Since the recent losses of several atmospheric instruments with good vertical sampling capabilities (SAGE II, SAGE III, GOMOS, SCIAMACHY, MIPAS), the scientific community is left with very few sounders delivering concentration profiles of key atmospheric species for understanding atmospheric processes and monitoring the Earth's radiative balance. The situation is so critical that, less than five such instruments will be on duty (most probably only 2 or 3) at the horizon 2020, whereas their number topped more than 15 in the years 2000. In parallel, recent inter-comparison exercises among the climate chemistry models (CCM) and instrument datasets have shown large differences in vertical distribution of constituents (SPARC CCMVal and Data Initiative), stressing the need for more accurate vertically-resolved data at all latitudes. In this frame, the Belgian Institute for Space Aeronomy (IASB-BIRA) proposed a small mission called ALTIUS (Atmospheric Limb Tracker for the Investigation of the Upcoming Stratosphere), which is currently in preliminary design phase (phase B according to ESA standards). Taking advantage of the good performances of the PROBA platform (PRoject for On-Board Autonomy) in terms of pointing precision and accuracy, onboard processing ressources, and agility, the ALTIUS concept relies on a hyperspectral imager observing limbscattered radiance and solar/stellar occultations every orbit. The objective is twofold: compared to scanning instruments, the imaging feature allows to better assess the tangent height of the sounded air masses (through easier star tracker information validation by recognition of scene details), while its spectral capabilities will be good enough to exploit the characteristic signatures of many molecular absorption cross-sections (O3, NO2, CH4, H2O, aerosols,...). The payload will be divided in three independent optical channels, associated to separated spectral ranges (UV: 250-450 nm, VIS: 440-800 nm, NIR: 900-1800 nm). This approach also offers better risk mitigation in case of failure of one channel. In each channel, the spectral filter will be an acousto-optical tunable filter (AOTF). Such devices offer reasonable étendue with good spectral resolution and excellent robustness and compactness and TeO2-based AOTF's have already been used in space missions towards Mars and Venus (MEX and VEX, ESA). While such TeO2 crystals are common in VIS-NIR applications, they are not transparent below 350 nm. Recent progresses towards UV AOTF's have been made with the advent of KH2PO4-based (KDP) filters. Through collaboration with the Moscow State University (MSU), several experiments were conducted on a KDP AOTF and gave confidence on this material. Here, we present the general concept of ALTIUS and its optical design with particular attention on the AOTF. Several results obtained with optical breadboards for the UV and VIS ranges will be exposed, such as the O3 and NO2 absorption cross-section measurements, or spectral images. These results illustrate the spectral and optical performances to be expected from an AOTF-based hyperspectral imager. Their implications for the ALTIUS mission will be discussed.