Researchers at Johns Hopkins Medicine, Baltimore, Maryland, have identified a genetic variation that may influence cognition and IQ, and the finding could advance the development of treatments focused on cognition in people with schizophrenia or other serious mental illness. In this video, clinical pharmacologist Kristin Bigos, PhD, assistant professor of medicine at the Johns Hopkins University School of Medicine, discusses what led up to the research, how the findings were reached, and further studies being conducted.
Read the transcript:
Part 1: Leading up to This Research
It's long been known that glutamate is involved in cognition and memory. We've also long known that the glutamate system is really important in the development of schizophrenia and healthy brain development.
About a decade ago, one of the first whole‑genome‑wide analyses of schizophrenia was published by Pharmacogenomics Consortium. One of the top genes associated with risk for schizophrenia was the gene GRM3, which codes for the metabotropic glutamate receptor 3.
That led a number of investigators and drug companies to develop agonists for mGluR3. At the time, it was hard to find things that were specific for mGluR3, so the agonists that were being tested were substrates for both mGluR3 and a similar metabotropic receptor, mGluR2.
The problem is that mGluR2 and mGluR3 are not exactly the same, both in terms of their pharmacology and physiology, but also, where they're located in the synapse and how they work. mGluR3 is primarily presynaptic, although it is also postsynaptic and also on astrocytes. It's thought to decrease glutamate transmission, in general.
Those mGluR2/3 drugs ultimately failed in clinical trials. The one that got the furthest was a drug, pomaglumetad, which was developed by Eli Lilly. They conducted a phase 3 trial of that drug and looked at total symptoms of schizophrenia, and the drug did not improve total symptoms of schizophrenia.
Around that time, colleagues of mine were developing drugs to target specifically the mGluR3 pathway, and they did so by developing drugs that bound to an enzyme called GCPII, glutamate carboxypeptidase II.
GCPII regulates the amount of a chemical called NAAG, N‑acetylaspartylglutamate. NAAG is the endogenous substrate of mGluR3. It was thought that if we could alter NAAG amounts in the brain, that we might be able to specifically target mGluR3.
The problem with those drugs is that they looked a lot like glutamate and had trouble, because of that, getting into the brain. That had been worked on and somewhat shelved for a long time, but this mGluR3 pathway and the GRM3 gene kept coming back.
It's now been associated with hundreds of thousands of people in a genome‑wide association study in the follow‑up study of PGC2 and now PGC3, and so we kept coming back to—we thought that this could be an important mechanism for targeting symptoms of schizophrenia, if we could target GCPII, if we could get drugs into the brain that altered GCPII.
We really wanted to know if you altered GCPII, what would happen. That's really ultimately what led us to start to study NAAG and the gene FOLH1, folate hydrolase 1, which codes for GCPII, glutamate carboxypeptidase II.
One last thing is that around the same time, a few years ago, the GRM3 gene was also associated with cognition statistically in a GWAS study of human cognition. And so we thought, well maybe it's not about schizophrenia symptoms in general, maybe it's more about cognition. And so if we look specifically at measures of cognition in schizophrenia, that might tell us a little bit more about how altering NAAG via GCP II inhibition may improve schizophrenia symptoms.
Part 2: Study Methods (4:23)
We did a series of studies over a number of years on a bunch of different patient populations, postmortem brain samples, looking at genetics, all sorts of things, to try to get at the sense of how could we measure changes in GCPII without having a drug that got into the brain, but changed GCP II, ultimately, altering NAAG levels.
We used something called imaging genetics, that my lab has worked on for a number of years, where you can map different genetic variants to brain circuitries related to mental illness and specific domains, in this case, cognition.
We looked for variants in the folate hydrolase 1, FOLH1 gene, that were associated with differential expression of FOLH1 in the human brain. This was a postmortem brain sample that is at the Lieber Institute for Brain Development.
We looked at several hundred normal, control brain samples and also a few hundred patients with schizophrenia. We looked at mRNA expression in the prefrontal cortex of these brains and found that there is missense mutation in exon‑2 of FOLH1 that's associated with increased mRNA expression in the DLPFC.
We then used that SNP in further studies to understand how it worked in living, behaving humans. We took that SNP that was associated with high FOLH1 levels and we measured that genotype in patients with recent onset psychosis and also healthy volunteers.
We measured NAAG levels in human brains using magnetic resonance spectroscopy. We did this at a high‑field strength, 7 Tesla MRI, which gives us a better resolution of separating NAAG from NAA. GCPII or FOLH1 cleaves glutamate from NAAG into NAA, and that's how it regulates the amount of NAAG.
We looked and found that both patients with psychosis and healthy subjects who carry this FOLH1 missense mutation had less NAAG levels in their brains. They have more FOLH1, so it was cleaving more NAAG into NAA and glutamate.
We then were interested in what was different about the people who carried the variant and didn't. We looked at measures of cognition and we found that both healthy subjects and those with psychosis who carried this particular genetic variant had a lower IQ score.
They also had a lower composite score of looking at different cognitive domains and combining those to a composite, overall cognition. Because we had NAAG levels, we didn't need to use the genetics as a surrogate for this.
We looked directly at correlating NAAG levels with different cognitive domains and found that those patients and healthy subjects who had higher NAAG levels performed better on tasks of visual memory.
We also looked at the ability of differences in NAAG levels or genetic variation associated with NAAG levels specifically on measures of brain activity in functional MRI. Again, going back to the imaging genetics, we can map genetic variation to alterations in functional brain activity during a task.
In this study, we specifically looked at the N‑back working memory task, which has been long studied in schizophrenia, but also healthy volunteers. In this study we just measured brain activity during this N‑back working memory task in healthy volunteers.
We did that, in part, because we wanted to look at people who had high levels of performance and then be able to compare people who all had high levels of performance—in this case, our cut‑off was above 80% correct—to be able to compare between different genotypes.
Basically, what we have the subjects do is they have to remember a series of numbers. In a zero‑back condition, they simply press the number of the button on the screen.
In the two‑back condition, they have to remember two numbers back. They see the first number, they wait. They see the second number, they wait. On the third number, they push the number that corresponds with the first number.
We compare brain activity during the two‑back compared to the zero‑back, and we can isolate the working memory network. You can see prefrontal cortical activity and parietal cortical activity, and also, some hippocampal activity.
What we do is we measure this BOLD activity during fMRI while the subjects are performing this two‑back and zero‑back working memory task. And then we measured, as I said before, this folate hydrolase 1, missense mutation, and we compare the carriers who have the mutation versus the carriers that don't.
What we found was that the carriers of the missense mutation that's associated with higher FOLH1 levels in human brain and lower NAAG levels in human brain have increased parietal cortical activity during the working memory task.
We interpret this as being less efficient or having cortical inefficiency. It's what we see when we compare patients with schizophrenia to healthy volunteers, and even siblings of patients with schizophrenia to healthy volunteers.
This was interesting because it provides potentially a mechanism for how NAAG is involved in improving cognition or in cognition at all. It suggests that it may be, in part, increasing the efficiency of the cortex while doing something like a working memory task.
Part 3: Impact on Future Treatment (10:48)
As I mentioned, there has been lots of interest in developing GCPII inhibitors to increase NAAG levels in human brain.
Doing the series of studies that we did suggested that having more NAAG in human brain would produce better cognitive functioning. Specifically, in this study, in patients with psychosis and schizophrenia. But also, the relationships all held in healthy volunteers. It may be that NAAG is changing cognition or improving cognition, or could improve cognition in other diagnoses that have cognitive impairment.
There are now efforts to understand the association between NAAG levels and overall cognition in specific cognitive domains in other disorders, such as ADHD or mild cognitive impairment or cognitive aging.
There's also been increased efforts in developing GCP I inhibitors, again, drugs that might be able to get into the brain better.
Part 4: Future Research (11:52)
We are conducting more research. As I said, we are starting some studies measuring NAAG levels in other patient populations to see if they are lower, and if they are in general associated with specific cognitive domains or just general cognition.
We are also testing new GCPII inhibitors different ways. One way we're trying to improve their brain penetration is using intranasal formulations. To do so, we are looking at primate models. We're starting studies looking at brain penetration of GCPII inhibitors and its ability to increase NAAG levels by measuring NAAG levels in MRI.
I'll just add that one of the reasons we've spent so much time looking at different cognitive domains and different brain circuitries related to mental illness is to develop biomarkers for future trials.
It may be that one of the reasons that previous drugs have failed is that they looked at total symptoms of schizophrenia. While we're not sure, because we haven't measured anything like this before in humans, it may be that these drugs specifically improve cognition and not just total symptoms of schizophrenia.
It may be that they only improve certain cognitive domains, and if we don't measure those domains in a certain way, we may miss effects that could improve their lives.
The other reason that we've looked at genetics in different patient populations was to try to select the best populations to move forward in the first‑in‑humans clinical trials of the GCPII inhibitors. Both from a benefit standpoint, we want to pick patients—both population, or people who have low NAAG levels, or people who have genetic variations associated with low NAAG levels—who would be most likely to benefit and most likely to respond in clinical trials.
Zink CF, Barker PB, Sawa A, et al. Association of missense mutation in FOLH1 with decreased NAAG levels and impaired working memory circuitry and cognition. The American Journal of Psychiatry. 2020 December 1;[Epub ahead of print].
Kristin L. Bigos, PhD, is Assistant Professor, Department of Medicine, Department of Psychiatry and Behavioral Sciences, and Department of Pharmacology and Molecular Sciences at Johns Hopkins School of Medicine in Baltimore, Maryland. Her research is focused on the development of neuropsychiatric drugs and using precision medicine in the treatment of mental illness. She was previously an Investigator at the Lieber Institute for Brain Development in Baltimore.