Influence of Biocorona Formation on the Transformation and Dissolution of Cobalt Nanoparticles under Physiological Conditions.

Cobalt (Co) nanoparticles (NPs) are produced in different applications and unintentionally generated at several occupational and traffic settings. Their diffuse dispersion may lead to interactions with humans and aquatic organisms via different exposure routes that include their transformation/dissolution in biological media. This paper has investigated the particle stability and reactivity of Co NPs (dispersed by sonication prior to exposure) interacting with selected individual biomolecules (amino acids, polypeptides, and proteins) in phosphate-buffered saline (PBS). No or minor adsorption of amino acids (glutamine, glutamic acid, lysine, and cysteine) was observed on the Co NPs, independent of the functional group and charge. Instead, phosphate adsorption resulted in the formation of a surface layer (a corona) of Co phosphate. The adsorption of larger biomolecules (polyglutamic acid, polylysine, lysozyme, and mucin) was evident in parallel with the formation of Co phosphate. The dissolution of the Co NPs was rapid as 35-55% of the particle mass was dissolved within the first hour of exposure. The larger biomolecules suppressed the dissolution initially compared to exposure in PBS only, whereas the dissolution was essentially unaffected by the presence of amino acids, with cysteine as an exception. The formation of Co phosphate on the NP surface reduced the protective properties of the surface oxide of the Co NPs, as seen from the increased levels of the released Co when compared with the nonphosphate-containing saline. The results underline the diversity of possible outcomes with respect to surface characteristics and dissolution of Co NPs in biological media and emphasize the importance of surface interactions with phosphate on the NP characteristics and reactivity.

Introduction

Co metal is widely used in high wear-resistant alloys and as a binder in hard metals because of its excellent wear resistance, magnetic and catalytic properties, as well as strength at high temperatures. Nanoparticles (NPs) of Co and Co oxides are further used as pigments, as catalysts, in magnetic fluids, and as contrast agents for medical imaging.[1] Their use and potential dispersion via, for example, wear processes at occupational settings, from implant materials, or tire studs at traffic settings may result in adverse effects on humans and the environment. Repeated exposure to Co ion concentrations exceeding 20 μg/L is reported to result in risks for systemic toxicity.[2] Exposure to Co can furthermore induce asthma and acute illness including fever, anorexia, malaise, breathing difficulties, and interstitial pneumonitis.[3] However, Co is also a vital component of vitamin B12, an essential compound for the well-being of humans, plants, and animals.[4]

Nanosized particles need special attention as they, in many cases, pose a greater risk than larger sized particles on human health and the environment.[5,6] Tires with studs made of tungsten carbide cobalt (WC-Co) are frequently used during the winter season in the Northern countries. As a consequence, wear particles of W and Co, some of nanometer size, have been observed in road dust.[7−10] Workers in hard metal industries have further experienced negative health effects from airborne Co,[11] and WC-Co particles have shown higher reactivity and generation of free radicals (indicative of a higher toxic potency) compared with WC particles and also release of Co ions.[12−15]

For risk assessment, it is important to understand the dynamics of reactions taking place at the outermost surface of NPs as it influences their mobility, stability, and dissolution pattern.[16−22] Surface adsorption of humic and fulvic acids, amino acids, and proteins[21] to metal NP surfaces has been shown to take place instantaneously upon environmental entry, forming an eco- or biocorona.[23] This corona formation has further shown to influence NP dissolution, agglomeration, and cell interactions.[21,23,24]

The adsorption of biomolecules (the biocorona) influences the corrosion properties, characteristics, and reactivity of the surfaces and wear particles of CoCrMo alloys in hip implants.[25] Literature findings furthermore report that proteins (albumin and fibrinogen) enhance (catalyze) both the release of Co and the corrosion process in saline (NaCl).[26] The surface oxide composition is important as minor dissolution has been observed in cell media for Co3O4 NPs, whereas complete dissolution was evident within 24 h for Co metal NPs (surface oxide mainly composed of CoO).[27] The effects of biomolecule adsorption, forming biocoronas, on the dissolution and reactivity of Co NPs are however poorly understood. This lack of knowledge hampers risk assessments and accurate interpretations of toxicological findings, both from human and environmental perspectives.

This paper focuses on the interaction between Co NPs and different biomolecules of varying structure and properties under simulated physiological conditions with the aims to: (i) deduce the trends in the adsorption of biomolecules of different characteristics, (ii) explore if the formation of a biocorona prevents the agglomeration of Co NPs in solution and, hence, influence their stability and mobility, and (iii) investigate the influence of biomolecule characteristics and biocorona formation on the dissolution of Co NPs.

Phosphate-buffered saline (PBS) is used in this study to mimic the ionic strength (ca. 170 mM, mainly from NaCl) and pH of blood (pH 7.4). The selection of amino acids (lysine, glutamine, glutamic acid, and cysteine) is based on their different characteristics in terms of net charge and functional groups at the same pH. Lysine is positively charged and has an amine group as the functional group on the side chain, glutamine is uncharged with an amide as the functional group, glutamic acid is negatively charged with a carboxylate functional group, and cysteine is uncharged with thiol (SH) as the functional group that under certain conditions forms strong complexes with both metallic surfaces and metal ions.[29,30] Two oppositely charged polypeptides were selected to investigate the effect of biomolecule size and number of charged entities compared with the selected amino acids: positively charged poly-l-lysine and negatively charged poly-l-glutamic acid. A comparative study using polypeptides and amino acids provides insight on how larger biomolecules with the same functional groups affect the interaction with Co NPs. Studies on the natural proteins lysozyme and mucin, for instance, of relevance for the inhalation route, were performed in parallel to investigate how the differences in protein structure and flexibility affect the interaction with Co NPs. Mucin is a flexible, high-molecular-weight (several MDa) glycoprotein produced in epithelial tissues of the mucous membrane that lines most of the cavities in the human body, for example, the digestive and respiratory tracts. Mucin is also important for lubrication and may act as a chemical barrier that hinders the corrosion process of metallic surfaces.[31,32] Lysozyme is a small compact globular protein that is present in saliva, tear fluids, and mucus. Mucin and lysozyme are oppositely charged at pH 7.4 (mucin–net negatively charged; lysozyme–net positively charged). Previous studies show that the net charge of the protein can influence the metal release process of metallic surfaces.[33−35] These studies revealed that metal surface interactions of negatively charged protein bovine serum albumin (BSA) significantly enhanced the extent of metal release from both stainless steel and silver, whereas the interaction with positively charged proteins such as lysozyme predominantly resulted in a minor increase of the amount of released metal from stainless steel and a significant reduction of released silver. Co has a high affinity to phosphate and readily forms insoluble complexes.[36] It is hence possible that Co preferentially forms such complexes with phosphate rather than with the biomolecules of this study. These aspects are important to address because Co phosphate complexes present on the surface, and in solution, most likely will result in different surface characteristics of the Co NPs (e.g., charge and barrier effects on dissolution) compared with the corresponding interactions with different biomolecules.

In this study, attenuated total reflection Fourier transform infrared (ATR–FTIR) spectroscopy was employed to probe the adsorption of biomolecules to a layer of Co NPs deposited onto the ATR crystal in situ. The released amount of Co ions in solution was determined for different particle loadings and biomolecule concentrations using atomic absorption spectroscopy (AAS). Solution speciation estimations were performed using the joint expert speciation system (JESS) to deduce the trends of Co ion complexation to amino acids and phosphate. The zeta potential (apparent surface charge) was estimated for the Co NPs in PBS, 10 mM NaCl, and PBS buffer without chloride. Photon cross-correlation spectroscopy (PCCS) was employed to assess the particle stability and changes in particle size. The particle morphology and composition (native dry particles and particles after exposure in PBS and in PBS-containing biomolecules) were assessed by means of transmission electron microscopy (TEM) combined with energy-dispersive spectroscopy (EDS). The composition of the outermost surface was determined by X-ray photoelectron spectroscopy (XPS).

Results and Discussion

Particle Characteristics under Dry Conditions

The TEM micrographs of unexposed Co NPs under dry conditions show agglomerated particles of polyhedral structure typically sized between 10 and 50 nm (Figure ). The Brunauer–Emmett–Teller (BET) surface area was determined to be 10.7 ± 0.2 m2/g.[27]

Figure 1

Unexposed Co NPs. (A) TEM micrographs; the scale bar equals 50 nm. (B) XRD diffraction pattern, (C) XPS spectrum of Co 2p3/2 with fitted Gaussian peaks and cumulative fit added in addition to raw data (dashed curve), and (D) XPS spectrum of O 1s with fitted peaks and cumulative fit added in addition to raw data (dashed curve).

Compositional analyses by means of X-ray diffraction (XRD) revealed CoO (main diffraction peaks at 42.5, 36.6, 61.7, 74.0, and 77.9—2θ)[37,38] to be the predominating crystalline oxide. Its presence on the surface was supported by high-resolution XPS measurements (Co 2p3/2: 780.1, 782.1, 785.5, and 786.5 eV; O 1s: 529.4 eV),[27,39] though phase identification using this technique is only indicative and not conclusive. The XPS findings further indicate the possible presence of Co3O4 (Co 2p3/2 peaks at 779.7, 781.5, 783.8, and 789.8 eV), an observation supported by recent analyses by the authors using Raman spectroscopy.[39] Its presence is not surprising as the surface of CoO has been shown to readily oxidize to Co3O4 in air.[39,40] As no diffraction peaks related to Co3O4 were observed, these results indicate its presence in an amorphous state and/or in a minor quantity.

Particle Characteristics after Sonication

Metal NPs have in general strong van der Waals forces, which make their driving force for agglomeration relatively high. Sonication was therefore required to prepare the particle dispersions for testing. However, as previous findings have shown this step to largely influence dissolution and the surface characteristics of metallic particles,[41] such effects were investigated in this study. This knowledge is essential to properly connect NP characteristics prior to exposure with the observed transformations after exposure of the same NPs. Co NPs in both non-sonicated and sonicated solutions showed bands in the ATR–FTIR spectra (Figure ), at 650–680, 550–600, and 380 cm–1, all attributed to Co3O4.[37,42] The band at 675 cm–1 can also be attributed to CoO (Tang et al. 2008). The broad plateau between 400 and 520 cm–1 includes several bands from both Co3O4 (e.g., 519, 482 cm–1) and CoO (e.g., 455 cm–1).[37] The band at 620 cm–1, attributed to Co3O4 (Tang, Wang et al. 2008), was visible prior to sonication but disappeared after the sonication steps, which may indicate a change of the Co3O4 (spinel) structure.

Figure 2

ATR–FTIR spectra for non-sonicated and sonicated Co NPs in ultrapure water and ethanol, respectively.

Changes in the outermost surface oxide composition of the Co NPs as a result of sonication were also observed by means of XPS. The binding energy of the main Co 2p3/2 peak attributed to Co3O4 shifted from 779.7 to 781.8 eV, and both the main Co 2p3/2 peak at 780.1 eV and the O 1s peak attributed to CoO at 529.4 eV disappeared. This corroborates the results of ATR–FTIR spectroscopy that showed sonication to change the composition and possibly also the thickness of the surface oxide.

Sonication further influenced the dissolution of the Co NPs. 8% of the total Co mass was dissolved after sonication in the stock solution. As a result, the administered dose contained 92% NPs and 8% dissolved ions, findings in line with previous observations on differences between the nominal and the administrated dose of metal NPs.[36]

Changes in the Surface Characteristics of Co NPs as a Result of Interactions with PBS and Biomolecules

The absorption spectra, based on the ATR–FTIR measurements of Co NPs in PBS in the presence and absence of amino acids, are presented in Figure together with the background spectra for PBS. The concentration of biomolecules used is relatively close to what is expected under physiological conditions.[43] The PBS solution showed (Figure B) peaks at 1077 and 989 cm–1, assigned to the monoprotonated (HPO42–) species,[44] and a peak at 1640 cm–1 related to the bending mode of H2O.[45]

Figure 3

ATR–FTIR spectra collected after 5 min exposure of Co NPs in (A) saline and in PBS containing 146 mg/L lysine and (B) in PBS, in PBS containing 146 mg/L lysine, glutamine, glutamic acid, or cysteine, and pure PBS solution with MQ water as the background. The PBS solution spectra are added for comparison (no Co NPs present). All spectra have been offset for clarity.

In the presence of Co NPs, all conditions showed bands at 1110 and 960 cm–1 that are attributed to the stretching vibration of P–O bonds.[36] The background spectra for each solution (without Co NPs) were subtracted in each case (Supporting Information, Figure S1). A shift between 25 and 40 cm–1 compared with the P–O bands in PBS was evident for all cases and attributed to the formation of Co phosphate surface complexes in which Co ions influence the vibration of the original HPO42−.[46] No evident peaks that could be attributed to any amino acid interactions with the Co NPs were observed, such as δas(NH3+) at 1626–1629 cm–1 and δs(NH3+) at 1526–1527 cm–1 for lysine, or 1668–1687 cm–1 for ν(C=O) of glutamine (see Figures and S1), though some overlap with the broad H2O band at 1645 cm–1 cannot be excluded.[47] From the results, it is evident that the surface interactions are dominated by the formation of Co phosphates and that any adsorption of the small-sized amino acids is minor or not possible to observe with the ATR–FTIR technique. Measurements in saline and water showed the adsorption of the amino acids (see Supporting Information, Figure S2). A minor increase in peak intensities of the phosphate bands was observed with time (data not shown), indicative of a nearly complete adsorption already within a few minutes. The dominance of phosphate over amino acids in terms of surface affinity is corroborated by equilibrium calculations that show a higher affinity between Co and phosphate ions compared with Co ions and lysine or glutamine molecules in solution (Table ).

Table 1

JESS Chemical Equilibrium Calculations (pH 7.4, 37 °C) of the Chemical Speciation of Co(II) Ions (0.1, 1, 10 mg/L) in PBS and in PBS Containing Different Amino Acids (146 mg/L—1 mM)a

solution	0.1 mg/L Co2+	1 mg/L Co2+	10 mg/L Co2+
PBS	91% CoHPO4, 4% CoCl+, 4% Co2+	91% CoHPO4, 4% CoCl+, 4% Co2+	76% Co3O4(s), 22% CoHPO4, 1% CoCl+, 1% Co2+
lysine in PBS,	69% CoHPO4, 24% CoLys+, 3% CoCl+, 3% Co2+	69% CoHPO4, 24% CoLys+, 3% CoCl+, 3% Co2+	68% Co3O4(s), 22% CoHPO4, 8% CoLys+, 1% CoCl+, 1% Co2+
functional group: amine (−NH3+),
pKa: 10.67
glutamine in PBS,	88% CoHPO4, 4% CoCl+, 4% Co2+, 4% CoGln+	87% CoHPO4, 4% CoGln+, 4% CoCl+, 4% Co2	76% Co3O4(s), 22% CoHPO4, 1% CoGln+, 1% CoCl+, 1% Co2+
functional group: amide (−CONH2),
uncharged
glutamic acid in PBS,	86% CoHPO4, 5% CoGlu, 4% CoCl+, 4% Co2+	86% CoHPO4, 5% CoGlu, 4% CoCl+, 4% Co2+	75% Co3O4(s), 22% CoHPO4, 1% CoGlu, 1% CoCl+, 1% Co2+
functional group: carboxylate (−COO–),
pKa: 4.15
cysteine in PBS,	76% CoCys, 17% CoHPO4, 5% CoCys22–, 1% CoCl+	72% CoCys, 16% CoHPO4, 5% CoCys22–, 5% Co2Cys32–	54% CoCys, 23% Co2Cys32–, 15% CoHPO4, 3% CoCys22–, 3% Co3Cys42–
functional group: thiol (−SH),
pKa: 8.14

Only components with a fraction of Co ions exceeding 1% are included in the table.

According to the equilibrium calculations in Table , Co ions in solution have a higher affinity to cysteine (thiol functional group) compared to phosphate. However, no or minor adsorption of cysteine was possible to be observed by means of ATR–FTIR spectroscopy. This indicates a difference in affinity between Co ions and cysteine in solution compared with its affinity to adsorb at the oxidized Co NP surface. From an electrostatic (Coulomb) perspective, a positively charged amino acid is expected to adsorb onto the negatively charged Co NP surface (see zeta potential measurements below), whereas an uncharged amino acid would not adsorb. However, the high ionic strength of the PBS solution results in very short-range electrostatic interactions, and the driving force may hence not be strong enough for any exchange between the adsorbed phosphate species at the surface and any amino acid molecules in solution.

Zeta potential measurements were conducted to investigate how the adsorption of phosphate affected the surface charge of the Co NPs (Figure ). The most negative potential (p < 0.05, Student’s t test) was measured in the phosphate buffer (0.02 mM phosphate). All measurements in PBS resulted in very broad potential distributions, from negative to positive potentials, making Coulomb interactions between negatively charged phosphate groups and some positively charged Co NPs plausible (Figure S3). The results clearly show that the adsorption of multivalent phosphate anions (Figure ) generates a more negatively charged surface (Figure ).

Figure 4

Average zeta potentials of Co NPs (10 mg/L) at pH 7.4 in 10 mM NaCl (considering a higher concentration of NaCl corresponding to a high ionic strength that causes a screening effect in the zeta potential measurements, only 10 mM NaCl is used here), PBS (>150 mM ionic strength), and phosphate buffer (20 mM). The error bars represent the standard deviation from at least three independent measurements. The stars indicate statistically significant differences (p < 0.05, Student’s t test). Note—broad zeta potential distributions (see Supporting Information, Figure S3).

ATR–FTIR spectra are presented in Figure for Co NPs after interactions with polylysine, polyglutamic acid, lysozyme, and mucin in PBS for different time periods. From the spectra, it is evident that the amide bands (amide I and II) from the adsorbed biomolecules dominate the spectral region between 1200 and 1800 cm–1, and the band present at 1625 cm–1, attributed to the water bending mode, grows in the negative direction with time. Curve fitting using Gaussian functions (details given in Figure S5) was used to resolve the negative spectral band in addition to the bands corresponding to the adsorbed proteins and polypeptides (amide bands).

Figure 5

ATR–FTIR spectra for the Co NP layer in (A) 146 mg/L polylysine in PBS, 146 mg/L polyglutamic acid in PBS; (B) 146 mg/L lysozyme in PBS, 146 mg/L mucin in PBS.

The observed vibrational peak positions for Co NPs exposed to different polypeptide and protein solutions are summarized in Table .

Table 2

Peak Positions of Amide and Phosphate Bands (cm–1) Observed in the ATR–FTIR Spectra of Co NPs (10 mg/L) Exposed in PBS Containing Polypeptides or Proteins (146 mg/L)

Co NP + polylysine (cm–1)	Co NPs + polyglutamic acid (cm–1)	Co NPs + lysozyme (cm–1)	Co NPs + mucin (cm–1)	assignment
1660	1670	1675	1675	amide I
1645	1650	1654	1650	C=O stretch[48]
1545	1555	1542	1545	amide II
1515	1525	1520	1525	N–H bend, N–C stretch[48]
≈1625	≈1625	≈1625	≈1625	H2O bend
1070	1095	1090	1100	phosphate[44]
987	957	964	961

The position of the amide I bands (the main band between 1645 and 1655 cm–1) indicates mainly an α-helix or unordered secondary structure of the adsorbed biomolecules.[48] The very weak band at approximately 1670 cm–1 may indicate that some of the molecules have other secondary structures, such as turns.[48] The fitting procedure became exceedingly difficult to perform in the case of the 2 h exposure in mucin because of the large negative water bending band at ≈1625 cm–1. It should be noted that additional uncertainties exist, such as contributions from side chains[47] and difficulties in fitting in the case of broad bands composed of several vibrational modes.[49]

As evident from Figure , the amplitude of the negative water band grew with time, which indicates that less water was probed in the spectra compared with the background (Co NPs in PBS) collected before the introduction of biomolecules and subtracted from each spectrum when the biomolecules were introduced (Figure ). As will be shown in the next section, the dissolution of the Co NPs was relatively rapid (at least 20% of the total mass dissolved within 1 h). As the dissolution of the surface oxide of the Co NPs results in a thickness reduction of the Co NP film, the effective refractive index of the medium above the ATR crystal is reduced.[50] This makes the probing depth of the IR beam in the ATR measurements shorter, which in turn makes the amount of probed water smaller compared to the background that is collected at the start of the experiment.[50] In all, this results in an FTIR spectrum with a water band in the negative direction. This assumption was confirmed as the stretching vibrations of water (≈3000–3700 cm–1) followed the same trend as the bending mode at 1625 cm–1. An increased adsorption of both biomolecules and phosphate anions (the biocorona) also results in a reduction of water molecules that are associated to the Co NP surface and thus in a lower amplitude of the band related to water in the spectra. This effect was however minor as the adsorption of polypeptides and proteins was fast and did not increase in amplitude after 2 min of exposure, whereas the negative water band amplitude continued to increase within the exposure time frame (up to 2 h) (Figures and 6). An increased amplitude of the negative water bending mode band was also observed for the amino acid solutions (Figure S5). The adsorption of proteins and biomolecules was irreversible with respect to rinsing, as evident from the lack of changes in the ATR–FTIR spectra upon extensive rinsing using PBS (data not shown).

Figure 6

Fitted peak areas of the amide bands of adsorbed polypeptides and proteins and the presence of H2O (bending mode) for the Co NPs in PBS. (A) Polyglutamic acid, (B) polylysine, (C) lysozyme, and (D) mucin.

Peaks attributed to phosphate adsorption (1110 and 960 cm–1) were also observed in Figure , similar to the ATR–FTIR spectra of Co NPs in PBS with different amino acids (Figure ). In the case of polylysine, these peak positions resemble the positions of the solution spectra of PBS, whereas their positions were shifted for the other biomolecules in a similar way as observed for the amino acids (Figure ). One plausible explanation for the lack of peak shifts in the presence of polylysine is the ionic bonding between the positively charged amine group (NH3+) of the adsorbed polylysine side chain and the negatively charged phosphate anion (HPO42–). This conformation might not induce a change in the vibration mode of HPO42– in the same way as observed for the Co phosphate complexes observed for Co NPs in PBS.

Several intermolecular interactions can influence the adsorption of biomolecules on surfaces, including ionic (electrostatic) interactions (both repulsive and attractive), hydrogen bonding, hydrophobic interactions, hydration forces, acid–base interactions, and van der Waals forces.[51] The most important driving forces for protein adsorption are often regarded to be hydrophobic and ionic interactions, combined with an entropy gain originating from the release of small molecules (e.g., water and counterions) from the interface between the biomolecule and the surface and in some cases conformational changes of the protein during the adsorption.[52,53] Hence, the size of the biomolecule will largely influence its ability to adsorb onto surfaces. In the case of larger biomolecules compared with amino acids, a higher amount of contact points between the Co NPs will result in a gain in entropy upon adsorption, as opposed to the amino acids. Adsorption was evident for the polypeptides and proteins regardless of their charge. From this, it follows that the gain in entropy upon adsorption originates from dehydration from the hydrophilic Co NPs as a result of displaced water molecules and other small molecules. Adsorption took place for both the negatively charged polyglutamic acid and the net negatively charged mucin on the average negatively charged Co NP layer deposited onto the ATR crystal. Some positively charged Co NPs may have been present in the Co NP layer (Figure S3). However, the electrostatic interactions are, as discussed above, expected to be of a very short range because of the high ionic strength of the solutions.

The ATR–FTIR results were corroborated by TEM findings, as illustrated for unexposed Co NPs and Co NPs exposed in PBS, in PBS + lysine, and in PBS + polylysine, after 1 and 24 h (Figure ). EDS data (not shown) indicate a predominance of Co, P, and O (sometimes also Na) (indicative of Co phosphate) when exposed in PBS (P/Co atomic ratio: 0.36–0.42) and in PBS + lysine (P/Co atomic ratio: 0.24–0.56), whereas the exposure in polylysine resulted in particle surface layers predominantly rich in Co, O, and C (no or minor content of P) as well as features (no evident particles left) with the P/Co atomic ratio between 0.10 and 0.30. The latter observation is in concordance with the lack of peak shifts for the phosphate peaks in the case of polylysine, as observed with ATR–FTIR spectroscopy (see Figure ), which implies the formation of Co phosphate species in solution or their presence within the outermost part of the biocorona rather than that adsorbed to the Co NP surface. Unexposed Co NPs showed only the presence of Co and O (Figure ). Similar observations were made after 24 h with the Co particles still left after 24 h for all conditions, which means no complete dissolution within this time frame.

Figure 7

TEM images of Co NPs. (A) Before exposure. (B,C) After 1 and 24 h in PBS. (D,E) After 1 and 24 h in PBS + 146 mg/L lysine. (F,G) After 1 and 24 h in PBS with 146 mg/L polylysine. The scale bar equals 50 nm.

Changes in Particle Stability and Dissolution Related to the Adsorption of Phosphates and Biomolecules

Studies were conducted to assess whether the adsorption of a “corona”, that is, of phosphate species and biomolecules of different characteristics, would influence the extent and pattern of Co NP dissolution and the colloidal stability of Co NPs.

Scattered light intensity levels from the NPs in solution were rapidly and substantially reduced within 24 h of exposure for all biomolecule conditions and also for Co NPs exposed in PBS only, as illustrated in Figure for the selection of biomolecule-containing solutions. A reduction in scattered light intensity indicates the agglomeration and sedimentation of NPs, showing that the van der Waals forces were higher than any repulsive electrostatic or steric forces from the adsorbed phosphate and/or polypeptides and proteins. Increasing scattered light intensities were observed for some early time points (<1 h), indicative of larger sized agglomerates momentarily entering into the path of the laser beam during the PCCS measurements.

Figure 8

Changes in scattering light intensity with time (every 10 min up to 1 h and 1 day) measured by means of PCCS for Co NPs (10 mg/L) exposed in PBS with and without biomolecules (146 mg/L). (A) PBS + cysteine, (B) PBS + polyglutamic acid/polylysine, (C) PBS + lysozyme/mucin.

Rapid sedimentation disabled the possibility for detailed particle size distribution measurements with time. The particle sizes in solution were in the range from 400 to 5000 nm during the first 1 h of exposure. However, there was not a clear tendency of changes in particle size with time. After 24 h, the intensity was reduced by 40–180 times to levels less than 5000 counts/s. These low and largely varying light intensities between different sampling times and biomolecule conditions up to 24 h, and poor curve fitting for correlation functions in PCCS, suggest the predominant formation of micrometer-sized agglomerates and almost complete sedimentation within 24 h.

Because of the rapid particle agglomeration and sedimentation for all exposure conditions already within 24 h, differences in dissolution behavior are only displayed after 1 and 24 h. The release data after 168 h are presented in Figures S6 and S7 in the Supporting Information.

The adsorption of phosphate on Co NPs (ATR–FTIR, Figures and 4) resulted in 4.5–6 times higher released amount (per mass) of Co in the case of PBS compared with saline (Figure A). It is possible that the adsorbed phosphate can destabilize surface Co species by weakening the Co–O bonds, which is the opposite trend of metals such as Fe when it comes to the effect of phosphate on dissolution.[54] Complexation of phosphate to free Co ions in solution can in addition enhance the extent of dissolution as the system is pushed further away from equilibrium conditions in terms of dissolved Co ions.

Figure 9

Co dissolution per mass (%) for different Co NP loadings in PBS (pH 7.4) with and without amino acids. (A) Saline, PBS; (B) PBS with and without 14.6 and 146 mg/L amino acids, 1 h; (C) PBS with and without 146 mg/L amino acids, 1 h; (D) PBS with and without 146 mg/L amino acids, 24 h.

No significant differences in soluble Co were observed for the different amino acid concentrations or when compared with PBS at a particle loading of 1 mg/L (Figure B). The released fractions of Co in solution after 1 and 24 h from the Co NPs in PBS with and without the presence of amino acids (146 mg/L) are presented in Figure C,D for three different particle concentrations (1, 10, and 20 mg/L Co NPs). Except for the exposure with cysteine, the results show 20–60% of the Co NP mass to be dissolved within 1 h followed by considerably slower release rates with time (24 h—Figure D and 168 h—Supporting Information), independent of particle and amino acid concentrations (Figure B–D).

Observed dissolution rates are in general agreement with the reported dissolution rates in physiological cell media of the same Co NPs.[27]

Speciation modeling calculations (Table ) showed that Co ions have the highest affinity to phosphate when compared with all amino acids investigated, except for cysteine. These calculations support higher fractions of released Co in solution in the presence of cysteine (Figure ). This may be related to a higher affinity between cysteine and soluble Co ions, from which follows an increased complexation and hence driving force for dissolution.[55] As no significant differences in agglomerate size (PCCS) could be discerned for the Co NPs (10 mg/L) in PBS with and without cysteine (146 mg/L), differences in surface area do not play a significant role for the comparison of dissolution between different biomolecules. It should be noted that Table does not include insoluble Co phosphates as they are missing from the JESS database. Recent predictions using Visual MINTEQ show that insoluble Co3(PO4)2 dominates the formation of Co phosphate salts in PBS.[27] The lack of reduced amounts of Co observed in solution (Figure ) indicates however the minor precipitation of Co complexes/compounds up to 24 h. The phosphate concentration in this study (20 mM) is higher than that under more realistic conditions (ca. 0.38 mM)[43] and can hence amplify the importance of phosphate relative to more realistic conditions. Nonetheless, PBS is a very widely used buffer for NP transformation investigations, and Co phosphates are still expected to form under physiological conditions.[4]

The released fractions of Co per mass from the Co NPs exposed for 1 and 24 h (168 h in Supporting Information) in PBS in the presence of polypeptides, mucin, or lysozyme are presented in Figure for two different biomolecule concentrations and Co NP loadings. Under these conditions, 15–45% of the Co NP mass was dissolved into solution within 1 h, that is, to a generally lower extent than that observed in the case of PBS and amino acids (35–55%). Previous literature findings show dissolved amounts of Co (from 100 μg/L Co NPs in PBS) of 15.6% after 2 h.[27] However, as the sonication time of that study was only 3 min, compared to 15 min in this study, the observed findings show that a prolonged sonication time can result in increased dissolution rates also in the stock solution and alter the surface properties of the sonicated NPs compared to dry conditions (Figure ). The results are in concordance with the literature, showing effects of sonication on particle reactivity and surface composition.[27] The only exception was lysozyme after 1 h for the lower particle loading and the biomolecule concentration. It is hypothesized that the small size of lysozyme in solution can increase its rate of exchange between solution and the surface and therefore enhance the extent of metal release at a larger rate than the other, larger, biomolecules such as mucin and polypeptides.

Figure 10

Dissolution of Co NPs of different concentrations (1 and 10 mg/L) in PBS with and without polypeptides or proteins of different concentrations (14.6 or 146 mg/L): (A) 1 mg/L Co NPs in PBS containing 146 mg/L polypeptides/proteins, (B) 10 mg/L Co NPs in PBS containing 146 mg/L polypeptides/proteins, (C) 1 mg/L Co NPs in PBS containing 14.6 mg/L polypeptides/proteins, (D) 10 mg/L Co NPs in PBS containing 14.6 mg/L polypeptides/proteins.

The higher particle loading (10 mg/L) generally resulted in lower amounts of released Co after 1 h compared to the lower loading (1 mg/L), possibly because of the formation of larger particle agglomerates that facilitates the adsorption and biocorona formation of the relatively large polypeptides and proteins. Higher loadings will, in addition, generally result in a higher released Co ion concentration in solution. Higher concentrations in solution should reduce the dissolution rates compared with the findings for lower loadings, as equilibrium conditions will be reached at lower NP concentrations.[56] These effects were not observed any more after 24 h for which more (1 mg/L Co NPs), or similar/slightly reduced (10 mg/L Co NPs), amounts of Co per mass were dissolved compared with the exposure in PBS only. Exposure up to 1 week (168 h) showed no or only minor effects of the adsorbed biomolecules on the extent of Co dissolution (Figures S6 and S7). The results illustrate that the formation of a biocorona of polypeptides or proteins combined with phosphate species at least initially hinders (up to 1 h) the dissolution of Co NPs in PBS.

Even though no large differences in the dissolution pattern could clearly be discerned related to the charge or number of charged entities, the results imply that the negatively charged amino acids and polypeptides with the same functional group (carboxylates), glutamic acid and polyglutamic acid, resulted in somewhat higher released quantities after 1 h compared with the positively charged lysine and polylysine (amine groups) (Figure ). These findings are in line with previous findings that show the metal–surface interactions of negatively charged proteins (BSA) to significantly enhance the extent of metal release from both stainless steel (of different grades) and silver, whereas the interaction with positively charged proteins (lysozyme) predominantly resulted in a minor increase of the amount of released metals from stainless steel and a significant reduction of released silver from massive silver.[33−35]

Figure 11

Released amount of Co per mass from Co NPs: (A) 1 mg/L and (B) 10 mg/L after 1 h in PBS with and without the presence of amino acids, polypeptides, or proteins in a concentration of 146 mg/L. The stars indicate statistically significant differences (p < 0.05, Student’s t test).

The correlation between the charge of biomolecules and dissolution was, in this study, less evident with a higher release of Co after 1 h in the case of lysozyme (net positively charged) compared to mucin (net negatively charged). However, it is hypothesized that the small compact globular structure of lysozyme increases its exchange rate between solution and the surface and therefore enhances the metal release process compared to the other more flexible biomolecules (mucin and polypeptides) as observed for lower particle and biomolecule concentrations.

Initially hindered release processes in the presence of adsorbed polypeptides/proteins are in general concordance with the trends often observed for corrosion and metal release.[57] The adsorbed molecules will temporarily block the surface areas for ongoing dissolution/corrosion processes. However, slower corrosion/release processes such as ligand exchange that take place after certain time periods can result in enhanced dissolution,[57] even though some exceptions from this trend have been reported for corrosion.[58] There is moreover a possibility that the biomolecules may slowly replace the adsorbed phosphate.[21]

The observed trend of initially slower dissolution in the presence of proteins is opposite to what has been observed for the massive Co metal interacting with BSA in saline.[26] This implies that different proteins may have different dominating modes of interaction with the Co surface, as BSA was hypothesized to catalyze corrosion reactions leading to the faster dissolution of NPs. These effects were not observed in this study.

Conclusions

This paper focuses on the interaction between Co NPs and biomolecules of different properties in PBS in order to understand their potential adsorption and biocorona formation and how it influences particle stability, particle characteristics, and the dissolution of Co NPs. The sonication step that was performed to disperse the Co NPs in PBS prior to exposure to the biomolecules resulted in a slightly altered composition of the surface oxide of the Co NPs compared with dry conditions and induced some particle dissolution in the stock solution. This preparatory step may be the reason why the dissolution rates of the Co NPs in PBS were faster compared with the previously reported rates.

The high affinity of phosphate to Co ions and the Co NP surface resulted in the formation of strong surface complexation that largely hindered the adsorption of small amino acids. The presence of a phosphate-rich corona reduced the protective properties of the CoO/Co3O4-containing surface oxide on the Co NPs, assessed from increased dissolution.

Compared to the influence of PBS, the presence of amino acids of different characteristics had no or minor influence on the dissolution of Co NPs, with cysteine as the main exception. The dissolution of Co NPs was rapid with 15–40% dissolved within 1 h of exposure. In addition to phosphate adsorption, the larger biomolecules (polylysine, polyglutamic acid, mucin, and lysozyme) adsorbed to the Co NPs forming a biocorona, which resulted in the reduction of the dissolution rate during the first hour. These effects did, however, not last after longer exposure periods (24–168 h). The presence of negatively charged biomolecules possibly increased the release of Co compared with the positively charged biomolecules, with lysozyme being an exception.

The results imply that Co NPs will not be relatively long-lived if internalized by organisms, for example, in the human lung or within the gills of a fish. The presented results underline the diversity of possible outcomes with respect to the surface and particle characteristics and their dissolution pattern depending on the presence of different types of biomolecules, as well as the importance of phosphate for the resulting NP surface composition and properties. These aspects are important to consider as the media for eco- and nanotoxicological assays, for example, typically vary with respect to both phosphate and biomolecule contents.[21]

Materials and Methods

Investigated Particles

Co NPs (product code CO-M-028M-NP.1000N, LOT # 1211392979-814, <100 nm, metal purity-99.9%) were purchased from American Elements, USA. According to previous particle characterization, the average primary particle size was 25 ± 8.8 nm (based on TEM imaging), the BET surface area was 10.7 m2/g, and CoO was the main constituent of the surface oxide.[27]

Chemicals and Solution Preparations

PBS was used to simulate the human blood serum, commonly used for bioaccessibility studies.[59] Its chemical constituents include 8.77 g/L NaCl (CAS: 7647-14-5, Lot: 17E314108, assay 100.1%, VWR Chemicals), 1.28 g/L Na2HPO4 (CAS 7558-79-4, Lot: # BCBV2075, assay ≥99.0%, Sigma-Aldrich), and 1.36 g/L KH2PO4 (CAS 7778-77-0, Lot: 16K154102, assay 99.8%, VWR Chemicals) and has an ionic strength corresponding to 0.15 M NaCl and 0.02 M phosphate buffer. The 20 mM phosphate buffer (PBS buffer without chloride) included 1.28 g/L Na2HPO4 and 1.36 g/L KH2PO4. NaCl (150 mM) contained 8.77 g/L NaCl. All salts were dissolved in ultrapure water (18.2 MΩ cm resistivity, Millipore, Sweden). The pH was adjusted to 7.4 by means of 50% NaOH (batch no. 08D280509, made in EC-EMB 45053, EC label: 215-185-5, Prolabo). l-Lysine (MW: 146.19 g/mol, CAS 56-87-1, Lot: #BCBT1990, assay ≥98%, PCode: 101813210, Sigma-Aldrich), l-glutamine (MW: 146.14 g/mol, CAS: 56-85-9, Lot: # BCBJ2148V, BioUltra, PCode: 101455953, Sigma-Aldrich), l-glutamic acid (MW: 147.13 g/mol, CAS 56-86-0, Lot: #BCBH3883V, Assay ltra3, ECPCode: 101420757, Sigma-Aldrich), and l-cysteine (MW: 121.16 g/mol, CAS 52-90-4, Lot: #SLBQ4967V, assay ≥98%, PCode: 1002305403, Sigma-Aldrich) were all dissolved directly in the PBS solution to a concentration of 146 mg/L (1 mM). Poly-l-lysine hydrobromide (MW: 30 000–70 000 g/mol, CAS: 25988-63-0, Lot: #SLBS0150V, PCode: 1002509720, Sigma-Aldrich) and poly-l-glutamic acid sodium (MW: 15 000–50 000 g/mol, CAS: 26247-79-0, Lot: #096K5103V, PCode: 1001412838, Sigma-Aldrich) were dissolved in PBS to a concentration of 146 mg/L. Lysozyme from chicken egg white (lyophilized powder, protein 5, MW: 14.1 kDa, CAS: 12650-88-3, Lot: #SLBT5160, PCode: 1002515139, Sigma-Aldrich) and mucin from bovine submaxillary glands (Type I-S, MW: 7 MDa, CAS: 84195-52-8, Lot: #SLBS0651V, PCode: 1002519367, Sigma-Aldrich) were dissolved directly in PBS to a concentration of 146 mg/L.

The selected concentrations of the amino acids are relevant to physiological conditions.[60] The investigated concentrations of the polypeptides and proteins were the same as for the amino acids, in order to have comparable conditions. Even though these concentrations are different from real biological conditions, this fundamental study only reflects the conditions using one kind of amino acid, polypeptide or protein separately in each solution and not combined.

Biomolecule Adsorption Studies

The ATR–FTIR technique, using a Bruker Tensor 37 FTIR spectrometer with a platinum ATR–IR accessory, was employed to study the adsorption of different biomolecules to the Co NPs. The ATR–IR accessory consists of a diamond crystal with an angle of incidence for the IR beam of 45°. For investigations of dry Co NPs, a deuterated triglycine sulfate (DTGS) detector equipped with a polyethylene window was used with a relatively low wavenumber (>ca. 250 cm–1). A  mercury cadmium telluride  detector with a ZnSe window was used for the in situ adsorption experiments on Co NP films in order to have a greater sensitivity compared with the DTGS detector, however limiting the working wavenumber range to >ca. 650 cm–1.

A total of 256 scans were collected with a resolution of 4 cm–1 in the experiments using amino acids and 512 scans in the studies using polypeptides and proteins. The Co NPs (a particle loading of 25 mg in 10 mL ethanol) were dispersed by means of tip sonication for 15 min (Branson Sonifier 250, constant mode, output 2) that delivered an acoustic energy per liter of 1.18 × 106 J/L.[41] Approximately 400 μL of the sonicated Co NP solution was transferred onto the ATR–IR crystal by means of a pipette directly after dispersion, followed by drying for approximately 2 h at ambient air conditions. This procedure allowed ethanol evaporation and NP layer formation on the crystal. All spectra were collected when such NP layers were exposed to different aqueous solutions. Background spectra were collected for the Co NP layer in PBS only, followed by measurements in the biomolecule solutions of interest. Separate spectra with biomolecules in PBS showed no peaks related to the bulk solution species of the biomolecules at the studied concentrations of this study. This means that the detected signals of biomolecules originate from their adsorption to Co NPs. All experiments were performed in duplicate, and a spectrum was collected every fifth min up to 1 h for the experiments in the amino acid solutions and up to 2 h in the case of the polypeptide and protein solutions. At the end of the experiments, PBS was used to rinse the Co NP layer to assess if the adsorption of the biomolecules was irreversible or not.

The limit of detection (LOD) in terms of adsorption of an equivalent layer of amino acids was estimated to ca. 10%, that is, when amino acids adsorbed to an extent exceeding 10% of the surface, a signal can be detected (see Supporting Information for details).

Metal Release Measurements

A stock solution with a concentration of 1 g NPs/L was prepared in an acid-cleaned glass vial by dispersing 6 mg Co NPs (weighted by XP26 DeltaRange Microbalance, Mettler Toledo) in 6 mL of ultrapure water (18.2 MΩ cm resistivity, Millipore, Sweden) using tip sonication (Branson Sonifier 250, constant mode, output 2) for 15 min. To prepare the biomolecule–Co NP solutions, amino acids, polypeptides, or proteins solutions (ca. 20 mL) of each concentration were transferred into an acid-cleaned 60 mL PMP Nalgene jar from which an appropriate volume of the freshly prepared Co NP stock solution was added to Co NP concentrations of 10 and 1 mg NPs/L. For a sample with 1 mg/L Co NP loading, 0.02 mL of NP stock solution was added into a 19.98 mL solution containing amino acids and polypeptides or proteins. For the 10 mg/L Co NP loading, 0.2 mL of the NP stock solution was added into 19.8 mL of the biomolecule-containing solution. At the beginning of each experiment, a dose sample (stock solution added into ultrapure water and acidified directly after preparation) was prepared in an acid-cleaned plastic vial (25 mL) in order to determine the actual added amount of Co NPs (the administered dose) as it may differ from the nominal dose as shown for metal NPs.[41] Each set of experiments to be analyzed for the released Co concentration in solution using AAS included a dose sample, a blank sample (biomolecule solution without Co NPs), and triplicate samples of the biomolecule-exposed Co NP solutions (see below). The dose samples were immediately acidified to pH < 2 using 65 vol % ultrapure HNO3. Samples with biomolecule-Co NPs were placed in an incubator (Stuart S 180) at bilinear agitation conditions (37 °C, 12° inclination, 22 cycles/min) for 1, 24, and 168 h (1 week). After exposure, solutions of 4.5 ± 0.1 mL were transferred to centrifuge tubes and ultracentrifuged (Beckman Optima L-90K Ultracentrifuge, 50 000 rpm, 1 h) to remove the nondissolved particles from the solutions. Note that the ultracentrifugation treatment effectively makes the exposure time in solution somewhat longer than the nominal investigation time periods for dissolution (1 h, 24 h, and 1 week)

Nanoparticle tracking analysis (Nanosight NS300, Malvern, Sweden) measurements confirmed that no detectable Co NPs were left in solution after ultracentrifugation. After the centrifugation step, 3.7 mL of the top volume of the centrifuged solution was transferred to acid-cleaned 10 mL vials and acidified to a pH < 2 by using 65% ultrapure HNO3 for subsequent AAS measurements.

Chemical Analysis of Released Co in Solution by means of AAS

Graphite furnace AAS (PerkinElmer, AAnalyst 800) was used to analyze the amount of released Co in solution for exposures with a particle loading of 1 mg/L. Standard solutions with known Co ion concentrations (10, 30, 60, and 100 μg/L) were prepared in 1% ultrapure HNO3 together with a blank sample (1% ultrapure HNO3). Mg(NO3)2 was used as a matrix modifier during the analysis. The LOD was approximately 3.6 μg/L (the mean value of blank samples + 3 times the standard deviation of the blank samples). A calibration standard was used for quality control and run every fourth sample to ensure no memory effects. Calibration and analysis were repeated if the quality control sample deviated more than 10%. Flame AAS was employed to assess the released Co in solution for higher particle loading (10 mg/L). This calibration curve included a blank sample (ultrapure water) as well as 1, 3, and 10 mg/L Co ion standard solutions. The 3 mg/L solution was used as quality control and run every fourth sample. The LOD was approximately 0.20 mg/L. Calibration and analysis were repeated if the quality control deviated more than 10%. Before the AAS analysis of the samples, Co ions in the biomolecule control samples were made and measured, which indicated that the AAS method worked well. The results showed the described method to be valid for all Co-biomolecule exposures with recoveries >83%. The dose sample results were compared with those of digested Co NP samples (digestion with 65% HNO3 and H2O2), and a complete dissolution of the Co NPs was proved in the dose samples.

Changes in Particle Morphology and Composition Assessed Using TEM Combined with EDS

The morphology and structure of unexposed and precipitated Co NPs from the different solutions were analyzed using TEM (Hitachi HT7700 microscope operating at 100 kV) after the metal release investigations in PBS, lysine, and polylysine solutions for 1 and 24 h. The precipitated particles were transferred via a pipette directly from the jars onto TEM copper grids, coated with holey carbon films (Ted Pella), and left to allow the aqueous solution to evaporate under ambient air condition. All images were collected in bright-field mode.

TEM was also performed using a JEOL 200 kV 2100F field emission microscope operated in scanning beam mode (scanning transmission electron microscopy) combined with EDS microanalysis using a windowless silicon drift detector X-MaxN TLE from Oxford Instruments and the Aztech software. A beryllium sample holder was used when performing the chemical analyses.

Compositional Analyses of the Outermost Surface Oxide Using XPS

XPS (Kratos AXIS UlraDLD, Kratos Analytical) measurements were performed for compositional analysis of the outermost surface of the Co NPs (pristine-unexposed NPs) and Co NPs after sonication in ultrapure water and in ethanol, respectively, with the aim to investigate if the sonication treatment would change the surface oxide composition. Dry NPs were transferred with a spoon directly onto an adhesive carbon tape for XPS analysis. Co NPs dispersed in ultrapure water and in ethanol were dispersed in the same way as the NP stock solutions for the metal release experiments (described above), pipetted onto a carbon tape (total volume about 200 μL), and dried at ambient air conditions. High-resolution spectra of Co 2p, O 1s, and C 1s were acquired on two different areas (each ≈0.4 mm2) using a monochromatic Al X-ray source (150 W), a pass energy of 20 eV, and a linear background correction, averaging nine scans for each element and area. All binding energies were corrected to the adventitious carbon C 1s peak (C–C, C–H) set at 285.0 eV.

Particle Size and Zeta Potential in Solution

Measurements of the particle size in solution were performed using dynamic light scattering spectroscopy by means of PCCS (Nanophox, Sympatec GmbH, Germany). The main purpose was to investigate the stability of Co NPs dispersed in PBS with and without amino acids, polypeptides, or proteins. All measurements were performed at an ambient laboratory temperature (25 °C) after 10, 20, 30, 40, 50, and 60 min exposure in solution. A volume of 1 mL was pipetted into the PCCS cuvettes and measured for 180 s. All measurements were run twice. Standard latex particles sized 100 nm (Malvern instruments) were used as quality control prior to the analysis to ensure the accuracy of the measurements.

The zeta potential of the Co NPs was determined using laser Doppler microelectrophoresis using a Zetasizer Nano ZS instrument (Malvern Instruments, U.K.) at 25 °C. Triplicate measurements were performed in PBS, in 10 mM NaCl (pH 7.4), and in phosphate buffer (20 mM) without NaCl.

Speciation Modeling of Released Co in Solution

The JESS version 8.3[60] model was used for chemical equilibrium speciation calculations of Co in PBS and amino acid solutions. The temperature was set to 37 °C and the redox potential to 300 mV based on measurements using an Inlab redox electrode (Mettler Toledo, Sweden).

Statistical Analysis

Student’s t test analysis was done in EXCEL, with a double-side mode to compare the difference between the results of several groups of metal release samples.