Background

Over and beyond the various organic [16] and inorganic [710] nanoparticles (NPs), many hybrid organic–inorganic [1116] NPs have been intensively researched in biomedical, environmental, and energy applications for their size, shape, charge, and surface chemistry including diverse functionalities. Specific organic-functional groups in many hybrid NPs offer usually unique properties in terms of the accessability and bioactivity of targeting cells or biomolecules in bionanotechnology without some post-functionalization in nanotechnology [1720].

One candidate of organic–inorganic NPs with covalent-bonded primary amines, namely 3-aminopropyl-functionalized magnesium phyllosiliate [i.e., Mg-aminoclay, formulated as [H2N(CH2)3]8Si8Mg6O12(OH)4] was developed by one-pot sol–gel reaction under ambient conditions by Mann et al. [21, 22], showing unique interactions of organic-pendents with cell or other molecules [23, 24] in biomedical fields, as well as with heavy metals [25] in environmental applications. This aminoclay structure is composed of tetrahedral brucite (MgO) in the middle, sandwiched by octahedral silica (SiO2) as the unit structure in the vertical direction (i.e., 2:1 trioctahedral clay) and a repeated tetrahedral/octahedral structure in pairs, known as the 1:1 dioctahedral structure. Diverse high-density primary amines [–(CH2)3NH2] in octahedral structures have been coined aminoclays [26], according to the cationic metals used in their preparation [27, 28].

Recently, organo-building blocks of Mg- and Ca-aminoclays were tested for possible use as drug-delivery carriers, and were found to result in neither cytotoxicity nor inflammation [29]. Further, protonated clusters of Mg-aminoclay with positively charged zeta potential in the wide pH range of 2.0–12.0 [30] were tested as biodistribution and elimination pathways in in vivo mice after Cy 5.0 conjugation with organo-building blocks in delaminated Mg-aminoclay. The results showed fast elimination or excretion of Mg-aminoclay in mice after oral or intravenous injection, respectively, without toxicity [31]. With the exceptions of transparent Mg- and Ca-aminoclays in aqueous solutions, other colored aminoclays have not been tested for cytotoxicity to determine the feasibility of their use in biomedical applications. Importantly, the lack or weak fluorescent-emission intensity of aminoclays has driven research to explore fluorescent imaging for promising drug-delivery-carrier and simultaneous bio-imaging applications in diagnostics and therapeutics.

Carbon nanodots (CD) in zero-dimensional (0D) carbon materials [3234] in the form of biocompatible and non-toxic fluorescent NPs with properties distinct from those of one-dimensional (1D) carbon nanotubes (CNTs) [12, 35, 36] and two-dimensional (2D) graphene [3739], are especially intriguing for their photostability and bio-imaging, and contrast-agent applicabilities, Also, CD has a potential of their mass production from organic molecules using eco-friendly preparation methods [34].

Results

CD conjugated FeAC (CD-FeAC) NPs

In the present study, characterizations of water-solubilized FeAC NPs [27, 4042] and photoluminescent (PL) CD-conjugated FeAC (CD-FeAC) NPs was performed to determine their cytotoxicities in designed cell lines (Table 1). Particularly, the uptake of CD-FeAC NPs in HeLa cells, which is two different-sized and multifunctional NP platforms, was bio-imaged. Schematically (Fig. 1), mesolamellar-stacked FeAC NPs were delaminated in aqueous solution by repulsion of protonated-amine-enriched organo-building blocks of FeAC sheets. CD NPs, contrastingly, were dispersed in aqueous solution by 10 min bath sonication. 1-ethyl-3-(3-dimethylaminopropyl)-carbodiimide/N-hydroxysulfosuccinimide (EDS/NHS) conjugation at low temperature afforded organic pendent groups of organo-building blocks of FeAC (i.e., CD-FeAC) NPs without reforming mesolamellar layers.

Table 1 Designed system of cell lines used in this study
Fig. 1
figure 1

Schematic representation of CD conjugation with organo-building blocks of delaminated FeAC NPs in PBS buffer (0.01 M and pH 7.2) by amide linkage

Rudimental characteristics of FeAC, CD, and CD-FeAC NPs

The PL spectra of FeAC, CD, and CD-FeAC NPs at 290–410 nm excitation wavelengths were recorded with emission spectra (Fig. 2). The PL intensity of FeAC NPs was very weak (Fig. 2a), whereas CD NPs, at an 8.4 % internal quantum yield, showed suitable PL (Fig. 2b). The CD-FeAC NPs also showed good PL data for the purposes of bio-imaging (Fig. 2c). In corresponding Raman spectra, due to the stong PL interference in the CD NPs, CD-FeAC NPs peaks showed indistinguishable D and G bands (Additional file 1: Figure S1) [43]. Transmission electron microscopy (TEM) images of FeAC, CD, and CD-FeAC NPs dispersed in aqueous solution showed successful conjugation of CD NPs with FeAC NPs, with clear contrasts (Fig. 3), compared to only carbon coated copper grid. FeAC NPs displayed the amorphous phase in the entire several-layer morphology, with distinct contrasts (Fig. 3a, b), which result is consistent with the relevant previous study [4042] and CD NPs manifested 2–5 nm spherical and semicrystalline sizes but with some aggregated NPs with ~20 nm size (Fig. 3c, d), in the obtained atomic force microscopy (AFM) image showing a rough root mean square (RMS) of 1.865 (Additional file 1: Figure S2). As for CD-FeAC NPs, CD NPs seemed to be uniformly distributed in the FeAC sheets, and FeAC NPs had an dispersion ability of CD NPs (Fig. 3e, f), especially in light of the containment of CD NPs in the FeAC matrix [44].

Fig. 2
figure 2

Fluorescent emission spectra of a FeAC, b CD, and c CD-FeAC NPs according to excitation wavelength ranging from 290 to 410 nm with 20 nm wavelength intervals

Fig. 3
figure 3

Transmission electron microscopy (TEM) images of a, b FeAC, c, d CD, and e, f CD-FeAC NPs dispersed in PBS buffer at 2.5 mg/mL. Note that in a, b, the arrows indicate the edges of the organo-building blocks of FeAC NPs, and that in cf, the red dotted circles stand for CD NPs. Scale bar a 100 nm, b 10 nm, c 50 nm, d 5 nm, e 100 nm, f 10 nm

For additional information relevant to in vivo mice experimentation and further clinical trials, the aqueous behaviours of FeAC, CD, and CD-FeAC NPs also are needed. Thus, their surface chemistries (e.g., zeta potentials) and hydrodynamic sizes were measured in PBS buffer and serum-free RPMI media at neutral pH (Table 2). The surface charges of FeAC NPs in the PBS buffer and serum-free RPMI media were, respectively, approximately +8 and +2.3 mV of the zeta potentials, whereas that of CD was approximately –30 mV. These results were ascribed, respectively, to the abundance of protonated amine groups (i.e., cationic clusters) and carboxylated groups. Accordingly, the zeta potentials of CD-FeAC NPs in PBS buffer and nutrient media were approximately −4.0 and −7.0 mV, respectivley. The hydrodynamic sizes of FeAC, CD, and CD-FeAC NPs averaged ~202/~337, ~10.5/~20.61, and ~362.3/~519.4 nm in PBS buffer/serum-free RPMI media at neutral pH, respectively. Generally in serum-free RPMI media, the hydrodynamic size was increased due to the strong ionic effects related to the aggregation behavior of NP colloidals. As a result, the NP’s aggregation was induced. In comparison with TEM imaging analysis, DLS data showed relatively less aggregates. It may be related to drying effect for TEM sample preparation [41, 45].

Table 2 Averaged zeta potential values (mV) and hydrodynamic diameters (nm) of FeAC, CD, and CD-FeAC NPs in PBS buffer and serum-free RPMI media (2.0 mg/mL) at neutral pH, as determined by dynamic light scattering (DLS)

XRD patterns and FT-IR spectra of FeAC, CD, and CD-FeAC NPs

For identification and confirmation of the crystalline/amorphous phase, the power X-ray diffraction (XRD) patterns of FeAC, CD, and CD-FeAC NPs were recorded (Fig. 4a). The regular distance in the mesolamellar-structured FeAC at d001 was calculated to ~14.26 Å at 2θ = 6.24°, and in the broad peaks at higher angles, the distances were ~7.83, ~3.92, ~2.77, and ~1.50 Å at 2θ = 11.35, 22.69, 32.34, and 62°, respectively (Fig. 4ai), resulting in the 1:1 dioctahedral phyllosilicate [4042]. In Fig. 4aii, the CD NPs show conventional peaks at 2θ = 24.04 and 42.87° corresponding the assignment of (002) and (101) planes of graphitic carbon. The interlayer spacing for those planes was calculated as 3.71 and 2.11 Å, respectively, indicating that the interlayer spacing of (002) was slightly shifted relative to that (3.44 Å) in bulk graphite [33]. The XRD diffraction peaks in CD-FeAC NPs matched only those in FeAC NPs (Fig. 4aiii). This can be explained by the facts that the CD NPs peaks were relatively weak and hidden.

Fig. 4
figure 4

a Power X-ray diffraction (XRD) patterns and b Fourier transform infrared (FT-IR) spectra of i FeAC, ii CD, and iii CD-FeAC NPs in KBr-pellet mode. In b iii, the red dotted circles indicate amide bondings (C=O stretching, N–H deformation, and C–N stretching)

The covalent bindings of organic-functional groups were detected by the Fourier transform infrared (FT-IR) spectra (Fig. 4b). The FT-IR peak assignments in FeAC NPs were as follows: –OH (3384 cm−1), –CHx (3000 cm−1), –NH3 + (2020 cm−1), H2O (1609 cm−1), –CH2 (1487 cm−1), Si–C (1121 cm−1), Si–OH (1034 cm−1), Si–O–Si (1014 cm−1), Si–O–C (770 cm−1), Fe–O–Si (679 cm−1), and Fe–O (481 cm−1) (Fig. 3bi), which are in line with those reported [4042]. The CD NPs FT-IR peak assignments were –OH (3397 cm−1), C=O (1599 cm−1), and C–O–C (1446 cm−1) (Fig. 4bii), indicating the abundance of oxygen-rich groups and the resultantly hydrophilic surface and partial oxidation state of the CD NPs. Significantly, in the case of CD-FeAC NPs, most of the FeAC and CD NPs peaks were recorded as overlapped. The amide-bonding characteristics in CD NPs at C=O (1620 cm−1) and N–H and C–N (1499 cm−1) were confirmed by the slight peak shift at C=O (1599 cm−1) [4548].

Cytotoxicity results of FeAC, CD, and CD-FeAC NPs

Based on the characterization data for FeAC, CD, and CD-FeAC NPs, MTT cytotoxic assay [29, 49] was tested according to the sample loading concentrations (Fig. 5a–c). FeAC NPs resulted in negligible cytotoxicity in normal cells but a slight (20 %) cytotoxic effect in cancer cells up to 1000 μg/mL. CD showed no cytotoxicity up to 1000 μg/mL. In CD conjugated FeAC (CD-FeAC) NPs, it was also reduced cytotoxicity due to biocompatible CD property and decreasing accessibility with Fe source in FeAC NPs. To see FeAC NPs cytotoxicity in detail, FeAC NPs were evaluated for other cancer and normal cell lines (Fig. 6), it resulted in slightly cytotoxic effects by reduced cell viability (%). Beyond mitochondria-based MTT assay, chromatin-based NR assays of FeAC, CD, and CD-FeAC according the loading concentrations in HeLa cells showed similar trends (Fig. 7).

Fig. 5
figure 5

3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) cell viability (%) of a FeAC, b CD, and c CD-FeAC NPs loading concentrations where yellow dotted lines indicate as 100 % cell viability guides

Fig. 6
figure 6

MTT cell viability assays (%) of FeAC NPs loading concentrations for a A549, b WI-38, c WM-266-4, and d CCD-986SK. Note that yellow dotted lines indicate as 100 % cell viability guides

Fig. 7
figure 7

NR cell viability assays (%) of FeAC (a), CD (b), and CD-FeAC (c) NPs loading concentrations in HeLa cells. Note that yellow dotted lines indicate as 100 % cell viability guides

NPs’s observation of cellular uptake in HeLa cells

Remarkably, the cytotoxicity of CD-FeAC NPs in HeLa cells was slightly decreased at <1000 μg/mL, owing to the biocompatible CD conjugation, compared to that of FeAC treatment. Confocal microscopy images of the uptaken CD-FeAC NPs in HeLa cells and RAG cells showed a clear blue emission (Fig. 8a; Additional file 1: Figure S3a) in comparison with photos of fresh HeLa cells (Fig. 8b) and RAG cells (Additional file 1: Figure S3b). Because FeAC NPs without fluorescent materials were discerned by contrast in cross-sectioned TEM image, the intracelluar location of only FeAC clusters in cross-sectioned HeLa cells was confirmed as the cytoplasm (Fig. 8c), indicating that the uptaken FeAC NPs showed no acute cytotoxicity to the living cell’s morphology. Furthermore, it was confirmed by elemental mapping of the uptaken FeAC NPs into a single HeLa cell, showing markedly contrasted or sharp elemental peaks of Fe and Si in the presence of FeAC NPs (Fig. 9). In addition, confocal microscopy images of DAPI stained nucleus in the absence and presence of FeAC NPs were observed (Fig. 10). Only DAPI stained HeLa cells showed slightly scattered blue emission in nucleus sites (Fig. 10a) but clear blue emission in the presence of FeAC NPs showed spherical or elliptical morphology (Fig. 10b–e), interestingly, FeAC NPs may show a negligible change while nucleus staining, retaining nuclear integrity.

Fig. 8
figure 8

Confocal microscopy images of CD-FeAC NPs treatment in HeLa cells (a) and fresh HeLa cells (b) and where is fluorescent image in upper-left panel, bright field image in upper-right panel, and overlapped image in bottom-left panel, and c cross-sectioned transmission electron microscopy (TEM) image of FeAC NPs treatment in HeLa cells. In b, the arrows and red dotted circles indicate the boundary of cell membrane and the uptaken FeAC NPs, respectively, in a single HeLa cell

Fig. 9
figure 9

Cross-sectioned transmission electron microscopy (TEM) image (a) and its Fe elemental mapping with green color (b), and scanning transmission electron microscopy (STEM) image (c), d enlarged image of c, and its energy-dispersive X-ray (EDX) analysis (e) in Z-contrast mode of a single HeLa cell with treatment of FeAC NPs (200 μg/mL)

Fig. 10
figure 10

Confocal microscopic images of DAPI stained HeLa cells in the presence of FeAC NPs at 0 μg/mL (a), 250 μg/mL (b, c) and 500 μg/mL (d, e). Scale bars (a, b, d) 50 μm, (c, e) 20 μm

Discussion

The cytotoxic results of Mg- and Ca-aminoclays were researched as non-toxic nanomaterial for bio-medical and bio-imaging [2931]. In this study, cytotoxicity of FeAC resulted in slightly cytotoxic effects by reduced cell viability (%), due to the oxidation of H2O2 in cancer cells as a result of coordinated Fe3+ source to hydroxyl free radicals (∙OH) like Fenton-like reaction [5052]. These ∙OH induced death in the cancer cell or apoptosis in vitro due to the radical-induced DNA- and cell-membrane damage. CD conjugation why it reduced cytotoxicity in FeAC NPs may be related to decreasing probability in accessibility of Fe source with cells. In detail, the reduced toxicity may be related to CD blocked the entrance or contact sites of FeAC for reactive oxygen species (ROS) generation, rather than the negatively charged surface property in CD-FeAC NPs.

In generally, anticancer agents delivered into cytosol and then, reached nucleus and directly effected in cell death [34]. So, tracking studies of CD-FeAC or FeAC NPs carriers into cells is important. As shown in Fig. 8c, FeAC NPs were existed into nucleus neighbors in HeLa cells abundantly, indicating CD-FeAC or FeAC NPs carriers can be successfully delivered target compounds into cytosol or nucleus in cells. In addition, nucleus site was negligibly damaged by FeAC NPs (Fig. 10). It is indicated that intact FeAC NPS have little cytotoxic effects for nucleus sites as well as an ability of transfection carriers [5355]. As therapeutic agents, thus, taking into consideration the practical concentration in this study, <500 μg/mL concentration of CD-FeAC is suitable to development of transfection reagent [29, 31]. In conclusion, CD-FeAC NPs can be a useful diagnostic and therapeutic agent in providing drug-delivery-carrier with simultaneous fluorescent bio-imaging and tracking functionalities, although the use of only FeAC NPs can be given up the simultaneous bio-imaging.

Conclusions

In summary, it has dealt that CD-FeAC NPs plays roles both in bio-imaging and as a drug-delivery carrier into human cells with little cytotoxicity as simple preparation and inexpensive sources. In close future, cinnamic acid derivatives as anticancer agent can be loaded for target cargos [56] by amide bonding between carboxylic groups in cinnamic acid derivatives and amine groups in FeAC NPs. Taking into consideration that, as noted above, the practical applied dosage of NPs is <500 μg/mL, CD-FeAC NPs is feasible for trapping of targeting drugs and proteins, because it shows cytotoxicity only to cancer cells. Conclusively, a CD-FeAC-based hybrid agent for imaging/selective anticancer platform in in vivo is currently in the planning stage.

Experimental section

Preparation of Fe-aminoclay (FeAC) NPs

Synthesis of FeAC NPs was carried out according to the method available in the literature [4042]. To a 500 mL-beaker solution containing 200 mL of ethanol, 8.4 g (31.08 mmol) of FeCl3∙6H2O salt (Sigma-Aldrich, USA) was added. After complete dissolution by 10 min magnetic stirring, 13.0 mL (58.73 mmol) of 3-aminopropyltriethoxysilane (APTES, Sigma-Aldrich, USA) was added to the ferric (Fe3+) ethanolic solution. In the course of mixing preparatory to the sol–gel reaction, brown slurry was formed. After 6 h equilibrium in production, the FeAC product was collected by 10 min 6000×g centrifugation. Then, after two ethanol-washing steps, the FeAC product was oven-dried at 50 °C for 1 day. Finally, preparatory to its use, the FeAC product was ground by pestle and mortar into a brown powder.

Preparation of carbon nanodots (CD) NPs

Organic waste solution (100 g of dried animal feces’s wastes per 10 L of three double distilled water) at 45 °C was treated by 40 kHz ultrasound for 90 min (Ultrasonics UC-05, Lab Companion, Korea) [33, 34]. It was then centrifuged at 2500 rpm for 5 min to remove large or agglomerated particles. The CD-containing supernatant was filtered twice through a 0.22 μm pore sized membrane to remove any remnant large particles, and finally dried at 45 °C.

Conjugation of CDs with FeAC (CD-FeAC) NPs

For conjugation on the surface of FeAC NPs, CD NPs (5.0 mg) were dissolved in 1.0 mL of phosphate-buffered saline (PBS buffer, 0.01 M and pH 7.2). To induce formation of labile intermediates on the CD surfaces, 1-ethyl-3-[3-dimethylaminopropyl]carbodiimide (EDC) (2.2 mg, 11.2 μmol) and N-hydroxysulfo-succinimide (sulfo-NHS) (2.44 mg, 11.2 μmol) were added, and the solution was stirred in darkness for 30 min at room temperature (RT). The thus-activated CD solution (200 μL) subsequently was mixed with 1 mL of FeAC solution (10.0 mg/mL in PBS buffer) and stirred for 3 h at RT. After completion of amide-bonded conjugation, the reacted sample was washed twice with PBS buffer and incubated in 0.1 M Tris–HCl buffer (pH 7.4) for 1 h at RT with shaking (250 rpm), to cap the unreacted sulfo-NHS functionality [31, 40]. The final product (10 mg/mL) was re-suspended in PBS buffer and stored at 4 °C until use.

Characterizations of FeAC, CD, and CD-FeAC NPs

Transmission electron microscopy (FE-TEM, Tecnai TF30 ST, FEI company, USA) images were examined. Samples had been prepared for TEM imaging by dropping a tiny pipetted amount on a carbon-coated Cu grid (300-mesh) and oven-drying at 50 °C. After 2 h bath-sonication of 2.0 mg/mL of FeAC and CD-FeAC NPs in PBS buffer and serum-free Roswell Park Memorial Institute (RPMI) media, respectively, their hydrodynamic diameter sizes and zeta potential values were measured using particle size analyzer (Zetasizer nano zs, Malvern, UK). For CD’s height and surface roughness, after 100 μL of the CD solution was placed on a silicon wafer and air-dried overnight, it was scanned by atomic force microscope (AFM, VEECO Instrument, USA). Disorder-induced D and first-order graphite G bands in samples were recorded by Raman microscopy system (NT-MDT NTEGRA Systems, USA), utilizing photoluminescence (PL) spectroscopy with a changeable UV transilluminator (DUT-260, Core Bio Systems, Korea) within the 290–410 nm excitation range. In order to check intrinsic property of quantum-like structure of CD, the quantum yield of pristine CD at 3 mg/mL was examined by spectrofluorometer (FP-8500, Jasco, Japan) following the literature [33].

The powder X-ray diffraction (XRD) patterns from 3° to 70° were obtained in 0.01 increments by micro-area X-ray diffractometry (D/MAX-2500, Rigaku, 40 kV and 300 mA) [42]. Additionally, the organic functional groups in the samples were determined according to the recorded Fourier transform infrared (FT-IR) spectra using FT-IR spectrophotometer (FT-IR 4100, Jasco, Japan) (composition: 90 wt% KBr plus 10 wt% sample in KBr-pellet mode).

Cytotoxic evaluations of by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) and neutral red (NR) assays

In order to examine cytotoxic effect of samples, two assays such as mitochondria—MTT and chromatin-based NR means were approached. The designed cell line systems were organized with cancer and normal cells (Table 1). The cytotoxicities of FeAC, CD, and CD-FeAC were measured using the EZ-Cytox Cytotoxicity assay Kit (MTT assay, Daeillab Service, South Korea) [29] and neutral red (NR) solution (Sigma-Aldrich, USA) [49]. Cells were seeded in 96-well plates to a concentration of 1 × 104 cells per well with RPMI culture media. After 24 h, the RPMI culture media was replaced with various concentrations of FeAC, CD, and CD-FeAC NPs in cell free culture media prepared by two-times washing, followed by immediate 2 h bath-sonication. Cells were incubated for 24, 48, and 72 h at 37 °C, after which they were washed twice with PBS buffer. Next, 100 μL of EZ-Cytox solution (MTT assay) or NR solution (NR assay) with 25 μg/mL in serum-free RPMI media was added and incubated for 3 h. The NR solution was washed with PBS buffer before adding 100 μL of lysis solution including acetic acid, ethanol, and water (1:50:49). The absorbance spectra of formazan (MTT assay) and the released NR (NR assay) were measured with the Infinite® 200 PRO (TECAN) at wavelengths of 450 and 542 nm, respectively. Cytotoxicity is expressed corresponding to MTT and NR release by untreated control cells.

Cellular uptake observation by confocal microscopy and cross-sectioned transmission electron microscopy (TEM)

HeLa cells and RAG cells were cultured on an 8-well chamber slide with a concentration of 2 × 104 cells per well and on a 6-well chamber slide with a concentration of 2 × 105 cells per well for confocal microscopy (LSM510 META NLO, Carl Zeiss, Germany) and field emission transmission electron microscopy (FE-TEM, Tecnai TF30 ST, FEI company, USA), respectively. After 24 h incubation, cells were exposed to FeAC or CD-FeAC NPs for a further 24 h, and were then washed several times with PBS buffer. After drying the PBS on the 8-well chamber slide, the slide was covered with cover glass using fluorescent mounting media (DaKo). The fluorescence of CD-FeAC NPs in cells was measured under LSM510 META non-linear optic (NLO) confocal microscopy at excitation 345 nm and emission 460 nm wavelength. Cells in the 6-well plate were collected using trypsin–EDTA and washed twice with PBS buffer. Another observation of the nucleus in the FeAC NPs-treated and absence of FeAC NPs with HeLa cells, 4′,6-diamidino-2-phenylindole (DAPI, 1 μg/mL, 10 min-incubation) staining protocol was followed and observed under confocal microscopy.

As for the cross-sectioned TEM imaging, it followed the procedure available in the literature [45]. In detail, the FeAC NPs-treated HeLa cells were fixed in a 2.5 % paraformaldehyde-glutaraldehyde mixture buffered with phosphate (0.01 M and pH 7.2) for 2 h, post-fixed in 1.0 % osmium tetroxide in the same buffer for 1 h, dehydrated in graded ethanol and propylene oxide (PPO), and embedded in Epon-812. Ultra-thin sections, cut by the ULTRACUT E (Leica, Austria) ultramicrotome, were stained with uranyl acetate and lead citrate and examined under CM 20 electron microscopy (Philips, Netherlands).