Vaccine Research Lines Converge

Vaccinologists working on different diseases met in Germany to exchange ideas—and found that their approaches already have a great deal in common

By Philip Cohen, PhD

Experts gathered at the Vaccine Congress 2005 in Berlin this September to compare notes on their progress battling many diseases around the globe. The meeting was titled New Approaches to Vaccine Development: From the bench to the field (www.vaccine-berlin2005.org). Over three days, speakers talked about challenges facing vaccine research, development, and delivery, from basic science to clinical trials, as well as regulatory and safety issues. The infectious agents under the spotlight ranged from relatively new emergent threats such as the SARS-associated coronavirus, to the ongoing battle against major killers like HIV, tuberculosis, and malaria, to the endgame health officials are engaged in to eradicate poliovirus outbreaks from the planet.

Within those talks, the scientific tales told were impressive not only for the cutting edge biology on display but also for the extent to which different areas of research are converging, says Congress President Stefan Kaufmann of the Max Planck Institute for Infection Biology in Berlin. Vectors first developed for one disease are now being eyed for multi-disease inoculations, for example, and interest in new adjuvants is soaring across the board. While such cross fertilization is not unusual in scientific research, Kaufmann thinks a sense of urgency has also driven researchers to look for new approaches in related and unrelated diseases. "The clock of infections is ticking, and we don't want to lose any more time," he says.

TB or not TB

About 3 billion doses of the live-attenuated vaccine for tuberculosis (TB), Mycobacterium bovis bacilli Calmette-Guerin (BCG), have been dispensed and have played a major role in protecting children from severe infections with M. tuberculosis. Even so this vaccine, first developed more than 80 years ago, needs improvement. BCG contains rather than eradicates M. tuberculosis infection, so doesn't prevent the most prevalent form of TB infection in adults, pulmonary tuberculosis; in some populations it appears to have little or no effect, which appears especially true in developing country settings.

The urgency for improved TB vaccines is increasing with the emergence of M. tuberculosis strains resistant to multiple drugs. And more and more people are becoming co-infected with M. tuberculosis and HIV making treatment further complicated since antiretrovirals can trigger TB outbreaks and BCG can cause disease in immunocompromised people. "HIV and TB are a dangerous liaison," says Kaufmann. But the reasons for poor protection afforded by BCG in some settings aren't yet clear. Failure may be related to the local mycobacterial strains and their effect on immunity, or interaction with BCG. It's also possible that genetic differences between BCG and M. tuberculosis or differences in their immunogenicity limit how effective vaccination can be.

Kaufmann began his efforts to improve BCG with an analysis of what limits its immunogenicity. When this mycobacterium invades antigen-presenting cells, like macrophages or dendritic cells (DCs), it takes up residence in vacuoles within cells called phagosomes. Bacterial protein antigens from this compartment are presented in the context of MHC II, which primarily stimulates CD4+ T cells. As a result BCG antigens rarely enter the MHC I pathway, eliciting few CD8+ T cells, which are known to be major contributors to TB immunity.

It was clear that the road to greater immunity lay outside of the phagosome. So Kaufmann's team set to build BCG antigens an escape vehicle. They genetically engineered recombinant BCG (rBCG) to carry a protein from Listeria monocytogenes called listeriolysin, which forms protein pores in phagosome membranes. Since this protein operates optimally at acidic pH, the investigators also knocked out the gene for BCG urease C, an enzyme that normally buffers the phagosomal environment to near neutrality, to create their DureC hly+ rBCG strain.

Mice vaccinated with this new rBCG strain showed superior protection when challenged by M. tuberculosis. Most strikingly, animals vaccinated with rBCG and then challenged with the multi-drug resistant clinical M. tuberculosisisolate Beijing/W had a 100-fold reduction of bacterial load in their lungs compared to mice vaccinated with the parental BCG, which showed no significant protection above control vaccinations against this pathogen. The improved strain also had an increased safety profile; immunodeficient SCID mice survived 80 days after inoculation with 10colony forming units of DureC hly+ rBCG while the immunocompromised rodents receiving similar doses of BCG died in less than 25 days.

The rBCG vaccine does elicit CD8+ T cells, including some with antigen-specific cytotoxicity. It turns out, however, that this improved immunogenicity is only partly due to the intended listeriolysin-aided leakage of rBCG antigens to the cytoplasm. Kaufmann's team found that the microbe also triggers profound levels of apoptosis when it infects mouse or human macrophages. In vivo, apoptotic cells would release vesicles with bacterial antigens which could be presented in the context of MHC I and MHC II after they are engulfed by neighboring DCs. How DureC hly+ rBCG triggers apoptosis is unclear, although Kaufmann suspects it may be initiated by the listeriolysin-aided release of cellular proteases from the phagosome. Such proteases are known to activate apoptotic signal cascades. The researchers have patented their rBCG strain and licensed it to Vakzine Projekt Management GmbH, which is preparing to start clinical trials at the end of this year.

Boosting BCG immunity by creating a leaky phagosome appears to be in vogue. In a later talk Jerry Sadoff of Aeras Global TB Vaccine Foundation reported that his team has developed a similar strategy using the pore-forming protein perfringolysin O from Clostridium perfringens. This protein has the advantage of operating at near neutral pH and thus not requiring the genetic background of a urease C mutation for optimal function. Sadoff's team is now in the process of generating immunogenicity data.

William Jacobs of the Albert Einstein College of Medicine in New York has attacked the problem of improving TB vaccination from the opposite direction. Rather than building up BCG, his team is stripping down wild-type pathogenic M. tuberculosis. In the early 20th century, BCG was attenuated by more than 200 serial passages in bile salts of a virulent bacterial strain resulting in the loss or rearrangement of more than 100 genes. Today, with the entire genome of M. tuberculosis sequenced and the ability to knock out genes at will, Jacobs is making progress toward reducing this microbe's virulence while preserving its immunogenicity by creating strains with designer combinations of mutations.

One approach Jacobs described is combining mutations in different biosynthetic pathways, resulting in slow growth in vivo. A promising mutant dubbed mc26020 contains deletions in genes for lysine and pantothenate synthesis and shows dramatic attenuation. When interferon-g (IFN-g) knockout (GKO) mice, which are very susceptible to TB infection, were inoculated with 105 colony forming units of this mutant all the animals survived after 400 days of observation. In contrast, mean survival times for BCG-infected GKO mice were less than 100 days.

When used as a vaccine, immunological protection provided by mc26020 was similar to BCG in a mouse model of TB challenge. But Jacobs is also working to improve immunogenicity of his strain. His team identified a gene called NlaA (Nurim-like antiapoptotic factor), which when knocked out in M. tuberculosis strains causes more apoptosis in infected cells and therefore presumably better antigen presentation.

Along with their use as TB vaccines, the rBCG and mc26020 strains may be valuable as platforms for the delivery of other antigens. Jacobs mentioned that he is collaborating with Barton Haynes (see An Interview with Barton Haynes), team leader for the recently awarded CHAVI grant from the US National Institutes of Health (NIH), to explore using mc26020 as an AIDS vaccine. The collaborators are currently planning to test immunogenicity in mice, guinea pigs, cattle, and non-human primates.

Jaap Goudsmit of the Netherlands-based company Crucell also argued that rBCG strains would be valuable vaccine vectors for other diseases. His company collaborates with Aeras (which provides rBCG strains), IAVI, and the NIH to develop vaccines against TB, HIV, and malaria using its proprietary adenovirus vaccine vector serotype 35(Ad35). "An rBCG vaccine followed by an Ad35 boost is the strategy of choice for combined pediatric vaccine against HIV, malaria, and TB," he says. Goudsmit thinks there are a number of arguments in favor of childhood vaccination against HIV: it could lower mother to child transmission of HIV, guarantee that individuals would be vaccinated before they became sexually active, and deal with the issue of pre-existing immunity against vectors. "We eventually will need to combine vaccinations or we'll run into problems since researchers for these different diseases are all using the same vectors," he says.

Bacterial balancing acts

The challenge in these attempts to supercharge bacteria as vaccine vectors is to strike the right balance between a hardy vector that stimulates immunity and one that poses no danger for its human host. "Ideally you want wild-type ability to withstand stress, wild-type ability to invade cells before the vector displays attenuation," says Roy Curtiss III of the Biodesign Institute at Arizona State University in Tempe. Curtiss reported on a clever genetic strategy to maximize the efficiency of Salmonella spp. vectors by placing virulence factors or other essential genes under control of an arabinose-inducible promoter. As a result, the bacteria can be grown in arabinose media with wild-type stress resistance and invasive powers, which wane after a few generations of growth in vivo. He showed one example of this regulated attenutation strategy in which the phosphomannose isomerase gene, which adds sugars to the protective surface LPS-O-antigen, was placed in this cassette. Growth in vivo causes the shut down of the isomerase gene in the arabinose cassette. As a result, new sugars are not added to the bacterial LPS-O-antigen and the Salmonella strain becomes highly attenuated after seven generations of growth. "We got high immunogenicity but some mice die," says Curtiss. To quicken the attenuation, Curtiss is looking to add more genes to the regulated cassette. One attractive set of candidates are genes that drive synthesis of the bacterial wall component diaminopimelic acid (DAP). Loss of DAP results in bacterial lysis, attenuating the strain and also increasing antigen delivery to the cell. Curtiss recently received a $14.8 million grant from the Bill & Melinda Gates Foundation's Grand Challenges in Global Health Initiative to develop a new pneumonia vaccine for newborns based on this technology.

Alexander Schmidt at the University of Muenster, Germany and his colleagues have avoided the attenuation question altogether by using a microbial vector that is not only safe but beneficial in some people. The probiotic Eschericia coli Nissle 1917 strain has long been used to treat inflammatory bowel disease and recently has received attention as a vector to deliver therapeutic molecules (FEMS Immunol. Med. Microbiol. 43, 373, 2005). Oddly, the Nissle 1917 field isolate carries a number of genetic markers associated with pathogenic strains and is known to have adjuvant properties. "To a microbiologist and the immune system at first glance it may look like a bad guy, but behaves very well indeed," says Schmidt. "I've taken it myself. Besides tasting like E. coli it has no ill effects whatsoever."

To this strain Schmidt has added a recombinant version of the E. coli gene for AIDA-1 (Adhesin Involved in Diffuse Adherence), a protein transporter which delivers an adhesin to the surface of some strains. His team substituted the adhesin N-terminus of the protein for antigens from Shiga toxin of enterohemorrhagic E. coli (EHEC) and the OspA protein from the lyme disease bacteria Borrelia burgdorferi. After 8 to 9 days of oral immunization, the researchers were able to measure specific mucosal and systemic antibody responses against these antigens despite poor bacterial colonization of the murine intestine. His team is now working to develop disease models in the mouse to test for protection.

Adjuvant extravaganza

Another hot topic at the meeting was designing adjuvants and other strategies to give more kick to vaccines. Hermann Wagner at the Technical University of Munich described the use of microspheres to co-deliver adjuvant and antigen. Wagner's team is hoping to extend the success his and other labs have had using ligands for Toll-like receptors (TLRs) chemically coupled to antigens to boost immune responses (see Toll bridge to immunity, andResearch Briefs). "The problem is that the chemistry for these complexes must be optimized for each antigen, which isn't ideal," says Wagner.

So instead, Wagner's team took a solution containing the TLR9 ligand CpG and a model protein, ovalbumin (OVA), and trapped it inside polylactide microspheres. They found these microspheres are readily ingested by DCs and make their way to endosomes where TLR9 resides. In a proof-of-concept study, the spheres were injected subcutaneously into mice and the animals developed OVA-specific CD8+ T cell responses, good antibody responses and were protected against infection by a L. monocytogenes strain expressing OVA.

Annie Mo of Antigenics presented her company's strategy of inducing cellular immunity by complexing antigenic peptides with the heat shock protein HSP70. HSP70 contains a peptide-binding pocket and can act as a chaperone for antigenic peptides. It can also bind the CD91 cell surface receptor, facilitating peptide uptake by APCs through endocytosis. The peptides carried by HSP70 are then processed and presented in the context of MHC molecules. Antigenics has already developed a number of cancer immunotherapies now in clinical trials which involve purifying HSP70-peptide complexes from a patient's own tumors. But the ability of in vitroreconstituted HSP70-peptide complexes to induce strong immune responses to viral antigens has been demonstrated in animal models for antigens from influenza virus and HIV. Mo described her company's first HSP-based polyvalent vaccine program on an infectious disease, a therapeutic vaccine for herpes simplex virus type 2 (HSV-2).

The HSV-2 vaccine is composed of a species-matched recombinant HSP70 and synthetic peptides derived from HSV-2 proteins. To optimize the system, they used a 35-mer peptide from HSV-2 glycoprotein B that contains a known CD8+ T cell epitope. Preliminary experiments showed that peptide concentrations as low as 0.02 nanomolar complexed with HSP70 elicited a measurable antigen-specific T cell response. And the specific response to the peptide improved more than 3-fold when it was just one of a peptide pool. "That was good news," says Mo. "It suggested a polyvalent peptide pool would work." Indeed, in a mouse model a polyvalent HSV vaccine delivered by HSP70 increases the frequency of IFN-g secreting T cells 4-fold over peptide alone. After the meeting, Antigenics announced they have started a Phase I trial of a recombinant human HSP complexed with 32 synthetic peptides representing HSV-2 proteins.

A specific class of adjuvants gaining in interest are those that elicit mucosal immunity. Mucosal sites are a major point of entry for pathogens and recently have become a focus of AIDS research with the discovery that much of the crucial immunological damage wrought by HIV occurs in gut mucosa within the first two weeks of infection. One of the best studied and most potent mucosal adjuvants is cholera toxin (CT), but this molecule is also highly toxic making it impractical as a component of human vaccines.

Jan Holmgren at Goteberg University in Sweden is interested in developing non-toxic alternatives to CT. One approach he discussed was to isolate the B subunit of CT (CTB), a non-toxic molecule which, while less powerful than CT, can be used to promote mucosal immunity. Holmgren tried to further boost its effectiveness by linking recombinant CTB to CpG oligonucleotide that mimics bacterial DNA and engages TLR9. They found that in mice treated intravaginally the CTB-CpG conjugate elicited more than 10-fold stronger production of MIP-1a, MIP1-b and RANTES in the vaginal mucosa over each component separately or the two mixed together. There was also very strong expression of these chemokines by human lymphocytes exposed to the CTB-CpG conjugate. Their analysis shows the effect is dependent on TLR9, suggesting CTB helps to deliver CpG to this receptor in endosomes by binding its ligand, the GM1 receptor.

Holmgren, Cecil Czerkinsky's team of INSERM in France, and their colleagues also presented a poster on linking CTB to a model antigen, OVA, for use as a vaginal immunogen. In a mouse model, CTB-OVA applied vaginally activated antigen-specific IFN-g secreting CD4+ and CD8+ T cells and elicited antigen-specific cytotoxic T lymphocytes. OVA or CTB delivered alone or mixed together failed to trigger a comparable immune response.

The site of delivery of mucosal vaccines is known to have a powerful effect in many mammals including mice, rats, macaques and humans. In general, the strongest IgA production is at the site of administration and then weakens in proportion to distance. Nasal application is the exception to this rule since it generates a strong immune response at distal sites including the vaginal tract. Holmgren and Czerkinsky reported on another promising delivery route: under the tongue. Using CT mixed with OVA, they found that sublingual administration in mice triggered immune changes throughout the body. There were significant IgA and IgG antibody responses both in serum and in mucosal tissues including the lungs and genital tract. The CT/OVA mixture also prompted the release of a mixture of cytokines mediating cell and antibody immunity and generated cytotoxic T cell responses peripherally and in the lungs. But for this robust response the use of the whole cholera toxin was important. When less toxic CTB was linked to OVA and delivered sublingually it induced immunological tolerance.

These researchers see the sublingual route of mucosal vaccination as promising. Compared to ingested antigen there is little degradation so much less antigen is needed. Compared to nasal administration there is less risk of neural toxicity. And as opposed to injection it is less invasive; no device is required to deliver the antigen and there's no issue of needle phobia.

Carlos Guzman of GBF-German Research Center for Biotechnology talked about new derivatives of another mucosal adjuvant, macrophage-activating lipopeptide of 2 kilodaltons (MALP-2). MALP-2 is a TLR ligand that binds the heterodimer TLR2/6 on macrophages. MALP-2 also promotes the activation and maturation of DCs and his group has recently demonstrated that it can directly stimulate B cells.

Guzman reported on a structure-function analysis aimed at dissecting the role of the fatty acids and peptide components of this molecule, as well as the impact of its stereochemistry. The studies revealed that the chiral center and fatty acid acylation were critical for MALP-2 function as an adjuvant. The peptide was a different story—when Guzman substituted polyethyleneglycol (PEG) for the peptide, the resulting molecule was as good an adjuvant as MALP-2 by a number of immunological criteria, but was more stable and soluble and could be manufactured at lower cost.

Viruses...good, bad, and bacterial

There aren't many examples of beneficial viral infections. Which is why one of the more intriguing mysteries about AIDS surrounds people co-infected with HIV and GB virus C (GBV-C)—a number of studies have suggested people with the dual infection have slower disease progression, higher CD4+ T cells counts and survive longer than those infected with HIV alone, suggesting GBV-C affords some protection. Bernhard Fleckenstein and Heide Reil at Friedrich-Alexander-University Erlangen-Nürnberg and their colleagues are investigating the mechanism of this protection by studying how the two viruses interact in peripheral blood mononucleocytes (PBMCs) in cell culture.

His team found that coinfection of human PBMCs with the viruses resulted in an average of 90% inhibition of HIV replication compared to single HIV infection. GBV-C inhibited all HIV subtypes and strains, including R5 and X4 isolates. Fleckenstein showed preliminary work suggesting that some of that inhibition is due to the GBV-C protein E2. When a recombinant version of this protein is simply added to the media of HIV-infected cells it inhibits HIV replication. Adding monoclonal antibodies against E2 protein abolishes this inhibition. It isn't yet clear what the precise mechanism of E2 inhibition is, how strain specific these effects are, or if other GBV-C proteins have similar anti-HIV activity. If E2 and other GBV-C proteins prove to be a powerful inhibitors of HIV activity, Fleckenstein speculates it may be possible to use them therapeutically.

For flu researchers, fears of a new pandemic are being fed by the recent discovery that the devastating 1918 pandemic strain was most likely an avian influenza that jumped to humans—a concern given the increasing reports of outbreaks of avian flu in humans. However little is actually known about how such pandemic strains form and cross the species barrier. To begin addressing this question, Hans-Dieter Klenk of Philipps University in Germany characterized the ability of avian and human flu viruses to infect human lung tissue. The flu virus protein hemagglutinin (HA) binds cells through sugar residues on surface proteins. The sugars recognized by human virus HA are 2-6 linked sialic acids while bird virus prefers 2-3 linked sialic acids.

The researchers infected cultures of human tracheo-bronchial eptihelium with the viruses using a system that maintains the morphology and function of the lung cells, including cilia beating. They found that 2-6 linked sugars are abundant on non-ciliated tracheo-bronchial cells and during initial infection human virus infects predominantly these cells. In contrast, 2-3 linkages were seen on ciliated cells and they were in sufficient numbers to allow infection of these cells by avian flu. "Ciliated cells are the entry site of avian flu viruses into the human respiratory tract," says Klenk. However, the avian virus did not spread through the culture while the human virus reached high enough concentration to infect even ciliated cells containing the non-preferred receptor. This suggests that human lung ciliated cells may serve as a milieu where avian and human flus can recombine.

Mammalian viruses form the basis of many vaccines now in development. But in the search for new tools, David Hone and his colleagues at Aeras have reached further down the viral evolutionary ladder. In their system, the f8 bacteriophage forms the basis of their delivery system for genetic sequences for the TB antigens 85A and 85B.The f8 phage contains a doublestranded RNA genome in three segments: small (S), medium (M), and large (L). The researchers removed phage genes from the two smaller segments and substituted those for TB antigens. The L segment, which contains all sequences necessary for the construction of the viral nucleocapsid, was left intact. These recombinant nucleocapsids replicate to copy numbers of 150 to 200 inside Shigella bacteria.

The plan is to deliver the nucleocapid-carrying Shigella orally and have the bacteria invade macrophage and DCs. The bacteria naturally escape endosomes and are engineered with a cell wall defect so that they lyse as they try to divide, spilling the f8 particles into the cytoplasm. There, finding themselves in a pool of ribonucleotides, the nucleocapsids spontaneously start spooling out single-stranded RNA from their interior that contains the antigen gene sequence, which is translated into protein by the host cell. Hone says they have already used the system in a mouse model to deliver the TB antigens. While preliminary, the results look promising. "We appear to be inducing IFN-g at the same level as an adenovirus vector with the same antigens," he says. This hybrid bacterial/phage vector could potentially be produced at low cost and, since the antigen genes are never present as DNA, carries virtually no risk for chromosome integration, one concern for DNA-based vectors. His team is now planning to determine the strength and protectiveness of the response elicited by vector and characterize the T cell subsets that respond.