Introduction

Saccharide (sugar) recognition using synthetic receptors or chemosensors in aqueous media has been extensively investigated because the process is very important for a range of biomedical applications [1,2,3,4,5,6,7,8,9,10,11,12,13]. Nevertheless, further practical applications in this field are currently hampered by numerous drawbacks. Methods to monitor biologically important oligosaccharides as tumor markers in real-time are highly desirable [14, 15]. Unfortunately, methods that sense oligosaccharides in a selective and sensitive manner are currently inefficient because the target oligosaccharides in the blood exhibit conformational complexity, extensive solvation (hydration), and extremely low concentration [6, 12, 16].

To date, two strategies have been successfully used for saccharide sensing. Boronic acid-incorporated chemosensors capture mono- and disaccharides through dynamic boronate formation [17,18,19,20,21,22,23,24,25]. Second, water-soluble supramolecular cages, or artificial lectins, are functional examples in which oligosaccharides are placed in the cavity through hydrophobic effects and multiple CH-π interactions [26,27,28,29,30,31,32,33,34,35,36,37,38]. However, neither method performs satisfactorily in real-life applications. The inherent problem of the supramolecular approach is its lock-and-key or “rigid” mechanism; the cage cavity must be expanded in a step-by-step manner to fit the target oligosaccharide closely, which is unrealistic. An attractive alternative for creating “smarter” chemosensors with oligosaccharide-sensing control is an induced-fit approach, which is flexible and dynamic. Nature adopts induced-fit mechanisms for sugar chain recognition. For example, lectins utilize the cooperativity of highly ordered hydrogen-bonding networks and CH-π interactions [39], which can serve as a blueprint for designing smart dynamic chemosensor systems.

In 2010, we discovered that curdlan (Cur) functions as a selective and sensitive oligosaccharide chemosensor through interactions within the dynamic hydrogen-bonding networks of the Cur polymer backbone [40]. Cur (Fig. 1a, top) is the simplest glucan, as it possesses a linear backbone and consists of β-(1,3)-linked D-glucose units. The most intriguing characteristic of native glucans is their reversible denaturing/renaturing behavior, which leads to the formation of random coils in DMSO and triplexes in aqueous solutions [41,42,43]. Therefore, we focused on the reversibility of Cur, hypothesizing that oligosaccharides become trapped after splicing into Cur strings during the renaturing process. As a first prototype shown in Fig. 1b, the modified Cur glucan bearing 4-dimethylaminobenzoate was developed as a reporter to sense the tetrasaccharide acarbose (Fig. 1c) in aqueous media in a selective and sensitive manner [40]. Although acarbose is used worldwide to treat type 2 diabetes and obesity, the medication causes unwanted side effects; thus, methods to monitor acarbose concentrations in blood, which is an aqueous solution, in real-time are highly desirable [44]. During our acarbose-sensing investigation using modified Curs, we established that morphological changes in the modified Cur induced by switching the solvent from DMSO to an aqueous medium play a significant role in sensing [45]. Namely, as shown in Fig. 1d, the random-coiled string of the modified Cur in DMSO changes to a globule in aqueous media, which selectively captures acarbose in dynamic and flexible strings, generating Cur-saccharide coaggregation. Based on the sensing mechanism, the limits of detection (LODs) were significantly improved from 10 mM for DABz-Cur [40] to 5 mM for H2Por-Cur [45], 200 μM for AlPor-Cur [45], 100 μM for AlTPP-Cur [46], and 5 μM for TPE-Cur [47] by varying the reporter (see structures in Fig. 1b). Previous studies have noted the appreciable efficacy of the fluorescent reporter TPE compared to the circular dichroism chromophores DABz and Por. Therefore, exploring novel dynamic polymer backbones that can trap oligosaccharides is very important.

Fig. 1
figure 1

Chemical structures of (a) glucans, curdlan (Cur), schizophyllan (SPG), and highly branched glucan (6BG3); (b) modified Cur; (c) acarbose; and d Schematic representation of the reversible dynamic morphological changes in modified Cur in aqueous media: random coil-to-globule (top) and -cur-saccharide coaggregation (bottom) processes. e TPE-6BG3

Full size image

In this study, to expand the range of glucans and explore this new possibility, we focused on the highly branched glucan 6BG3 (Fig. 1a, bottom), which is a 1,6-glucose-branched β-1,3-glucan, rather than other branched glucans, such as schizophyllan (SPG) (Fig. 1a, center). Recently, we established that 6BG3 also possesses denaturing/renaturing characteristics based on random coil-to-globule conversion, as observed for Cur [48]. These findings suggest that 6BG3 can function as a dynamic, flexible, and induced-fit type of oligosaccharide chemosensor. This prompted us to investigate the type of oligosaccharide recognized by 6BG3 in aqueous media, the extent of recognition, and the underlying mechanism. Herein, we report the oligosaccharide-sensing behavior of the fluorophore-probed 6BG3 (Fig. 1e) in aqueous media. According to previous reports on Cur chemosensors, the TPE fluorophore is suitable for the present purpose. This comparative study of 6BG3 vs. Cur provides new insights into the factors that govern oligosaccharide-sensing outcomes when using glucans.

Experimental procedure

Materials

Spectroscopic-grade DMSO, Milli-Q H2O, and commercially available nonaminosaccharides were used as received. 6BG3 (degree of glucose branching = 90%) was supplied by DS Wellfoods Co., Ltd., and was dried at 60 °C under high vacuum prior to use. The number-average weight (Mn) and polydispersity index (PDI) of 6BG3 were previously determined to be 3300 and 3.9, respectively [48]. Glucosamine and validamycin A, available in acidic form, were neutralized by adding an appropriate amount of aqueous KOH to DMSO containing TPE-6BG3. All polymer concentrations are expressed in chromophore units unless noted otherwise.

AFM measurements

A 1:9 (v/v) DMSO-H2O solution of native 6BG3 or TPE-6BG3 was dropped onto the mica surface, predried under N2 gas flow, and dried under high vacuum prior to AFM observation.

Saccharide-sensing procedure

Sample solutions of TPE-6BG3 in 1:9 (v/v) DMSO-H2O were prepared; a portion of the stock DMSO solution containing TPE-6BG3 and the target saccharide was diluted with H2O. The resulting mixture was further stirred for 10 min and subsequently subjected to fluorescence spectroscopy.

Synthesis of TPE-6BG3

The synthetic scheme was shown in Fig. 2. 6BG3 (120 mg, 0.39 mmol in glucose units) was added to dry DMSO (2 mL) in a three-necked flask. The solution was heated at 70 °C for 1 h, followed by the addition of N-methyl-2-pyrrolidinone (NMP) (18 mL). The resulting highly viscous solution was stirred at 90 °C overnight. After cooling to room temperature, TPE-COOH (84 mg, 0.22 mmol) was dissolved in the solution. Subsequently, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride (EDC) (425 mg, 2.22 mmol) and N,N-dimethyl-4-aminopyridine (DMAP) (220 mg, 2.22 mmol) were added to the DMSO/NMP solution, and the mixture was stirred for 3 days. To promote the reaction, EDC (426 mg, 2.22 mmol) and DMAP (220 mg, 2.22 mmol) were added every 24 h. After three days, the reaction mixture was slowly poured into methanol (200 mL), and a white precipitate was generated. This precipitate was collected, washed with methanol (3 × 50 mL), and dried under high vacuum to afford TPE-6BG3 in 64% yield (135 mg, 0.25 mmol in monomer unit) as a white solid. 1H NMR (400 MHz, DMSO-d6, 25 °C) δH 7.71 (H7), 7.12 (H8), 6.97 (H8), 5.22–2.98 (sugar protons); 13C NMR (150 MHz, DMSO-d6, 25 °C) δC 143.1 (CAr), 140.1 (CAr), 131.1–127.4 (CAr), 103.6 (C1), 86.6 (C3), 77.1, 76.8, 75.1, 74.2, 73.0, 70.5, 68.9, 61.4 (C6); IR ν 3426, 2895, 2360, 2135, 1716, 1645, 1445, 1026, 750, 700 cm–1. The degree of substitution (DS) for TPE-6BG3 was determined by comparing the molar extinction coefficient (ε) of authentic anionic TPE-COOH and the TPE-COOH liberated from TPE-6BG3 upon hydrolysis. A 5:95 (v/v) aqueous KOH-DMSO solution of TPE-6BG3 (0.094 g/L) was stirred for hydrolysis and monitored via UV/vis spectroscopy. After hydrolysis (21 h) was completed, the absorbance at 310 nm was 0.57579. Thus, the concentration of liberated TPE-COO- from TPE-6BG3 was estimated to be 4.89 × 10–5 M based on the ε = 11772 of TPE-COO-. When the DS of TPE-6BG3 was 1, the concentration of liberated TPE-COO- was calculated to be 1.41 × 10–4 M ( = 0.094/666.51). Thus, the DS can be estimated as 0.35 ( = 4.89 × 10–5/1.41 × 10–4). TPE-6BG3, with a DS of 0.35, was soluble in DMSO and in 1:9 (v/v) DMSO-H2O.

Fig. 2
figure 2

Synthetic scheme of TPE-6BG3

Full size image

Results and discussion

For this purpose, the best-performing TPE chromophore for aggregation-induced emission (AIE) [49] was attached to the 6BG3 backbone as a reporter to afford TPE-6BG3 with a degree of substitution (DS) of 0.35 (Fig. 1e) (see NMR spectra in the Supplementary Information (SI)). First, we investigated the morphological changes in TPE-6BG3 using atomic force microscopy (AFM). An aqueous solution (1:9 (v/v) DMSO-H2O) of native (nonmodified) 6BG3 was drop-cast onto mica and then completely dried. The AFM image of 6BG3 shown in Fig. 3a reveals a dispersed/diverse structure, as previously reported [48], indicating that a renatured macromolecular structure similar to that of other glucans was formed [45,46,47]. In contrast, the AFM image of mica obtained from the aqueous solution of TPE-6BG3 exhibited a dotted, globular morphology in the renatured state (Fig. 3b). The height and width of the globules were estimated to be 8.6 ± 5.9 nm (n = 13) and 125 ± 32 nm (n = 12), respectively (Fig. S2 in the SI). These morphological changes (random coil-to-globule conversions induced by modifications of the chromophore on the glucan backbone) are very likely to be common in glucan chemistry.

Fig. 3
figure 3

AFM images of (a) native 6BG3 (data from ref. 48) and (b) TPE-6BG3 prepared from 1:9 (v/v) DMSO-H2O solutions drop-casted on mica surfaces

Full size image

Next, the photophysical properties of TPE-6BG3 were analyzed in solution to investigate the effects of these morphological changes on the fluorescence behavior. As shown in Fig. 4a (black line, inset), the DMSO solution of TPE-6BG3 displayed negligible fluorescence across the wavelength range, indicating that the phenyl rings in the TPE luminogen moved freely on the randomly coiled 6BG3 backbone. Interestingly, the fluorescence spectrum for the aqueous DMSO solution of TPE-6BG3 (Fig. 4a, red line) was intense due to the aggregation of the TPE luminogen based on its AIE character, which is consistent with globule formation in aqueous media; the appended TPE chromophores aggregated in the globules formed by the dynamic 6BG3 backbone, affecting high emission. The fluorescence quantum yield (ΦF) of TPE-6BG3 in aqueous DMSO was sufficient for detection (0.30), in contrast to its undetectable ΦF in DMSO. Direct evidence of globule formation was obtained using dynamic light scattering (DLS) measurements. As shown in Fig. 4b (black line), the DLS data obtained in aqueous DMSO indicated the presence of monodisperse particles with a hydrodynamic diameter (dh) of 43 nm. To elucidate the origin of the excited species, the fluorescence lifetimes of TPE-6BG3 were obtained. The decay profiles obtained in aqueous media were not subjected to regular deconvolution fitting, as the number of excited species cannot be determined due to the globular ensemble of the TPE fluorophores. Therefore, in this situation, the stretched exponential function (Kohlrausch decay) model (Eq. 1) appears to be more suitable for polymeric systems [50, 51]:

$$I(t)={I}_{0}\exp [\left(\right.-(t/\tau )]^{\beta }$$
(1)

where I0 and I(t) express the photon counts at the initial and arbitrary times, respectively, τ is the average lifetime, and β is the dispersion factor (0 < β < 1). As shown in Fig. 4c (left), the decay profiles of TPE-6BG3 in aqueous DMSO were fitted using Eq. 1 to obtain τ and β as 2.7 ns and 0.86, respectively. Next, the changes in optical and particle characteristics upon complexation with oligosaccharides were investigated.

Fig. 4
figure 4

a Fluorescence spectra (λex 378 nm) of TPE-6BG3 (36 μM) in DMSO (black) and in 1:9 (v/v) DMSO-H2O at 25 °C. b DLS profiles of 1:9 (v/v) DMSO-H2O solutions of TPE-6BG3 (9 μM) without (black) and with acarbose (200 μM, red). c Fluorescence lifetime decays (black) of 1:9 (v/v) DMSO-H2O solutions of TPE-6BG3 (9 μM) without (left) and with acarbose (200 μM, right). The orange lines show the fitting results

Full size image

To explore the selectivity of the dynamic globulization of TPE-6BG3, we tested its response to various saccharides, ranging from mono- to hexasaccharides (see structures in Fig. 5a); the globule was retained at least during the spectroscopic examinations. As shown in Fig. 5b, the fluorescence intensities of all the examined saccharides decreased for TPE-6BG3 upon complexation with saccharides (see Figs. S3–6 in the SI for detailed data). Notably, acarbose and its analog, validamycin A, induced markedly enhanced fluorescence, indicating that robust interactions occurred between the dynamic 6BG3 backbone and acarbose. Interestingly, the data in Fig. 5c (red regression line) show that the fluorescence changes were strongly correlated with the length of the saccharide skeleton. Therefore, the extensive hydrogen bonding network in the 6BG3 backbone plays a significant role in the sensing behavior. Among the saccharides tested, acarbose and validamycin A possess a valienamine structure (red moiety in Fig. 5a) and may interact strongly with the dynamic 6BG3 globule. Thus, the responses to glucosamine, which bears a hydrogen-bond donor amino group, and hydrophobic fucose, which has a methyl group, were compared to those to the corresponding monosaccharide, glucose. Based on the results, the former compounds induced more pronounced changes in fluorescence intensity. This result further supports the hypothesis that the high selectivity of the 6BG3 globule for acarbose (valienamine + maltotriose) led to the difference. We found that 6BG3, a glucan, also exhibited high selectivity toward the acarbose skeleton. This discovery of acarbose sensitivity could apply to general glucans, such as SPG and other branched glucans. Titration experiments were also conducted to determine the sensitivity for acarbose using the TPE-6BG3 chemosensor. As shown in Fig. 5d, the AIE intensities gradually decreased upon complexation of acarbose with TPE-6BG3, indicating that the dynamic globules expanded (vide infra). In Fig. 5e, the slope of the linear decrease, which was obtained from Δint versus the concentration of acarbose (µM), was calculated to be –3.51 and was used as the sensitivity parameter, Δint = -y [acarbose (μM)]; the value obtained is smaller than that (–5.94) of TPE-Cur (DS = 13%) [47]. Furthermore, the LOD for acarbose obtained using TPE-6BG3 was determined to be 10 μM (see Fig. S7 and the present LOD definition in SI); the LOD obtained using TPE-Cur was 5 μM, as described above. The differences between 6BG3 and Cur are most likely responsible for the looser renaturing behavior [48] based on the highly branched glucose on the glucan backbone.

Fig. 5
figure 5

a Chemical structures of the tested saccharides (mono- to hexasaccharides). b ΔInt. (484 nm) of TPE-6BG3 (36 μM) induced upon complexation of various saccharides (200 μM); ΔInt. represents the delta fluorescence intensity. c ΔInt. data plotted as a function of saccharide length. d Fluorescence spectra (λex 378 nm) of TPE-6BG3 (36 μM) in the absence (black) and presence (10, 15, 50, 100, 150, and 200 μM, colored lines) of acarbose in 1:9 (v/v) DMSO-H2O at 25 °C; the excitation wavelengths applied were those at which comparable absorbances were obtained. e ΔInt. data were plotted as a function of the acarbose concentration; ΔInt. = –3.51 [acarbose (μM)]. f Schematic illustration of the changes in globule expansion-based AIE

Full size image

To obtain mechanistic knowledge on the selective sensing of acarbose by TPE-6BG3, the abovementioned analyses were performed with acarbose. DLS analysis revealed a dh of 53 nm (Fig. 4b, red line). The expanded globule structure resulted from the formation of 6BG3-acarbose coaggregates, as illustrated in Fig. 5f. This finding is further supported by the gradual decrease in the AIE intensities upon the addition of acarbose, with a decreased ΦF of 0.27. The decrease in fluorescence intensity upon the addition of acarbose can be explained by the restricted rotation of the phenyl rings in the TPE luminogen, which relaxed due to the expansion of the dynamic globule. Furthermore, the decay profile in the presence of acarbose was subjected to Eq. (1), which yielded τ and β values of 2.5 ns and 0.83, respectively. The shorter τ and high degree of β strongly indicate that the degree of freedom increased for the appended TPE chromophore, reinforcing the expansion of flexible and dynamic globular coaggregation.

Conclusions

In conclusion, we developed a glucan-based dynamic, induced-fit oligosaccharide chemosensor. The highly branched glucan 6BG3 traps acarbose, a clinical drug, in aqueous media. Acarbose sensing was achieved in a selective and sensitive manner via globule-to-glucan-saccharide coaggregation conversion. This sensing behavior may be general for all glucans. Therefore, the results obtained herein can serve as a guideline for the construction of new smart chemosensors based on a dynamic induced-fit approach, which may enable the sensing of complicated oligosaccharides and sugar chains, which are challenging to recognize using conventional lock-and-key-type chemosensors.