So most experiments are done on mice (and rats) because they breed fast, die fast, are easy to care for, and because they've been used before in research so they are well understood and there are good tools for manipulating their biology.
But if you read studies done on mice and rats then you have to understand the differences between humans and rodents.
Here's an example of a standard lab rodent feed: http://labdiet.com/pdf/5001.pdf
Calories come from: 28.5% protein, 13.5% fat, 58% carbohydrate. This is a good diet for mice and rats, it keeps them healthy and slender. Now this is not what is usually recommended for humans (and just as well since we are rather different creatures). For 1-3 year-olds the official recommendations are for 5-20% protein, 30-40% fat, and 45-65% carbohydrate while for adults it is 10-35% protein, 25-35% fat, and 45-65% carbohydrate (US numbers, page 15, but official recommendations are pretty much the same in most of the Western world). So even official recommendations, which by paleo standards are considered a bit fat-phobic, have a minimum fat recommendation that is twice as high as what mice and rats thrive on. So what's the big difference between humans and mice? Well, for one thing the efficiency of de novo fatty acid synthesis is about 2.5-5 times as good in mice as in humans (sorry, I'm just going to use 'mice' from here on out). Mice are just much better at making fat from carbohydrates. And this feed has a very low level of sucrose (less than 2% of carbohydrates), a negligible amount of fructose, and no alcohol (fructose, whether as a monomer or from sucrose, and alcohol can only be metabolised in the liver and in a fed state the energy from that metabolisation can only be channelled into fatty acid synthesis - humans seem to also fail at moving that fat efficiently to peripheral fat stores. Getting mice to develop non-alcoholic fatty liver disease is quite hard and even when transgenic mice are employed it is hard to get the same phenotype as in humans).
Standard rodent chow isn't enjoyable either. The mice feed because they are hungry, not because the feed gives them a particularly great reward feeling. If mice are fed cafeteria junk food instead of standard chow they actually gain more weight than they do on a very high calorie density feed (with lots of fat).
Laboratory mice are originally bred from house mice. These are the vermin that were so adapted to eating grains that they moved in to human settlements almost as soon as agriculture was invented. The egyptians elevated cats to sacred animals because the mice were such a problem. So they've had agricultural grains as main calorie for as long as humans have (and probably with grains as a much larger component of their diet for millions of years) and they have a shorter generational time and a larger effective population which means that they have had much greater opportunity to adapt to that diet.
tl;dr There are always problems associated extrapolating from a model animal to humans but with regard to whether grains are good for you or not mice are pretty much useless.
Extra-reading: Slate has a great three-part article on lab mice and model animals http://www.slate.com/articles/health_and_science/the_mouse_trap/2011/11/lab_mice_are_they_limiting_our_understanding_of_human_disease_.html