Near-infrared fluorescent proteins.

Abstract

Fluorescent proteins with emission wavelengths in the near-infrared and infrared range are in high demand for whole-body imaging techniques. Here we report near-infrared dimeric fluorescent proteins eqFP650 and eqFP670. To our knowledge, eqFP650 is the brightest fluorescent protein with emission maximum above 635 nm, and eqFP670 displays the most red-shifted emission maximum and high photostability.

Full text

Page 1

brief communications
nature methods |  VOL.7  NO.10 |  OCTOBER 2010 |  827
localization, cell migration, embryogenesis and other studies
involving deep-tissue imaging. Although an infrared protein that
requires biliverdin injection for fluorescence has been reported
recently4, fully genetically encoded markers are preferable. Here
we report near-infrared fluorescent proteins based on the dimeric
far-red fluorescent protein Katushka2.
For whole-body imaging experiments, high reporter expression is
usually required, and low cytotoxicity is crucial5. According to our
experience, Katushka is not cytotoxic, as we have generated stable
lines of transgenic Xenopus laevis2 and Danio rerio (Supplementary
Fig. 1) expressing Katushka in muscle cells at high levels.
However, a recent report has suggested that more optimal
low-toxicity fluorescent protein variants can be selected using
a bacterial expression system5. We performed a similar screen
(Online Methods) and identified a low-cytotoxicity Katushka
variant named Katushka-9-5, which had spectral characteristics
nearly identical to those of Katushka and had minimal negative
impact on bacterial growth, in contrast to Katushka or monomeric
near-infrared fluorescent
proteins
Dmitry Shcherbo1,6, Irina I Shemiakina1,6,
Anastasiya V Ryabova2, Kathryn E Luker3,
Bradley T Schmidt3, Ekaterina A Souslova1,
Tatiana V Gorodnicheva4, Lydia Strukova1,
Konstantin M Shidlovskiy1, Olga V Britanova1,
Andrey G Zaraisky1, Konstantin A Lukyanov1,
Victor B Loschenov2, Gary D Luker3,5 &
Dmitriy M Chudakov1
fluorescent proteins with emission wavelengths in the near-
infrared and infrared range are in high demand for whole-body
imaging techniques. here we report near-infrared dimeric
fluorescent proteins eqfP650 and eqfP670. to our knowledge,
eqfP650 is the brightest fluorescent protein with emission
maximum above 635 nm, and eqfP670 displays the most
red-shifted emission maximum and high photostability.
Fluorescent proteins are reliable in vivo reporters for whole-body
imaging techniques, and the development of far-red light–emitting
proteins offered hope that a further red-shift of the emission is
possible without dramatic loss of brightness1,2. The next goal was
to reach the near-infrared and infrared range, in which living
tissues are even more transparent owing to low absorbance and
light scattering3. Use of such fluorescent proteins for whole-body
imaging would facilitate investigation of metastasis and tumor
1Shemiakin-Ovchinnikov Institute of Bioorganic Chemistry, Russian Academy of Science, Moscow, Russia. 2A.M. Prokhorov General Physics Institute, Russian Academy
of Science, Moscow, Russia. 3Center for Molecular Imaging, University of Michigan Medical School, Ann Arbor, Michigan, USA. 4Evrogen JSC, Moscow, Russia.
5Department of Microbiology and Immunology, University of Michigan Medical School, Ann Arbor, Michigan, USA. 6These authors contributed equally to this work.
Correspondence should be addressed to D.M.C. (chudakovdm@mail.ru).
Received 13 NovembeR 2009; accepted 9 august 2010; published oNliNe 5 septembeR 2010; doi:10.1038/Nmeth.1501
0
0.2
0.4
0.6
0.8
1.0
Wavelength (nm)
Fluorescence (a.u.)
Wavelength (nm)
Absorbance (a.u.)
0
0.2
0.4
0.6
0.8
1.0
300400500600700400500600700800900
d
0
Fluorescence (a.u.)
5001,000
Time (s)
1,5002,000
0.2
0.4
0.6
0.8
1.0
0
eqFP650
eqFP670
eqFP650
eqFP670
eqFP650
eqFP670
mNeptune
E2-Crimson
Time (s)
0100 200 300 400 500 600 700
0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
Fluorescence (a.u.)
eqFP650
eqFP670
mNeptune
E2-Crimson
b
c
f
e
a
Surviving embryos (%)
Katushka-9-5
Katushka
Katushka-9-5
Katushka
0
10
20
30
40
50
60
70
P = 0.05
P = 0.002
E2-Crimson
mNeptune
Neptune
eqFP650
0
2
4
6
8
10
12
Tailbud stage
Tadpole stage
Fluorescence (a.u.)
figure 1 | Cytotoxicity and spectral characteristics. (a) Katushka and
Katushka-9-5 cytotoxicity in microinjected X. laevis embryos.
Percentages of surviving embryos by the tailbud and tadpole stages are
shown (means ± s.d.; n = 5 experiments each performed on 60 embryos).
Significance was analyzed by Student’s t-test. (b) Flow cytometry
analysis of HeLa cells 48 h after transient transfection with vectors
encoding E2-Crimson, mNeptune, Neptune and eqFP650. Fluorescence
brightness was normalized to the relative efficiency of excitation by
the 488 nm laser line used. Emission was collected at 660–700 nm.
Means ± s.d. are shown (n ≥ 3 transfection experiments). (c) Normalized
fluorescence excitation (solid lines) and emission (dashed lines) spectra
for eqFP650 and eqFP670. (d) Normalized absorption spectra for eqFP650
and eqFP670. (e,f) Normalized photobleaching curves for eqFP650,
eqFP670, mNeptune and E2-Crimson using widefield fluorescence
microscopy under metal halide illumination (e) or laser-scanning
confocal microscopy (f). Error bars, s.d. (n = 4 experiments).
© 2010 Nature America, Inc. All rights reserved.

Page 2

828 |  VOL.7  NO.10 |  OCTOBER 2010 |  nature methods
brief communications
(m)Neptune, a recently reported monomeric far-red fluorescent
protein6 (Supplementary Fig. 2). Neither Katushka-9-5 nor
Katushka demonstrated any visible cytotoxicity when expressed
in mammalian HeLa cells (Supplementary Fig. 3). To compare
cytotoxicity of Katushka-9-5 and Katushka in vivo, we micro-
injected plasmids in the animal poles of X. laevis embryos at the
two-cell stage (20 pg per blastomere), and tracked the death rate
by the tailbud stage (stage 25) and by the tadpole stage (stage 42).
These experiments demonstrated lower toxicity of Katushka-9-5
compared to Katushka (Fig. 1a).
Because of its lower toxicity, we selected Katushka-9-5 for the
development of fluorescent protein variants with further red-
shifted emission. Based on the crystal structure of the related
far-red fluorescent protein mKate7 (Protein
Data Bank (PDB): 3BXB), we selected
amino acid positions in proximity to the
chromophore (including Met14, Leu16,
Met44, Thr62, Tyr121, Ser148, Ser165,
Met167, Arg203 and Leu205) for tar-
geted mutagenesis (numbering in accord-
ance with Aequorea victoria GFP, avGFP;
Supplementary Fig. 4). We performed
semisaturated (two to five amino acid
variants) site-directed mutagenesis of these
positions in several combinations and random
mutagenesis. We manually screened bac-
terial libraries containing these mutants
using a fluorescence stereomicroscope
equipped with a 650LP emission filter and
a built-in spectrofluorimeter. We screened
more than a million individual colonies
and selected two red-shifted variants,
named Entacmaea quadricolor (eq)FP650
and eqFP670, that had substitutions
N24G,M44A and M14T,N24G,M44C,S14
8N,S165N, respectively. Notably, Met44 is
also substituted by a small amino acid in
two recently published far-red fluorescent
proteins, mNeptune6 and E2-Crimson8
(Supplementary Fig. 4). This mutation
has been shown to create a cavity filled
by a water molecule forming a hydrogen
bond with the chromophore acylimine
oxygen6 and appears to be a universal
solution for providing a bathochromic
shift of the fluorescence emission for
DsRed-like chromophores.
Characteristics of eqFP650 and eqFP670
are summarized in Table 1 and Figure 1.
Both proteins were characterized by fast
maturation in Escherichia coli at 37 °C and
demonstrated no residual short-wavelength
fluorescence of intermediate or alterna-
tive chromophore forms, in contrast to
E2-Crimson, which exhibits a second
bright blue emission peak, and mNep-
tune, which has a pronounced green peak
(Supplementary Fig. 5). As evidence of
complete chromophore maturation, the
absorbance spectra of freshly purified proteins were single peaks
with minor absorption bands at 340–360 nm characteristic of red
fluorescent proteins (Fig. 1d). In transiently transfected HeLa cells,
eqFP650 produced the brightest signal in its spectral class (Fig. 1b).
In HEK-293T cells that provide high expression, eqFP650 produced
bright signal 14 h after transfection.
eqFP650 and eqFP670 had only one amino-acid substitution in
the outer surface of the beta-barrel (Gly24, characteristic for avGFP
and E2-Crimson, Supplementary Fig. 4) and both inherited the
dimeric nature of Katushka-9-5, as well as its low toxicity in bacte-
rial expression experiments (Supplementary Fig. 2). In eukaryotic
cells, eqFP650 and eqFP670 were evenly distributed and demon-
strated no aggregation or toxicity (Supplementary Fig. 6).
table 1 | Key characteristics of far-red fluorescent proteins
mcherry Katushka
Excitation peak (nm)587
Emission peak (nm) 610
Fluorescence
quantum yield
Molar extinction
coefficient (M−1 cm−1)
at excitation maximum
Brightnessa (a.u.)15,840
Molar extinction
coefficient (M−1 cm−1)
at 635 nm
Quantum yield
in infrared
(700–900 nm)
Brightness in
infraredb (a.u.)
Photostability,
widefieldc (s)
Photostability,
confocalc (s)
pKa4.5
Reference9
rfP639 e2-crimson mneptune eqfP650 eqfP670
588605599
639646 649
0.180.12 0.18
588
635
0.34
592
650
0.24
605
670
0.06 0.22
72,00065,00069,00058,50057,50065,00070,000
22,100
1,700
12,420
~2,000
7,080
12,640
10,350
7,900
15,600
4,300
4,200
15,7001,000
0.040.07~0.05 0.030.05 0.07 0.03
40 119~100379395301 471
601NDND 19216 1901,289
4877 ND14 2967>700
5.5
2
ND
10
ND
8
5.8
6
5.7 4.5
This work This work
Proteins are ordered according to the position of emission maximum. ND, not determined.
aCalculated as the product of the molar extinction coefficient and quantum yield. bCalculated as the product of the extinction
coefficient at 635 nm, quantum yield and emission fraction between 700 nm and 900 nm. cTime to bleach 50% of fluorescence
signal brightness.
a
E2-CrimsonmNeptuneeqFP650
c
Excitation filter, 605/30 nm
b
Excitation filter, 570/30 nm
620
Emission filter wavelength (nm)
20 nm bandwidth
640
660
680
700
720
740
760
780
0
0.2
0.4
0.6
0.8
1.0
Relative fluorescence/
luminescence (a.u.)
Relative fluorescence/
luminescence (a.u.)
Radiant efficiency
*
*
0
0.2
0.4
0.6
0.8
1.0
660
680
Emission filter wavelength (nm)
20 nm bandwidth
700
720
740
760
780
*
*
Katushka
eqFP650
eqFP670
mNeptune
E2-Crimson
eqFP670
mNeptune
E2-Crimson
Katushka
eqFP650
figure 2 | Whole-mouse imaging with IVIS Spectrum system (Caliper). (a) Representative
fluorescence reflectance images (excitation filter, 605/30 nm and emission filter, 660/20 nm) of
mice injected intramuscularly with HEK 293T cells expressing E2-Crimson, mNeptune or eqFP650.
Asterisk denotes background fluorescence in mice injected with E2-Crimson cells. Scale bar, 1 cm.
The color bar indicates radiant efficiency ×10–6; minimum is 0.001, and maximum is 0.006.
(b,c) Fluorescence efficiency from cell implants imaged with 570/30 nm (b) or 605/30 nm
(c) excitation filters and various emission filters, normalized to photons from firefly luciferase to
control for transfection efficiency and numbers of implanted cells. Means ± s.e.m. are shown
(n = 6–10 per point). *P < 0.05 (Student’s t-test) for eqFP650 relative to other proteins.
© 2010 Nature America, Inc. All rights reserved.

Page 3

nature methods |  VOL.7  NO.10 |  OCTOBER 2010 |  829
brief communications
eqFP650 generally retained the high brightness of Katushka and
was characterized by a strong bathochromic shift, with excitation
and emission peaks at 592 nm and 650 nm, respectively (Fig. 1c
and Supplementary Fig. 7). To date, eqFP650 is the brightest flu-
orescent protein with emission maxima above 635 nm and should
be an optimal genetically encoded marker for in vivo imaging.
eqFP670 is characterized by lower brightness but stronger
bathochromic shift, with excitation and emission peaks at 605 nm
and 670 nm. eqFP670 is to our knowledge the first GFP-like
fluorescent protein with such long-wavelength emission, approxi-
mately half of which falls in the infrared part of the spectrum.
Compared to Katushka, eqFP670 had fourfold greater infrared
brightness (Table 1). Furthermore, eqFP670 had high pH stability
(Table 1) and extremely high photostability (Fig. 1e,f) and should
allow for accumulation of the fluorescent signal over long exposure
times. The combination of Asn148 and Asn165 in eqFP670 is unique
and has not been encountered in other fluorescent proteins. This
implies tight packing around the chromophore, which probably
forms the basis of the high photostability and pH resistance.
To investigate fluorescence detection in deep tissues by whole-
animal imaging, we transiently transfected HEK 293T cells
with a plasmids encoding either Katushka, eqFP650, eqFP670,
mNeptune6 or E2-Crimson8 and injected cells intramuscularly
into the gluteal region of mice. We also transfected these cells
with firefly luciferase plasmid to normalize for transfection effi-
ciency and total numbers of injected cells. Representative images
of mice injected with cells expressing E2-Crimson, mNeptune
or eqFP650 showed that eqFP650 had the highest fluorescence
(Fig. 2a), and quantification at various emission wavelengths
showed higher fluorescence from eqFP650 at two excitation
wavelengths (Fig. 2b,c). No cell implant was detectable with
640 nm excitation and infrared detection, probably owing to
the high background signal.
To compare efficiency of infrared imaging, we used another
imaging system. We subcutaneously (2-mm depth) injected equal
portions of mature Katushka, eqFP650, eqFP670, E2-Crimson,
Neptune and mNeptune protein. Injection of protein samples
provided higher local concentration of a fluorescent protein than
injection of transiently transfected cells, allowing for better signal-
to-noise ratio. In this artificial system, all the proteins tested had
comparable infrared signal (Fig. 3). This system, however, did not
account for multiple influences on fluorescent protein signal in
living cells that were present in the in vivo experiment shown in
Figure 2, such as protein maturation rate, protein turnover, mRNA
stability and transcription rate. We did not attempt to quantify
protein expression and do not know how it may have varied.
Although we demonstrated that eqFP650 now may be the pref-
erable near-infrared fluorescent protein for in vivo cell-labeling
experiments, we believe that development of longer-wavelength
fluorescent proteins such as eqFP670 with high brightness and
of true monomeric variants for protein fusions will increase the
sensitivity and capabilities of deep-tissue imaging techniques.
methods
Methods and any associated references are available in the online
version of the paper at http://www.nature.com/naturemethods/.
Accession codes. GenBank: HQ148301 (eqFP650) and HQ148302
(eqFP670).
Note: Supplementary information is available on the Nature Methods website.
acKnowledgments
We thank J.M. Steele for technical assistance with mouse experiments and
B.S. Glick (University of Chicago) for providing the plasmid encoding E2-Crimson.
This work was supported by grants from the Molecular and Cell Biology program
of the Russian Academy of Sciences, the Howard Hughes Medical Institute
(55005618), the US National Institutes of Health (R01CA136553, R01CA136829
and P50CA093990), Rosnauka (02.512.12.2053) and the Russian Foundation for
Basic Research (08-04-01702-à and 10-04-01042).
author contributions
D.S., I.I.S. and L.S. developed fluorescent proteins. A.V.R. and V.B.L.
performed experiments on a Biospec imaging system. K.E.L., B.T.S. and
G.D.L. performed experiments on a Caliper IVIS spectrum imaging system.
A.G.Z. performed experiments on Xenopus embryos. T.V.G. and E.A.S. grew cells
and performed microscopy experiments. K.M.S. generated transgenic zebrafish.
O.V.B. performed flow cytometry analysis. K.A.L., G.D.L. and D.M.C. designed
and planned the project and wrote the manuscript.
comPeting financial interests
The authors declare competing financial interests: details accompany the full-text
HTML version of the paper at http://www.nature.com/naturemethods/.
Published online at http://www.nature.com/naturemethods/.
reprints and permissions information is available online at http://npg.nature.
com/reprintsandpermissions/.

1. Hoffman, R.M. J. Biomed. Opt. 10, 41202 (2005).
2. Shcherbo, D. et al. Nat. Methods 4, 741–746 (2007).
3. Deliolanis, N.C. et al. J. Biomed. Opt. 13, 044008 (2008).
4. Shu, X. et al. Science 324, 804–807 (2009).
5. Strack, R.L. et al. Nat. Methods 5, 955–957 (2008).
6. Lin, M.Z. et al. Chem. Biol. 16, 1169–1179 (2009).
7. Pletnev, S. et al. J. Biol. Chem. 283, 28980–28987 (2008).
8. Strack, R.L. et al. Biochemistry 48, 8279–8281 (2009).
9. Shaner, N.C. et al. Nat. Biotechnol. 22, 1567–1572 (2004).
10. Kredel, S. et al. Chem. Biol. 15, 224–233 (2008).
KatushkaeqFP650eqFP670
mNeptuneNeptuneE2-Crimson
figure 3 | Whole-mouse infrared imaging with the Biospec system.
Infrared fluorescence of subcutaneous injections of equal amounts
of protein samples into a living mouse after normalization using the
extinction coefficient at the absorption maximum wavelength of each
protein. Excitation at 635 nm with photodiodes was used, and emission
was measured over the integrated 700–900 nm range. Scale bar, 1 cm.
© 2010 Nature America, Inc. All rights reserved.

Page 4

nature methods
doi:10.1038/nmeth.1501
online methods
Cloning and gene construction. Cloning and gene construction
were performed as described in reference 2. Briefly, purification
of PCR products and products of restriction digests was per-
formed by gel electrophoresis and extraction using the QIAquick
gel extraction kit (Qiagen). Plasmid DNA was purified using the
QIAprep Spin Miniprep kit (Qiagen). Site-directed mutagenesis
was performed by overlap-extension PCR11. Random mutagenesis
was performed using the Diversity PCR Random Mutagenesis kit
(Clontech). For bacterial expression, a fluorescent protein gene was
cloned into the pQE30 vector (Qiagen) using BamHI and HindIII
restriction sites. For expression in eukaryotic cells, a fluorescent
protein gene was cloned instead of TurboGFP into the pTurboGFP-
N vector (Evrogen) using AgeI and NotI restriction sites.
Characterization of fluorescent proteins in vitro. Fluorescent
proteins were characterized as described in reference 12. Briefly,
proteins were expressed in E. coli XL1 Blue strain (Invitrogen),
purified using Talon metal-affinity resin (Clontech) and desalted
using gel-filtration columns (Bio-Rad). Cary 100 UV/VIS
Spectrophotometer and Varian Cary Eclipse Fluorescence spec-
trophotometer were used to measure absorption and excitation-
emission spectra. All measurements were performed in 100 mM
NaCl with 20 mM Tris-HCl (pH 7.5). Molar extinction coefficients
were calculated based on the chromophore absorption in native
proteins and proteins alkali-denatured with an equal volume of
2 M NaOH, considering that in alkaline conditions DsRed-like
chromophore converts to the GFP-like one13 with an extinction
coefficient of 44,000 M−1 cm−1 at 452 nm. Absorbance spectra
were measured immediately after denaturation. Quantum yields
were determined by direct comparison with mCherry. To mea-
sure pH stability, buffers in the pH range from 3 to 10 were used.
An aliquot of purified protein was diluted in the corresponding
buffer solution. After 1 h incubation at room temperature (25 °C),
the fluorescence brightness was measured. In each sample, actual
final pH was measured using a microelectrode (Sartorius).
Photostability. Selected proteins underwent a photostability
comparison test in LSM 510 confocal scanning microscope
(Carl Zeiss) and epifluorescence Leica AFLX 6000 microscope.
Proteins with 6His tags were bound to Talon metal affinity resin
beads, placed on a glass slide and exposed to light. The following
parameters were used for bleaching by a 561 nm laser line in a
confocal microscope: 63× oil-immersion objective, 1.5× zoom,
4 s per image scan rate and maximal power. For bleaching by a
mercury arc lamp, a TexasRed filter set was used, which passes
the 546 nm peak of a mercury lamp. Bleaching times were cor-
rected on the molar extinction coefficients of the corresponding
fluorescent protein at 546 nm or 561 nm.
Animal imaging and data analysis. HEK 293T cells were grown in
DMEM (Invitrogen), 10% FBS, 1% glutamine and 0.1% penicillin-
streptomycin-gentamicin. Cells were transfected with 8.7 μg of
plasmid encoding various fluorescent proteins and 1 μg of plasmid
encoding firefly luciferase by calcium phosphate precipitation as
described previously14. Cells were used for experiments 2 d after
transfection. The plasmid for E2-Crimson was provided cour-
tesy of B. Glick (University of Chicago). All animal procedures
were approved by the University of Michigan Committee on Use
and Care of Animals. Mice were anesthetized with isoflurane and
injected with 2.5 × 106 cells intramuscularly into the right and left
gluteal musculature of each mouse. Cells were injected in 100 μl
of a 1:1 mixture of F12 medium and matrigel (BD Biosciences).
Imaging was performed approximately 1 h after injection.
Bioluminescence and fluorescence imaging were performed on
an IVIS Spectrum (Caliper) system. Fluorescence images were
acquired with excitation and emission filters indicated in Figure 2,
using 3-s acquisition, small binning and 12.5-cm field of view.
After fluorescence imaging, bioluminescence imaging of firefly
luciferase was performed as described previously15. Data for fluo-
rescence and bioluminescence were quantified as efficiency and
photons, respectively. To account for differences in transfection
efficiency and actual numbers of injected cells, pseudocolor dis-
plays for fluorescence images were normalized for photons for
firefly luciferase in each cell implant. Data were also graphed as
ratios of fluorescence efficiency to firefly luciferase photons and
presented as mean values ± s.e.m. (n = 4–6 samples per point).
11. Ho, S.N., Hunt, H.D., Horton, R.M., Pullen, J.K. & Pease, L.R. Gene 77,
51–59 (1989).
12. Shcherbo, D. et al. Biochem. J. 418, 567–574 (2009).
13. Gross, L.A., Baird, G.S., Hoffman, R.C., Baldridge, K.K. & Tsien, R.Y.
Proc. Natl. Acad. Sci. USA 97, 11990–11995 (2000).
14. Smith, M.C. et al. Cancer Res. 64, 8604–8612 (2004).
15. Luker, G.D. et al. J. Virol. 76, 12149–12161 (2002).
© 2010 Nature America, Inc. All rights reserved.