DNase-seq tells you about accessibility, not actually gene activity (in effect, the degree to which DNAse can chew up a particular stretch of DNA is dependent on how coiled up that stretch is as less-coiled/bound stretches are more accessible). It's true that increases in accessibilty around transcription start sites will generally correlate to increased expression, but this isn't a 1:1 correlation. It's probably better to interpret highly accessible regions as having increased potential (versus some appropriate background) for transcription.