This is not entirely true, E. Coli is known to be able to metabolise glucose. The bacteria were "grown" in a solution that included glucose as it's main component. There were also many populations of the bacteria that were being evolved seperately (they NEVER mixed). Suddenly, in one population, a bacteria emerged that could metabolise citrate. This gave that bacteria a massive advantage, because it could now consume two types of food and it had no competition for the citrate (unlike glucose, which all the other bacteria could consume as well).
This also allowed the total population in that group to explode (there's now more food in total, glucose + citrate).
Another cool thing is that this smashes the "Irreducible Complexity" argument. The ability to metabolise citrate is developed by two separate mutations, which, on their own achieve nothing. Some of the populations developed the first mutation and some developed the second one, but none of them had previously developed both. This shows that the "preliminary" mutations were not harmful to the bacteria, so they just "hung around" until one of them was lucky enough to get the second mutation too.
Anyway, look up Lenski's work, I'm sure his papers (and those of his students/colleagues) are better at explaining it all than me...