B R I E F C O M M U N I C AT I O N S
Cell-selective metabolic labeling
of proteins
John T Ngo1, Julie A Champion1, Alborz Mahdavi1,
I Caglar Tanrikulu1, Kimberly E Beatty1, Rebecca E Connor1,
Tae Hyeon Yoo1, Daniela C Dieterich2, Erin M Schuman2 &
David A Tirrell1
Metabolic labeling of proteins with the methionine surrogate
azidonorleucine can be targeted exclusively to specified cells
through expression of a mutant methionyl-tRNA synthetase
(MetRS). In complex cellular mixtures, proteins made in cells
that express the mutant synthetase can be tagged with affinity
reagents (for detection or enrichment) or fluorescent dyes (for
imaging). Proteins made in cells that do not express the mutant
synthetase are neither labeled nor detected.
Time-dependent changes in cellular proteomes can be monitored
via a variety of powerful electrophoretic and spectroscopic methods. Traditionally, radiolabeled amino acids have been used to label
proteins synthesized during an amino acid ‘pulse’; labeled proteins
can be distinguished from preexisting (unlabeled) proteins through
electrophoretic separation followed by radiographic detection1. More
recently, mass spectrometry has enabled the use of stable isotopes in
amino acid pulse labeling2.
In 2006, investigators introduced the BONCAT (bio-orthogonal noncanonical amino acid tagging) strategy for selective enrichment and
identification of newly synthesized proteins in cells3,4. The BONCAT
approach reduces sample complexity and permits direct analysis of the
primary protein synthesis response to stimuli. Bio-orthogonal functional
groups5 are introduced into proteins by pulse labeling with reactive, noncanonical amino acids. Labeled proteins are selectively modified with
affinity tags for enrichment6,7; removal of unlabeled proteins simplifies
subsequent analysis and identification by mass spectrometry.
All of these methods suffer from limitations when experiments
are performed in systems that contain multiple cell types. Because
incorporation of amino acids is nonspecific with respect to cell identity, proteins from all cell types are labeled. In studies of interactions
between different cell types in a single organism, the origin of the
identified proteins can be difficult to ascertain because the cells share
a common genome. When interactions between cells of different
genomes are studied, detection of low-abundance proteins can be
problematic. In infection studies, for example, the protein content of
the larger host cells can overwhelm that of the pathogen8 and limit
detection and identification of the proteins of primary interest. In
complex bacterial communities where hundreds of organisms can
occupy a common biological niche9, probing the proteome of a single
species in its natural context is an even greater challenge.
To address these difficulties, we describe here a versatile method
for cell-selective protein labeling in mixed cellular environments.
To achieve selective labeling, we used noncanonical amino acids
that are excluded by the endogenous protein synthesis machinery
(Fig. 1a). These amino acids face discrimination by the quality control
mechanisms found at the level of aminoacyl-tRNA synthetases10; they
are not charged to tRNA and are not used in protein synthesis. By
screening libraries of methionyl-tRNA synthetase (MetRS) mutants
from Escherichia coli11,12, we have identified a mutant synthetase
(NLL-MetRS) (Supplementary Fig. 1) that efficiently appends azidonorleucine (2, Fig. 1b) to cognate tRNA. Cells bearing the mutant
MetRS are able to use 2 as a surrogate for methionine (1) in protein
synthesis. Wild-type cells are inert to 2; proteins made in these cells
use only methionine and are not labeled (Fig. 1a). In coculture, protein labeling is restricted to mutant cells.
To validate this approach, we first confirmed that incorporation of
2 into newly synthesized proteins is dependent on expression of the
mutant synthetase. An E. coli strain (DH10B/pJTN1) constitutively
expressing a plasmid-borne copy of NLL-MetRS was pulse labeled
with 2 and compared to a control strain (DH10B/pQE-80L) that did
not express the enzyme. Separate cultures of the two strains were
grown in minimal medium containing the 20 canonical amino acids.
When the cell density reached an optical density at 600 nm (OD600)
of 0.5, cells were pulse labeled with 1 mM 2 for 10 min. Control
cells were pulsed in the same fashion with 1, or incubated with the
protein synthesis inhibitor chloramphenicol before labeling with 2.
Cell lysates of each culture were probed for incorporation of 2 via
Cu(I)-catalyzed ligation13,14 to biotin-FLAG-alkyne (4) followed
by western blotting with protein detection by anti-FLAG antibody
(Supplementary Methods). The results of these experiments indicated that only proteins synthesized in cells constitutively expressing
the NLL-MetRS (DH10B/pJTN1) were labeled with 2 and susceptible
to ligation to 4 (Supplementary Fig. 2). A second control strain, in
which the wild-type synthetase was overexpressed, was also inert to
labeling (Supplementary Fig. 1).
The behavior observed in separate cultures was maintained when
cells were incubated in coculture to simulate a complex, mixed cellular environment. Two different heterologous proteins were used as
markers for the cells of origin to distinguish between NLL-expressing
and wild-type E. coli. An E. coli strain (DH10B/pJTN2) expressing
the NLL-MetRS was programmed to express green fluorescent protein (GFP) upon induction with IPTG. The control strain (DH10B/
pJTN3) carried an IPTG-inducible gene for the marker protein
dihydrofolate reductase (DHFR). Both marker proteins were tagged
with His6 to enable Ni-affinity purification and detection with His5
antibody. Individual cultures of these bacterial strains were grown
to OD600 = 1.0, and a third culture was created by mixing cells in a
1Division
of Chemistry and Chemical Engineering and 2Division of Biology, California Institute of Technology, Pasadena, California, USA. Correspondence should be
addressed to D.A.T. (tirrell@caltech.edu).
Received 28 October 2008; accepted 28 May 2009; published online 9 August 2009; doi:10.1038/nchembio.200
NATURE CHEMICAL BIOLOGY
VOLUME 5 NUMBER 10 OCTOBER 2009
715
B R I E F C O M M U N I C AT I O N S
Labeling of individual and mixed cultures was
performed as described earlier. After pulse
a
labeling with 2 and induction of protein synthesis, cells were collected by centrifugation
Cell expressing
mutant MetRS
and washed before Cu(I)-catalyzed labeling
React with probes
Cell-selective
of cells with dimethylaminocoumarin-alkyne
specific for 2
labeling with 2
(5). After washing with phosphate-buffered
saline to remove excess dye, cells were imaged
by fluorescence microscopy. The results
(Fig. 2b) were consistent with those of the
Noncanonical amino acid (2)
Proteins are not
western analysis; the coumarin fluorescence
Wild-type cell
labeled
was confined to cells that express NLL-MetRS.
Reactive probe
Control cells expressing GFP were inert with
O
O
N
N
H HN
O
respect to labeling, as indicated by the absence
N
N
b
NH
HN
N
N
of coumarin emission from these samples
S
N
H
N
O
O
O
S
N
(Supplementary Fig. 3).
H
N
O
Cell-selective protein labeling can also be
OH
GGADYKDDDDK
OH
OH
OH
O
N
H N
H N
H N
H N
H
accomplished
in systems containing mixtures
O
O
O
O
O
N
O
H
of bacterial and mammalian cells. Mouse
5
6
4
1
2
3
alveolar macrophages were infected with
E. coli cells that constitutively express the
Figure 1 Cell-selective labeling of proteomes with azidonorleucine. (a) Schematic representation of
NLL-MetRS (DH10B/pJTN1) or with control
incorporation of azidonorleucine exclusively in cells expressing NLL-MetRS. (b) Structures of amino
acids and probes used in this study: methionine (1), azidonorleucine (2), azidohomoalanine (3),
bacterial cells that express a GFP marker probiotin-FLAG-alkyne (4), dimethylaminocoumarin-alkyne (5) and TAMRA-alkyne (6).
tein (DH10B/pJTN4). Before infection, 2 mM
2 was added to the macrophage medium; to
initiate infection, bacteria were added to the
volumetric ratio of 1:2 (DH10B/pJTN2:DH10B/pJTN3). To initiate culture medium and co-incubated for 35 min at 37 °C. Cells were
labeling, 1 mM 2 was added to each of the three cultures, and expres- fixed, permeabilized and subjected to Cu(I)-catalyzed conjugation to
sion of marker proteins was induced with 1 mM IPTG for 3 h. Cell TAMRA-alkyne (6, Invitrogen) (Supplementary Methods). Bacteria
lysates from all three samples were subjected to Cu(I)-catalyzed azide- were both bound and internalized by macrophages, as confirmed by
alkyne ligation to 4, and marker proteins were isolated by Ni-affinity confocal microscopy and three-dimensional analysis (Supplementary
purification (Supplementary Methods). Western analysis (Fig. 2a) Movie 1). Macrophage-associated bacterial cells that express the NLLof isolated proteins with His5 antibody revealed the expected expres- MetRS exhibited strong fluorescence emission from 6 (Fig. 3a). The
sion patterns from the three cultures. In contrast, analysis of blots control bacterial strain was bound and internalized by macrophages
with streptavidin-HRP (for detection of conjugation to 4) revealed (as seen by detection of GFP, Supplementary Fig. 4) but exhibited no
exclusive modification of GFP, the marker protein synthesized in cells conjugation to 6 above background (Fig. 3a). To confirm protein synexpressing NLL-MetRS. DHFR isolated from control cells exhibited thesis by macrophages during the infection period, cells were treated
no such modification. N-terminal protein sequencing indicated with azidohomoalanine (3) in medium lacking 1 (Fig. 3b). Both 2 and
10–20% replacement of 1 by 2 in the GFP marker protein.
3 are susceptible to ligation to alkyne-functionalized probes; however,
To demonstrate further the utility of this approach, we used fluo- in contrast to 2, 3 is activated by wild-type MetRS and does not disrescence microscopy to distinguish proteins made in cells express- criminate between cell types4. As shown in Figure 3, both bacterial
ing the NLL-MetRS from those made in cells that do not express the cells and macrophages were labeled with 3, whereas labeling with 2
mutant synthetase. The control strain (DH10B/pJTN4) expressed an was observed only in bacterial cells that express the NLL-MetRS.
IPTG-inducible GFP, whereas the strain (DH10B/pJTN5) constituNewly synthesized bacterial proteins can be enriched from such cultively expressing the NLL-MetRS carried an IPTG-inducible DHFR. tures by affinity chromatography. Using an E. coli strain that expresses
2
2
2
F i g u r e 2 Cell-selective protein labeling in mixed populations. (a) Western
blot detection of marker protein expression with His 5 antibody (left)
and with streptavidin-HRP (right). DHFR was made in cells lacking the
NLL-MetRS (wild type, WT), and GFP was made in cells expressing the
NLL-MetRS (NLL); both proteins contain multiple ATG (methionine)
codons (7 in DHFR and 8 in GFP). Azidonorleucine was added to all three
cultures upon induction of protein synthesis. Samples from the mixed
culture contain both proteins (as shown by His5 antibody detection), but
only GFP is labeled with azidonorleucine and susceptible to labeling with
4. (b) Fluorescence images of cells expressing GFP but not NLL-MetRS
(WT), cells expressing DHFR and NLL-MetRS (NLL) and a mixed culture
of the two (mixed). Azidonorleucine was added to all three cultures upon
induction of protein synthesis. Cells from all three cultures were treated
with 5, but only cells expressing the NLL-MetRS were labeled. Note that
the marker proteins co-expressed with NLL-MetRS are different in the
western blotting and fluorescence imaging experiments.
716
a
WT
NLL
Mixed
WT
NLL
Mixed
His6-GFP
His6-DHFR
Streptavidin-HRP
His5 antibody
b
Wild-type GFP
Constitutive NLL
Mixed
GFP/coumarin overlay
2
10 µm
VOLUME 5 NUMBER 10 OCTOBER 2009
NATURE CHEMICAL BIOLOGY
B R I E F C O M M U N I C AT I O N S
Mitotracker
TAMRA
b
Overlay
Mitotracker
TAMRA
Overlay
Wild type
Constitutive NLL
a
10 µm
Wild type
c
L
FT
W1
W3
W5
E
β-actin
Bacterially
expressed
GFP
10 µm
Figure 3 Cell-selective labeling in mixtures of bacterial and mammalian cells. (a) Fluorescence images of mixed cultures containing bacteria attached
to or internalized by mouse alveolar macrophages. Infection was performed in medium containing azidonorleucine. Bacterial cells constitutively
expressing the NLL-MetRS were labeled by TAMRA-alkyne (constitutive NLL), whereas cells lacking the NLL-MetRS (WT GFP) are visible only in the
GFP channel (Supplementary Fig. 4). Macrophages were labeled with Mitotracker Deep Red (Invitrogen) and exhibited very low TAMRA background
emission. In all cases, conjugation of TAMRA-alkyne was confined to bacterial cells expressing the NLL-MetRS. (b) Fluorescence images of macrophage
infection with wild-type bacteria performed in the presence of azidohomoalanine. Protein synthesis by macrophages is indicated by strong TAMRAalkyne emission from both bacterial cells and macrophages. (c) Macrophages were infected with bacterial cells that express GFP under induction with
IPTG and that constitutively express the NLL-MetRS. Infection was performed in medium containing IPTG and azidonorleucine to facilitate bacterial
synthesis and labeling of GFP. Total cell lysate from the infection was subjected to conjugation with alkyne-functionalized biotin; labeled proteins
were enriched with streptavidin avidity. Bacterially expressed GFP and mammalian β-actin were followed by immunoblots. Analyses of the lysate (L),
unbound flow-through (FT), washes (W1, W3, W5) and eluent (E) reveal a separation of bacterial and mammalian representative proteins. Bacterially
expressed GFP was labeled with azidonorleucine and thus subject to conjugation to biotin and enrichment with streptavidin. Proteins originating from
macrophages, including β-actin, were not labeled with 2 and therefore were not conjugated to alkyne-functionalized biotin.
the NLL-MetRS constitutively and GFP under induction with IPTG,
we infected macrophages in medium containing 2 mM 2. Immediately
upon infection, IPTG was added to initiate bacterial synthesis of
GFP. After 35 min at 37 °C, the total cell mixture was collected by
centrifugation and lysed. Proteins were subjected to Cu( I)catalyzed azide-alkyne ligation with alkyne-functionalized biotin
(Supplementary Methods). Biotinylated proteins were selectively
enriched by collection on neutravidin-agarose beads. After five
washes, proteins were eluted from the resin with 2 mM free
biotin and 2% SDS (w/v). To examine the extent of enrichment,
the lysate, resin flow-through, washes and eluent were analyzed
by immunoblot (Fig. 3c). The mammalian protein β-actin was
detected with anti-β-actin and served as a representative macrophage protein. The bacterial marker GFP was detected with
anti-His5 antibody. No actin was detected in the eluent, which
indicates at least 50-fold depletion of the mammalian marker.
In contrast, comparison of the GFP band intensities in the
eluent and lysate confirmed good recovery of the affinity-tagged
bacterial protein.
The results described here illustrate the use of mutant aminoacyltRNA synthetases to enable cell-specific protein labeling with noncanonical amino acids. In mixed cellular systems, newly synthesized
proteins in selected cells can be labeled with affinity reagents or fluorescent dyes for enrichment, identification and visualization. This
approach will enable unambiguous determination of the cellular origins of proteins made in complex multicellular systems and will provide new insight into intercellular communication. We are expanding
the studies described here by engineering new amino acid/synthetase
NATURE CHEMICAL BIOLOGY
pairs and by using the azidonorleucine/NLL-MetRS pair to examine
a variety of intercellular interactions.
Note: Supplementary information and chemical compound information is available
on the Nature Chemical Biology website.
ACKNOWLEDGMENTS
We gratefully acknowledge the US National Institutes of Health (R21 DA020589
and R01 GM62523) and the US Army Research Office Institute for Collaborative
Biotechnologies for support of this work. We thank K. Boulware for assistance in
generating the three-dimensional movie of infected macrophages. E.M.S. is an
investigator of the Howard Hughes Medical Institute.
Published online at http://www.nature.com/naturechemicalbiology/.
Reprints and permissions information is available online at http://npg.nature.com/
reprintsandpermissions/.
1. Beynon, R.J. & Pratt, J.M. Mol. Cell. Proteomics 4, 857–872 (2005).
2. Mann, M. Nat. Rev. Mol. Cell Biol. 7, 952–958 (2006).
3. Dieterich, D.C., Link, A.J., Graumann, J., Tirrell, D.A. & Schuman, E.M. Proc. Natl.
Acad. Sci. USA 103, 9482–9487 (2006).
4. Dieterich, D.C. et al. Nat. Protoc. 2, 532–540 (2007).
5. Prescher, J.A. & Bertozzi, C.R. Nat. Chem. Biol. 1, 13–21 (2005).
6. Beatty, K.E., Xie, F., Wang, Q. & Tirrell, D.A. J. Am. Chem. Soc. 127, 14150–14151
(2005).
7. Beatty, K.E. et al. Angew. Chem. Int. Ed. 45, 7364–7367 (2006).
8. Zhang, C.G., Chromy, B.A. & McCutchen-Maloney, S.L. Expert Rev. Proteomics 2,
187–202 (2005).
9. Macpherson, A.J. & Harris, N.L. Nat. Rev. Immunol. 4, 478–485 (2004).
10. Ibba, M. & Söll, D. Science 286, 1893–1897 (1999).
11. Link, A.J. et al. Proc. Natl. Acad. Sci. USA 103, 10180–10185 (2006).
12. Yoo, T.H. & Tirrell, D.A. Angew. Chem. Int. Ed. 46, 5340–5346 (2007).
13. Tornøe, C.W., Christensen, C. & Meldal, M. J. Org. Chem. 67, 3057–3064 (2002).
14. Lewis, W.G. et al. Angew. Chem. Int. Ed. 41, 1053–1057 (2002).
VOLUME 5 NUMBER 10 OCTOBER 2009
717