The vast majority of deep-sea ecosystems are sustained by exported organic material from the productive, sunlit surface ocean. Bacteria dominate benthic communities both in biomass and abundance, and have been recognized as the key players in the remineralization of organic material. Since most sediment bacteria remain however uncultivated and represent unknown taxa, we have very limited knowledge of their metabolic capabilities and enzymatic machinery. Here we studied deep-sea surface sediments along a seafloor depth gradient from 1200 m to 5500 m at the Arctic long-term ecological research station HAUSGARTEN. We applied Illumina 16S rRNA gene surveys based on DNA and cDNA, as well as metagenomic and -transcriptomic sequencing to elucidate total and active bacterial community composition and gain insight into the carbohydrate processing and uptake capabilities of deep-sea benthic bacteria. We identified specific taxa of interest and quantified their cellular abundance using CAtalyzed Reporter Deposition–Fluorescence In Situ Hybridization. Results from the different molecular approaches were in good agreement and suggested similar community structures with the same dominant members. Interestingly, typically predominant sediment taxa, i.e. the JTB255 marine group, the Sh765B.TzT29 group or the OM1 clade, were underrepresented in the active part of the community, while other usually low-abundant taxa, i.e. Flavobacteriia and the SAR202 clade, were overrepresented. At low taxonomic resolution, communities along the slope were similar, yet showed high turnover at species level. Although, the repertoire of carbohydrate-active enzymes (e.g. polysaccharide hydrolases) appeared unchanging along the depth gradient, the relative contribution of distinct enzyme-coding genes varied. Specific glycoside hydrolases involved in polysaccharide degradation of algae material (e.g. for laminarin; xylan) had higher counts at shallow depth, while others responsible for the breakdown of bacterial cell walls (e.g. for components of peptidoglycan) were more strongly represented at deep stations. Our findings indicate an adaptation of the communities to differences in organic matter quality.