ViroBlogy

Big News Day: HIV Vaccine Conference, Tuesday 13th September
The first plenary session of the conference had as its theme “Novel approaches in clinical evaluation through global collaboration” – and it was graced by the presence of no fewer than three scientists in full military dress uniform complete with medals, from the USA and Thailand (Nelson Michael and Jerome Kim and Punnee Pitisuttithum), reflecting the significant involvement of both countries’ military in the RV144 efficacy trial.

It was probably fitting, however, that it was led off by Pontiano Kaleebu of the MRC/UVRI Research Unit on AIDS from Uganda, on Africa’s contributions to AIDS vaccines.  He said that Africa had been crucial to the endeavour for a number of good scientific and societal reasons, but principally because most infections are there:  some countries are up to 15% of total population being HIV+ and sub-Saharan Africa contributed 17% of global infections in 2010.  Factors influencing vaccine efficacy that were unique were the great diversity of viruses, the mainly heterosexual transmission of viruses, diverse HLA alleles and significant preexisting vector immunity.

Small Buddha used for offerings, temple of the Reclining Buddha

South African scientists – largely drawn from the University of Cape Town, he says, modestly – had been responsible for the only vaccines designed in Africa, which were now in clinical trial.  Africa had been part of much work on epidemiology and variation of HIV-1.  Africa and Africans had contributed to understanding transmission events, mechanisms of early viral control and immune escape, and had helped in the addition of new broadly neutralising MAb derived from patients in African cohorts.

The first vaccine trial in Africa was done in 2000 in Uganda, and there have been many since:  30 trials or 17% of all trials have been done in Africa, mainly by  IAVI (13)  and the HVTN (11).

He noted that financing had declined and that the  reduced vaccine pipeline was a challenge, as many well-established sites have no vaccine to trial.  Another challenge was that new prevention successes means lower viral incidence, so trials have to be bigger – and may be impossible in certain cohorts.  There was also a challenge in the up and downstream HIV research imbalance in Africa – where there was no research infrastructure in many centres so sampes got shipped out, while clinical trials were large and well serviced.

His conclusion was that Africa had made significant contributions to vaccine research and development, but that challenges such as those mentioned could threaten further work.

Dr Punnee Pittisutithum of Mahidol University in Bangkok described how Thailand had a national plan established as early as 1993, and revised in 2006, to transfer technology, and to collaborate with a variety of institutions and countries, in HIV vaccine research and development and prevention efforts.  A collaboration in 1997 with Japan had used recombinant BCG as a subtype E vaccine prime, boosted with live vaccinia virus.  There had since been14 preventive trials including 2 efficacy trials.  In 1997 they had established a plan to monitor circulating virus – and now 1765 Thai viruses had been sequenced, and they had a very good idea of variation and currently circulating viruses.
They had an impressive infrastructure to set up and monitor clinical trials, which accounted for the success of the Thai trials over the years.  The partially successful RV144 efficacy trial had resulted in a study of correlates of protection involving 35 investigators from many institutes, including in Thailand.  She also made a point – as many others did subsequently of thanking the 16 402 Thai men and women who participated in the RV144trials.

Jerome Kim of the Walter Reed Army Institute of Research and Deputy Director of the US Military HIV Research Program spoke next, on correlates, sieve analysis and clinical development of the RV144 trial.

His first news was that there was a correlation of high Ab concentration to Env in vaccinees with a low risk of infection – resulting from 4000-odd samples analysed in the last two years for correlation. The work was the result of 35 investigators from 25 institutions collaborating on samples gathered during the trial.

A unique finding was that the gp120 and ALVAC vaccines were novel immunogens – the gp 120 from Vaxgen (also used in previous trials) used a N-terminal 11 aa replacement from a HSV gD glycoprotein epitope, which may have affected presention of Env HIV epitopes: its presence changed the binding of mAbs directed against gp120 up to 10-fold better for conformation-specific epitopes in the V2 and V3 loop epitopes, but not for linear epitopes.
There had been a good reaction to V2 peptides by intracellular cytokine staining (ICS) assays, with the response by CD4+ T-cells mainly.  One unfortunate finding was that gp120 Ab binding dropped 1.5 logs from 12 to 24 weeks post the last vaccination.

Sieve analysis – a new term to me, but denoting analysis of breakthough HIV or other infections in vaccine trials for selection pressure by the vaccine – looked at the gp120 V2 loop sequence in placebo and vaccine arm infectees – and saw selection.

It was left to the ununiformed Barton Haynes – Duke Human Vaccine Institute – to actually break the big news on correlates in the RV144 case control study.  Bart gave a lesson in the careful exposition of a complex topic of potentially huge significance in the HIV vaccine world, by starting with an explanation of what they had discovered – correlates of risk rather than of protection – and exactly what it meant.

He explained that a correlate of risk may be causal, or that it could be a surrogate marker.  Their team had looked at 41 infected and  205 uninfected vaccinees and 40 placebo recipients,  followed for 3 yrs.  There were only 41 cases, so the statistical study was only powered to pick up strong correlates and could miss weak ones.

Pilot work noted that most Ab responses were directed to the gp120 V2 loop.  In brief, their work found that there were two correlations associated with infection rate: the first was that serum IgG binding to a scaffolded V2 loop correlated inversely with infection rate.  The second was that Env binding of serum IgA correlated positively with infection rate: the statistical analysis showed a 43% reduction in infection rate associated with high serum IgG to the V2 region and a 54% increase in infection rate with high serum IgA binding.  Their hypothesis to explain this result was that high V2 IgG Ab levels were protective and low plasma IgA was associated with protection.

Ongoing analysis wase focused on looking at 9 PBMC-produced cytokines – and medium to high cytokine levels seemed to be correlated with protection.  Epitope mapping of IgA binding find C1 peptide response was correlated with infection, so maybe plasma monomeric IgA can block antibody dependent cytotoxicity (ADCC) caused by IgG binding?  He noted that only monomeric plasma IgA was collected in this trial, and mucosal dimeric IgA was to be collected next.

Haynes thought the way forward is to see if correlates of risk are causal correlations, by new clinical trials, or by having the trial Abs tested in non-human primates (NHPs) for passive protection.  They were presently testing the IgA association with risk by looking at mAbs in macaques binding all sorts things in Env, to see if IgA inhibited other Ab binding.

While the result was undoubtedly interesting – and unexpected – I was not convinced that it was as significant as it was made out to be, for a number of reasons – and a number of people I spoke to at the conference agreed.  Thus, in my slightly jaundiced opinion, the relatively weak correlations with risk applied for just one region in Env for this one vaccine combination with a unique monomeric gp120 product, for one subtype of HIV-1, in one population, for a trial in which the efficacy was only 30-odd percent, and which even then was only just significant.  They also tested only plasma antibodies, which may not give the full picture, and got next to no cellular response, which could correlate with the total absence of virus control seen in both placebo or vaccine recipients who became infected.

Accordingly, I think the massive obsession with analysing the results of the trial may turn out in retrospect to have been a bit of a waste, compared to testing new products which have markedly better responses in preclinical trials.  Another view can be seen here.

Giuseppe Pantaleo of the Centre Hospitalier Universitaire Vaudois in Lausanne closed out the session, with a presentation on poxvirus vector-based vaccines beyond the RV144 trial.    He pointed out that modified vaccinia Ankara (MVA) was the result of 571 passages in chick embryo fibroblast (CEF) cells; that the ALVAC canarypox vaccine resulted from 200 such passages, and vaccinia Copenhagen had had the deliberate deletion of 18 ORFS to result in the NYVAC vaccine.  The two former vaccines did not replicate in humans; however, replication competent poxviruses give appropriate innate immune cytokine responses and CD4 help.  To this end, NYVAC is known to grow in human primary keratinocytes, is highly attenuated, has no effect on dendritic cell (DC) maturation, and one gets higher levels and longer persistence of expressed antigens, cross presentation of Ag by MHC I and II receptors and stimulation of memory T-cell responses.

They tested replication competent and a replication deficient NYVAC and DNA expressing the same Ag and compared them, with  a boost of type C gp120.  They compared the effect of DNA priming or not, and scarification or intramuscular injection for NYVAC, with DNA and NYVAC both expressing Env and a Gag/Pol polyprotein.  The gag gene in constructs makes particles and a trimeric secreted gp140.  Pantaleo noted that the DNA plus regime elicited much more cellular immunity and a predominantly Gag/Pol response, while NYVAC alone gives 70-80% response to Env.  In the DNA+ group there was a balanced CD4 and CD8 T-cell response.  High Elispot results get long term and durable response.  There was no difference between scarification and im immunisaiton, and no increase of immune response with protein boost.  There was also no difference between rep and non rep NYVAC.  In the no DNA group the rep virus was lots better.

Most HIV-1 neutralising Ab response was in the DNA- groups and SHIV neutralisation was restricted to the DNA- group.

In ADCC assays for the DNA+ group there was no advantage in boosting with protein, and response decayed later with some animals being negative; in the DNA- group responses were considerably higher, there was an advantage to boosting,and all animals were positive.  In cross-type titring assays there was good cross-binding of IgG, with the DNA- groups being better.

The lesson from this was that a greater magnitude of T-cell responses do not necessarily correlate with neutralising (NAb) responses.

For plasma IgA responses they see the same distribution as for IgG.  In the DNA- groups they get very little response up to 3 months, then good responses 8-9 months, which then wane after 12.  Their response would be to boost with poxvirus plus Env at 12 months.  Pantaleo thinks we need compressed regimens to reduce the time of reduced protection, that we should try Env-only regimens, and that we should tailor vectors for optimal Ab responses.

My opinion on this is that one should try for Env-specific Ab responses AND Gag- and other protein-specific T-cell responses, elicited at the same time by immunisation in different limbs.

There were 6 Tuesday afternoon sessions, in two sets of three, so some judicious choices were needed.  I went with Oral Sessions 3 and 5, entitled Novel Immunogens and Inserts, and Acute Infection/Viral Diversity, respectively.

Oral 03: Novel Immunogens and Inserts
Two stand-out talks for me were one by A Flamar, and another by M Zhou – with a third on my favourite virus-like particles, by L Yang.

The Flamar talk reported targetting to CD40 receptors of five 19-32 aa peptides containing a string of known highly conserved CD4 and CD8 T-cell epitopes from Gag, Nef and Pol covalently linked to a lipid tail for antigen presenting cell (APC) uptake.  These have been tested and found to be therapeutic already.  The targetting is done using a MAb targetting CD40 with the HIV5 pep attached to the heavy chain C-terminus. The epitopes are 2 from Gag, 2 from Nef and one from pol.  The MAb is a humanised one with mouse Vh and Vkappa portions, which binds monocytes and APCs specifically.  The immunogen expands HIV peptide-specific CD4 and 8 T-cells from HIV+ patients.  They get broad peptide-specific responses and CD4 and CD8 polyfunctional responses.  The latter are CTL-characteristic and can kill target cells as well as suppressing HIV replication in vitro.

They are presently humanising the V region for clinical manufacture and testing in mice and NHP.

The Zhou talk discussed the use of mimotopes – peptide sequences mimicking native epitopes – displayed via phage surfaces, which mimic a membrane-proximal or MPER gp41 epitope, and bind Ab from an elite controller of virus load.  The M13 display library is made by env-specific PCR and fragmentation followed by cloning, and is bound by immobilised IgG.  “Panned” phage is eluted and amplified.  They get epitopes localised to gp41, inclusing the MPER region, using EC26-2a4 Ab.  The core epitope overlaps the known binding site of the broadly-neutralising MAb 2F5 but is distinct: the sequence is  NEQELLELDK.  They used this as an immunogen after a env DNA prime as phage plus adjuvant x3 – and got neutralising Ab back.  This is a genuinely exciting result, as it builds on much speculation regarding just how good Abs directed against this region are at neutralising a wide rtange of HIV variants – and answers some of the questions about how difficult they are to make.  These two sorts of immunogen – one aimed at T-cell responses, the other at neutralising Ab response – may yet be a valuable adjunct to other vaccines containing more conventional ingredients.

The third talk by Yang was very useful in that it demonstrated the possibility of using insect cells – which can be cultured very reliably at large scale, and are already used to make a major human vaccine (GSK’s anti-HPV Cervarix) – to make genuine HIV virus-like particles, with a Gag shell inside a membrane, studded with processed Env spikes.  I was especially interested as we have already used the same technology to make Gag-only VLPs, which are a real possibility as a subunit vaccine.  However, routine production of VLPs is complicated by the fact thats VLP are usually produced in low quantity, there is poor cleavage of Env, there is often recombinant baculovirus contamination, and poor batch consistency.

They used VLPs Drosophila S2 cells that had been stably transfected with plasmids encoding HIV-1 Gag and Env.  S2 cells are good as they grow up to high density and can easily be cultured in suspension – in a WAVE bioreactor in this case, which is scalable up to hundreds of litres.  They get glycosylated Env which undergoes appropriate cleavage and ends up as spike protein on budded Gag-containing VLPs.  They get 23 million cells per ml, and 8 mg of gp120 / litre – which is an excellent yield for VLPs.  Appropriate Mabs bind the spikes, indicating correct conformation.  Upon immunisation with a DNA prime and VLP plus CpG boost, they get a good Ab response which is weakly neutralising.  An ADCC test was also positive.  The T-cell response was a relatively poor CD4 but good CD8 cell.  The result is not entirely new – there were two posters at the conference describing the same thing, and our group has used stably transfected insect cells to make baculovirus-free VLPs – but they have investigated production at scale, and have shown appreciable and appropriate immunogenicity for what may be a valuable future component of heterologous prime-boost regimens.

Oral 5: Acute Infection/Viral Diversity
While I probably should have been more interested in acute infections, and there were several most worthy talks on this, I am inexorably drawn as a result of my history in virology and with HIV, to studies of virus diversity and especially of virus evolution over time and between individuals.  So there were really only two talks in it….

L Yin presented a fascinating account of the use of deep pyrosequencing to look at the evolution of viral diversity in peripheral blood cells in single individuals over time.  The study used pyrosequencing of cell-associated virus – which of course, reflects the whole history of the individual’s infection as integrated DNA – to look at diversity and both real and inferred longitudinal variation, given multiple blood samples from the individuals over time.  They used samples from children infected at birth, for time spans of 18 months to 6 years.  In six children, 4 of the 6 showed big differences in virus populations, while 2 did not not.  The biggest diversity was for R5 coreceptor binding sequences, illustrating immune selection of viruses.

Their conclusion was that deep sequencing was a robust method for evaluation of complexity and population structure and for evaluation of the virus historical record in an individual.  It was also easily possible to compare cell-associated and plasma-isolated virus, as DNA and cDNA respectively.

V Novitsky from the Essex lab presented on the dynamics of changes in Gag sequences in the global epidemic; how they change over time and the probable age of HIV subtype C in particular.

They sampled databases for sequences of 500 bases and up for gag, and found only 1800 -odd suitable sequences: these were mostly from South Africa, and Zambia, Malawi and Botswana.  The sequences were reduced for dating purposes to 433 by criteria such as <10 per year per country, while  966 were used for diversity.  They arbitrarily defined 9 groups of about 150 viruses over 20 years from 1983.  Interestingly, there was no clustering by year of sampling, or extinguishing of lineages.  SA had profound founder effects for 2 groups of viruses;  for diversity of Gag over time, one could see significant increase over time.  The p17 C terminus had the highest changes for this protein, p24 less so and spread throughout the sequence, with the most changes in the rest of Gag (p15).  Only 20 aa positions over 500 show consistent selection pressure changes, meaning the consensus sequence of Gag fom Subtype C HIV-1 remains pretty much same over more than 25 years.  They estimated the time of viral diversification from other subtypes to have been around 1959 – with a hefty uncertainty.

While the results may not seem exciting – and indeed, some people said “What’s new?” – the fact that Gag consensus sequences have essentially been stable over a protracted period is interesting; so too is the fact that lineages do not seem to have disappeared as one sees with influenza viruses with immune selection.  A very interesting virus, HIV-1….