Suppose a photon enters a density perturbation. What happens? The photon initally gains energy? To escape the density pertubration the photon has to lose energy. If there is no change in the state of the density pertubation, then there is no net change in photon energy.
But what happens if, during the photon transit time, the density perturbation decreases in amplitude. That means the photon has a smaller density pertubation to climb out of, than the one it initially entered. In this case, the photon experiences a net energy gain. In the opposite case where the density pertubation increases in amplitude then the photon experiences a net energy loss.
The end result of this will be an observed anistropy in the temperature of the Universe (e.g. not all microwave background photons will have exactly the same temperature). The temperature distribution of the photons be modified by the statistical occurence of photons traversing density enhancements. Now this does not happen to all photons, but since there are so many photons per density enhancement, then many photons are subject to this modification. This is what we measure in the WMAP data as a temperature anisotropy.
The r.m.s. tempearture fluctuations as measured by WMAP are on the order of 10-5. This is very small, but its sufficient to produce the observed large scale structure.
In addition, the angular scale over which the fluctuations are measured is quite sensitive to the geometry of the Universe. The WMAP scale strongly indicate that the Universe is flat. This is a critical Universe where the total mechanical energy is 0. What is intriguing about this situation, is that it seems there is now a need for "dark energy" to help insure this condition of 0 total mechanical energy. Weird ... (more on this weirdness later)
Mapping the angular scale of the flucutations on to the large scale geometery of the Universe.