The Compatibility Gene, page 22
There’s one very special thing about trophoblast cells which looks to be important for resolving this: although they lack HLA-A and -B, they have at their surface a peculiar HLA protein that’s almost never seen anywhere else in the body: HLA-G. The shape of HLA-G is very similar to the -A, -B, and -C proteins, but HLA-G differs from these other HLA proteins in that it doesn’t vary much between each of us (it’s one of the non-classical HLA proteins).23
The HLA-G gene was identified in the late 1980s, but it took many years to find out where it was used in the body.24 Early evidence for the protein being used in the placenta sparked controversy – due to different views as to what constitutes proof of the presence of HLA-G. The problem is that there’s such huge variability in HLA-A, -B and -C proteins that it is very difficult to get any reagent or process to reveal the presence of HLA-G specifically and not any other HLA protein.25 Eventually, a consensus was reached that HLA-G is indeed on placental trophoblast cells, and so the next question was: what does it do there?
Several of its features indicated that it would not do the same job as the other, more common, HLA proteins. For example, HLA-G stays at the surface of cells for a very long time, while other HLA proteins turn over to give an up-to-date report on what’s being made inside each cell.26 From 1995, attention focused on whether or not HLA-G on trophoblast cells would affect the NK cells that Bulmer, Moffett and Starkey had found to be abundant in the uterus during pregnancy. In 1996, several teams independently found that HLA-G was capable of switching off the killing action of NK cells.27 The implication was that HLA-G on trophoblast cells marks these cells as special – specifically telling the mother’s NK cells to leave these cells alone; these foetal cells are non-self but they are not dangerous. For such an important discovery, repetition of the experiment in different labs is needed to build confidence in the community, so it helped that different teams observed NK cells being switched off by HLA-G. But in fact, the teams disagreed over the way in which HLA-G did it. Researchers were at odds over how NK cells were able to detect the presence of HLA-G or, specifically, which receptors on NK cells could bind to HLA-G.
One possible cause of discrepancy was that each team was using their own lab’s cells genetically altered to make HLA-G. To test whether or not this was a problem, one group requested a sample of the cells being used in another lab so that it could carry out a direct comparison. The request alone is enough to spark some feeling of ill trust, but things got far worse when the wrong cells were sent out. Somewhere along the line, one team’s cells had been mixed up, so that, in fact, experiments thought to be done on cells having HLA-G had actually used cells genetically altered to make a different HLA protein instead. Not to name or shame any particular person or team, this anecdote shows how science progresses through everyday human errors, which in fact occur far more frequently than strokes of genius or even serendipitous breakthroughs.
In the end, it became clear that some of the data that had been published were plain wrong. No one’s publications were ever formally retracted, or even officially corrected; just everyone in the community knew where the errors were. We know now that HLA-G can switch off immune cells in several ways, but it still remains unclear whether it affects all NK cells or only a fraction of them.28 In any case, Moffett – and many others – think that the whole idea of it being critically important to switch off uterine NK cells has been one big red herring, a decade-long diversion because our thinking took a wrong turn.
As we’ve just discussed, it does seem to make sense that trophoblast cells – which lack normal HLA proteins – have the special HLA-G protein to switch off NK cells that would otherwise kill cells that are missing HLA proteins. Well, sort of – it still seems strange that so many NK cells are present in the uterus during pregnancy. They surely can’t be there just to be turned off?
Moffett thinks that Medawar’s question of how a mother’s immune system gets switched off might have been the wrong thing to ask all along. Instead – Moffett thinks – we should be questioning why immune cells accumulate at the foetal–maternal interface in the first place. She’s right; because a closer look at these uterine NK cells shows us that these cells aren’t what they seemed at first.
NK cells from blood take their name – Natural Killers – for being good at killing diseased cells such as tumours, but it turns out that NK cells from the uterus are only weakly able to kill other cells. In fact, this was something Moffett reported early on but the observation was largely ignored for well over a decade. Everyone raced to work out how uterine NK cells were switched off without carefully considering whether or not they really need to be switched off.
Eventually, others caught up with Moffett. Several research teams – including Jack Strominger at Harvard, who had earlier worked with Bjorkman and Wiley to get the shape of the HLA protein – also found out that uterine NK cells were not good at killing.29 In fact, Strominger established that the activity of hundreds of genes is different in uterine NK cells compared to blood NK cells.30 The uterine cells get to keep the name ‘Natural Killer’ because they share many features with their blood counterparts – and they can deliver a lethal hit if pushed – but they don’t seem to have a killer instinct. It may not be so important after all for trophoblast cells to use HLA-G as protection against NK cells in the uterus, because these immune cells can’t kill very well anyway. And if killing isn’t their thing, what do the NK cells in the placenta really do?
Yaqub (Jacob) Hanna, a Palestinian Arab working with Ofer Mandelboim, an Israeli Jew, both at the Hebrew University in Jerusalem, discovered that NK cells – far from being involved in combat – actually secrete growth factors and other proteins which stimulate the invasion of trophoblast cells into the mother’s uterus. The implication of this is that, far from killing other cells, NK cells in the uterus can help shape the structure of the placenta during early pregnancy.31 Other researchers found that uterine NK cells also have a constructive role in mice (despite there being many differences between pregnancy in mice and that in people).32 One study, for example, even found that a bone-marrow transplant – which provides an abundance of immune cells – can reverse certain reproductive problems in mice.33 So, instead of being agents of destruction, NK cells in the uterus might actually aid blood flow in the placenta and help pregnancy succeed.34
This idea remains open to debate because it’s very hard to test directly what NK cells really do inside a woman’s uterus – and because uterine NK cells are hard to obtain in large numbers. Hanna and Mandelboim’s study, for example, had to use tissue from more than 550 elective abortions.35 To increase the numbers of uterine NK cells, scientists can culture them in the lab before beginning experiments. Some of the cell’s properties could very well change when grown in the lab, and they may well behave differently from when they are in a uterus. But there is evidence that Hanna and Mandelboim’s findings are relevant to NK cells in their natural environment.
Even though mouse anatomy is very different from human, mouse NK cells still interact with trophoblast cells in the uterus during pregnancy. And the activity of NK cells can influence how dilated the maternal uterine blood vessels become during pregnancy.36 Mice don’t get eclampsia or pre-eclampsia, but the level of blood supply in the uterus can directly affect their reproductive success in other ways. In mice, a high level of blood flow in the uterus can better support larger babies or an increased litter size. For that reason, many scientists think that NK cells help blood flow in a placenta; and that activating these immune cells is a benefit – not a hindrance – to pregnancy. So – even if anatomic details vary – there is evidence that pregnancy and immune-system genes are linked in many species.
If NK cells are there to help – and don’t need to be switched off – where does this leave HLA-G? What does this special HLA protein really do after all? Moffett, Mandelboim – and many others – simply say: we just don’t know.37 But the diversion of studying whether or not HLA-G can switch off NK cells has turned out to not be in vain. An ability of HLA-G to ward off immune cells, even to some extent, led research teams to discover that tumour cells – and perhaps other diseased cells – can usurp HLA-G for their own benefit. That is, some tumours make HLA-G themselves – to shield against an immune cell attack.38 This indicates that HLA-G could, in fact, be a target for anti-tumour drugs – or perhaps used as a diagnostic marker for especially dangerous tumours.39 Another potential medical use for HLA-G is that its ability to inhibit an immune response could be exploited to aid organ transplantation. Time will tell if these clinical applications prove viable.
All this information about trophoblast cells and NK cells in hand – fascinating as it is – doesn’t answer Moffett’s original question of why some women have pre-eclampsia and some don’t. So, after Loke retired in 2002, Moffett decided that an altogether different approach was needed to directly test for the importance of our immune system in pregnancy. She decided to find out whether or not particular immune genes – or combinations of genes between each parent – make pregnancy more or less likely to be successful.
A specific idea about what to look for came to her in thinking through the details of how cells interact in the placenta: on the surface of trophoblast cells, there will be the baby’s HLA-C proteins – which include those inherited from the father. These trophoblast cells contact the mother’s uterine NK cells, and the HLA proteins they have could either weaken or strengthen the activity of the NK cells – depending on how the receptors on the mother’s NK cells react to the versions of HLA-C inherited by the baby. This could influence the level of secretion of growth factors from the NK cells – which impacts blood flow in the placenta, in turn influencing whether or not pregnancy was successful. In this way, Moffett reasoned, the combination of the mother’s NK cell receptor genes and the HLA-C genes inherited by the baby – including those from the father – could affect the success of pregnancy.
Family histories and population-based studies had already indicated that susceptibility to pre-eclampsia could be inherited, but nobody knew which genes were important. Moffett’s idea was nice – but plenty of nice ideas fall by the wayside when tested rigorously. As Darwin’s friend Thomas Huxley said: many a beautiful theory was killed by an ugly fact. To test her idea properly, Moffett had to set up a genetic study to find out if maternal NK cell genes and foetal HLA-C correlate with the success of pregnancy. To do this, genes were analysed in blood taken from 200 women with pre-eclampsia and a similar number of women who had normal pregnancies. Their babies’ genes were analysed using umbilical-cord blood or mouth swabs.
Moffett found that no particular version of HLA-C on its own correlated to whether or not mothers had pre-eclampsia.40 But the risk of pre-eclampsia was increased when certain versions of HLA-C genes were inherited by the baby and the mother had particular NK cell receptor genes. One way that these data can be interpreted is that certain combinations of genes between parents can lead to trophoblast cells switching off NK cells to some extent.41 HLA-C is able to switch off NK cells – as we’ve discussed in the context of ‘missing self’. So HLA proteins inherited by the baby could dampen activity of the mother’s NK cells – depending on the specific versions of HLA-C inherited and which NK-cell receptor genes the mother has. This could lower the NK-cell secretion of growth factors, leading to insufficient blood flow in the placenta and in turn, problems in pregnancy. This is a plausible scenario – consistent with the genetic analysis of parents and babies – but in truth, it’s not really known how these genes influence the frequency of pre-eclampsia. Even without understanding exactly how this works, these results show that differences in our immune-system genes can influence who gets born.
Defects in the placenta can cause other problems in pregnancy, not just pre-eclampsia – for example, recurrent miscarriage. Up to 3 per cent of couples in the UK have three or more consecutive miscarriages, which is more frequent than would be expected by chance – indicating that some couples are prone to miscarry. There are many issues that can underlie recurrent miscarriage, but one involves an insufficient blood flow in the placenta.42 Moffett tested whether or not any particular combination of immune-system genes would be unusually frequent in couples who suffered recurrent miscarriages and discovered that – just like for pre-eclampsia – particular combinations of HLA-C and NK cell receptor genes correlated with the risk.43 This time, her analysis revealed that a receptor protein that increases NK cell activity was protective.44 This is, once again, consistent with the idea that activating uterine NK cells is good for pregnancy.
Moffett also found that poor growth of the baby – a condition formally called foetal growth restriction – similarly correlates with particular combinations of NK receptor genes and HLA-C.45 The genetic link here again fits with the idea that activation of NK cells – and not too much inhibition – is important for a successful pregnancy. Altogether, Moffett’s series of genetic studies indicate that pregnancy is wired to be more successful with couples having particular combinations of immune-system genes.
It’s not that if you have this or that genetic inheritance you must have children with this or that other person, because these effects only slightly increase or decrease the relatively small risk of there being particular problems. As Isaac Asimov said, while thinking about the behaviour of gases: you can’t tell what an individual molecule is going to do, but if you deal with trillions, quadrillions and quintillions, you can tell, very accurately, what they’re going to do on the average.46 Similarly here, these small effects don’t predict who exactly will have problems in pregnancy – but they shape humanity overall.
We are only at the beginning of understanding this, but already there are many implications. First, there are potential medical benefits, as these discoveries seed new ideas for solving problems in fertility and pregnancy. Although it’s not easy to predict which couples are likely to have a problem in pregnancy – because these immune-system genes only contribute a little to the overall risk – it could help to diagnose problems by checking the activity of uterine NK cells during pregnancy. The difficulty with this is in how to assess uterine NK cell activity. NK cells in blood taken from a mother’s arm are obtained more easily than uterine NK cells, but it’s not yet clear whether or not blood cells could report useful information that correlates to the state of cells in the uterus. It’s also not clear – if a problem is detected – how best to manipulate the activity of NK cells in the uterus. Administration of hormones could alter the number of NK cells in the uterus, but we don’t know yet whether or not the number of NK cells, rather than their state of activation, can influence pregnancy outcome.47 Upcoming clinical trials will assess the possibility of using drugs that manipulate NK cells to help with problems in pregnancy.48
Aside from seeding new ideas for medicine, these discoveries say something fundamental about human nature. It could just so happen that reproduction has co-opted use of these highly variable immune-system genes, and we shouldn’t read into it any more than that, in the same way that it doesn’t matter much that the vas deferens tube traffics sperm a longer way round than necessary. But to me, this is not like the vas deferens tube situation and it is almost certain that this does matter; that this genetic link between reproductive success and our ability to fight disease persists because it is beneficial.
There’s not much cost to the vas deferens taking a detour on its way from the testicles to the urethra. So there’s little pressure for the path this tube takes to be as short as possible. In contrast, there’s an immense selective pressure on genes that influence the success of pregnancy or survival from disease, because these processes are so vitally close to what gets inherited; they determine directly who gets born and who lives. All other things being equal, genes that decrease the risk of a mother or baby dying at birth must propagate rapidly in the population.
This would be especially true historically – before medical interventions helped with difficult births. Sadly, even in the twenty-first century, around one in every hundred mothers dies in childbirth in countries where medical provision is poor.49 This gives an estimate of the minimum frequency by which mothers die naturally during or shortly after childbirth. It indicates how strongly genes would be favoured if they could protect – even slightly – against maternal mortality, including those that protect against eclampsia.
Similarly, genes that can provide protection against infectious disease – especially against an illness which can be fatal before having children – must also propagate rapidly in the population (all else being equal). For as long as that disease was prevalent, such a gene or set of genes would rapidly increase in frequency in subsequent generations. Even protection against diseases that are not fatal can still impact the success of one’s children and hence be selected for, through the generations. So variation in immune-system genes across all humanity is certain to be affected by their role in both reproduction and survival from disease.
This plays out as follows: some combinations of compatibility genes will be especially protective against a particular disease, and those versions propagate in the population. But successful pregnancy will have other requirements for variations of these very same genes. Versions of compatibility genes – and other immune-system genes – that favour successful reproduction will also be favoured in subsequent generations. These two pressures on the same sets of genes leads to a balance in what gets selected overall: a balance between versions of these genes that help us survive disease and those that help in pregnancy. In short, the outcome is to keep these genes diverse.
Despite this leap in understanding human nature, Medawar’s paradox remains unsolved: we actually still don’t fully understand how a baby is protected from the mother’s immune system. But, by trying to find the answer, we know that uterine immune cells can help – not hinder – pregnancy. Many genes that regulate pregnancy and birth do not vary much between us. Yet the most variable of all our genes help construct this most intimate of contacts between people.
