Angewandte
Chemie
versus EcACP (see Figure S10) revealed major differences
between these carrier proteins and is consistent with the
localization of the dye on the protein surface.
cross-linked EcACP–EcKASII showed a strong fluorescence
signal (lex = 408 nm and lem = 515 nm) (see Figure S14A),
although EcACP labeled with 4 itself gave a low fluorescence
signal (see Figure S14B). The titration with EcKASII of
EcACP labeled with 4 led to the appearance of a strong
fluorescence signal, albeit only after incubation overnight
(see Figure S14B). The incubation of EcKASII with 4 alone
did not result in fluorescence (see Figure S14C), thus
suggesting that protein–protein interactions between ACP
and ketoacyl synthase are required for the binding of 4 within
EcKASII. As expected, hACP loaded with 4 did not cross-link
with EcKASII (data not shown). These results indicate that
the solvatochromic fluorescence response with 4 is indeed
a result of its transition from sequestration within EcACP to
the active pocket of EcKASII. The increased fluorescence
response therefore provides a direct response for chain-
flipping activity.
We next examined the dynamic movement of a seques-
tered acyl substrate from within a type II ACP into the active
site of a partner enzyme, also referred to as chain flipping.[17]
This phenomenon has never been directly detected, but the
protein–protein interactions governing this activity remain at
the forefront of pathway engineering. Recently, two elegant
studies revealed snapshots of these protein–protein interac-
tions[17] in the form of cocrystal structures of E. coli ACP with
partner enzymes (LpxD[18] and FabA[1]). The dynamic inter-
action between FabA and ACP was also investigated by NMR
titration experiments; however, since the population of
protein species involved in chain flipping is most likely
small, this event has to date not been measured. The size of
ACP–partner-protein complexes pushes standard solution
protein NMR spectroscopy to its limits and requires novel
NMR spectroscopic techniques, such as solution-state
TROSYor solid-state magic-angle-spinning sedimentation.[19]
Indeed, when we added unlabeled E. coli ketoacyl synthase II
(EcKASII, FabF) to a solution of our labeled EcACP, severe
signal broadening was observed (see Figure S9). Further-
more, protein NMR spectroscopy still often requires rela-
tively large amounts of protein and can be time-consuming
and costly. Thus, our solvatochromic approach to visualization
of the chain-flipping mechanism offers a facile and orthogonal
way to study ACP–partner-protein interactions.
We chose to examine ketoacyl synthase II, which extends
fatty acids up to eighteen carbons.[20] Previously, we showed
that EcACP loaded with chloroacrylic or a-bromo moieties
can successfully interact and cross-link with EcKASII.[21] Both
EcACP and hACP loaded with 1 were titrated with increasing
concentrations of EcKASII, and we only observed a signifi-
cant increase in fluorescence for EcACP (Figure 2c). As
a negative control, EcACP was titrated with a protein that is
not its partner, bovine serum albumin (BSA), which showed
no effect (see Figure S11). Human ACP does not functionally
interact with EcKASII (Figure 2d), as determined by a mech-
anistic cross-linking experiment (not shown).[21a] These find-
ings indicate that either EcKASII induces hydrophobicity
around the pantetheine probe by compaction or EcKASII is
able to extrude the cargo from the ACP for binding within its
hydrophobic active site. Computational docking studies
indicate that the active site of EcKASII can accommodate
probe 1 (see Figure S12), thus suggesting that chain-flipping
activity is responsible for the increase in fluorescence.[22]
As mentioned, we recently introduced mechanistic cross-
linking as a tool to capture protein–protein interactions
between carrier proteins and their partners.[21a] We hypothe-
sized that cross-linking could be used to conclusively deduce
the source of the increase in the fluorescence signal upon the
addition of EcKASII to EcACP loaded with probe 1.
Therefore, the solvatachromic cross-linking probe 4 was
prepared, and its loading onto EcACP resulted in a yellow
colored protein with a fluorescence response. The addition of
EcKASII resulted in a cross-linked complex with increased
fluorescence, as visualized by SDS-PAGE (see Figure S13),
with a molecular weight of 54 kDa. Isolated solvatochromic
Single-molecule fluorescence microscopy[23] is emerging
as a valuable technique for studying biological processes, such
as protein trafficking in living organisms and protein–protein
interactions.[24] We reasoned that solvatochromic response
from chain-flipping activity could be applied to single-
molecule fluorescence microscopy, whereby single binding
events between EcACP and EcKASII could be quantified. In
this study, we attached EcKASII to a quartz surface and
flowed DMN-labeled EcACP over the surface (Figure 3a)
while recording a video of fluorescence events (Figure 3b).
Many bright events were observed, unlike when the pante-
theine probe 1 was flowed over the same surface. More than
85% of the brightest spots acquired by the camera (lex =
488 nm and lem = 525/50 nm) and selected by the software
were distinct peaks with a clear signal intensity over back-
ground and a defined lifespan (Figure 3b), whereas the
corresponding experiment with probe 1 alone resulted in
some bright spots, of which less than 20% were peaks
(possibly originating from aggregation). When labeled
EcACP was flowed over an unmodified poly(ethylene
glycol) (PEG, Mw ꢀ 5 kDa) surface, approximately 60% of
the observed bright spots were peaks, but with very different
intensities and “ON” time (lifespan) distribution (Figure 3b).
In the case of labeled EcACP–EcKASII, the lifespan times
were short and peak-intensity distributions were narrow, thus
indicating that the interactions with the partner protein
(KASII) changed the environment of the probe attached to
EcACP significantly. Conversely, in the case of labeled
EcACP flowing over an unmodified PEG surface, peak
intensities were substantially increased, and the lifespan times
showed more variation (Figure 3b). We also modified EcK-
ASII with the cyanine dye Cy5 and performed a colocalization
experiment with two different lasers (lex = 488 nm and lex =
640 nm) in which either the Cy5-labeled EcKASII or the
DMN-labeled EcACP was excited. This experiment is nor-
mally used for tight binding events; we were investigating
transient interactions between ACP and KS, thus complicat-
ing the analysis. However, a nontrivial population (ca. 8%) as
compared to the statistical probability (ca. 3%) showed
bright fluorescence events in exactly the same location, in
both channels, thus suggesting that we were indeed observing
ACP–partner-protein interactions (see Figure S15). The sig-
Angew. Chem. Int. Ed. 2014, 53, 14456 –14461
ꢀ 2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim