THE JOURNAL • PERSPECTIVE • www.fasebj.org The Path to New Therapies for Schizophrenia and Bipolar Illness Edward M. Scolnick1 Stanley Center for Psychiatric Research, Broad Institute, Cambridge, Massachusetts, USA ABSTRACT: Schizophrenia and bipolar illness are two of the most serious forms of mental illness. Until relatively recently, almost nothing was known about the molecular pathogenesis of either illness. The single largest risk factor that predisposes people to schizophrenia or bipolar illness is genetic risk. Heritability is high, and the incidence is significantly higher in identical twins than in nonidentical twins. Despite decades of work aimed at identifying the genes involved in these two illnesses, virtually no progress had been made until the past decade. With the knowledge and technologies that have been gained from the Human Genome Project, it has been possible to begin to understand the underlying genetics and to use the new information to begin the effort to discover new and better medicines to treat these illnesses. This article will describe the past decade of work toward this goal and articulate both the promise that now exists and what is still needed to bring dramatic and tangible change to patients.—Scolnick, E. M. The path to new therapies for schizophrenia and bipolar illness. FASEB J. 31, 1254–1259 (2017). www.fasebj.org KEY WORDS: genetics • pharmacology • progress In the fall of 2004, I became an associate member of the bipolar illness for many years. Virtually no reproducible Broad Institute, founded on the principle of collaboration results had been achieved. The Broad Institute had a su- among scientists at the Massachusetts Institute of Tech- perb group of scientific experts in human genetics, a field nology, Harvard, and Harvard-affiliated hospitals. The in which I had no particular training. David Altshuler, Broad Institute had been created in 2003 as an outgrowth of Mark Daly, and Shaun Purcell were mentors to the pro- the Genome Center at the Whitehead Institute. The Genome gram and were essential if the new effort was going to Center was overseen by Eric Lander, who soon became succeed. director of the new Broad Institute. I had retired from my A few important facts were known about these two positionaspresidentofresearchatMerckResearchLabo- psychotic illnesses, and several epidemiologic observa- ratories at the end of 2002 and then from the company in tions had been made about variables that conferred risk. September 2004. The Broad Institute was interested in Babies born to mothers in populations that had endured expanding its program on the genetics of psychotic illnesses, famine had a higher incidence of schizophrenia (1, 2). Ep- and I was fortunate to meet and have a partnership with idemiology had determined that moving from one country Pamela Sklar, M.D., Ph.D., assistant professor of psychiatry to another also increased risk. But, the single largest risk in the Partners Psychiatry Department, as well as an asso- factor for a person to become ill with one of these illnesses ciate member of the Broad Institute. Sklar had been study- was genetic risk. If a family member had either schizo- ing the genetics of bipolar illness, and without her help, the phrenia or bipolar illness, the first-degree relatives of the new program would not have become a reality. proband (patient) had a 7- to 10-fold increased risk of having one of these illnesses compared with the average person not related to the patient. This increased risk was EPIDEMIOLOGY OF SCHIZOPHRENIA true for both schizophrenia and bipolar illness (3). In addition, if a person had schizophrenia, the first- Research in the field of psychiatry had been trying to gain degree relatives also had an increased risk of bipolar ill- traction in deciphering the genetics of schizophrenia and ness. Likewise, if a person had bipolar illness, the person’s first-degree relatives had an increased risk of schizophre- ABBREVIATIONS: D2, dopamine 2; FMRP, fragile X mental retardation 1 nia. The reciprocal overlap in risk was documented in protein; gnomAD, Genome Aggregation Database; GWA, genome- wide association; LOF, loss of function; MHC, major histocompati- an elegant epidemiologic study by investigators at the bility complex; SNP, single-nucleotide polymorphism Karolinska Institute (Stockholm, Sweden) (4). Despite the 1 Correspondence: Stanley Center for Psychiatric Research, Broad In- fact that the illnesses run in families, many genetic linkage stitute, 75 Ames St., Cambridge, MA 02142, USA. E‐mail: scolnick@ studies in extended families had failed to yield consistent broadinstitute.org results at finding any genes that predisposed to either doi: 10.1096/fj.201700028 illness. Such studies usually involved relatively small 1254 0892-6638/17/0031-1254 © FASEB Downloaded from www.fasebj.org to IP 18.9.61.111. The FASEB Journal Vol.31, No.4 , pp:1254-1259, June, 2017 samples of cases. The results were puzzling, because the collaborators in this initial effort were Dr. Patrick Sullivan heritability of each disease is in the range of 50–80%. (University of North Carolina Medical School, Chapel Hill, However, a possible interpretation was that many genes NC,USA)andDr.MichaelO’Donovan (Cardiff Univer- were involved and that no single gene had a strong sity, Cardiff, United Kingdom). Sullivan’smajorsourceof enough effect to allow that gene to be detected in linkage cases and controls was from a collection in Sweden with studies. collaborators at the Karolinska Institute, with the guidance I had no training in human population genetics, but I of Dr. Christina Hultman. The collection from Sweden was learned from my colleagues, Altshuler, Daly, and Purcell, based on identifying thousands of cases, mainly from that new methods were being developed to study the ge- hospital case records. Validation of this rather simple netic underpinnings of many human diseases that had methodological approach was achieved by detailed study complex genetics (i.e., where many genes, each with a of a small number of cases. small effect size contributed to risk for an illness, such We fully recognized that the identification of cases was illnesses had a polygenic rather than a monogenic risk). based on a collection of symptoms that defined schizo- The methods were an outgrowth of the sequencing of the phrenia and not on a chemical, biological, or physical test human genome project and an ongoing effort to provide a insuchpatients.Itwasnotatallcertainthattherewouldbe detailed map of the human genome. The map was being sufficient homogeneity in the cases, using this approach, to constructed by defining and mapping so-called common allow a signal to be detected in an association study. This single-nucleotide polymorphisms (SNPs), that is, single- limitation was part of the broader problem in studying base changes differing among the human population but schizophrenia and bipolar illness, since the diagnosis of shared by a significant percentage of the human pop- either disease was solely based on symptomatic criteria. ulation. Although there is no hard and fast definition of We began the effort mainly on patients with schizophre- common SNPs, it is generally taken to mean those shared nia, as more cases were available, and initiated efforts to by 5% or more of the population. collect sufficient cases of bipolar illness while the initial Why was this mapping important as a tool to study studies were conducted. the underlying polygenic architecture of many human Because of the generous funding from the Stanley diseases? Earlier work by Stacey Gabriel, Altshuler, and Medical Research Institute, in 2007, we were able to begin Daly had discovered that large blocks of genes in the in earnest the genetic studies. For 2–3 yr as the first data human genome were in linkage disequilibrium (i.e., became available, we realized that this was an even more their linkage to one another was only rarely interrupted formidable problem than we had thought, and we were by recombination) (5). Thus, by mapping these common not certain that the approach was going to yield positive variant SNPs and knowing they provided signposts for data. We were buoyed by calculations made by Daly, who other genes in these relatively stable segments, one showed in several ongoing association studies in other could then perform a different kind of genetic study— diseases, such as Crohn’s disease, type 2 diabetes, and an association study instead of a linkage study. In an multiple sclerosis, that the yield of clear associations was a association study, one could look for the frequency of function of the sample size. In each of those studies, his association of marker SNPs in control populations analysis suggested that if we persisted and analyzed without the disease of interest versus the frequency of enough samples, then positive results would emerge. This associationofsuchmarkerSNPsintheDNAofthe was a high-risk strategy; however, it was clear that unless population with a given illness. we could begin to identify unambiguously the genes that Although it was not certain that such an approach put people at risk for these diseases, there would not be would yield positive results for schizophrenia or bipolar tangible progress in the field. It was always obvious that it illness, it was clear that this was a new approach worth was impossible to carry out biochemistry on living brain trying. Statistical estimates suggested that there was an tissue, imaging studies did not have the molecular reso- 80% chance that real associations could be found. The lution to unravel the abnormal biochemistry, and chemical discovery of the multiple genes that were the basis for measurement of peripheral body fluids or tissues would many common human diseases was a rapidly emerging not give information about abnormal brain biochemistry; field, and association studies were begun, not only in thus, we had two choices: we could give up our effort, or schizophrenia and bipolar illness but also in diabetes, we could put our faith in Daly’s calculations and persist in macular degeneration, multiple sclerosis, Crohn’sdisease, the effort.