N-Doped Carbon Quantum Dots as Fluorescent Bioimaging Agents

: Graphene quantum dots, carbon nanomaterials with excellent ﬂuorescence characteristics, are advantageous for use in biological systems owing to their small size, non-toxicity, and biocom-patibility. We used the hydrothermal method to prepare functional N-doped carbon quantum dots (N-CQDs) from 1,3,6-trinitropyrene and analyzed their ability to ﬂuorescently stain various bacteria. Our results showed that N-CQDs stain the cell septa and membrane of the Gram-negative bacteria Escherichia coli , Salmonella enteritidis , and Vibrio parahaemolyticus and the Gram-positive bacteria Bacillus subtilis , Listeria monocytogenes , and Staphylococcus aureus. The optimal concentration of N-CQDs was approximately 500 ppm for Gram-negative bacteria and 1000 ppm for Gram-positive bacteria, and the exposure times varied with bacteria. N-Doped carbon quantum dots have better light stability and higher photobleaching resistance than the commercially available FM4-64. When excited at two different wavelengths, N-CQDs can emit light of both red and green wavelengths, making them ideal for bioimaging. They can also speciﬁcally stain Gram-positive and Gram-negative bacterial cell membranes. We developed an inexpensive, relatively easy, and bio-friendly method to synthesize an N-CQD composite. Additionally, they can serve as a universal bacterial membrane-staining dye, with better photobleaching resistance than commercial dyes. resistance than the commercially available membrane dye FM4-64. In addition, N-CQDs can emit red and green ﬂuorescence when exposed to two different wavelengths. Our results show that N-CQDs can be synthesized using a one-step hydrothermal process and their optical properties make them suitable as biological ﬂuorescent labeling agents.


Introduction
Graphene quantum dots (GQDs) are a three-dimensional material made of graphene or graphite and other graphite derivatives and are synthesized using top-down methods [1][2][3][4]. The material usually has a layered structure, with a lateral size of up to 20 nm [4,5]. The physical and chemical characteristics of GQDs are the same as those of graphene. Graphene quantum dots formed from graphene sheets have a lateral size of <20 nm; however, as the sheets are prepared using the bottom-up method, the maximum size reached is less than 10 nm [6]. Carbon nanodots (CDs) of a diameter less than 10 nm are also called carbon quantum dots (CQDs) [7].
Carbon quantum dots have potential applications in biomedicine [3,[8][9][10][11][12] owing to their attractive fluorescent properties such as photostability and excitation-related emission. Recently, they have been increasingly used as nanometers in sensing, photocatalysis, and optoelectronics [3,[13][14][15][16][17][18]. Carbon quantum dots not only exhibit excellent optical properties of traditional semiconductors but also offer advantages of low cytotoxicity and less environmental hazards [19][20][21][22][23][24][25]. They can be multifunctionally conjugated with other nanoparticles. Such quantum dots have favorable water solubility, biocompatibility, and size; moreover, they are easy to modify and inexpensive to produce. Carbon nanodots, GQDs, and polymer dots possess similar photoelectrochemical properties, despite differences in size and the internal and surface structures [10,26,27]. For these materials to fluoresce effectively, their size and surface chemical groups must be carefully modulated to fine-tune the electronic structures.
The most attractive feature of CQDs is their photoluminescent (PL) property [28,29]. Due to the π-π* transition of the sp2-conjugated carbon and the n-π* transition of hybridization with N, S, P, N, and other atoms, the absorption peak of CQDs falls within the ultraviolet-visible (UV-vis) region [30].
Carbon nanodots from different materials have been developed for fluorescent labeling in various bacterial species. Wang et al. used fluorine-doped carbon nitride (FC 3 N 4 ) powder as the precursor and ethylene glycol as the solvent to synthesize quantum dots at a large scale using a simple glycol-assisted ultrasonic method [31]. They produced a highquality FC 3 N 4 quantum dot of uniform size 1.5-2.0 nm. Doping FC 3 N 4 with fluorine can adjust its band gap, thereby changing the emission peak position (434.5 nm) and increasing the fluorescence intensity. It could be used as a blue, fluorescent probe for Escherichia coli; however, the fluorescent images differed in brightness and were unsuitable for detecting bacterial infection. Weng et al. used a simple dry heating method to synthesize mannosemodified fluorescent CQDs (Man-CQDs) from solid citrate amine and mannose [32]. The average Man-CQD particle size was 3.1 ± 1.2 nm; Man-CQDs showed high solubility, with excitation and emission wavelengths of 365 and 450 nm, respectively. Fluorescent Man-CQDs can stain E. coli K12 flagella. The combination of E. coli FimH and its ligand mannose enabled to quantitatively detect bacteria in laboratory samples with as few as 450 colony-forming units.
Das et al. heated gram shells at 315 • C for 3 h to synthesize CDs that could bind to bacterial cell surfaces and performed multicolor imaging of E. coli cells after incubating the cells with the CDs for 3 h [33]. Pal et al. prepared N-doped CDs from a mixture of curcumin and polyethyleneimine using a hydrothermal method and obtained multicolor-stained cell membranes in live E. coli and Staphylococcus aureus by incubating the cells with the N-doped CDs for 3 h [34]. Baig and Chen synthesized CDs from egg whites and used them for successful multicolor imaging of E. coli and S. aureus [35]. These CDs at a lower volume, just 1 µL of a 0.1 mg mL −1 CD solution, offered faster labeling.
In addition, N, S-co-doped CDs were synthesized by microwave heating a tris-acetateethylenediamine buffer and a thiourea mixture, and they showed bioactive labeling properties [36]. These CDs could effectively label E. coli, S. aureus, and Pseudomonas aeruginosa, but were unable to efficiently label Klebsiella pneumoniae, likely due to the high diversity of bacterial membrane structures. Wang et al. prepared excitation-independent CDs using sodium citrate, urea, and thiourea that stained live Xanthomonas axonopodis cells with a blue fluorescence under UV after 3 h of incubation [37]. Yang et al. autoclaved a glycerol and 3-[2-(2-aminoethyl amino)ethylamino]propyl-trimethoxysilane mixture for 12 h at 260 • C to synthesize CDs and functionalized these CDs with positively charged moieties of quaternary ammonium lauryl betaine on the CD surface [38]. This type of CD could efficiently label the Gram-positive species Micrococcus luteus, S. aureus, and Bacillus subtilis, but not some Gram-negative bacteria such as E. coli, P. aeruginosa, and Proteus vulgaris. Nandi et al. carbonized the 6-O-acylated fatty acid ester of D-glucose to synthesize amphiphilic CDs and successfully labeled both Gram-positive and Gram-negative bacteria. These CDs only non-specifically stained whole cells; they did not specifically stain bacterial membranes [39].
Imaging bacteria is important for various research purposes. Several parameters can affect imaging outcomes, including the surface charge of the CDs and the structural characteristics of the cellular membranes. It is important to develop a universal fluorescent dye for use in both Gram-positive and Gram-negative bacteria that have similar staining loci. Here, we used 1,3,6-trinitropyrene (TNP) to prepare functional N-doped CQDs (N-CQDs), using the hydrothermal method, for the universal staining of cell membranes in various bacteria. For this purpose, we analyzed the fluorescent staining of Gram-positive bacteria B. subtilis, S. aureus, and L. monocytogenes and the Gram-negative bacteria E. coli, S. enteritidis, and V. parahaemolyticus with two fluorescent colors.

One-Pot Hydrothermal Synthesis of N-CQDs
To synthesize N-CQDs, we adapted previously published hydrothermal methods [42,43]. Two gram of pyrene was added to 240 mL of 16 M nitric acid and stirred well at 200 rpm and 80 • C for 48 h. One liter of ddH 2 O was added to the solution and stirred at 150 rpm for another hour. The solution was then filtered and dried to obtain TNP. Next, 2 g of TNP was mixed with 70 mL of dimethylformamide and ultrasonicated for 1 h. The solution was hydrothermally reacted in a stainless-steel autoclave at 180 • C for 12 h to obtain N-CQDs. To separate large particles and obtain a supernatant, the N-CQD sample was centrifuged three times at 13,000 rpm for 5 min each. The supernatant was filtered using a membrane filter (0.22-micrometer) and dried under heating at 50 • C. The yield of the N-CQD composites was approximately 69 wt.%.

N-CQD Characterization
A transmission electron microscope (JEM3010; JOEL, Tokyo, Japan) operating at 200 kV was used to analyze the N-CQD samples. The chemical composition of the samples was analyzed using an X-ray photoelectron spectrometer (XPS, Fison VG ESCA210; West Sussex, UK) equipped with a Mg-Kα radiation emitter. The C1s, N1s, and O1s spectra were de-convolved using a nonlinear least squares fitting algorithm with a symmetric Gaussian function. A Jasco FP-8200 spectrometer was used to record the UV-vis spectrum of the N-CQD suspension, with the wavelength scan rate set to 60 nm/min.
The PL emission spectrum of each suspension was recorded using a fluorescence spectrometer (F-7000 FLS920P; Hitachi, Tokyo, Japan) at 360, 490, and 557 nm. All experimental data represent three independent experiments.

N-CQD Staining
Bacterial cells were grown to OD 600~1 .0, and 3 mL aliquots were centrifuged and pelleted at 12,800× g (rcf) for 3 min. The cell pellets were then washed with 1× TBS buffer and stained with 50 µL of N-CQDs at 0, 500, 1000, and 5000 ppm for 3 min at 25 • C in the dark. For DAPI and N-CQD co-staining, the bacteria were incubated with 5 µL of DAPI for 3 min, and then with N-CQDs for another 3 min in the dark. The bacterial cultures were centrifuged at 90× g (rcf) for 20 s, and supernatants were used for microscopic observation. For FM4-64 staining, the bacteria were stained with 50 µL of FM-4-64 for 3 min in the dark and centrifuged at 90× g (rcf) for 20 s; the supernatants were used for microscopic observation.
Eight microliters of the stained cells were spot on a glass slide and covered with a poly-L-lysine-treated coverslip. A fluorescence microscope (Axioscope 5 Pol TL LED; Carl Zeiss, Göttingen, Germany) fitted with a Zeiss Axiocam 305 color camera (Carl Zeiss) and an attached X-Cite 120PCQ fluorescence illuminator (Excelitas Technologies Corp., Pleasanton, CA, USA) was used to capture microscopic imagines. All experimental data represent three independent experiments.

Analysis of Fluorescent Labeling Different Bacteria
Stained cells were visualized using red TRITC-A-Basic-000, green FITC-A-Basic-000, and blue BFP-A-Basic-000 fluorescent filter sets (IDEX Health & Science, LLC, Rochester, NY, USA) at the excitation and emission wavelengths of 542 ± 10 and 620 ± 26, 475 ± 17.5 and 530 ± 21.5, and 390 ± 9 and 460 ± 30 nm, respectively. A fluorescence microscope was used to acquire the images. At a 4000 mini second exposure time, green images of all the bacteria were acquired. At 450, 760, 1500, 400, 1500, and 640 mini second exposure times, red images of B. subtilis, E. coli, S. enteritidis, S. aureus, L. monocytogenes, and V. parahaemolyticus, respectively, were acquired. At a 35 mini second exposure time, blue images of all the bacteria were acquired.
To quench the experiment, the cells stained with FM4-64 and N-CQDs were irradiated with light by coupling with a red TRITC-A-Basic-000 filter for 3 and 30 s, respectively, and then the cells were re-imaged using the TRITC-A-Basic-000 filter, as described above [44]. All experimental data represent the results of three independent experiments.

Structural and PL Analysis of N-CQDs
In this study, we used a hydrothermal method to synthesize N-CQD nanomaterials. Transmission electron microscope (TEM) and X-ray diffraction (XRD) were used to analyze their crystal structures by analytical scanning, as shown in Figure 1a  Next, we performed the XRD analysis of the N-CQD samples; it showed a peak at 25.2 • , attributed to the (002) crystal plane of the graphite structure (Figure 1d). The Raman spectrum analysis showed that N-CQDs have two peaks at 1359 cm −1 (D) and 1598 cm −1 (G), whereas the 2D peak was at 2700 cm −1 , as shown in Figure 2a. Generally, in Raman analysis, the G band at~1580 cm −1 is a single crystallite of graphite, but the D band at~1350 cm −1 is attributed to amorphous carbon or deformation vibrations of a hexagonal ring. The I D /I G ratio was approximately 0.95 (Figure 1a), and it showed that the N-CQD composite has mixed crystallinity with highly amorphous carbons. In the UV-vis absorbance spectrum analysis, an n-π* transition absorption peak between 300-350 nm is because of C=O/C=N bonding. In Figure 2b, a band-gap transition absorption band is caused by the n-π * transition of the non-bonded electrons of atoms near 500-600 nm in the center or at the absorption edge of the surface molecule. The PL spectrum indicates the optimal excitation wavelength and the strongest emission range of phosphor. As shown in Figure 2c, the spectrum revealed a broad range at approximately 500-620 nm and a sharp peak at 562 nm. Therefore, it is suitable to be excited at around 490 and 557 nm.
The excitation spectrum analysis of the N-CQD composite at 490 and 557 nm (commonly used for in bio-fluorescence microscopy) is shown in Figure 2d. Our results showed that N-CQDs had an excitation wavelength of 490 nm, corresponding to an emission wavelength of approximately 560 and 600 nm (at FTIC filter range excitation/emission: 465-495 nm/515-555 nm), and an excitation wavelength of 557 nm corresponding to an emission at 613 nm (at TRITC filter range excitation/emission: 527.5-552.5 nm/590-650 nm).

X-ray Photoelectron Spectroscopy (XPS) Analysis of N-CQDs
To determine the surface chemical composition of N-CQDs, we performed the XPS analysis. As shown in Figure 3a, the C-C/C=C peak was at 284.5 eV, and the C-N and C-OH peaks were at 285.2 and 286.2 eV, respectively. In Figure 3b, the characteristic peaks at 398.6, 400.6, and 406.0 eV indicated the presence of pyrrole/pyridine N, quaternary N, and NO x , respectively. In addition, the high-intensity peaks at approximately 531.8 and 533 eV indicated C-OH and C=O, respectively (Figure 3c). In Table 1, the molar percentage of C-C/C=C, C-N, C-OH, and C=O was 57.09, 36.21, 4.16, and 2.54%, respectively. Among other functional groups, the proportion of the C-N groups was as high as 36.21%, indicating the presence of N dopants in the carbocyclic structure. The ratio of quaternary N atoms in pyrrole and pyridine was 36.97%:42.04%, and NO x comprised 20.99%. These findings indicate that in the subsequent process, the formation of CD included four or five carbocyclic rings and N dopants and that NO x might have existed at the edge of carbocyclic rings. N-Doping into four or five carbocyclic rings tends to be positively charged and attracted to negative lipid bilayer membranes.

Fluorescent Bioimaging Using N-CQDs
To elucidate the effects of N-CQDs on bacterial cell staining, we used 0, 500, 1000, and 5000 ppm N-CQDs to stain the Gram-positive bacterium B. subtilis and Gram-negative bacterium E. coli at 25 • C. A Zeiss fluorescence microscope was used to analyze the cells, as shown in Figures 4 and 5. The best staining of B. subtilis and E. coli was observed with 1000 ppm N-CQDs with the TRITC filter and 500 ppm N-CQDs with the FITC filter, respectively. We found that N-CQDs stained B. subtilis and E. coli cell membrane and septa (Figure 6a-d). The staining results of Gram-positive and Gram-negative bacteria at different concentrations of N-CQDs can be attributed to their different membrane structures.  In addition, we tested the ability of the N-CQD composite to stain Gram-positive bacteria S. aureus and L. monocytogenes, and Gram-negative bacteria S. enteritidis and V. parahaemolyticus (Figure 7). The results were similar to those of the E. coli and B. subtilis strains; N-CQDs could only stain the cell membrane and septa ( Figure 6). In addition, we coupled N-CQDs with DAPI dye, a chromosomal dye, and found that DAPI stained the inner side of the cells but not the cell membrane, whereas N-CQDs stained the cell membrane but not the inner side of the cells (Figure 8).
The XPS of N-CQDs showed that, in the C1s peak, C-C or C=C, C-N, C-OH, and C=O accounted for 57.09, 36.21, 4.16, and 2.54%, respectively (Table 1). In the N1s peak, pyrrolic/pyridinic N, quaternary N, and NO x corresponded to 36.97, 42.04, and 20.99%, respectively (Table 1). Functional groups such as C-N, C-OH, C=O, and NO x can interact with proteins in the membranes and septa; the hydrogen bonds in C-OH, C=O, and NO x can interact with NH 3 in polypeptides; and C-N can interact with COOH in polypeptides. In addition, we found the percentage of NO x in the total N content was around 20.99%, which might lead to an increase in the interaction between NO x and NH 3 and cause N-CQDs to be bound to the cell membrane. Gram-negative bacteria have double lipid bilayer membranes, which may make it more difficult for N-CQDs to penetrate. However, N-CQDs can still stain both Gram-positive and Gram-negative bacteria. This suggests that N-CQDs can interact with not only the outer lipid bilayer membranes of Gram-negative bacteria but also the outer layer peptidoglycans in Gram-positive bacterial cell membranes.  Compared with the findings of our previous study using PEG6000/CDs as the bioimaging dye, here, we found that N-CQDs can stain both Gram-positive and Gramnegative bacterial membrane and septa, whereas PEG 6000 /CDs stain the entire Gramnegative bacteria, including the membrane. In addition, we found that the NO x content in N-CQDs was around 20.99%, but PEG 6000 /CDs lack NO x . The C-OH and O-C=O content in PEG 6000 /CDs was over 60% in the C1 S peak, whereas that in N-CQDs was only around 7%. This suggests that NO x may play a key role in membrane binding but may not penetrate cellular membranes. Finally, to understand N-CQD photo-stability, we performed a fluorescence quenching experiment to analyze FM4-64 and N-CQDs in B. subtilis cells exposed to fluorescent light at 557 nm for 3, 10, and 30 s. FM4-64 dye was bleached after 10 s of irradiation, whereas the cells retained the N-CQD stain, indicating that an N-CQD is more stable to irradiation as a potential bio-labeling agent (Figure 9). An N-CQD has 10-20 more aromatic condensed rings than FM4-64, which only has two aromatic carbon rings that respond to fluorescent light emission. This might explain why FM4-64 is easier to quench than an N-CQD and suggests that it interacts with cell membranes in both Gram-positive and Gram-negative bacteria because of its high content of NO x and low content of C-OH and O-C=O. Perhaps NO x is better for interacting with cell membranes, but it does not easily penetrate cell membranes.

Conclusions
We developed a simple and easy-to-use hydrothermal method for the synthesis of an N-CQD composite and evaluated its function in the biological labeling of different bacteria. Our results indicate that the fluorescence properties of N-CQDs facilitate the staining of the cell membrane and septa of both Gram-positive and Gram-negative bacteria. In addition, our results showed that the best staining concentration for Gram-positive bacteria is approximately 1000 ppm and that for Gram-negative bacteria it is approximately 500 ppm. The N-CQD composite has better light stability and higher photobleaching resistance than the commercially available membrane dye FM4-64. In addition, N-CQDs can emit red and green fluorescence when exposed to two different wavelengths. Our results show that N-CQDs can be synthesized using a one-step hydrothermal process and their optical properties make them suitable as biological fluorescent labeling agents.