-
Rapid labeling of intracellular His-tagged proteins inliving
cellsYau-Tsz Laia,1, Yuen-Yan Changa,1, Ligang Hua,1, Ya Yanga,
Ailun Chaoa, Zhi-Yan Dub, Julian A. Tannerc, Mee-Len Chyeb,Chengmin
Qianc, Kwan-Ming Nga, Hongyan Lia, and Hongzhe Suna,2
Departments of aChemistry and cBiochemistry, bSchool of
Biological Sciences, University of Hong Kong, Pokfulam, Hong Kong,
People’s Republic of China
Edited by Kenneth N. Raymond, University of California,
Berkeley, CA, and approved February 3, 2015 (received for review
October 12, 2014)
Small molecule-based fluorescent probes have been used for
real-time visualization of live cells and tracking of various
cellularevents with minimal perturbation on the cells being
investigated.Given the wide utility of the (histidine)6-Ni
2+-nitrilotriacetate(Ni-NTA) system in protein purification,
there is significant interestin fluorescent Ni2+-NTA–based probes.
Unfortunately, previous Ni-NTA–based probes suffer from poor
membrane permeability andcannot label intracellular proteins. Here,
we report the design andsynthesis of, to our knowledge, the first
membrane-permeablefluorescent probe Ni-NTA-AC via conjugation of
NTA with fluoro-phore and arylazide followed by coordination with
Ni2+ ions. Theprobe, driven by Ni2+-NTA, binds specifically to
His-tags geneti-cally fused to proteins and subsequently forms a
covalent bondupon photoactivation of the arylazide, leading to a
13-fold fluo-rescence enhancement. The arylazide is indispensable
not only forfluorescence enhancement, but also for strengthening
the bindingbetween the probe and proteins. Significantly, the
Ni-NTA-ACprobe can rapidly enter different types of cells, even
plant tissues,to target His-tagged proteins. Using this probe, we
visualized thesubcellular localization of a DNA repair protein,
Xeroderma pigmen-tosum group A (XPA122), which is known to bemainly
enriched in thenucleus.We also demonstrated that the probe can
image a geneticallyengineered His-tagged protein in plant tissues.
This study thus offersa new opportunity for in situ visualization
of large libraries of His-tagged proteins in various prokaryotic
and eukaryotic cells.
His-tagged protein | Ni-NTA | fluorescent probe | live cell |
photoactivation
Chemical and biochemical labeling of proteins can
elucidateprotein function, localization, and dynamics as well as
otherbiological events in live cells (1–3). Small molecule-based
fluo-rescent labeling of recombinant proteins holds particular
prom-ise as an alternative to fluorescent protein fusion technology
(4, 5)without deleterious perturbation of protein functions. A
repre-sentative technique is small tag-based fluorescent imaging,
in whicha protein of interest is genetically fused with a short
peptide thatbinds site-specifically to a designed synthetic
fluorescent probe.Over recent decades, tremendous progress has been
made in usingsmall molecule-based probes to monitor cellular events
(6, 7). Inparticular, metal-chelation labeling of a protein is
attractive owingto its simplicity and high specificity. Among these
probes, FlAsHand its derivatives, including ReAsH and SplAsH, are
successfulsmall molecule-based metal-containing probes that can
light upintracellular proteins fused with a tetracysteine motif (1,
8, 9).However, this system suffers from drawbacks, including a
highbackground that requires extensive washing (10), and
unsuitabilityin a cellular oxidizing environment (11).
Nevertheless, the pioneeringdevelopment of biarsenical-based
fluorescent probes inspiredresearchers to design various probes
that target other tag-ging systems.Given the wide utility of the
(histidine)6-Ni
2+-nitrilotriacetatesystem (Ni-NTA) in molecular biology and
biotechnology foraffinity chromatography-based protein
purification, this systemhas also been exploited to
site-selectively label large libraries ofexisting
hexahistidine-tagged (His-tagged) proteins via conjuga-tion with
fluorophores (12–16). Various NTA-based fluorescent
probes have been developed via conjugation of fluorophoreswith
mono-NTA (12, 17) or di-, tri-, and tetra-NTA derivativesto either
mimic the concept of FlAsH or overcome the weakbinding nature of
His-tag with Ni2+-NTA (Kd = 13 μM) (11, 18–21). Despite significant
increases in stability of the multiplechelator heads/His-tag
adducts compared with mono-NTA,negative charges of these moieties
might prevent them fromentering cells. Indeed, all reported
Ni-NTA–based fluorescentprobes that target His-tagged proteins are
exclusively limited tolabel membrane proteins because none can
cross cell membranesto label intracellular proteins in live cells
(15, 17). A previousclaim that a di-NTA dibromobimane conjugate
entered cells totarget polyhistidine-containing proteins is not
convincing be-cause the fluorescence measurement was carried out
using wholecells and results were not corroborated with in vivo
imaging data(22). It is thus possible that the observed
fluorescence upon treat-ment of whole cells with the probe could be
derived from in-teraction with polyhistidine-containing proteins on
cell membranes.Recently a di-NTA–based fluorescent probe containing
an
α-chloroacetamide that targets the Cys-appended His-tag
(Cys-His6-tag) was reported to be able to label intracellular
Cys-His6–tagged proteins in live cells (16). However, the di-NTA
fluorophoreconjugate itself could hardly enter cells, unless the
conjugate wasattached to a cargo consisting of a cell-penetrating
peptide andfluorescent quencher (a dabsyl-appended icosapeptide
with a dabsylgroup, a tetra-His, and an octa-Arg), enforcing the
probe to entercells to label Cys-His6–tagged instead of His6-tagged
proteins.
Significance
The hexahistidine-Ni2+-NTA system is used extensively in
pro-tein purification, and large numbers of His-tagged protein
li-braries exist worldwide. The application of this tagging
systemto image proteins in live cells would offer significant
oppor-tunities to track cellular events with minimal spatial
andfunctional perturbation on a protein of interest.
However,previously reported Ni-NTA–based probes suffer from
poormembrane permeability and have been limited to label mem-brane
proteins only. Here we report, to our knowledge, thefirst small
fluorescent probe, Ni-NTA-AC, that can rapidly crosscell membranes
to specifically target His-tagged proteins invarious types of live
cells, even in plant tissues. The probe willprovide new
opportunities for in situ analysis of variouscellular events.
Author contributions: Y.-T.L., Y.-Y.C., L.H., Y.Y., A.C.,
Z.-Y.D., M.-L.C., C.Q., H.L., and H.S.designed research; Y.-T.L.,
Y.-Y.C., L.H., Y.Y., A.C., Z.-Y.D., and K.-M.N. performed
re-search; Y.-T.L., Y.Y., and A.C. performed syntheses; Y.-T.L.,
Y.-Y.C., L.H., Z.-Y.D., J.A.T.,M.-L.C., C.Q., H.L., and H.S.
analyzed data; and Y.-T.L., Y.-Y.C., L.H., M.-L.C., H.L., andH.S.
wrote the paper.
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.1Y.-T.L., Y.-Y.C., and
L.H. contributed equally to this work.2To whom correspondence
should be addressed. Email: [email protected].
This article contains supporting information online at
www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental.
2948–2953 | PNAS | March 10, 2015 | vol. 112 | no. 10
www.pnas.org/cgi/doi/10.1073/pnas.1419598112
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://crossmark.crossref.org/dialog/?doi=10.1073/pnas.1419598112&domain=pdfmailto:[email protected]://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplementalhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplementalwww.pnas.org/cgi/doi/10.1073/pnas.1419598112
-
Furthermore, the probe is relatively large and complicated to
make.The consequential high cost of application precludes its use
in la-beling many existing His-tag libraries of proteins. Moreover,
therelatively slow labeling (80% yield within an hour) also hinders
itsuse for real-time imaging. Therefore, it is preferable to design
smalland simple fluorescent probes with good membrane permeability
torapidly label intracellular His-tagged proteins.Herein, we
present the design, synthesis, and application of
a fluorescent probe Ni-NTA-AC that exhibits excellent mem-brane
permeability and can rapidly enter cells to image
intracellularHis-tagged proteins (Fig. 1A). The probe targets a
His-taggedprotein specifically through Ni2+-NTA with ∼13-fold
fluorescence“turn-on” responses. An arylazide was incorporated into
the probeinitially with the purpose of overcoming weak binding
betweenNi2+ and histidines. Unexpectedly, the arylazide was also
in-dispensable for fluorescence enhancement. We have demon-strated
that our probe could image His-tagged proteins in differenttypes of
cells, even in plant tissues. The ability to rapidly
visualizeintracellular proteins genetically fused with a His-tag
offers greatpotential for spatial and functional analysis of
libraries of existingHis-tagged proteins in different types of live
cells.
Results and DiscussionDesign and Synthesis of a Fluorescent
Probe Ni-NTA-AC. Previouslymany fluorescent probes used the Ni-NTA
system, includingdi-, tri-, or tetra-NTA derivatives conjugated
with fluorophores,but could only label membrane proteins due to
poor mem-brane permeability (11, 18). We reasoned that negative
chargesmight prohibit crossing of cell membranes, although
introductionof multi-NTA to the probes did overcome the weak
binding of
Ni2+-NTA to the His-tag. We therefore designed a probe (Fig.1A)
consisting of a mono-Ni-NTA moiety, a small membrane-permeable
fluorophore (a coumarin derivative) (23), and anarylazide moiety.
Ni2+-NTA targets the His-tag to achieve specificlabeling of a
protein of interest, and the arylazide group providesan additional
covalent bond between the probe and its targetprotein upon
photoactivation (24) to resolve the intrinsic weakbinding of
Ni2+-NTA to His-tag. A linker between mono-Ni-NTAand the
fluorophore was designed to allow flexibility in
facilitatingefficient protein labeling.We first synthesized a
coumarin-based ligand NTA-AC via
a three-step synthesis with an overall yield of 64% (Fig. 1A
andSI Appendix, Figs. S1 and S2) by conjugating the NTA moietywith
a coumarin fluorophore and arylazide, and the purity ofNTA-ACwas
confirmed by both NMR and electrospray ionization massspectrometry.
The ligand exhibited amaximumabsorption at∼342nm (e = 11,100
M−1·cm−1) and emitted at 448 nm (Φ = 0.056; SIAppendix, Fig. S3).
The lowquantumyield ofNTA-AC is attributedto the presence of an
azide at the seventh position of the coumarinmoiety, which quenches
the fluorescence of coumarin as reportedpreviously (25, 26). The
probeNi-NTA-ACwas then generated bysubsequent incubation of NTA-AC
with Ni2+ (as NiSO4) inbuffered aqueous solution. As shown in Fig.
1B, upon addition ofequimolar amounts of Ni2+ to NTA-AC, the
fluorescence wassignificantly quenched by ∼70%, in sharp contrast
with the 5%reduction observed in previously reported NTA-DCF
conjugate(17), thus Ni-NTA-AC only has very weak emission at 448
nm. Thetitration data were nonlinearly fitted using the
Ryan–Weberequation (27), which gave rise toKd of 38± 13 nM. To
evaluate thebinding stoichiometry, a Job’s plot was constructed by
monitoringfluorescence changes upon complexation ofNTA-AC with Ni2+
at448 nm excited at 342. Maximum fluorescence changes wereobserved
at amolar ratio of Ni2+:NTA-AC of 0.5, indicative of theformation
of Ni-NTA-AC complex with a ratio ofNTA-AC:Ni2+ of1:1 (Fig. 1C);
this was further verified in electrospray ionizationmass spectrum
(a peak at m/z 558.6), in agreement with the cal-culated value of
558.9.
Evaluation of Ni-NTA-AC Probe in Labeling His-Tagged Proteins
inVitro. To examine the feasibility that Ni-NTA-AC can labela
His-tagged protein in vitro, we applied the functional domainof
Xeroderma pigmentosum group A (XPA122) as a showcasestudy.
Xeroderma pigmentosum group A serves as a classic formof XP
proteins, which is important for repairing DNA damagecaused by UV
radiation (λ ≤ 310 nm); and the functional do-main, XPA122, serves
as the site of damaged-DNA bindingto initiate DNA repair (28).
Proteins with (denoted as His-XPA122) or without (XPA122)
genetically fused His-tag at itsN terminus were overexpressed and
purified as described pre-viously (SI Appendix) (29). The
interaction of Ni-NTA-AC withthe protein was first investigated by
fluorescence spectroscopy.Incubation of 10-M equivalents of
His-XPA122 with Ni-NTA-ACled to an increase in fluorescence
intensity with time, reachinga plateau at ∼9 min, where an
approximate 13-fold increase influorescence intensity was observed
(Fig. 2A). In contrast, noobvious fluorescence changes (less than
50% increases) werenoted upon mixing of Ni-NTA-AC with XPA122 under
identicalconditions (SI Appendix, Fig. S4). Similarly, premixing of
the li-gand NTA-AC (without coordination of Ni2+) with
His-XPA122under identical conditions resulted in no fluorescence
enhance-ment at all (SI Appendix, Fig. S5). These combined results
in-dicate that Ni-NTA-AC selectively targets the His-tag of
theprotein through Ni2+, resulting in fluorescence turn-on
responses.Nonspecific binding is negligible under the conditions
used. Al-though the underlying mechanism of fluorescence turn-on
responsesof the probe toward His-tagged proteins is not fully
understood, itis possible that the sandwich-like structure of the
probe withNi2+-NTAmoiety weakly bound to the fluorophore is
abolished upon
Fig. 1. (A) Schematic diagram demonstrating intracellular
labeling of His-tagged proteins using Ni-NTA-AC (Top) and the
synthetic scheme of NTA-AC(Bottom). The probe enters the cells
rapidly and targets His-tagged proteinswith significant
fluorescence turn-on. (B) Normalized fluorescence changesof NTA-AC
(5 μM) upon addition of Ni2+ (as NiSO4). Approximately 70%decrease
in fluorescence is noted upon Ni2+ chelation. (C) Job’s plot of
thefluorescence changes (λex = 342 nm, λem = 448 nm) upon
complexation ofNTA-AC with Ni2+ in Tris buffer. Maximum
fluorescence changes were ob-served at a molar ratio of Ni2+:NTA-AC
of 0.5, indicative of the formation ofNi-NTA-AC complex with a
ratio of NTA-AC: Ni2+ of 1:1.
Lai et al. PNAS | March 10, 2015 | vol. 112 | no. 10 | 2949
CHEM
ISTR
YCE
LLBIOLO
GY
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdf
-
binding to His-tagged proteins (30), then subsequent
interactionof the fluorophore with protein targets might lead to
fluores-cence enhancement. Moreover, the arylazide is also likely
to playan important role in fluorescence turn-on because it was
reportedpreviously that attachment of arylazide to coumarin at the
seventhposition led to fluorophore quenching. However, such an
effect isabolished upon the reduction of arylazide (25, 31).The
binding properties of Ni-NTA-AC toward His-XPA122
were also studied by isothermal titration calorimetry
(ITC),which gave rise to a dissociation constant of 7.1 ± 0.6 μM
andbinding capacity of 1.4 (SI Appendix, Fig. S6A), consistent
withthe weak binding strength of Ni2+-NTA to histidine residues.The
nonintegral stoichiometry of Ni-NTA-AC binding to the His-tagged
protein is attributable to the fact that one His-tag possiblybinds
1–2 Ni2+ ions (32, 33), verified by our MS data (Fig. 3C).The weak
binding might lead to dissociation of the probe fromlabeled
proteins in the complex environments of live cells. Totackle this
problem, we incorporated an arylazide with a ratio-nale of
providing additional binding between the probe andtargeted proteins
upon photoactivation. The role of arylazidewas examined
subsequently. First, we examined the time re-quired for arylazide
activation. A mixture of equimolar amountsof His-XPA122 and
Ni-NTA-AC was irradiated with 365-nm UVlight using a 4-W long-wave
compact UV lamp (720 μW/cm2) atdifferent time intervals. Samples
were analyzed by SDS/PAGEgel electrophoresis and fluorescence
signals were quantified. Asshown in SI Appendix, Fig. S7, a
fluorescent band could obviouslybe seen after 1 min of UV exposure
and the fluorescence in-creased with exposure time and maximized at
10 min, suggestingthat the efficiency of photoactivation of
arylazide increasescorrespondingly. A mixture of Ni-NTA-AC and 10-M
equivalentsof His-XPA122 was thus subjected to irradiation under UV
light(365 nm) for 10 min to ensure complete photoactivation
ofarylazide. As expected, over 10-fold fluorescence
enhancementcompared with the fluorescence of Ni-NTA-AC was
observableowing to the probe binding to His-XPA122 (Fig. 2B).
Uponaddition of 40-M equivalents of EDTA to the mixture to strip
offNi2+ from the probe–protein complex, the observed
fluorescenceintensities, instead of being decreased, were slightly
increased(∼30%), attributable to recovery of the fluorescence being
quenchedby Ni2+ (Fig. 2B). Subsequently, we carried out a similar
experimentexcept under darkness to prevent arylazide from
photoactivation.As shown in Fig. 2B, ∼60% decrease in fluorescence
were noted,indicating that without photoactivation of arylazide,
removal ofNi2+ from the probe–His-XPA122 complex abolishes the
probe’sability to image His-tagged proteins.The capability of
arylazide to strengthen binding between the
probe and His-tagged proteins upon photoactivation was
alsoinvestigated by observation of the probe–protein complex
under
denatured conditions. The Ni-NTA-AC–labeled His-XPA122was
irradiated with UV light (365 nm) for 10 min, then subjectedto
SDS/PAGE gel electrophoresis. An intense blue fluorescentband
corresponding to His-XPA122 was observable in the SDS/PAGE,
corroborating with strong binding of the probe to His-XPA122 even
under denatured conditions (Fig. 2C). In contrast,no corresponding
blue fluorescent band was detected whenNi-NTA-AC (12 μM) was mixed
with equimolar amounts ofXPA122 or His-XPA122 in the presence of
40-M equivalentsof EDTA, which removes Ni2+ from the probe (Fig.
2C), inline with the above observations that the probe targets
His-tagof XPA122 through Ni2+ and subsequent photoactivation ofthe
arylazide group strengthens binding between the probeand the
protein.We further examined the role of arylazide within the probe
by
synthesizing a coumarin-based ligand without arylazide
attached—namely, NTA-C, for comparison via a three-step synthesis
(SIAppendix, Figs. S8–S10). Addition of equimolar amounts of
Ni2+
led to ∼50% fluorescence reduction for NTA-C (SI Appendix,
Fig.S11A). Unexpectedly, incubation of Ni-NTA-C with
His-XPA122under similar conditions resulted in no fluorescence
enhancement
Fig. 2. (A) Fluorescence spectra of Ni-NTA-AC (1 μM) at
different time intervals after addition of His-XPA122 (10 μM).
(Inset) Time-dependent fluorescencechanges (λem = 448 nm) of
Ni-NTA-AC upon binding to His-XPA122. (B) Normalized fluorescence
of Ni-NTA-AC incubated with His-XPA122 under variousconditions.
Without photoactivation of arylazide (under dark), addition of
excess EDTA (40-M equivalents) to the mixture of Ni-NTA-AC and
His-XPA122 leadsto the fluorescence decrease by ∼60% owing to
chelation of Ni2+ by EDTA from Ni-NTA-AC, dissociating the probe
from His-XPA122. After binding of Ni-NTA-AC toHis-XPA122 and upon
photoactivation of arylazide, addition of excess EDTA showed a
slight increase (∼30%) in fluorescence, corresponding to
fluorescence re-covered from Ni2+ quenching. (C) SDS/PAGE of
protein labeling by equimolar amounts of Ni-NTA-AC (or Ni-NTA-C)
under different conditions. Lane 1: His-XPA122;lane 2: His-XPA122
in the presence of excess EDTA (50 μM); lane 3: His-XPA122 and
Ni-NTA-C (without arylazide); lane 4: XPA122 (without His-tag).
Fig. 3. Labeling efficiency of Ni-NTA-AC to His-XPA122. (A)
SDS/PAGE ofprotein labeling upon incubation with different amounts
(0–10 M equiv-alents) of Ni-NTA-AC monitored by Coomassie Blue and
fluorescence stain-ing. (B) Labeling yield of Ni-NTA-AC to
His-XPA122 determined by SDS/PAGEin Fig. 3A. (C) MALDI-TOF MS
spectra of His-XPA122 (10 μM) in the absenceand presence of 1- and
2-M equivalents of Ni-NTA-AC. The peak at m/z of14,981 is
assignable to the intact protein (calculated 14,979), and peaks
atm/z of 15,549 and 16,100 correspond to the protein bound to one
and twoprobes, respectively. The labeling efficiency was evaluated
by comparing thepeak areas of intact His-XPA122 and the probe bound
His-XPA122, to be46% and 70%, respectively, upon the addition of 1-
and 2-M equivalents ofthe probes to the protein solutions.
2950 | www.pnas.org/cgi/doi/10.1073/pnas.1419598112 Lai et
al.
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfwww.pnas.org/cgi/doi/10.1073/pnas.1419598112
-
(SI Appendix, Fig. S11B), although Ni-NTA-C still bound
toHis-XPA122 with an affinity similar to the probe Ni-NTA-ACbased
on ITC titration data (SI Appendix, Fig. S6B). Similarly,no evident
blue fluorescence was detected on SDS/PAGE whenNi-NTA-C (12 μM) was
mixed with equimolar amounts of His-XPA122 (Fig. 2C). These results
demonstrate that arylazide iscrucial not only for strengthening
binding between the probeand its targets, but also for significant
fluorescence turn-onresponses toward His-tagged proteins.We further
evaluated the labeling efficiency of Ni-NTA-AC
to His-tagged proteins by monitoring reactivity toward
His-XPA122. His-XPA122 was incubated with 0, 0.2, 0.5, 1.0, 2.0,
5.0,and 10 M equivalents of Ni-NTA-AC, and subsequently
UV-irradi-ated and subjected to SDS/PAGE (Fig. 3A). By comparing
thefluorescent intensities of the protein bands with respect to
thoseof Coomassie blue staining, we determined that 1-M
equivalentof Ni-NTA-AC could label ∼50% of His-XPA122 (Fig. 3B).
Wealso examined labeling efficacy using MALDI-TOF mass
spec-trometry. Equimolar amounts of Ni-NTA-AC and His-XPA122were
incubated and then subjected to UV irradiation, followedby MALDI-MS
analysis. As shown in Fig. 3C, two peaks at m/z14,981 and 15,549
were observed, corresponding to the intactprotein and the protein
with one probe bound (calculated 14,979and 15,541, respectively).
The intensity of the peak at m/z 15,549further increased upon
incubation with 2-M equivalents of theprobe to the protein
solution. A weak peak at m/z 16,100 wasassignable to the protein
with two probes bound. By comparingthe peak areas, we found that
∼46% and 70% of His-XPA122were labeled upon addition of 1- and 2-M
equivalents ofNi-NTA-AC, respectively (Fig. 3C), consistent with
the resultsfrom SDS/PAGE (Fig. 3B).
Evaluation of Ni-NTA-AC Probe Membrane Permeability and
Toxicity.We then investigated the cell permeability of Ni-NTA-AC in
livemammalian cells. We genetically fused a His-tag to the N
ter-minus of red fluorescent protein (RFP; His-RFP) and
transientlyexpressed this fusion protein in HeLa cells. The
fluorescence
responses to the His-RFP transfected cells upon administrationof
Ni-NTA-AC were monitored by confocal microscopy using
redfluorescence as a reference. As shown in Fig. 4 A and B and
SIAppendix, Fig. S12, blue fluorescence appeared upon treatmentof
cells with Ni-NTA-AC and reached saturation within 2 min,indicating
that the probe can rapidly cross cell membranes, andfluorescence
turn-on upon binding to His-RFP can be initiatedquickly; this also
indicates that the fluorescent probe Ni-NTA-ACcan readily label
cytosolic proteins such as His-RFP in HeLa aslong as the proteins
are fused with a His-tag. In contrast, no bluefluorescence was
observed after treatment of cells with NTA-AC,i.e., without Ni2+
coordination, indicating that NTA-AC itselffailed to enter cells
(SI Appendix, Fig. S13). This observation is inline with our
rational design of the probe considering chargemight be important
in determining probe membrane perme-ability. In the absence of
Ni2+, the charged hydrophilic NTAmoiety remains exposed, thereby
prohibiting NTA-AC fromcrossing cell membranes. Upon coordination
of Ni2+ to the NTAmoiety, the overall charge reduces dramatically
and, moreover,a sandwich-like structure is probably formed owing to
the weakinteractions between Ni2+-NTA and the fluorophore as well
asflexible linker (30), leading to Ni2+-NTA moiety to be
“buried,”which enables Ni-NTA-AC to cross the hydrophobic cell
mem-branes. It should also be noted that neither interaction
betweenNTA-AC and histidines nor the fluorescence turn-on effectwas
possible without Ni2+-binding to NTA-AC.The potential toxicity of
the probe in bacterial and mammalian
cells was also examined. The viability of E. coli reached ∼99
±1% even when 100 μM of Ni-NTA-AC was incubated with thecells (SI
Appendix, Fig. S14). The viability of HeLa cells in-vestigated by
MTT assay showed that over 90% cells remainedalive upon incubation
with 25 and 50 μM Ni-NTA-AC, againconfirming that the probe
exhibits no toxicity toward the cellsunder the conditions used (SI
Appendix, Fig. S15). We also ex-amined the potential toxicity of UV
irradiation to the cells bylight microscope. No morphological
changes were observed un-der illumination conditions over 40 min,
indicating that the
Fig. 4. Ni-NTA-AC labeling of protein in prokaryoticand
eukaryotic cells. (A) Images of
Ni-NTA-AC–labeledHis-RFP–transfected cells at different times.
Ni-NTA-ACenters the cells rapidly and labels intracellular
His-RFPprotein in 2 min. (B) Relative fluorescent intensityplotted
against incubation time. (C) Images of E. colicells with or without
His-XPA122 overexpression aftertreatment with Ni-NTA-AC (10 μM) for
30 min (n = 5).Only cells expressing His-tagged proteins show
bluefluorescence. (Scale bars: 5 μm.) (D) Images of
His-XPA122-transfected HeLa cells after incubation withNi-NTA-AC
(25 μM) for 30min. The signals are enrichedin the nucleus, where
XPA protein is located (29). HeLacells without transfection served
as a control, showingno fluorescence under identical treatments (n
= 5).(Scale bar: 10 μm.) (E) SDS/PAGE and Western blottinganalysis
of cells used for confocal imaging in D. Lane 1:purified His-XPA122
and Ni-NTA-AC; lane 2: cell lysateof HeLa cells without His-XPA122
transfection; lane 3:nuclear extract of His-XPA122 transfected HeLa
cells. Ablue band appeared in the nuclear extract sample
ofHis-XPA122 transfected cells matches with the band ofpurified
His-XPA122, confirming the occurrence of la-beling of His-tagged
protein by Ni-NTA-AC in trans-fected cells but not in untransfected
cells. (F) Images offluorescent labeling of
His-RFP-His-XPA122–transfectedHeLa cells after treatment with
Ni-NTA-AC (25 μM;n = 5). The protein was expressed all over the
cells, incontrast to fluorescent labeling using Ni-NTA-AC. Theblue
and red fluorescence are colocalized and shownin purple in the
overlay image. (Scale bar: 10 μm.)
Lai et al. PNAS | March 10, 2015 | vol. 112 | no. 10 | 2951
CHEM
ISTR
YCE
LLBIOLO
GY
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdf
-
imaging conditions are apparently nontoxic to live cells, and
thusapplicable for live-cell imaging.
Evaluation of Ni-NTA-AC Probe for Labeling of His-Tagged
Proteins inCells. We first investigated the photoactivation
conditions ofarylazide for imaging experiments because this is
crucial forfluorescence enhancement. We illuminated
His-RFP-transfectedHeLa cells with different light sources,
including 405-nm laser,a DAPI filter cube equipped in the
illumination system of theconfocal microscope, and the UV lamp
using 4-W long-wavecompact UV lamp (720 μW/cm2) at different time
intervals be-fore confocal imaging. As shown in SI Appendix, Fig.
S16, HeLacells with His-RFP transfection exhibited the largest blue
fluo-rescence increase upon 1- to 2-min irradiation by a DAPI
filtercube, whereas irradiation for 2 min by UV lamp gave rise
tolower photoactivation efficiency, similar to that activated by
405-nmlaser. Thus, in all cell-imaging experiments, it is not
necessary toapply an external light source for arylazide
photoactivation. In-stead, an illumination system equipped with a
DAPI filter cubeof the confocal microscope was used for optimal
photoactivationwithin 1 min before image capturing. Cellular
autofluorescencewas minimized through optimization of imaging
parameters us-ing the cells without treatment of Ni-NTA-AC.To
examine the feasibility of the probe in imaging His-tagged
proteins in live cells, we investigated the applicability of
Ni-NTA-ACfor labeling of His-XPA122 in E. coli cells. The probe was
in-cubated with E. coli cells overexpressing His-XPA122 for 30
minat 37 °C, then subjected to confocal imaging (n = 5)
afterwashing with Hepes buffer. As shown in Fig. 4C, E. coli
cellsoverexpressing His-XPA122 were stained with blue
fluorescence,and subsequent SDS/PAGE analysis of the cell lysates
showedthat only one protein band with molecular weight of ∼15
kDaexhibited blue fluorescence, verifying that indeed His-XPA122was
the only protein labeled (SI Appendix, Fig. S17). In contrast,cells
without His-XPA122 expression exhibited no blue fluores-cence,
revealing the feasibility of the probe in labeling
intracellularHis-tagged proteins in live bacterial cells. The
viability and mem-brane integrity of these cells were also examined
by propidiumiodide staining. As shown in Fig. 4C, no cells were
stained in red bypropidium iodide, suggesting that they are live
cells.
Imaging His-Tagged Proteins in Live Mammalian Cells Using
Ni-NTA-ACProbe. We further investigated the capability of the probe
to labelHis-tagged proteins inside mammalian cells. HeLa cells with
orwithout His-XPA122 transfection were supplemented with 25
μMNi-NTA-AC in HBSS buffer for 30 min at 37 °C, washed,
andsubjected to confocal imaging. As shown in Fig. 4D, an
intenseblue fluorescence located mainly at the nuclei was observed
onlyin the cells transfected with His-XPA122, but not in those
cellswithout transfection. To further confirm the identity of the
labeledprotein, cells with or without His-XPA122 transfection
weretreated with the probe and irradiated at 365 nm by UV lamp
for10 min to ensure complete photoactivation of arylazide, then
thenuclei of transfected and untransfected HeLa cells were
extrac-ted and concentrated, and subsequently subjected to
SDS/PAGEfor fluorescence imaging and Western blotting (Fig. 4E).
Thepurified His-XPA122 labeled with Ni-NTA-AC (5 μM) was alsoused
for comparison. The blue fluorescent and Western blottingbands from
the nuclei of His-XPA122–transfected cells were similarto those of
purified His-XPA122, confirming that the labeled pro-tein was
indeed His-XPA122 expressed in HeLa cells. In contrast,no
corresponding bands were observed for the nuclei of untrans-fected
cells (Fig. 4E).Fluorescent proteins have been widely used to study
protein
function and localization as well as other biological events in
thephysiological context of living cells when genetically fused to
theprotein of interest (34). However, the use of fluorescent
proteinsmight potentially interfere with the proper localization or
function
of the protein of interest due to its large size (1), in
particular forrelatively small proteins. Therefore, small
molecule-based fluores-cent probes have significant advantages; to
demonstrate this, weincorporated His-tagged RFP into the N terminus
of His-taggedXPA122 to generate a His-RFP-His-XPA122 plasmid, then
trans-fected HeLa cells to investigate the cellular localization of
theprotein under identical conditions. Both blue fluorescence (due
toNi-NTA-AC labeling) and red fluorescence (due to the expressionof
RFP) were observed with colocalization (Fig. 4F). Interestingly,the
fluorescence signals, instead of being enriched in the nucleus
asfound for His-XPA122 (Fig. 4D), were distributed evenly
through-out the cells, suggesting that RFP disrupts XPA122
localization.The perturbation of protein localization may be
attributed to therelatively large size of RFP (27.5 kDa) (5)
compared with XPA122(15 kDa). By contrast, using the Ni-NTA-AC
probe, His-XPA122was found to be enriched in the nucleus, in
agreement with thepreviously reported intracellular localization of
XPA, a protein in-volved in the recognition of DNA damage during
nucleotide exci-sion repair processes (16, 29). Taken together, we
demonstrate thatthe new fluorescent probe Ni-NTA-AC can be
preferentially appliedto track the abundances and localization of
His-tagged proteins inlive mammalian cells, in particular for small
proteins.
Application of Ni-NTA-AC Probe for Imaging of His-Tagged Protein
inPlant Tissues.We finally show that Ni-NTA-AC can label proteinsin
other eukaryotic systems. Transplastomic Nicotiana tabacumvar.
Xanthi (tobacco) plants expressing His-tagged Brassicajuncea
chitinase BjCHI1 (His-BjCHI1) generated as describedpreviously were
tested (35). Protoplasts were extracted fromleaves of 4-wk-old
wild-type and His-BjCHI1 transplastomic to-bacco (SI Appendix, Fig.
S18) according to previously reportedprocedures (36). Upon
incubation of protoplasts with Ni-NTA-AC (10 μM) for 30 min, blue
fluorescence was detected in thechloroplasts (SI Appendix, Fig.
S19), where His-BjCHI1 wasexpressed and subsequently accumulated
(35). Colocalization ofblue fluorescence with red autofluorescence
of chloroplasts byconfocal microscopy was observed only in cells
expressing His-BjCHI1, but not in wild-type cells, indicating
labeling of His-BjCHI1 in the chloroplasts by Ni-NTA-AC (SI
Appendix, Fig.S19). Furthermore, we added Ni-NTA-AC (10 μM) into
PBSbuffer for root immersion of 7-d-old transplastomic
seedlingsovernight, which were then washed and blotted dry before
im-aging. Confocal microscopy of the abaxial surface of
trans-plastomic tobacco leaves revealed blue fluorescence in
tobaccoleaves expressing His-BjCHI1 (Fig. 5), suggesting that the
probewas taken up through the roots of seedlings and the
His-taggedprotein inside living leaves was subsequently detected.
Theseresults clearly demonstrate that the probe, Ni-NTA-AC, can
Fig. 5. Confocal imaging of guard cells from the abaxial surface
of atransplastomic tobacco leaf expressing His-BjCHI1 (Lower) in
comparisonwith the wild-type (Upper) leaf (n = 5). Seedlings (Left)
were incubated inNi-NTA-AC–containing buffer (10 μM) overnight.
Wild-type tobacco servedas a negative control. (Scale bars: 10
μm.)
2952 | www.pnas.org/cgi/doi/10.1073/pnas.1419598112 Lai et
al.
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfwww.pnas.org/cgi/doi/10.1073/pnas.1419598112
-
readily be extended to label His-tagged proteins in
variouseukaryotic cells, including plant tissues.
ConclusionsThe wide prevalence of a (histidine)6-Ni
2+-nitrilotriacetate sys-tem in protein purification has led to
significant interest in thedevelopment of small Ni-NTA–based
fluorescent probes to im-age His-tagged proteins. Given the
existing large libraries of His-tagged proteins, such probes would
have significant impact infunctional investigation of intracellular
proteins under physio-logically relevant conditions. Here, we have
to our knowledgedeveloped the first small cell-permeable
fluorescent probeNi-NTA-AC that enables rapid labeling (2 min) and
observation ofintracellular His-tagged proteins in different types
of live cellsand even in plant tissues with a plethora of potential
applica-tions. The probe exhibits high specificity and labeling
efficiencytoward intracellular His-tagged proteins. Background
stainingresulted from binding of the probe to endogenously
expressedhistidine-rich proteins in certain type of cells might
lead toa lower sensitivity compared with fluorescent protein
labeling,a similar problem that was also encountered by FlAsH.
Never-theless, it will not prohibit the use of the Ni-NTA-AC probe
whenthe protein of interest is overexpressed. Moreover, the
probeposed less perturbation than RFP on protein function and
locali-zation in live cells owing to its small size. Further
efforts to
develop differently colored NTA-AC analogs are underway inour
laboratory. A recently developed green probe also exhibitsexcellent
membrane permeability and turn-on fluorescence uponbinding to
His-RFP in COS-7 cells (SI Appendix, Fig. S20). Thestrategy we
reported here should find many possible applica-tions, such as in
situ visualization of subcellular localization ofproteins, tracking
protein trafficking, and monitoring proteindynamics in live
cells.
Materials and MethodsNTA-AC was synthesized from
2-(7-azido-4-methyl-2-oxo-2H-chromen-3-yl)acetic acid by the
conversion of aromatic amine to arylazide, and
subsequentlyamidation was performed to connect the metal-chelating
NTA to the fluo-rophore, which yielded NTA-AC. NTA-C was
synthesized similarly. Ni-NTA-AC(and Ni-NTA-C) was prepared by the
addition of 1 M equivalent of NiSO4 toNTA-AC (or NTA-C) in buffered
aqueous solution at pH 7.2. Details of thesynthesis of NTA-AC and
NTA-C, preparation of the probe (Ni-NTA-AC),characterization, and
cell experiments are described in SI Appendix.
ACKNOWLEDGMENTS. We thank Prof. Peter J. Sadler for helpful
commentsand acknowledge the assistance of Li Ka Shing Faculty of
Medicine FacultyCore Facility, The University of Hong Kong. Funding
was provided by TheResearch Grants Council of Hong Kong Grants
704612P and 703913P, theUniversity of Hong Kong emerging Strategic
Research Theme (e-SRT) on In-tegrative Biology, the Wilson and
Amelia Wong Endowment Fund, anda university scholarship (to
Y.Y.).
1. Giepmans BNG, Adams SR, Ellisman MH, Tsien RY (2006) The
fluorescent toolbox forassessing protein location and function.
Science 312(5771):217–224.
2. Sletten EM, Bertozzi CR (2009) Bioorthogonal chemistry:
Fishing for selectivity in a seaof functionality. Angew Chem Int Ed
Engl 48(38):6974–6998.
3. Uttamapinant C, Sanchez MI, Liu DS, Yao JZ, Ting AY (2013)
Site-specific protein la-beling using PRIME and chelation-assisted
click chemistry. Nat Protoc 8(8):1620–1634.
4. Tsien RY (1998) The green fluorescent protein. Annu Rev
Biochem 67:509–544.5. Shaner NC, Steinbach PA, Tsien RY (2005) A
guide to choosing fluorescent proteins.
Nat Methods 2(12):905–909.6. Ueno T, Nagano T (2011) Fluorescent
probes for sensing and imaging. Nat Methods
8(8):642–645.7. Marks KM, Nolan GP (2006) Chemical labeling
strategies for cell biology. Nat Methods
3(8):591–596.8. Hoffmann C, et al. (2010) Fluorescent labeling
of tetracysteine-tagged proteins in
intact cells. Nat Protoc 5(10):1666–1677.9. Adams SR, Tsien RY
(2008) Preparation of the membrane-permeant biarsenicals
FlAsH-EDT2 and ReAsH-EDT2 for fluorescent labeling of
tetracysteine-tagged proteins.Nat Protoc 3(9):1527–1534.
10. Stroffekova K, Proenza C, Beam KG (2001) The
protein-labeling reagent FLASH-EDT2binds not only to CCXXCC motifs
but also non-specifically to endogenous cysteine-richproteins.
Pflugers Arch 442(6):859–866.
11. Soh N (2008) Selective chemical labeling of proteins with
small fluorescent moleculesbased on metal-chelation methodology.
Sensors 8(2):1004–1024.
12. Guignet EG, Hovius R, Vogel H (2004) Reversible
site-selective labeling of membraneproteins in live cells. Nat
Biotechnol 22(4):440–444.
13. Meredith GD, Wu HY, Allbritton NL (2004) Targeted protein
functionalization usingHis-tags. Bioconjug Chem 15(5):969–982.
14. Hauser CT, Tsien RY (2007) A hexahistidine-Zn2+-dye label
reveals STIM1 surface ex-posure. Proc Natl Acad Sci USA
104(10):3693–3697.
15. Hintersteiner M, et al. (2008) Covalent fluorescence
labeling of His-tagged proteinson the surface of living cells.
ChemBioChem 9(9):1391–1395.
16. Uchinomiya S, Nonaka H, Wakayama S, Ojida A, Hamachi I
(2013) In-cell covalentlabeling of reactive His-tag fused proteins.
Chem Commun (Camb) 49(44):5022–5024.
17. Goldsmith CR, Jaworski J, Sheng M, Lippard SJ (2006)
Selective labeling of extracel-lular proteins containing
polyhistidine sequences by a fluorescein-nitrilotriacetic
acidconjugate. J Am Chem Soc 128(2):418–419.
18. Jing C, Cornish VW (2011) Chemical tags for labeling
proteins inside living cells. AccChem Res 44(9):784–792.
19. Uchinomiya S, Ojida A, Hamachi I (2014) Peptide tag/probe
pairs based on the co-ordination chemistry for protein labeling.
Inorg Chem 53(4):1816–1823.
20. Lata S, Gavutis M, Tampé R, Piehler J (2006) Specific and
stable fluorescence labelingof histidine-tagged proteins for
dissecting multi-protein complex formation. J AmChem Soc
128(7):2365–2372.
21. Kapanidis AN, Ebright YW, Ebright RH (2001) Site-specific
incorporation of fluores-
cent probes into protein: Hexahistidine-tag-mediated fluorescent
labeling with
(Ni(2+):nitrilotriacetic acid (n)-fluorochrome conjugates. J Am
Chem Soc 123(48):
12123–12125.22. Krishnan B, Szymanska A, Gierasch LM (2007)
Site-specific fluorescent labeling of
poly-histidine sequences using a metal-chelating cysteine. Chem
Biol Drug Des 69(1):
31–40.23. Uttamapinant C, et al. (2010) A fluorophore ligase for
site-specific protein labeling
inside living cells. Proc Natl Acad Sci USA
107(24):10914–10919.24. Melcher K (2004) New chemical crosslinking
methods for the identification of tran-
sient protein–protein interactions with multiprotein complexes.
Curr Protein Pept Sci
5(4):287–296.25. Chen B, et al. (2013) Fluorescent probe for
highly selective and sensitive detection of
hydrogen sulfide in living cells and cardiac tissues. Analyst
(Lond) 138(3):946–951.26. Thevenin BJ, et al. (1992) A novel
photoactivatable cross-linker for the functionally-
directed region-specific fluorescent labeling of proteins. Eur J
Biochem 206(2):
471–477.27. Bai YC, et al. (2008) Ultraviolet absorbance
titration for determining stability con-
stants of humic substances with Cu(II) and Hg(II). Anal Chim
Acta 616(1):115–121.28. Cleaver JE (2005) Cancer in xeroderma
pigmentosum and related disorders of DNA
repair. Nat Rev Cancer 5(7):564–573.29. Kuraoka I, et al. (1996)
Identification of a damaged-DNA binding domain of the XPA
protein. Mutat Res 362(1):87–95.30. Kamoto M, Umezawa N, Kato N,
Higuchi T (2008) Novel probes showing specific
fluorescence enhancement on binding to a hexahistidine tag.
Chemistry 14(26):
8004–8012.31. Chen Y, et al. (2013) A ratiometric fluorescent
probe for rapid detection of hydrogen
sulfide in mitochondria. Angew Chem Int Ed Engl
52(6):1688–1691.32. Valenti LE, De Pauli CP, Giacomelli CE (2006)
The binding of Ni(II) ions to hexahistidine
as a model system of the interaction between nickel and
His-tagged proteins. J Inorg
Biochem 100(2):192–200.33. Knecht S, Ricklin D, Eberle AN, Ernst
B (2009) Oligohis-tags: Mechanisms of binding to
Ni2+-NTA surfaces. J Mol Recognit 22(4):270–279.34. Lam AJ, et
al. (2012) Improving FRET dynamic range with bright green and
red
fluorescent proteins. Nat Methods 9(10):1005–1012.35. Guan Y,
Ramalingam S, Nagegowda D, Taylor PWJ, Chye M-L (2008) Brassica
juncea
chitinase BjCHI1 inhibits growth of fungal phytopathogens and
agglutinates Gram-
negative bacteria. J Exp Bot 59(12):3475–3484.36. Papadakis AK,
Siminis CI, Roubelakis-Angelakis KA (2001) Reduced activity of
anti-
oxidant machinery is correlated with suppression of totipotency
in plant protoplasts.
Plant Physiol 126(1):434–444.
Lai et al. PNAS | March 10, 2015 | vol. 112 | no. 10 | 2953
CHEM
ISTR
YCE
LLBIOLO
GY
Dow
nloa
ded
by g
uest
on
July
2, 2
021
http://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdfhttp://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1419598112/-/DCSupplemental/pnas.1419598112.sapp.pdf