Front Matter

The side chain of L258 was identified as the major factor to discriminate between

sn-1 andsn-3 orientation. To verify and apply this model, the size of this residue was

changed by mutating it to the smaller alanine, the more polar serine, and the bulky

phenylalanine. As predicted by the model, decreasing its size decreased repulsion of

bulkysn-2 substituents, thus allowing the substrate to move into the His gap and

shifting stereoselectivity towardsn-1. This effect is most prominent for the bulky

and rigid phenyl substituent. For this substrate, a switch in stereopreference from

sn-3 tosn-1 occurred. For the rigid but smaller amide substrate and for the flexible

ester, geometry and stereoselectivity did not change considerably compared to wild-

type ROL. For all substrates, enantiomeric excess values of L258A and L258S are

similar. Therefore, the stereoselectivity-determining interaction was assumed to be

steric rather than polar.

Increasing the size of residue 258 by replacing leucine by phenylalanine had an

even more dramatic effect on the geometry of binding and thus of stereoselectivity.

Since rigid substrates are pushed out of the gap insn-1 orientation,sn-3 stereose-

lectivity is considerably increased. Flexible substrates are still accommodated in the

gap. Hence, only moderate changes of stereoselectivity towards flexible substrates

were observed.

While the sequences of ROL and RML are highly similar, and notably L258 is

conserved, the stereoselectivity of both enzymes is different: in contrast to ROL,

RML hydrolyzes flexible and rigid substrates insn-1. The only nonconservative

difference in the substrate binding site-A89 in ROL, W88 in RML-has been shown

not to play a role in stereoselectivity: replacing A89 in ROL by tryptophan did not

affect stereoselectivity (Scheib et al., 1999). Therefore, it is not obvious which re-

sidues mediate stereoselectivity. Using protein modeling and engineering, a second

layer of residues was identified which are relevant for function (Beer et al., 1996).

Although these residues are not in contact with the substrate, they stabilize the cat-

alytic machinery by a hydrogen network. The authors postulated that H144 in ROL

plays a key role in the positioning of the active site H257. Replacing E265 by as-

partate resulted in a breakdown of the active site geometry, supposedly due to a new

H-bond between H144 and the carboxyl group of D265. Also, packing of the aspar-

tate side chain close to H144 should increase steric hindrance between both residues.

These second layer residues also seemed to play a role in stereoselectivity. As a

major difference between ROL and RML, triacylglycerol substrates were pushed

out of theHis gapin RML relative to ROL. This is caused by interaction with

G266, which is located close to C 2 of glycerol in a short loop, theG-elbowloop

(Figure 6). While G266 is conserved in both enzymes, the G-elbow loop is shifted

toward the binding site in RML as compared to ROL. The neighboring residues to

G266 vary in Mucorales lipases. ItsN-terminal neighbor (T265 in RML, E265 in

ROL) interacts with H143 or H144 in RML and ROL, respectively. Because of pack-

ing restraints, the side chain of E265 points away from H144 in ROL. In RML,

however, T265 is small enough to orient its side chain toward H143 forming a hy-

drogen bond. Hence, Caof T265 in RML is located closer to H143 than E265 in

ROL, thus shifting the G-elbow loop toward the substrate binding site. Since the

substrate binds less deeply in RML, the effect on stereoselectivity is similar to in-

creasing the size of theHis gap: flexible and rigid substrates bind with a similar

geometry and stereoselectivity shifts towardsn-1. Thus, the observed differences

5.4 Modeling and engineering of stereoselectivity 95