That's a good question. Let's see whether I can explain it in a way that makes intuitive sense.
First, I call your attention to some simple Newtonian physics. Let's suppose I had two particles scattering off each other in three dimensions. Both energy and momentum are conserved. If I project the collision of the particles onto a two-dimensional screen, momentum and energy is also conserved in the projection, but each particle in the projection behaves as if it had more mass than it did in three dimensions. The "extra mass" is proportional to the momentum that I projected out for each particle. Momentum in the extra (third) dimension behaves as a mass term in the two dimensional universe.
Now, we live in a three (space) plus one (time) dimensional universe. The reason for supposing that there are extra physical dimensions is to solve what's known as the "heirarchy problem": why is gravity so much weaker than the other forces? One solution is to suppose that matter particles and the three gauge forces are constrained to operate on a 3+1 dimensional slice of a higher-dimensional space, while gravity is permitted to propagate through the "bulk" (the rest of the space). Gravity, it is supposed, is every bit as strong as the other forces, but we only see a tiny slice of a force that is diluted through a far larger volume.
These extra dimensions, it is supposed, are "compactified". That is, if you travel a miniscule distance in one of these extra directions, you get right back to where you were. Compactified dimensions are also called "Kaluza-Klein" dimensions.
Gravitons, of course, are themselves massless. (If they weren't, gravity wouldn't obey an inverse square law.) But remember two things I said: gravitons are permitted to travel through the bulk, and momentum in an extra dimension appears as mass, when projected onto any subspace of the bulk. With the right kind of machine, I can kick a graviton off in an orthogonal direction, and it will appear in my lab as a hugely massive particle. A linear electron-positron collider is such a machine.
Gravitons are quantum particles, of course. As a result of the compactification scale of the extra dimensions, I can't simply give a graviton any old momentum in the orthogonal direction. The momenta are quantized. (In a semi-classical picture, I must fit a half-integer number of wavelengths around the compactified dimension.) This means that I will have a series of discrete states corresponding to the normal modes of the graviton in the extra dimension. This series is called a "Kaluza-Klein graviton tower", and it's analogous to the series of (electromagnetic) spectral lines that are produced by atoms.
The expected energy scale of the first K-K graviton tower may be in reach of the next generation linear collider that is on the drawing boards.