Q&A With Paul Patterson. Questions by Victoria L. Wilcox
Q: As you think about the research looking for biological mechanisms that could transduce maternal infection into risks for schizophrenia, what do you think are the major insights of the field so far?
A: In terms of the mechanism of how maternal infection alters fetal brain development, the first thing to say is that it's not caused by infection of the fetus, but rather it's caused by the activation of the mother's immune system. In their recent paper (Moreno et al., 2011), Moreno and colleagues provide some evidence in favor of that—that is, they don’t see evidence of the virus in the fetus, but that's been published previously by us (Shi et al., 2005). I think the most compelling evidence, though, is that you can get very similar, maybe the same, behavioral abnormalities in the offspring without using the virus at all; you can just activate the mother's immune system. And we do that with double-stranded RNA, poly I:C, without a pathogen.
It should be noted that the behavioral abnormalities are consistent with those seen in schizophrenia, but many of them are also consistent with symptoms in autism. This is not surprising, because maternal infection is a risk factor for both schizophrenia and autism. There was an important study in 2010 that analyzed over 10,000 cases of autism in the Danish medical registry and found an association with viral infection in the first trimester (Atladóttir et al., 2010). This is similar to findings by Alan Brown and colleagues in schizophrenia, where they found the risk factor for infection was likely to be in the late first, early second trimester (Brown et al., 2004).
In the schizophrenia epidemiology, the infections can be of almost any kind; there's evidence for viral, bacterial, and parasitic toxoplasma infections as increasing risk. This is also consistent with the idea that it's the mother's immune response that's important, not the particular kind of infection.
Q: What are some of the challenges to this field? Can they be overcome and, if so, how?
A: On the behavioral side, of course, you're dealing with mice or rats, which have only analogous behaviors to those in humans. In that sense, the behaviors that we and others look at are consistent with those in human schizophrenia and autism, but, of course, these are rodent behaviors. In that sense, Moreno et al. provide a new piece of data that is, I think, very important, which is they provide evidence for what I call the rodent equivalent of hallucinations. They see activation of the sensory cortex, in the absence of sensory input, which can also be evoked by injecting a hallucinatory drug. The surrogate marker for neuronal activation that they use is early gene activation.
The hallucinogenic drug induces these genes in the cortex, but the key point is that there is a stronger induction in the offspring of mothers who were infected compared to controls. This is what is seen in schizophrenia: schizophrenics are more sensitive than controls to hallucinogens. Moreno further reports that the behavioral response to hallucinogens is also greater in the offspring of infected mothers. Our group has unpublished data very similar to these using the poly I:C maternal immune activation model.
A key point about this rodent version of hallucinations is that it is a behavior that one can measure in rodents that is more specific to schizophrenia than other behaviors that we and others have measured. Hallucinations can also be found in bipolar disorder, but they are very prominent in schizophrenia. This is a so-called positive symptom, rather than a negative symptom, which is what everybody is assaying normally.
Q: I'm glad you mentioned poly I:C because that leads into my next question. The Moreno study uses “mousified” influenza virus, whereas a lot of the studies seem to use poly I:C to simulate the flu. What are the pros and cons of the different approaches?
A: The influenza model is more like the human situation, where one is studying an actual infection of the mother. The poly I:C approach has the advantage that one can control the timing of the mother’s immune response very precisely, whereas in the infection, the mother is sick for many days after infection—in other words, for a good part of embryonic development. In contrast, the poly I:C drives up the cytokines in the mother very transiently, so one is looking at a very narrow time window of the effect.
Also, one can control the dose very closely with poly I:C, while it is harder to control the dose with infection of the virus, which is dependent on the breathing pattern of the animal. One can use different concentrations of the virus, but it is never clear what dose the animal is actually absorbing in the respiratory tract. We've done infection with different doses of influenza virus, and we do see a dose-response curve in the behavior of the offspring, but it's difficult to specify exactly what dose the mouse is getting. Nonetheless, the cytokine response to poly I:C is certainly not identical to influenza infection.
Q: What stands out to you about the results of the Moreno study?
A: I mentioned that, to my mind, the most interesting result is the evidence that the offspring of infected mothers are more sensitive to the hallucinogen. They also report that there are changes in the serotonin and glutamate receptors in the brains of the offspring, which is useful, and adds to the neuropathological findings of others in this model.
Q: What do the results imply about the role of glutamate and serotonin in the relationship between the infection in the mother and schizophrenia in her offspring, and is that important in terms of the neuropathology?
A: Moreno et al. make the point that they think the change in these receptors might be related to the behavioral changes that are observed in the offspring, which could be true. One would have to, of course, investigate that in much more detail to see if there's really a relationship between changes in, say, a glutamate receptor and the behaviors.
Q: Are there any limitations or caveats about the study that are important to keep in mind?
A: I don't think it's a terribly important part of the paper, but the negative result that they do not detect the virus is not as compelling as the positive result that similar behaviors can be observed in the offspring without using a virus; one can just stimulate the immune system in the mother.
Q: This is a mouse model, and some people are skeptical of them. Obviously, there are differences and similarities between mice and humans, so what, if anything, does this study based on a mouse model mean for humans with schizophrenia? What's needed to connect those dots?
A: A key advantage is that, in the rodents, one can study the mechanism of how maternal infection alters fetal brain development—this can't be done in humans.
Q: You can't just infect pregnant women with a virus, for instance.
A: What's known so far is that it's the mother's immune response that's important. Second, we found that the rise in the cytokine interleukin-6 that occurs during infection is critical for the changes in behavior of the offspring. If one blocks the interleukin-6 response, you don't see the abnormal behaviors in the offspring. Conversely, if one simply injects IL-6 alone in pregnant mice, the abnormal behaviors are found in the offspring, so you don't need poly I:C or the virus; transiently increasing this cytokine is sufficient.
Another point about the mechanism is that the placenta is at least one of the sites of action of the maternal immune response (Hsiao and Patterson, 2010). Cytokines are upregulated in the placenta, there are endogenous immune cells in the placenta that are activated, and this results in endocrine changes in the placenta that undoubtedly affect the fetus. In addition, maternal poly I:C administration (or maternal LPS, to mimic bacterial infection) increases cytokine levels in the fetal brain, and these changes are prolonged until at least the early postnatal period (see Patterson, 2009). In other words, the first steps are being taken toward understanding the cellular and molecular pathways of how maternal infection alters fetal brain development, which ultimately changes behavior.
That's one point: the mechanism. Another feature of animal work is the ease of testing potential therapeutics. For example, the Meyer group in Zurich and the Weiner group in Tel Aviv recently published important papers on one type of intervention, which is treating adolescent offspring of immune-activated mothers with antipsychotic drugs before they exhibit certain behavioral abnormalities and neuropathology (Meyer et al., 2010; Piontkewitz et al., 2010). This prevents the onset of at least some of the abnormal behaviors and neuropathology, in particular, the enlargement of the ventricles, which is a cardinal neuropathology in schizophrenia. This, I think, adds fuel to the fire concerning the question of prodromal treatment of teenagers at extremely high risk of schizophrenia. Many people argue that such treatment is ineffective or perhaps even unethical, but the rodent results, both in rats and mice, are very compelling that adolescent treatment can prevent abnormal behavior and pathology in the offspring.
Such results also suggest that other novel treatments could be tested in these offspring because, clearly, not all the abnormal behaviors are permanently set during embryonic development. One might have predicted that since fetal brain development is altered, it's too late to do anything postnatally. These results indicate otherwise.
Q: Are there any other potential approaches to treatment or some kind of preventive intervention that could occur prior to the prodrome, such as vaccinating the mothers?
A: We're particularly interested in modulating the immune system of the offspring in the early postnatal period. That's another place that is potentially important, particularly in autism, because you would like to intervene in autism as early as possible. Since the offspring of infected or immune-challenged mothers display the cardinal symptoms of autism, which we haven't talked about yet, you could test early interventions. These symptoms are 1) repetitive or stereotyped behaviors; 2) deficits in communication, which in the animals is tested by recording ultrasonic vocalizations; and 3) deficits in social interaction. These behaviors can be readily assayed in the rodents and have parallels in human autism. There's also a neuropathology that's commonly found in autism, which is a localized deficit in Purkinje cells.
Regarding maternal vaccination, one issue arises from the CDC recommendation that all pregnant women be vaccinated for the flu. The rodent results raise the question: Is this really safe? Vaccination activates the immune system, and that's what poly I:C does. In my mind, it raises the quantitative question: Is vaccination strong enough to activate the mother's immune system to the extent that it could alter fetal brain development as we see in the mice with poly I:C or IL-6 injection?
Q: That's a sobering thought.
A: And this hasn't been clarified. In fact, a review by the Canadian CDC flu experts (Skowronski and De Serres, 2009) concludes that it is not yet clear that maternal vaccination is both safe and effective. It seems that more studies need to be done. In fact, the studies that are quoted by the U.S. CDC regarding maternal vaccination did not look for schizophrenia in the offspring, because, of course, that would involve following people for 20 or more years.
Q: Interesting. Is there anything else you want to say regarding what are the biggest unanswered questions that remain in the field?
A: In terms of the mechanism of how maternal infection changes fetal brain development, we're only taking the very first small steps to understanding that. We don't know what the cytokines do in the fetal brain; we're not exactly sure where they're acting, or which are the key changes in brain development that lead to abnormal behaviors. There are many, many steps between changes in fetal brain development and actually seeing how that plays out in behavior. In fact, that will be very difficult to do because the behaviors that we're looking at are complicated. There is also a lot to be done in terms of testing potential new treatments for both the autism- and the schizophrenia-like behaviors and pathology.
Q: Do you think the findings from the Moreno study and other studies exploring this area could lead to a treatment?
A: There's nothing in the Moreno study that speaks to that question directly, but the results on the antipsychotic drugs show that you can test the drugs, and you can get effects, so the field is wide open for that.
Q: Is there anything I haven't asked you about that you would like to say about this topic, about research in this area, or about the Moreno study in particular?
A: Two points: First, I think everybody recognizes the importance of genes in schizophrenia and autism, but the animal studies that we're talking about so far don't really speak to the question of genetic risk. Therefore, another area of future work that's very important is to combine the susceptibility genes with the environmental risk factors, such as maternal infection. There have been a couple of studies on this published so far, but there's much more that can be done. People have looked at maternal immune activation combined with the DISC1 gene knockout (Abazyan et al., 2010). There's also a recent paper from Dan Ehninger and Alcino Silva on the tuberous sclerosis mouse model (Ehninger et al., 2010). Mutations in the tuberous sclerosis genes can enhance the chance of developing autism features. This paper shows a synergism between the poly I:C environmental factor and the tuberous sclerosis mutation.
Q: So there could be some gene-environment interactions involving maternal infection and perhaps some gene that makes humans or animals more vulnerable.
A: Right. This hasn't been studied in humans yet—that is to say, people who are doing genetic studies in humans haven't been looking at the combined effects with environmental risk factors. That may be part of the puzzle for why the effect sizes of most mutations are so small, because maybe the mutations have to occur in combination with an environmental risk factor.
Q: That makes sense to me.
A: That's point number one: the genetics of it. Point number two: it's frustrating that so much less attention is paid to environmental risk factors compared to genetic risk factors, because the effect size of maternal infection is so great. I think the epidemiology deserves a lot more attention than it has been getting.
Q: Do you think maternal infection could be related to some of the epidemiological factors that have been found to be related to schizophrenia, such as social density?
A: Yes, the point has been made that a number of other risk factors found by epidemiology, which are seemingly so disparate and unrelated, could in theory all be related to this maternal infection. That would include being born and raised in an urban environment and being born in winter-spring months, both of which would enhance the probability of maternal infection. Also, obstetric complications and maternal stress increase the risk for schizophrenia, and these involve increases in IL-6 levels, which the mouse studies showed to be critical for schizophrenia-like behaviors in the offspring.
Q: In addition, certain immigrant communities face a higher risk.
A: It's conceivable that when people are transplanted into a different environment, they could be more susceptible to infection by microbes they have not encountered before. Obviously, there are many other factors that are new to the immigrant as well, but it is interesting to think about unifying hypotheses.
Q: Thank you so much for taking the time to talk with me; I really appreciate it.