288
between protein modules, linkers are also decisive in the engineering
of proteins with novel function. In particular, desired traits can
often be obtained by covalently connecting different modules via
suitable linkers and by thus generating hybrid (or, chimeric) proteins. Some cases call for flexible linkers that bring modules into
spatial proximity, but allow reorientation and relative movements;
in other cases, rigid connectors are required to fix modules at discrete distances and defined angular orientations. Additional considerations in linker design include proper folding, intracellular
trafficking, chemical and biological stability of resultant hybrid
proteins [3]. Structural information, where available, and multiple
sequence alignments often provide valuable clues as to which linker
suffices for a specific engineering purpose. Nonetheless, a priori it
is difficult to select among many possible linker variants the one(s)
best suited for meeting the above criteria. As a corollary, often
multiple linkers are constructed and empirically tested for best
performance, which incurs considerable expenditure of time and
effort.
These concepts are exemplified in signal receptors and their
engineering. In particular, sensory photoreceptors constitute the
group of signal receptors that impart sensation of light [4].
Photosensor modules absorb light of appropriate quality and in
response modulate the biological activity of effector (or, output)
modules. Across different photoreceptor classes, the linkers connecting these modules are often of α-helical or coiled-coil conformation; modifications to the linker as confined as exchange,
addition or deletion of single residues can profoundly affect receptor activity and regulation [5–7]. As a case in point, we constructed
the photoreceptor YF1 by recombining the blue-light-responsive
light–oxygen–voltage (LOV) photosensor module of Bacillus subtilis YtvA (BsYtvA) with the Bradyrhizobium japonicum FixL
(BjFixL) histidine kinase effector [8] (Fig. 1). The crystal structure
of homodimeric YF1 in its dark-adapted state showed its two LOV
photosensor modules to be connected to the effector module via a
parallel coiled-coil linker, denoted Jα [9]. Length variations of Jα
revealed heptad (i.e., seven-residue) periodicities of catalytic activity and regulation by light [8]. The original YF1 construct derived
almost its entire linker from the parental protein BjFixL, but
equally one could have used the corresponding linker of the other
parental protein BsYtvA, or hybrids of both linkers. Given that the
parental linkers are 23 and 27 residues long, there are (23 + 1)·(27
+ 1) = 672 possible combinations for connecting BsYtvA and
BjFixL if one restricts hybrid fusions to these linker segments
(Fig. 1). Although comprehensive interrogation of all linker combinations could provide invaluable insight into signaling mechanisms and engineering principles, manual construction and separate
testing of each individual variant is prohibitively cumbersome. We
hence sought to assess all possible linker variants in parallel. To this
Robert Stabel et al.
between protein modules, linkers are also decisive in the engineering
of proteins with novel function. In particular, desired traits can
often be obtained by covalently connecting different modules via
suitable linkers and by thus generating hybrid (or, chimeric) proteins. Some cases call for flexible linkers that bring modules into
spatial proximity, but allow reorientation and relative movements;
in other cases, rigid connectors are required to fix modules at discrete distances and defined angular orientations. Additional considerations in linker design include proper folding, intracellular
trafficking, chemical and biological stability of resultant hybrid
proteins [3]. Structural information, where available, and multiple
sequence alignments often provide valuable clues as to which linker
suffices for a specific engineering purpose. Nonetheless, a priori it
is difficult to select among many possible linker variants the one(s)
best suited for meeting the above criteria. As a corollary, often
multiple linkers are constructed and empirically tested for best
performance, which incurs considerable expenditure of time and
effort.
These concepts are exemplified in signal receptors and their
engineering. In particular, sensory photoreceptors constitute the
group of signal receptors that impart sensation of light [4].
Photosensor modules absorb light of appropriate quality and in
response modulate the biological activity of effector (or, output)
modules. Across different photoreceptor classes, the linkers connecting these modules are often of α-helical or coiled-coil conformation; modifications to the linker as confined as exchange,
addition or deletion of single residues can profoundly affect receptor activity and regulation [5–7]. As a case in point, we constructed
the photoreceptor YF1 by recombining the blue-light-responsive
light–oxygen–voltage (LOV) photosensor module of Bacillus subtilis YtvA (BsYtvA) with the Bradyrhizobium japonicum FixL
(BjFixL) histidine kinase effector [8] (Fig. 1). The crystal structure
of homodimeric YF1 in its dark-adapted state showed its two LOV
photosensor modules to be connected to the effector module via a
parallel coiled-coil linker, denoted Jα [9]. Length variations of Jα
revealed heptad (i.e., seven-residue) periodicities of catalytic activity and regulation by light [8]. The original YF1 construct derived
almost its entire linker from the parental protein BjFixL, but
equally one could have used the corresponding linker of the other
parental protein BsYtvA, or hybrids of both linkers. Given that the
parental linkers are 23 and 27 residues long, there are (23 + 1)·(27
+ 1) = 672 possible combinations for connecting BsYtvA and
BjFixL if one restricts hybrid fusions to these linker segments
(Fig. 1). Although comprehensive interrogation of all linker combinations could provide invaluable insight into signaling mechanisms and engineering principles, manual construction and separate
testing of each individual variant is prohibitively cumbersome. We
hence sought to assess all possible linker variants in parallel. To this
Robert Stabel et al.