DISCUSSION
Assembly line PKSs, which represent some of the largest known proteins, contain modules and domains connected by loops that are experimentally challenging to study. To better understand the dynamics of PKSs and facilitate their engineering, we sought to better characterize these loops. As the insertion of repetitive sequences naturally occurs throughout the DNA encoding PKS assembly lines, cataloguing where insertions persist provides a measure for the tolerance of the encoded loops to modification.
Out of all of the loops in assembly line PKSs, the most tolerant to insertion is at the updated module boundary, between KS and FSD[4-6]. Even 137 unstructured residues at this location seems not to impair the function of the nannocystin module NcyB_1c. Connecting modules at this junction may preserve how its component domains structurally and functionally work together, as supported by recentin vivo and in vitro PKS engineering studies [29, 30]. Although this junction is upstream of the AT domain, AT swapping has been naturally observed and accomplished through engineering (by including FSD)[4, 31]. More studies are needed to elucidate the connection between the AT region and the downstream domains of the module.
The AT-DH loop of γ/δ-modules is more sensitive to modification than the AT-KR loop of β-modules. Out of 599 examined modules, there are no insertions in the AT-DH loop, nor are there deviations from its length of 5-7 residues when n1 is present. One explanation is that this loop is structured at the interface between KS and DH. This would mean that this connection is not hinge-like as in the highly related metazoan fatty acid synthase[32]. Other connections between enzymatic domains, in particular DH-KRs and ER-KRc, are also resistant to modification. Perhaps these loops help position the catalytic sites of the module relative to ACP.
The loops upstream and downstream of ACP (15 ± 5 and 13 ± 6 residues, for β-δ modules) may need to be short to prevent docking with non-cognate enzymes, such as AT of the downstream module; however, these tethers limit the degrees of freedom available to ACP. Surprisingly, δ-modules, which contain the most enzymatic domains, possess the shortest linkers (8 ± 1 and 14 ± 2 residues). As ACP cannot stretch to access its AT in the assembly line as presented (Figure 1), AT may need to access another conformation. One possibility is that ATs adopt the position observed in the electron microscopy reconstructions of a construct from the pikromycin PKS[8]. Another possibility is that a hinge between AT and the processing enzymes (immediately after the LPTYxFx5W motif) facilitates these docking interactions[32]; however, this requires large, asymmetric motions throughout the assembly line that would hamper polyketide production compared to a more rigid, symmetric assembly line. In the yeast fatty acid synthase, active sites are fixed around reaction chambers in which the only mobile domain is ACP[33].
Those studying and engineering a PKS may wish to make changes to its polypeptides. Sequence alignments do not always sufficiently indicate the tolerance of loops to modification. As the repetitive sequences catalogued here are not likely to be functionally relevant, they serve as beacons illuminating where changes can be made both between and within domains. The reported repetitive sequence insertions could help in positioning 3C protease cut sites on each end of an ACP to liberate it from an assembly line for analysis by mass spectrometry, installing purification tags upstream of Class 1a NDDs or elsewhere, and adding domains such as fluorescent proteins or polyketide processing enzymes at desired locations.
As essential as loops are to the proper functioning of assembly line PKSs, they have been relatively ignored compared to PKS domains. However, they are informative as to how far domains can move from one another and provide restraints for structural biologists attempting to elucidate assembly line architecture. Some loops may be more structured than previously thought, such as the residues at the AT/DH interface and those downstream of KRs from γ/δ-modules. With the abundant sequence information that is now available the tolerance of loops to modification can be evaluated through bioinformatics. The resistance of domains such as ACP and KS to these modifications also reveals the importance of their surfaces for domain-domain and docking interactions. Complementing what is known about PKS domains by studying the loops that connect them and are present within them has advanced our understanding of PKS assembly lines as well as our ability to engineer them.