Taking the Gritty Approach

Despite their commendable track record and promising future, particle-based recombinant vaccines have never been the sole focus of a scientific meeting. That just changed.

By Andreas von Bubnoff

Vaccines have long been made from live-attenuated or killed forms of targeted pathogens. And the approach has worked rather well: the smallpox vaccine, for example, was made this way and used to end a terrifying disease. But, for safety and other reasons, vaccine developers today tend to favor recombinant vaccines, in which parts of the pathogen are synthesized from scratch and used as immunogens. The resulting vaccines contain either soluble proteins or entire virus-like particles (VLPs), which are similar in size to viruses.

Between the two, VLP-based vaccines have by far the better track-record. Indeed, as Martin Friede of the World Health Organization (WHO) pointed out at a recent conference on virus-like particle & nanoparticle vaccines, both of the approved recombinant vaccines—against hepatitis B (approved in 1983) and HPV (approved in 2006)—are based on VLPs. What’s more, another VLP-based vaccine, GlaxoSmithKline’s malaria vaccine candidate RTS,S, might well be approved in the near future. By contrast, no non-VLP recombinant vaccine has yet been approved for general use, Friede said.

Not, however, for want of effort. In fact, Friede noted, the bulk of research on recombinant vaccines has focused on non-VLP-based approaches. Between 1983 and 2006, 1284 of the 1339 published clinical trial reports studied non-VLP vaccines. Only 55 evaluated VLP vaccines. In fact, this meeting—held Nov. 28-30 in Cannes, France—was the first solely focused on particle-based recombinant vaccines, according to meeting organizer John Herriot and Gregory Glenn, Chief Medical Officer of Novavax, a company that develops VLP-based flu vaccines, who was on the scientific advisory panel for the conference.

It rained heavily throughout the conference, but there was little reason to go outside: the roughly 140 attendees were treated to talks on particle-based vaccines addressing everything from flu and HIV to the obscure afflictions of salmon.

Particulate matters

Many VLP vaccine candidates are built from viruses that infect bacteria (bacteriophages), or those that infect plants, animals, or even humans. Given the abundance of alternatives, Friede said, there’s little reason to use human viruses. Preexisting immune responses to those viruses can, for example, affect immune responses to any vaccines based on them. 
That’s unlikely to be an issue for vaccines made using synthetic particles. Such vaccines can also have better safety and purity profiles, as they can be made without resort to biological processes. 
But what accounts for the greater immunogenicity of particle-based vaccines? In the opening talk of the conference, Martin Bachmann of the University of Zurich said one important consideration is size. The particles can’t be too large, as only particles smaller than 200 nm can directly enter lymphatic vessels from vaccination sites, so that they can be transported to the lymph nodes where they activate B cells.

Another important feature is what Bachmann calls “repetitiveness.” That is, the particles display the antigen many times on their surface. This is important for a good antibody response, he said, because it enables several B cell receptors (BCRs) to bind an antigen at the same time, which boosts the activation of B cells. Repetitiveness is also important for activation of the complement system, which in turn increases B cell activation, he said. As it happens, both of these features—a diameter less than 200 nm, and repetitiveness—are also typical of viruses (HIV is about 100-150 nm across, for example). Further, Bachmann noted, some particle-based vaccines also carry toll like receptor (TLR) ligands on their surface, which activate the innate immune response and so improve both humoral and cellular responses.

VLPs against HIV

VLPs seem to be quite popular with AIDS vaccine researchers. Richard Compans from Emory University presented work on one VLP-HIV vaccine concept to induce antibodies to the membrane proximal external region (MPER) of gp41, which is important for fusion of the viral membrane with the target cell membrane. The VLPs consist of a sphere of plasma membrane that contains the HIV core protein inside and HIV Env on the surface. The recombinant HIV genes are expressed in baculovirus, which infects cultured insect cells, and the VLPs are purified from infected cells.

The researchers boosted the number of Env spikes on the VLPs to ten times the number found on HIV, which typically displays about 10-15 of them. The antibodies induced by these VLPs were not very potent, but became more so when VLPs were also engineered to carry a membrane-anchored form of bacterial flagellin, which activates TLR5 (MBio 2, e00328-10, 2011).

In another study, the researchers made the gp41 part of Env more accessible to the immune system by replacing the gp120 part of the Env spikes with the smaller hemagglutinin (HA1) subunit of the flu virus. Immunization of guinea pigs with the resulting “HA gp41 VLPs” induced antibodies that were specific to the gp41 part of Env and that neutralized tier 1 and 2 viruses (PLoS One 6, e14813, 2011).

One advantage of the approach, Compans said, is that gp41 is presented in its native conformation on the VLP. Compans also described an improved way to deliver the VLP vaccine: using a device with several microneedles, each a fraction of a millimeter long and coated with a VLP-based flu vaccine displaying the HA protein. In mice, this induced 10-fold higher immune responses than did an intramuscular injection with the same amount of vaccine.

Tsafrir Mor from Arizona State University also presented a VLP-based approach to make an HIV vaccine candidate that elicits a gp41-specific antibody response. In this case, the VLPs were made in tobacco plants. Aside from being relatively inexpensive, Mor said, the approach ensures that prions and other contaminants that can be introduced by animal cell lines are not a concern.

Like Compans, Mor and colleagues managed to increase the number of Env spikes on the surface of their VLPs about 10-fold compared to HIV. To expose gp41, they removed the outer gp120 portion of Env. The remaining gp41 retained its transmembrane section, Mor said, to better recapitulate the protein’s native shape, especially in regions close to the plasma membrane.

It seems to have worked. Mor and colleagues found that their VLPs, which only contained HIV gp41 and Gag, bound two broadly neutralizing antibodies (bNAbs), 2F5 and 4E10, that target the MPER of gp41. In addition, intranasal immunizations of mice elicited antibodies to the gp41 MPER and to Gag in the serum and in mucosal tissues such as the vaginal mucosa.

Fabien Pitoiset, a graduate student from the Université Pierre et Marie Curie in Paris, reported on a VLP HIV vaccine candidate based on murine leukemia virus (MLV), a retrovirus that infects mice. While the VLPs still contain MLV Gag proteins, the researchers replaced the MLV Env on their surface with HIV Env gp140, which includes the entire part of the HIV Env spike that lies outside the plasma membrane. To increase the number of HIV Env spikes on the VLP surface, they replaced the transmembrane part of gp41 with a transmembrane domain from vesicular stomatitis virus. They produced their “retroVLPs” in a mammalian cell line called 293.

The retroVLPs activated cultured dendritic cells (DCs), and subcutaneous injection into mice resulted in uptake of the VLPs into DCs in lymph nodes and the activation of these DCs. Pitoiset said this resulted in “good” cellular and humoral immune responses to HIV Env, which were further improved if the VLPs carried single stranded RNA inside, which is known to activate innate immune responses through TLR7 and 8.

Pitoiset and colleagues plan to include other TLR ligands such as bacterial flagellin in their VLPs, and to insert conserved parts of internal HIV proteins such as Gag, Nef and Tat into the MLV Gag proteins of their VLPs to improve cellular immune responses.

One challenge for HIV vaccine development is that it takes many years for HIV-infected people to develop the kinds of bNAbs that have been shown to neutralize a wide range of HIV strains. Researchers are currently trying to design vaccines to induce similar antibodies. Such antibodies are highly affinity-matured: their potency and breadth of neutralization derives from extensive hypermutation, which makes them very different from their germ line precursors. This means that people injected with vaccines devised to elicit bNAbs may not be able to produce the desired antibodies until much later—perhaps even a few years after vaccination. Of course, this would be very far from ideal.

One way to begin to address this problem is to develop a vaccine that contains proteins or peptides that preferentially bind the germ line precursors of HIV-specific bNAbs (see Vaccines to Antibodies: Grow Up! IAVI Report, July-Aug. 2010). At the meeting, John O’Rourke, a research assistant professor in Bryce Chackerian’s lab at the University of New Mexico in Albuquerque, presented initial steps toward the development of such a vaccine.

In collaboration with Gary Nabel’s lab at the National Institute of Allergy and Infectious Diseases’ Vaccine Research Center, O’Rourke performed a VLP-based screen to find peptides specific to the germ line precursors of bNAbs that target the CD4 binding site: VRC01, VRC04 and CHAVI31. The researchers made a library of VLPs, where every VLP presents one of billions of different random peptides on its surface.

To make the VLPs, they used bacteriophage protein MS2 which, when expressed in the bacterium Escherichia coli, self-assembles into a 90-subunit VLP, and inserted billions of random 6-10 amino acid peptides into the MS2 protein. They then screened this library for VLPs that best bound the germ line versions of the bNAbs.

O’Rourke said their approach resulted in the generation of VLPs that each contained the RNA molecule that encoded the antibody-binding peptide presented on its surface. This clever arrangement allowed researchers to quickly identify the sequence of the peptides that were displayed on the VLPs that bound the germ line antibodies most strongly.

Using this system, O’Rourke’s lab has identified several VLP-associated peptides that bind strongly to the germ line precursors of all three bNAbs. Nabel’s lab has shown that one VLP that binds to the CHAVI31 germ line precursor, and two that bind to the VRC04 germ line precursor, can also activate BCR signaling in vitro.

O’Rourke said immunization experiments of humanized mice are underway to identify the kinds of immune responses these VLPs generate. The ultimate goal, he said, is to develop a vaccination regimen that can kick start the affinity maturation process and guide it towards the production of bNAbs. This could be accomplished, he thinks, by priming with a germ line binding vaccine, and boosting with a vaccine carrying antigens that are more similar to the actual HIV components.

Toward broader flu vaccines

Flu was a focus of the meeting as well. Peter Pushko, president and CSO of the company Medigen, presented an approach that, he said, for the first time makes it possible to present proteins from three different flu strains on the surface of one VLP. This could help reduce costs, Pushko said, because it enables researchers to manufacture just one particle against three different strains of seasonal or pandemic influenza. He said production of the VLPs, from generating antigens with the right sequence to the final VLP, only takes 6-7 weeks, which would make it easier to quickly respond to new flu strains.

To make such trivalent VLPs, the researchers expressed several flu proteins in baculovirus-infected insect cells: Hemagglutinin from three different flu strains, neuraminidase, and the matrix protein M1. Pushko showed that such VLPs containing HA from three different pandemic viruses were immunogenic in ferrets, which are a well-established animal model in flu vaccine development. They also protected ferrets from challenge with each of the three corresponding pandemic strains. The same was the case with VLPs that contained HA subtypes derived from three different seasonal influenza viruses (Vaccine 29, 5911, 2011).

Synthetic approaches

Many vaccine candidates described at the conference were made in cell lines, plants or bacteria. But synthetic particle vaccines are no less intriguing. Arin Ghasparian, CEO of the Zurich-based company Virometix, reported that his company has been developing synthetic VLPs that are 20-30 nm in size and contain 72 monomers. Each monomer has a lipid portion and a protein portion. The protein portion has the shape of a “coiled coil,” which can bind to the coiled coil portion of other monomers. Make a mix of the monomers, and the VLPs assemble themselves into particles with peptide exteriors and lipid-lined interiors. Immunogens that are to be displayed on the surface of the VLP are attached to the protein part of the monomer before VLP self-assembly takes place.

Ghasparian said one advantage of this system is that it takes only a week to make a desired monomer. Another is that the resulting VLPs are uniform, well defined, easy to purify, and stable outside the fridge. Ghasparian presented preliminary results obtained using a Virometix VLP that displays a 24 amino acid part of the HIV Env V3 loop on its surface. This stretch of V3 has previously been shown to bind to F425-B4e8, one of the few V3-specific mAbs that can neutralize HIV strains from different subtypes. In addition, the structure of this V3-antibody pair has been determined (J. Mol. Biol. 375, 969, 2008).
Ghasparian and colleagues grafted onto their VLPs a version of this V3 peptide that had been stabilized in the conformation recognized by the F425-B4e8 antibody. They found that immunization of rabbits with these VLPs induced V3-loop-specific antibodies that could neutralize the same HIV strain that had served as the source of the V3. It also induced antibodies that neutralized other HIV strains, but only after removal of the HIV Env V1 and V2 loops, which in the intact Env spike occlude the V3 loop (Chemobiochem. 12, 2829, 2011). This means that the V3-VLPs are not useful for development into an HIV vaccine candidate. But it shows that, in principle, this VLP system can be used to develop vaccines that elicit functional immune responses in vivo, Ghasparian said.

Ghasparian and colleagues also immunized rabbits with VLPs that contain the malaria circumsporozoite antigen that is also used in the RTS,S vaccine candidate. They found that this induced antibodies that bound to whole malaria sporozoites.
Julianna Lisziewicz, president and CEO of the company Genetic Immunity, presented a therapeutic vaccine approach that’s based on synthetic nanoparticles and intended to induce an immune response that kills infected cells in HIV-infected individuals. The immunotherapy, as Lisziewicz calls it, is named DermaVir and consists of synthetic nanoparticles with diameters between 70 nm to 300 nm. Each carries a single plasmid DNA inside that encodes all 15 HIV proteins, two of which are nonfunctional to ensure that the resulting HIV particles cannot integrate or replicate (Vaccine 29, 744, 2011). This, Lisziewicz said, makes DermaVir the most complete vaccine modality in terms of the number of HIV epitopes used.

Before DermaVir is used, a rough sponge is rubbed over the application site to disrupt the outermost epidermal layer (stratum corneum). Then the skin area is covered for three hours with about 1013 nanoparticles, which are applied as a liquid and contained by a patch. To improve the immune response, the patches are applied at different locations, so that responses are simultaneously activated inside a number of different lymph nodes.

Lisziewicz said data from mice, rabbits and monkeys show that some of the particles are taken up by Langerhans cells, the antigen presenting cells (APCs) of the skin. They then carry the particles with them as they migrate to lymph nodes, mature into DCs and express the HIV genes from the nanoparticles. The resulting HIV VLPs contain all HIV proteins but cannot integrate into the genome or replicate.

Evidence from mice, monkeys, and human trials shows that these HIV VLPs can induce HIV-specific central memory CD8+ T-cell responses with broad epitope specificity and “high proliferative capacity,” Lisziewicz said. This, she added, is the same type of immune response that keeps HIV replication under control in long term nonprogressors (J. Virol. 86, 6959, 2012). A recently completed phase II trial called GIEU006, she said, showed that treating HIV-infected patients, who were not on highly active antiretroviral therapy (HAART), with DermaVir four times in 24 weeks resulted in a 70% reduction of viral load, compared with placebo-treated patients. To Lisziewicz, this suggests that the CD8+ T-cell response induced by DermaVir kills HIV-infected cells. The effect of DermaVir on viral load is, however, less potent than HAART. Lisziewicz hopes that this can be improved by customizing the HIV sequences in the vaccine DNA so that as many epitopes as possible overlap between the vaccine and the HIV strains the patient is infected with (and the parts of these HIV strains that are presented to the immune system).

A recent phase I trial showed that DermaVir also induces an HIV-specific CD8+ central memory T-cell response in people on HAART (PLoS One 7, e35416, 2012). Next, Lisziewicz plans a clinical trial to study if such people continue to show undetectable viral loads after they interrupt treatment. “We are hoping to show that at least some of these patients will respond,” and retain undetectable viral loads after treatment interruption, she said, adding that such patients could then be given DermaVir once a year to keep the virus suppressed.

One advantage of DermaVir treatment is that it has fewer side effects than HAART. And because DermaVir kills HIV infected cells, Lisziewicz hopes that it will reduce the HIV reservoir, setting the stage for the successful application of a future curative therapy.

Synthetic particle vaccines are also being used to bridge a gap in vaccine development: the need for vaccines that protect newborns from many infectious diseases. Ofer Levy of Boston Children’s Hospital and Harvard Medical School said that some established vaccines, for example polysaccharide vaccines such as PneumoVax, don’t work in newborns because their immune systems are immature—featuring incompletely developed lymph nodes, for example, and lacking certain elements of the complement response.

To address this issue, Levy and colleagues are working on vaccines for newborns based on polymerosomes, 100-150 nm sized micelles formed by molecules that have both hydrophobic and hydrophilic segments. Because DCs are critical to a good vaccine response, Levy wants to test vaccines in an in vitro cell culture system that contains neonatal monocyte-derived DCs (MoDCs), which come from cord blood monocytes.

Another reason many vaccines don’t work in newborns is that their immature APCs, such as DCs, are not easily activated. But Levy presented preliminary results suggesting that this barrier might be overcome with the use of his polymerosomes. He found that the nanoparticles are taken up by the cultured neonatal MoDCs, and even induce a cytokine response, which becomes stronger if the polymerosomes contain the TLR7/8 agonist resiquimod (R-848).

Next, Levy wants to develop a polymerosome-based HIV vaccine for newborns, in part because birth is the most reliable of relatively rare points of contact people have with health care providers in many developing countries. “In Africa, if you want to get immunizations into the population, it’s going to be at the point of birth,” he said.

He plans to put the HIV Gag protein and resiquimod inside the polymerosomes, and test whether that can induce antigen-specific immune responses in an in vitro culture system that simulates immunization of newborns. The culture will contain the neonatal MoDCs that take up the polymerosomes, and lymphocytes taken from cord blood of the same newborns that served as the source for the MoDCs. This way, Levy said, he can test if the polymerosome vaccine can induce lymphocyte proliferation and transition to a memory cell phenotype. If this works, Levy plans to test the system in monkeys.

When asked about possible safety concerns in newborns, Levy conceded that there is a higher safety bar when developing vaccines for newborns. Still, he said, the good safety and efficacy track record of BCG, a live-attenuated vaccine that activates multiple TLRs and is commonly given as a neonatal vaccine to prevent tuberculosis, provides some proof of concept and reassurance regarding safety issues. “When I started talking about this topic seven years ago,” Levy said, “people used to tell me ‘are you crazy? You are going to give a TLR agonist to a newborn?’ Well, guess what, BCG activates TLR2, 4 [and] 8. So on a daily basis all over the world, newborns are being injected with Toll 2, 4 and 8 agonists.”