More genetics done right -- GWAS and HIV controllers

About 1 in 300 people infected with HIV don't go on to develop AIDS. or do so much more slowly than expected -- this has been known for some time.  It was once thought that these people would never develop AIDS, so this subset of people was termed the "long-term non-progressors" but in fact many eventually do develop AIDS, and this group are now called "HIV controllers."

Now The International HIV Controllers Study reports the results of a genomewide association study (GWAS) undertaken to try to determine what is different about HIV controllers compared with those who develop disease (the ScienceExpress paper is here, and a story in The Independent about the report is here).  They included 900 HIV controllers from around the world and 2600 'progressors', people for whom the natural history of HIV infection was as expected. 

Not surprisingly, the significant genetic differences between controllers and progressors were within the major histocompatibility complex (MHC), a part of the immune system that distinguishes self from non-self, self from viruses and bacteria.  Specifically, they found 5 single nucleotide polymorphisms (SNPs) within a stretch of the MHC that codes for HLA-B, a gene for a protein involved in defending against viruses. It presents fragments of HIV on the surface of infected cells, and this presentation is recognized by CD8+ T-cells.  While there is extensive variation in MHC genes among humans, the HLA-B protein in controllers regularly differed from that of progressors by 5 amino acids, which changes the shape of  the protein and how it binds to HIV.  

Altogether, these results link the major genetic impact of host control of HIV-1 to specific amino acids involved in presentation of viral peptides on infected cells. Moreover, they reconcile previously reported SNP and HLA associations with host control and lack of control to specific amino acid positions within the MHC class I peptide binding pocket, where the HIV fragment is 'housed' on presentation. Although variation in the entire HLA protein is involved in the differential response to HIV across HLA allotypes, the major genetic effects are condensed to the positions highlighted in this study, indicating a structural basis for the HLA association with disease progression likely mediated by the conformation of the peptide...

(The illustration is from Wikimedia Commons and is of the backbone structure of HLA-B*5101 complexed with HIV's immunodominant epitope KM2 1e28. Peptide is shown in yellow in the binding pocket. Beta 2 microglobulin is show in the lower left, and membrane attachment site is in the lower right. 3D Structure is derived from Maenaka, K. et al. (2000) Nonstandard peptide binding revealed by crystal structures of HLA-B*5101 complexed with HIV immunodominant epitopes. J.Immunol. 165: 3260-3267)

The authors of the International HIV Controllers Study write that there are other differences between controllers and progressors that may be important in whether or how quickly they progress to disease, but that they believe that the differences they found in the HLA-B gene, and what they mean for viral/peptide interaction, are the most likely explanation.  They are cautious about over interpreting, even when talking to the media, which is not always the case. As quoted in the The Independent:

Dr Walker emphasised that the discovery is not like a "light switch" that turns someone into an HIV controller. It is one factor among several that increases the chances of someone being able to survive for many years with HIV and not antiretroviral treatment, he said.

"We've identified a major determinant but there are other factors that will influence the pathway. We've not identified the precise mechanism to explain HIV controllers but we know that of all the genetic influences involved, this is by far the most important," Dr Walker said.

And they can't yet describe the specifics of how the altered protein works, but that is the focus of current studies, and one can assume that they will ultimately be successful now that the target of investigation is clearer.

As regular MT readers know, we aren't great fans of GWAS in general, but this is one example of a successful study.  This is because the genetic effect is large enough to be detectable. The investigators used clever techniques to dissect out (statistically) the effects of particular variants in the HLA-B gene.  The strongest signal was in an allele called HLA-B*5701.  This was replicated and in fact was the only usefully strong signal in this study.  That shows that GWAS works, because that allele was already known to be involved in slow progression.  Indeed, it's screened for in treating HIV patients because those bearing that allele can over-react to a drug called abacovir (the drug is not used on such patients).  So, one must be tempered about this finding and the usefulness of the GWAS approach.

Beyond this, the study found weak signal (semi-imaginary?) in a CCR5-CCR2 region, where variation has been suggested in other studies to affect HIV sensitivity.  Nothing else in the genome generated any signal.  And these two sites were responsible for 19 and 5 percent of resistance.  So it is a typical GWAS result.  Nonetheless, it focuses attention on the HLA region as 'the' region to think about, at least at this stage.  And it shows that focused, problem-specific GWAS studies can do their job, even if there were other ways to find these genes--why?  because their signal is strong.  And it confirms the general challenge to us, that many genes with minor effect are probably involved here as with other complex traits....what to do about them is the next challenge.

A discussion of the paper in Nature concludes:

However, it will be a long time before this work gives rise to treatments or vaccines. "We're a long way from translating this, but the exciting part is that this GWAS led us to an immune response. That has to be good news for vaccines, because they manipulate the immune response," says Walker. "We're cautiously optimistic that this will help us develop ways of inducing better responses, because we now know what it is that we're trying to induce."

A general flu vaccine! (?)

The influenza virus is a tough critter to combat. It changes shape too often for effective vaccines to train the immune system to recognize all the possible shapes. That's why we need a new vaccine each year, targeted to the strain that the authorities guess will be predominant.

The problem is that the immune system recognizes the overall shape of the virus in order to attack it. A key molecule that immune systems 'see' is the protein hemagglutinin (HA), which is located on the virus surface (the blue and red mushrooms in the figure). The name comes from the fact that this kind of molecule can cause blood cells to coagulate when exposed to it, the details of which are unrelated to this post.


The HA molecule binds to a sugar-related molecule on host cells' surfaces, and the virus then pops into the cell where it can proliferate. There are many variants, or strains, of HA, numbered H1 to H16 etc., and a vaccine must be targeted to the specific characteristics of the strain that you want to be protected against. The strain variation targeted by traditional flu vaccines is in the blue or 'head' part of the HA molecule as seen in this figure.

A general flue vaccine would not have to be targeted to any given HA variant. It would in principle lead your immune system to recognize all strains, but that's not how flu vaccines have worked so far. However, a new tactic may be about to bear fruit. Part of the HA molecule is apparently not as visible to the immune system (shown as the red stems under the blue head in the figure), but it doesn't vary nearly as much as the head of the molecule--apparently, if it changes, the virus's protein coat won't form properly. This conservation of structure makes this a vulnerable region in the virus. The new strategy would target the stem, perhaps along with a seasonal supplement for a specific HA strain when/if needed. The hope is that this will generate broad long lasting, or even lifetime, immunity after a single vaccination in early childhood. Here is a release from Gary Nabel at NIH who is one of the investigators, explaining the approach, although we cannot yet find a detailed description of how the vaccine allows the immune system to 'see' the HA stem.

This is an evolutionary approach to vaccination because it targets the part of HA that is conserved, presumably by strong natural selection. You attack the pathogen at a place it cannot easily vary, so it can't quickly evolve a defense. Exactly why the HA stem section cannot vary very much is probably known, but not by us--perhaps it has to do with the proper assembly of the virus's protein packaging--all the coat proteins must fit tightly together, etc. There would thus be no stem-variant viruses that could take the place of those destroyed by the vaccine-boosted immune system.

Why natural immunity does not also generate antibodies that recognize the stem is an interesting question. Virologists may know the answer.....but we don't.

More on bugs

We've written before about how it seems that genomewide association studies are finding more and more genes associated with the immune system. And that's for a wide spectrum of conditions, from schizophrenia to macular degeneration to inflammatory bowel disease, and now, Alzheimer's Disease, according to a story in the NYTimes. These genes still explain only a small fraction of risk, but it is certainly starting to look like a trend.

The new hypothesis [about Alzheimer's] got its start late one Friday evening in the summer of 2007 in a laboratory at Harvard Medical School. The lead researcher, Rudolph E. Tanzi, a neurology professor who is also director of the genetics and aging unit at Massachusetts General Hospital, said he had been looking at a list of genes that seemed to be associated with Alzheimer’s disease.

To his surprise, many looked just like genes associated with the so-called innate immune system, a set of proteins the body uses to fight infections. The system is particularly important in the brain, because antibodies cannot get through the blood-brain barrier, the membrane that protects the brain. When the brain is infected, it relies on the innate immune system to protect it.

And, when researchers exposed the protein that constitutes plaque in the brains of many Alzheimer's patients to microbes, it was a fairly efficient killer. While a good fraction of people with dementia are found not to have these plaques upon autopsy, and many people without dementia do have them, the possibility that the innate immune system might fight infection of the brain as well as kill brain cells seems to be real. But it's a complex trait, and like any other complex trait, may well turn out to have more than one cause, one of which may be an innate immune system doing its job too efficiently.

The innate immune system attacks common features of microbes. That's different from the 'adaptive' immune system, which generates a huge array of different antibody molecules by random rearrangement of chromosomal segments (see Mermaid's Tale for some facts about these different systems). Variation in the latter is generated during your life, and inherited variation is not thought to be very important (since the system generates millions of new antibody configurations during your life). Variation in the innate immune system is inherited and works directly, the way genes are usually thought to work: the protein coded by the genetic variant (allele) that you inherit does its job at finding and poking holes in bacteria (or doesn't, depending on its configuration).

Infectious disease can be a strong effect. Whether or not it is systematic and durable enough to relate to natural selection over many generations, it may be that variation in susceptibility can lead to substantially different risk of disease to persons exposed to a given kind of pathogen. If that's the case, then mapping studies -- like GWAS, comparing cases and controls to find parts of the genome that seem to be related to case status -- might be able to detect the stronger risk-effects of genetic variation related to infectious disease. At least, it seems that for a wide variety of diseases a substantial fraction of GWAS 'hits' involve immune or inflammatory genes.

This could be a misleading surmise on our part. The 'immune' system does all sorts of jobs related to molecular recognition, and may involve a larger fraction of the genome than has been thought, so that its involvement in a given disease may not reflect infection, but some other function. Since as we said above these mapping 'hits' only account for a fraction of the case-control contrast, infection can't be the only causal factor.

Secondly, some studies find immune system 'hits' in diseases expected to involve infection, but not for other diseases (like heart disease) that are not. So not everything need be infectious. On the other hand, even something like heart disease risk can in part be due to infection, and instances and mechanisms are known that do that.

So there are interesting things to be found. If infection does turn out to be important, and if we keep over-using antibiotics, we may face a much more serious threat from that direction than from all the claimed 'genetic' risk that so much money is being spent to find by GWAS, biobanks, and other means. If that is the case, hopefully the research effort can be pried from the genetic vested interests in time to address the real problem before it's too late.

Evolution for real, and for keeps

Of all the things we should be spending research funds furiously on, infectious disease should be at or near the top of the list. Yet another story has appeared on the proliferation of antibiotic resistant species that are causing death and disease, mainly in hospital patients at present. But there's no reason to expect this to stay contained. And here we face real instances of genetic and evolutionary determinism. Since proper nutrition and exercise would also be a way to avoid massive numbers of otherwise early deaths, because hale people have at least better chances of surviving infection, we should refocus effort, away from luxury and made-up 'genetic' disease to these major societal problems.

There is, in fact, evidence that an increasing number of supposedly old-age, lifestyle-related diseases are going to turn out to be related to infection after all--either directly, or indirectly. Here we are learning from many avenues of research, including, yes we're free to acknowledge it, from GWAS approaches (case-control studies that searched for regions of the genome whose variation was associated with risk). Some cancers like stomach and cervical, among others, turn out to be due largely to bacterial or viral infection. Some heart disease is like that, too. But the genes identified by some of the more successful GWAS turn out to have connection with inflammatory gene networks. These include problems with intestinal, eye, and even psychiatric functions.

Direct infection may cause the disorder (the GWAS would be identifying genes whose variation somewhat protects, or makes one somewhat more susceptbible to the disorder). Or in a substantial number of diseases, the effect may be indirect: autoimmune diseases in which the immune system, after a real infection, mistakes your own cells for attacking bacteria; or perhaps infection in which the successful immune response then leads to chronic inflammation that itself may be damaging (these are called autoimmune disorders).

In genetic studies susceptibility variants may be brought detectably to the fore by the strong point-cause effects of infection, and that could explain why a disproportionate fraction of detected genes--of GWAS genomic 'hits'--seem to be involve the immune system. Since the susceptibility variants are not nearly fixed in the population--it's the fact of their variation that makes them detectable by association studies--they are probably not the result of a history of widespread or very serious infection with strong natural selection effects. And the signals usually only account for a small fraction (usually very small) of what are complex traits involving many contributing genes and environmental exposures.

But antibiotic resistance is a kind of classic Darwinism. Pathogens have simple genomes, huge population sizes, and rapid generational turnover. Antibiotics slam them with a sledgehammer kind of very strong selective effect. This is a cocktail of factors ready-made for the rapid evolution of the favored few--the lucky bacteria or viruses whose makeup makes them resistant. The wipe-out of their susceptible peers leaves an open field for the resistant ones to proliferate, and the rapid turnover and great population growth rate favors quick fixation of the new genetic variant in the population. Further, the dense population of very similar food items (humans, particularly hospitalized humans, at least so far) favors rapid growth and spread of the pathogen population, too.

Here is where evolutionary and genetic theory meet in classically relevant ways. The standard deterministic models work well, because the signal can be so much stronger than the statistical 'noise' of chance events in the lives and times of individual bacteria.

Whereas genetic susceptibility to disease in the host (us) is fairly similar in most individuals, we represent in a sense a standing target for the bugs to adapt to. We do have an immune system designed to evolve as fast as microbes do (it's called our 'adaptive' immune system, and is too intricate to go into here--but it is covered in our book and of course in many sites on the web), even this system may be unable to cope with the hardy pathogens that evolve in survivors of our antibiotic assault.

For whatever reason even the adaptive immune system in vertebrates, including humans, may not have been exposed to environments so changeable as those we're likely to subject ourselves to in the near future if we're not careful. Whatever the reason, if we're slow to respond we're going to teach ourselves a clear lesson in evolutionary genetics.

Going mano a mano with vaccines against non-infectious disease

Many chronic conditions now thought to be due to passive (genetic) rather than active environmental exposures may turn out to be due to infection, or to somatic changes of various kinds that build up with age.  Many of the chronic disease candidate genes identified by genomewide association studies (GWAS) and other approaches seem to relate to immune function or 'inflammation', whatever that may include (e.g., recent schizophrenia results)  Autoimmune diseases also may involve exogenous pathogens. 

Progress in vaccine production research may bring a host of infectious diseases under preventive control (see our "getting the bugs out" post)--and this would be preventive with respect to infection-related chronic diseases, as well. Here it's molecule against molecule--antigen against antibody in a molecular mano a mano combat.

But the same ideas may actually apply even more broadly. As we discuss at length in our book, life is about molecular recognition. In the case of infectious disease, it's molecules on bacteria and viruses that the immune system recognizes. But even most non-infectious diseases involve undesirable molecular recognition problems (too much or too little signal molecule or response to it, for example). Networks of interacting gene products (protein and RNA-based) are being identified.

Malfunctioning networks become dangerous if they affect too many cells, such as cells early in development that are the ancestors of major segments of the celllular/embryological descent tree that makes major organs. Thus, one mutant liver-precursor cell could have devastating effects, while the same mutation occurring in a cell in a mature liver will have no discernible effect (because the rest of the liver cells will be normal).

In complex non-infectious diseases, signaling malfunctions can amplify if a cell or set of cells produces too much or too little signal, it can induce other nearby cells to start doing the same or at least start hyper- or hypo-responding. Some traits, like epilepsies, may amplify from a single or small number of cells in this way. Many other diseases may be similar in this regard, such as diabetes or hormone-dependent diseases in which signal or receptor concentrations may be off-level.

To date, we're much better at treating infection than we are genetic diseases, but targeting a genetic network in some molecular-recognition way may be an eventual treatment approach to such complex diseases. If, as must usually be the case, the disease involves some sort of molecular misbehavior, some kind of nanochip, say, to test relative levels of various molecular components may be able to detect something going wrong. If the effect circulates so that the nano-detector 'sees' it, then perhaps some antibody-like 'vaccine' can target the protein that is produced in excess, bringing its level back to within safe limits.

So here could, at least in principle, be a way in which we can meld infectious disease strategies and network or systems biology concepts, to provide detection and treatment of complex diseases. To do that, individual genotyping would not be needed; all we would need would be a micro-implant that could detect component (or even sets of components) levels that were out of health range.

This may be dreamworld thinking at present, but it is not a stretch to think that infectious disease approaches rather than 'genetic' approaches (in the sense of inherited variation) will lead the way to major health advances (assuming we can afford them!).