ARTICLE

Received 18 Oct 2016 | Accepted 14 Mar 2017 | Published 19 May 2017 DOI: 10.1038/ncomms15269 OPEN

A high quantum yield molecule-protein complex

fluorophore for near-infrared II imaging
Alexander L. Antaris1,*, Hao Chen2,3,*, Shuo Diao1, Zhuoran Ma1, Zhe Zhang3, Shoujun Zhu1, Joy Wang1,
Alexander X. Lozano1, Quli Fan3, Leila Chew1, Mark Zhu3, Kai Cheng3, Xuechuan Hong2, Hongjie Dai1
& Zhen Cheng3

Fluorescence imaging in the second near-infrared window (NIR-II) allows visualization of

deep anatomical features with an unprecedented degree of clarity. NIR-II fluorophores draw
from a broad spectrum of materials spanning semiconducting nanomaterials to organic
molecular dyes, yet unfortunately all water-soluble organic molecules with 41,000 nm
emission suffer from low quantum yields that have limited temporal resolution and
penetration depth. Here, we report tailoring the supramolecular assemblies of protein
complexes with a sulfonated NIR-II organic dye (CH-4T) to produce a brilliant 110-fold
increase in fluorescence, resulting in the highest quantum yield molecular fluorophore thus
far. The bright molecular complex allowed for the fastest video-rate imaging in the second
NIR window with B50-fold reduced exposure times at a fast 50 frames-per-second (FPS)
capable of resolving mouse cardiac cycles. In addition, we demonstrate that the NIR-II
molecular complexes are superior to clinically approved ICG for lymph node imaging deep
within the mouse body.

1 Department of Chemistry, Stanford University, Stanford, California 94305, USA. 2 State Key Laboratory of Virology, Key Laboratory of Combinatorial

Biosynthesis and Drug Discovery (Wuhan University), Ministry of Education, Wuhan University School of Pharmaceutical Sciences, Wuhan 430071, China.
3 Molecular Imaging Program at Stanford (MIPS), Bio-X Program, and Department of Radiology, Canary Center at Stanford for Cancer Early Detection,

Stanford University, Stanford, California 94305-5344, USA. * These authors contributed equally to this work. Correspondence and requests for materials
should be addressed to Z.C. (email: [email protected]) or to H.D. (email: [email protected]) or to X.H. (email: [email protected]).

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 1

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269

W
ithin the last 5 years, the development of new wavelength organic dyes similar in functionality and biocompat-
fluorophores emitting in the second near-infrared ibility to those in the visible and NIR-I should allow NIR-II
widow (NIR-II) at wavelengths ranging from 1,000 to imaging to transition rapidly into a clinical setting and the
1,700 nm has allowed visualization of deep anatomical features broader research community. Donor–acceptor–donor (D–A–D)
with an unprecedented degree of clarity1–11. Light scattering and dyes are a promising class of small-molecule fluorophores
attenuation as well as background autofluorescence all decrease for bioimaging whose emission extends past 1,000 nm11,17.
when imaging at progressively longer wavelengths, which have Traditionally used for semiconducting organic electronics such
enabled non-invasive through-skull imaging of brain vasculature, as OLEDs and dye sensitized solar cells, this highly tunable dye
detecting tumours at depths of B4 mm in the brain, resolving architecture is built around a strong electron acceptor such as
blood flow dynamics resulting from cardiac waveforms, and benzo[1,2-c:4,5-c0 ]bis([1,2,5]thiadiazole) (BBTD) to which strong
assessing blood flow anomalies in cardiovascular disease and electron donors are wired through p-spacers to lower the energy
traumatic brain injury1–3,10. However, the majority of NIR-II gap. Molecular engineering of the energy difference between the
fluorophores suffer from low quantum yields as the generation of highest occupied molecular orbital (HOMO) and lowest
long-wavelength photons require low bandgap materials in which unoccupied molecular orbital (LUMO) levels allows wavelength
non-radiative decay processes tend to dominate over radiative tuning of emission peaks spanning the NIR-II. However, the
photon emission12. In the visible or NIR-I (B750–1,000 nm) hydrophobic nature of these dyes requires additional
regions quantum yields of B80% and B10%, respectively, are functionalization to increase aqueous solubility. We recently
common yet those in the NIR-II are typically in the range reported the first organic dye for NIR-II imaging, a carboxylated
of B0.01–1.4% with the exception of lead sulfide quantum D–A–D dye termed CH1055, which required PEGylation if not
dots1,13–16. The dearth of high quantum yield NIR-II conjugated to a hydrophilic protein2. Although CH1055-PEG
fluorophores has hindered fully reaching the potential of deep- demonstrated excellent pharmacokinetics, a low quantum yield of
imaging penetration depths in this window and caused poor B0.3%, while typical of NIR-II fluorophores, leaves ample room
temporal resolution as long exposure times are needed to for further improvement.
compensate for low brightness. A clinically translatable high Herein, we demonstrate that changing the functional groups
quantum yield organic NIR-II fluorophore would open up many from carboxylic to sulfonic acid results in a completely water-
exciting possibilities for biomedical fluorescence imaging. soluble organic NIR-II dye (CH-4T) that readily forms supra-
While most current NIR-II fluorophores consist of inorganic molecular assemblies with plasma proteins to produce a brilliant
semiconducting nanomaterials, the development of long- increase in fluorescent brightness18. The fluorescence intensity of

a CH1055 CH-4T
b
O O
CH-PEG

HOOC COOH HO3S SO3H

N N
H H
N N

SO3H
N N H2N N N
S S S S
CH- 4T

N N HBTU, DIPEA, DMSO N N

N N
H H
N N DI PBS
HOOC COOH HO3S SO3H FBS
O O

c d ~50 nm e
50,000 1.0 CH-4T
45,000 CH-4T
Normalized fluorescence (a.u.)
Fluorescent intensity (a.u.)

CH-4T/PBS CH-PEG FBS

40,000 CH-4T/FBS 0.8 CH-4T in PBS
CH-PEG/PBS
35,000 CH-4T in FBS
CH-PEG/FBS
30,000 0.6
25,000
20,000 0.4
1.10–1.28 g cm–3

15,000
10,000 0.2
5,000
0 0.0
900 1,000 1,100 1,200 1,300 1,400 1,500 900 1,000 1,100 1,200 1,300 1,400 1,500
Wavelength (nm) Wavelength (nm)

Figure 1 | Synthesis and optical characterization of CH-4T. (a) Chemical structure of CH1055 and the one-step sulfonation to produce CH-4T. (b) NIR-II
fluorescent images (10 ms, 1100LP) of CH-PEG and CH-4T in deionized water, FBS and PBS at equivalent absorbance of OD 0.1 at 808 nm. Colourbar next
to b ranges from 1,000 to 15,000. (c) Fluorescent emission spectra of CH-PEG and CH-4T in FBS, PBS after excitation at 808 nm. (d) Normalized
fluorescent emission spectrum of CH-PEG in PBS and CH-4T in FBS, PBS demonstrating B50 nm hypsochromic shift. (e) NIR-II fluorescent image
(1000LP, 10 ms) of centrifuge tubes after sucrose DGU with free CH-4T dye on the left and CH-4T pre-mixed with FBS on the right. Dotted white line
indicates buoyant density region of unbound CH-4T as seen in free dye band in left tube. Bright regions in right centrifuge tube correspond to
CH-4T–protein complexes that are found at the higher buoyant densities typical of proteins (B1.10–1.28 g cm  3).

2 NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269 ARTICLE

CH-4T increases B50-fold in serum compared to phosphate- as well as ion pairing between the sulfonate groups
buffered solution (PBS) and revealed an impressively high and cationic amino acid residues such as histidine, lysine and
quantum yield of up to B5%. Further, simply heating the dye arginine19.
to 70 °C for 10 min in serum boosted its quantum yield up to To further verify the formation of CH-4T complexes with
11%, affording the highest reported quantum yield to date for a plasma proteins, the free dye and CH-4T pre-mixed with FBS
clinically suitable NIR-II fluorophore. Optimizing the brightness (CH-4T/FBS) were subjected to isopycnic density gradient
of CH-4T through protein complexation prior to injection ultracentrifugation (DGU) (Fig. 1e; see Methods for complete
allowed for the fastest video-rate imaging in the NIR-II with DGU details). A linearly changing sucrose density gradient
B1.5–2 ms exposure times and ultra-fast 50 frames-per-second (1.10–1.28 g cm  3) was established along the length of the
(FPS) dynamic imaging capable of resolving cardiac cycles with centrifuge tube that corresponds to the average buoyant density
unprecedented temporal resolution. In addition, we demonstrate range of proteins. During ultracentrifugation, macromolecules
that lymph nodes deep within the mouse (B5–8 mm) are clearly migrate to a position in the gradient where their buoyant density
imaged in the NIR-II using the CH-4T complex with imaging matches that of the surrounding gradient. Free CH-4T dyes
performance superior to indocyanine green (ICG). formed a distinct band near B1.10 g cm  3, while for CH-4T/FBS
a heterogeneous distribution of protein-bound fluorescence-
enhanced CH-4T was observed throughout the density gradient
Results with a bright region at the bottom of the tube (B1.20 g cm  3).
Synthesis and optical characterization. Synthesis of the sulfo- The absence of fluorescence in the buoyant density range
nated small-molecule organic dye (CH-4T, 1.4 kDa, Fig. 1a) was corresponding to free CH-4T in the centrifuge tube containing
achieved through amide bond formation between the carboxy- CH-4T/FBS indicates near-zero free dye and a strong protein
lated CH1055 and taurine by O-(benzotriazol-1-yl)-N,N,N0 ,N0 - binding affinity20.
tetramethyluronium hexafluorophosphate (HBTU), N,N-diiso-
propylethlamine (DIPEA) coupling in DMSO (see Supplementary
Methods for complete synthesis details). Taurine is a common Optimizing the brightness of CH-4T–protein complexes.
biomolecule found throughout the body containing both an To develop an ultra-bright NIR-II probe by exploiting CH-4T–
amine and sulfonate group, thus allowing facile modification of protein interactions, common serum proteins were added in
CH1055. After conjugation, CH-4T was purified with HPLC and excess to CH-4T to observe the increase in fluorescence
the eluted fractions analysed by matrix-assisted laser desorption enhancement under the NIR-II camera. Both human and bovine
ionizing time-of-flight mass spectrometry (MALDI-TOF-MS) serum albumin (HSA and BSA) demonstrated strong fluorescence
to collect CH1055 conjugated to four taurine moieties enhancement by 17-fold yet the intensity was B2  lower than
(Supplementary Fig. 1). An nuclear magnetic resonance (NMR) the dye in serum (Fig. 2a). Even weaker CH-4T fluorescence
spectrum further confirmed the synthesis and purity of CH-4T occurred when mixed with IgG antibodies. A fluorescence titra-
(Supplementary Fig. 2). An ultraviolet-visible absorbance spec- tion was performed by increasing the HSA concentration while
trum revealed a strong peak at B738 nm (Supplementary Fig. 3). keeping CH-4T at a constant 1 mM to determine the approximate
The relative fluorescent brightness of CH-4T was investigated by binding stoichiometry. As seen in Fig. 2b, the maximum fluor-
matching the absorbance at 808 nm (optical density (OD) 0.1) in escence intensity occurred at approximately a 2:1 HSA:CH-4T
deionized water (DI), PBS and fetal bovine serum (FBS) and molar ratio.
imaging the vials on an indium-gallium-arsenide (InGaAs) As the brightness of CH-4T seems highly dependent on
camera under 808 nm excitation. The fluorescent brightness of optimal binding between the dye and protein surfaces, heating of
CH-4T in DI water was marginally higher than in PBS, yet the complexes was attempted with the hopes of exposing CH-4T
a drastic 35-fold increase in fluorescence in the range of to typically inaccessible hydrophobic domains located in the
900–1,500 nm was observed after mixing free CH-4T dye with protein interior. After heating for 10 min, the brightness of the
FBS (10 ms, 1100LP; Figs 1b and 2e for higher image dynamic vials containing the CH-4T and FBS mix increased steadily with
range; see Methods for details on CH-4T spectroscopy). In temperature until B75–85 °C. However, heating to temperatures
contrast, CH1055-PEG mixed at equivalent concentrations in above B85 °C caused a precipitous drop in fluorescent intensity.
identical biological media produced a uniform NIR-II brightness Significant aggregation was observed for temperatures above
across all solutions (Fig. 1b). A fluorescent emission spectrum B90 °C, while a minimal amount of aggregation at lower
collected for CH-4T and CH-PEG in both PBS and FBS under an temperatures was removed through centrifugation post-heating
excitation of 808 nm (using a NIR-II spectrophotometer that (30 min, 15,000 r.p.m.). Heating virtually every dye–protein
compensated for the variability in the detector quantum efficiency complex caused a marked increase in fluorescent intensity with a
at different wavelengths (Fig. 1c)) demonstrated a higher fluor- similar B50 nm hypsochromic shift from B1,050 to B1,000 nm.
escence brightness of CH-4T in serum than CH-PEG by 16-fold, After returning to room temperature, the thermally stabilized
again suggesting the CH-4T dye became brilliantly bright in the dye–protein complexes maintained their boosted brightness when
presence of serum proteins. CH-4T mixed with FBS from compared to solely mixed solutions.
various sources all produced virtually the same level of fluores- The CH-4T–protein complexes represented the brightest
cence enhancement regardless of manufacturer (Supplementary fluorescent agent based on organic molecules with an emission
Fig. 4). peak in the B1,000–1,100 region of the NIR-II. Comparing the
To investigate CH-4T binding to biomacromolecules, the brightness of fluorophores in vials and cuvettes with the NIR-II
normalized fluorescence emission spectrum of CH-PEG in PBS camera may cause errors due to inter-filter effects, while the
was compared to CH-4T in both PBS and FBS. The fluorescence hypsochromic emission peak shift coupled with the nonlinear
emission peak of CH-PEG and CH-4T in PBS occurred at camera quantum efficiency as a function of wavelength
B1,055 nm with emission from 900 to B1,400 nm as expected necessitates a more rigorous measurement. The relative fluores-
from previous CH1055 spectroscopic measurements. However, cence brightness of CH-4T/FBS heated to 70 °C for 10 min,
CH-4T in serum yielded a hypsochromic shift of B50 nm termed CH-4T/FBS-HT, was 28-fold, 36-fold and 22-fold brighter
which is indicative of non-covalent binding between the dye than carbon nanotubes (QY ¼ 0.4%; IR-26 ¼ 0.5%), CH-PEG
and proteins through hydrophobic van der Waals’s interactions (QY ¼ 0.3%; IR-26 ¼ 0.5%) and IR-26 (QY ¼ 0.5%) in DCE,

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 3

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269

a b 18
CH-PEG PBS 16

Fluorescence enhancement
CH-PEG FBS 14

(vs CH-4T/PBS)
12
CH-4T PBS
10
CH-4T HSA 1 µM CH-4T
8
CH-4T BSA 6

CH-4T FBS 4
2
CH-4T HS
0
0 4.0×105 8.0×105 1.2×106 0 2 4 6 8 10 12 14 16
Integrated intensity [HSA] (µM)

c d
4.0
67,500,000 HiPCO SWCNTs
QY = ~11%

Integrated fluorescence (a.u.)

CH-4T/HSA
3.5 60,000,000 CH-4T/FBS
Fluorescence enhancement

CH-4T/HSA heated
52,500,000 CH-4T/FBS heated
3.0
(vs CH-4T/FBS)

45,000,000
2.5 37,500,000
30,000,000
2.0
22,500,000
1.5 15,000,000

1.0 7,500,000
0
20 30 40 50 60 70 80 90 100 0.00 0.02 0.04 0.06 0.08 0.10 0.12 0.14
Temperature (°C) Absorbance at 808 nm (OD)

e
CH-4T
CH-4T CH-4T CH-4T
SWCNTs CH-PEG FBS
PBS HSA FBS
heated

Figure 2 | Ultra-high quantum yield CH-4T–protein complexes. (a) Fluorescent brightness of CH-PEG, CH-4T in protein solutions at 5:1 protein-to-dye
molar ratio (b) Fluorescent titration for constant 1 mM of CH-4T in increasing HSA receptor concentration. (c) Plot demonstrating fluorescence brightening
through heating for 10 min of CH-4T pre-mixed with FBS and horn sonicated FBS. (d) Plot of the integrated fluorescence spectrum of CH-4T pre-mixed with
HSA, FBS before and after heating at 70 °C for 10 min at five different concentrations. Linear fits were used to calculate quantum yield by comparing the
slopes to reference HiPCO SWCNTs (QY ¼ 0.4%) (e) Photographs and corresponding NIR-II fluorescent images (10 ms, 1000LP) of carbon nanotubes,
CH-PEG and CH-4T in HSA, FBS and heated FBS. Photographs were taken at an absorbance value of OD 1 at 808 nm, while NIR-II fluorescent images were
at the lower concentration of OD 0.1. Colourbar next to e ranges from 4,000 to 40,000.

respectively, with a 110-fold brightness difference between Ultra-fast in vivo imaging with CH-4T. To investigate the
CH-4T/PBS and CH-4T/FBS-HT when measured on a wave- in vivo optical properties of free CH-4T dyes and the brightness
length-corrected NIR-II spectrophotometer (see Methods; optimized CH-4T–protein complexes, video-rate imaging
Supplementary Fig. 5; Supplementary Tables 1 and 2 for (1000LP, 35 ms exposure time) of mouse hindlimb vasculature
enhancement calculations). Due to recent discrepancies in the was performed after an intravenous injection of free CH-4T and
reported quantum yield of the IR-26 reference fluorophore CH-4T/FBS-HT at equivalent dosages (90 mg, 100 ml at OD 3.2 at
(QYIR26B0.05–0.5%; see Discussion), the absolute quantum 808 nm) under 808 nm excitation at a power density of
yields of CH-4T/PBS, CH-4T/FBS and CH-4T/FBS-HT are in 140 mW cm  2. After 10 min post-injection, a clear difference in
the range of 0.0098–0.098%, 0.48–4.8% and 1.8–10.8%, respec- the fluorescent intensity of the femoral vasculature was discerned
tively. The NIR-II fluorescent images (1000LP, 10 ms) in Fig. 2e as seen in Fig. 3a,b indicating that the higher quantum yield of
qualitatively shows the disparate brightness levels between the pre-made complexes of CH-4T/FBS-HT compared to CH-4T that
brilliantly fluorescent CH-4T–protein complexes and current complexed with serum proteins in vivo in circulating blood upon
NIR-II contrast agents such as single-walled carbon nanotubes injection of free CH-4T generated a brighter contrast agent. The
(SWCNTs) and CH-PEG, all with matching absorbance at B2  brighter femoral artery of CH-4T/FBS-HT injected mice
808 nm (OD 0.1). compared to those injected with free dye correlates with the

4 NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269 ARTICLE

a CH-4T b CH-4T/FBS-heated c CH-PEG

35 ms exposure 35 ms exposure 50 ms exposure
10 mins PI 10 mins PI 10 mins PI

d 32,000
e CH-4T f CH-4T/FBS-heated
CH-4T 2 ms exposure 2 ms exposure
Fluorescence intensity (a.u.)

28,000 CH-4T/FBS-heated 50 FPS 50 FPS

24,000 2 sec PI 2 sec PI
20,000
16,000
12,000
8,000 Femoral artery
4,000 ROI intensity
0
0

50

0
10

15

20

25

Time (s) 256

h CH-4T/FBS-heated
g

Pixels
1,600 Femoral artery
Fluorescence intensity (au)

1,400 Heart beat

1,200 at 50 FPS Time (ms)
0 9,400 1
1,000 256
i
800 CH-4T

Pixels
600
400
200 0 Time (ms) 9,400 1
0
0

0
50

00

50

00

50

00
1,

1,

2,

2,

3,

Time (ms)

Figure 3 | Free CH-4T versus CH-4T–protein complexes for NIR-II hindlimb vasculature imaging. NIR-II fluorescent images of mouse hindlimb
vasculature 10 min post-injection of (a) free CH-4T, (b) CH-4T heated at 70 °C for 10 min in FBS (CH-4T/FBS-HT) and (c) CH-PEG. All injection doses
were normalized to an absorbance of OD 3.2 at 808 nm with an exposure time of 35 ms, 50 ms for CH-4T, CH-PEG, respectively. Scale bar in a of 0.5 cm
and corresponds to a–c,e,f. Colourbar next to c ranges from 0.5 to 5  104 and corresponds to a–d. Fluorescent intensity of integrated ROI region on the
hindlimb femoral artery plotted for 250 s post-injection during video-rate imaging of free CH-4T and CH-4T/FBS-HT (35 ms exposure time). NIR-II
fluorescent images 2 s post-injection of e free CH-4T and f CH-4T/FBS-HT during ultra-fast 50 FPS imaging. Colourbar next to f ranges from 500 to –5,000
and corresponds to e,f,g. Fluorescent intensity of integrated ROI region on the femoral artery for 3 s post-injection demonstrating a clearly resolvable
cardiac cycle. The fluorescence intensity of a single vertical row of pixels during 50 FPS imaging in e,f was plotted against time for h CH-4T/FBS-HT and
i free CH-4T during 9.4 s post-injection. Pink bars in f,h mark the edges of the hindlimb. The femoral artery is clearly visible as the solid red streak in h only
after injection of pre-bound CH-4T in heated FBS. Colourbar next to i ranges from 0 to 2,000 and corresponds to h,i. All NIR-II images and fluorescent
intensity profiles for e–i were collected at a 2 ms exposure time. All NIR-II imaging was performed with a 1,000 nm long-pass emission filter.

B2–3  boosted solution brightness difference observed Fig. 3a,b). During the first 10 s post-injection, a strikingly stronger
between CH-4T/FBS-HT and the free dye mixed in serum. For rise in fluorescence was observed from the femoral artery of mice
comparison, much weaker fluorescence was observed at any time injected with the pre-brightened complex compared to free
from a mouse injected with an equivalent CH-PEG dosage (100 ml CH-4T. In contrast, the gradual increase in fluorescent intensity
at OD 3.2 at 808 nm; Fig. 3c), even at higher exposure times of observed from the femoral artery of the mouse injected with free
50 ms, since CH-PEG showed no complexing with serum proteins CH-4T occurred as the brightening kinetics of the free dye
to afford enhanced fluorescence in vivo. required longer periods of time to binds to serum proteins during
While the CH-4T/FBS-HT complex yielded a B2  increase circulation to reach its full fluorescent potential brightness.
in the steady-state vessel fluorescent intensity compared to free By dividing the fluorescent intensity of the femoral artery of the
dye, the benefits of pre-optimizing the brightness of the CH-4T– pre-bound versus free dye (Supplementary Fig. 6), earlier time
protein complexes are highly desired for imaging immediately points demonstrate a B33-fold higher brightness that equili-
post-injection. In Fig. 3d, the integrated intensity of a background brates to the steady-state brightness difference of B2-fold at
subtracted region of interest (ROI) region in the femoral artery B200 s. After this point, the vessel intensities of mice injected
was plotted for the first 250 s post-injection for CH-4T and CH- with free CH-4T have formed stable complexes and reached their
4T/FBS-HT (position of ROI region indicated by red arrows in steady-state maximum fluorescent brightness. The brightness

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 5

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269

difference between the femoral artery of CH-4T and the pre- To demonstrate the enhanced clarity of features deep inside the
brightened complex injected mice now matches the brightness body by imaging in the NIR-II, the lumbar lymph nodes were
difference seen at longer time points post-injection in Fig. 3a,b. In visualized after the belly was stretched to reduce the distance
strong contrast to CH-4T, CH-PEG at an equivalent dose between the lumbar lymph nodes and the surface of the mouse
produced a barely detectable increase in fluorescent signal in (Fig. 4d). The lumbar lymph nodes sit on either side of the spine,
the femoral artery after injection that remained at a constant low laterally of the descending aorta bifurcation when observing
fluorescent intensity (Supplementary Fig. 7). the mouse in a supine position and are connected to a lymphatic
The markedly higher brightness of pre-bound CH-4T circulat- channel responsible for draining the footpad and leg
ing in the vascular system during the first seconds post-injection (see Supplementary Fig. 10 for lymphatic system details). Without
can be exploited for high temporal resolution imaging of contortion, the distance between the lumbar lymph nodes and the
hemodynamic processes. A stark difference between injecting surface approaches B1–1.5 cm, the bulk of which is mostly
free CH-4T and CH-4T/FBS-HT (250 mg CH-4T/mouse) is comprised of the intestines as shown in the illustration in Fig. 4e.
clearly seen in the selected frames at equivalent time points Stretching the belly provides a fixed imaging depth of
during ultra-fast 50 FPS video imaging using exposure times of B0.5–0.8 cm and allows sufficient lymph node fluorescent signal
only 1.5–2 s (Fig. 3e,f) with an instrument overhead time of in both windows (see Methods for details).
B19 ms. At 2 s post-injection, no fluorescence is noted from the ICG (50 mM) was injected into the footpad and digits on
vasculature of mice injected with free CH-4T due to the slow both sides; afterwards the mouse was fixed physically and
brightening kinetics of the dye in vivo. In contrast, the femoral stretched on the imaging platform until the lumbar lymph
artery is clearly resolved during 50 FPS imaging when injecting nodes were visualized with a silicon detector in the NIR-I
CH-4T/FBS-HT (Fig. 3e). Plotting the integrated ROI intensity of (300 ms, 785–900 nm; for imaging set-up details see
the femoral artery for 3 s post-injection of CH-4T/FBS-HT allows Supplementary Figs 11 and 12). The subsequent injection of
for precise imaging of a cardiac cycle where intensity peaks CH-4T/HSA-HT (50 mM) on the same mouse in an identical
corresponding to ventricular ejections (Fig. 3g) are resolved with manner and imaging on the InGaAs detector (400 ms, 1,100 þ nm)
much higher resolution compared to previous attempts demonstrated a striking increase in the resolution of
with lower quantum yield semiconducting NIR-II polymers the lumbar lymph nodes as seen in Fig. 4f. Imaging with the
(see Supplementary Fig. 8 for comparison)1. The fluorescent CH-4T–protein contrast agent allowed the resolution of indivi-
intensity from a single row of pixels crossing the hindlimb was dual nodes with a signal-to-background (SBR) ratio of B13 as
selected from each frame for the first 9.4 s of the 50 FPS videos opposed to the diffuse conjoined features observed in the NIR-I
that generated the images seen in Fig. 3e,f and plotted as a with an SBR of B2.5. The enhanced clarity is reflected in the
function of time (Fig. 3h,i). The pink position markers denote the fluorescence cross-sectional intensity profile of the lumbar lymph
edges of the hindlimb in Fig. 3f,h. Plotting the fluorescent nodes in both windows in Fig. 4g as well as the B5  lower
intensity line profile of a single row of pixels over time clearly background autofluorescence levels in the NIR-II. Imaging in
shows that the blood flow front along with the femoral artery, both windows was performed with a 785 nm laser (16 mW cm  2)
visualized as the strong red streak in Fig. 3h, can easily be to provide a constant excitation power density for both dyes.
discerned with the CH-4T/FBS-HT contrast agent yet virtually no
fluorescence can be observed during the first B10 s when
injecting unbound CH-4T in Fig. 3i. Discussion
Imaging in the second near-infrared window allows optical
imaging to peer deeper into the body with more clarity than
Deep lymph node imaging with CH-4T complexes. To further any other wavelength regime22–27. However, the development
investigate the advantages of high-brightness contrast agents for of small-molecule organic dyes is necessary preceding the
biomedical fluorescence imaging, the imaging quality of lym- widespread use of this imaging technique. Inorganic
phatic vasculature and nodes were compared in both the first and nanomaterials are excellent for fluorescence-based preclinical
second NIR windows. The popliteal and sacral lymph nodes were animal imaging and while many current nanomaterial NIR-II
imaged in nude mice in a prone position with fluorophores probes show strong promise for clinical translation, organic
injected in both footpads as seen in Fig. 4a. ICG dispersed in DI NIR-II dyes similar in structure and pharmacokinetics to the
water at 50 mM was utilized as a NIR-I contrast agent and lymph FDA approved NIR-I organic fluorophores such as ICG and
node-to-background ratios of 8.6±2.9 were observed (Fig. 4b) methylene blue should expedite the transition of NIR-II imaging
which is consistent with reported values21. Although ICG can into the clinic28–30. Researchers without the specialized
form complexes when mixed with proteins such as HSA, the knowledge of nanomaterial-based fluorophores would benefit
benefits of the premixed ICG/HSA complex compared to free from organic NIR-II dyes that could be incorporated into
ICG in terms of lymph node detection efficiency is unclear21,22. bioimaging procedures currently relying on the plethora of
Lymph node fluorescence from ICG and ICG/HSA-HT likewise visible and NIR-I dyes. CH-4T’s complete aqueous solubility
produced nearly identical intensities as seen in Supplementary combined with its high serum quantum yield makes it an
Fig. 9. However, a very notable difference in brightness of attractive NIR-II clinical candidate capable of being utilized with
B8-fold is observed between free CH-4T injected on the left ease for biomedical vascular imaging procedures. CH-4T can
footpad and CH-4T/HSA-HT injected on the right side footpad dramatically reduce fluorescent dye doses by B25-fold for
(Fig. 4c) at equivalent 50 mM dosages immediately post-injection. biomedical imaging compared to previous, low quantum yield
The disparate fluorescent intensities of the popliteal and sacral contrast agents such as CH-PEG. CH-4T can also uniquely enable
lymph nodes for free CH-4T and CH-4T/HSA-HT demonstrated the fastest temporal resolution in the NIR-II spectral window
the need for premixing to form ultra-bright complexes for NIR-II The D–A–D NIR-II dye architecture shows strong potential for
lymphography. CH-4T/HSA-HT resulted in lymph node-to- developing an array of NIR-II fluorescent probes through the
background ratios of 29±7.6 with significantly sharper lymph molecular engineering of their constituent components. The
node features with a popliteal/sacral FWHM of 2.52±0.23 mm/ addition of the negatively charged sulfonate groups translated
1.61±0.11 mm compared to 4.86±0.17 mm/4.4±0.29 mm with into a dramatic boost in the in vivo quantum yield of CH1055
ICG in the NIR-I. that allowed for video-rate imaging at the fastest frame rate along

6 NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269 ARTICLE

a b c
ICG, NIR-I CH-4T, 1,100+ nm

SC
12
9 3 7 32 17
4 2

PO
HSA
PBS
heated

d e f g
Normal position
ICG, 785–900 nm 200 300
~1.5 cm

intensity on Si detector (a.u.)

ICG

Fluorescence intensity on
175

InGaAs detector (a.u.)

CH-4T/HSA-heated 250
150

Fluorescence
200
125
lumbar
Spine 100 150
lymph nodes
75
Lumbar LN CH-4T/HSA-heated, 1,100 nm 100
50
Stretched position 50
25
~0.5 cm
0 0
0 2 4 6 8 10 12 14 16
Position (mm)

Figure 4 | Lymph node imaging with ultra-bright CH-4T–protein complexes. (a) Photograph depicting nude mouse in prone position for imaging popliteal
and sacral lymph nodes in b,c. Fluorophore injection sites indicated by blue arrows next to footpad. (b) NIR-I fluorescent image B30 min post-injection of
50 mM ICG (50 ms exposure; colourbar ranges from 4,000 to 40,000). (c) NIR-II fluorescent image B30 min post-injection of CH-4T/PBS injected in left
foot and CH-4T/HSA-HT in right foot at equivalent dosages of 50 mM CH-4T (100 ms exposure; colourbar ranges from 4,000 to 55,000). 1 cm scale bar in
b corresponds to both b,c. (d) Photograph depicting nude mouse in a stretched position to narrow the distance between lumbar lymph nodes and mouse
surface. Stars indicate position of lumbar lymph nodes and white box corresponds to imaging field of view (e). Illustration of mouse anatomy surrounding
lumbar lymph nodes demonstrating a change in imaging depth from B1.5 to B0.5 cm. White triangle corresponds to the spine with red and blue circles
representing descending aorta and inferior vena cava, respectively. Brown lines represent intestines. (f) NIR-I/II fluorescent images of lumbar lymph nodes
after 50 mM of ICG and CH-4T/HSA-HT were injected sequentially into both footpads at an exposure time of 300 and 400 ms, respectively. ICG and CH-4T
imaging performed with a silicon and InGaAs camera, respectively, with identical 785 nm excitation laser conditions. Colourbar spans 0–1,500 for ICG and
0–8,000 for CH-4T/HSA-HT. Scale bars, 1 cm. (g). White arrows correspond to direction of the g fluorescent cross-sectional intensity profile of lumbar
lymph nodes on each detector.

with the lowest exposures time to date for imaging at NIR-II However, designing faster fluorescent NIR-II imaging systems
wavelengths1. Small changes in the molecular structure of through hardware and software modifications could theoretically
CH1055 yielded a dramatic increase in the optical performance yield frame rates surpassing 500 FPS given the 1.5–2 ms exposure
of the D–A–D dye, while further optimization of the electron time of the CH-4T complexes. Such a high-brightness dye
acceptor, p-spacers, electron donors and functional groups is additionally allows the detection of deeper features in the body as
likely to further increase the maximum attainable NIR-II evidenced by the clear resolution of the lumbar lymph nodes. As
quantum yield. the excitation power density drops off when penetrating deeper
The advent of a high quantum yield NIR-II fluorophore into the body, high quantum yield NIR-II fluorophores are
enables many exciting possibilities for NIR-II medical imaging in necessary to generate sufficient fluorescent signal. Although
terms of exposure time, imaging speed and penetration depth. previous materials such as carbon nanotubes or CH-PEG can
Carbon nanotubes, encapsulated organic dyes and Ag2S quantum clearly discern more shallow features with a high degree of clarity
dot contrast agents have required exposure times of 100 ms and due to minimal scattering, they have an insufficient quantum
may even exceed 200 ms for video-rate imaging2–5. The yield to detect deep features (Supplementary Fig. 13). The high
instrument overhead time for normal video-rate imaging is quantum yield NIR-II dye also lowers the requisite excitation
B80 ms and results in a frame rate of 5.3 FPS at a typical NIR-II power density as evidenced by the imaging of the lumbar lymph
exposure time of 100 ms. The donor–acceptor NIR-II polymer nodes with ICG and CH-4T/HSA-HT at an equivalent power
with a nominal 1.4% quantum yield allowed ultra-fast videos at density of 16 mW cm  2 and with similar exposure times.
B20 ms exposure times and frame rates of 25 FPS1. The CH-4T Previously, a 17.5-fold higher power density was required for
complex was able to easily capture ultra-fast videos under imaging femoral blood vasculature in the hindlimb with carbon
1.5–2 ms exposure times at 50 FPS, which is virtually at the nanotubes in the NIR-II (808 nm excitation, 140 mW cm  2)
maximum possible frame rate of our imaging system (52 FPS) compared to IR800 in the NIR-I (785 nm excitation,
given the instrument overhead time of 19 ms under ultra-fast 8 mW cm  2)4.
imaging conditions. At this juncture, we would like to highlight the variability in the
Even though CH-4T has a significant higher brightness than reported absolute quantum yield of IR-26, the reference
the NIR-II polymer, the achievable frame rate with CH-4T is only fluorophore that serves as the basis for all reported NIR-II
double that of the polymer as the instrument overhead time is quantum yields. The first reported quantum yield of IR-26 as
now the main factor limiting the time in-between frames1. 0.5% in 1,2-dichloroethane (DCE), the most widely applied value,

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 7

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269

has been called into question by recent reports placing the in normal tissue just days post-injection. Unlike current NIR-II
quantum at both 0.05 and 0.1% (refs 31–34). The reported nanomaterial fluorophores, CH-4T demonstrated complete
quantum yield for CH-4T and complexes is based on using excretion from the body, albeit at a slower rate than CH-PEG
HiPCO SWCNTs as a reference, whose quantum yield was (Supplementary Fig. 15). CH-4T elicited no observable cellular
similar to IR-26 (ref. 1). HiPCO SWCNTs were selected as a toxicity (Supplementary Figs 16 and 17), while CH-4T/FBS-HT
quantum yield reference as nanotubes are simple to disperse in and CH-4T/HSA-HT showed a marginal reduction in cell
water, demonstrate excellent stability unlike IR-26 that degrades viability at higher concentrations (Supplementary Figs 18 and 19).
in 1,2-dichloroethane over relatively short periods to time, and However, in a similar manner to the pioneering work by
are the most common NIR-II contrast agent5. The brightness of Frangioni and colleagues, tailoring the interaction strength
NIR-II dyes will be listed relative to other NIR-II fluorophores as between NIR-II fluorophores and proteins to maintain a high
well as in absolute quantum yield ranges henceforth until the quantum yield while employing strategies such as polyionic yet
discrepancies in the reported quantum yield of IR-26 are resolved. net-neutral probes may expedite excretion kinetics and reduce
As this work marks a transition in the reference absolute non-specific in vivo binding42,43. The same rational design
quantum yield, the absolute quantum yield of CH-4T is 1.1–11% principles developed for NIR-I fluorophores may be applied to
with respect to IR-26 ¼ 0.05–0.5%. Changing the absolute NIR-II organic dyes to develop the next generation of contrast
quantum yield of the reference only serves to scale the and molecular imaging agents.
brightness of the entire NIR-II field and does not affect the Hemodynamic imaging with a high frame rate in the NIR-II
relative brightness between NIR-II emitting materials. Regardless, poses several advantages over other vascular flow imaging
CH-4T is by far the brightest organic fluorophore of any NIR-I or modalities such as laser speckle flowgraphy and photoacoustic
NIR-II dye emitting past 1,000 nm (Supplementary Fig. 14) with a imaging (PI). Laser speckle flowgraphy computes blood
27-fold higher quantum yield than HiPCO SWCNTs, a classic velocity based on the measurement of laser speckle blurring
NIR-II imaging probe. (termed normalize blur (NB)) from coherent light scattering from
Tuning the interactions between organic NIR-II dyes consist- diffusing sources44,45. Modulation of the laser backscattered
ing of large, hydrophobic p-systems and plasma proteins is a interference pattern occurs as red blood cells travel through
route to brilliantly fluorescent dyes emitting past 1,000 nm. In a capillaries. Laser speckle flowgraphy has significant clinical
similar fashion to ICG’s B4  increase in brightness when potential as it requires no exogenous contrast agents yet this
exposed to serum proteins, the 110-fold fluorescence increase imaging modality currently suffers from a number limitations.
between CH-4T/FBS-HT and CH-4T in PBS most likely occurs Measurement of laser flux depends highly on tissue properties
through the combination of multiple processes35,36. First, the that can vary from person to person or in disparate disease states.
amphiphilic character of CH-4T may form aggregates in aqueous A correlation between the unit-less NB and capillary blood flow
solution through van der Waals forces that reduce brightness velocities occurs only under a narrow set of conditions44,45. This
through self-quenching which is known to occur in ICG29. disallows facile blood velocity comparisons among people or
The dispersion of aggregates by CH-4T individualization through even between an individual’s eyes during retinal vasculature
protein binding should contribute to the serum boost in quantum imaging. Photoacoustic imaging can additionally measure blood
yield. Second, CH-4T geometrical confinement by hydrophobic velocity without requiring exogenous contrast agents by
pockets should promote a rigid dye conformation that minimizes exploiting hemoglobin’s strong visible absorbance. However,
the torsional rotations that result in non-radiative decay12. Rigid with minimum resolvable features on the scale of B10–100 mm,
molecules have shorter fluorescent lifetimes and generally higher photoacoustic imaging cannot compete with fluorescence
quantum yields as they dissipate less energy though internal imaging in terms of spatial resolutionx46. CH-4T’s penetration
conversion compared to those with freely rotating parts37. depth (1–3 mm) combined with high temporal (50 FPS) and
Internal conversion is the most important competing non- spatial (B1 mm) resolution enables high-fidelity hemodynamic
emissive process for low energy-gap NIR-II dyes thus increasing imaging of vessel sizes ranging from microvasculature to major
rigidity through protein binding will favourably increase vessels. Application of high frame-rate NIR-II imaging with CH-
quantum yield. Heating magnifies these effects as CH-4T 4T for hemodynamic visualization within tumours, intravascular
presumably becoming entangled in a rigid conformation as tracking of circulating tumour cells or post-trauma cerebral
increased thermal energy allows access to typically inaccessible microvascular imaging may provide key insight into disease
hydrophobic pockets. At higher temperatures, CH-4T may pathophysiology. In addition to providing absolute blood
become completely encased in hydrophobic domains as proteins velocities, CH-4T’s high-brightness and long-wavelength
re-fold around the dye after heating to minimize the number of emission may improve other biological optical imaging
hydrophobic side-chains exposed to water. In addition, donor– modalities. For instance, one-photon confocal imaging with
acceptor dyes are subject to twisted intramolecular charge transfer CH-4T may push the penetration depth limit past that achievable
(TICT) in polar environments, especially if comprised of strong with the current palette of visible/NIR-I/NIR-II fluorophores.
electron donors and acceptors38–41. Molecular simulations predict Even with the superior NIR-I silicon CCD detector specifica-
the core of CH1055 can exist in non-planar conformations that tions (for example, pixel number, dark current, readout noise and
may contribute to a red-shifted TICT conformation that can so on), NIR-II imaging on the InGaAs detector with longer
dampen fluorescence12. The observed hypsochromic shift of wavelength probes improves imaging metrics. As no current
B50 nm can be explained through CH-4T adopting a planar silicon detector has suitable sensitivity past B1,000 nm, biome-
configuration and the alignment of p-orbitals when binding to dical fluorescent imaging in the NIR-II depended on the
proteins via hydrophobic or charge interactions in conjunction emergence of InGaAs focal plane array detectors. However,
with minimizing solvent relaxation. similarly to the benefits garnered by pushing probe emission
Developing a fundamental understanding of the manner in out of the visible (B350–600 nm) and into the NIR-I
which NIR-II organic dyes interact with biological tissue is of (B650–900 nm) on Si cameras, the development of bright,
critical importance for creating probes with high SBR ratios as long-wavelength NIR-II probes (41,000 nm) boosts imaging
well as high excretion levels. CH-4T demonstrated an impress- performance metrics. Internal comparisons within a broad
ively high quantum yield thorough protein binding and the dye wavelength range on an InGaAs detector have dramatically
was excreted through the biliary system with little residual dye left illustrated the imaging improvements by developing longer

8 NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269 ARTICLE

wavelength probes from 850 nm (ICG) to 1,100 nm (HiPCO 1  PBS were added through a syringe with a needled pressed against centrifuge
SWCNTs) to 1,500 nm (laser vaporization SWCNTs)14. ICG and tube wall to not cause layer mixing. After adding all density layers, the centrifuge
tube was placed at 10° from horizontal to allow for interlayer diffusion to smooth
laser vaporization SWCNTs both emit in a region with similarly out density steps. After 1 h, CH-4T samples were loaded on top of the gradient.
reduced InGaAs quantum efficiency of B30–40% compared to Any excess space in centrifuge tube was filled with 1  PBS until meniscus reached
B85% at B1,100 nm with HiPCO SWCNTs. This indicates the B3 mm from top of centrifuge tube. The gradient was run in an ultracentrifuge
marked differences in imaging quality at progressively longer (Beckman Coulter, Optima L-90K) using a Ti-55 rotor at 50,000 r.p.m. for 50 h.
wavelengths arise from reduced scattering/autofluorescence
rather than variable detector spectral sensitivities. Fundamental Preparation of CH-4T heated complexes. Prepared 784 mM CH-4T, 78.4 mM
advances in fluorophore chemistry have largely driven the NIR-II HSA/BSA in 1  PBS solution. Added 10 ml of 784 mM CH-4T into 200 ml of
78.4 mM HSA/BSA solution to keep the molar ratio of CH-4T:HSA at 1:2 for
imaging field, making InGaAs detectors more commonplace for optical characterization (1:1 ratio for in vivo HSA imaging). (HSA: Sigma Aldrich,
biomedical imaging. While newer generations of InGaAs Lot# SLBN5035V, SLBL6442V; BSA: Sigma Aldrich, Lot# SLBR6762V). Keep
detectors can uniformly increase sensitivity/SBR of all NIR-II overall protein concentration as low as possible to prevent possible gelling after
probes, relative imaging metrics within the NIR-II spectral range heating. For FBS, use maximum dye loading of 50 mg CH-4T/100 ml FBS
(see Supplementary Fig. 4 for manufacturer/batches). Vortex the solution to mix
will still be governed by wavelength dependent light-tissue evenly. Then add the dye–protein complex in a sealed eppendorf into different
interactions. Improving NIR-II fluorescent probes (that is, temperature water baths for 10 min (B70–75 °C for in vivo complexes). Discard if
increasing both quantum yield and NIR-II emission solution gels or denatures by turning white. Temperatures kept under 80 °C should
wavelength) should proceed in parallel with the ongoing not gel. If not used within a few hours, store the heated dye–protein solution at
4 °C for long-term storage. CH-PEG heated complexes were prepared by the same
innovations in improving InGaAs cameras.
procedure for comparison.
In summary, we have developed the first NIR-II contrast agent
based on a small-molecule organic dye with an ultra-high
Spectral characterization of CH-4T. Absorbance spectra of CH-4T and com-
quantum yield of 1.1–11% (IR-26 ¼ 0.05–0.5%). The negatively plexes were taken on a ultraviolet-visible-NIR Cary 6000i spectrometer that was
charged sulfonate groups enabled supramolecular binding to background corrected for each biological media such as water, FBS and HSA
serum proteins capable of a 110-fold fluorescence enhancement protein solutions. The NIR-II fluorescence emission spectrum was captured on a
compared to CH-4T in PBS. Heating the CH-4T–protein home-built spectroscopy set-up by exciting CH-4T and complexes with an 808 nm
laser diode with a power output of B160 mW. The excitation laser was filtered
complex caused a marked boost in fluorescence intensity with a combination of an 850 (Thorlabs)/1,000 (Thorlabs)/1,100 (Omega)/1,200
indicating that brightness increases are possible through optimiz- (Thorlabs)/1,300 (Omega) nm short-pass filters. Samples were added to either a
ing dye–protein interactions. In vivo vascular imaging was 1 mm or 1 cm path-length cuvette and the resulting emission filtered through a
performed at 50 FPS, the fastest frame rate to date in the 910 nm long-pass filter (Thorlabs) to reject the incident excitation laser light.
The emitted fluorescence was collected on a spectrometer (Acton SP2300i) coupled
NIR-II at an exposure time of 2 ms to unambiguously resolve to a linear liquid nitrogen cooled InGaAs detector array (Princeton Instruments,
vascular hemodynamics. In addition, imaging of deep lymph OMA-V). After collecting the raw acquisition data, a correction file was applied
nodes at a depth of B5–8 mm in both the first and second NIR to correct for the variable InGaAs quantum efficiency as a function of detection
windows clearly demonstrated the benefits of imaging deep wavelength as well as the variable 910 nm long-pass filter extinction features
anatomical features at longer wavelengths. CH-4T illustrates for  R 1;500 the NIR-IIspectral
across R 
region. Fluorescence enhancement defied as:
1;500
900 ICH4T=X dl = 900 ICH4T=PBS dl , where ICH  4T/X corresponds to the
the first time that high quantum yield dyes in the NIR-II are fluorescent emission spectrum of protein-bound CH-4T. All fluorescent
possible. Future work will strive to modify CH-4T to enable enhancement values were derived from measurements on the wavelength-corrected
covalent linkable NIR-II fluorophores whose brightness can be photospectrometer unless specifically stated otherwise.
utilized for ultra-high SBR in vivo molecular imaging.
Measuring NIR-II quantum yield. The fluorescence quantum yield of CH-4T and
complexes were measured in a similar manner as described in previous publica-
Methods tions1,2. Briefly, HiPCO SWCNTs were utilized as a reference fluorophore due to
Animal handling. All vertebrate animal experiments were performed under the
their high aqueous stability. The quantum yield of HiPCO SWCNTs has previously
approval of Stanford University’s Administrative Panel on Laboratory Animal
been determined as 0.4% based on an IR-26 quantum yield of 0.5% in DCE
Care. Eight-week-old female NU/NU mice were obtained from Charles River for all
(please see Discussion for further details on IR-26 quantum yield). A serial dilution
imaging studies and housed at the Research Animal Facility of Stanford under our
of five solutions of HiPCO SWCNTs as well as CH-4T and complexes with an
approved animal protocols. Before vessel or lymphatic imaging, all mice were
OD o0.1 at 808 nm was measured to confirm absorbance values at 808 nm and
anaesthetized in a rodent anaesthesia machine with 2 l min  1 O2 gas mixed with
the fluorescent emission spectrum was collected on a wavelength-corrected NIR-II
3% isoflurane. Tail vein injection of contrast agents were carried with a catheter
spectrometer in a 1 cm quartz cuvette (Starna) in the manner specified above. The
and synchronized with a camera that started continuous image acquisition
fluorescent emission spectrum was integrated and plotted against the OD value at
simultaneously. The injected dose was a 200–300 ml bolus in a 1  PBS solution at
808 nm and a linear fit was applied to verify the linearity between fluorescent
specified concentrations. During the time course of imaging the mouse was kept
brightness and concentration. For the brightest samples, inter-filter effects were
anaesthetized by a nose cone delivering 2 l min  1 O2 gas mixed with 3% isoflurane.
seen BOD 0.1 at 808 nm, thus lower concentration ranges were utilized. By
The sample sizes of mice were selected based on previously reported studies. Mice
comparing the slope of the linear fit between HiPCO SWCNTs and CH-4T and
were randomly selected from cages for all experiments. No blinding was per-
complexes, the quantum yield was determined based on the following supporting
formed. All groups within study contained n ¼ 3 mice.
equation (1):
 
Synthesis of CH-4T. Dissolve 1 mg CH1055 (1.03 mmol) in 100 ml of dry DMSO. slopesample nsample 2
QYsample ¼ QYref   ð1Þ
Then 5 mg taurine (40 mmol), 7 ml DIPEA (5.17 mg, 40 mmol) were added. Stirred sloperef nref
for 2 min and then add HBTU 5 mg (13.2 mmol). The reaction solution was stirred
overnight at room temperature under a nitrogen atmosphere. After the reaction where QYsample is the QY of CH-4T and complexes, QYref is the QY of HiPCO
finished, 100 ml of water was added and stirred for 1 hr to quench excess HBTU. SWCNTs in water (0.4% based on IR-26 ¼ 0.5% in DCE1), nsample and nref are the
Finally, Dionex Summit high-performance liquid chromatography (HPLC) system refractive indices of HiPCO SWCNTs, CH-4T and complex solutions which are
(Dionex Corporation, Sunnyvale, CA, USA), 340U four-channel ultraviolet-visible both water (1.33) in this case.
absorbance detector, Dionex C4, 9.4 mm  250 mm semi-preparative column,
gradient elution starting from 5% acetonitrile and ending up with 95% acetonitrile
(in water with 0.1% TFA) at 42 min, 3 ml min  1 flow rate, 254 nm and 650 nm Video-rate and ultra-fast NIR-II imaging. Video-rate and ultra-fast NIR-II
detection wavelength was used to purify the reaction. Overall, 1.3 mg CH1055- imaging was carried out in a similar manner as previous publication1,2. Briefly, all
4Taurine (yield 93%) was produced as a green solid. MALDI-TOF-MS was used to NIR-II images were collected on a 319  256 pixel two-dimensional InGaAs array
identify the product. MALDI-TOF-MS calculated for: [C62H64N10O16S6] (M.W.): (Princeton Instruments 2D OMA-V 1.7; pixel number: 319  256; readout noise:
1,396.3. Found: 1,396.2. See Supplementary Information for additional details. 50 electron r.m.s.; dark signal: 5,000 electrons per second per pixel) utilizing a
home-built imaging set-up. The excitation was provided by an 808 nm diode laser
(RMPPC Lasers) through an optical fibre and collimator at a power density of
DGU of CH-4T and CH-4T complexes. Sucrose density gradients were prepared 140 mW cm  2. The 808 nm laser was filtered with 850/1,000 nm short-pass filters
in 4 ml centrifuge tube. A total of 600 ml of 55, 50, 45, 40, 35, 30% w/w sucrose in (Thorlabs). Fluorescence emission was collected with a 910 nm long-pass filter

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 9

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269

(Thorlabs) combined with either a 1,000 nm long-pass filter (Thorlabs) or 1,100 nm 15. Chen, J. et al. Direct water-phase synthesis of lead sulfide quantum dots
long-pass filter (Thorlabs) as specified. A ‘  2.5’ magnification lens set (‘C’ coated, encapsulated by b-lactoglobulin for in vivo second near infrared window
200 mm and 75 mm, Thorlabs) was used for the mice hindlimb NIR-II imaging. imaging with reduced toxicity. Chem. Commun. 52, 4025–4028 (2016).
Changing the relative position of the two NIR achromats can modify the 16. Kong, Y. et al. Highly fluorescent ribonuclease-A-encapsulated lead sulfide
magnifications into ‘  1’ for whole body imaging. For ultra-fast imaging, the quantum dots for ultrasensitive fluorescence in vivo imaging in the second
camera is set to continuously expose using LabView software with a 19-ms near-infrared window. Chem. Mater. 28, 3041–3050 (2016).
overhead time caused by camera readout. Images and videos were processed with 17. Qian, G. et al. Band gap tunable, donor  acceptor  donor charge-transfer
MATLAB. Imaging the mouse in a supine position with the leg and chest region
heteroquinoid-based chromophores: near infrared photoluminescence and
well secured on the imaging platform with medical tape prevented any motion
electroluminescence. Chem. Mater. 20, 6208–6216 (2008).
artifacts resulting from breathing motion or heartbeat. Selecting a 3  3 pixel
18. Bhasikuttan, A. C., Mohanty, J., Nau, W. M. & Pal, H. Efficient fluorescence
region corresponding to the femoral artery in each video frame allowed for analysis
of the fluorescent intensity in hindlimb vasculature. The femoral vasculature enhancement and cooperative binding of an organic dye in a supra-biomolecular
mapped to the same pixel group throughout imaging, indicating that the leg is host-protein assembly. Angew. Chem. Int. Ed. 46, 4120–4122 (2007).
secure and that the periodic patterns do not arise from breathing induced artifacts. 19. Megyesi, M. & Biczok, L. Considerable fluorescence enhancement upon
Please see Supplementary Sections 12 and 13 for detailed schematic of optical path supramolecular complex formation between berberine and p-sulfonated
with component product information and catalogue numbers. calixarenes. Chem. Phys. Lett. 424, 71–76 (2006).
20. Ifft, J. B. Sedimentation equilibrium of proteins in density gradients. Biophys.
Chem. 5, 137–157 (1976).
Sacral/popliteal and lumbar lymph node NIR-I/II imaging. Sacral/popliteal and 21. Hutteman, M. et al. Optimization of near-infrared fluorescent sentinel lymph
lumbar lymph node NIR- II imaging was performed using the whole body set-up as node mapping for vulvar cancer. Am. J. Obstet. Gynecol. 206, 89.e1–5 (2012).
described in the above paragraph. A 1,344  1,024-pixel Hamamatsu silicon 22. Bardhan, N. M., Ghosh, D. & Belcher, A. M. Carbon nanotubes as in vivo
CCD detector (Hamamatsu C8484-03G02; pixel number: 1,344  1,024; readout bacterial probes. Nat. Commun. 5, 4918 (2014).
noise: B7 electron r.m.s.; dark current: 0.01 electrons per second per pixel) with a 23. Iverson, N. M. et al. In vivo biosensing via tissue-localizable near-infrared-
785 nm long-pass filter (Semrock) for NIR-I imaging was added in parallel to the fluorescent single-walled carbon nanotubes. Nat. Nanotechnol. 8, 873–880 (2013).
NIR-II set-up for NIR-I imaging comparisons as described in previous publica-
24. Yi, H. J. et al. M13 phage-functionalized single-walled carbon nanotubes as
tions4. The 785 nm laser (filtered with 780 nm band-pass filter) was used both for
nanoprobes for second near-infrared window fluorescence imaging of targeted
NIR-I and NIR-II imaging with a power density of 16 mW cm  2. All contrast
tumors. Nano Lett. 12, 1176–1183 (2012).
agents (ICG, CH-4T, CH-4T/HSA-HT) were injected into each digit and the
footpad under anaesthesia. Mice were then immediately placed on the imaging 25. Diao, S. et al. Chirality enriched (12,1) and (11,3) single-walled carbon
platform. A gentle massage was performed on the injection site for faster diffusion nanotubes for biological imaging. J. Am. Chem. Soc. 134, 16971–16974 (2012).
into the lymphatic system. For deep lumbar lymph node imaging, ICG was first 26. Ghosh, D. et al. Deep noninvasive imaging and surgical guidance of
injected as described above. The mice were taped securely in a position where the submillimeter tumors using targeted M13-stabilized single-walled carbon
lumbar lymph nodes were visualized on the silicon NIR-I detector. CH-4T nanotubes. Proc. Natl Acad. Sci. USA 111, 13948–13953 (2014).
complexes were subsequently injected B10 min later for imaging in the NIR-II. 27. Tao, Z. M. et al. Biological Imaging using nanoparticles of small organic
Fluorescent intensity cross-sectional profiles in the NIR-I versus NIR-II were molecules with fluorescence emission at wavelengths longer than 1000 nm.
analysed in Image J. Please see Supplementary Sections 12 and 13 for detailed Angew. Chem. Int. Ed. 52, 13002–13006 (2013).
schematic of optical path with component product information and catalogue 28. Hutteman, M. et al. Randomized, double-blind comparison of indocyanine
numbers. green with or without albumin premixing for near-infrared fluorescence
imaging of sentinel lymph nodes in breast cancer patients. Breast Cancer Res.
Treat. 127, 163–170 (2011).
Data availability. Data supporting the findings of this study are available within 29. Desmettre, T., Devoisselle, J. M. & Mordon, S. Fluorescence properties and
the article (and its Supplementary Information files) and from the corresponding
metabolic features of indocyanine green (ICG) as related to angiography. Surv.
author upon reasonable request.
Ophthalmol. 45, 15–27.
30. Kirchher, A.-M., Briel, A. & Mader, K. Stabilization of indocyanine green by
References encapsulation within micellar systems. Mol. Pharm. 6, 480–491 (2009).
1. Hong, G. et al. Ultrafast fluorescence imaging in vivo with conjugated polymer 31. Murphy, J. E. et al. PbTe colloidal nanocrystals: synthesis, characterization, and
fluorophores in the second near-infrared window. Nat. Commun. 5, 4206 multiple exciton generation. J. Am. Chem. Soc. 128, 3241–3247 (2006).
(2014). 32. Kopainsky, B., Qiu, P., Kaiser, W., Sens, B. & Drexhage, K. H. Lifetime,
2. Antaris, A. L. et al. A small-molecule dye for NIR-II imaging. Nat. Mater. 15, photostability, and chemical structure of IR heptamethine cyanine dyes
235–242 (2016). absorbing beyond 1 mm. Appl. Phys. B 29, 15–18 (1982).
3. Hong, G. et al. Through-skull fluorescence imaging of the brain in a new near- 33. Semonin, O. E. et al. Absolute photoluminescence quantum yields of IR-26 dye,
infrared window. Nat. Photonics 8, 723–730 (2014). PbS, and PbSe quantum dots. J. Phys. Chem. Lett. 1, 2445–2450 (2010).
4. Hong, G. et al. Multifunctional in vivo vascular imaging using near-infrared II 34. Hatami, S. et al. Absolute photoluminescence quantum yields of IR26 and IR-
fluorescence. Nat. Med. 18, 1841–1846 (2012). emissive Cd1  xHgxTe and PbS quantum dots -method- and material-inherent
5. Welsher, K. et al. A route to brightly fluorescent carbon nanotubes for challenges. Nanoscale 7, 133–143 (2015).
near-infrared imaging in mice. Nat. Nanotechnol. 4, 773–780 (2009). 35. Benson, R. C. & Kues, H. A. Fluorescence properties of indocyanine green as
6. Yi, H. et al. M13 phage-functionalized single-walled carbon nanotubes as related to angiography. Phys. Med. Biol. 23, 159–163 (1978).
nanoprobes for second near-infrared window fluorescence imaging of targeted 36. Mordon, S., Devoisselle, J. M., Soulie-Begu, S. & Desmettre, T. Indocyanine
tumors. Nano Lett. 12, 1176–1183 (2012). green: physicochemical factors affecting its fluorescence in vivo. Microvasc. Res.
7. Dang, X. et al. Layer-by-layer assembled fluorescent probes in the second near- 55, 146–152 (1998).
infrared window for systemic delivery and detection of ovarian cancer. 37. Siegel, J. A. Principles of fluorescence spectroscopy. Choice Curr. Rev. Acad. Lib.
Proc. Natl Acad. Sci. USA 113, 5179–5184 (2016). 44, 1196–1196 (2007).
8. Sordillo, L. A., Pu, Y., Pratavieira, S., Budansky, Y. & Alfanoa, R. R. Deep 38. Reja, S. I., Khan, I. A., Bhalla, V. & Kumar, M. A TICT based NIR-fluorescent
optical imaging of tissue using the second and third near-infrared spectral probe for human serum albumin: a pre-clinical diagnosis in blood serum.
windows. J. Biomed. Opt. 19, 056004 (2014). Chem. Commun. 52, 1182–1185 (2016).
9. Hong, G. et al. Near-Infrared II fluorescence for imaging hindlimb vessel 39. Sasaki, S., Drummen, G. P. C. & Konishi, G. Recent advances in twisted
regeneration with dynamic tissue perfusion measurement. Circ. Cardiovasc. intramolecular charge transfer (TICT) fluorescence and related phenomena in
Imaging 7, 517–525 (2014). materials chemistry. J. Mater. Chem. C 4, 2731–2743 (2016).
10. Zhang, X.-D. et al. Traumatic brain injury imaging in the second near-infrared 40. Rettig, W. Photophysical and photochemical switches based on twisted
window with a molecular fluorophore. Adv. Mater. 28, 6872–6879 (2016). intramolecular charge transfer (TICT) states. Appl. Phys. B 45, 145–149 (1988).
11. Sun, Y. et al. Novel benzo-bis(1,2,5-thiadiazole) fluorophores for in vivo NIR-II 41. Cao, C. et al. A twisted-intramolecular-charge-transfer (TICT) based
optical imaging. Chem. Sci. 7, 6203–6207 (2016). ratiometric fluorescent thermometer with a mega-Stokes shift and a positive
12. Hof, M., Hutterer, R. & Fidler, V. Fluorescence Spectroscopy in Biology: temperature coefficient. Chem. Commun. 50, 15811–15814 (2014).
advanced Methods and Their Applications to Membranes, Proteins, DNA, and 42. Choi, H. S. et al. Targeted zwitterionic near-infrared fluorophores for improved
Cells (Springer, 2005). optical imaging. Nat. Biotechnol. 31, 148–153 (2013).
13. Umezawa, K., Nakamura, Y., Makino, H., Citterio, D. & Suzuki, K. Bright, 43. Choi, H. S. et al. Synthesis and in vivo fate of zwitterionic near-infrared
color-tunable fluorescent dyes in the visible-near-infrared region. J. Am. Chem. fluorophores. Angew. Chem. Int. Ed. 50, 6258–6263 (2011).
Soc. 130, 1550–1551 (2008). 44. Senarathna, J., Rege, A., Li, N. & Thakor, N. V. Laser speckle contrast imaging:
14. Diao, S. et al. Fluorescence imaging in vivo at wavelengths beyond 1500 nm. theory, instrumentation and applications. IEEE Rev. Biomed. Eng. 6, 99–110
Angew. Chem. Int. Ed. 54, 14758–14762 (2015). (2013).

10 NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms15269 ARTICLE

45. Sugiyama, T., Araie, M., Riva, C. E., Schmetterer, L. & Orgul, S. Use of laser Additional information
speckle flowgraphy in ocular blood flow research. Acta Ophthalmol. 88, Supplementary Information accompanies this paper at http://www.nature.com/
723–729 (2010). naturecommunications
46. Hu, S. & Wang, L. V. Photoacoustic imaging and characterization of the
microvasculature. J. Biomed. Opt. 15, 011101 (2010). Competing interests: The authors declare no competing financial interests.

Reprints and permission information is available online at http://npg.nature.com/

Acknowledgements
reprintsandpermissions/
This work was partially supported by the Office of Science (BER), US Department of
Energy (DE-SC0008397) (Z.C.), NCI Cancer Center Nanotechnology Excellence Grant
How to cite this article: Antaris, A. L. et al. A high quantum yield molecule-protein
(CCNE-TR) U54 CA119367, the Calbrain Programme, a Neurotechnology Programme
complex fluorophore for near-infrared II imaging. Nat. Commun. 8, 15269 doi: 10.1038/
of California (H.D.), CA151459, the National Natural Science Foundation of China
ncomms15269 (2017).
81573383, 81373254 and 21390402, NSFHP (2014CFB704) (X.H.), International S&T
Cooperation Programme of China (2015DFA30440, 2014DFB30020) (X.H.), the Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in
National Science and Technology Major Project of the Ministry of Science and Tech- published maps and institutional affiliations.
nology of China No. 2012ZX10004801-003-011 (X.H.), Key Project of Chinese Ministry
of Education No. 313040 (X.H.), Academic Award for Excellent PhD Candidates funded
by Ministry of Education of China (No. 5052012306001), the Fundamental Research This work is licensed under a Creative Commons Attribution 4.0
Funds for the Central Universities (C.Q., H.C., X.H.) and Innovation Seed Fund of International License. The images or other third party material in this
Wuhan University School of Medicine (X.H.). article are included in the article’s Creative Commons license, unless indicated otherwise
in the credit line; if the material is not included under the Creative Commons license,
Author contributions users will need to obtain permission from the license holder to reproduce the material.
Z.C., H.D. and X.H. conceived and designed the study, supervised the project and wrote To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/
the manuscript. A.L.A. and H.C. designed the study, performed all the experiments and
wrote the manuscript. S.D., Z.M. helped with infrared. S.Z., L.C., J.W., A.X.L. helped
prepared dye–protein complexes as well as perform optical characterization. r The Author(s) 2017

NATURE COMMUNICATIONS | 8:15269 | DOI: 10.1038/ncomms15269 | www.nature.com/naturecommunications 11