日本大阪大學(xué)一個研究小組日前研發(fā)出一種可自主發(fā)光的蛋白，植入這種蛋白的癌細(xì)胞在實(shí)驗(yàn)鼠體內(nèi)肉眼可見，這種發(fā)光蛋白未來或可應(yīng)用到癌癥的早期診斷中。

　　大阪大學(xué)教授永井健治領(lǐng)導(dǎo)的研究小組將一種水母的發(fā)光蛋白與熒光蛋白相結(jié)合，研發(fā)出一種可自主發(fā)出明亮光線的新型蛋白。將含有這種蛋白的癌細(xì)胞移植到實(shí)驗(yàn)鼠體內(nèi)，觀測其在暗箱中的運(yùn)動情況，可肉眼辨認(rèn)出明顯發(fā)光的癌細(xì)胞。

　　綠色熒光蛋白是當(dāng)代生物學(xué)的重要“標(biāo)識”工具，2008年，三名科學(xué)家因在發(fā)現(xiàn)和研究綠色熒光蛋白方面作出貢獻(xiàn)而獲得諾貝爾化學(xué)獎。不過此前的綠色熒光蛋白必須用紫外線照射才能發(fā)光，而這種新蛋白可自主發(fā)出亮光，有望在早期癌癥診斷中發(fā)揮作用。相關(guān)論文已發(fā)表在《自然—通訊》雜志網(wǎng)絡(luò)版上
 

Abstract

　　The use of fluorescent proteins has revolutionized our understanding of biological processes. However, the requirement for external illumination precludes their universal application to the study of biological processes in all tissues. Although light can be created by chemiluminescence, light emission from existing chemiluminescent probes is too weak to use this imaging modality in situations when fluorescence cannot be used. Here we report the development of the brightest luminescent protein to date, Nano-lantern, which is a chimera of enhanced Renilla luciferase and Venus, a fluorescent protein with high bioluminescence resonance energy transfer efficiency. Nano-lantern allows real-time imaging of intracellular structures in living cells with spatial resolution equivalent to fluorescence and sensitive tumour detection in freely moving unshaved mice. We also create functional indicators based on Nano-lantern that can image Ca2+, cyclic adenosine monophosphate and adenosine 5′-t**hosphate dynamics in environments where the use of fluorescent indicators is not feasible. These luminescent proteins allow visualization of biological phenomena at previously unseen single-cell, organ and whole-body level in animals and plants.

　　Subject terms:　　Biological sciences　　Biotechnology　　Cell biology　　

　　Introduction

　　Although fluorescence imaging is widely used, its dependence on external illumination prevents its universal application. For example, fluorescence imaging cannot be easily used to study light-dependent biological processes, such as visual photoreception or photosynthesis. Although optical recording of the light-sensitive retina has been successfully performed using 2-photon excitation with a 930-nm femtosecond laser1, this method is not versatile because many biological molecules have significant absorption at both visible and infrared wavelengths2. Therefore, methods based on fluorescence with both 1-photon and 2-photon excitation cannot always be used to study light-dependent biological processes. Moreover, fluorescence is incompatible with non-invasive deep tissue imaging of whole organisms and other applications where the cellular substrate is autofluorescent (for example, chloroplasts of photosynthetic plants), saturated with photopigments (porphyric hepatocytes, melanocytes or retinal pigment epithelia) or extremely photosensitive. Experimental problems also arise when external illumination is required, as for biological technologies such as optogenetics, chromophore-assisted light inactivation and photolysis of caged compounds, which prevents simultaneous use of fluorescence imaging. Finally, the general power density of external illumination for live-cell microscopy with fluorescence (sub W?cm?2) sometimes causes phototoxic effects in visualized substrates, which alters cellular behaviour and ultimately leads to cell death. In contrast, chemiluminescence generates a visible light signal through a localized chemical reaction without the need for external illumination. Because the kcat values of conventional luciferases range from 0.1 for firefly luciferase (quantum yield=0.5)3 to approximately 4.4 for Renilla luciferase (RLuc)8 (quantum yield=0.053)4, transient expression of micromolar concentrations of luciferases, for example, RLuc8, generates a power density of emitted light that is 1/103 that of the general power density required for fluorescence emission in live-cell imaging (approximately 0.1?μW?cm?2). Therefore, although chemiluminescent proteins, including aequorin and luciferases, have been used to image living cells and organisms5, 6, the light output from these proteins is insufficient to provide temporal and/or spatial resolution equivalent to fluorescence.

　　In the case of luminous organisms, such as the sea pansy Renilla reniformis, the problem of low brightness has been solved in nature by the phenomenon of bioluminescence resonance energy transfer (BRET), in which the excited energy of a luminescent substrate, coelenterazine, bound to RLuc is efficiently transferred intermolecularly to the acceptor Renilla green fluorescent protein (quantum yield=0.3) by a F?rster resonance energy transfer (FRET) mechanism, thereby increasing the emitted photon number approximately six-fold7. Based on this natural intermolecular BRET, intramolecular BRET probes, such as aequorin-GFP8 and BAF-Y9, have been developed. Although these probes allow for live-cell imaging with improved resolution in space and time, they still underperform compared with fluorescent protein-based probes because of low brightness.

　　To address this problem, we obtained a brighter RLuc by random mutagenesis and fusion to a yellow fluorescent protein (YFP) with high BRET efficiency. The fusion protein showed much brighter luminescence than BAF-Y, enabling not only real-time imaging of intracellular structures in living cells but also sensitive tumour detection in freely moving mouse. Moreover, we developed Ca2+, cyclic adenosine monophosphate (cAMP) and adenosine 5′-t**hosphate (ATP) indicators based on this bright luminescent protein. These luminescent indicators will allow visualization after the optical control of cellular or enzymatic activity at the single-cell, organ and whole-body level in animals and plants.

　　Results

　　Design and application of the bright luminescent protein

　　To improve brightness, we designed a chimeric protein based on eBAF-Y9 (Supplementary Note 1), which is a fusion of enhanced YFP and an enhanced RLuc, RLuc8 (4). Because BRET efficiency, and thus brightness, depends on the photochemical and physical properties of the donor and acceptor, we honed these parameters by improving donor brightness, maximizing spectral overlap between the donor emission and acceptor absorbance using Venus10 and optimizing the spatial arrangement of the donor and acceptor in the fusion construct (Supplementary Note 1, Supplementary Fig. S1 and Supplementary Table S1). The resulting protein, which we called Nano-lantern, reminiscent of a light source with nanometre scale (Fig. 1a), exhibited 5.3 and 2.9 times greater luminescence than RLuc8 and eBAF-Y, respectively, over the entire emission range (Fig. 1b). The improved brightness of Nano-lantern should generate power densities in the range of 1?μW?cm?2 (versus 0.1?μW?cm?2 for RLuc8) following transient overexpression in the micromolar range in human cells, and thus increase imaging potential.

　　Figure 1: Development of the bright luminescent protein Nano-lantern.
 

　　(a) Schematic of the domain structure of Nano-lantern. Numbers represent the relative amino-acid position in the original protein. (b) Emission spectra of equimolar amounts of the luminescent proteins Nano-lantern, VRL10.3, eBAF-Y9, RLuc8_S257G, RLuc8 and RLuc. Emission spectra measurements were performed at least in t**licate, and the averaged data are shown. (c) Luminescence (left) and fluorescence (right) imaging of HeLa cells expressing Nano-lantern targeted to cytoplasm, mitochondria and histone H2B. The exposure times for the luminescence images were 3?s, 3?s and 1?s, respectively. The exposure time for all fluorescence images was 1?s. The reference fluorescence signal was captured by exciting Venus with light at 490?nm. Scale bars, 50?μm.

　　Indeed, when Nano-lantern was expressed in HeLa cells, a luminescence image with quality almost comparable to that of fluorescence images was obtained. Nano-lantern and fusions with defined localization tags enabled visualization of cell compartments and organelles, such as the cytoplasm, mitochondria and nucleus (histone H2B), in living cells by using low magnification lens (X20 dry objective) with brief, 1–3-s exposures (Fig. 1c). Nano-lantern also allowed for visualization of finer structures, including microfilaments, microtubules and their tips (EB3) with a high magnification lens (X60 oil-immersion objective) and longer, 3–60?s, exposures (Supplementary Fig. S2). Therefore, the enhanced luminescent signal (μW?cm?2) from Nano-lantern enables imaging at high spatial resolution until at least the subcellular level, with a quality approaching that of fluorescent protein (FP)-based imaging. This is because although the power density of the emitted light of Nano-lantern is still ~1/102 smaller than that of general fluorescence emissions (at ~100?μW?cm?2), image quality is defined by the signal-to-noise ratio (S/N ratio). In fluorescence imaging, this is compromized by high noise levels from stray excitation light and autofluorescence from both the sample and the medium. In chemiluminescence imaging, quality does not suffer from low signal levels because of the very low noise level. Because external illumination is not needed, an adequate S/N ratio is preserved.

　　Currently, chemiluminescence is commonly used for non-invasive tumour detection in living mice for tumour metastasis research and anti-cancer-drug screening11. Brighter luminescence should enable detection of smaller tumours. Indeed, when adenocarcinoma colon26 cell lines that stably express Nano-lantern or RLuc8 (colon26/Nano-lantern and colon26/RLuc8) was subcutaneously implanted into BALB/c nude mice (Fig. 2a), and we detected as few as 103 Nano-lantern-expressing cells 1 day after transplantation (Fig. 2b), which is 1 order of magnitude better than RLuc8-expressing tumours (Fig. 2c).

　　Figure 2: Luminescence imaging of mice with Nano-lantern-expressing tumours.
 

　　(a) Position of colon26 cells expressing Nano-lantern (green circles) or RLuc8 (red circles) in the transplanted mice (dorsal view). (b,d) Representative luminescence images of the indicated numbers of injected cells at 600?s (b) and 1?s exposures (d). (c) Comparison in relative light units (RLU, pixel-integrated counts per s) of 1 × 107 cells of colon26/RLuc8 and Nano-lantern. Measurements were performed in t**licate, and the averaged data and s.d. are shown. (e) Consecutive frames of video-rate images of Nano-lantern-expressing tumour cells in an unshaved mouse. The luminescent signal every 60?ms is shown in green in the still images and as a series of pseudo colours (blue, cyan, green, yellow, red and magenta) in the merged image. Scale bars, 1?cm.

　　The increased brightness of Nano-lantern allowed for imaging of 105 and 106 cells at 0.1 and 1?Hz sampling rates, respectively (Fig. 2d). Surprisingly, we could even obtain video rate (30?Hz) imaging of tumours 17 day after implantation of 106 of tumour cells in freely moving unshaved mice (Fig. 2e and Supplementary Movie 1). In previous reports of luminescent tumour cell imaging in non-anaesthetized mice12, successful images were only obtained with nude mice, larger tumour samples and prolonged exposure times (160 or 200 versus 33?ms in this report). Increased sensitivity and faster imaging without the need for anaesthesia using Nano-lantern greatly facilitates the analysis of tumour growth and metastasis in living animals.

　　Functional indicators based on Nano-lantern

　　Having demonstrated the benefit of the improved luminescent properties of Nano-lantern, micrometre spatial resolution in living cells and improved tumour resolution in whole animals, we sought to develop functional indicators for bioactive molecules such as Ca2+, cAMP and ATP based on Nano-lantern. Thus far, two types of chemiluminescence-based functional indictors have been reported; one uses the complementation of split luciferase (CSL), in which the luciferase molecule is split, with one portion fused to the sensor domain of a bioactive molecule and the other to a peptide that binds the sensor domain. This type of indicator emits a signal upon bioactive molecule-binding-mediated reconstitution of the split luciferase13. The other uses a BRET-based indicator composed of the sensor domain of a bioactive molecule flanked by an intact chemiluminescent donor and a fluorescent acceptor, in this case, a cellular response is detected as a change in BRET efficiency14. Although CSL-type indicators have a larger signal change, their absolute intensity is often too low to be imaged. Conversely, BRET-based type indicators have sufficient signal intensity for imaging; however, its signal change upon target molecule binding is generally small. To overcome these drawbacks, we sought to insert a sensor domain into Nano-lantern to create a hybrid CSL/BRET indicator that would exhibit a signal change with sufficient intensity for imaging (Supplementary Fig. S3a).

　　A Ca2+ indicator based on Nano-lantern

　　Given the novel hybrid approach, we first sought to develop a Nano-lantern-based indicator for Ca2+, one of the most important second messengers in living cells. We cloned the Ca2+-sensing domain (CaM-M13) of yellow cameleon, an established FRET-based Ca2+-indicator15, and searched for the optimal insertion site in Nano-lantern that would yield a large signal change and preserve high intensity emission. Other important properties, including Ca2+ affinity and reversibility of the catalytic activity of luciferase, were also examined. Ultimately, of the sites tested, we found that insertions into the non-structural Gly228/Gly229 loop in RLuc8 that links the hydrolase and cap domains were well tolerated (Supplementary Fig. S3b and Supplementary Note 2). The resulting construct, named Nano-lantern (Ca2+), showed a 300% signal change at 35% brightness of the parental protein at saturating Ca2+ concentrations (Fig. 3a). The Ca2+ affinities of Nano-lantern (Ca2+) and its derivatives (Supplementary Fig. S4 and Supplementary Table S2) were almost identical to that of the parental yellow cameleon series, suggesting that reconstitution of split-Rluc8 in Nano-lantern (Ca2+) did not interfere with the Ca2+-induced conformational change in the CaM-M13 domain (Supplementary Note 2).

　　Figure 3: Development of a Nano-lantern-based luminescent Ca2+ indicator.
 

　　(a) Schematic representation of the domain structures of CaM-M13@91_Nano-lantern and CaM-M13@228_Nano-lantern (Nano-lantern (Ca2+)). (b) Relative brightness of recombinant Nano-lantern, CaM-M13@91_Nano-lantern and CaM-M13@228_Nano-lantern (Nano-lantern (Ca2+)), with or without Ca2+. Measurements were performed at least in t**licate, and the averaged data and s.d. are shown. (c) A series of pseudo-coloured ratio images of HeLa cells expressing Nano-lantern (Ca2+) taken at video rate, following histamine stimulation. Scale bar, 10?μm. (d) Time course of the L/L0 ratio change at an ROI (white box in (c)) in a HeLa cell expressing Nano-lantern (Ca2+). Number indicates the time point of each image in (c). Representative data from at least five experiments are shown in (d) and (e). (e) A typical time course of the spatial L/L0 ratio change in ROIs (cyan, magenta and yellow boxes in (c)) in another HeLa cell. (f) Time course of the L/L0 ratio change in a rat hippocampal neuron co-expressing Nano-lantern (Ca2+) with the ChR2. ChR2 was repeatedly photo-activated in illumination sessions (~2?s), consisting of 15 cycles of camera exposure (100?ms) and pulses of blue light (0.8?ms) that were delivered during periods (30?ms) when camera exposure was off. Numbers indicate the time points of each image in (g). Representative data from five measurements are shown. (g) A series of pseudo-coloured ratio images of a rat hippocampal neuron expressing Nano-lantern (Ca2+). Scale bar, 10?μm.

　　To compare the performance of this Ca2+ indicator to a known standard, we expressed Nano-lantern (Ca2+) in HeLa cells. Following stimulation with 10?μM histamine, an acute Ca2+ spike (S/N=70) followed by Ca2+ oscillations with smaller amplitudes (S/N=29) were detected with a sampling rate as fast as 30?Hz (Fig. 3c). We could even observe the propagation of a cytoplasmic Ca2+ wave from one end of the cell to the other, as was previously observed using yellow cameleon15 (Fig. 3e and Supplementary Movie 2). Histamine-induced Ca2+ oscillations were abolished by the addition of the histamine receptor antagonist cyproheptadine, confirming that the luminescent signal change reflects changes in cytoplasmic [Ca2+] (Supplementary Fig. S5). Previous CSL-based Ca2+ indicators13 were not fully reversible in response to physiological or stimulated Ca2+ release. We confirmed by co-imaging Nano-lantern (Ca2+) with R-GECO116 that the kinetics of Nano-lantern (Ca2+) parallel those of fluorescence-based probes (Supplementary Fig. S6). Encouragingly, Nano-lantern (Ca2+), when expressed in dissociated rat hippocampal neurons (Supplementary Fig. S7), and imaged at 10?Hz, could report spontaneous cytoplasmic [Ca2+] spikes and transient [Ca2+] changes due to reversible stimulation (50?mM KCl).

　　To realize the theoretical advantage of Ca2+ imaging using luminescence probes over fluorescence-based indicators, we performed Ca2+ imaging combined with the control of neuronal activity using optogenetic tools. Ectopic expression of photoreceptor proteins, such as halorhodopsin and channelrhodopsin2 (ChR2), has enabled the functional analysis of neuronal cells and their networks with ms time resolution17. Yet, these tools are spectrally incompatible with fluorescent Ca2+ indicators. Excitation light can be used either to control the photoreceptor or to measure the Ca2+ transient, but not for both in the same cell at the same time. Because luminescence imaging can be performed in the absence of external light, it can be partnered with an optogenetic strategy without undesired interference due to the activity of the photoreceptor. To demonstrate this, we co-expressed Nano-lantern (Ca2+) and ChR2 in dissociated rat hippocampal neurons. The luminescent signal was imaged every 100?ms, and ChR2 was activated with a 0.8-ms pulse of blue light (438?nm) during the dead time of the CCD chip for data transfer18. In response to 15 cycles of activation light, a rapid increase was observed in the luminescent signal (Fig. 3f and Supplementary Movie 3). We also detected a gradual decrease in the signal to the basal level when the illumination was terminated. Catalytic consumption of substrate (coelenterazine-h) causes a gradual decrease in both the baseline and the amplitude of transients. These results suggest that these synergistic neurobiology tools can now be combined into a single experimental approach. In the future, it may also be possible to combine other optical control systems, such as chromophore-assisted light inactivation of target molecules19 and photolysis of caged compounds20, with luminescent functional indicators.

　　Improved cAMP indicators based on Nano-lantern

　　Imaging intracellular cAMP, another important second messenger, using fluorescent indicators has been challenging because of the small signal change (1–15%) with current FP-based indicators21, 22, 23. To examine whether the CSL/BRET hybrid strategy could be applied to construct a high-performance cAMP indicator, we cloned a series of cAMP-binding motifs from the PKA regulatory subunit or EPAC1, into Nano-lantern at the 228/229 position (similar to in Nano-lantern (Ca2+)) (Supplementary Fig. S8a). Recombinant proteins were purified and screened for intensity changes in response to cAMP. Among the nine constructs tested, three cAMP-recognition motifs gave a large signal change with distinct cAMP affinities (Kd=0.4, 1.6 and 3.3?μM), and one of these three, Nano-lantern containing a fragment of EPAC1 (G170 to A327 with Q270E), which we named Nano-lantern (cAMP1.6) (from its Kd), showed the largest signal increase (130%) upon cAMP stimulation (Fig. 4a, Supplementary Fig. S8b–d and Supplementary Note 3). In contrast, FRET-based indicators with an identical cAMP-recognition domain showed only a small signal change (11.7% for CFP/YFP-based constructs; Supplementary Fig. S8e and Supplementary Note 3).

　　Figure 4: Development of a Nano-lantern-based luminescent cAMP indicator.

　　(a) Schematic of the domain structure of Nano-lantern (cAMP). (b) Relative brightness of recombinant Nano-lantern and Nano-lantern (cAMP) with or without cAMP. These measurements were performed at least in t**licate, and the averaged data and s.d. are shown. (c) A cAMP image of ~100,000 D. discoideum cells expressing Nano-lantern (cAMP1.6) during aggregative morphogenesis. Scale bar, 1?mm. (d) Space and time plot along the yellow lines shown in (c) for 60?min. (e) Time course of the luminescence intensity change at the ROIs indicated in (c). These are typical data from five measurements.

　　To examine the ability of Nano-lantern (cAMP1.6) to detect intracellular cAMP concentration changes, we used Dictyostelium discoideum cells because cytoplasmic cAMP concentration (hereafter [cAMP]) in Dictyostelium cells increases in response to extracellular cAMP stimulation24, 25, which indirectly increases [cAMP] through the action of the cell-surface receptor and cAMP synthase (adenylate cyclase A). Cells stably expressing Nano-lantern (cAMP1.6) whose Kd was best fitted to the estimated range of oscillating cellular [cAMP] (Supplementary Fig. S8d and Supplementary Table S3)26 showed a rapid increase in luminescent signal (40% increase in initial intensity), followed by a gradual decay in response to 100?μM cAMP stimulation (Supplementary Fig. S9). These luminescence signal changes do indeed reflect cytoplasmic cAMP dynamics, as we confirmed that this transient cAMP-stimulated signal increase was lost in adenylate cyclase A mutant cells (Supplementary Fig. S10). During morphogenesis, populations of Dictyostelium cells form circular and spiral aggregation patterns27. This is achieved by the repeated intracellular synthesis and intercellular relay of cAMP (at a 6–10-min interval), which functions as a chemoattractant molecule. Direct imaging of this with fluorescence reporters has never been previously possible because of their limited dynamic range. When cells stably expressing Nano-lantern (cAMP1.6) were cultured on an agar plate (106 cells cm?2), we observed a propagating spiral wave of cAMP with a 6-min periodicity that had a larger signal change (Fig. 4c, and Supplementary Movie 4) than that detected with a currently available FRET-based indicator (Supplementary Fig. S11 and Supplementary Movie 5).

　　We were able to successfully detect cAMP at the single-cell level in Dictyostelium amoebae (cell volume, approximately 1/20 that of HeLa cells) with FRET-based indicators (Supplementary Fig. S11b); however, this was not possible with Nano-lantern (cAMP). In fact, approximately 20 single cells were required to obtain a luminescence image at high magnification (Supplementary Fig. S9) due to the low absolute intensity in this organism because of its lower expression compared with that in mammalian cells. However, these results still demonstrate that luminescent indicators have a clear performance advantage over fluorescence-based indicators when performing macroscopic analysis of populations of living cells for prolonged periods without obvious cell injury.

　　Visualizing ATP production in plant chloroplasts

　　Luminescence imaging can be applied in systems where fluorescence observation is impossible. A typical example is the chloroplast of photosynthetic plants, which converts the energy of sunlight into chemical energy in the form of NADPH and ATP. Fluorescence-based imaging cannot provide the needed spatiotemporal resolution for this process due to the strong autofluorescence of chlorophyll and the intrinsic photosensitivity of photosynthesis. Therefore, as another demonstration of luminescence-based functional imaging, we sought to visualize the dynamics of ATP production and utilization in chloroplasts. To this end, we developed a Nano-lantern-based ATP indicator using the ε subunit of the bacterial FoF1-ATP synthase similar to that found in the fluorescent ATP indicator ATeam28. This was also inserted into the 228–229 position in a modified Nano-lantern (VRL10.0_S257G) (Supplementary Note 4). The generated construct, which we named Nano-lantern (ATP1), exhibited a 200% increase in light output upon the addition of ATP with a Kd of 0.3?mM (Fig. 5a, Supplementary Fig. S12 and Supplementary Table S4).

　　Figure 5: Development of a Nano-lantern-based luminescent ATP indicator.

　　(a) Schematic of the domain structure of Nano-lantern (ATP1). (b) Relative brightness of recombinant Nano-lantern and Nano-lantern (ATP1) with or without ATP. These measurements were performed at least in t**licate, and the averaged data and s.d. are shown. (c) Time course of the luminescence intensity change in a plant leaf expressing Nano-lantern (ATP1) with (dashed line) or without (solid line) the mitochondrial ATP synthesis inhibitor oligomycin A. Representative data from five measurements are shown. (d) Plant leaf luminescence images of CT-Nano-lantern (ATP1), chloroplast autofluorescence and merge. Scale bar, 20?μm. (e) A typical pattern of the time course of the luminescence intensity change in a plant leaf expressing CT-Nano-lantern (ATP1) with (dashed line) or without (solid line) the photosynthesis inhibitor 3-(3,4-dichlorophenyl)-1,1-dimethylurea with weak light irradiation (0.3?mW?cm?2). Representative data from five measurements are shown. (f) A typical pattern of the time course of the luminescence intensity change in a plant leaf expressing CT-Nano-lantern (ATP1) (solid line) or in a wild-type plant (dashed line) with pulsed light (8.2?mW?cm?2) indicated as a white rectangle in the scheme above the graph. Representative data from five measurements are shown.

　　When expressed in a live plant leaf, Nano-lantern (ATP1) was able to detect the decrease in cytoplasmic ATP levels resulting from the addition of a mitochondrial ATP synthase inhibitor, oligomycin A (Fig. 5c). To visualize ATP production in chloroplasts during photosynthesis, transgenic Arabidopsis plants were constructed that could target Nano-lantern (ATP1) to the chloroplast stroma by using a transit peptide fusion29 (CT-Nano-lantern (ATP1)). The luminescent signal for CT-Nano-lantern (ATP1) in the mesophyll of leaf cells overlapped with the red autofluorescence signal of the chloroplasts (Fig. 5d), and an increase in ATP levels was detected after weak light irradiation. This response was effectively suppressed by treatment with the photosynthesis inhibitor, 3-(3,4-dichlorophenyl)-1,1-dimethylurea30 (Fig. 5e), demonstrating that this response in ATP levels was photosynthesis dependent.

　　Further monitoring of the regulation of chloroplast ATP levels was possible by pulsing strong photo-irradiation (Fig. 5f and Supplementary Movie 6). During photo-irradiation, ATP levels increased rapidly in the first few seconds, and then continued to increase at a moderate rate. We speculate that this slower increase is due to downregulation of the photosynthesis rate by energy-dependent quenching (qE quenching)31, a rapid mechanism used to modulate the amount of light accepted and to prevent the harmful effects of excess light irradiation. To our knowledge, this is the first visualization of qE quenching-mediated downregulation of ATP synthesis in live plant cells.

　　Taken together, these results demonstrate the potential of Nano-lantern-based indicators as powerful tools for functional observation in biological materials with strong autofluorescence and photosensitivity. Other cells with these properties include porphyric hepatocytes and erythrocytes, which accumulate the photo-sensitizing substance porphyrin, and retinal ganglion cells, which absorb light in animal retinas. Nano-lantern-based indicators may be useful for analysing the native properties of these cells, which cannot always be determined with existing FP (because of the need for external illumination) or chemiluminescent indicators (because of their poor indicator performance).

　　Discussion

　　By optimizing the BRET efficiency between Venus and improved RLuc8, we developed Nano-lantern, the brightest luminescent protein reported to date. This enhanced brightness enables luminescent imaging on a micrometre scale in single cells, and tumour imaging in freely moving, unshaved animals. We expect that the increased brightness of Nano-lantern will make more advanced applications possible, such as high-throughput drug screening and single-cell tracking in living animals and plants. In addition, Nano-lantern can help realize the other advantages of low-intensity light imaging, including diminished photobleaching and phototoxicity, which allow for prolonged observation without harming the cellular substrate. However, rapid consumption of the luminescent substrate by Nano-lantern hampers prolonged observation of biological events, especially, in the case of whole-body imaging. Therefore, to facilitate indefinite observation, a method that is applicable to the entire organism and provides a continuous luminescent substrate must be developed.

　　An important aspect of this study is the development of a simple method for the construction of high-performance luminescence-based indicators by inserting the sensor domain of bioactive molecules into the non-structural Gly228/Gly229 loop of the RLuc8 moiety in Nano-lantern. However, saturation screening to determine better insertion sites might also improve the functional indicator performance in terms of the rate constants for the association and dissociation of the ligand with the indicator, both of which are important properties for reliable reporting of bioactive molecule dynamics.

　　Finally, a further increase in brightness is clearly desirable. The efficiency of intermolecular BRET between RLuc and Renilla green fluorescent protein was reported to be 1.0 (ref. 7), suggesting that Nano-lantern, the BRET efficiency of which is estimated to be approximately 0.2, could be improved. If this was achieved, luminescent proteins might allow imaging on even smaller spatial or temporal scales, possibly even at the single-molecule level.