UC Davis

Autism spectrum disorder (ASD) is a neurodevelopmental disorder characterized by social and communication deficits with restricted interests and repetitive behaviors that currently affect over 2% of children in the US. Autism etiology is a combination of genetic predisposition and exposure to environmental, chemical, and biological insults during gestation and during the first years of life. Particularly, maternal immune dysregulation during pregnancy can act as an insult by changing the fetal-neuro-immune homeostasis and result in neurodevelopmental disorders such as autism in genetically susceptible individuals.

Maternal Autoantibody Related Autism (MAR-ASD) is s subtype of autism in which some women develop autoantibodies against proteins necessary for healthy brain development, and during pregnancy, these pathogenic autoantibodies can traverse the developing blood-brain-barrier and cross-react with fetal brain proteins in neuroprogenitor cells, altering neurodevelopment and resulting in altered brain anatomy and impaired behaviors associated with autism in the exposed offspring. In addition, MAR ASD is associated with increased brain volume and more severe behavioral deficits in the affected progeny. This dissertation aims to characterize the maternal autoantibody profiles against 8 critical in fetal brain including CRMP1, CRMP2, GDA, LDHA, LDHB, NSE, STIP1, and YBOX, as well as defining their relationship of these autoantibodies with an ASD diagnosis in their children. Further, the development of an ELISA assay to determine the technical and clinical validity of these autoantibody patterns to be potentially used as ASD-risk biomarkers will also be described herein. Chapter 1 provides an in-depth literature review of maternal-immune dysregulation as an “insult” in the gestational environment that can trigger neurodevelopmental alterations and contribute to the etiology of autism. This chapter focused on epidemiological and experimental data that suggests that maternal anti-brain autoantibodies during gestation can increase ASD risk in their offspring and their potential clinical use for women at risk of having a child with a neurodevelopmental disability. In Chapter 2, we expanded upon our previous findings and described neuron-specific enolase (NSE) as a new protein target for maternal autoantibodies, and we also performed epitope mapping and that demonstrated differential antibody binding sequences between the case and control groups (ASD vs TD). Of interest, some of the epitope sequences recognized by the ASD groups had homology with epitopes from pathogens such as hepatitis virus and Vibrio cholera.

Chapter 3 was focused on the creation of an ELISA assay designed to characterize the profiles of ASD-specific maternal autoantibody profiles against the eight brain targets noted above, and to correlate these profiles with child outcome. We used plasma samples from mothers enrolled in the retrospective Childhood Autism Risks from Genetics and Environment (CHARGE) study to assess IgG reactivity to each protein by ELISA by establishing a positive cut-off based on the Youden index, a method that is used broadly in clinical tests. We then analyzed the data using machine learning sub-group discovery techniques and reported that reactivity to combinations composed of two or more of the target proteins was specific for an ASD diagnosis. We termed those combinations as maternal autoantibody related (MAR) ASD patterns. In that study, the primary patterns associated with ASD were CRMP1+GDA, CRMP1+CRMP2, and NSE+STIP. In addition, reactivity to CRMP1 increased the odds for a more severe autism diagnosis.

In Chapter 4, we validated the MAR ASD pattern in a new prospective cohort, the Early Markers of Autism (EMA) study, where samples were collected mid-gestation, allowing us to assess, for the first time, the ability to predict risk of a child developing ASD following gestational exposure to MAR autoantibodies. In addition, the EMA study allowed us to subdivide the ASD group into two categories: ASD without intellectual disability (ASD no-ID) and with ID (ASD+ID). Furthermore, we found that while CRMP1+CRM2 was the primary pattern for both ASD phenotypes, the presence of these autoantibodies during gestation significantly increased the risk for an ASD+ID outcome and could serve as a biomarker to facilitate the diagnosis of ASD with and without ID. In addition, while not statistically significant due to the low number of samples for each pattern, we made some interesting demographic observations relative to the ASD group and some of the MAR ASD patterns that will serve as a for future studies.

In conclusion, the 4 chapters described herein demonstrate that MAR ASD patterns are clinically relevant and highly specific for autism, with the prospect to facilitate diagnosis for children at high risk of neurodevelopmental disorders. While these patterns still require substantial clinical validation before they can be made available to the community, they are setting the stage for ongoing and future studies including the development of animal models that will serve to elucidate the pathological significance and the mechanisms of action for the autoantibodies that make up the patterns for MAR ASD.