An answer to another question here provides a great explanation (with pictures) of what a single frequency looks like in the spatial domain.
Note that a single sample in the frequency domain affects all pixels in the spatial domain. That frequency produces a sinusoid with one frequency in the x dimension and another in the y dimension.
If you want to change the overall intensity of the image, you can do so by manipulating the DC value in the frequency domain. This is the value at zero frequency in x and y (upper left sample in the frequency domain). The DC value represents a wave with zero frequency in the spatial domain, so it is just a constant offset over the entire spatial image, which is basically the overall intensity.
If a single frequency affects the entire spatial image, how do you manipulate the frequency samples to do meaningful things in the spatial domain? Well, one way to think about it is that hard edges and quick intensity changes in the spatial domain correspond with higher frequencies while smooth changes in intensity in the spatial domain correspond with low frequencies.
So, if you want to emphasize edges in the image (sharpen), increase the magnitudes of higher frequencies. Similarly, if you want to smooth the edges (blur), decrease the magnitudes of higher frequencies.