Albumin-seeking NIR dyes covalently combine with albumin and increase brightness

We proposed a chemoselective albumin-seeking strategy for NIR dyes binding to albumin in vivo. Previous studies have proved the covalent bonding between chlorine-containing (Cl-containing) cyanine dyes and albumin [27]. IR-780, IR-808 and IR-783 were cyanine dyes possessing chlorocyclohexene that could first enter the albumin pocket through supramolecular interactions and underwent covalent binding via a nucleophilic substitution reaction between the -SH group in albumin and the Cl-C bond in these dyes [27, 28]. While the Cl-free dye (ICG) bound albumin only through non-covalent supramolecular interactions [29]. Covalent binding of Cl-containing cyanine dyes and bovine serum albumin (BSA) enhances the stability and brightness of free cyanine dyes (Fig. 1a) [13, 30, 31]. We selected the Cl-containing NIR cyanine dyes that can covalently bind to albumin with optimal albumin-binding abilities of IR-780 > IR-808 > IR-783, and ICG as a comparison which not covalently bind to albumin (Fig. 1b) [27]. First, we compared the brightness characteristics of the four probes in vitro binding to albumin. The dyes exhibited the highest and lowest brightness in DMSO and PBS, respectively. Dyes in BSA were incubated at a 1:1 molar ratio at 60 °C for 2 h. The absorption and fluorescence spectra showed that the free dyes and Dye@BSA in PBS exhibited absorption peaks at 779–799 nm with NIR-II tail emission (Fig. S1). IR-780, IR-808, and IR-783 were associated with a higher brightness forming Dye@BSA than free dyes in PBS (Fig. 1c). In addition, compared with dyes in PBS, Dye@BSA has higher NIR-II quantum yield (Fig. S2, S3, Table S1). Next, the binding proportions of the four dyes to BSA were analyzed. Mass spectrometry illustrated that IR-780, IR-808, IR-783 bound to BSA in a 1:1 ratio (Fig. 1d). The bio-layer interferometry (BLI) experiments quantitatively compared binding affinity of IR-780 (KD = 2.80E + 04 nM) > IR-808 (KD = 9.50E + 04 nM) > IR-783 (KD = 4.30E + 05 nM) > ICG (KD =8.10E + 05 nM) (Fig. S4, Table S2). Gel electrophoresis showed that IR-780 bound almost completely to BSA, IR-808 bound mostly to BSA with a small fraction of free IR-808, IR-783 bound only partially to BSA, and ICG did not form any stable complex with BSA (Fig. 1e). As albumin is the most abundant protein in the blood, and previous studies have shown that increasing the BSA: dye ratio enhances the binding proportion [27, 32, 33]. Therefore, we further incubated BSA with dyes at a ratio of 10:1. The gel electrophoresis revealed an enhanced binding of Cl-containing cyanine dyes to BSA, with IR-780 and IR-808 binding almost completely to BSA and IR-783 exhibiting an increased binding to BSA, whereas ICG binding proportion remained unchanged (Fig. 1f).

Fig. 1

Chlorine-containing albumin-seeking dyes covalently bind to albumin and increase brightness. (a) Schematic diagram of the covalent binding site of chlorine-containing cyanine dye and albumin. (b) Structural formula of IR-780, IR-808, IR-783, and ICG. (c) Brightness of four dyes (50 nM) in DMSO, PBS, and BSA with incubating at 60 ºC of 1:1 molar ratio (mean ± SEM, n = 3). (d) Mass spectrometric analysis of four dyes in BSA with 1:1 reaction molar ratio at 37 ºC for 2 h. (e-f) Gel electrophoretic analysis of fluorescence intensity of four dyes in BSA at 1:1 and 10:1 reaction molar ratio at 37 ºC for 2 h (mean ± SEM, n = 3)

Albumin-seeking dyes highly affinity and selectively bind to albumin in situ with enhanced brightness and circulation time

Before applying albumin-seeking dyes to stroke imaging, we observed the systemic metabolism and circulation of the four dyes in mice. After intravenous injection, IR-780 and IR-808 accumulated in the liver and then distributed throughout the whole body maintaining high brightness for approximately 72 h with a long metabolism time. In contrast, IR-783 and ICG accumulated in the liver and cleared rapidly from the body within 24 h (Fig. S5).

To verify the metabolism of dyes in the circulation and the formation of dye@albumin complexes in vivo, blood samples were collected from mice intravenously injected with the same dose of the four dyes. The serum was separated and subjected to gel electrophoresis, followed by NIR fluorescence imaging and gel staining with Coomassie Brilliant Blue. IR-780 and IR-808 maintained higher brightness in the circulation for a longer period compared to IR-783 and ICG, which gradually decreased within 7 days after the tail venous injection of the probes (Fig. 2a, b). In contrast to the results of brightness in vitro binding to albumin, IR-808 exhibited the highest brightness in circulation, followed by IR-780. It may be attributed to the hydrophobicity of IR-780, resulting in self-assembly and lower brightness when injected into the mice than IR-808 [34]. The serum gel electrophoresis results showed that IR-780, IR-808, and IR-783 formed dye@albumin complexes of approximately 66 kDa in situ, whereas ICG remained in its free form and was rapidly metabolized (Fig. 2c). Figure 2d shows the band patterns of various proteins in the serum collected at different time points. These results indicated the high selectivity and stability of IR-780, IR-808, and IR-783 binding to albumin. Consistent with the serum fluorescence brightness results, the IR-808@albumin band was the brightest and the IR-780@albumin band was the second brightest. Only a portion of IR-783 formed complexes with albumin in the bloodstream, whereas the unbound portion of the dye was quickly metabolized and excreted. ICG did not bind to endogenous albumin and exhibited the fastest metabolism, reaching its highest brightness after 1 min, which decreased rapidly 10 min post-administration (Fig. 2e). These results illustrate that albumin-seeking probes can selectively and stably bind albumin in vivo with increasing brightness and circulation time.

Fig. 2

Comparison of circulation metabolic behavior of albumin-seeking dyes in vivo. (a) Serum was collected at time points after tail vein injection (50 µM, 60 µL) of IR-780, IR-808, IR-783 and ICG and imaged under the NIR-II window (80 ms, 1100 LP). (b) Normalized fluorescence intensity of serum after injection of dyes (mean ± SEM, n = 3 for each group). (c) Gel electrophoresis analysis of dyes and albumin binding changed over time in mice serum. (d) Coomassie brilliant blue showed protein distribution of serum after gel electrophoresis. (e) Normalized fluorescence intensity of albumin complex band of IR-780@albumin, IR-808@albumin, and IR-783@albumin and free ICG (mean ± SEM, n = 3 for each group)

Albumin-seeking dyes monitoring BBB disruption in stroke mice in real-time

Albumin is normally scarce in the brain tissue. However, blood albumin enters the brain through the damaged BBB post-stroke [35, 36]. Therefore, the albumin in brain serves as a biomarker for BBB injury [25, 26]. Although the dye-albumin complex enhanced the brightness and circulation time of free dye, introducing exogenous albumin may trigger immune or inflammatory reactions [37]. Therefore, the probe that can bind to endogenous albumin in situ may offer safe imaging and good capabilities for monitoring BBB disruption.

We explored albumin-seeking dyes for visualization of BBB disruption after stroke in a photothrombotic stroke (PTS) model with infarcts in the left cerebral cortex of mice. At 30 min post stroke, IR-780, IR-808, IR-783, and ICG were intravenously injected at the same dose respectively, and real-time dynamic NIR-II imaging was performed without the need for craniotomy (Fig. 3a). As shown in Fig. 3b, cerebral vessels were visualized but no fluorophore leakage into the stroke region at 1 min after probes injection. Fluorescence signals gradually accumulated in the stroke area in the IR-780 and IR-808 groups with leakage observed as early as 10 min, indicating the accumulation in the brain parenchyma through the damaged BBB. IR-808 exhibited a brighter imaging than IR-780, consistent with the relative brightness of the serum. IR-783 have weaker brightness than IR-780 and IR-808, and showed a low fluorescence signal at 1 min. Due to the relatively weak binding affinity of IR-783 with albumin, free IR-783 cleared rapidly from the body [27]. Therefore, unbound IR-783 was rapidly metabolized and thus showed lower fluorescence signals in cerebral vascular at 10 min / 30 min / 1 h than those at 1 min. After a long period of injection into the blood circulation, IR-783 could bind to albumin and greatly enhance its fluorescence intensity, thus showing a higher fluorescence signal at 3 h than that at 10 min / 30 min / 1 h and accumulation of fluorescence signals in the BBB leakage region was observed but with lower brightness than IR-780 and IR-808. ICG showed higher brightness in the blood vessels 1 min after injection but was rapidly metabolized within 10 min, with minimal signal accumulation in the stroke area. To minimize the influence of brightness in the blood vessels, we quantified the fluorescence signal in the BBB disruption area by calculating the difference between the signal in the BBB disruption area and corresponding signal in the healthy contralateral area (Fig. 3c). IR-780, IR-808, and IR-783 showed the largest difference in the accumulation of fluorescence signals at 6 h, whereas ICG exhibited minimal signal accumulation in the brain tissue, which gradually decreased after peaking at 1 h.

In addition, we validated the imaging ability of the BBB using the dye@BSA complex synthesized in vitro (Fig. S6a). IR-780@BSA, IR-808@BSA, and IR-783@BSA complexes exhibited similar brightness and circulation times in monitoring BBB disruption. ICG@BSA exhibited a behavior similar to free ICG. The brightness of the BBB disruption area was observed to be IR-780@BSA > IR-808@BSA > IR-783@BSA (Fig. 3d). Although IR-780 demonstrated the strongest binding ability to albumin in vitro, it did not achieve similar effects as IR-780@BSA complex in vivo, possibly because of its hydrophobicity leading to self-assembly, and insufficient binding in vivo [38]. In contrast, IR-808 has similar effects as the IR-808@BSA complex, indicating that IR-808 can serve as an ideal albumin-seeking probe for monitoring BBB disruption after stroke without the need for exogenous albumin, thus offering better biocompatibility. To confirm that BBB imaging using IR-808 relies on its in vivo binding to albumin, we used an albumin-escaping probe, IR-808AC [39], which clearly visualized blood vessels but did not accumulate nor show BBB disruption and was completely metabolized from the blood vessels within 10 min (Fig. S6b).

Fig. 3

Albumin-seeking dyes real-time monitoring BBB disruption in mice with stroke. (a) Schematic diagram of photothrombotic stroke model and NIR-II imaging for BBB disruption monitoring post-stroke (Created with BioRender.com). (b) NIR-II dynamic imaging (40 ms, 1100 LP) recorded at indicated time points post-injection of stroke mice with IR-780, IR-808, IR-783, and ICG (50 µM, 60 µL). (c-d) Difference in fluorescence signal intensity on the left and right sides post-injection of free dyes and dye@BSA (mean ± SEM, n = 3 for each group)

The albumin-seeking dyes allows for immediate assessment, multi-timepoint monitoring after stroke and a real-time, high-resolution visualization of BBB disruption. Among the four dyes tested, IR-808 exhibited the most remarkable visualization of BBB disruption. Therefore, IR-808 is a promising probe for accurate assessment of stroke injury and we conducted detailed imaging using IR-808 in our subsequent investigations.

IR-808 enables evaluation of BBB injury in strokes with various severities

We first displayed the imaging of IR-808 monitoring BBB leakage in NIR sub-windows. Fig. S7a presents the imaging results of IR-808 in NIR-I window (850–900 nm) and three sub-NIR-II windows (> 1100 nm, > 1200 nm, and > 1300 nm). IR-808 could provide sufficient brightness under the longer NIR-II sub-window by increasing the exposure time and improves signal-to-background ratio (SBR) compared with in NIR-I window (Fig. S7b). We next observed changes in cerebral perfusion and IR-808 fluorescence signal accumulation in acute-phase of stroke. Dynamic continuous imaging of cerebral arteries and veins in the mice was observed 2.8 s post IR-808 injection. Compared with healthy side, a reduction in cerebral perfusion was observed in the infarct area of stroke mice (Fig. S8a, b). Leakage of fluorescent signal was observed in the stroke area 5 min after injection of IR-808, and fluorescent signal accumulation increased within 1 h (Fig. S8c, d).

We further explored the efficiency of IR-808 to detect and distinguish BBB injury of sham mice and three severity levels of stroke. As described in the methods, the severity of stroke injury was graded into three groups using the PTS model [40]. TTC results showed a progressive increasing infarct volume at 24 h after stroke in mild, moderate and severe stroke groups (Fig. 4a). Western blotting analysis of protein expression in brain tissue showed a gradual decrease in tight junction-related proteins (ZO-1, occludin) and basement membrane-related protein (laminin) and an increase in matrix metalloproteinase (MMP-9) levels in the sham, mild, moderate, and severe stroke groups (Fig. 4b), indicating an increased degree of BBB disruption. NIR-II imaging shows the fluorescence signals were similar in both sides and did not leak through the intact BBB in the sham group. In stroke groups, fluorophores leakage initially in the infarct core and gradually accumulates in the peri-infarct tissue as stroke time progressing, indicating dynamic aggregation through leakage BBB into brain tissue (Fig. 4c). Relative fluorescence intensity of the affected side increased progressively after stroke. In the moderate to severe stroke, the relative fluorescence intensity peaked at 12 h, whereas in the mild stroke it increased until the 24-h peak (Fig. 4d). The BBB leakage area increased over time in stroke groups and was larger in the worse stroke groups (Fig. 4e). In addition, we analyzed statistical differences in BBB disruption between stroke severity at time points. The relative fluorescence intensity was significantly higher in the more severity stroke groups at 6 h (P < 0.01) and 12 h (P < 0.05) (Fig. S9a). The BBB damage area was significantly different at 6 h between the three stroke groups (P < 0.01) and was significantly lower in mild stroke than in moderate-to-severe strokes at 12 h (P < 0.01) (Fig. S9b). At 24 h, relative fluorescence intensity and damage area were significantly higher in severe stroke than mild stroke (P < 0.05). These results indicate that IR-808 allows dynamic assessment of strokes, as well as good differences in leakage area and relative fluorescence intensity measured by IR-808 in various severity strokes.

Fig. 4

IR-808 dynamic monitoring of BBB disruption with varying severity. (a) Coronal slices and whole-brain TTC staining in sham, mild, moderate, severe stroke groups and statistical analysis of infarct volumes quantified in slices (mean ± SEM, n = 6 for each group). (b) Western blotting analysis of BBB injury protein expression in the brains of mice with stroke with incremental severity (mean ± SEM, n = 3 for each group). (c) IR-808 NIR-II imaging for in vivo multi-timepoint monitoring of BBB disruption in strokes of different severity (40 ms, 1100 LP). (d) Quantification of the relative FL intensity (Ratio of fluorescence intensity of affected side/healthy side-1) at time post-injection (n = 6 for each group). (e) Quantification of the area of fluorescence signal accumulation in the BBB damage area at time post-injection (n = 6 for each group). *P < 0.05, ****P < 0.0001

BBB disruption evaluated by IR-808 correlates well with Evans blue measurement

Currently, BBB disruption in experimental animals is mainly assessed through intravenous injection of Evans blue (EB), followed by ex vivo imaging and quantification [41]. To verify the reliability of the IR-808 in vivo evaluation of BBB injury, we performed in terms of both IR-808 NIR-II imaging in vivo and EB ex vivo measurement and quantification in sections at 6 h, 12 h, and 24 h, respectively. IR-808 leakage area and fluorescence intensity as well as EB leakage area and quantification were analyzed in the same mice (Fig. 5a).

Figure 5b present IR-808 in vivo and EB ex vivo imaging of mild, moderate, and severe strokes at 6 h, 12 h, and 24 h post-stroke. The relative fluorescence intensity of IR-808 correlated well with EB quantification at 6 h (r = 0.975, P < 0.001), 12 h (r = 0.964, P < 0.001), and 24 h (r = 0.851, P < 0.001) (Fig. 5c). This indicates that the quantitative evaluation of BBB injury by using IR-808 in vivo imaging is of high reliability. We further analyzed the findings in the brain sections (Fig. 5d). EB leakage and accumulation of IR-808 fluorescence signal can be quantified in sections (Figs. S10, S11). The fluorescence intensity quantification of IR-808 counted in sections correlated well with the relative fluorescence intensity in vivo (6 h: r = 0.921; 12 h: r = 0.857; 24 h: r = 0.879, all P < 0.001), demonstrating that in vivo NIR-II imaging has a good accuracy in reflecting IR-808 leakage into brain tissue (Fig. S12a). And the NIR fluorescence intensity in the sections correlated with EB quantification validating the results of in vivo imaging (Fig. S12b). Furthermore, the size ratio of BBB disruption in whole brain measured by IR-808 to EB at 6 h, 12 h and 24 h was 1.045 ± 0.039, 1.005 ± 0.047, and 1.396 ± 0.063, respectively, indicating that the IR-808 is reliable in assessing the area of leakage (Fig. S12c, d). Comparing multiple time points in vivo and ex vivo with EB, our results demonstrate the accuracy of IR-808 NIR-II imaging for repeat assessment of BBB injury in vivo.

Fig. 5

Correlation of IR-808 with Evans blue evaluation of BBB injury. (a) Schematic diagram showing the quantification of IR-808 and Evans blue (EB) on BBB disruption evaluation in the same mice (Created with BioRender.com). (b) IR-808 in vivo NIR-II imaging and its ex vivo visible EB imaging in mice with mild, moderate, and severe stroke at 6 h, 12 h, and 24 h. Orange dotted line outlines BBB leakage areas. (c) Correlation of relative FL intensity of IR-808 with Evans blue quantification at 6 h, 12 h, and 24 h (n = 12 each time point). (d) The corresponding images of brain sections under NIR and visible scanning in panel b

In addition, to explore the correlation between BBB leakage assessed by IR-808 and final infarct volume, we also assessed the correlation at 6 h, 12 h and 24 h of dynamic monitoring in various severities stroke. The relative fluorescence intensity at 6 h (r = 0.903, P < 0.001), 12 h (r = 0.825, P < 0.001), and 24 h (r = 0.659, P = 0.003) as well as the area of BBB leakage at 6 h (r = 0.880, P < 0.001), 12 h (r = 0.785, P < 0.001), and 24 h (r = 0.779, P < 0.001), correlated with the final infarct volume (Fig. S13). These results suggest that monitoring BBB disruption by IR-808 allows for early prediction of the final infarct volume during dynamic monitoring.

Dual-channel NIR-II dynamic imaging simultaneously evaluates BBB injury and cerebral perfusion

Brain neuropathology changes complex after stroke, and changes in BBB damage often accompanied with dynamic changes in cerebral perfusion. Thus, dual-channel NIR-II imaging for simultaneous evaluation of BBB injury and cerebral perfusion has a high clinical applicability. IR-808 under 808 nm excitation showed the area of BBB disruption. To assess cerebral perfusion while monitoring the BBB, we employed another NIR-II probe, CO-1080@BSA, under 980 nm excitation and was intravenously injected at 6 h after stroke (Fig. 6a, b) [42]. Dynamic cerebrovascular images were captured beginning with injection of CO-1080@BSA. NIR-II images showed BBB leakage (808 nm, IR-808) and dynamic changes of the cerebrovascular arterial to venous phase (980 nm, CO-1080@BSA) in the two NIR-II imaging channels (Fig. 6c). We quantified the dynamic fluorescence signals of the cerebral blood vessels in the BBB leakage area and corresponding healthy side. Compared with that in the healthy side, the cerebral perfusion was significantly reduced in the BBB leakage area (Fig. 6d). We performed principal component analysis (PCA) on a series of 40-frame dynamic images to distinguish the arteries (red) and veins (blue) in the mouse brain vasculature and merged them with BBB leakage (green) (Fig. 6e). PCA revealed reduced arteries in the core of the ischemic area, while arterial vessels were still present in the surrounding BBB injury area. The real-time dual-channel NIR-II imaging system can observe changes in cerebral perfusion dynamics while evaluating BBB injury, which helps to insight the mechanism of neurological injury after stroke, and has good potential for in vivo and clinical application.

Fig. 6

Dual-channel NIR-II imaging of BBB disruption and cerebral perfusion after stroke. (a) Schematic diagram of dual-channel for BBB and cerebral perfusion imaging (Created with BioRender.com). (b) The absorption spectra of IR-808 and CO-1080@BSA. (c) Dual-channel images of the BBB and cerebral vascular (100 ms, 1200 LP) after stroke at time post CO-1080@BSA injection. IR-808 was used to evaluate BBB disruption using 808 nm laser excitation, while CO-1080@BSA was used to dynamically evaluate the cerebral vascular using 980 nm excitation at 6 h post stroke. (d) Fluorescence intensity (normalized to the right-side peak) versus time in the left (infarct side) and right (contralateral) sides of the brain of mice with stroke showed a significant reduction in perfusion in the stroke area. (e) Dynamic sequential images of vessels with CO-1080@BSA were analyzed by principal component analysis and merged with images of the BBB leakage area (green) to resolve arterial (red) and venous (blue) cerebral vessels

NIR-II BBB imaging provides real-time assessment of thrombolytic therapy efficacy and favorable biosafety

Thrombolytic therapy is a crucial reperfusion therapy in acute ischemic stroke, but reperfusion may aggravate BBB disruption [43, 44]. Therefore, it is necessary to evaluate the BBB and cerebrovascular perfusion in real time. We performed NIR-II BBB imaging in a mouse thrombolytic therapy model, achieving simultaneous dynamic assessment of cerebrovascular perfusion and BBB injury before and after thrombolysis (Fig. S14a). After stroke, middle cerebral artery (MCA) occlusion and reduced perfusion on the affected side were observed using IR-808 at 808 nm channel (Fig. S14b). After thrombolysis, MCA revascularization and perfusion recovery were observed at 980 nm channel using CO-1080@BSA (Fig. S14c). Meanwhile, BBB injury can be evaluated before and after thrombolysis in real-time, with the potential to guide treatment (Fig. S14d-f). To further verify the universality of our method, we successfully performed dynamic imaging of BBB disruption in a mouse permanent middle cerebral artery occlusion model (Fig. S15) and a rat PTS model (Fig. S16). These results demonstrate the potential wide applicability of albumin-seeking probe combined with NIR-II imaging for stroke BBB evaluation.

Considering its potential clinical applications, we tested complete metabolism in stroke brain tissue, effect on BBB recovery after stroke, and systemic safety. Fluorescence signals accumulating in stroke brain tissue were completely metabolized 2 weeks after IR-808 injection (Fig. S17a-b). At 14 d, the effects of the retention of IR-808 on brain tissue function were evaluated by behavioral tests and H&E staining. Behavioral tests results showed no significant difference in unilateral sensorimotor dysfunction and cognitive function between IR-808 group and PBS group (Fig. S17c). H&E staining showed the histopathological changes of peri-infarct tissue at 14 d after stroke was not significantly different between IR-808 group and PBS group (Fig. S17d). These results indicate that IR-808 retention did not impair brain tissue function. In addition, BBB leakage was measured by EB in stroke mice injected with IR-808 or PBS at 7 d, 14 d, and 21 d after stroke, respectively. The results showed no significant difference in EB leakage between the IR-808 and PBS groups, indicating that the use of IR-808 has no impact on BBB repair after stroke (Fig. S17e). In addition, serum from mice co-injected with IR-808 and CO-1080@BSA showed blood clearance under 808 nm and 980 nm excitation, respectively (Fig. S18a). Liver function, kidney function, and blood routine indices were within the normal range in mice treated with IR-808 or IR-808 plus CO-1080@BSA (Fig. S18b). H&E staining showed no significant histological damage to the heart, liver, spleen, lungs, kidneys, or brain compared to the PBS group (Fig. S18c). These findings suggest that IR-808 and the co-injection of IR-808 plus CO-1080@BSA was safe and promising for clinical translation to in vivo NIR-II imaging of stroke.