A mid-latitude filament was observed before and after its eruption with SOHO (EIT, SUMER, CDS and LASCO) and with ground based observatories (Meudon and Pic du Midi) in the context of a coordinated MEDOC campaign in Orsay. The eruption was followed by a large Coronal Mass Ejection well observed by LASCO. Few hours before its eruption, the filament is partially heated (as seen in 195 \AA with EIT). The physical conditions of the filament before its eruption have been investigated by spectroscopic analysis of SUMER (the Lyman series L4 to L9) and of CDS combined with the Multi-channel Subtractive Double Pass Spectrograph (MSDP) spectra of H\alpha. Five hours before the eruption, large broadening of chromospheric and transition region lines (CDS) were observed in the main body of the filament suggesting strong turbulence as well as opposite Dopplershifts one each side of the filament (H\alpha and He I) which could be interpreted as twist motions. The optical thickness of the filament is rather large. During the eruption the twist is largely developed as observed in 304 \AA. The Doppler shifts of the filament estimated from Lyman lines are in good agreement with the velocity of the front edge of the CME bright loop. We notice that both, the filament and the bright loop, are deviated towards the equator. This implies that they belong to the same global expansion event constrained to remain in the equatorial streamer. An X-ray bright point observed close to a filament footpoint could be the signature of reconnection process linked to the destabilization of the filament. It is interpreted in the frame work of new MHD modelling of lateral filament footpoints (Aulanier \& Demoulin 1998).