virus away from the cellular debris after an infection is complete, so that the progeny can then carry out the next round of infection. Blocking this enzyme activity with antibody may confer some benefit. Other possible conserved antigens are the NS1 protein (a nonstructural nuclear protein), M1 (the matrix protein), NP, and M2 (which is the ion channel).

Innate and Adaptive Immunity
Analysis of the immune system to identify its smallest critical element has stripped it of some of its power. We have spent decades identifying epitopes, purifying the essential antigens, and focusing on the details of antigen processing and presentation. In the late 1990s it was discovered that another arm of the immune system, the innate immune system, has a limited repertoire of receptors, which are called “toll”-like receptors (TLRs). (Toll was a gene found in the fruit fly [Drosophila] that involves body-part segmentation and orientation.)

As nature has a propensity for recycling motifs, the toll gene has been duplicated dozens of times and has mutated and been adapted by mammalian cells to serve as antennae on the outside of antigen-presenting cells. TLR antennae specifically recognize unique elements of pathogens that are not part of mammalian biology; they actually provide the first line of defense, an inflammatory response. TLRs also control the initiation of adaptive immune response, which is the T-cell and B-cell response that immunologists have been working with for 60 years.

The adaptive immune system has a random, highly dispersed repertoire of antigen receptors that reshuffle and adapt over time to antigens through a series of clonal selections and expansions. There are limitations to immunological memory, however, and the system cannot distinguish between what is dangerous and what is not without the support of the very effective innate immune system.

Figure 5 shows a series of pathogen-associated molecular patterns (PAMPs), which are elements of pathogens belonging to bacteria or viruses (e.g., porins, lipopolysaccharides, lipoproteins, peptidoglycans, and flagella) that are not found in mammalian cells. Because these are all elements or pathogenic components specific to bacteria or viruses, they are recognized by TLRs as being foreign. TLRs also recognize the DNA sequence CpG (cytosine and guanine separated by a phosphate) as being foreign.

In classical antigen processing and presentation (Figure 6, that is, without TLRs, extracellular antigens are taken into the cell in an endocytic vesicle, where protease containing lysosomes break up the antigen into peptides by partial protein degradation. The major histocompatibility complex (MHC) class II presentation system comes in through the Golgi apparatus, a small organelle inside the cell, picks up these peptides, and carries them back to the surface of the cell for presentation. The MHC class II presentation system allows cells to tell surrounding cells if they are healthy or infected.
Figure 7 shows the antigen-presenting system with TLRs present. The antigen now is coupled to a pathogen pattern, or is part of a pathogen, and is picked up by a TLR and carried into an endocytic vesicle. The TLR then sends signals to the nucleus to create a battery of cytokines that stimulate the neighboring T and B cells. These signals also cause the antigen-presenting cell to express special structures (CD80/86) that interact with T cells to drive an immune response. These PAMPS are recognized as components of bacteria or viruses and, in the presence of a TLR, send signals to the nucleus of the dendritic cell to trigger the synthesis of cytokines, which activate the adaptive immune system.

The M2 Ion Channel Protein
In the context of the influenza virus, the M2 ion channel protein is particularly important. There are only about 20 copies of M2 ion channel on the surface of the virus, many fewer than of HA or NA. The ion channel is important in virus replication, because during the uncoating of the virus a pH change must take place before it can start its infection. The M2 ion channel is actually a hole that protons go through during the pH change.

VaxInnate has explored two methods of making the influenza M2e vaccine: (1) hooking four copies of the 24 amino acid M2 protein to flagellin produced in E. coli; and (2) hooking a copy of M2e to Pam3-Cys (e.g., tri-palmitoyl-cysteine, a synthetic adjuvant linked to M2e, the antigenic protein) to make Pam3Cys.M2e. The flagellin fusion is thought to be the better method, because the M2e 24 amino acid peptide is almost invisible to the immune system and arouses almost no immune response. Only when the antigen is coupled with a PAMP does something meaningful happen in terms of immunity.

Human Influenza Multivalent Vaccine Strategy
An advantage of driving the immune response with the TLR coupling system is that the M2e peptide appears to be relatively conserved and is a relatively stable target that should not change year after year. If we look at historical data for the 24 amino acid sequences in H1, H2, and H3 influenza subtypes over time, it appears that the protein is indeed relatively conserved and relatively stable. In the VaxInnate consensus vaccine, there is a very nice cross-reactivity with H1, H2, and H3 human subtypes.

Knowing that the H5, H7, and H9 strains currently show a number of mutations in the M2e 24 amino acid sequence in avian subtypes, three separate consequence vaccines would be required to develop a multivalent (quadravalent) vaccine with four different versions of the 24 amino acid peptide. This could confer immunity for H1, H2, and H3, as well as H5, H7, and H9 strains. Once developed, the vaccine could be made in advance and stockpiled for use in pandemic situations. In animal models this seems to work well.

Prospects for the Immediate Future
Refinements in egg-based systems and cell-culture systems look like promising second-generation product advancements. Just using a new cell-culture system will not be enough, but it would get us away from eggs, which would have some advantages in terms of speed and yield. But the process would still be very slow. The approaches for rDNA-based vaccines and new models for M2e vaccines I’ve just described may not be available in time to counter a pandemic. Thus, the key to survival in the next few years will be good surveillance, good public health planning, and a publicly agreed strategy for how to behave when a pandemic appears.

ARCHES (Academic Resources for Computing and Higher Education Services). 2000. Flora of the Skin, Nose, and Throat. Images taken from Brock Biology of Microorganisms, 9th ed, by Madigan/Martinko/Parker. Available online at:
Dubensky, T.W. Jr., J.M. Polo, and M.A. Liu. 1998. Live virus vaccines: something old, something new, something borrowed... Nature Medicine 4(12): 1357–1358. doi:10.1038/3939.
Gilbert, S.F. 2006. A Companion to Developmental Biology, 7th ed. Available online at:
Isin, B., P. Doruker, and I. Bahar. 2002. Functional motions of influenza virus hemagglutinin: a structure-based analytical approach. Biophysical Journal 82(2): 569–581.
Kost, T.A., J.P. Condreay, and D.L. Jarvis. 2005. Baculovirus as versatile vectors for protein expression in insect and mammalian cells. Nature Biotechnology 23(5): 567–575.
Li, H., N. Sethuraman, T.A. Stadheim, D. Zha, B. Prinz, N. Ballew, P. Bobrowicz, B.K. Choi, W.J. Cook, M. Cukan, N.R. Houston-Cummings, R. Davidson, B. Gong, S.R. Hamilton, J.P. Hoopes, Y. Jiang, N. Kim, R. Mansfield, J.H. Nett, S. Rios, R. Strawbridge, S. Wildt, and T.U. Gerngross. 2006. Optimization of humanized IgGs in glycoengineered Pichia pastoris. Nature Biotechnology 24(2): 210–215. Epub 2006 Jan 22.
Medzhitov, R. 2001. Toll-like receptors and innate immunity. Nature Reviews. Immunology 1: 135–145. Available online at:
NLM (United States National Library of Medicine). 2006. Genetics Home Reference Handbook for Gene Therapy: Table of Illustrations. Available online at:
Schultz-Cherry, S., J.K. Dybing, N.L. Davis, C. Williamson, D.L. Suarez, R. Johnson, and M.L. Perdue. 2000. Influenza virus (A/HK/156/97) hemagglutinin expressed by an alphavirus replicon system protects chickens against a lethal infection with Hong Kong-origin H5N1 viruses. Virology 278(1): 55–59.
SciELO (Scientific Electronic Library Online). 1997–2005. SciELO Brazil. Available online at:
Smith, T. 2003. The Creation of Relenza for the Treatment of Influenza. Hosted by Available online at:
UIUC (University of Illinois at Urbana-Champaign). 2004. Roy J. Carver Biotechnology Center at UIUC. Solid-phase Peptide Synthesis (SPPS). Last edited November 2004. Available online at:
WHO (World Health Organization). 2005. WHO Guidance on Development of Influenza Vaccine Reference Viruses by Reverse Genetics. Available online at:

About the Author: Alan Shaw is president and CEO of VaxInnate.