If we are to understand autism and eventually have effective treatments, preventions and cures, scientists must discover the genes involved. In a remarkable six-month period, we have seen the publication of two NAAR-funded studies that have identified "candidate" genes for autism. Significantly, both of these genes appear to have functions that fit into the known neurobiology of autism. NAAR is delighted to report on these important research findings by 1999 NAAR Awardees, Drs. Flavio Keller and Christopher Stodgell and their respective colleagues.
Of all the "genetically complex" diseases that are being studied, autism has the highest "heritability" (that is, it is the most genetic). The most commonly talked about method of searching for these genes is the genome screen or "linkage studies". This method has been used to find genes in other brain diseases such as Alzheimer's disease and Huntington's Chorea and has been described as the "workhorse" of human genetic mapping. Unfortunately, for most of the complex brain disorders such as bipolar disorder, Tourette's disorder and autism, genes that appear to be associated with these diseases have not yet been identified. The reasons for this are not entirely clear. Risch and his colleagues at Stanford speculate that autism has 15 or more genes involved. The search goes on and new methods and types of analyses are underway.The geneticists nevertheless remain optimistic that genes will be discovered for autism due to the high heritability of the disease.
Here is the story of two research groups who have found genes that are associated with autism. One group came to this finding while working on the causes of elevated serotonin blood levels in autism, while the other has been proceeding for several years on a measured and evidence-based path towards the gene they are studying.
The Reelin Gene
Many parents of children with autism have heard of the work of Dr. Karl Reichelt from Oslo, Norway. For over twenty years he has been publishing his work on abnormal peptides found in the urine of autistic children. He has speculated that some of these could be due to the inadequate breakdown of some "exogenous" substances (that is, substances which we take into the body such as foods). He specifically has speculated that gluten and casein may be involved which has, in turn, led many parents to alter their children's diets. In a 1999 publication, Reichelt and his colleagues reported a significantly high blood level of a particular tripeptide, PyroGlu-Trp-GlyNH2, in individuals with autism. This tripeptide is three amino acids: glutamic acid, tryptophan, and glycine. (Amino acids are the building blocks which, when put together, form peptides. Long chains of peptides are known as proteins.)
This peptide is particularly interesting because it was shown to be a potent stimulator of serotonin uptake into the platelets. Many individuals with autism have abnormally high serotonin in the blood. A collaborator of Dr. Reichelt's research team, Dr. Flavio Keller from the Campus Bio-Medico University in Rome, Italy, submitted a proposal to NAAR to replicate the work of the Norwegian group and look for that particular tripeptide in the plasma of normal and autistic individuals. After finding it he would then correlate levels of the tripeptide with the serotonin blood level in individuals with autism and their first- degree relatives. Dr. Keller's goal was to look for a biologic marker for autism and perhaps a clue to a candidate gene.
What happened next was one of those findings that seems to be so common in scientific research. Dr. Keller and his research group could not find this tripeptide in their Italian subjects. Thinking perhaps that the tripeptide was particular to the Scandinavian population, Dr. Keller asked Dr. Reichelt to send him specimens from his Norwegian subjects. Two different laboratories in Italy were each again unable to find this tripeptide. Dr. Keller was unable to replicate Dr. Reichelt's work. At this point the study might have ended as being merely a negative finding.
Fortuitously, however, while working on this research, the Italian group was also pondering possible sources of this tripeptide. They searched databases of proteins and realized that there could be thousands of sources. They also added some other peptides that Dr. Reichelt had found (though unpublished). When they combined these peptides there was only one protein which could be responsible for creating these peptides and that was the protein reelin.
Although, to some extent, reelin came "out of thin air", it was an intriguing lead. Reelin is a "signaling protein" which plays a pivotal part in the migration of several types of neurons and in the development of neuronal connections. There had also been several recent scientific reports implicating reelin in schizophrenia and more specifically in information processing in schizophrenic patients. There is a mouse model (the "reeler mouse") that has a deletion of the reelin gene and, interestingly, this animal model has much in common with autism. Both the reeler mouse and individuals with autism have been shown to have decreased purkinje cell numbers in the cerebellum as well as dysplasia of the dentate nucleus with reduced cell counts in adults. In the brain stem, both have dyspastic inferior olives and cytoarchitectonic alterations in the facial nucleus. There are also abnormalities in the hippocampus, entorhinal cortex and amygdala. The most characteristic feature of the reeler mouse is inverted cortical laminations, which means that the layers of cells in the cortex are layered upside down. This is not present in autism although some very recent work in autism is finding abnormal pathology consistent with abnormal migration and layering in the cortex of autistic brains. A further "coincidence" is that the reelin gene is located on chromosome 7q22--a region in which four genetics teams have each found to be an area of interest on the genome and therefore possibly involved in autism, despite not reaching statistical significance. If the reelin gene were found to be abnormal, it could be contributing to the linkage findings even if not strongly enough to emerge as a singular cause of autism.
Dr. Keller and his team decided to study the gene in two ways. One was through a case control association study where Dr. Keller and his colleagues compared individuals with autism to a control group. Although this method can be a powerful tool in studying genes, one of the hazards is that the investigator must be careful in selecting an appropriate control group. For example, many genes are more common in certain ethnic groups and so factors such as that must be taken into consideration.
What they found was an area on the reelin gene with a variable number of trinucleotide repeats. (This is the type of gene abnormality found in Fragile X Syndrome in which parts of the gene repeat over and over like a skipping record.) This type of variation in the sequence of a gene is called a "polymorphism". A polymorphism does not necessarily signify a disease, but it can serve as a marker to check whether a particular gene is associated with the disease.
Dr. Keller then compared the polymorphism that they found in the reelin gene in two groups. One group was comprised of 95 individuals with autism of Italian descent and the other was comprised of 186 unaffected individuals of the same racial and ethnic background. They found that 17.9% of the autistic group had at least one allele with more than 11 trinucleotide repeats (a "long allele") compared to 7.1% of the controls. This difference is statistically significant.
The second type of study that Dr. Keller and his colleagues undertook was a transmission disequalibrium test. In this type of study, they looked at allele transmission from parent to child within families. Because they are studying the transmission from parent to child, the control issue is not the same as in the case control association study. As such, they were able to add 89 American individuals with autism for a total of 172 trios (subject and two parents) and 395 first- degree relatives. What they found was that the transmission of the long alleles was significantly more common in the affected child than in the unaffected child.
How could transmission of a long allele of reelin lead to autistic disorder? An answer may come from Fragile X Syndrome. In Fragile X Syndrome, stretches of repeated trinucleotides are present in a region of the gene called the promoter region. This region does not code for the protein itself, but its function is to control how much of the protein is produced. If too many trinucleotide repeats are present, then the promoter does not work properly and less protein is produced. In Fragile X, an individual is a "carrier" if there is a certain number of repeats; if the number of repeats is even higher than this threshold number, then the person manifests the disease at a level proportionate to the number of repeats.
As in all initial scientific findings, this report should be considered preliminary and must be replicated by other, independent research groups. It is, however, encouraging that a research team from Queens University in Ontario, Canada (Zhang et al) presented preliminary findings at a recent scientific meeting noting a finding of the trinucleotide repeat in the same region of the reelin gene in autistic patients. And another research group from the University of Minnesota (Fatemi et al) has measured reelin levels in postmortem samples of autistic cerebellar tissue and found a 43% reduction compared to controls. Dr. Keller's findings therefore offer the promise of a new avenue for research efforts and the possibility of identifying a candidate gene in autism.
HOXA 1 Gene
A few months before Dr. Keller's publication in the March 2001 issue of Molecular Psychiatry, we had our first wave of excitement when the autism research team at the University of Rochester reported a significant linkage with another gene. To some extent, this was less of a surprise as the Rochester group had been working towards this for some time. In fact, the preliminary work for this finding was the science feature article of NAAR's very first NAARRATIVE in Summer 1997. This work also received a great deal of popular press when it was published as a feature article in Scientific American in February 2000. However, for those readers unfamiliar with the work of Dr. Patricia Rodier, Dr. Christopher Stodgell and their colleagues, I will briefly review some of their previous work.
While undertaking research on thalidomide survivors and, specifically, looking at an eye movement disease caused by the thalidomide exposure, researchers Miller and Stromland made an extraordinary observation. Of the 86 known thalidomide survivors willing to be examined in Sweden, their external birth defects indicated that the mothers of 15 took the medication between days 20-24 in their pregnancies. In 4 of these 15 cases, the children born had autism. Among the 71 survivors injured at other times during their gestation, there were no cases of autism. Although these numbers are small, the ratios were striking. This led Dr. Rodier, an expert in embryology, to hypothesize that, although thalidomide is not the cause of autism in the general population, these findings strongly suggest that at least a form of autism is caused by injury to the developing nervous system during days 20-24 in utero. The implication is that although we may not yet know what causes autism, we can begin by knowing when it is caused. As such, scientists can narrow down the search for causes by taking into account the timing and areas of the brain that were affected by thalidomide in this select population of individuals with autism.
Dr. Rodier and her colleagues, including Dr. Christopher Stodgell, NAAR's 1999 Roland D. Ciaranello, M.D. Memorial Fellow in Basic Research, have pursued this hypothesis in several different ways. Clinically, they have noted abnormalities in the cranial nerves of autistic children (that is, the nerves that innervate the face including the ears, mouth, eyes and facial muscles). They have also made post-mortem examination of the cranial nerve nuclei in the brainstem. An animal model has been created by using the chemical valproic acid. This is the same substance as the medication Depakote. Valproic acid is similar to thalidomide in that it interrupts neural tube closure, the biologic process in the brain that takes place between days 20 and 24 in utero. Like exposure to thalidomide, exposure to valproic acid early in pregnancy increases the rate of autism in the offspring. The rat model created has many of the same neuroanatomical brain findings as are noted in autism. This animal model has been adopted by many other research groups as the best animal model for autism at this time.
Another animal model is the "knockout" mouse that is lacking the Hoxa1 gene. This is a gene that expresses only in very early brain development. It is known to be involved in those same processes during the same period of time (days 20-24 in utero). This knockout mouse also has many similarities to the brain abnormalities found in autism. These findings led the Rochester group to investigate whether an abnormality of the HOXA1 gene was what was actually involved in the injury found in individuals with autism.
The research results of Drs. Rodier and Stodgell and their colleagues were published late last year in the journal Teratology. They found that there was a variant in the HOXA1 gene. This variation was a single base substitution of guanine for adenine in the string of histidine repeats that is thought to be critical for the function of the gene product. Although this gene is highly conserved across species (hox genes are even found in insects!), this is the first report of a variant in any mammalian species.
Unlike the Italian study, in which ethnically similar controls could be identified and compared to people with autism, the ethnic diversity in the U.S. makes direct comparisons of controls to people with autism almost impossible. Therefore, this study used unrelated controls only to determine whether the newly-discovered variant of HOXA1 is present across many ethnic groups or whether it is seen in only a few. The answer is that it is widespread in people of European, Middle Eastern, or African origin, but not in those of Asian ancestry.
To test whether the gene is related to autism, the in-vestigators studied transmission rates of the two versions of the gene to affected and unaffected children in families having a child with an autism spectrum disorder diagnosis. In a sense, this amounts to comparing the genotypes of children to those of their parents. (It is the same approach used in the Italian study when the American families were added.) The results showed a highly significant increase in the transmission of the version of the gene with the substitution to affected offspring, suggesting that it plays some role in autism. Another interesting result was that this effect was greatest when the variant was inherited from the mother of the affected child rather than from the father. This has been found to be true of some genes inherited in other developmental disabilities as well.
Put together, these observations create a dizzying array of findings that seem to suggest several possibilities but which do not provide much in the way of conclusions. As with all science, reproducibility is essential and necessary. In this case, we are fortunate that the Rochester group is part of the Collaborative Programs of Excellence in Autism (CPEAs), a network of research groups investigating autism funded by the National Institutes of Child Health and Human Development (NICHD) and Deafness and Communication Disorders (NIDCD). This network is now attempting to replicate these findings on a larger scale. We are hopeful that this study will be completed soon. With larger
numbers, some of the trends observed, such as the ethnic and gender effect, can be studied.
At this point, these investigators believe that the HOXA1 may be an autism susceptibility gene. That is, having the variant in the HOXA1 gene doesn't in itself cause autism but may make an individual more vulnerable to developing the disorder. Possibly, this could be a vulnerability to an environmental insult. Another explanation could be a vulnerability to another abnormal gene that might not itself be sufficient to cause autism. This is similar to the finding that has emerged this year in Rett syndrome. Rett syndrome researchers have discovered a "regulator" gene (that is, a gene that regulates the functioning of other genes) which is abnormal in about 80% of Rett cases. HOXA1 is also a regulator of other genes, acting during the early development of the embryo.
It is difficult at this point to assess the importance and the impact of the reelin and HOXA1 gene findings. On the one hand, they may be significant breakthroughs in autism research. Alternatively, even if replicated by other scientists, they may prove to be only incidental findings. A few things are certain, however. With such a highly heritable disease as autism, identifying the genes involved will be essential in finding causes and cures. Even incidental findings provide precious clues as to where science should focus next. As findings are added, brick by brick, we create the foundation on which answers will emerge.
The reelin gene study was coordinated by Drs. Flavio Keller and Antonio M. Persico of the University "Campus Bio-Medico” in Rome and involved a collaboration between European and American Autism Research and Clinical Centers.
References: Ingram, J.L., Stodgell, C.J., Hyman, S.L., Figlewicz, D.A., Weitkamp, L.R., Rodier, P.M. (2000) Discovery of allelic variants of HOXA1 and HOXB1: Genetic susceptibility to autism spectrum disorders. Teratology, 62, 393-405.
Persico A.M., D'Agruma L., Maiorano N., Totaro A., Militerni R., Bravaccio C., Wassink T.H., Schneider C., Melmed R., Trillo S., Montecchi F., Palermo M., Pascucci T., Puglisi-Allegra S., Reichelt K.L., Conciatori M., Marino R., Quattrocchi C.C., Baldi A., Zelante L., Gasparini P., and Keller F. (2001) Reelin gene alleles and haplotypes as a factor predisposing to autistic disorder. Mol. Psych. 6,150- 159.
Rodier, P.M. (2000) The early origins of autism. Scientific American, 282, 56-63.