As important as the genetic code itself is the decoding that takes place at ribosomes via translation of mRNA into proteins. Antibiotics may significantly interfere with this process 5. Particularly important is the ability of aminoglycosides (e.g., gentamicin) to suppress premature stop mutations in mammalian mRNA and to restore physiological amounts of protein that otherwise could not be translated and synthesized. This action of antibiotics such as gentamicin can be used for therapeutic purposes in genetic diseases that arise from point mutations which introduce premature stop codons (UAA, UAG, UGA) into coding sequences 67. For example, aminoglycoside antibiotics are currently being considered for treatment of Duchenne's muscular dystrophy patients who carry a nonsense mutation in the dystrophin gene 8 as well as for the treatment of cystic fibrosis (CF) patients 9. In CF patients, mutations in the cystic fibrosis transmembrane conductance regulator (CFTR) gene, containing a premature stop codon, are responsible for decreased production of CFTR chloride channels. In vitro, aminoglycoside antibiotics were shown to be capable of restoring the appearance of functional CFTR channels in mutant cells, and in a pilot study with nine patients carrying stop mutations, local gentamicin application produced improvement in a physiologic parameter measured in nasal epithelia 9. Interestingly, recent studies of autism patients confirmed previous evidence that linked autism to a region of chromosome 7q near the CFTR gene 10.

We now hypothesize that antibiotics could a) trigger the autism syndrome in susceptible infants by causing the stop codon readthrough, i.e., a misreading of the genetic code of a hypothetical vulnerable gene, and/or b) improve autism symptoms by correcting the premature stop codon mutation in a hypothetical polymorphic gene linked to autism.

These speculative hypotheses are inspired by the observations by Sandler et al. 4 who linked the onset of autism symptoms to previous antimicrobial therapy and the improvement of these symptoms to treatment with vancomycin. These authors did not specify whether aminoglycosides were used in children prior to onset of autism in their study; they primarily focused on antimicrobial therapy-induced diarrhea which they thought might relate causally to the onset of autism symptoms. Thus, our hypothesis that aminoglycoside antibiotics may precipitate autism in genetically susceptible individuals would be supported if a link could be established between amynoglicosides and the onset of autism symptoms, or if it could be found in a post-hoc analysis of existing clinical data from a larger pool of autism patients. Genetic susceptibility is a known factor in the occurrence of side effects of aminoglycosides, which produce lasting, sometimes permanent alterations despite even a transient use. For example, inherited susceptibility to aminoglycoside ototoxicity is based on a mutation in mitochondrial DNA (i.e., 12S rRNA) 11. It is possible that similar aminoglycoside-sensitive mitochondrial mutations could be involved in autism as well; namely, an autistic phenotype was recently observed in a child from a family with the mitochondrial DNA G8363A transfer RNA mutation 12.

Paradoxically, our hypothesis postulates that in some cases, i.e., premature stop codon mutation in a hypothetical autism-linked gene, antibiotics might be helpful in treating autism symptoms. Sandler et al. 4 found a transient improvement of autism symptoms in children treated with vancomycin. Since vancomycin is poorly absorbed (and it is not an aminoglycoside), these authors concluded that it must have produced a local effect (i.e., on the intestinal flora) and they proposed the existence of a "gut-brain" connection in autism. We hypothesize that aminoglycoside antibiotics might be helpful because they would correct a premature stop codon mutation in a hypothetical autism-linked gene product. This effect could not only be achieved in the central nervous system (CNS), i.e., for a CNS-specific RNA, but also in the periphery. For example, insulin-like growth factor (IGF-1), a growth-promoting peptide hormone that has neurotrophic properties in the CNS and is synthesized in the periphery including the intestine 13, has been shown to cross the blood-brain barrier; when injected systemically IGF-1 increases the proliferation and survival of neurons in the CNS of adult rats 14. Thus, if a similar peripherally-synthesized protein plays a role in autism when mutated, the premature stop codon mutation in the RNA that is responsible for its synthesis might be susceptible to correction by local (e.g., enteric) administration of an effective antibiotic.