(a,d) MALLS analysis of the CNOT3 and CNOT2 NOT-box domains. The molecular weight of the protein in solution is indicated in the elution profile. (b) Interfaces between the CNOT3 NOT-box α-helices and the β-barrel. Residues along the interfaces are shown as sticks. (c) Hydrophobic core of the CNOT3 NOT-box β-barrel. (e) Structure of the CNOT2 tetramer as observed in the crystal. The tetramer consists of two pairs of dimers (orange-yellow vs. purple-rose) with a perpendicular orientation to each other. (f) Structure-based sequence alignments of the NOT2 and NOT3 NOT-box domains. Secondary structure elements as determined from the CNOT2 and CNOT3 structures are shown above the alignment. Residues conserved in all aligned sequences are shown with a yellow background, and residues with >70% similarity are highlighted in orange. A NOT2-specific insertion is boxed in purple. Black squares mark residues that form the interface between the NOT-Box N-terminal α-helices and the β-barrel. Gray squares mark residues that form the hydrophobic core of the β-barrel. The species abbreviations are the same as those in Supplementary Fig. 2. (g) Pulldowns showing that CNOT2-C and CNOT3-C exclusively form heterodimers in solution. MBP-tagged CNOT2 and His6-tagged CNOT3 were coexpressed in E. coli. The heterodimers were copurified using MBP pulldown followed by Ni-affinity purification (lanes 1 and 2) or Ni-affinity purification followed by MBP pulldown (lanes 3 and 4).