Introduction

Nanobiosensor have played a key role in the fast cancer diagnosis and in clinical medicine and gained a great deal of attention in the areas of biological sciences [1, 2]. Recently, glyconanoparticles (GNPs) have been of increased interest as biofunctional nanosensors, which effectively combine the unique chemico-physical and optical properties of various nanoparticles with the characteristics of the carbohydrate coating [3,4,5]. Carbohydrates on the surface of NPs not only enhance the water solubility, biocompatibility and stability, but also endow these NPs with recognition ability. Thus, the GNPs have shown great potential for biomedical applications, especially in cellular labeling and imaging [6, 7]. Therefore, research on the biological effects associated with carbohydrate and develo** new GNPs have become a hot topic in recent years.

A variety of nanoparticles (NPs), including gold, iron oxide, and semi-conductor quantum dots (QDs), have been used as the carriers of carbohydrates to prepare GNPs since the first synthesis of carbohydrate-functionalized gold nanoparticles in 2001[8]. Among them, QDs have been found as an attractive biomedical resource and widely used in vitro and in vivo for multiple color imaging and targeted drug delivery due to their broad absorption, high quantum yield, and long-term photostability [9,10,11]. Besides, many researches indicate that QDs have great potential in cancer detection and treatment [12, 13]. Good water solubility, biocompatibility, and biological targeting of QDs, all of which are essential for biomedical application, could easily be achieved through surface modification or conjugation with carbohydrates. Therefore, glyco-quantum dots (glyco-QDs) have attracted significant attention and many different glyco-QDs have been developed by coating with a variety of carbohydrates. For example, Shin-Ichiro Nishimura et al. synthesized a series of carbohydrate-capped Cd-based quantum dots (CdSe/ZnS and CdSeTe/CdS) with a variety of common sugars, such as α-glucose, α-mannose, lactose, N-acetyl-β-D-glucosamine, and N-acetyl-β-D-galactosamine. In vivo near-infrared (NIR) fluorescence imaging of these glyco-QDs revealed the importance of the terminal sialic acid residues for achieving prolonged in vivo lifetime [14]. Recently, Richichi's group reported a kind of CdSe/ZnS QDs-based fluorescent glyconanoprobe as nanoprobes for carbonic anhydrase IX imaging in cancer cells [15]. However, to date, the majority of glyco-QDs related research in biomedicine focused on Cd-based QDs including CdS, CdSe, and CdTe QDs, which are often added by ZnS and/or ZnSe as protective shells. Unfortunately, many studies have identified that highly cytotoxic Cd2+ ions in QDs can been released in cellular environment, which would ultimately increase the cytotoxicity effect of these QDs in a biological environment [16, 17]. So, the broad potential applications of Cd-based QDs in clinical medicine have been questioned by many biomedical researchers due to the high toxicity of heavy metal cadmium [18]. Accordingly, the potential high toxicity of fluorescent GNPs based on Cd-based QDs severely hinders their biomedical application. These results motivated the development of Cd-free glyco-QDs with less toxicity.

Currently, there has been a growing effort to prepare environmentally friendly Cd-free and less toxic QDs with performance comparable to or even better than existing Cd-based QDs [19]. Among these Cd-free QDs, copper indium sulfide (CuInS2) QDs have shown the greatest application potential for biological applications. CuInS2 QDs are I-III-VI2 semiconductor nanocrystals and does not contain any toxic heavy metals. They can provide PL emission ranging from the visible to the NIR, and have broad absorption, size-tunable photoluminescence (PL), excellent optical, and chemical stability [20, 21]. However, the conventional synthesis of CuInS2 QDs usually employs toxic organic solvents and the hydrophobic QDs are not appropriate for their application in biomedical fields. In addition, the use of organic solvents in the synthetic process is harmful to environment. In the view of hydrophilicity, toxicity and environment protection, the aqueous synthesis of high quality CuInS2 QDs is desired. In recent years, many efforts have been devoted to the aqueous synthesis of CuInS2 QDs for biological applications. Therefore, CuInS2 QDs could offer the opportunity to realize the potential of glyco-QDs without the toxicity limitations in biological environment. However, there is little research on the preparation and application of GNPs materials based on CuInS2 QDs in the biomedical field [22].

In general, the synthetic protocols for the preparation of glyco-QDs reported so far can be mainly described as: covalent conjugation, thiolated-glycans exchange, adsorption, and biotin-streptavidin chemistry [23]. Among them, valence conjugation is generally regarded as a straightforward and simple method. By following this route, functional carbohydrates are directly covalently linked on the surface of QDs. For instance, Penadés et al. directly used the thiol-ending saccharides as cap** ligand and stabilizer to form the GNPs based on CdS QDs in degassed water under room temperature by a straightforward single step method [24]. Obviously, the direct synthesis of high-quality semiconductor glyco-QDs in aqueous medium is less harmful to the environment and health of people compared with the conventional organic synthesis of QDs. And, the resultant QDs are better water-soluble and biocompatible. Nevertheless, to the best of our knowledge, there is almost no literature on the synthesis of glyco-QDs based on ternary CuInS2 QDs in aqueous solution. Therefore, there is a high need to develop nontoxic and water-soluble glyco-CuInS2 QDs capped by carbohydrates as bio-compatible labeling and imaging probes in various biological applications.

Therefore, we developed in this work a simple, green and reproducible methodology for the construction of glyco-QDs in water via direct synthesis methods. These target glyco-QDs were synthesized by the “direct” reaction of thiol-ending monosaccharides (SH-fructose, SH-galactose, SH-mannose, SH-glucose) with metal salts precursors (CuCl2·2H2O, InCl3·4H2O and CS(NH2)2) in water. The as-prepared glyco-QDs not only exhibited well-separated dual-wavelength emission in the visible light region (500 ~ 590 nm) and the NIR range (~ 827 nm) upon excitation spectral scan (215 nm to 515 nm), but also displayed the high photostablity, good water-solubility, favorable photoreversibility and noncytotoxicity. Moreover, these glyco-CuInS2 QDs exhibited excellent biocompatibility and good biorecognition ability. The investigation of cellular imaging in tumor cells (HeLa, A549, MKN-45) indicated their reversible dual-color (green and red) imaging function and excellent membrane-targeting properties originating from a large number of carbohydrates on the surface of QDs. More importantly, they could penetrate uniformly into the interior of three-dimensional (3D) multicellular tumor spheroids (MCTS) due to their high negative charge, which realized the deep penetration depth of QDs in in vitro spheroid models. In summary, these glyco-CuInS2 QDs have successfully achieved 2D and 3D spatiotemporal visualization of the cell membrane, thus providing a huge potential in the application of commercial probes for cell membrane imaging.

Results and discussion

Design and synthesis of four kinds of glycol-CuInS2 QDs

The most common synthetic protocols for the preparation of glyco-QDs are ligand exchange with hydrophilic QDs or surface modification with carbohydrate [25,26,27]. However, the former requires the use of toxic organic solvents and the other one undergoes complex multistep process. Therefore, toxic organic solvent-free technology and simple method for hydrophilic synthesis of QDs are desired. Thus, we proposed a simple, green and reproducible design methodology of the construction of low cost glyco-QDs by the “direct” reaction of neutral monosaccharides (D-fructose and D-glucose mainly from honey, D-galactose from animal milk and D-mannose from fruit peels) with metal salts precursors in aqueous media. Scheme 1 illustrates the synthetic routes to four kinds of glycol-CuInS2 QDs, and the detailed synthesis procedure are described in Additional Information. The thiolation of monosaccharide is a key intermediate in the total synthesis of glyco-CuInS2 QDs. As shown in Scheme 1, four thiol-functionalized monosaccharides (SH-fructose, SH-galactose, SH-mannose, SH-glucose) were obtained by using DCC/DMAP esterification procedure with 3-mercaptopropionic acid. Subsequently, four environmentally friendly glyco-CuInS2 QDs were synthesized by using only these thiol-functionalized monosaccharides as stabilizing agents and cap** ligands in aqueous medium by reaction with CuCl2·2H2O, InCl3·4H2O and CS(NH2)2 under an Ar atmosphere.

Scheme 1
scheme 1

Schematic illustration of the synthesis and biolabelling process of four kinds of glycol-CuInS2 QDs

Full size image

In order to explore the optimal conditions for the synthesis of glyco-CuInS2 QDs, the molar ratios between Cu, In, S and the thiol-ending saccharide ligands, pH of the reaction solutions and reaction temperatures were systematically investigated. Additional file 1: Fig. S3–S5 show these changes in fluorescence intensity at 827 nm of the synthesized glyco-CuInS2 QDs at the different molar ratios, pH and reaction temperatures, respectively. Meanwhile, the optimal synthesis conditions are summarized in Additional file 1: Table S1. The experimental results show that the intensities of fluorescence of Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs reach the strongest when the ratios of Cu, In, S to ligands were 1:1:2:36, 1:1:2:24 or 1:1:2:12 at pH 11 ~ 12 and 150 °C, respectively. Moreover, the emission wavelength of these glyco-CuInS2 QDs do not change significantly with the change of the above reaction conditions. Thereupon, we synthesized a series of structurally well-defined glyco-CuInS2 QDs with natural carbohydrates on the surface.

Structural characterization of glyco-CuInS2 QDs

These synthesized glyco-CuInS2 QDs were characterized by TEM, DLS, XRD, FT-IR and XPS to demonstrate whether CuInS2 QDs were successfully encapsulated by the peripheral sulfhydryl monosaccharides, respectively. The effect of monosaccharide composition on crystal structure and morphology of CuInS2 QDs was first confirmed. As shown in the TEM images of glyco-CuInS2 QDs (Fig. 1 a1–a4), all monosaccharide-capped CuInS2 QDs have regular spherical shape and good monodispersity with mean diameters in the range of 3–4 nm in aqueous phase (Fig. 1b1–b4). Meanwhile, these insets to Fig. 1 a1–a4 show clear crystal lattice fringes with lattice spacing of 2.52 Å, 2.48 Å, 2.87 Å and 2.55 Å, respectively, which indicates the single crystal nature of the QDs. Figure 1c shows the XRD patterns of these synthesized glyco-CuInS2 QDs. The XRD spectra in Fig. 1c confirm the crystal structure of these glyco-CuInS2 QDs. The XRD pattern of glyco-CuInS2 QDs consisted of three major peaks with 2θ values of 27.9, 46.3, and 54.8, respectively, and all diffraction peaks corresponding to the (112), (024), and (116) indices of the tetragonal crystal structure (CuInS2, JCPDS no. 47-1372), respectively, suggesting that the crystal structure is well-maintained after cap** with monosaccharides. According to the Debye–Scherrer formula: D = Kλ/Bcosθ (Where θ is the diffraction angle; λ is the X-ray wavelength; B is the half-height width of the diffraction peak; K = 0.89), the sizes of the prepared quantum dots were calculated to be 3.53 nm, 3.89 nm, 4.11 nm and 3.40 nm, respectively, which were consistent with the TEM and particle size test results. Furthermore, the clear lattice fringe (inset of Fig. 1a) indicated that the crystal structure of these CuInS2 QDs are not affected after cap** with four monosaccharides. In addition, the broad diffraction peaks suggest that they have small sizes [28]. Besides, no other diffraction peaks were observed, which emphasizes the absence of any impurities in the prepared samples.

Fig. 1
figure 1

a1–a4 TEM images of Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs at a scale bar of 5 nm, respectively (Inset: HR-TEM images of glyco-CuInS2 QDs and their lattice spacing). b1–b4 Particle size distribution maps for Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs (Inset: Path diagram of glyco-CuInS2 QDs irradiated by laser pointer under daylight). c1 XRD diagram of Glu-CuInS2 QDs synthesized with four monosaccharides as ligands. d1–d4 FTIR spectra of Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs for each of four sulfhydrylated monosaccharides and their synthetic glyco-CuInS2 QDs

Full size image

Moreover, the IR spectrum reveal the composition of as-synthesized glyco-CuInS2 QDs. The FTIR spectra of the sulfhydrylated monosaccharides and sugar-capped CuInS2 QDs are compared in Figure d1-d4. The IR absorption bands of four sulfhydrylated monosaccharides at 3500–3000 cm−1 are ascribed to the stretching vibration of -OH, whereas the peaks at 3000–2900 cm−1, 2500–2600 cm−1 and 1750 cm−1 are assigned to -CH2-, -SH and C = O of the ester groups, respectively. Because the CuInS2 QDs are capped with monosaccharides, typical peaks of sugars can be found in the IR spectrum of glyco-CuInS2 QDs. More importantly, the almost disappearance of the -SH stretching vibrational peak indicates that these sulfhydrylated monosaccharides may combine on the surface of the CuInS2 through the -SH groups [29]. SEM–EDS spectrum of glyco-CuInS2 QDs had been performed in Additional file 1: Fig. S6 to further determine the elemental composition of these QDs. These results indicate that all of glyco-CuInS2 QDs contain C, O, Cu, In, and S elements. Meanwhile, the distribution of elements of CuInS2 QDs are also studied with EDX elemental map**s, the bright points indicate the high concentration of the elements in Additional file 1: Fig. S7. The results show that C, O, Cu, In, and S elements are homogeneously distributed throughout sample, suggesting that CuInS2 QDs are successfully capped with four monosaccharides, respectively.

Besides, XPS was used to further confirm the composition and the valence state of all elements structure in these monosaccharide-capped CuInS2 QDs (Additional file 1: Fig. S8). The XPS survey scans for glyco-CuInS2 QDs clearly displayed the C 1 s, O 1 s, Cu 2p, In 3d, and S 2p signals at around 285, 531, 952, 452 and 163.3 eV, respectively [44]. The dual-emission phenomenon has a wide range of applications as chemosensors and markers for bioimaging [31, 45, 46]. Besides, glyco-CuInS2 QDs mainly emit NIR fluorescence at excitation wavelengths in the range 215–250 nm and green fluorescence in the range of 400–500 nm. As shown in Fig. 4a2-d2, these glyco-CuInS2 QDs show different green fluorescence emission maxima at 582 nm (Fru-CuInS2 QDs), 500 nm (Gal-CuInS2 QDs), 502 nm (Man-CuInS2 QDs) and 553 nm (Glu-CuInS2 QDs) under 515, 415, 395 and 455 nm excitation, respectively. And, all of glyco-CuInS2 QDs exhibit the same NIR fluorescence emission maxima at 827 nm under 215 nm excitation. The fluorescence quantum yields (QY) and lifetimes (τ) of these QDs for the green fluorescence emission in water were measured, respectively. As shown in Additional file 1: Table S2, the QYs of Fru-CuInS2, Gal-CuInS2 and Glu-CuInS2 QDs exceed 10% in water and Fru-CuInS2 QDs have highest QY (Φf = 15.05) compared with three other QDs (Φf,Gal-CuInS2 = 10.05%, Φf,Man-CuInS2 = 4.60%, Φf,Glu-CuInS2 = 12.10%). Furthermore, the average fluorescence lifetimes for the maximum emission of glyco-CuInS2 QDs in water are 4.27 ns, 4.71 ns, 3.96 ns and 6.61 ns, respectively (Additional file 1: Table S2 and Fig. S13). Unfortunately, the QYs and lifetimes of QDs for the NIR emission at 827 nm cannot be obtained due to the lack of suitable excitation source (~ 215 nm) in instrument testing.

Notably, the green and NIR emissions are well separated by > 235 nm, which is extremely beneficial for biosensing or bioimaging. Surprisingly, these glyco-CuInS2 QDs have very large Stokes shift of up to 612 nm for NIR emission (Additional file 1: Fig. S14), which allows for efficient separation of absorbance and emission maxima. More interestingly, the fluorescence emissions of glyco-CuInS2 QDs can be tuned from green to NIR by switching the excitation wavelength. As shown in Fig. 4a3-d3, lower-energy emission peak at 827 nm is much higher than the higher-energy emission peak around 500 nm under 215 nm excitation. And then, the green light intensity gradually increases and the NIR light intensity decreases with the increase in the excitation wavelength from 215 to 515 nm. Ultimately, the green fluorescence is in a dominant position. In addition, the approximate luminescence color changes of glyco-CuInS2 QDs were identified with the help of chromaticity color coordinates in the CIE chromaticity diagram (Fig. a4–d4). The CIE chromaticity diagram results indicate that different light colors4 are produced on the standard swatches at different excitation wavelengths in the range of 215 to 515 nm, while the light colors are excessive from red to green. Therefore, the luminescence colors of these glyco-CuInS2 QDs could be easily adjusted through changing the excitation wavelength.

Moreover, the photostability of glyco-CuInS2 QDs is extremely important feature used as fluorescence probes in biological staining and cell imaging. Therefore, the luminescent stabilities of four glyco-CuInS2 QDs have been investigated using fluorescence spectroscopy. As shown in Additional file 1: Fig. S15, all glyco-CuInS2 QDs in aqueous solution show good photostability because there are no obvious changes between the fluorescence spectra of glyco-CuInS2 QDs at the first day, the seventh days and the fourteenth days. Furthermore, light illumination stabilities of glyco-CuInS2 QDs in the aggregated state are also investigated under the illumination of two light sources and times. Firstly, all these solid-state glyco-CuInS2 QDs were prepared by adding dropwise the concentrated solutions on the glass slides for drying. Then, these glass slides were viewed under a fluorescence microscope. Significantly, all these solid-state QDs exhibited strong fluorescence signals irradiated by altering the excitation wavelengths in both of the green and red channels (Fig. 5). Besides, the luminescent brightness did not decrease significantly after 24 h of illumination. These test results also identify that all glyco-CuInS2 QDs have excellent light-stabilities.

Fig. 5
figure 5

Fluorescence microscopy images of glyco-CuInS2 QDs, obtained from green a and red b emission channels continuously irradiated for 0 h, 8 h, 16 h, 24 h respectively (the images also include the emission spectra of QDs at different excitation wavelengths)

Full size image

Formation and dual emission mechanisms of glyco-CuInS2 QDs

The formation mechanism was summarized in Fig. 6. As shown in Fig. 6a, Cu+ and In3+ cations were released from metal precursors and coordinated with thiol-ending glycosides to form Cu-In thiolate [CuIn(SR)x]n after being mixed with thiol-ending monosaccharides (RS-H). Subsequently, [CuIn(SR)x]n was decomposed to QDs by thermolysis process. The formation of glyco-CuInS2 QDs could be explained by a nucleation-growth mechanism following the LaMer model [47]. Meanwhile, based on the hard-soft acid–base (HSAB) theory, Cl− is hard base, while Cu+ and In3+ are soft and hard acid, respectively. So, Cu+ is much easier to be released from CuCl than In3+ ions from InCl3 due to the weaker interaction between Cu-Cl. On the other hand, RSH is soft base. So, the Cu-rich [CuIn (SR)x] n is formed because of the stronger coordinating ability between Cu+ and RS− than that between In3+ and RS− [48]. Thus, highly Cu-rich glyco-CuInS2 QDs are finally obtained with high density of surface defect states.

Fig. 6
figure 6

a Schematic illustration of formation mechanism of glyco-CuInS2 QDs; b Schematic diagram of semiconductor QDs photoluminescence (Solid lines represent radiative transitions, dashed lines represent non-radiative transitions)

Full size image

Dual emission mechanism of glyco-CuInS2 QDs can further be elucidated based on their absorption and emission behavior. As shown in Fig. 6b, the energy band structure usually consists of a low-energy valence band (VB) filled with electrons and an empty high-energy conduction band (CB). The photon energy is transferred to the QDs if they can absorb incident light under laser excitation. Thus, under the action of high-energy rays, the electron jumps from the fundamental energy level of the valence band to the higher energy level of the conduction band. Subsequently, the photogenerated electrons are transferred nonradiatively to the intra-bandgap states (1). Meanwhile, the surface defects on the QDs capture some electrons, resulting in the intra-bandgap states (2). All electrons from two states undergo radiative and nonradiative recombination with holes [49]. Therefore, the emission monitored in the visible light region of 500 ~ 600 nm (high energy band) and the emission recorded at 827 nm (low energy band) may be attributable to visible excitonic emission and near-infrared surface defect emission, respectively. Indeed, this dual emission mechanism is consistent with previous published works on the two distinct transitions in CuxInS2 QDs [50, 51].

The pH-sensitive qualities of fluorescence

Due to the aggressive proliferation of tumor cells and the rapid formation of irregular blood vessels, nutrients and oxygen are rapidly consumed at the tumor site, causing lactic acid metabolites produced by glycolysis in the tumor cells to accumulate in the tumor interstitium, ultimately leading to a drop in the pH of the extra-tumor cellular environment to 6.2–6.9 [52]. Therefore, the high brightness of fluorescence probes in a weak acid solution is of great importance for in vivo imaging. Thus, it is essential to examine the effects of pH on the fluorescence intensity of glyco-CuInS2 QDs. As shown in Additional file 1: Fig. S16, the fluorescence of four glyco-CuInS2 QDs exhibit pH-sensitive response, and the maximum fluorescence intensity can be obtained when the pH is in the range of 6 to 7, which will be beneficial to the in vivo imaging.

The dual-color live-cell membrane imaging of glycol-CuInS2 QDs

Motivated by these excellent optical performances of glycol-CuInS2 QDs, we further study their imaging performances in living organisms. Three tumor cells (HeLa, A549, MKN-45) were selected for testing prior to performing cell imaging experiments. Firstly, the detection of cytotoxicity is a critical prerequisite for its application in cell imaging. Three kinds of tumor cells were incubated in concentrations of 20, 40, 60, 80 and 100 μg/mL of Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs aqueous solutions for 24 h, and their effects on cell viability were determined by MTT [53, 54], respectively. As shown in Additional file 1: Fig. S17, the survival rates of three tumor cells are above 90% for all culture concentrations, which indicated that all glycol-CuInS2 QDs possesses low cytotoxicity towards the living cells.

Besides, their abilities to enter tumor cells are also essential for biological imaging. Therefore, we investigated how glycol-CuInS2 QDs entered three kinds of tumor cells before performing cell staining. The dynamical interaction process of glycol-CuInS2 QDs with tumor cells at different time points through a live cell workstation (Olympus DP80) were recorded in Fig. 7a1–d1, respectively. It can be observed that the aggregated QDs particles slowly enter the cell membrane with time migration from 0 to 100 min. These results indicate that all glycol-CuInS2 QDs enter target cells by endocytosis, which occurs at the cell surface. The low cytotoxicities and excellent cell-entry abilities of these glycol-CuInS2 QDs provide strong in vivo evidences for live cell staining.

Fig. 7
figure 7

a1–d1 the images of 0–100 min migration of aggregated glyco-CuInS2 QDs towards the cell membrane, respectively; a2–a5 the imaging images of HeLa cells treated with 100 μg/mL glyco-CuInS2 QDs for 2 h, 6 h and 12 h, respectively; b2–b5 the images of A549 cells treated with 100 μg/mL glyco-CuInS2 QDs for 2 h, 6 h and 12 h, respectively; c2–c5 the imaging images of MKN-45 cells treated with 100 μg/mL glyco-CuInS2 QDs for 2 h, 6 h and 12 h, respectively (scale bar 10 μm); d2–d5 mean brightness values of HeLa, A549, MKN-45 cells after incubation with glyco-CuInS2 QDs for 2 h, 6 h and 12 h

Full size image

Subsequently, cell imaging experiments were performed to further demonstrate the practical application of four glycol-CuInS2 QDs. The HeLa, A549, MKN-45 cells were individually isolated in 12-well plates and incubated with glycol-CuInS2 QDs at a concentration of 100 μg/mL for 2, 6 and 12 h, respectively. As shown in Fig. 7, all glycol-CuInS2 QDs are highly concentrated in HeLa, A549, MKN-45 cell membrane and the fluorescence signals are continuously enhanced with the incubation time increased. For example, almost no fluorescent signals are observed in membrane of three tumor cells because only a few Fru-CuInS2 QDs enter cells after 2 h in Fig. 7a2–c2. With the entry of more QDs in cell membrane after 6 h incubation, the increasing fluorescent signals can be observed. Finally, significant fluorescent signals are observed after 12 h incubation, suggesting that Fru-CuInS2 QDs can be effectively captured by cell membrane. Meanwhile, Gal-CuInS2 QDs, Man-CuInS2 QDs and Glu-CuInS2 QDs also exhibit similar imaging abilities, respectively (Fig. 7a3–c3, Fig. 7a4–c4 and Fig. 7a5–c5). Furthermore, all glycol-CuInS2 QDs show same dual-colorful and tunable fluorescence signals by altering the excitation wavelengths in green and red emission channels.

Besides, methods of imaging provide the means to visualize in space and time, but there is still a great need to make quantitative measurements. Image J is an open-source image processing platform for multidimensional image data, built for the specific needs of scientific images [55]. In this paper, we made use of the Image J software toolkit, which has utility in scientific image analysis, to quantify the imaging fraction. For example, as shown in Fig. 7d2, the specific fluorescence intensities of HeLa, A549 and MKN-45 cells incubated with Fru-CuInS2 QDs are 1.450 a.u, 3.069 a.u. and 5.748 a.u after 2 h of incubation, and increase to 75.656 a.u., 59.634 a.u. and 64.705 a.u. after 6 h of incubation, and significantly increase to 243.413 a.u., 214.783 a.u. and 225.490 a.u. after 12 h of incubation, respectively. Meanwhile, the specific fluorescence intensities of HeLa, A549 and MKN-45 cells incubated respectively with Gal-CuInS2 QDs, Man-CuInS2 QDs and Glu-CuInS2 QDs also exhibit the similar growth trends with the increasing of incubation time (Fig. 7d3–d5).

Most notably, confocal imaging experiments show the excellent specific cytomembrane location for living tumor cells. According to a previously reported literature [56], based on the recognition of glyco-cell membrane receptor proteins, nanodots with more glycosyl modifications are more recognized by the receptor, entered the cells more and fluoresced brighter. Thus, these glycol-CuInS2 QDs can specifically identify cancer cells with glyco-receptor protein overexpression on the cell membrane for fluorescence imaging. The cell membrane is involved in various cellular processes and biological functions, such as cell signalling, cell adhesion, endocytosis, cytosol and selective permeation of substances, therefore, the cell membrane is indispensable to the cell, and its observation can also yield much information related to cellular state and disease. In view of this, the synthesis of cell membrane-targeted fluorescent bioprobes is of great importance [57]. Furthermore, in order to determine whether the fluorescence emitted by the two channels remains in the same position, a fluorescence co-localization assay was performed [58]. From the scattering trend of pixel points in the co-localized pixel plots in Fig. 8, we can find that the fluorescence of glycol-CuInS2 QDs emitted by the red and green channels are in a somewhat linear relationship and favor the red channel, indicating that the red and green fluorescence are co-localized in a certain proportion, but the green fluorescence is slightly darker than the red fluorescence. We then carried out co-localization analysis by a series of different parameters, Pearson's correlation coefficient (PCC) [59]. Figure 8a4–d4 shows the overlap rates of these glycol-CuInS2 QDs in different cell imaging with red and green channel imaging. Reassuringly, all overlaps are above 90%, indicating high targeting specificity.

Fig. 8
figure 8

a1–d1 Pixel plots of imaging confocal after treatment of HeLa cells with Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs, respectively; a2–d2 Pixel plots of imaging confocal after treatment of A549 cells with Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs, respectively; a3–d3 Pixel plots of imaging confocal after treatment of MKN-45 cells with Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs, respectively; a4–d4 Plots of threshold versus Pearson correlation coefficient in fluorescence confocal detection (inset: overlap of different channel imaging positions for the three cells)

Full size image

The imaging of glycol-CuInS2 QDs in multicellular tumor spheroids

The studies of QDs imaging had been carried out in vitro with cancerous cell lines. However, only research in 2D monolayers model is unable to evaluate some very important abilities for glycol-CuInS2 QDs in cancer imaging, for example the ability of QDs to penetrate and retain within a tumor. At present, the challenge associated with QDs is their poor penetration depth, constraining them to a peripheral accumulation in in vivo tumors. Multicellular tumor spheroids (MCTS) have gained increased recognition as a useful three-dimensional (3D) tumor spheroid model for evaluating the effect of exogenous molecules on tumor progression in cancer research because they may closely mimic some physiological characteristics of solid tumors. However, to date, only few studies were conducted to investigate the QDs imaging in 3D in vitro spheroids models. Therefore, MCTS are used to evaluate the tumor penetration capacity of four glycol-CuInS2 QDs in this study, respectively (Additional file 1: Video S1–S4). Figure 9a shows the transport of glycol-CuInS2 QDs inside tumor spheroids at different time points, and a time-dependent penetration of glycol-CuInS2 QDs in multicellular tumor spheroids can been observed from 20 to 180 min. At 20 min post-treatment, red fluorescent spots in the confocal images indicate a distribution of glycol-CuInS2 QDs in the outer region of the 3D model. As the time increases, enhanced fluorescence signal in the inner region can be observed, suggesting more homogeneous uptake. Finally, there glycol-CuInS2 QDs penetrated almost to the center of the 3D spheroids after 180 min of incubation. Besides, as shown in Fig. 9b, the fluorescent signals of Gal-CuInS2 and Man-CuInS2 QDs are observed dispersed evenly throughout tumor cells, indicating the penetration capacities are higher than those of Fru-CuInS2 and Glu-CuInS2 QDs. Previous studies have shown that the internalization ability of QDs to HeLa cellular spheroids is affected by the QDs surface charges. In general, positively-charged QDs cannot reach the necrotic zone since they can be absorbed by the proliferating cells of the spheroid. While the negatively-charged QDs can diffuse more quickly, allowing them to penetrate deeply into the tissues [56, 60]. Thus, our results are in excellent agreement with the experimental results reported in the previous study. Figure 9c shows the schematic representation of glyco-CuInS2 QDs penetration in 3D MKN-45 spheroid. Glyco-CuInS2 QDs can succeed in penetrating uniformly into the interior of 3D MCTS and reach the necrotic zone due to their high negative charge (zeta potential values ranging from − 23.9 to − 30.1 mV). which overcame the problem of poor penetration depth of existing QDs in in vitro spheroid models. These results indicate that glycol-CuInS2 QDs have excellent tumor penetration capacity, which is of particular importance since the fluorescence probes used for cancer imaging must efficiently penetrate tumor tissues to reach all of the viable cells [61, 62]. Therefore, these glycol-CuInS2 QDs have great potential to be used as a fluorescent probe for intratumoral analysis.

Fig. 9
figure 9

a Fluorescence expression of glyco-CuInS2 QDs in 3D MKN-45 cell microspheres at different times (Scale bar: 500 µm); b the corresponding quantitative fluorescence intensity values; c Schematic representation of glyco-CuInS2 QDs penetration in 3D MKN-45 spheroid

Full size image

The dual-color bioimaging of glycol-CuInS2 QDs in zebrafish

Finally, we investigated the imaging abilities of glycol-CuInS2 QDs in living organisms. Based on the genetic similarity of 87% to humans, the zebrafish has outstanding advantages as a model organism because its experimental.results are applicable to humans in most cases. Nowadays, zebrafish has become one of the most valuable models in vertebrate developmental biology and shows great potential for use in other disciplines [63]. Thus, using zebrafish, it is possible to study fundamental questions in the life sciences and to reveal the molecular mechanisms of embryonic and tissue organ development. Therefore, zebrafish is chosen as model organism in our study. Firstly, zebrafishes were incubated with glycol-CuInS2 QDs aqueous solution (1 μM) in a 96-well plate for 1 h. Subsequently, two-channel fluorescence signals were recorded by the fluorescence microscope. As shown in Fig. 10, all glycol-CuInS2 QDs in zebrafish display strong fluorescence signals in green and red channels under different excitation sources of fluorescence microscopy. The results indicate that these glycol-CuInS2 QDs can be absorbed through the skin of zebrafish and are low toxicity in biological systems. Moreover, we used Image J software to perform accurate calculations of the fluorescence intensity of the imaging (Additional file 1: Fig. S18) and used plug-ins to measure the fluorescence position in live imaging of different channels of zebrafish (Fig. 10e1-e4). The trends of pixel point dispersion in the QDs show the proportional overlap in the imaging of the two channels, and the overlap rate is quantified in Fig. 10f, which surprisingly remains above 91.56% for all glycol-CuInS2 QDs, showing a high degree of overlap. Therefore, these results clearly demonstrate that these glycol-CuInS2 QDs synthesized by aqueous method have great potential in in vivo bioluminescence imaging.

Fig. 10
figure 10

Microscopic images of two channels of zebrafish after 1 h incubation with glyco-CuInS2 QDs. a1–d1 bright field; a2–d2 green channel; a3–d3 red channel; a4–d4 merged. Scale bars are 600 μm. e1–e4 Pixel plots of imaging confocal after treatment of zebrafish with Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs samples; (f) Plot of threshold versus Pearson's correlation coefficient in fluorescence confocal detection (inset: four samples treated with zebrafish overlap of the red and green channel imaging positions)

Full size image

Conclusions

We developed an eco-friendly green approach to the aqueous-phase synthesis of QDs-based fluorescent nanobiosensor by the “direct” reaction of thiol-ending natural monosaccharides with metal salts precursors. Compared with existing synthesis methods through surface modification of QDs by carbohydrate, the “direct” strategy is simpler and more low-cost. The formation of glyco-CuInS2 QDs could be explained by a nucleation-growth mechanism following the LaMer model. As-prepared glyco-CuInS2 QDs (Fru-CuInS2, Gal-CuInS2, Man-CuInS2 and Glu-CuInS2 QDs) showed monodisperse spherical morphology (< 5 nm), good water solubility and non-cytotoxicity. These glyco-QDs exhibited excitation-wavelength-dependent visible/NIR dual emission with good photostablity and favorable photoreversibilit, which may be attributable to visible excitonic emission and near-infrared surface defect emission, respectively. Confocal microscopy experiments showed that glyco-QDs had highly specific targeting ability for cancer cell membranes (HeLa, A549, MKN-45) due to their good biorecognition ability originating from glycocluster on the surface of the CuInS2 QDs. Therefore, these glyco-QDs can be used as a potential plasma membrane imaging reagent for dual-color (green and red) imaging of cell membrane. More importantly, these glyco-QDs could succeed in penetrating uniformly into the interior of 3D multicellular tumor spheroids (MCTS) because of their high negative charge, thus suggesting the possibility to exploit such nanoprobes for the tumor bioimaging with excellent tumor penetration capacity. Therefore, our prepared glyco-QDs shows great commercial prospects in the field of biomedicine due to their low cost, simple manufacturing process and deep tumor penetration capacity. Besides, these glyco-CuInS2 QDs can exhibit more efficient recognition of tumor cells by coating with tumor-targeting agents such as cRGD.