**5. The caspase fold of metacaspases**

Caspases, metacaspases and paracaspases have a conserved pattern of tridimensional organization and are then considered as structural homologues. The degree of this conservation is variable, but the overall structure is related to a conformation named caspase fold that is characterized by a core formed by a contiguous six‐stranded β‐sheet (β1–β4, β7 and β8) and helices α1–α5 region, present in every caspase structure. Also, the presence of three well‐ ordered loops (L1, L2 and L4) is well characterized [67].

As it was early discussed, metacaspases are divided into two groups based on the presence of a pro domain or a linker region. Type I metacaspases possess an N‐terminal prodomain with length of about 80–120 amino acids [32], with two CXXC‐type LSD1‐like zinc finger structures as well as a proline/glutamine rich region [45, 68]. The type II metacaspases do not have prodomains but, otherwise, contain a large loop (linker region) between the p10 and p20 domains which ranges about 90–150 amino acids [32, 69].

The p10 and p20 domains are present in all of these proteins. For p10, there is a SGCXDXQTSADV consensus sequence, as well as other conserved short sequences [40]. The p20 domain contains the conserved catalytic dyad histidine/cysteine as a remarkable feature, where the two amino acids are distant from each other in about 29–47 amino acids. There are also other conserved regions/amino acids that together give about 80% of consensus on the entire sequence [32]. A noticeable signature is the motif DSCHSG in the surroundings of the catalytic Cys, which is highly conserved among all type II plant metacaspases [70].

Differently of metacaspases, all caspases contain a conserved QACXG (where X can be R, Q, or G) pentapeptide active‐site motif. The catalytic residues histidine 237 and cysteine 285, and those involved in forming the P1 carboxylate binding pocket on caspase 1 (Arg‐179, Gln‐283, Arg‐341 and Ser‐347), are also conserved in all other caspases, except for the conservative substitution of the threonine for the serine 347 in caspase 8. This explains the requirement for an aspartate in the substrate P1 position. The residues that form the P2–P4 binding pocket are not well conserved, suggesting that they may determine the substrate specificities of the different caspases [71]. The metacaspases do not have these features and present cleavage specificity for lysine/arginine on the P1 position on the substrate, so their binding residues seem to be of opposite chemical nature from those of caspases.

The type II metacaspases present autoprocessing sites, whose cleavage seems to be necessary for their full activation: the residues Lys 260 and Arg 214, on the wheat type II metacaspase [70]; Lys 269, from MCIIPa [46]; and the Arg 214, from AtMC9 [45]. Caspases, otherwise, have different cleavage site, always after Asp residues. Caspase 1 is cleaved after Asp‐103, Asp‐119, Asp‐297 and Asp‐316 [71]; caspase 7, after Asp‐23, Asp‐198 [72]; and caspase 3, after Asp‐9/ Asp‐28 and Asp‐175 [73].

Actually, there are only two metacaspases with elucidated structural organization. MCA2, from *T. brucei*, the first metacaspase with elucidated structure, is topologically more related to other metacaspases, such as YCA1, than to caspases, presenting the same number and organization of β sheets and α helix, although the enzyme showed 65% of secondary elements similarly to caspase 7. Also, the β‐sheet region of MCA2 is sized two strands longer than caspases. As for YCA1, this characteristic prevents dimerization. Autocatalytic processing occurs only *in vitro* and is not required for enzyme activity, which was also shown to be dependent of calcium as in other metacaspases [74].

The crystal structure from the yeast metacaspase Yca1 shows the general patterns of the caspase fold, with the well‐ordered loops, being L1 and L4 in opposing sides of the substrate interaction site, and the conserved caspase core. Concerning β conformations, yeast YCA1 presents two β‐strands (β5 and β6) which are absent in caspases (caspase 3 and caspase 9), and these are located in a way that blocks dimerization. As a result, YCA1 cannot form dimer, as caspases do. The catalytic dyad consists on the residues Cys276 and His220 which are well conserved among other proteins with caspase fold. Also, the identity between the YCA1 and the caspases 3 and 9 is lower than 12%, and the sequence divergence greatly affected the root‐ mean‐square deviation (RMSD) analysis. The superimposition of YCA1 and caspase 3 showed a higher structural variation than the superimposition of caspase 3 and 9. On *in vitro* assays, the presence of Ca2+, and not of other divalent cations, was required to an enhanced capability of auto processing. Under this condition, two peptide fragments were formed revealing that the processing occurred after Arg72 and Lys86. When compared with MCA2, YCA1 shows 24.9% of amino acid sequence identity and a very similar structure through RMSD analysis [75].
