Autophagy is a conserved cell-survival pathway wherein old, damaged or harmful cellular components are surrounded by a double membrane vesicle called the autophagosome for lysosomal degradation and recycling. All eukaryotes have a conserved BECN homolog, a key coiled-coil domain (CCD)-containing autophagy protein. Mammals are unique as they have two BECN homologs, BECN1 and BECN2, both of which have the same domain architecture and function in autophagy. BECN2 has been shown to also function in non-canonical autophagy. My research focused primarily on investigating selected interactions of the BECN2 CCD. We investigated interactions of the BECN2 CCD with two CCD-containing proteins known to bind to BECN1: UVRAG, an important autophagy protein, and TAB2, a protein important for inflammatory responses. We show that the BECN2 and UVRAG CCDs interact, but were unable to purify stable complexes for structural studies. In comparison, the CCDs of BECN2 and TAB2 bind with an affinity tighter than that of BECN2 homodimerization, forming a well-folded elongated heterodimer. These preliminary results provide information regarding interactions that enable BECN2 to regulate autophagy, in a manner analogous to BECN1. We also show that, unlike BECN1, BECN2 facilitates non-canonical autophagy involving ATG9A-bearing vesicles, via interactions with the STX6 SNARE domain. We show that the STX6 SNARE domain binds to BECN2 residues 181-250 (BECN2(181-250)) within the BECN2 CCD. The STX6 SNARE is disordered in the absence of interacting partners but becomes helical when in complex with BECN2. The BECN2(181-250):STX6 SNARE complex is also more stable than either the BECN2(181-250) or the BECN2 CCD homodimer. We determined the 2.65 Å X-ray crystal structure of the STX6 SNARE bound to BECN2, showing that the complex is a heterotrimeric helical bundle, consisting of one helix comprising BECN2(181-250) and two STX6 SNAREs that are anti-parallel to each other. The heterotrimer interface is stabilized by 15 layers of three residues, each contributed by a different helix, of which, six layers are entirely hydrophobic, including two consisting of three leucines each. We verified the importance of these hydrophobic layers using point mutations and affinity pulldowns, showing that mutations of the hydrophobic layers either significantly or completely disrupt the interaction between BECN2(181-250) and STX6 SNARE domain. This 3-helix bundle likely represents an intermediate during the formation of the full SNARE complex, thereby providing mechanistic insights into the process by which ATG9A-bearing vesicles transport lipids to the growing phagophore. Together these studies help explain the role of the BECN2 CCD in canonical and non-canonical autophagy, providing clues as to why mammals have two BECN paralogs.