Researchers use animals in place of humans, for research purposes; in the case of autism it is usually the unfortunate mouse, but sometimes rats.
The Jackson Laboratory in the US is the source for more than 8,000 strains of genetically defined mice used for research purposes.
SFARIgene has a fascinating on-line database that lists all the mouse/rat models of autism and the research linked to them. Most importantly it also lists all the “rescue lines”, the research showing therapies that improved the mouse’s autism.
For example, you can look up the model of human Fragile-X, which is called Fmr1, and then see the long list of drugs that helped that particular type of mouse.
There are already well over 200 different mouse/rat genetic models of autism and 1,000 rescue lines.
So while medicine has no approved drugs to treat human autism, autistic mice appear to be better placed.
There remains the question of how close humans are to mice. They are more closely related than you might think, but there are still big differences.
There are also induced models of autism, where the scientists have not tinkered with a specific single gene; these might closer relate to most human autism. You will find a model of advanced paternal age, a model of diesel exhaust particles, and all kinds of other things.
One very widely used model is called the Maternal Immune Activation (MIA) model. In the research you may find it called Polyinosinic:polycytidylic acid, or just poly(I:C).
In the MIA model the pregnant mouse is injected will an immune stimulant (Polyinosinic:polycytidylic acid) that triggers a big immune response, which affects the development of her pup. The pup is born with features that resemble human autism.
There is a similar model where the mother is given an infection rather than induced inflammation.
Depending on the gestational age at which MIA or infection is administered, the offspring can be studied in the context not only of autism, but also schizophrenia. This should not be surprising if you have read the post discussing the overlapping polygenic nature of autism and schizophrenia.
You can even induce temporary autism using proprionic acid. Proprioic acid is produced naturally in your intestines when the food you eat reacts with the bacteria that live there. Proprionic acid is a SCFA (short chained fatty acid), you need to have some SCFAs, but as it often the case, too much may not be good for you. In the case of a mouse, when injected with a large dose of poprionic acid, its behaviour changes to that of autism. This is entirely reversible over time, or faster still, by administering the antioxidant NAC (N-acetyl cysteine).
Researchers create a mouse model that matches as closely as possible the human condition they are trying to treat. Then they can investigate various drugs that might be of therapeutic benefit. In some cases a large number of drugs from a library of compounds are tried on the off chance of stumbling upon one that is effective.
An alternative approach is when a researcher has a theory that a specific drug should be effective, he then tests it in several different mouse models of autism. If the drug is effective in several mouse models that would suggest it might be beneficial in some humans. This is how Ben-Ari advanced his bumetanide research and Catterall his low dose Clonazepam research; the difference is that Ben-Ari has moved on to humans, as regular readers know.
Those of you who look at the SFARgene database will see how hundreds of so very different things, both genetic or environmental, lead to the same autism.