In vivo nanoparticle-mediated radiopharmaceutical-excited fluorescence molecular imaging

Abstract

Cerenkov luminescence imaging utilizes visible photons emitted from radiopharmaceuticals to achieve in vivo optical molecular-derived signals. Since Cerenkov radiation is weak, non-optimum for tissue penetration and continuous regardless of biological interactions, it is challenging to detect this signal with a diagnostic dose. Therefore, it is challenging to achieve useful activated optical imaging for the acquisition of direct molecular information. Here we introduce a novel imaging strategy, which converts γ and Cerenkov radiation from radioisotopes into fluorescence through europium oxide nanoparticles. After a series of imaging studies, we demonstrate that this approach provides strong optical signals with high signal-to-background ratios, an ideal tissue penetration spectrum and activatable imaging ability. In comparison with present imaging techniques, it detects tumour lesions with low radioactive tracer uptake or small tumour lesions more effectively. We believe it will facilitate the development of nuclear and optical molecular imaging for new, highly sensitive imaging applications.

ARTICLE

Received 16 Sep 2014 | Accepted 19 May 2015 | Published 30 Jun 2015

Zhenhua Hu1,2,*, Yawei Qu3,*, Kun Wang1,2,*, Xiaojun Zhang4,*, Jiali Zha3, Tianming Song1, Chengpeng Bao1, Haixiao Liu1, Zhongliang Wang5, Jing Wang6, Zhongyu Liu7, Haifeng Liu3 & Jie Tian1,2,8

Cerenkov luminescence imaging utilizes visible photons emitted from radiopharmaceuticals to achieve in vivo optical molecular-derived signals. Since Cerenkov radiation is weak, non-optimum for tissue penetration and continuous regardless of biological interactions, it is challenging to detect this signal with a diagnostic dose. Therefore, it is challenging to achieve useful activated optical imaging for the acquisition of direct molecular information. Here we introduce a novel imaging strategy, which converts g and Cerenkov radiation from radioisotopes into uorescence through europium oxide nanoparticles. After a series of imaging studies, we demonstrate that this approach provides strong optical signals with high signal-to-background ratios, an ideal tissue penetration spectrum and activatable imaging ability. In comparison with present imaging techniques, it detects tumour lesions with low radioactive tracer uptake or small tumour lesions more effectively. We believe it will facilitate the development of nuclear and optical molecular imaging for new, highly sensitive imaging applications.

1 Key Laboratory of Molecular Imaging, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China. 2 Beijing Key Laboratory of Molecular Imaging, Beijing 100190, China. 3 Department of Gastroenterology, General Hospital of Chinese Peoples Armed Police Forces, Beijing, 100039, China.

4 Department of Nuclear Medicine, Chinese PLA General Hospital, Beijing, 100853 China. 5 School of Life Science and Technology, Xidian University, Xian 710071, China. 6 Department of Nuclear Medicine, Xijing Hospital, Fourth Military Medical University, Xian 710032, China. 7 Anal-colorectal Surgery Institute, No. 150 Central Hospital of PLA, Luoyang 471031, China. 8 The State Key Laboratory of Management and Control for Complex Systems, Chinese Academy of Sciences, Beijing 100190, China. * These authors contributed equally to this work. Correspondence and requests for materials should be addressed to H.L. (email: mailto:[email protected]

Web End [email protected] ) or J.T. (email: mailto:[email protected]

Web End [email protected] ).

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 1

& 2015 Macmillan Publishers Limited. All rights reserved.

DOI: 10.1038/ncomms8560 OPEN

In vivo nanoparticle-mediated radiopharmaceutical-excited uorescence molecular imaging

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

In vivo Cerenkov luminescence imaging (CLI) is an emerging technique that can be used to visualize the biological distribution of radiopharmaceuticals through Cerenkov

luminescence (CL)113. Similar to bioluminescence and uorescence molecular imaging (BLI and FMI), CLI offers high surface resolution and high throughput13,14. Unlike positron emission and single-photon emission computed tomography (PET and SPECT), CLI can image both b and b particle emitting isotopes11,15. Since many clinically approved nuclear tracers produce Cerenkov light, CLI shows great promise for clinical translation and has been recently applied for human thyroid and lymphatic node imaging16,17.

Despite these achievements, CLI faces many challenges. Cerenkov luminescence is weak with an intensity that is several orders of magnitude less than that of standard uorescence imaging17,18. This is because most of the radiated energy is in the form of g photons and/or other emitted particles generated during radioactive decay. The CL spectrum is weighted towards ultraviolet and blue15,19, which leads to high absorption in biological tissues and limits the imaging depth. Like g radiation in nuclear imaging, CL is not triggered light but is continuously generated during radioactive decay regardless of biological interactions. To achieve activatable imaging of nuclear agents, whose optical signal is activated on relevant biological interactions, the uorescent conversion of CL has been proposed1923. However, the overall optical signal must be killed for energy transformation, which limits imaging sensitivity.

To overcome these challenges, especially, to achieve activatable imaging and enhance the optical signal, we present a novel in vivo imaging strategy, radiopharmaceutical-excited uorescence imaging (REFI). It utilizes europium oxide (EO) nanoparticles to convert g-radiation (major) and CL (minor) into uorescence.

This is similar to a process called radioluminescence, which uses ionizing radiation (X- or g-radiation) to irradiate nanophosphors and generate light. Phosphorous materials have been used in radiation detectors for many years. A recent publication reported measurements of Eu(II) and Eu(III) in the new CsSrI3:Eu scintillator using X-ray absorption spectroscopy24. However, adopting this concept for bioimaging is quite recent

and underexplored2528. There is no comparison between this approach and other imaging techniques for highly sensitive in vivo tumour detection yet. Our technique combines the merits of CLI and FMI to boost the light intensity and signal-to-background ratio and achieve internal activatable imaging via the excitation from two different electromagnetic radiations. To investigate the excitation mechanisms, the signal enhancement, the emission spectrum, the tissue penetration ability and the difference in outcomes between REFI, CLI, FMI and PET, a series of in vitro phantom and in vivo xenograft studies are conducted. The benets of REFI for highly sensitive tumour imaging are demonstrated in two scenarios. We believe that by combining the strengths of CLI and FMI, signal-enhanced optical imaging through internal radionuclide excitation with dual electromagnetic radiations will benet preclinical research and clinical diagnosis in the future.

ResultsEO morphology and spectrophotometry. Scanning electron microscopy (SEM) showed EO nanoparticles with peanut-like morphology (Fig. 1a). The average diameter of EO nanoparticles was 8522 nm. The z potential was 28.6 mV. The excitation prole is shown in Fig. 1b. There were multiple characteristic absorption peaks at 301, 323, 365, 384, 396, 467 and 536 nm. The emitted uorescence spectra of EO excited by 400- and 535-nm lasers are shown in Fig. 1c,d. The peak emission was 620 nm in both cases. The absolute quantum yield of EO was 39%.

The mechanism of radiopharmaceutical excitation. The radio-pharmaceutical imaging tracer 18F-FDG (uorodeoxyglucose) emits g-radiation (511 keV), b particles and CL, whereas

99mTc-MDP (methylene diphosphonate) only emits g-radiation (140 keV). Therefore, we employed both radioactive tracers to reveal the mechanism of radioactive tracer-excited uorescence from EO.

Our previous study illustrated that an excitation range from ultraviolet to blue can excite EO to generate red emission (620 nm). This indicated that CL (350450 nm weighted) was also

a b

Emission intensity (AU)

2.5105

2.0105

1.5105

1.0105

5.5105

0.0

0.0

200 300 400 500

600

Excitation wavelenght (nm)

Emission wavelenght (nm)

c

d

Emission intensity (AU)

4105

4105

4105

4105

0

Emission intensity (AU)

2.5105

2.0105

1.5105

1.0105

5.5105

600 700 800 900 1,000 Emission wavelenght (nm)

500

600 700 800 900 1,000

Figure 1 | EO nanoparticle morphology and uorescence characterization. (a) SEM visualization of EO nanoparticles with mean size of 85 nm. Scale bar, 500 nm. (b) The optical excitation spectrum with a 620 nm lter displaying the characteristic absorption peaks of EO at 301, 323, 365, 384, 396, 467 and 536 nm. (c,d) The emission spectra of EO excited by 400- and 535-nm lasers displaying the peak emission with both excitation wavelengths at 620 nm.

2 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560 ARTICLE

a b

EO EO

2.5

1.5106 0.6106

0.5 0.2

Radiance(p s1 cm2 sr1)

Radiance(p s1 cm2 sr1)

18F-FDG

99mTc-MDP

I

II

III

IV

V

I

1.0

II

III

IV

V

c

d

6 p s1 )

3

2

1

6 p s1 )

Optical signal intensity

of Eu 2O 3(10

Optical signal intensity

of Eu 2O 3(10

8

0

18F-FDG

4 99mTc-MDP

I

II

III

IV

V I II III IV V

Figure 2 | The study of the radiopharmaceutical excitation mechanism. (a,b) EO is excited by 18F-FDG (a) and 99mTc-MDP (b) with different radiation blocking conditions. From row I to V: normal view, black box blocking CL, aluminium foil blocking b, lead torus blocking g, normal view. (c,d) The emission intensity of each condition is plotted for using 18F-FDG (c) and 99mTc-MDP (d), respectively. Both cases reveal obvious intensity decrease in Condition IV.

The slight intensity decrease from Condition I to V in (c) indicates the relatively shorter half life of 18F.

capable of exciting EO. However, it was not known whether it was the only radiation from radionuclides causing the excitation. Figure 2 demonstrates that neither blocking Cerenkov light (Row II) nor blocking b radiation (Row III) signicantly inuenced the induction of uorescence from EO. Conversely, the uorescent signal decreased dramatically when the g radiation from 18F-FDG or 99mTc-MDP was blocked (Row IV). Therefore, g radiation was the major cause of EO excitation. Both high (511 keV) and low energy (140 keV) g-rays were able to achieve EO excitation.

The uorescent emission of EO increased linearly with increasing activity of radioactive tracers (Fig. 3a,b). The excitation efciency of 99mTc-MDP was better than that of 18F-FDG. Fluorescent emission strongly correlated with tracer activity with R2 values of 0.97 (18F-FDG) and 0.99 (99mTc-MDP). For both

18F-FDG and 99mTc-MDP excitation, the emitted uorescent intensity of EO increased exponentially with an increase in its mass (Fig. 3c,d). Fluorescent emission strongly correlated with the amount of EO with an R2 value of 0.99. With increasing excitation distance, the EO emission intensity decreased exponentially (Fig. 3e,f). Fluorescent emission strongly and inversely correlated with distance with R2 values of 0.99 for

18F-FDG excitation and 0.97 for 99mTc-MDP excitation. The uorescent signal was almost undetectable when the excitation distance between 18F-FDG and EO was over 25 mm.

The fractions of g- and Cerenkov excitation in REFI are dependent on radioactivity, g- photon energy, EO mass, and excitation distance. For a single study (100 mCi 18F-FDG

10 mg EO with 10 mm distance), we employed a lead partition and two sets of mirrors (Fig. 3g,h) to obtain dual-radiation excitation and CL radiation excitation separately. The induced mean optical signal of EO was (100.55) 104 p s 1 and

(4.60.1) 104 p s 1, respectively (Fig. 3i). Therefore, the

fractions of g- and CL excitation were 95.4 and 4.6% in this case, which proved g-radiation was the major source of excitation.

Different excitation efciencies. Figure 3b indicates that different radioisotopes with the same activity caused different

excitation efciencies, and 99mTc-MDP with lower g photon energy (140 keV) was better than 18F-FDG (511 keV), when the radiotracer was separated with EO for excitation. However, when we co-mixed different radiotracers, 140 mCi 18F-FDG,

99mTc-MDP and 131I-NaI (131I-sodium iodide, major g energy 364 keV), with 10 mg EO, respectively, and investigate their emission spectra, we found the opposite (Fig. 4a). 18F-FDG EO

showed the highest signal, and 99mTc-MDP EO was the lowest.

It was a surprise that 18F and 99mTc showed different relative excitation efciency in conditions of separation and mixture. We hypothesized that the photoelectric interactions and Compton scattering of europium and g-photons were related to the distance between radionuclide and europium. When mixing the 18F with EO, both b particles and g-photons transferred energy to extranuclear electrons of EO and resulted in visible photons.

However, 99mTc only emits g photons, thus its excitation efciency was lower than 18F in case of mixing with EO. When radiotracer was separated from EO for certain distance, the probability of b particles reaching EO was almost zero. It was g photons that dominated the excitation for both isotopes. However, the probability of Compton scattering for 511 keV g photon might be higher than that for 140 keV g photon, which did not induce optical signal. The probability of photoelectric effect for 140 keV g-photon might be higher than that for 511 keV g-photon, which induced optical signal. That was the reason why separating and mixing radiotracers with EO showed different excitation efciency.

To further explore the relationship between emission intensity and excitation photon energy below 140 keV, we built a hybrid X-ray and optical system (Supplementary Fig. 1a). X-ray with the energy range of 40 to 130 keV was used to excite the EO (Supplementary Fig. 1b), and the emission intensity increased linearly with an R2 value of 0.95 (Supplementary Fig. 1c).

Optical signal enhancement and spectral red shift. With the presence of 10 mg EO, the optical signal enhancement showed a marked effect (Fig. 4a). The emission intensity was two orders of

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 3

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

a

Radiotracer EO

e

3.0

2.01051.0

1 mCi18F-FDG

520 Ci18F-FDG

1 mCi99mTc-MDP 520 Ci 99mTc-MDP

1.6

1.2105

18F-FDG

99mTc-MDP

Radiance(p s1 cm2 sr1)

0.8

f

Radiance(p s1 cm2 sr1)

5 p s1 )

Optical signal intensity

of Eu 2O 3(10

3

18F-FDG

5 p s1 )

R2=0.99

R2=0.97

b

Optical signal intensity

of Eu 2O 3(10

20

99mTc-MDP

6 p s1 )

Optical signal intensity

of Eu 2O 3(10

10

18F-FDG

2

1

2 99mTc-MDP

0

R2=0.99

5 10

15 20

25

0

20 40 60

R2=0.97

Excitation distance (mm) Excitation distance (mm)

0

1,000 2,000

Activity (Ci)

0

g

h

c

1.0

Radiance

(10 6p s 1cm 2sr 1)

6.0

4.0104

2.0

Radiance(p s1 cm2 sr1)

1.2

0.8106

0.4

Radiance(p s1 cm2 sr1)

5.0

3.0105

1.0

Radiance(p s1 cm2 sr1)

18F-FDG 18F-FDG

0.1

EO

Mirrors

Lead partition

18F-FDG

99mTcMDP

d

i

Optical signal intensity

of Eu 2O 3(10

Optical signal intensity

of Eu 2O 3(10

5 p s1 )

18F-FDG

5 p s1 )

8

4

0

Intensity (104 p s1 )

120 100

80 60 40 20

0

Dual radiation excitation

12

99mTc-MDP

R2=0.99

8

4

0

R2=0.99

0.0 0.1 0.2 0.0 0.1 0.2 Mass of Eu2O3 (g) Mass of Eu2O3 (g)

CL excitation

EO

Figure 3 | The investigation of different factors inuencing the excitation. (a) The emission intensity of EO depends on the radioactivity and the radiotracers. (b) The linear relationship between emission intensity and radioactivity. 18F-FDG and 99mTc-MDP shows different excitation efciencies. (c) The emission intensity of EO depends on its mass. (d) The exponential relationship between emission intensity and EO mass. (e) The emission intensity of EO depends on the excitation distance. (f) The inverse relationship between emission intensity and excitation distance. (g,h) Using mirrors and a lead partition to achieve g and CL dual excitation (g) and pure CL excitation (h), respectively, to measure the fractions of g- and Cerenkov excitation.

(i) The quantication reveals that CL contributed a small portion of the dual-radiation excitation.

magnitude more than the normal Cerenkov luminescence, even with the only 2-min exposure (Note that it took 5-min exposure time for the rest of our studies of REFI and CLI.). This together with our previous mechanism studies (Fig. 2) proved that EO was utilized to achieve uorescent conversion of g-radiation emitted from radiopharmaceuticals to boost the overall optical signal. Besides that, all three combinations (18F-FDG, 99mTc-MDP or

131I-NaI EO) showed similar spectra between 570 and 740 nm.

The peak of the emission prole was at 620 nm followed by a smaller peak at 700 nm. It is notable that for wavelength o570 nm, the overall optical signal was weaker than Cerenkov signal emitted from 18F or 131I along. This indicated that the blue Cerenkov light also contributed the excitation of EO, therefore part of its energy was shifted from blue to red. The radio-pharmaceutical induced uorescence showed greater signal intensity and shifted the optical spectrum from blue towards red. Both improvements suggested better ability to penetrate tissues. Since the emission proles were similar for all three radiotracers, and the most clinically used 18F-FDG demonstrated the best excitation efciency in the case of co-mixing EO, we focused on the 18F-FDG with EO for the rest of our studies.

Figure 4b,c shows the difference in tissue penetration ability between CLI and REFI. The Cerenkov luminescent signal of the control well (50 mCi 18F-FDG) was nearly ablated (from 6.4 to0.4 105 p s 1) after covering with a piece of porcine gastric

mucosa tissue (1-mm thick). However, the REFI signal was greater in Wells 1 to 4, which contained a range of amounts of EO(0.150.60 mg) mixed with 18F-FDG (50 mCi each). The signal intensity of Well 4 (50 mCi 18F-FDG 0.6 mg EO) was

approximately fourfold greater (1.8 105 p s 1) than that of

the control (Fig. 4d).

Combining the merits of CLI and FMI. CLI does not need external excitation as FMI does, which excludes background reection and autouorescence. Thus, an improved signal-to-background ratio is one of the strengths of CLI. REFI relies on radiation from a radiopharmaceutical to excite EO internally. Combining REFI with CLI confers the absence of external excitation and gives REFI the merit of a high signal-to-background. In both in vitro and in vivo phantom studies, the signal-to-background ratio of REFI was signicantly larger than that of

4 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560 ARTICLE

a

b

1,400 1,200 1,000

800 600 400 200

10

8 6 4 2 0

18F-FDG

131I-Nal 131I-Nal+EO

Intensity (104 p s1 cm2 sr1)

18F-FDG+EO 99mTc+EO

PBS

Control0 mg EO +18F-FDG

Well 20.30 mg EO +18F-FDG

Well 40.60 mg EO +18F-FDG

Well 30.45 mg EO +18F-FDG

0.2 0.4 0.6 0.8 (106 p s1 cm2 sr1)

450 500 550 600 650 700 750 800 850

Wavelength (nm)

Well 10.15 mg EO +18F-FDG

c d

200,000

100,000

Optical signal intensity (p s1)

Control0 mg EO +18F-FDG

150,000

Well 10.15 mg EO +18F-FDG

Well 30.45 mg EO +18F-FDG

Well 20.30 mg EO +18F-FDG

Well 40.60 mg EO +18F-FDG

Radiance

50,000

Control

Well 1

Well 2

0

Well 3

Well 4

Radiance

1.0 2.0 3.0 4.0

(106 p s1 cm2 sr1)

Figure 4 | The emission proles and the tissue penetration assessment. (a) The emission spectra of 18F-FDG, 18F-FDG EO, 131I-NaI, 131I-NaI EO,

99mTc-MDP EO and PBS (each radiotracer, 140 mCi; EO, 10 mg). The spectra of 18F-FDG and 131I-NaI indicate the Cerenkov light is weighted towards blue

(o 500 nm), and the spectra of 18F-FDG EO, 131I-NaI EO and 99mTc-MDP EO show a similar prole with peak emission at 620 nm. The EO

absorbance of part of the Cerenkov light is indicated by the difference between the emission spectra of radiotracer with and without EO from the wavelength of 500570 nm. (b,c) The comparison between Cerenkov luminescence (Control, 50 mCi 18F-FDG) and 18F-FDG radiation-excited uorescence (well 14) before (b) and after (c) covering with a piece of 1-mm thick porcine gastric mucosa tissue. (d) The quantication of the optical intensity of each well after covering the tissue.

FMI (in vitro Po0.001, in vivo Po0.01, Fig. 5). Regions of interest are shown with black circles, and quantications are listed in Fig. 5 and Supplementary Table 1.

The activatable imaging strategy of external excitation in FMI is advantageous in terms of molecular specicity. Conversely, radioactive probes are constantly emitting CL during decay regardless of their interactions with biological molecules. This excludes CLI from the benets of biologically specic, activatable imaging. However, REFI incorporates g-radiation and CL to excite EO internally for uorescent emission. This strategy conferred the merit of FMI for activatable imaging to REFI. For in vitro phantom imaging (Fig. 5ah), the activated REFI signal was distinguished from conventional CLI by either optical signal enhancement (Fig. 5c,d, Po0.001) or 620-nm ltering (Fig. 5e,f, Po0.001). For in vivo phantom imaging (Fig. 5in), PET showed no signicant differences (P40.05) between the two implanted capillary tubes (Fig. 5i), which indicated that the two tubes were lled with the same dose of

18F-FDG (50 mCi). However, the signal intensity was signicantly different between CLI and REFI with no ltering (Fig. 4j, Po0.05) or ltering (Fig. 4k, Po0.001).

In vivo validation of the superiorities of REFI. EO nanoparticles were directly injected into the tumours of Bcap-37 (human breast cancer cell) xenograft mice to simulate tumour-targeted nanoparticle delivery. Ten hours later, 18F-FDG was tail-vein injected. Figure 6a shows a comparison of REFI, CLI, and FMI in vivo. All three imaging techniques successfully detected optical

signals from tumours. However, REFI exhibited both high optical intensity (REFI versus CLI: 8.431.35 versus 3.690.84, unit: 105 p s 1, Po0.01) and high signal-to-background ratio (REFI versus FMI: 1.740.17 versus 0.940.14, Po0.01). Therefore, REFI offered the best tumour to normal tissue contrast in all cases. The superiorities of REFI were consistent with the results of the previous phantom studies. We also performed a comparison between REFI and CLI, in which EO was tumour injected after the tail-vein injection of 18F-FDG. The signal enhancement effect was signicant too (Po0.01,

Supplementary Fig. 2).

REFI and CLI were compared using U87MG (human gliomablastoma cell) and Bcap-37 xenograft mice models to verify the passive accumulation of EO nanoparticles in tumours via enhanced permeability and retention (EPR) effect after intravenous (i.v.) injection.

For in vivo imaging of U87MG xenografts, REFI with no ltering or 620-nm ltering showed signicantly stronger optical signal (both cases Po0.001) compared with that of CLI (Fig. 6b).

The mean intensity of REFI (no ltering: 135.712.1 105 p s 1,

620 nm ltering: 115.311.6 104 p s 1) was about 27 times and

15 times greater than that of CLI (no ltering: 5.11.6

105 p s 1, 620-nm ltering: 7.92.1 104 p s 1) with no ltering

and 620 nm ltering, respectively.

For in vivo longitudinal imaging of Bcap-37 xenografts, the consistent phenomenon was observed. For both tumour and bladder, the optical signal of REFI was signicantly stronger than that of CLI at all time points (Fig. 6c). This indicated that EO particles may partially excrete through kidney and urinary

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 5

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

a

b

i

j

Depth 2 mm

4.0

2.0

Radiance 10 8

(p s 1cm 2sr 1)

BQML 1e+07

8e+06

4e+06

6.0

4.0

2.0

2.5

1.5

0.5

Radiance 10 5

(p s 1cm 2sr 1)

REFI

CLI

schematic view

FMI

c

d

0

2.0

1.0

Radiance 10 6

(p s 1cm 2sr 1)

PET

Open

k l

1.2

0.8

0.4

Radiance 10 4

(p s 1cm 2sr 1)

Radiance10 8

(p s 1cm 2sr 1)

CLI

REFI

Open

620 nm filtering

REFI

CLI

e

f

2.0

1.0

Radiance 10 6

(p s 1cm 2sr 1)

620 nm filtering

FMI

CLI

Open

REFI

620 nm filtering

m

*

***

g

4

3

2

1

Optical signal intensity

(107 p s1 )

1.5

1.0

0.5

0.0

Optical signal intensity

(106 p s1 )

3

2

1

***

***

Optical signal intensity

(107 p s1 )

Optical

signal intensity

(106 p s1 )

0.2

0.1

0.0

CLI REFI

0 CLI REFI CLI REFI

0 CLI REFI

Open 620 nm filtering

Open 620 nm filtering

n

* *

h

***

(Signal intensity

-background)

/background

80

40

(Signal intensity

-background)

/background

80

40

0 FMI

REFI

0 FMI

REFI

Figure 5 | The in vitro and in vivo phantom comparison between different imaging techniques. (a) A schematic illustration of the in vitro tissue-mimicking phantom. (b) FMI of the 100 mCi 18F-FDG 1 mg EO mixture. (c,d) With no ltering, CLI using 18F-FDG only (c) and REFI using the mixture

of EO and 18F-FDG (d) show obvious different signal intensity. (e,f) With 620-nm ltering, CLI (e) and REFI (f) also demonstrate intensity differences. (g) The optical intensity comparison between CLI and REFI with no ltering and 620-nm ltering. (h) The signal-to-background ratio comparison between FMI and REFI. (i) The PET image of the in vivo phantom shows no signicant difference between the two implanted glass tubes (left tube: 50 mCi of

18F-FDG 0.15 mg of EO mixture, right tube: 18F-FDG, 50 mCi). (j,k) REFI and CLI show signicant differences with no ltering (j) or 620-nm ltering

(k). (l) FMI of the in vivo phantom. (m) The quantitative comparison of CLI and REFI in optical intensity with no ltering and 620 nm ltering. (n) The comparison of FMI and REFI in signal-to-background ratio.

system. However, the intensity differences were not as great as in the U87MG xenograft experiment. This may have been because the injection of EO was performed 24 h earlier than the injection of 18F-FDG in the U87MG group, whereas the EO and 18F-FDG were injected together in the Bcap-37 group. This may have resulted in further passive accumulation of EO inside U87MG tumour tissues.

We also performed the comparison between FMI using ICG (indocyanine green) and REFI using 11C-CHO (Choline) with EO through HepG2 (human hepatic cancer cell) orthotopic liver tumour mice models. The results further proved that REFI offered much better tumour to normal tissue contrast than FMI did (P o 0.01, Supplementary Fig. 3).

Multimodality comparison for tumour detection. To investigate the advantages of REFI for sensitive imaging of tumour lesions, 4T1-luc2 (luciferase-expressing mouse adenocarcinoma cell) xenografts were used for multimodality imaging comparison. PET, CLI, FMI and REFI were applied for two scenarios, and REFI offered the best imaging performance in both cases. The same dose of 18F-FDG was i.v. injected into the xenografts for PET, CLI and REFI. All quantications of optical intensity and signal-to-background ratio are listed in Figs 7 and 8, and Supplementary Table 2.

Six days after subcutaneous injection of 4T1-luc2 tumour cells, the tumour lesions were clearly visible on the upper and lower back of each mouse (Fig. 7a) with average diameter of

6 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560 ARTICLE

a

b

2.5

1.5

0.5

CLI REFI

CLI REFI

Time after the injection (min)

***

***

Radiance 10 5

(p s 1cm 2sr 1)

1.5

1.0

0.5

Radiance 10 8

(p s 1cm 2sr 1)

6.0

4.0

2.0

Radiance 10 5

(p s 1cm 2sr 1)

Optical signal intensity

(106 p s1 )

15

10

5

Open

0 CLI

CLI

REFI

REFI

CLI

REFI

FMI

Open

620 nm filtering

***

**

Optical signal intensity

(105 p s1)

8,000

7,000

6,000

5,000 15 10

5 0

6.0

4.0

2.0

Radiance 10 5

(p s 1cm 2sr 1)

***

***

(Signal intensity-

background) /background

2.5

2.0

1.5

1.0

0.5

0.0

1.5

1.0

0.5

0.0

620 nm filtering

FMI REFI REFI

CLI CLI

FMI

REFI of tumor REFI of bladder

Optical signal intensity

(106 p s1 )

CLI of tumor CLI of bladder

c

1.2

0.8

0.4

CLI

REFI

Radiance 10 5

(p s 1cm 2sr 1)

Optical signal intensity

(106 p/s)

2.0

1.6

1.2

0.8

40 40

70 100 130 min

55 60 80 80

110 120

Figure 6 | The comparison between different optical techniques for in vivo imaging xenografts. (a) After direct intratumoural injection with EO(0.05 mg) and tail-vein injection of 18F-FDG (800 mCi) into the Bcap-37 xenografts, REFI shows the best tumour to normal tissue contrast among all three imaging modalities. (b) After tail-vein injection with EO (0.1 mg, 24 h prior) and 18F-FDG (500 mCi) in the U87MG-xenografted mice, REFI shows signicantly greater signal than CLI did with both no ltering and 620-nm ltering. This demonstrates the passive accumulation of the EO nanoparticle in the tumour tissue. (c) The longitudinal observation comparing CLI and REFI in Bcap-37 xenografts. EO (0.1 mg) and 480 mCi of 18F-FDG are mixed and injected via the tail-vein. Both tumour and bladder show greater optical signal compared with that in the control mice at all time points.

**

a b

c

White-light

Intensity (104 p/s)

Intensity (104 p/s)

PET Sagittal

Upper-axial

Lower-axial

Max.

Min.

REFI CLI

Open ** **

10

12

6

Radiance

(10 4p s 1cm 2sr 1)

8

8

6 6

4 4 2 2 0 0

10

Intensity

5

4

REFI

CLI

Upper tumor (open)

REFI CLI

Lower tumor (open)

d

e

REFI CLI

4.0

620 nm

***

FMI QD620

*

*

11

10

9

8

Radiance

(10 3p s 1cm 2sr 1)

Intensity (104 p s1 )

1.8 1.81.5 1.51.2 1.20.9 0.90.6 0.60.3 0.30.0 0.0

Intensity (104 p s1 )

6

4

2

Radiance

(10 9p s 1cm 2sr 1)

(Signal-background)

/background

3.5 3.5

3.0 3.0

2.5 2.5

2.0 2.0

1.5 1.5

0.5 0.50.0 0.0 REFI REFI

FMI FMI

(QD620) Upper tumor

(QD620) Lower tumor

(Signal-background)

/background

1.0 1.0

REFI REFI

CLI CLI

Upper tumor (620 nm)

Lower tumor (620 nm)

Figure 7 | The multimodality comparison for in vivo imaging the dual-tumour xenografts. (a) Six days after the subcutaneous injection of 4T1-luc2 tumour cells, two tumour lesions are visible on the back of a mouse model (red arrows). (b) Three slices of PET (280 mCi 18F-FDG) in sagittal and axial directions show clear 18F-FDG uptake in the lower tumour (red arrows) but no signicant uptake in the upper one (white arrows). The position of the three

PET slices are indicated in (a) with black dotted lines. (c) Without ltering, REFI (280 mCi 18F-FDG with 0.1 ml, 1 mg ml 1 EO) shows optical signal of both tumours and brown adipose tissue (left mouse), but CLI does not visualize the upper back tumour (right mouse). (d) With 620 nm ltering, two tumours are visualized in REFI (left mouse), but Cerenkov signal are nearly vanished. (e) FMI of QD620 (0.1 ml, 10 mg ml 1) does not show great tumour to normal tissue contrast due to the non-specicity of the uorescent probe. The black circle indicates the regions of interest (ROI) of the background for calculating signal-to-background ratio.

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 7

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

a

b c d

White-light

BLI

Open

**

1.4

1.2

0.8

REFI CLI 2.0

1.5

0.5

Radiance

(10 5p s 1cm 2sr 1)

Intensity (104 p s1 )

4.54.03.53.02.52.01.51.00.50.0

PET axial

Radiance

(10 7p s 1cm 2sr 1)

1.0

1.0

Min.

Max.

Intensity

REFI

CLI

Tumor (open)

e

f

FMI QD620

REFI CLI

620 nm

**

** *

Intensity (103 p s1 )

Radiance

(10 4p s 1cm 2sr 1)

3

2

1

2.0

1.5

0.5

7 6 5 4 3 2 1 0

Radiance

(10 9p s 1cm 2sr 1)

FMIRJ2-DG750 6

4

2

Radiance

(10 9p s 1cm 2sr 1)

(Signal-background)

/background

5

4

3

2

1

0

1.0

REFI

CLI

FMI (RJ2-DG750)

Figure 8 | The multimodality comparison for in vivo early detection of small tumour lesions. (a) Sixty-ve hours after tumour cell injection, a white light image shows a small tumour lesion (red arrow). (b) BLI conrms the location of the small tumour lesion. (c) Axial PET shows false-negative scan. (d,e) With no ltering and 620 nm ltering, CLI (right mouse) shows false-negative detection (black arrows), but REFI (left mouse) shows true-positive detection (red arrows). (f) FMI of untargetted QD620 shows multiple suspected lesions, and targeted RJ2-DG750 shows overestimation of the tumour region. The signal-to-background ratios of both uorescent probes are signicant lower than that of REFI.

Tumor (620nm)

REFI

(620 nm)

FMI (QD 620)

5.60.8 mm. PET can detect the lower back tumour (Fig. 7b red arrows), yet cannot detect the upper back one (Fig. 7b white arrows). This was because the brown adipose tissue (high

18F-FDG uptake) close to the upper tumour inuenced its uptake of the radiotracer. CLI conrmed this phenomenon (Fig. 7c, right mouse). Optical signal were detected from brown adipose tissue and lower tumour lesion, but upper tumour was still not visualized. However, with the presence of EO and 18F-FDG in the same tumour lesion, the internal radiopharmaceutical excitation was activated. Signicant optical signal enhancement (REFI versus CLI: upper tumour: Po0.01, lower tumour:

Po0.01) was detected in both tumours for REFI (Fig. 7c, left mouse). After 620-nm ltering, most of the Cerenkov luminescence was blocked (Fig. 7d, right mouse), including the optical signal from the brown adipose tissue (Fig. 7d, left and right mice). However, REFI offered signicant optical signal for upper (REFI versus CLI: Po0.001) and lower (REFI versus CLI: Po0.05)

tumour lesions (Fig. 7d, left mouse).

In previous phantom imaging and in vivo imaging studies, the FMI showed relatively worse performance partially due to the low quantum yield of EO. To make the comparison fair and thorough, the quantum dot 620 (QD620) was used in this study. It is untargeted, same as EO, but with absolute quantum yield of more than 80% (more than twice of EO). Besides that the dose of QD620 (0.1 ml, 10 mg ml 1) was 10 times higher than the dose of EO (0.1 ml, 1 mg ml 1) for in vivo imaging. However, REFI still demonstrated signicant better signal-to-background ratio with the same excitation and emission setting up (REFI versus FMI: upper tumour: 2.740.59 versus 0.550.02, Po0.05, lower tumour: 3.070.63 versus 0.430.05, Po0.05, Fig. 7e).

Sixty-ve hours after subcutaneous injection of 4T1-luc2 tumour cells, the tumour size was only 2.10.3 mm (Fig. 8a, red arrow). BLI visualized the tumour lesion (Fig. 8b), but PET offered a negative scan (Fig. 8c). This insufcient sensitivity of PET is probably due to the partial volume effects that caused signicant underestimation of radiotracer concentration in small lesions2931. CLI conrmed this again (Fig. 8d,e, black arrows). It

was really difcult to detect the week Cerenkov signal from such a small lesion. However, REFI successfully achieved a positive imaging in both no ltering and 620-nm ltering modes (Fig. 8d,e red arrows, REFI versus CLI: both Po0.01).

To make a comprehensive comparison between FMI and REFI, we employed two uorescent probes this time, the untargeted QD620 and targeted RediJect 2-DeoxyGlucosone 750 (RJ2-DG750). RJ2-DG750 is a probe for targeting of tumours that exhibit elevated glucose uptake rate in comparison with surrounding tissues. The FMI of QD620 showed multiple suspected lesions (Fig. 8f, white arrows), whereas the FMI of RJ2-DG750 showed an overestimated tumour lesion area (Fig. 8f, right mouse). Nevertheless, the signal-to-background ratios of FMI were still signicantly smaller than that of REFI (REFI versus FMI: 4.420.18 versus 0.640.14 (QD620), Po0.05, 4.420.18 versus 1.480.57 (RJ2-DG750), Po0.01).

To further validate above ndings that REFI was more sensitive than PET for small tumour detection, another in vitro phantom and in vivo breast cancer mice model studies were conducted (Supplementary Figs 4 and 5). The results also proved that REFI was indeed more sensitive in small tumour detection.

Biodistribution of EO and toxicity evaluation. Cytotoxicity assay of EO was performed. Human umbilical cord mesenchymal stem cell morphology after 24 h of incubation with 50 and 400 mg ml 1 EO was captured using uorescence microscopy (Fig. 9a). Neither signicant change of cell morphology nor cell aggregation was observed in the experimental sample compared with that of the control sample (Fig. 9a) without nanoparticles.

The biodistribution of EO was measured 40 min after the tail-vein injection of EO (0.1 ml, 1 mg ml 1). Vital organs, such as heart, kidney, liver, lung and spleen, were ground using tissue-grinding pestles. Each sample (0.1 ml), including the sample of blood and urine, was mixed with 18F-FDG (0.1 ml, 100 mCi) respectively, so that the REFI signal was induced. By comparing the optical intensity with the control samples obtained

8 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560 ARTICLE

a

Control 50 g ml1 400 g ml1

b Blood Spleen Lung Liver Kidney Heart Urine

Radiance

(10 5p s 1cm 2sr 1)

EO concentration:

Intensity (EO injected - control) (103 p s1 )

45

40

35

30

25

20

15

10

5

3.5

EO injected group

Control group

3.0

2.5

2.0

0 Blood Spleen Lung Liver Kidney Heart Urine

c

EO injected group

Control group

Kidneys Liver Lungs Spleen Tumor

Heart

Figure 9 | Biodistrubition of EO and toxicity evaluation using MTTassays and H&E microscopy. (a) In vitro cytotoxicity assay indicated no signicant cell morphologic changes and no obvious cell aggregation. Scale bar, 200 mm. (b) The optical signal difference of the EO-injected group and control group (left image, upper row and bottom row) indicated the different concentrations of EO in tissues, blood and urine samples. The quantitative measurements showed that liver, kidneys and urine had a similar EO concentration, which proved the reason of signal enhancement in the bladder during in vivo REFI imaging. The high EO concentration in the spleen was probably because of phagocytosis by macrophages. (c) Compared with the control tissues, the kidneys, liver, lungs, spleen, heart and tumour of the EO tail-vein-injected group (Bcap-37 xenografts) did not show obvious structural changes.

from mice without the injection of EO (pure CL signal), the signal difference reected the biodistribution of EO per 0.1 ml (EO concentration). The results showed the spleen and heart had the highest EO concentration. Blood, liver, kidney and urine were very similar, and lung had the lowest EO concentration (Fig. 9b). This proved that EO was partially accumulated in liver, and partially excreted in urine through kidney.

Haematoxylin and eosin (H&E) microscopy (Fig. 9c) showed no obvious structural change in kidneys, lungs, spleen, liver, heart or tumour of the experiment group with EO injection. Compared with QD620, the EO nanoparticles were much less toxic, as all mice injected with QD620 died within the next 4 h, but all mice injected with EO lived more than 4 weeks except killed mice.

DiscussionThis study establishes REFI as a novel imaging strategy. Using EO nanoparticles and clinical radiopharmaceuticals, the g- (major)

and Cerenkov (minor) radiation can be converted to uorescence to achieve internally excited optical imaging. Our series of in vitro studies identied the excitation source and illustrated the relationship between the emission intensity and various parameters, such as radioactivity, EO mass and excitation distance. The excitation efciency of different radiotracers and X-ray photon energies was investigated. The optical signal

enhancement, spectra red shift and signal tissue penetration were also demonstrated.

REFI utilized both g- and Cerenkov radiation from radioactive tracers to achieve internal activatable imaging. This unique feature eliminates the adverse effects of autouorescence and reection from external excitation as occurs in conventional uorescence molecular imaging and enhances emission signals through uorescent conversion of g-radiation to overcome the challenge of detecting a weak Cerenkov signal from radioisotopes. Comparison of REFI, CLI and FMI through phantom in vitro and Bcap-37, U87MG and 4T1-luc2 xenograft in vivo studies clearly demonstrated that REFI combined the merits of CLI and FMI with superb signal-to-background ratio and internal activatable imaging ability. Especially in the 4T1-luc2 xenograft study of small tumour detection, REFI showed signicant better signal-to-background ratio than FMI did, no matter it was QD620 (higher quantum yield) or RJ2-DG750 (better targeting specicity) that was applied in FMI. Furthermore, in comparison of CLI, the optical signal of REFI was boosted remarkably and was spectrally shifted towards the deeper tissue-penetrating red range. All these features beneted the optical imaging performance in our in vitro and in vivo studies.

REFI employed both radiotracer and EO nanoparticles, and the internal signal activation highly relies on the distance of them. Therefore, if tumour lesion shows similar uptake of one tracer but

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 9

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

higher uptake of the other, in comparison with surrounding normal tissues, the tumour to normal tissue contrast will stand out. Our dual tumour 4T1-luc2 xenograft study demonstrated this phenomenon through multimodality comparison of REFI, PET and CLI. The uptake of 18F-FDG in upper back tumour was inuenced by brown adipose tissue nearby, but REFI was able to visualize the tumour unaffectedly, whereas PET and CLI failed.

The sensitivity of PET is superior to many other imaging modalities, but it is still limited by the low resolution partially due to the partial volume effects2931. For small tumour lesion with the size smaller or close to its spatial resolution, the underestimation of radiotracer uptake becomes signicant. The high supercial resolution of optical imaging, the signal enhancement effect and the better signal-to-background ratio empowered REFI to detect small subcutaneous tumour lesions more effectively. Our 4T1-luc2 xenograft study indicated that REFI was able to detect tumour lesions with the size o2 mm, which were o3 days after tumour transplanting in nude mice. This suggested a great potential of applying REFI for highly sensitive early tumour detection and tumour metastasis imaging, even with passive delivery of the nanoparticle. Different from bioluminescence imaging, REFI does not require incorporation of immunogenic proteins32,33, such as luciferase and GFP (green uorescent protein). It thus can be applied to a wider range of animal tumour models and holds better clinical translation potential.

Other studies of Cerenkov-induced uorescence imaging (SCIFI) have already demonstrated the mechanism of utilizing Cerenkov light solely for excitation and applications of SCIFI in tumour marker detection using targetted uorescent probes19,22. The fundamental difference between REFI and SCIFI (or other similar approaches) is the utilization of g-radiation for optical excitation and the resulted signal enhancement, as the overall optical signal of SCIFI was even weaker than Cerenkov light because of the inevitable energy loss during excitation. There were also studies of using radioluminescent nanoparticles and radiotracers, such as 18F-FDG, for biomedical imaging applications23,2527, but these works remained in phantom studies and did not reveal the dual-radiation excited imaging mechanism. To the best of our knowledge, this is the rst in vivo small animal tumour model study using REFI or other similar approaches.

The EO nanoparticle is not the only choice for applying the strategy of REFI. The lanthanide-doped nanoparticle is likely to provide similar uorescence conversion of both g- and Cerenkov radiation from radioisotopes26,27. These can be excited by collimated X-rays (approximately the same energy magnitude as g-radiation from 99mTc-MDP) and have already been applied in

X-ray/optical imaging modalities, such as X-ray luminescence computed tomography27,34,35. Therefore, our imaging technique converts external high-energy (X-ray) and low-energy (optical light) electromagnetic excitation to internal g- and CL excitation via radiopharmaceuticals for in vivo optical molecular imaging.

We believe that REFI and its principle hold great promise for in vivo activatable imaging that can detect molecular targets or events and provide quantitative information of pathological processes on molecular levels with proper modication of the EO or other lanthanide-doped nanoparticle to overcome the issue of target specicity. REFI, with its intrinsic superiorities, will benet the imaging of tumour-to-tumour molecular heterogeneity, which can further facilitate the development of precision medicine and personalized patient care. Therefore, this imaging technique will expand the applicability of activatable nuclear and optical molecular imaging.

In conclusion, through utilizing EO nanoparticles mediators, we have achieved internal conversion of continuous radio-pharmaceutical radiation to activatable uorescence for molecular

imaging. REFI can merge the advantages of nuclear and optical molecular imaging techniques in vivo. This creates a strong motivation for further modication of the EO nanoparticle to obtain biomarker specicity and application of this highly sensitive imaging technique to early and small tumour detection in the near future.

Methods

Nanoparticles and radionuclide tracers. The EO nanoparticle (Eu2O3, 99.9% metal basis, molecular weight 351.91) was purchased from the Aladdin Chem

istry Co. Ltd. 18F-FDG, 99mTc-MDP and 131I-NaI was provided by the department of nuclear medicine, Chinese PLA General Hospital, Beijing, China. The QD620 was purchased from China Beijing Beida Jubang Science & Technology Co. Ltd, and the RJ2-DG750 probe was purchased from PerkinElmer. The absolute quantum yield of EO and QD620 was measured using FLS980 uorescence spectrometer (Edinburgh Instruments). Both kinds of nanoparticle were dissolved in PBS for measurements.

SEM and spectrophotometry. The size and morphology of EO nanoparticles were determined by SEM (Hitachi S-4700). As-prepared EO powder samples were dispersed and dropped on a copper grid for scanning. The particle size was quantied using Image J. The EO nanoparticles were characterized for uorescent properties using an EnSpire Multimode Plate Readers (PerkinElmer). The samples were read using a transparent 96-well plate. The excitation prole was obtained using a 620-nm emission ltre. The uorescence proles were obtained with excitation at 400 and 535 nm.

Mechanistic study of radiopharmaceutical excitation. The imaging system used for these studies was the IVIS Spectrum system (Caliper Life Sciences). The imaging parameters were binning: 4, exposure: 5 min and aperture: f1, unless otherwise indicated.

In the identication of the excitation source experiment, two 1.5-ml tubes lled with 0.2 g EO and 600 mCi of 18F-FDG were placed 15 mm apart inside the imaging chamber and a sequence of images were acquired. The rst image was taken in the normal view. The second was taken with a black cardboard box covering the

18F-FDG tube. The third and fourth images were taken with an aluminium foil and a lead torus placed between the two tubes, respectively. Finally, a normal view of the two tubes was acquired again. Images of tubes with 990 mCi of 99mTc-MDP and0.20 g of EO were acquired following the same protocol.

In the emission intensity versus radioactivity experiment, EO (10 mg) was

excited using 0.1 ml of 18F-FDG and 0.1 ml of 99mTc-MDP with 11 different activities (2, 5, 10, 19, 30, 63, 140, 240, 521, 1,010 and 1,990 mCi). The images were analysed quantitatively to obtain the relationship between the emission intensity and radioactivity. The excitation distance was 10 mm.

In the emission intensity versus EO mass experiment, 18F-FDG and 99mTc-MDP of 100 mCi of were used to excite EO in eight different amounts (0.0012, 0.0052,0.0099, 0.0511, 0.1082, 0.1501, 0.2042 and 0.2505 g). The excitation distance was

10 mm.

In the emission intensity versus excitation distance experiment, the excitation distance was set to 1, 3, 5, 7, 10, 15, 20, 25, 43 and 60 mm, respectively. The radioactivity of 18F-FDG and 99mTc-MDP was 250 mCi in each acquisition, and 10 mg of EO was used.

For the assessment of g and Cerenkov excitation fraction, a pair of mirrors was set next to the two EP tubes containing 18F-FDG (100 mCi) and EO (10 mg), respectively. The tips of two EP tubes were 12 mm apart. After taking optical images, a lead partition (12-mm thick) was put in between the EP tubes to block g-radiations from 18F. However, the CL was partially reected to the EO because of the two mirrors reection. Then optical images were taken again for pure CL excitation.

Emission spectra assessment. 18F-FDG, 99mTc-MDP and 131I-NaI (each140 mCi) were mixed with EO nanoparticles (10 mg), respectively. Their emission spectra together with the Cerenkov luminescent spectra of 18F and 131I (both 140 mCi) were measured using the IVIS system (exposure: 2 min). PBS was used as the Control.

X-ray excitation of EO nanoparticles. During the X-ray excitation process, the voltage of the X-ray tube (Hamamatsu L9181-02 Microfocus X-ray Source, Japan) was increased gradually, while the tube current was kept constant. Therefore, X-ray photons with different emission energies (40, 60, 80, 100, 110, 120 and 130 keV) were obtained to excite EO nanoparticles (2 mg) inside an EP tube (Supplementary Fig. 1b), but the total photon number of the X-ray beam was kept constant for each excitation. To protect the EMCCD (DU888 , Andor, UK), the optical imaging

system was perpendicular to the X-ray tube (Supplementary Fig. 1a). The white light image was acquired with a 0.1-s exposure inside room light, and the X-ray-excited uorescent image was acquired with 5-s exposure inside a light sealed environment. The PBS was used as control.

10 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.

NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560 ARTICLE

Excitation efciency of different radiopharmaceuticals. 18F-FDG, 99mTc-MDP and 131I-NaI (each 100 mCi) were used to excite 10 mg EO, respectively (mixing excitation). The optical images were acquired using the IVIS Spectrum system (Caliper Life Sciences) with binning: 4, exposure: 2 min and aperture: f1. The excitation efciencies of using different radiotracers were compared (Supplementary Fig. 1d).

Biological tissue penetration assessment. The biological tissue penetration ability of the uorescence emission was evaluated using a piece of 1-mm thick porcine gastric mucosa tissue dissected from a freshly butchered porcine stomach.

18F-FDG of 50 mCi was injected into one well as the control. The other four wells were injected with the mixture of 18F-FDG (50 mCi) and EO (0.15, 0.30, 0.45 or0.60 mg). Images without and with tissue blocking were taken (binning: 8, exposure: 2 min).

Comparison of optical imaging techniques via phantoms. Two identical tissue-mimicking phantoms were made from high-density polyethylene to simulate the optical properties of mouse muscle36. Each was a cube that was 40 40 40 mm3

in dimension. A small hole (diameter 2.5 mm, length 20 mm) was drilled 2 mm beneath the surface (Fig. 4a). A 0.1-ml mixture of 18F-FDG (100 mCi) and EO(1 mg) or 0.1 ml 18F-FDG (100 mCi) alone was injected into each hole. CLI, REFI and FMI were performed for comparison. For CLI and REFI, both 620-nm ltered and unltered images were acquired (exposure 5 min). For FMI, the excitation and emission wavelengths were 465 and 620 nm (exposure 1 s). The experiment was performed in triplicate.

Two glass tubes were lled with 0.1 ml of 18F-FDG (50 mCi) or a 0.1-ml mixture of 18F-FDG (50 mCi) and EO (0.15 mg). They were then subcutaneously implanted into a euthanized nude mouse. PET (Genisys PET, SoeBiosciences), CLI, REFI and FMI were performed. The experiment was performed in triplicate.

Animal experiments. All animal experiments were conducted in compliance with the guidelines of the Institutional Animal Care and Use Committee of General Hospital of Chinese Peoples Armed Police Forces. All animal procedures were performed isourane gas anaesthesia (3% isouraneair mixture), and all efforts were made to minimize suffering. The Balb/c nude mice were obtained from the Laboratory Animal Center of the Chinese Academy of Medical Sciences. Seven- to eight-week-old mice were used for experiments. Female mice were applied for Bcap-37 and 4T1-luc2 studies, and male mice were applied for U87MG and HepG2 studies. The subcutaneous tumour and breast cancer mice models were established by subcutaneously injecting 5 106 tumour cells in nude mice. The orthotopic liver

tumour mice models were established by performing a laparotomy in mice under isourane gas anaesthesia and injecting 5 106 HepG2 cells into the liver.

In vivo validation of REFI. In the experiments of the intratumoural administration route, the control and experimental groups (three Bcap-37 xenografts per group) were tail-vein injected with 800 mCi of 18F-FDG. The tumour tissue of the experimental group received direct injection of 0.05 ml of EO solution (diluted in normal saline, 1 mg ml 1) 10 h prior to tail-vein injection. Forty minutes after the injection of 18F-FDG, CLI and REFI were performed for both groups (exposure5 min) and the FMI was performed for the experiment group (exposure 2 s).

18F-FDG (100 mCi) was tail-vein injected into three 4T1-luc2 xenografts. CLI was performed 40 min after the injection (binning 4, exposure 5 min and aperture f1). Then, the EO solution (25 ml, 1 mg ml 1) was locally injected into the tumour lesions, and REFI was taken immediately (Supplementary Fig. 2).

In the experiments of i.v. administration, EO solution (0.1 ml, 1 mg ml 1) was tail-vein injected into three U87MG-xenografted mice as the experiment group.

Twenty-four hours later, both control and experimental groups received injections of 0.1 ml of 18F-FDG (500 mCi). After another 40 min, CLI and REFI were conducted with 620-nm ltering and no ltering (exposure 5 min).

EO solution (0.1 ml, 1 mg ml 1) and 0.1 ml of 18F-FDG (480 mCi) were mixed and tail-vein injected into three Bcap-37 xenografted mice. The control group only received the injection of 18F-FDG (480 mCi). Longitudinal observations were performed at 40, 55, 60, 70, 80, 100, 110 and 130 min after the injection (exposure 5 min).

In vivo comparison between REFI and FMI using ICG. Three HepG2 liver tumour mice models were used to compare the FMI and REFI. Volume of 0.1 ml ICG solution (1 mg ml 1) was tail-vein injected into the mice. Twenty-four hours later,

EO (0.1 ml, 1 mg ml 1) and 11C-CHO (250 mCi) was tail-vein injected into the same mice. Forty minutes later, the abdominal cavity of the mice was surgically opened to expose the liver, and REFI was performed with binning 4, exposure 5 min, 620-nm ltering (Supplementary Fig. 3b). Then, FMI was performed with excitation 795 nm and emission 835 nm (binning 4, exposure 2 s, Supplementary Fig. 3a).

In vitro phantom comparison between REFI and PET. A glass bottle was lled with 12 ml 18F-FDG with the concentration of 3.75 mCi ml 1. Then, a glass capillary 1 mm in diameter was inserted inside. The tip of the capillary was lled

with 0.5 ml EO (1 mg ml 1) and 0.5 ml 18F-FDG (4.17 mCi ml 1). The capillary tip was used to simulate a very small tumour lesion with a volume of 1 mm3, and the glass bottle was used to simulate a mouse body. The small lesion had a slight higher uptake of 18F-FDG and EO than its surroundings. Then, the phantom was imaged using REFI (binning 4, exposure 5 min, no ltering) and PET respectively (Supplementary Fig. 4).

In vivo multimodality comparison. EO solution (0.1 ml, 1 mg ml 1) and 0.1 ml of

18F-FDG (280 mCi) were mixed and tail-vein injected into three 4T1-luc2 dual tumour mice models. The mice received subcutaneous injection of 4T1-luc2 tumour cells 6 days ago. Volume of 0.1 ml of 18F-FDG (280 mCi) was injected into another three identical models. For each mouse, PET was scanned 35 min after injection with 10 min data acquisition, and then REFI and CLI were immediately performed with no ltering and 620 nm ltering (exposure 5 min). For FMI, three identical models received i.v. injection of QD620 (0.1 ml, 10 mg ml 1) and were scanned 50 min later, using system with 465 nm excitation and 620 nm emission (binning, 4 and exposure, 2 s).

In the experiments of in vivo small tumour lesion imaging, for PET, CLI and REFI, the protocols were exactly the same as the protocols of the dual tumour mice models imaging study, except that all tumour mice models were imaged 65 h after subcutaneous injection of 4T1-luc2 cells into the left lower abdominal mammary fat pad. For FMI, same protocol was applied for using QD620. However, for targetted uorescence imaging, 0.1 ml (100 nmol ml 1) RJ2-DG750 was i.v.

injected into the mice models. The mice were scanned 2 h later with 745-nm excitation and 800-nm emission (binning, 4 and exposure, 2 s).

Six 4T1-luc2 orthotopic breast cancer mice models were established to perform the comparison. Thirty-six hours after tumour transplantation, BLI was applied to verify the location of the tumour. Then PET, CLI and REFI were performed to detect the tumour. Sixty-ve hours after tumour transplantation, the same procedure was applied again (Supplementary Fig. 5). For PET, CLI and REFI, the tail-vein injection of 100 mCi 18F-FDG was applied. For REFI (three of the six mice models), the tail-vein injection of 0.1 ml EO (1 mg ml 1) was applied.

Cytotoxicity assay. In vitro cytotoxicity was measured by performing MTT assays on human umbilical cord mesenchymal stem cells. Different concentrations of EO (0, 50, 200 and 400 mg ml 1, diluted in PBS) were added to the wells. The samples were observed 24 h later using a conventional uorescence microscope (Olympus IX71).

Biodistribution of EO. Three mice were killed 40 min after the tail-vein injection of EO as the experiment group, and another three mice without EO injection were killed as control. Heart, kidney, liver, lung and spleen were ground using tissue-grinding pestles, and 0.2 ml PBS was added during grinding for dilution. Then each sample (0.1 ml), including 0.1 ml blood and 0.1 ml urine, was mixed with 0.1 ml

18F-FDG (100 mCi), respectively. Samples from the control followed the same procedure. Then optical images were acquired using the IVIS Spectrum system. The EO concentration in each sample was dened as optical signal intensity (EO-injectedcontrol).

Hematoxylin and eosin staining. To evaluate the tissue toxicity of EO, the Bcap-37 xenografted mice were killed immediately after in vivo imaging for histological examination. Tumours and organs were xed in 4% formalin- and parafn-embedded sections (4-mm thickness) were prepared for H&E staining. The slices were examined using a digital microscope (Leica QWin).

Statistical analyses. Statistical comparisons were made using Students t-test and GraphPad Prism 5 software. P values o0.05 were considered to indicate signicance. Average and s.d. were calculated for experiments performed in triplicate.

No s.d. indicated relevant measurement performed only once.

References

1. Robertson, R. et al. Optical imaging of Cerenkov light generation from positron-emitting radiotracers. Phys. Med. Biol. 54, N355N365 (2009).

2. Ruggiero, A., Holland, J. P., Lewis, J. S. & Grimm, J. Cerenkov luminescence imaging of medical isotopes. J. Nucl. Med. 51, 11231130 (2010).

3. Xu, Y., Liu, H. & Cheng, Z. Harnessing the power of radionuclides for optical imaging: Cerenkov luminescence imaging. J. Nucl. Med. 52, 20092018 (2011).

4. Xu, Y. et al. Proof-of-concept study of monitoring cancer drug therapy with Cerenkov luminescence imaging. J. Nucl. Med. 53, 312317 (2012).

5. Li, C., Mitchell, G. S. & Cherry, S. R. Cerenkov luminescence tomography for small-animal imaging. Opt. Lett. 35, 11091111 (2010).

6. Hu, Z. et al. Experimental Cerenkov luminescence tomography of the mouse model with SPECT imaging validation. Opt. Express 18, 2444124450 (2010).

7. Robertson, R., Germanos, M. S., Manfredi, M. G., Smith, P. G. & Silva, M. D. Multimodal imaging with (18)F-FDG PET and Cerenkov luminescence imaging after MLN4924 treatment in a human lymphoma xenograft model. J. Nucl. Med. 52, 17641769 (2011).

NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications 11

& 2015 Macmillan Publishers Limited. All rights reserved.

ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms8560

8. Holland, J. P., Normand, G., Ruggiero, A., Lewis, J. S. & Grimm, J. Intraoperative imaging of positron emission tomographic radiotracers using Cerenkov luminescence emissions. Mol. Imaging 10, 177186 (2011).

9. Hu, Z. et al. Cerenkov luminescence tomography of aminopeptidase N (APN/ CD13) expression in mice bearing HT1080 tumors. Mol. Imaging 12, 173181 (2013).

10. Thorek, D. L. et al. Cerenkov imaging - a new modality for molecular imaging. Am. J. Nucl. Med. Mol. Imaging 2, 163173 (2012).

11. Hu, Z. et al. Three-dimensional noninvasive monitoring iodine-131 uptake in the thyroid using a modied Cerenkov luminescence tomography approach. PLoS ONE 7, e37623 (2012).

12. Liu, H. et al. Optical imaging of reporter gene expression using a positron-emission-tomography probe. J. Biomed. Opt. 15, 060505 (2010).

13. Mitchell, G. S., Gill, R. K., Boucher, D. L., Li, C. & Cherry, S. R. In vivo Cerenkov luminescence imaging: a new tool for molecular imaging. Philos. Trans. A Math. Phys. Eng. Sci. 369, 46054619 (2011).

14. Liu, H. et al. Intraoperative imaging of tumors using Cerenkov luminescence endoscopy: a feasibility experimental study. J. Nucl. Med. 53, 15791584 (2012).

15. Liu, H. et al. Molecular optical imaging with radioactive probes. PLoS ONE 5, e9470 (2010).

16. Spinelli, A. E. et al. First human Cerenkography. J. Biomed. Opt. 18, 20502 (2013).

17. Thorek, D. L., Ried, C. C. & Grimm, J. Clinical Cerenkov luminescence imaging of (18)F-FDG. J. Nucl. Med. 55, 9598 (2014).

18. Chin, P. T. et al. Optical imaging as an expansion of nuclear medicine: Cerenkov-based luminescence vs uorescence-based luminescence. Eur. J. Nucl. Med. Mol. Imaging 40, 12831291 (2013).

19. Thorek, D. L., Ogirala, A., Beattie, B. J. & Grimm, J. Quantitative imaging of disease signatures through radioactive decay signal conversion. Nat. Med. 19, 13451350 (2013).

20. Thorek, D. L., Das, S. & Grimm, J. Molecular imaging using nanoparticle quenchers of Cerenkov luminescence. Small 10, 37293734 (2014).

21. Dothager, R. S., Goiffon, R. J., Jackson, E., Harpstrite, S. & Piwnica-Worms, D. Cerenkov radiation energy transfer (CRET) imaging: a novel method for optical imaging of PET isotopes in biological systems. PLoS ONE 5, e13300 (2010).

22. Axelsson, J., Davis, S. C., Gladstone, D. J. & Pogue, B. W. Cerenkov emission induced by external beam radiation stimulates molecular uorescence. Med. Phys. 38, 41274132 (2011).

23. Ma, X. et al. Enhancement of Cerenkov luminescence imaging by dual excitation of Er3 , Yb3 -doped rare-earth microparticles. PLoS ONE 8, e77926 (2013).

24. Zhuravleva, M., Friedrich, S. & Melcher, C. L. The europium oxidation state in CsSrI3:Eu scintillators measured by X-ray absorption spectroscopy. Opt. Mater. 36, 670674 (2014).

25. Shen, J., Sun, L. D. & Yan, C. H. Luminescent rare earth nanomaterials for bioprobe applications. Dalton Trans. 24, 56875697 (2008).

26. Carpenter, C. M. et al. Radioluminescent nanophosphors enable multiplexed small-animal imaging. Opt. Express. 20, 1159811604 (2012).

27. Sun, C. et al. Synthesis and radioluminescence of PEGylated Eu(3 )-doped

nanophosphors as bioimaging probes. Adv. Mater. 23, H195H199 (2011).28. Chen, H. et al. Monitoring pH-triggered drug release from radioluminescent nanocapsules with X-ray excited optical luminescence. ACS Nano 7, 11781187 (2013).

29. Bellevre, D. et al. Staging the axilla in breast cancer patients with 18F-FDG PET: how small are the metastases that we can detect with new generation clinical PET systems? Eur. J. Nucl. Med. Mol. Imaging 41, 11031112 (2014).

30. Soret, M., Bacharach, S. L. & Buvat, I. Partial-volume effect in PET tumor imaging. J. Nucl. Med. 48, 932945 (2007).

31. Hoetjes, N. J. et al. Partial volume correction strategies for quantitative FDG PET in oncology. Eur. J. Nucl. Med. Mol. Imaging 37, 16791687 (2010).

32. Bhaumik, S. & Gambhir, S. S. Optical imaging of Renilla luciferase reporter gene expression in living mice. Proc. Natl Acad. Sci. USA 99, 377382 (2002).

33. So, M.K., Xu, C., Loening, A.M., Gambhir, S.S. & Rao, J. Self-illuminating quantum dot conjugates for in vivo imaging. Nat. Biotechnol. 24, 339343 (2006).

34. Pratx, G., Carpenter, C. M., Sun, C., Rao, R. P. & Xing, L. Tomographic molecular imaging of x-ray-excitable nanoparticles. Opt. Lett. 35, 33453347 (2010).

35. Pratx, G., Carpenter, C. M., Sun, C. & Xing, L. X-ray luminescence computed tomography via selective excitation: a feasibility study. IEEE Trans. Med. Imaging 29, 19921999 (2010).

36. Cong, W. et al. Practical reconstruction method for bioluminescence tomography. Opt. Express. 13, 67566771 (2005).

Acknowledgements

This paper is supported by the National Basic Research Program of China (973 Program) under Grant No. 2011CB707700, and the National Natural Science Foundation of China under Grant No. 81227901, 61231004, 61401462, 61302024, 81471700 and 81230033.

Author contributions

J.T. and H.F.L. initiated the project and contributed to the experimental design as the principal investigators. Z.H., Y.Q., K.W. and X.Z. conceived the experiments, characterized the EO nanoparticles and performed in vitro and in vivo experiments. X.Z., J.Z., T.S., C.B., H.L., Z.W. and J.W. assisted in in vivo PET and optical imaging. Z.L. assisted in toxicity evaluation. K.W. and Z.H. analysed the data and wrote the manuscript. All authors commented and contributed to the manuscript.

Additional information

Supplementary Information accompanies this paper at http://www.nature.com/naturecommunications

Web End =http://www.nature.com/ http://www.nature.com/naturecommunications

Web End =naturecommunications

Competing nancial interests: The authors declare no competing nancial interests.

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

Web End =http://npg.nature.com/ http://npg.nature.com/reprintsandpermissions/

Web End =reprintsandpermissions/

How to cite this article: Hu, Z. et al. In vivo nanoparticle-mediated radio-pharmaceutical-excited uorescence molecular imaging. Nat. Commun. 6:7560 doi: 10.1038/ncomms8560 (2015).

This work is licensed under a Creative Commons Attribution 4.0 International License. The images or other third party material in this article are included in the articles Creative Commons license, unless indicated otherwise in the credit line; if the material is not included under the Creative Commons license, users will need to obtain permission from the license holder to reproduce the material. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/

Web End =http://creativecommons.org/licenses/by/4.0/

12 NATURE COMMUNICATIONS | 6:7560 | DOI: 10.1038/ncomms8560 | http://www.nature.com/naturecommunications

Web End =www.nature.com/naturecommunications

& 2015 Macmillan Publishers Limited. All rights reserved.