The period immediately following the proposal of M-theory has been called the “second superstring revolution,” because it once again inspired a flurry of research into superstring theory. The research this time focused on understanding the connections between the existing superstring theories and between the 11-dimensional theory that Witten had proposed.
Witten wasn’t the first one to propose this sort of a connection. The idea of uniting the different string theories into one by adding an 11th dimension had been proposed by Mike Duff of Texas A&M University, but it never caught on among string theorists.
Witten's work on the subject, however, resulted in a picture where the extra dimension could emerge from the unifications inherent in M-Theory — one that prompted the string theory community to look at it more seriously.
In 1994, Witten and colleague Paul Townsend had discovered a duality between the 10-dimensional superstring theory and an 11-dimensional theory, which had been proposed back in the 1970s: supergravity.
Supergravity resulted when you took the equations of general relativity and applied supersymmetry to them. In other words, you introduced a particle called the gravitino — the superpartner to the graviton — to the theory. In the 1970s this was pretty much the dominant approach to trying to get a theory of quantum gravity.
What Witten and Townsend did in 1994 was take the 11-dimensional supergravity theory from the 1970s and curl up one of the dimensions. They then showed that a membrane in 11 dimensions that has one dimension curled up behaves like a string in 10 dimensions.
Again, this is a recurrence of the old Kaluza-Klein idea, which comes up again and again in the history of string theory. By taking Kaluza’s idea of adding an extra dimension (and Klein’s idea of rolling it up very small), Witten showed that it was possible — assuming certain symmetry conditions — to show that dualities existed between the existing string theories.
There were still issues with an 11-dimensional universe. Physicists had shown supergravity didn’t work because it allowed infinities. In fact, every theory except string theory allowed infinities. Witten, however, wasn’t concerned about this because supergravity was only an approximation of M-theory, and M-theory would, by necessity, have to be finite.
It’s important to realize that neither Witten nor anyone else proved that all five string theories could be transformed into each other in our universe. In fact, Witten didn’t even propose what M-theory actually was.
What Witten did in 1995 was provide a theoretical argument to support the idea that there could be a theory — which he called M-theory — that united the existing string theories. Each known string theory was just an approximation of this hypothetical M-theory, which was not yet known. At low energy levels, he also believed that M-theory was approximated by the 11-dimensional supergravity theory.