Chemical Composition of Nanoparticles Formed by Fetuin-A and Albumin in Supersaturated Ionic Solutions
To further study the chemical composition and characteristics of the protein-mineral nanoparticles formed in DMEM containing supersaturated amounts of precipitating ions, we used several types of spectroscopy. Earlier studies had shown that both NB-like specimens as well as serum calcium granules could be characterized as carbonate HAP ([2], [3]; see also the earlier study by Kajander and Ciftcioglu [5]), a type of mineral similar to that found in bones as well as in ectopic calcification [127], [128]. We sought therefore to verify whether nanoparticles containing either fetuin-A or albumin might reveal these same chemical characteristics so as to demonstrate that such simple protein-mineral complexes may indeed reproduce all the structural characteristics attributed to NB and calcium granules.
We first used energy-dispersive X-ray spectroscopy (EDX) in order to determine the elemental composition of the mineral phase associated with the protein-containing particles. As seen in Figure 12, the mineral nanoparticles seeded by BSF and/or HSA (labeled as “BSF-NLP,” “HSA-NLP,” and “BSF-HSA-NLP”) mainly showed peaks of calcium (Ca) and phosphorus (P) (Fig. 12A–C, compare their signals with those obtained for the control samples of CaCO3, Ca3(PO4)2, and HAP shown in Fig. 12J, K, and L, respectively).
10.1371/journal.pone.0008058.g012 Figure 12  Energy-dispersive X-ray spectroscopy of protein-mineral nanoparticles shows elemental compositions indistinguishable from those of calcium granules and NB.
Protein-mineral nanoparticles were obtained as in Fig. 9, from solutions containing BSF (A), HSA (B), or both (C), to which 0.3 mM each of CaCl2 and NaH2PO4 was added, followed by incubation in cell culture conditions for 1 month. EDX spectra were also obtained for calcium granules prepared by adding either CaCl2 (D), NaH2PO4 (E), or a combination of both (F) to FBS, followed by overnight incubation, centrifugation, and washing, as described in the  Materials and Methods . NB were cultured from 10% HS (G, “HS-NB”) or from 10% FBS (H and I, corresponding to “Nanons” and “DSM 5820”, respectively). In these specimens, major peaks of carbon, oxygen, calcium, and phosphorus were noted, concordant with the presence of a calcium phosphate mineral containing carbonate. The three controls CaCO3 (J), Ca3(PO4)2 (K), and HAP (L), diluted and washed in double-distilled water, were shown for comparison. The following Ca/P ratios were obtained: (A) 1.37; (B) 1.53; (C) 1.6; (D) 1.32; (E) 1.48; (F) 1.14; (G) 1.48; (H) 1.4; (I) 1.27; (K) 2.54; and (L) 1.48. Phosphorus was not detected in the CaCO3 samples shown in (J).   Peaks corresponding to carbon (C) and oxygen (O) were also observed in the protein-mineral particles, indicative of the presence of carbonate; in fact, in the experiments shown in Figure 12A–C, the oxygen:carbon ratios obtained for these three samples varied between 2.41 to 3.23, close to the ratio of 3 expected for carbonate. The intensities of the C and O peaks also varied greatly with the sample tested as well as with the aging within the same sample. Thus, consistently, we found that fresh samples of BSF-NLP or HSA-NLP contained more carbonate, while aged samples (after 1 week) contained predominantly calcium and phosphate—results that are consistent with earlier findings made on NLP and granules formed in the presence of whole serum [2], [3].
Occasionally, elements like sodium (Na), magnesium (Mg), and chlorine (Cl) were also detected in low amounts (Fig. 12A–C). As such, they can probably be interpreted as minor constituents of the protein-mineral particles, or, alternatively, as unrelated contaminants.
Overall, the EDX spectra obtained for the protein-mineral particles (Fig. 12A–C) were indistinguishable from those of calcium granules prepared in FBS (Fig. 12D–F) or the various NB controls cultured from either 10% HS or 10% FBS (Fig. 12G–I). That is, the main elements present in these samples also consisted of calcium and phosphorus, and to a lesser extent carbonate (Fig 12D–I). The calcium:phosphorus (Ca:P) ratios obtained for the protein-mineral particles varied between 1.37 to 1.60 (Fig. 12A–C) and were similar to the ones found for calcium granules and NB (shown in Fig. 12D–I, with Ca:P ratios varying between 1.14 to 1.48; see also refs. [3], [5]). Carbon and oxygen peaks (e.g. carbonate) as well as minor elements corresponding to copper (Cu) and aluminum (Al) were also noted at times in these samples (Fig. 12D–I).
As additional controls, we also included various chemically defined and commercially available compounds that were incubated and washed in DMEM (Fig. 12J–L). This treatment resulted in a slight widening of the signal peaks as compared to those obtained with the compounds dissolved in water (data not shown; compare with data shown in refs. [2], [3]), but the peak signal strengths were otherwise largely comparable.
Next, we used Fourier-transformed infrared spectroscopy (FTIR) in order to identify the main functional groups found in the protein-mineral particles, with a view toward assessing the similarities between these particles and the previously characterized calcium granules and NB. For the particles prepared in supersaturated solutions containing BSF and/or HSA, the FTIR spectra showed mainly peaks corresponding to phosphate and carbonate (Fig. 13A–C, compare the signals obtained with those seen in the controls of CaCO3, Ca3(PO4)2, and HAP shown in Fig, 13J, K, and L, respectively). Phosphate peaks were observed at wavelengths of 575 cm−1, 605 cm−1, 960 cm−1, and 1,000–1,150 cm−1 (Fig. 13A–C; see also refs. [129], [130]). Carbonate groups, on the other hand, were detected at 875 cm−1 and near 1,400–1,430 cm−1 (Fig. 13A–C; see also refs. [131], [132]). Some peaks seen in the controls were rarely observed in the protein-mineral particles. For example, note the presence of a carbonate peak at 650 cm−1 in the CaCO3 control shown in Figure 13J that was absent in the protein-mineral particles shown in Figure 13A–C.
10.1371/journal.pone.0008058.g013 Figure 13  Fourier-transformed infrared spectroscopy of protein-mineral nanoparticles reveals the presence of both carbonate and phosphate.
Protein-mineral nanoparticles were obtained as described in Fig. 9, by diluting BSF (A), HSA (B), or both proteins (C) into DMEM, then adding 0.3 mM each of CaCl2 and NaH2PO4, and incubating the solutions in cell culture conditions for 1 month. Calcium granules were prepared by adding either CaCl2 (D) or NaH2PO4 (E) into FBS, or a combination of both CaCl2 and NaH2PO4 (F) into HS, as described in the  Materials and Methods . NB were cultured from 10% HS (G, “HS-NB”) or 10% FBS (H and I, corresponding to “Nanons” and “DSM 5820”, respectively). The FTIR spectra of the protein-mineral nanoparticles revealed peaks similar to both calcium granules and NB as shown by the presence of phosphate peaks at 575 cm−1, 605 cm−1, 960 cm−1, and 1,000–1,150 cm−1 as well as carbonate peaks at 875 cm−1 and 1,400–1,430 cm−1. Some of the peaks corresponding to phosphate or carbonate were not detected in a few calcium granule samples such as the one shown in (D). In the various nanoparticle samples presented here, peaks corresponding to amide I, II, and III at 1,660 cm−1, 1,550 cm−1, and 1,250 cm−1, respectively, were observed and were attributed to the presence of serum proteins. Spectra for the controls CaCO3 (J), Ca3(PO4)2 (K), and HAP (L), diluted and washed in double-distilled water, were included as controls. Residual water (H2O) was observed at 1,650 cm−1 in some controls prepared in the absence of proteins (K and L); this peak could also have contributed to the intensity of the amide I peak seen at 1,660 cm−1 in the other samples shown.   In addition to the phosphate and carbonate signals, a peak corresponding presumably to the chemical bonds found in proteins were also detected in the protein-mineral particles. Thus, a peak corresponding to amide I was noticed at 1,660 cm−1 (Fig. 13A–C; see refs. [127], [133]). However, we also observed that the same peak corresponding to amide I at 1,660 cm−1 might also overlap with a peak corresponding to water that was found at 1,650 cm−1 in the control specimens prepared in the absence of proteins (Fig. 13K and L; see also refs. [134], [135]). We noticed that the peaks corresponding to amide II and III at 1,550 cm−1 and 1,250 cm−1, respectively, were usually not present in the protein-mineral particles, possibly due to the relatively low amount of proteins in these samples (Fig. 13A–C, see also refs. [127], [133]).
The FTIR spectra obtained for the protein-mineral particles (Fig. 13A–C) were similar to those of calcium granules prepared in serum (Fig. 13D–F) or the NB controls cultured from either 10% HS (Fig. 13G) or 10% FBS (Fig. 13H and I). That is, the spectra obtained for these samples consistently showed major peaks of phosphate and carbonate along with peaks corresponding to amide bonds found in proteins. Still, minor differences were noted between these samples. For instance, some samples of calcium granules such as the one shown in Figure 13D appeared to lack peaks of phosphate at 575 cm−1, 605 cm−1, and 960 cm−1 and a peak corresponding to carbonate also went missing at 875 cm−1. The phosphate peak observed for this particular sample of calcium granules at 1,000–1,150 cm−1 was also of lower intensity (Fig. 13D). In addition, both the calcium granules and NB specimens generally displayed a more prominent amide I peak as compared to the protein-mineral particles (compare the intensity of the amide I peak at 1,660 cm−1 in A–C with the same peak shown in D–I). This difference in the intensity of the amide I peak may simply reflect the fact that the calcium granules or NB were either prepared or seeded with whole serum while the protein-mineral particles were assembled in lower amounts of purified proteins. Consistent with this interpretation, the calcium granules and NB specimens displayed additional peaks corresponding respectively to amide II at 1,550 cm−1 (Fig. 13D–F, G, and I) and amide III at 1,250 cm−1 (Fig. 13I) that were not readily noticed with the protein-mineral samples.
For the control compounds, we also noted the presence of carbonate peaks associated with the calcium phosphate sample (Fig. 13K, where carbonate peaks can be seen at 875 cm−1 and 1,400–1,430 cm−1). These peaks might be due to the prolonged contact with ambient CO2 from air. In spite of these differences, it can be concluded that the protein-mineral particles studied here by FTIR can largely mimic both calcium granules and NB preparations, indicating close if not identical structural relationships between all these three types of entities.
We confirmed the presence of the major functional groups seen here with FTIR analysis by performing micro-Raman spectroscopy. As can be seen from the micro-Raman spectra shown in Figure 14A–C, the mineral particles prepared in supersaturated solutions containing fetuin-A and/or albumin revealed mainly peaks of phosphate at 440 cm−1, 581 cm−1, and 962 cm−1 as well as carbonate peaks at 1,080 cm−1. However, unlike the FTIR spectra and compared to the signals obtained for the control compounds (Fig. 14J–L), there was a clear dampening of signals seen with all the specimens of protein-NLP, calcium granules, and NB—a result that could be due to the presence (and interference) of proteins in the particle scaffold, a possibility that was pointed out earlier [3]. That is, the green laser used to study the Raman scattering effect makes the proteins fluoresce and such fluorescence usually dampens the signal that can be detected from the sample, thereby producing a high background and a low signal-to-noise ratio [127]. Possibly for this same reason, the phosphate and carbonate peaks were often found absent or markedly reduced from the micro-Raman spectra obtained from the various protein-NLP samples, as illustrated by Figure 14B and C, where there is the absence of the carbonate peak at 1,080 cm−1 (Fig. 14B) or of any substantial peak (Fig. 14C). Furthermore, some peaks that were present in the controls were rarely seen in the protein-NLP samples (BSF-NLP, HSA-NLP or BSF-HSA-NLP) such as the carbonate peaks seen in at 280 cm−1 and 712 cm−1 in CaCO3 (Fig. 14J) or the phosphate peak seen at 361 cm−1 in the Ca3(PO4)2 control (Fig. 14K). As seen earlier during the FTIR analysis, we also noticed that the Ca3(PO4)2 control contained small impurities of carbonate that may have originated from prolonged contact with CO2 from air (Fig. 14K, showing a small carbonate peak seen at 1,080 cm−1). In the case of calcium granules and NB specimens prepared from serum, these samples also produced variable phosphate and carbonate signals depending on the serum lots used (Fig. 9D–I, see also ref. [3]). For instance, calcium granules prepared in FBS, following the addition of calcium, produced small peaks of HPO4 2− and carbonate that were rarely seen in the controls (Fig. 14D, see refs. [136], [137] where the peaks found at 1,002 cm−1 and 1,150 cm−1 were attributed respectively to HPO4 2− and carbonate). In addition, a few NB specimens like “DSM 5820” did not produce any noticeable signal (Fig. 14G).
10.1371/journal.pone.0008058.g014 Figure 14  Micro-Raman spectroscopy of protein-mineral nanoparticles shows chemical compositions similar to those of calcium granules and NB.
Protein-mineral nanoparticles were prepared as in Fig. 9, by adding 0.3 mM each of CaCl2 and NaH2PO4 to DMEM containing BSF (A), HSA (B), or both proteins (C), followed by incubation in cell culture conditions for 1 month and processing for micro-Raman spectroscopy. Calcium granules were obtained by adding CaCl2 (D), NaH2PO4 (E), or both (F) to FBS, followed by overnight incubation and preparation for micro-Raman spectroscopy as described in the  Materials and Methods . Micro-Raman spectra were also acquired for NB that were initially cultured from 10% HS (G, “HS-NB”) or 10% FBS (H and I, “Nanons” and “DSM 5820”, respectively). These nanoparticle samples showed phosphate groups at 361 cm−1, 440 cm−1, 581 cm−1, 962 cm−1, 1,002 cm−1 (HPO4 2−), and 1,048 cm−1 and carbonate moieties at 280 cm−1, 712 cm−1, 1,080 cm−1, and 1,150 cm−1. The protein-mineral nanoparticles mainly showed peaks of phosphate and lower peaks of carbonate (A–C) while the calcium granules (D–F) and NB (G–I) samples showed carbonate and phosphate peaks of variable intensities. The three controls CaCO3 (J), Ca3(PO4)2 (K), and HAP (L), diluted and washed in double-distilled water, were included for comparison.   In spite of a significant dampening of signals seen here and together with the EDX and FTIR analyses, it can still be inferred that the micro-Raman data show that the nanoparticles prepared in supersaturated solutions containing BSF and/or HSA have a chemical composition largely similar to that of both calcium granules and NB.
To complete the chemical analysis of the protein-mineral nanoparticles, we used powder X-ray diffraction (XRD) spectroscopy to study the nature of the crystals found in these samples. For the protein-nanoparticles obtained from supersaturated solutions (BSF-NLP, HSA-NLP and BSF-HSA-NLP), no peak was obtained from the XRD analysis (Fig. 15A–C), indicating a predominance of amorphous phases here. Precipitation of calcium phosphate is usually described to proceed from an amorphous precursor phase which slowly transforms into various calcium phosphate compounds of increased crystalline complexity, ending with the crystalline apatite, which is the more thermodynamically stable end-product [138]. This transition does in fact also occur with protein-NLP, and, upon prolonged incubation or by using lower protein-to-calcium-phosphate ratios, several protein-NLP samples have been shown to progress to crystalline patterns, including the display of classic Ca10 apatite signals (Fig. 15D–F). Thus, in the case of these particles, the intensity of the peaks could be modulated by the amount of calcium, phosphate, and proteins added as well as the length of incubation. In general, the spectra will shift more rapidly toward the crystalline phases in the presence of higher amounts of precipitating ions, longer incubations, and lower amounts of proteins. It is likely that the presence of proteins in association with the mineral phase promotes the formation of an amorphous mineral phase that either blocks or slows the transition to apatite, a concept that has been advocated in the past by Mann and other researchers [58].
10.1371/journal.pone.0008058.g015 Figure 15  Powder X-ray diffraction spectra of protein-mineral nanoparticles reveal both amorphous and crystalline patterns.
Protein-mineral nanoparticles were obtained as described in Fig. 9, by diluting the proteins, separately (A and B) or together (C) into DMEM, followed by addition of the precipitating reagents CaCl2 and NaH2PO4 each to a final concentration of 0.3 mM and incubating the solutions in cell culture conditions for 1 month. The XRD spectra obtained for the protein-mineral nanoparticles obtained after 1 week of incubation represented mainly amorphous patterns as seen by the absence of diffraction peaks (A–C). Longer incubation of 1 month produced protein-mineral nanoparticles with crystalline peaks corresponding to HAP crystals (D–F, Ca10(PO4)6(OH)2). XRD spectra were also obtained for calcium granules that were prepared by adding either CaCl2 (G) or NaH2PO4 (H) into FBS or both CaCl2 and NaH2PO4 into HS (I), followed by sample preparation as described in the  Materials and Methods . Peaks corresponding to Ca10(PO4)6(OH)2 were obtained for both calcium granules prepared in FBS (G and H) while the ones prepared in HS usually gave amorphous patterns (I). XRD spectra showing the presence of HAP crystals (J), a calcium phosphate compound (K, Ca5(PO4)3OH), or an amorphous pattern (L) were also acquired for NB cultured in 10% HS for 1 month (J, “HS-NB”) or in 10% FBS for 1 month (K, “Nanons”) or 1 week (L, “DSM 5820”). Commercial grades of CaCO3 (M), Ca3(PO4)2 (N), and HAP (O), used as controls, were diluted and washed in double-distilled water.   In support of this notion, the calcium granules prepared in serum in the presence of high concentrations of calcium and phosphate ions also showed variable levels of crystallinity depending on the amount of serum present as well as the amounts of calcium and phosphate added (Fig. 15G–I). Of particular interest is the fact that calcium granules prepared from FBS often displayed crystalline peaks of Ca10 apatite, as indicated by the prominent peak at 31.8 degrees on the 2-θ scale (Fig. 15G and H), comparable to that seen with the commercially available HAP powder used for comparison (Fig. 15O), while the calcium granules prepared in HS showed a similar propensity to remain amorphous (Fig. 15I; see also ref. [3] for similar data). This may be attributed to higher protein concentrations associated with HS (averaging 60 mg/ml) compared to FBS (32 mg/ml; see  Materials and Methods ). While the FTIR and micro-Raman analyses suggest that carbonate should be present in the mineral phase of the particles, the signal for carbonate was not detected in our specimens (Fig. 15A–C and data not shown). This result might be due to the absence of this particular diffraction crystalline plane in the samples examined, or alternatively due to the predominance of the HAP signal.
For comparison, it should be noted that the various NB specimens tested also displayed variable amorphous and crystalline patterns (Fig. 15J–L). For example, in one experiment, NB that had been cultured from 10% HS showed complete conversion to the Ca10 apatite compound under the conditions tested (Fig. 15J) while other samples of “Nanons” and “DSM 5820” showed either low peaks of calcium phosphate crystals (Ca5(PO4)3OH in Fig. 15K) or only amorphous signals (Fig. 15L). The commercially available controls of CaCO3, Ca3(PO4)2, and HAP that had been incubated in DMEM were also included for comparison (Fig. 14M–O).
The presence of both amorphous and crystalline patterns in all the samples seen, including each of the protein-mineral complexes examined, reveals the possibility of a continuous progression from amorphous to crystalline dispositions that in turn can be modulated by the types and amounts of proteins and minerals present. In addition, both the presence of excess proteins (either whole serum or purified proteins like fetuin-A or albumin) and certain competing ions (magnesium or carbonate, data not shown; see also ref. [2]) are now known to result in smaller particle sizes. In fact, the XRD analysis of small mineral nanoparticles is often complicated by the fact that only large crystals over a certain dimension tend to diffract X-rays [139], [140]. Thus, it is not entirely clear whether the nanoparticles studied here are truly amorphous during at least portions of their development or perhaps comprise of small crystals that fail to diffract X-rays.