We demonstrate the application of continuous wave dynamic nuclear polarization (DNP) at 0.35 T for site-specific water accessibility studies on spin-labeled membrane proteins at concentrations in the 10–100 μM range. The DNP effects at such low concentrations are weak and the experimentally achievable dynamic nuclear polarizations can be below the equilibrium polarization. This sensitivity problem is solved with an optimized home-built DNP probe head consisting of a dielectric microwave resonator and a saddle coil as close as possible to the sample. The performance of the probe head is demonstrated with both a modified pulsed EPR spectrometer and a dedicated CW EPR spectrometer equipped with a commercial NMR console. In comparison to a commercial pulsed ENDOR resonator, the home-built resonator has an FID detection sensitivity improvement of 2.15 and an electron spin excitation field improvement of 1.2. The reproducibility of the DNP results is tested on the water soluble maltose binding protein MalE of the ABC maltose importer, where we determine a net standard deviation of 9% in the primary DNP data in the concentration range between 10 and 100 μM. DNP parameters are measured in a spin-labeled membrane protein, namely the vitamin B₁₂ importer BtuCD in both detergent-solubilized and reconstituted states. The data obtained in different nucleotide states in the presence and absence of binding protein BtuF reveal the applicability of this technique to qualitatively extract water accessibility changes between different conformations by the ratio of primary DNP parameters ϵ. The ϵ-ratio unveils the physiologically relevant transmembrane communication in the transporter in terms of changes in water accessibility at the cytoplasmic gate of the protein induced by both BtuF binding at the periplasmic region of the transporter and ATP binding at the cytoplasmic nucleotide binding domains.