Angewandte
Communications
Chemie
tive measurement of protein dynamics was performed with
labeled live mammalian cells and neurons by fluorescence
correlation spectroscopy (FCS).
ments.[10] Mindful of the potential photolability of AZ
owing to its photoreactive aryl azide moiety,[14] we chose
5Y/5T over 5Y/5Z (a natural surrogate pair of HPG/AHA),
because it enabled simultaneous double-click labeling with an
1:1 mixture of TER-N3 and BT reporters in the presence of
CuI catalysts (Figure 1D). Possible cross-reactivity between
azide/TCO was previously shown to be low.[15] As proof-of-
concept experiments, we chose 4 h incorporation windows to
achieve maximum protein labeling signals, although a shorter
incubation time may be suitable as well (Supporting Infor-
mation, Figure S9). As shown in Figure 1E, sequential
incubation of live HeLa cells with 5Y and 5T, followed by
(TER-N3 + BT) treatment, produced two populations of
labeled proteins as shown in both the in-gel fluorescence
profiles (left) and imaging results (right). Cross-background
labeling was not detected in cells treated with 5Y or 5T alone
(panels 1/2/4/5/7/8). By monitoring elF2a phosphorylation (a
cell stress marker[8a]), we detected no apparent cellular stress
under our labeling conditions due to formation of truncated
proteins (Supporting Information, Figure S9). While 5Y-
treated cells showed a diffuse labeling pattern (panel 1),
a small number of punctate stains were observed in 5T-
labeled cells (panel 5). Furthermore, the cellular distribution
of their fluorescence signals appeared to be slightly different.
This was despite the fact that in our earlier IF experiments
(Supporting Information, Figure S5), most analogues showed
similar staining patterns. We therefore evaluated whether
different puromycin analogues, possibly due to the minor
difference in their ribosome A-site affinity as a result of
structural variation, might cause the newly synthesized
proteins to degrade or aggregate differently (Supporting
Information, Figure S10); while protein degradation rates for
most analogues were similar, the molecular weight distribu-
tion, as well as the relative amounts of soluble/insoluble
proteins, varied slightly. We thus concluded that 5Y/5T pair is
indeed suitable for pulse-chase multiplexed labeling experi-
ments.
We first designed and synthesized two new PO analogues,
5Y and BY (Supporting Information, Scheme S1). Together
with the reported AY, each analogue has a terminal alkyne
substituted at the AA, 5’-OH and the base site, respectively
(Figure 1A). The puromycylation reaction of newly synthe-
sized proteins from HeLa cells, followed by SDS-PAGE/WB
analysis with anti-puromycin antibody (Figure 1B), and in-gel
fluorescence scanning of the corresponding fluorescently
labeled protein lysates (with respective click reporters;
Figure 1C), indicate successful metabolic incorporation of
all three analogues in a concentration-dependent manner
(Supporting Information, Figure S2), although the rate of BY
incorporation was significantly lower than that of PO, AY or
5Y (boxed in red; Figure 1B,C). Large-scale PD/LC-MS/MS
experiments showed similar trends, resulting in positive
identification of newly synthesized proteins from various
subcellular compartments (Supporting Information, Fig-
ure S3). Pretreatment of cells with cycloheximide (CHX,
a protein synthesis inhibitor[13]) nearly abolished the incor-
poration (Figures 1B,C; Supporting Information, Figure S4).
With these findings, and in consideration of synthetic
accessibility, the 5’-OH of PO was thus chosen as the
preferred site for most of our subsequent modifications; we
À
made 5Z/5N/5T, having N3, norbornene, and trans-cyclo-
octene (TCO), respectively, located at the 5’-OH site, as well
À
as AZ which has N3 at the AA site (further structural
changes at AA site were not tolerated; data not shown).
Similar metabolic incorporation experiments indicate all new
analogues (except BY) were nearly as effective as AY and 5Y
(Figures 1B,C; Supporting Information, Figure S2), a conclu-
sion also supported by the corresponding IF experiments
(Supporting Information, Figure S5). By taking advantage of
the clickable tag in each PO analogue, direct in-cell imaging
of PO-labeled newly synthesized proteins was carried out by
treatment of fixed cells with suitable fluorescent reporters
(Supporting Information, Figure S6); most labeled cells
except those from BY-incorporated experiments showed
varying degrees of fluorescence signals distributed through-
out the cytosol and nucleus, while DMSO-treated control cells
showed minimal background fluorescence. This indicates our
newly developed PO analogues were indeed suitable for
labeling and imaging of newly synthesized proteins from
mammalian cells. Representative analogues were further
shown to work well in other mammalian cells and primary
neurons (Supporting Information, Figures S7). By striking
a balance between labeling newly synthesized proteins with
these new puromycin analogues, and their cytotoxicity at
higher concentrations (as protein synthesis inhibitors; Sup-
porting Information, Figures S2 and S8), we found most
probes worked well at 10 mm, thus this concentration was
chosen for subsequent experiments.
We next assessed whether the imaging experiments could
be performed in live cells, by using the 5Z/5T pair (Figure 2),
which may be click-labeled by reporters containing dibenzyl-
cyclooctyne (DBCO) and tetrazine, respectively, under
copper-free conditions.[9] We first optimized the types of
fluorophores suitable for live-cell imaging experiments (Sup-
porting Information, Figures S11–13 and Movie S1); both
FITC-DBCO and TMR-DBCO were suitable for imaging 5Z-
labeled, newly synthesized proteins. For 5T-labeled proteins,
FITC-Tz and TMR-Tz were suitable reporters. Interestingly,
the two previously reported tetrazine-containing fluoro-
phores,[16] BT and CT, were found to be mostly trapped in
the endo/lysosomes under our conditions and not suitable for
live-cell imaging. We further confirmed the positive imaging
signals were from newly synthesized proteins by blocking
protein synthesis with CHX and preventing protein degrada-
tion with bortezomib (a proteasome inhibitor; Supporting
Information, Figure S11). Finally, we chose 5Z/5T and TMR-
DBCO/FITC-Tz for subsequent multiplexed imaging experi-
ments in live mammalian cells and neurons.
Encouraged by above results, we next investigated
whether these new PO analogues could be used in pulse–
chase experiments for multiplexed labeling of two temporally
defined sets of newly synthesized proteins (Figure 1D,E). The
AHA/HPG pair was previously used in similar experi-
As shown in Figure 2 (see also the Supporting Informa-
tion, Figure S14), sequential labeling of live HeLa cells with
Angew. Chem. Int. Ed. 2016, 55, 4933 –4937
ꢀ 2016 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
4935