Selective and sensitive visualization of endogenous nitric oxide in living cells and animals by a Si-rhodamine deoxylactam-based near-infrared fluorescent probe

A Si-rhodamine deoxylactam-based near-infrared fluorescent probe has been successfully developed for the imaging of endogenous NO in living cells and mouse models.


Introduction
Nitric oxide (NO), a ubiquitous messenger molecule in biological systems, is endogenously produced mainly by three NO synthases (eNOS, nNOS, and iNOS) from NADPH, O 2 and Larginine. [1][2][3][4][5] NO produced via the eNOS and nNOS pathways is on the physiologic level, and plays various physiological roles in the human body, such as the inhibition of platelet aggregation and adhesion, blood vessel relaxation and angiogenesis, and neurotransmission. However, upon induction by the tumor necrosis factor TNF-a and lipopolysaccharide (LPS), a high level of intracellular NO could be produced via the iNOS pathway, which completely transforms the biological actions of NO by its fast reaction with a superoxide radical (O 2 À c) to produce peroxynitrite (ONOO À ), an extremely powerful oxidant. 6-8 ONOO À , despite its positive functions in the immune response and redox regulation of cell signal, 6 could severely damage a wide variety of molecular targets, including lipids, DNA, proteins and enzymes, ultimately leading to mitochondria dysfunction and cell death. 7,8 Moreover, there is no biological defense system against ONOO À . As a result, ONOO À has been implicated in a variety of disease states, including diabetes, Alzheimer's disease, cancer, arthritis, autoimmune diseases, and other disorders. 4 Thus, the development of methods and tools that can quickly, sensitively, and selectively track NO generation in biological systems is very important for unraveling its precise roles in health and disease.
Among various cellular biology tools, uorescent probes, combined with uorescence microscopy, have shown the unique advantages for mapping the spatial and temporal distributions of biological molecules in biological systems, due to their sensitivity, visualization, and noninvasiveness. Accordingly, in the past decade, a large number of uorescent NO probes have been exploited, which typically utilize the specic reactions of NO with o-phenylenediamine (OPD) [9][10][11][12][13][14][15][16][17] and metalligand complexes. [18][19][20][21][22] Among them, OPD-based uorescent NO probes were the earliest developed and are the most versatile uorescent indicators for NO to date, and the corresponding sensing mechanism is based on the reaction of the OPD group with NO under aerobic conditions to form triazole derivatives, thereby turning on the uorescence by inhibiting the photoinduced electron transfer (PeT) process. Even so, this type of probe suffers from some limitations, such as possible interference by dehydroascorbic acid/ascorbic acid/methylglyoxal (DHA/AA/MGO), [23][24][25][26] pH-sensitive uorescence output near neutral pH, [9][10][11][12] and relatively long response time (commonly more than 5 min). As such, in recent years, some new strategies have been developed, such as diazo ring formation, 27 NOtriggered ring-opening of the OPD-locked rhodamine lactam, 28 reductive deamination of aromatic primary monoamines, 29 the formation of Se-NO bonds, 30 and the aromatization of Hantzsch ester. 31 Although improved in many aspects, most of these probes still suffer from relatively long uorescence response times, thus being unsatisfactory for real-time tracking of NO in biological systems (half-life of NO: 0.1-5 s). In addition, for tissue and living animal imaging, uorescent probes with excitation and emission wavelengths in the near-infrared (NIR) region (650-900 nm) are especially favorable due to the high tissue penetration and low phototoxicity of NIR light, as well as the small background autouorescence of biomacromolecules in the NIR region. However, as far as we know, only a few uorescent NO probes could meet this requirement. [32][33][34][35] In this work, we present an OPD-locked and Si-rhodamine deoxylactam-based uorescent NO probe, i.e. deOxy-DALSiR, which could not only overcome the above-mentioned limitations, as evidenced by its NIR excitation and emission wavelengths, negligible interference by DHA/AA/MGO, fast uorescence response within seconds, and stable uorescence output against pH changes, but also show some additional important sensing performances, including a baseline-level background uorescence, huge uorescence off-on ratio (6300-fold), and an ultra-low detection limit (0.12 nM). These features, combined with its good membrane permeability, low cytotoxicity, and high resistance to photoxidation, have made the probe an ideal indicator for the tracking of endogenous NO in vitro and in vivo.

Results and discussion
Design rationale and synthesis of deOxy-DALSiR The design of deOxy-DALSiR was inspired by OPD-locked rhodamine lactam (named DALR here), a uorescent NO probe reported in 2008 (Scheme 1A). 28 The easy synthesis, negligible background uorescence, and big uorescence offon ratio have since made the probe an attractive platform for constructing various uorescent NO probes. 34,[36][37][38][39][40] However, the N-acyltriazole intermediate, arising from the initial reaction of the probe with NO, suffers from slow hydrolysis, and as a result, the uorescence response time of the probe for NO is relatively long ($30 min). More seriously, although not detected in that report, the intermediate may suffer from a cysteine (Cys)induced native-chemical-ligation (NCL) and cyclization cascade reaction 41,42 leading to nonuorescent rhodamine lactam (Scheme 1A), given that N-acyltriazoles are highly active for the S-acylation reaction [43][44][45] and the resulting rhodamine lactam is highly stable in physiological pH. 46 In fact, the possibility has been conrmed by our experimental results (see below). Thus, the sensitivity of the probe for intracellular NO would be discounted due to the relatively high physiological concentration of Cys in living cells (30-200 mM). [47][48][49] Moreover, the visible excitation and emission wavelengths also make the probe unsuitable for in vivo imaging applications. We envisioned that the above limitations could be overcome by an OPD-locked Sirhodamine deoxylactam derivative, i.e. deOxy-DALSiR (Scheme 1B). According to our design principle, the reaction of deOxy-DALSiR with NO under aerobic conditions would bypass the Nacyltriazole intermediate to directly generate a stable N-alkyltriazole product deOxy-DALSiR-T, thereby avoiding interference from a water molecule or Cys. Not only that, but some additional important sensing advantages were also greatly expected for the probe: (1) it contains a potential NIR Si-rhodamine uorophore, 50 thus making it suitable for in vivo imaging applications; (2) it should have low background uorescence at a physiological pH due to its spirocyclic structure, and, even if it exists in its ring-opened form in poorly acidic conditions, the low background uorescence is also greatly expected due to PeT from the electron-rich OPD group to the excited Si-rhodamine uorophore; 51 (3) it should be able to resist interference by DHA/AA/MGO, because the locked or alkylated OPD group may lose its capability to condense with the three biological species to form uorescent quinoxaline or the 1H-quinoxaline-2-one heterocycle; 23-26 (4) it should respond rapidly to NO, because, unlike the OPD group of the widely used rhodamine or uorescein-based uorescent NO probes, 10,11 the OPD group of the probe does not contain any electron-withdrawing substituent, thus making it highly reactive with NO; 32 (5) it should show stable uorescence for NO in a wide pH range, because its triazole product deOxy-DALSiR-T lacks any acidic NH proton, thereby precluding the triazolate-induced partial uorescence quenching near neutral pH. 12 With these considerations in mind, we synthesized deOxy-DALSiR starting from the commercially available 3-bromo-N,Ndimethylaniline 1. As shown in Scheme 2, the reaction of 1 with n-BuLi in THF, followed by treatment with Si(CH 3 ) 2 Cl 2 , gave diaryl silyl ether 2; the treatment of 2 with 2-formylbenzoic acid in the presence of CuBr 2 provided Si-rhodamine lactone 3; the reaction of 3 with POCl 3 in 1,2-dichloroethane gave Sirhodamine chloride 4; the reaction of 4 with OPD in the presence of NEt 3 afforded the precursor DALSiR; the desired product deOxy-DALSiR was nally obtained by reducing DALSiR with BH 3 $THF solution. The detailed synthetic procedures are shown in the ESI. † The spectral response of deOxy-DALSiR to NO It was reported that the ring-opening transformation of rhodamine deoxylactams is more sensitive to pH than that of rhodamine lactams. 52,53 We envisioned that it should also be the case for deOxy-DALSiR due to the similar chemical structure. As shown in Fig. 1A, at pH ¼ 4 deOxy-DALSiR showed a strong absorption peak at 650 nm, indicating that it exists mainly as the ring-opened form in poorly acidic conditions; as the pH values increased, the absorption intensities gradually decreased, and nally reached the baseline level when pH $ 7, indicating that its ring-closed form dominates in poorly basic conditions. Thus, the chemical structure of deOxy-DALSiR is pH-dependent. Notably, whether existing as the ring-closed form or as the ring-opened form, deOxy-DALSiR showed negligible uorescence in the NIR region ( Fig. 1B), consistent with previous speculation that deOxy-DALSiR is always nonuorescent either due to its non-conjugated spirocyclic structure or the PeT quenching process (Scheme 1B). Importantly, in the pH region of 4-10, deOxy-DALSiR exhibited a dramatic uorescence off-on response for NO in the NIR region (l max ¼ 680 nm), strongly indicating that the probe has the potential to serve as an imaging tool for NO in complex biological environments.
Encouraged by the above results, we thoroughly evaluated the uorescence sensing performances of deOxy-DALSiR toward NO in the simulated physiological conditions (PBS, 50 mM, pH ¼ 7.4, containing 20% CH 3 CN). As shown in Fig. 2A, the probe itself showed no observable emission when excited at 645 nm; upon treatment with excessive NO, a huge uorescence off-on response up to 6300-fold was immediately observed at 680 nm. Remarkably, the uorescence off-on response was considerably rapid and could be completed within seconds (Fig. 2B), indicative of the great potential of the probe for realtime tracing of NO generation in biosystems. Further, the uorescence titration assay revealed a dose-dependent increase in the uorescence intensities, which reached saturation when   30 mM NO was used (Fig. 2C). In this case, an excellent linear correlation between the uorescence intensities and NO concentrations from 0 to 20 mM was achieved (Fig. 2D), and the detection limit was estimated to be as low as 0.12 nM based on 3s/k. To the best of our knowledge, this is the highest detection sensitivity for NO achieved for uorescent probes reported to date. Thus, it is very promising for the probe to detect trace amounts of intracellular NO. To establish the selectivity, we tested the uorescence responses of the probe toward various biologically relevant species, including reactive oxygen species (ROS: ClO À , H 2 O 2 , O 2 c À , 1 O 2 , cOH, NO 2 À , and ONOO À ), DHA/AA/ MGO, metal ions (K + , Ca 2+ , Na + , Mg 2+ , Al 3+ , Zn 2+ , Fe 2+ , Fe 3+ , Cu + , and Cu 2+ ), and Cys/GSH. As shown in Fig. 2E and S1 (ESI †), all of these competitive species, including DHA/AA/MGO, did not elicit any obvious uorescence enhancement of the probe, suggesting that the probe possesses fairly high specicity for NO. Taken together, these results reveal that deOxy-DALSiR should outperform many of the previously reported uorescent NO probes in terms of the greater uorescence off-on ratio, faster uorescence response rate, higher selectivity, and ultrasensitivity.
To conrm the reaction mechanism, we carried out HPLC-MS assays on deOxy-DALSiR aer treatment with NO. As shown in Fig. S2 (ESI †), the treatment of the probe with NO mainly produced a new product, which could be assigned to Nalkyltrizole deOxy-DALSiR-T based on the MS data (m/z: calcd 516.2578; found 516.2582). Further, we synthesized deOxy-DALSiR-T by treating deOxy-DALSiR with NaNO 2 in a mixture of AcOH and methanol (ESI †), and found that its absorption and emission maxima and proles are identical to those of deOxy-DALSiR treated with NO (Fig. S3, ESI †). Obviously, the results are in good agreement with our proposed reaction mechanism (Scheme 1B).
With deOxy-DALSiR-T in hand, we also tested the pH effects on its uorescence intensities, given that the widely used OPDbased uorescent NO probes commonly suffer from triazolateinduced partial uorescence quenching via PeT near neutral pH. [9][10][11][12] As shown in Fig. S4 (ESI †), due to the absence of any acidic NH proton in its N-alkyltriazole moiety, deOxy-DALSiR-T avoided the unfavorable deprotonated reaction, and thus displayed stable uorescence in a wide pH range of 4-10.
Comparison of the uorescence sensing performances between deOxy-DALSiR, DALSiR, and DALR It should be mentioned that DALR is a rhodamine lactam-based uorescent NO probe that was reported in 2008 (Scheme 1), 28 and DALSiR is a Si-rhodamine counterpart of DALR and also a synthetic precursor of deOxy-DALSiR (Scheme 2). To further illustrate the advantage of our design strategy, we compared the uorescence sensing performances of deOxy-DALSiR, DALSiR, and DALR for NO in the presence of Cys, given that the Nacyltriazole intermediates resulting from the initial reactions of the latter two with NO may suffer from a Cys-induced NCL and cyclization cascade reaction to generate the nonuorescent lactam products (Scheme 1A). As shown in Fig. 3A and S5 (ESI †), the addition of NO to the mixture of deOxy-DALSiR and Cys resulted in a nearly identical uorescence off-on response to that in the absence of Cys, indicating that Cys hardly interfered with the reaction of deOxy-DALSiR with NO; by sharp contrast, almost no uorescence response was observed when NO was added to the mixture of DALSiR (or DALR) and Cys, strongly indicating that the produced N-acyltriazole intermediate could be quickly consumed by Cys. Further, we performed a control assay by the initial treatment of the three compounds with NO, and then with Cys. As shown in Fig. 3B and S6 (ESI †), the uorescence of deOxy-DALSiR treated with NO was not affected by the subsequent addition of Cys, whereas the uorescence of DALSiR (or DALR) treated with NO dramatically decreased with the subsequent addition of Cys. The results further support the above speculation. Notably, the incomplete inhibition of the uorescence in the cases of DALSiR or DALR (Fig. 3B) is presumably due to the lower amount of (Si-)rhodamine dye produced via the hydrolysis of the N-acyltriazole intermediate before Cys attack (Scheme 1A). In addition, the absorption spectra studies also support the above speculation. As shown in Fig. S7 (ESI †), deOxy-DALSiR, DALSiR, and DALR all showed negligible absorbance in the visible and NIR regions, due to their spirocyclic structures; upon the addition of NO, a strong absorption band was immediately observed for all of the three compounds, indicating that NO could trigger the ring-opening of the three compounds; aer the subsequent treatment with Cys, the absorption intensity for the former remained almost unchanged, but for the latter two it decreased gradually within ten minutes, further supporting that the N-acyltriazole intermediate, resulting from the reaction of DALSiR (or DALR) with NO, could be captured by Cys. To conrm the reaction, we performed HPLC-MS assay on DALR pretreated with NO and then treated with Cys. As shown in Fig. S8 (ESI †), in addition to the lower amounts of unreacted DALR, a main new product was observed, which was proved to be the Cys-containing rhodamine lactam in terms of the MS data (calcd: 546.2427; found: 546.2423). Taken together, these results strongly suggest that the sensitivity of deOxy-DALSiR for intracellular NO should be superior to that of DALSiR or DALR, given the relatively high physiological concentration of Cys (30-200 mM) in mammal cells. [47][48][49] In addition, we also tested the effects of some other biological reductants, such as GSH, Hcy, ascorbate (Asc), dithiothreitol (DTT), and NaHSO 3 , on the uorescence response of deOxy-DALSiR toward NO. As shown in Fig. S9 (ESI †), the presence of these reductants virtually did not interfere with the uorescence response of deOxy-DALSiR toward NO, consistent with the case of Cys.

Imaging exogenous and endogenous NO in living cells using deOxy-DALSiR
Before the imaging assays, the cytotoxicity of deOxy-DALSiR and its alkyltrizole product deOxy-DALSiR-T was tested in HeLa cells by MTT assay. As shown in Fig. S10 (ESI †), aer 24 h of cellular internalization of 2-50 mM deOxy-DALSiR (or deOxy-DALSiR-T), >90% of the cells remained viable, indicative of the excellent biocompatibility of both the probe and its alkyltrizole product. Even so, in order to reduce the interference to cell proliferation and physiology, a low concentration of deOxy-DALSiR (2 mM) was used in the subsequent bioimaging assays. Subsequently, the specicity of deOxy-DALSiR for NO in living cells was evaluated. As shown in Fig. 4A, when the HeLa cells were treated with deOxy-DALSiR, they showed almost no uorescence; when the deOxy-DALSiR-loaded HeLa cells were treated with NOC-9 (a commercially available NO donor), strong intracellular uorescence was observed in the red channel. Thus, the probe is cell membrane-permeable and could image exogenous NO in the cellular environment. Further, when the deOxy-DALSiRloaded HeLa cells were treated with several representative ROS, such as H 2 O 2 , ClO À , and SIN-1 (a commercially available ONOO À donor), respectively, almost no uorescence was observed for each case, indicating that the probe still possesses high specicity for NO in cellular environments. Notably, the continuous irradiation of the deOxy-DALSiR-loaded HeLa cells in the absence and presence of NOC-9 in the imaging conditions for 60 min neither elicited any obvious uorescence enhancement for the former, nor resulted in any obvious uorescence decrease for the latter (Fig. 4B, details in Fig. S11, and ESI Videos S1 and S2 †). Thus, deOxy-DALSiR and its trizole product could tolerate photoxidation and photobleaching, respectively, thus being suitable for time-lapse and long-term bioimaging applications.
Encouraged by the above results, we explored the potential applications of deOxy-DALSiR for the imaging of endogenous NO in living cells. The assays were rst performed in RAW264.7 macrophages because these cells are known to express highlevel iNOS upon stimulation by LPS/INF-g. 54,55 As shown in Fig. 5A, the cells themselves were nonuorescent; when the cells were incubated with deOxy-DALSiR, a weak but clear intracellular uorescence was observed in the red channel; when the cells were pretreated with the NO synthase inhibitor aminoguanidine (AG) 56 and then treated with deOxy-DALSiR, the intracellular uorescence greatly decreased; when the cells were stimulated with LPS/INF-g and then treated with deOxy-DALSiR, a dramatic uorescence enhancement in the red channel was observed; when the cells were stimulated with LPS/INF-g in the presence of AG and then treated with deOxy-DALSiR, only weak  intracellular uorescence was observed. Thus, deOxy-DALSiR could image not only the basal NO but also the stimulationinduced NO in RAW264.7 cells, consistent with its high sensitivity found in chemical conditions. Further, the probe was utilized to monitor endogenous NO in pancreatic b-cells (INS-1), given that the excessive and sustained generation of NO derived from iNOS plays an important role in pancreatic b-cell death and the pathophysiological progression of diabetes. 57,58 In the assays, streptozotocin (STZ), a widely used chemical to trigger pancreatic b-cell damage and induce experimental diabetes by the production of ROS and NO, 57 was used as the inducer. As shown in Fig. 5B, similar to the case in the RAW264.7 macrophages, the probe could image both the basal and STZ-induced NO in pancreatic b-cells; moreover, the STZ-induced NO generation in pancreatic b-cells is dose-dependent and temporally regulated (Fig. S12, ESI †). Thus, deOxy-DALSiR holds great potential for studying diabetes pathogenesis.
It is known that ischemia-reperfusion leads to increased iNOS-mediated NO production and the subsequent formation of ONOO À via the diffusion-controlled reaction of NO and O 2 c À . 59 ONOO À is a powerful oxidant and its overproduction during ischemia could cause severe damage to endothelial cells. [60][61][62] Thus, we also performed the assay of visualizing NO production using deOxy-DALSiR in endothelial EA.hy926 cells aer oxygen-glucose deprivation (OGD), a widely used in vitro ischemic model. [63][64][65] As shown in Fig. 6, a time-dependent uorescence enhancement was observed in the endothelial cells over 0.5 to 2 h following OGD exposure, revealing that the probe is competent enough to monitor NO uxes during ischemia.
To probe the subcellular localization of deOxy-DALSiR, we performed costaining assays in HeLa cells. In the assays, NOC-9 was used to light up the probe in the cells, and Pearson's correlation coefficient (R) was used to analyze the scope of distribution between the two uorescent channels from the probe and commercial trackers. As shown in Fig. 7A, when the cells were costained with deOxy-DALSiR and a commercial MitoTracker followed by NOC-9 treatment, a poor overlapping image and low Pearson's correlation coefficient (R ¼ 0.20) were observed. By contrast, when the cells were costained with deOxy-DALSiR and commercial LysoTracker followed by NOC-9 treatment, a good overlapping image and high Pearson's correlation coefficient (R ¼ 0.82) were found (Fig. 7B), indicating that the probe was mainly located in the lysosomes. A feasible explanation is that deOxy-DASiR is easy to protonate in poorly acidic conditions (Fig. 1A), and thus could be trapped by the poorly acidic lysosomes (pH range: 4.5-5.5), consistent with the proton-driving lysosome localization of the alkylmorpholine-containing lysosomal probes. 66 The result is interesting because NO has been reported to play important roles in lysosome-related disorders and diseases, including lysosomal storage disorders, 67 Gaucher's disease, 68 and Danon disease, 69 and there are few uorescent NO probes that can target lysosomes so far. 17,35 In vivo imaging of NO generation in mouse models using deOxy-DALSiR Given that deOxy-DALSiR could operate in the NIR region, we further investigated its potential to image endogenous NO in living nude mice. The images were obtained using a Bruker In-Vivo FX Pro small animal optical imaging system with an excitation lter of 620 nm and an emission lter of 670 nm. As can be seen in Fig. 8A, when the mouse was intraperitoneally (i.p.) injected with deOxy-DALSiR for 30 min, almost no uorescence Fig. 6 The time-dependent fluorescence accumulation of deOxy-DALSiR (2 mM) in EA.hy926 endothelial cells over 0.5 to 2 h following OGD exposure. The images were obtained using DeltaVision Microscopy Imaging Systems, and the excitation and emission bandpasses of the standard Cy5 filter set were used. The scale bar: 50 mm.  signal was observed; when the mouse was i.p. injected with LPS for 12 h to induce inammation, followed by i.p. injection of deOxy-DALSiR, an obvious uorescence signal was observed in the abdominal region (Fig. 8B), indicating that deOxy-DALSiR could image endogenous NO in the inamed mouse model. Further, we tested the in vivo sensing performance of deOxy-DALSiR for NO in the STZ-treated mouse model. As shown in Fig. 8C, when the mouse was i.p. injected with STZ for 12 h, followed by i.p. injection of deOxy-DALSiR, a clear uorescence signal was also observed in the abdominal region, consistent with iNOS being highly expressed in the initial stage of diabetes. 57,58 Thus, deOxy-DALSiR holds great potential for studying the pathological roles of NO in living animals.

Conclusions
In summary, we reported an OPD-locked, Si-rhodamine deoxylactam-based near-infrared uorescent NO probe deOxy-DALSiR. The probe not only overcame the limitations suffered by most of the previously reported OPD-based uorescent NO probes, such as possible interference by DHA/AA/MGO, slow response rate, pH-sensitive uorescence output, and short excitation and emission wavelengths, but also avoided the severe interference from Cys suffered by the OPD-locked and rhodamine lactam-based uorescent NO probes developed later. More importantly, the probe could detect NO with a rapid response rate, huge uorescence off-on ratio, and ultra-low detection limit. These excellent sensing performances, coupled with the good cell permeability and low cytotoxicity, have enabled the probe to image endogenous NO not only in RAW 264.7 macrophages, pancreatic b-cells, and endothelial EA.hy926 cells, but also in living mouse models. The probe is greatly expected to be a useful imaging tool for studying NOrelated physiological and pathological functions.