Effect of iron oxide nanoparticles on the oxidation and secondary structure of growth hormone

Abstract

Oxidation of therapeutic proteins (TPs) can lead to changes in their pharmacokinetics, biological activity and immunogenicity. Metal impurities such as iron are known to increase oxidation of TPs, but nanoparticulate metals have unique physical and chemical properties compared to the bulk material or free metal ions. Iron oxide nanoparticles (lONPs) may originate from equipment used in the manufacturing of TPs or from needles during injection. In this study, the impact of lONPs on oxidation of a model protein, rat growth hormone (rGH), was investigated under chemical stress. Hydrogen peroxide (H₂O₂)- and 2,2′-azobis (2-methylpropionamidine) dihydrochloride (AAPH) oxidized methionine residues of rGH, but unexpectedly, oxidation was suppressed in the presence of lONPs compared to a phosphate buffer control. Fourier transform infrared (FTIR) spectroscopy indicated splitting of the α-helical absorbance band in the presence of lONPs, while CD spectra showed a reduced α-helical contribution with increasing temperature for both rGH and rGH-IONP mixtures. The results collectively indicate that lONPs can increase the chemical stability of rGH by altering the kinetics and preference of amino acid residues that are oxidized, although the changes in protein secondary structure by lONPs may lead to alterations of physical stability.

INTRODUCTION

The higher order structure (secondary, tertiary, quaternary) of therapeutic proteins (TPs) is important for their biological activity and stability. Chemical changes including oxidation of susceptible amino acids can lead to alteration in the secondary structure of TPs, resulting in increased aggregation, altered pharmacokinetics, reduced biological activity and increased immunogenicity.¹

The trace metal impurities arising from buffers, formulation components, and processing equipment, can oxidize TPs by catalyzing formation of reactive oxygen species (ROS). Soluble iron impurities can generate superoxide radicals, hydrogen peroxide (H₂O₂), and hydroxyl radicals leading to oxidation of TPs.^2,3 In addition, iron oxide nanoparticles (IONPs) may be present due to iron containing equipment used during the manufacturing process or produced by injection needles during TP administration.^4,5 A protein corona can develop around nanoparticles (NPs) leading to the formation of larger particle sizes and possibly aggregation.⁶ Protein-NP interactions may induce changes in protein secondary structure⁷ and further oxidation of TPs.⁸ It is not known, however, if alterations in the secondary structure of TPs induced by binding to IONPs can influence the extent of oxidation. We selected rat growth hormone (rGH) as a model protein for studying oxidation, since it is a very similar analogue to the TP human growth hormone, and is suitable for future in vivo immunogenicity studies using rodent models.⁹ Human growth hormone (hGH) is a subcutaneously administered pharmaceutical ¹⁰ widely used to treat growth hormone deficiency.^11,12 Since hGH is dosed daily over a period of years, the development of anti-hGH antibodies to oxidized protein is of concern. Pediatric studies have identified hGH antibodies in 75% of patients after 12 months of treatment¹³, and in at least one case a loss of hGH efficacy despite dose escalation in a child with a high titer of anti-hGH antibodies¹⁴. In this study, the oxidation and secondary structure of rGH were evaluated in the presence of IONPs. Overall, this study has demonstrated that IONPs markedly decreased the oxidation of rGH under chemically accelerated conditions; however, the IONPs induced alternations in rGH’s secondary structure that may have an impact on stability and bioactivity.

EXPERIMENTAL SECTION

Materials

The lONPs (γ-form) were purchased from US Research Nanomaterials Inc (Stock# US7558, Fe₂O₃ CAS# 1309–37-1, Houston, TX). The 2,2′-azobis (2-methylpropionamidine) dihydrochloride (AAPH), H₂O₂, potassium ferricyanide, 1,10-phenanthroline monohydrate, sodium citrate tribasic dihydrate, and ascorbic acid were purchased from Sigma Aldrich (St. Louise, MO). Ammonium bicarbonate, dithiothreitol (DTT), iodoacetamide (IAA), LC/MS-grade water (0.1% v/v formic acid), acetonitrile (0.1% v/v formic acid), monobasic sodium dihydrogen phosphate, sulfuric acid, hydrogen fluoride, dibasic sodium hydrogen phosphate, and the Pierce BCA protein assay kit (Product number: 23227) were purchased from Fisher Scientific (Waltham, MA). Trypsin/Lys-C (mass spectrometry grade) was purchased from Promega (Madison, Wl). Amicon ultra-0.5 centrifugal filter devices (10, 30, 100 kDa cut off membranes) were purchased from Millipore Inc., (Bedford, MA, USA). The 4-aminomethyl-benzylsulfonic acid (ABS) was synthesized according to a published method.¹⁵

Production of rat growth hormone (rGH)

Rat GH is a 22-kDa protein of 191 amino acids that contains two disulfide bonds.^16,17 The cDNA plasmid for the rGH protein sequence (Protein accession number- AAI66872) was synthesized by GenScript (New Jersey, USA) and inserted into pET-28a(+) using the Ndel and BamHI restriction sites.¹⁸ The pre-protein sequence (first 25 amino acids) was removed from the original cDNA for production in E. coli.; the resulting rGH amino acid sequence is shown in the Supplementary Data S1. The plasmid was transformed into DH5α competent E. coli cells, and a single colONy was expanded, mini-prepped, and the sequence confirmed by KanPro Research Inc. (Lawrence, KS).

The confirmed rGH plasmid was transformed into BL21 (DE3) E. coli, and rGH was produced by shake flask culture using a protocol modified from previously reported methods.^17,19,20 In brief, the E. coli cells were cultured in lysogeny broth (LB) media with kanamycin (40 μg/mL) and chloramphenicol (40 μg/mL) at 37 °C and 200 rpm shaking. Once an optical density (at 600 nm) of around 0.6 was reached, cultures were induced with isopropyl β-d-1-thiogalactopyranoside (IPTG) (1.0 mM). After 4 h, cultures were centrifuged at 1391 ×g for 10 min and the cell pellet was suspended in lysis buffer (50 mM tris-hydrochloride, 500 mM sodium chloride, pH 8). The mixture was then sonicated (50% power, 4 cycles × 30 sec, at 4 °C) until it was no lONger viscous using a Fisher Scientific Sonic Dismembrator Model 500 (Waltham, MA). Inclusion bodies of rGH were isolated by centrifuging at 6,000×g (at 4 °C) for 30 min, followed by a water wash, and then dissolved in 6-M urea lysis buffer. The proteins were refolded by following proprietary procedures (KanPro Research Inc.), and then dialyzed in phosphate buffer saline (PBS), containing 88 mM of mannitol. rGH was analyzed for purity using SDS-PAGE (Supplementary Data S2). Intact mass analysis of the protein confirmed the molecular weight of 21.9 kDa and also showed around 15% initial oxidation (+16 m/z) on two methionine residues (M1 and/or M5). This modification probably occurred during purification and refolding (data not shown). Detailed analysis of a tryptic digest of the protein (sequence coverage of ~92%) using mass spectrometry was previously reported.¹⁹

Chemical Oxidation of rGH and sample processing

Chemically accelerated oxidation of rGH was induced with H₂O₂ or AAPH, in the absence or presence of lONPs. The final concentrations of rGH, H₂O₂, AAPH and lONPs were 0.8 mg/mL, 2 mM, 10 mM and 1.7 mM, respectively. This mixture was incubated at 37 °C for 3 or 24 h while shaking at 100 revolutions/minute. For H₂O₂-induced oxidation, 10 μL of catalase solution (1 mg/mL) were added to quench the oxidation. AAPH was removed using 10-kDa Amicon ultra-0.5 centrifugal filters. The oxidized rGH was processed further for reduction of disulfide bonds, to disrupt secondary structure and improve tryptic digestion efficacy, using DTT and alkylation of cysteine residues using IAA to prevent reformation of disulfide bonds. The disulfide bonds were reduced by adding 100 μL of dithiothreitol (26-mM stock solution in 50-mM ammonium bicarbonate buffer) and incubating this mixture at 45 °C for 1 h. The reduced cysteine residues were alkylated by adding 200 μL of IAA (50-mM stock solution in 50-mM ammonium bicarbonate buffer) and incubating at 37 °C for 1 h.

Trypsin/Lys-C digestion of rGH

The reduced and alkylated rGH (0.4 mg) was precipitated with 700 μL of 0.5-M perchloric acid. The samples were centrifuged at 15000×g for 15 min and the resulting pellet was suspended in 400 μL of 50-mM ammonium bicarbonate buffer (pH 7.5). Four μg of trypsin/Lys-C were added, and the samples were incubated at 37 °C overnight. The proteolytic digests were centrifuged using 10-kDa centrifuge filters to remove trace amounts of undigested rGH and trypsin/Lyc-C. The tryptic peptides were stored at −20 °C until mass spectrometry analysis.

Mass spectrometry

A nanoAcquity UPLC (Waters Corporation, Milford, MA) connected to a Xevo Q-TOF (Waters Corporation, UK) was used for LC-ESI-MS experiments. Tryptic peptides were first trapped on an UPLC Symmetry C18 nanoAcquity trap column (5 μm, 180 μm × 20 mm, Water Corporation) and then separated on an UPLC Peptide CSH C18 nanoAcquity column (1.7 μm, 75 μm × 250 mm, Waters Corporation). For separation, we applied an optimized method and parameters that were previously reported.²¹ In silico tryptic digestions of rGH and prediction of fragment ions were performed with web based applications, Proteomics Toolkit (Institute of Systems Biology, Seattle, WA) and ProteinProspector 5.21.2 (The University of California, San Francisco, CA).

Fluorogenic derivatization

Tyrosine residues of rGH were oxidized with AAPH in the absence or presence of IONPs. The oxidation products were derivatized with ABS to produce a fluorescent benzoxazole according to our previously described protocol.^21,22 Briefly, rGH oxidized by AAPH was buffer exchanged with 100 mM sodium phosphate buffer (pH 9) using 10-kDa cutoff filter devices. Potassium ferricyanide and rGH were mixed in a molar ratio of 30:1. ABS (10 mM) was added, and the mixture was incubated at ambient temperature in the dark for 90 min. Fluorescence (360 nm-excitation/490 nm-emission) was measured using a SpectraMax GeminiXS microplate fluorometer (Molecular Devices, Sunnyvale, CA).

Fourier transform infrared (FTIR)

Spectra of rGH solutions (1 mg/mL) with varying concentrations of IONPs (0.42 to 1.47 mM) were acquired with a Bruker Tensor 27 FTIR spectrometer fitted with a Bio-ATR cell (Bruker, Billerika, MA). Scans were recorded from 600 to 4000 cm⁻¹ at a resolution of 1 cm⁻¹. The FTIR spectra were acquired after 10 min, 30 min, 1 h and 2 h incubation periods at ambient temperature. In addition, rGH and rGH-IONP mixtures were incubated at increasing temperatures from 15 to 85 °C at increments of 2.5 °C/step with a 120 s equilibration time, and 128 scans/step. The phosphate buffer-IONP-mixture was used for background subtraction of the FTIR spectra. The Salvitzky-Golay algorithm was used to calculate the second derivative spectra, baseline and atmospheric corrections, and 9-point data smoothing (Opus v6.5, Bruker Corporation, Billerica, MA).

Circular Dichroism (CD) spectroscopy

Far-ultraviolet (UV) CD spectra were acquired using a Chirascan spectropolarimeter (Applied Photophysics, Surrey, United Kingdom). The final concentration of rGH was 13 μM and the path-length for CD measurements was 0.1 cm. The spectra were acquired from 200 nm to 260 nm with increasing temperature from 10 to 90 °C at increments of 2.5 °C, with equilibration time of 120 s (n=3). The secondary structures were estimated from the CD spectra using the BeStSel web tool.²³

Measurement of iron content in IONPs

The iron content of the IONPs was determined using a 1,10-phenanthroline colorimetric assay.^24–26 A 10% w/v IONP-water dispersion was diluted 5-fold in water and centrifuged at 500×g for 30 min. The pellet of aggregated/settled IONPs was discarded and the supernatant was used for further analysis. Dynamic light scattering (Zetasizer Helix, Malvern Instruments Ltd., Malvern, UK) was used to measure particle size of the supernatant IONPs (~20 nm, 99.2% by number analysis, standard deviation of 3.8 nm) suspended in phosphate buffer (pH 7.4). The supernatant IONPs (1 mL) were treated with 0.5 ml of 8-N sulfuric acid (H₂SO₄), 0.4 ml of 15 % w/v ascorbic acid solution, and 0.2 mL of 48% hydrogen fluoride (HF) at 70 °C for 1 h. The IONP supernatant was reduced with ascorbic acid to convert Fe³⁺ to Fe²⁺ ions, so only Fe²⁺ ions would be available to form a colored complex after addition of 1,10-phenanthroline (0.5 ml, 100 mM). After incubating at 40 °C for 25 min, the reaction was quenched with 1 mL of 5% w/v boric acid, and then incubated at ambient temperature for a further 30 min. Samples were then diluted 20-fold with sodium citrate tribasic dihydrate buffer (10% w/v). Absorbance measurements at 510 nm were acquired with a SpectraMax Plus Microplate Spectrophotometer (Molecular Devices, Sunnyvale, CA), using a calibration curve of known Fe²⁺ concentrations. Fe²⁺ content was assayed in the same manner without the ascorbic acid reduction step. Color did not develop in any of the samples (data not shown), indicating that the IONPs did not have a significant Fe²⁺ content and were composed predominantly of Fe³⁺.

Statistical analysis

The ChemoSpec (R package) was used in RStudio (Version 1.1.383) (Boston, MA) to perform principle component analysis (PCA) on FTIR data.²⁷ Analysis of variance (ANOVA) and Tukey’s multiple comparison test were used to compare the degree of oxidation measured using the mass spectrometry and ABS derivatization technique. ANOVA and Tukey’s test were performed using GraphPad Prism 6 (La Jolla, CA).

RESULTS AND DISCUSSION

Identification of oxidized residues

In this study, H₂O₂ and AAPH were used to induce the chemical oxidation of rGH. Tryptic digestions were used to monitor the oxidation of methionine residues (Table 1). The MS/MS spectrum (Figure 1) of tryptic peptide no. 2 shows that M73 represents an oxidation target. The fragment ions y5, y6 and b3 indicate an increase in mass of 16 Da. In addition, a characteristic loss of methanesulfenic acid (CH₃SOH, 64 Da)²⁸ from the y5 ion was observed. This further confirmed the oxidation of M73 to methionine sulfoxide in the tryptic peptide no. 2. Other methionine-containing peptides, tryptic peptide no. 4, 3, 1, also showed oxidation of methionine. The MS/MS spectra for these tryptic peptides are described in the Supplementary Data S3, S4, S5, S6, S7 and S8, and the results are summarized in Table 1.

Table 1:

Oxidized residues (in red) and the associated tryptic peptides

No.	Oxidized residue	Tryptic peptide
1	M1, M5, F2	MFPAMPLSSLFANAVLR
2	M73	TDMELLR
3	M102	IFTNSLMFGTSDR
4	M124	DLEEGIQALMQELEDGSPR
5	M149	FDANMR

Figure 1:

Collision-induced dissociation (CID) of the oxidized peptide TDM(+16)ELLR obtained using LC-ESI-MS

The y11 and y13 ions indicated that M124 was oxidized on the tryptic peptide no. 4 (Supplementary Data S3). In the case of tryptic peptide no. 3; y7, y8, y9, y10 and y11 ion fragments were increased by 16 Da (Supplementary Data S4). Therefore, M102 was oxidized to methionine sulfoxide in the tryptic peptide no. 2.

Similarly, two peaks of the oxidized tryptic peptide no. 1 were observed in the chromatogram (data not shown) with the same m/z. The CID of the first peak (Supplementary Data S5) showed that either M1 or F2 were oxidized, because the fragment ions b2, b3, b4, b5 and b6 showed an increase in mass by 16 Da. The characteristic loss of methanesulfenic acid (CH₃SOH, 64 Da) from the a2 ion was observed. This may indicate oxidation of M1, rather than F2. However, any other fragment ion which could distinguish oxidation specifically on M1 or F2 was not found. While in the case of the second peak of tryptic peptide no. 1, which eluted earlier, we concluded that the M5 amino acid was oxidized because the y13 and b5 fragment ions increased in mass by 16 Da (Supplementary Data S6).

The tryptic peptide no. 1 was oxidized at both the M1/F2 and M5 amino acids. This can be shown by the increase in mass by 32 Da for the b5 fragment ion (Supplementary Data S7). The other fragment ions (y13, y15, b4) were increased in mass by 16 Da, indicating oxidation of M1/MF2 and M5.

In the case of tryptic peptide no. 5, oxidation of M149 was confirmed using the MS/MS fragmentation pattern (Supplementary Data S8). Oxidation of M149 can be shown by the increase in mass of the y4 and y5 fragment ions by 16 Da. Furthermore, a characteristic loss of methanesulfenic acid (CH₃SOH, 64 Da) from the y4 and y5 ions was observed. These fragments confirmed that tryptic peptide no. 5 was oxidized at the M149 residue. However, this oxidized tryptic peptide co-eluted with the native non-oxidized tryptic peptide no. 5. Hence the fragmentation pattern had several fragment ions which originated from the native peptide, and they were not assigned.

Quantification of oxidized residues in the presence of lONPs

Out of the 7 methionine residues (Table 1) found in rGH, 6 residues (M1, M5, M73, M102, M124, and M149) were monitored for oxidation in the presence of lONPs and H₂O₂. The peptide containing M179 was not found in the mass spectrometric analysis. According to the in silico tryptic digestion, M179 is part of a small 4 amino acid peptide. We hypothesize that this peptide was poorly retained on the trap column and, therefore, not detected.

The oxidation of M73 was significantly reduced (P≤0.01) after 24 h of incubation in the presence of H₂O₂ and IONPs compared to H₂O₂ alONe (Figure 2). However, at the earlier 3 h time point, oxidation of M73 was greater for the H₂O₂ + IONP than the H₂O₂ alONe or IONP alONe groups. This discordance at 3 h compared to 24 h may be due to non-equilibrium adsorption of the rGH to the IONP or temporal changes in the conformation of the IONP adsorbed rGH. The tryptic peptide no. 1 was also di-oxidized (M1/F2 and M5) by H₂O₂ (Supplementary Data S9). Similarly, IONPs suppressed oxidation of M102 and M124. In contrast, no difference in oxidation in the presence and absence of IONPs was observed for the peptide M149. Overall, the IONPs suppressed H₂O₂-induced oxidation of the rGH. These results were confirmed by monitoring H₂O₂ levels using the FOX assay in the presence of IONPs and without any protein. The H₂O₂ levels did not decrease significantly during 24 h of incubation in the presence of the IONPs (data not shown). This indicated that oxidant levels were not decreased by the IONPs and that suppression of rGH oxidation was due to interaction with the IONPs.

Figure 2:

Methionine (M73, M102, M124, M149) oxidation in the presence of IONPs and H₂O₂ using mass spectrometry

Caption: The results are presented as percent of the peptide oxidized (mean and standard deviation, n=2). Percent oxidation of various rGH residues was calculated based on the mass spectrometry derived area under the curve of oxidized and non-oxidized peptides.

3 h and 24 h groups were compared separately using 2-way ANOVA. The symbols above each column indicate the 3 or 24 h time points of rGH+IONP+H2O2.

NS: Not significant, ****: P≤0.0001, ***: P≤0.001, **: P≤0.01, * P≤0.05

The oxidation of methionine residues (M73, M102, and M124) was suppressed due to IONPs. However, M149 oxidation was not altered even in the presence of IONPs (Figure 2). This could be due to the location of methionine residues in the 3-D structure of rGH. M124 is a part of α-helix number 3 in rGH¹⁶, while other methionine residues are part of the non-helical polypeptide chains. In the case of M124, the 24-h sample containing IONPs differed significantly in the level of oxidation from the sample without IONPs (p≤0.0001). The H₂O₂-mediated oxidation of other methionine residues either was not significantly different (e.g. M149) in the presence of IONPs or was affected to a relatively lesser extent (e.g. M73, M102). These results indicate that the location of methionine residues in the 3-D structure of TP may alter the impact of IONPs on the extent of their oxidation.

After incubation with AAPH; M1, M5, and M149 were either not oxidized or much less oxidized compared to incubation with H₂O₂. However, in the case of M73, M102 and M124; similar results were observed as shown previously with H₂O₂-induced oxidation (Supplementary Data S10). Overall, the presence of IONPs reduced AAPH-induced oxidation of rGH.

The tyrosine residues (7 total) in rGH were oxidized by AAPH in the presence and absence of IONPs. The degree of oxidation was monitored using a fluorogenic derivatization method (Figure 3). Tyrosine oxidation was reduced in the presence of IONPs. Control groups such as buffer-IONPs, buffer, buffer-IONPs-AAPH were also studied (Supplementary Data S11). These controls confirmed that IONPs did not have any negative or positive impact on the fluorescence signal used in the assay for quantitation.

Figure 3:

Oxidation of tyrosine amino acids of rGH in presence of AAPH (10 mM) and IONPs (1.7 mM) after 3-h incubation at 37 °C

Caption: Samples were compared using 1-way ANOVA. The symbols above each column indicate comparison with rGH+IONP+AAPH. NS: Not significant, ****: P≤0.0001, ***: P≤0.001, **: P≤0.01, * P≤0.05

In another experiment, rGH was incubated with AAPH after removing the IONPs to monitor if the effect of IONPs on rGH was reversible. This study showed that even after removing the IONPs, tyrosine residues were not oxidized. Briefly, the IONPs were removed by filtration at the end of 3 h-incubation. Following that, the same amount of additional AAPH was added and incubated again for 3 h. A control sample was incubated without any additional AAPH. Differences between these 2 samples were not statistically significant as shown by ANOVA and Tukey’s multiple comparison test (Supplementary Data S12).

We considered that the reduced oxidation of rGH (Figure 3) in the presence of IONPs was due to loss of protein during the sample processing steps or due to binding to the IONPs. The possibility of protein loss of was eliminated by another experiment in which a rGH-IONP mixture and rGH alONe were filtered and the filtrate was measured for protein concentration using a BCA assay. There was no significant loss of the protein during filtration with and without IONPs (Supplementary Data S13). This suggests that the reduced oxidation of rGH seen in the presence of IONPs is due to IONP interaction and not due to loss of protein in the sample processing steps. In addition, the possibility of protein loss due to irreversible binding to IONPs was determined by incubating rGH with IONPs up to 24 h at room temperature and 37 °C. These samples were centrifuged (20,000×g) to remove the IONPs and the supernatant was measured for protein content (data not shown). After 24 h-incubation, there was minor protein loss (<10% at room temperature and <20% at 37 °C) due to irreversible binding to the IONPs. Thus, the interaction of IONPs with rGH was mostly reversible and minimum protein was lost due to the sample processing and irreversible binding with IONPs.

There is also the possibility that residual Fe⁺³ (ferric ions) may bind to rGH, stabilizing the protein structure to prevent oxidation even after removal of IONPs. Marra et al. reported the use of certain divalent metal ions (i.e. Zn⁺², Mg⁺² and Ca⁺²) to stabilize formulations of small peptides (10–30 amino acid) derived from sea snail venom.²⁹ However, these metal ions are not redox active, unlike ferric ions. In addition, Fe³⁺ binding usually occurs at specific sites in proteins which can provide the geometry necessary for complexation. Nevertheless, when Fe⁺³ are bound to these sites, they maintain their catalytic activity as shown by Stadtman ³⁰; therefore, any ferric ion chelation by rGH is unlikely to significantly reduce oxidation.

Soluble metal impurities from manufacturing processes can oxidize sensitive amino acids via metal-catalyzed reactions that may significantly impact the stability and shelf-life of TP formulations. AlONg with the soluble iron impurities, IONPs may be present as contaminants during TP manufacturing due to the iron shredding of manufacturing equipment.⁴ IONPs may also be shredded by the injection needle used during administration of TPs. Therefore, it is important to investigate the impact of IONPs on the oxidation of proteins. Whereas multiple studies have confirmed the iron-induced oxidation of TPs, our results show that the extent of H₂O₂- and AAPH-mediated oxidation of rGH is reduced in the presence of IONPs. This result is in marked contrast with the well-known effect of soluble iron impurities leading to increased oxidation of TPs.² Jayaram et al also reported that titanium dioxide nanoparticles generated ROS (hydroxyl radicals and superoxide radicals), which catalyzed oxidation of corona proteins around the nanoparticles.³¹

We did not measure the addition of soluble iron species (Fe³⁺ and Fe²⁺) to the protein solution from the dissolution of the IONP. The dissolution of iron oxide species is very slow except at extremes of temperature and pH³², so we expected a minimum contribution. The in vivo stability of IONP-based clinical contrast agents used for MRI imaging has been reported extensively, and complete dissolution of IONP was reported to require over 50 h at pH 0³³. As we observed an overall decrease in oxidized species with the IONPs, any oxidative contribution from IONP-derived Fe³⁺ and Fe²⁺ ions was minimum and overwhelmed by the opposing stabilizing effects of the IONPs.

FTIR studies showing IONP-induced changes in rGH secondary structure

The FTIR spectra of rGH were altered after incubation with various concentrations of IONPs, suggesting changes in the secondary structure of the protein. Incubation with 1.7-mM IONPs appears to lead to aggregation or intramolecular beta sheet formation resulting in the creation of new bands denoted as a and b, corresponding to 1627 and 1640 cm⁻¹ signals, respectively (Supplementary Data S14). Similar FTIR bands were reported previously as indicating protein aggregation (1620 cm⁻¹) and intramolecular beta sheets (1635 cm⁻¹).^34,35 Lower concentrations of IONPs (0.85 and 0.42 mM) lead to more subtle structural changes in the α-helical region (1653 cm⁻¹) (Figure 4 and Supplementary Data S14).³⁶ In the presence of 0.85 and 0.42 mM IONPs (1-and 2-h samples) the α-helical band is split into 2 bands at 1649 and 1657 cm^-1. This indicates that the interaction of rGH with the IONPs can lead to structural changes. In addition, there is an increase in intramolecular beta sheet indicated by increased intensity of the band around 1680 cm⁻¹ compared to pure rGH and the 0-h sample of the rGH-IONP mixture. The spectra of IONPs and phosphate buffer showed that they did not affect the FTIR bands of rGH (Supplementary Data S15).

Figure 4:

Second derivative FTIR spectra of rGH and rGH-IONP mixtures (0.85 mM IONPs) at 25 °C A) Unzoomed and B) Zoomed (1640–1700 cm⁻¹)

Caption: FTIR spectra were acquired as duplicates or triplicates. Mean representative spectra are shown. Blue: rGH alONe; Green: rGH+0.85 mM IONP (0 h); Black: rGH+0.85 mM IONP (1 h); Red:rGH+0.85 mM IONP (2 h)

In previously published reports, it was shown that the metal oxide nanoparticles can alter the secondary structure of TPs and enzymes. For example, zinc oxide nanoparticles increased the amount of α-helical structure of lysozyme as demonstrated by CD and FTIR.³⁷ In addition, airborne nanoparticles changed the expression of proteins by human bronchial epithelial cells.³⁸ Furthermore, alteration of human growth hormone structure due to metal ions (cobalt and iron) was reported previously.^39,40 Trace amounts of copper (>75 ppb) were responsible for carbonylation and aggregation of TPs.⁴¹ Formation of higher order clusters of the proteins and gold nanoparticles was also demonstrated.⁶ In another study, CD data showed alteration of secondary structure of Cytochrome c due to interaction with gold nanoparticles.⁴²

The rGH-IONP mixture and rGH were subjected to increasing temperatures from 15 to 85 °C and FTIR spectra were acquired (Figure 5). Both groups (with and without IONPs) showed alteration of FTIR spectra with increasing temperature indicating aggregation as manifested by increase in the 1620 cm⁻¹ peak. In the presence of IONPs, an additional band at around 1650 cm⁻¹ (marked with a dashed red line) was detected in the ‘y region’ (Figure 5). The intramolecular beta sheets are generally observed at 1635 and 1689 cm⁻¹ on the FTIR spectra.³⁴ In this study, band at 1620 cm⁻¹ was observed, which may indicate intermolecular beta sheet formation. Therefore, the transition temperature for rGH in presence of IONPs is monitored using the 2^nd derivative absorbance at 1620 and 1694 cm⁻¹ with increasing temperature from 15 to 85 °C (Supplementary Data S16). The transition temperature in the presence and absence of lONPs was similar (around 55 °C).

Figure 5:

FTIR spectra of A) rGH and B) rGH+IONP (1.7 mM) from 15 to 85 °C (x and y indicate altered regions due to IONPs, the dashed red line shows an additional feature observed at 1650 cm-1 due to IONPs)

Principal component analysis (PCA) on the FTIR spectra

Principal component analysis was used to differentiate the FTlR spectra generated using rGH and various combinations of the rGH-lONP mixture (Figure 6). The spectra of mixtures of rGH and lONPs based on various lONP concentrations (1.7, 0.85 and 0.42 mM) and incubation times (0, 10, 30 min, 1 and 2 h) were separated effectively using PCA. Principle component (PC)1 and PC2 were the major components with contribution of around 49 and 30%, respectively (Figure 6 A). lONger incubation (2.5 h) of rGH with 1.7-mM lONPs leads to aggregation and hence it was not included in the PCA. However, other time points (0, 10 and 30 min) of 1.7-mM lONP and rGH mixture were included. Pure rGH, 0-h (1.7, 0.85, 0.42 mM lONP and rGH) and 10-min (1.7 mM lONP and rGH) time points were in the same region of PC1 vs PC2 plot which constitutes the first cluster. There is a second cluster for the 1- and 2-h time points of rGH and 0.42 or 0.85 mM lONP mixtures. ln addition, the 30-min time point of rGH and 1.7-mM lONP mixture present in the third cluster. ln the case of PC3 vs. PC1 scores plot, pure rGH sample was separated from the other samples (Figure 6 B). However, the PC3 contribution was minor (10%) compared to PC1 and PC2 (Supplementary Data S17). Loading plots (Supplementary Data S18) indicate that PC1 had significant contributions to the α-helical FTlR spectral region (1645 to 1660 cm⁻¹).³⁶ ln addition, aggregation (1627 cm⁻¹) and intramolecular beta sheets (1640 cm⁻¹) region also contributed to PC1.

Figure 6:

Principal component analysis of FTIR spectra for rGH and rGH-IONP mixture at various concentrations of IONPs and incubation times A) Plot of PC1 and PC2, B) Plot of PC1 and PC3

Principal component analysis on the FTIR data acquired with increasing temperature from 15 to 85 °C showed two separate clusters for the rGH and rGH-IONP mixture (Figure 7). PC1 was the major component with contribution of around 82%, while PC2 had a lower contribution of around 18% (Supplementary Data S19). Loading plots indicate that PC1 and PC2 had contributions from the FTIR bands indicating aggregation (1627 cm⁻¹), intramolecular beta sheets (1640 cm⁻¹) and α-helical structure (1645 to 1660 cm⁻¹) (Supplementary Data S20). In this study, various amounts of IONPs were tested to analyze alteration of the FTIR spectra. PCA was successfully used to distinguish the FTIR spectra of different combinations of rGH-IONP mixtures and incubation time points. PCA may also be useful to distinguish various formulations and provide stability evaluation of TP formulations.

Figure 7:

Principal component analysis of FTIR spectra acquired for rGH and the rGH-IONP mixture at different temperatures (15 to 85 °C ramping with a 2.5 °C interval)

Circular Dichroism (CD) spectroscopy

CD spectroscopy was used to evaluate any change in rGH secondary structure due to IONPs. For this study, 3 IONP concentrations were tested (1.7, 0.85 and 0.42 mM). Higher concentration of IONPs (1.7 and 0.85 mM) interfered with the CD spectra (data not shown). Interference from 0.42 mM IONPs was minimal; therefore, this concentration was selected for further analysis. The spectra of rGH alONe and the rGH-IONP mixture were identical for lower temperatures (Supplementary Data S21); however, at higher temperatures (57.5, 60 and 62.5 °C) the CD spectra were different with the negative α-helical band at 208 nm clearly reduced in intensity (Figure 8). The thermal transition temperature (66 °C) calculated by measurements at 220 nm was similar for rGH samples with and without IONPs (Supplementary Data S22). The α-helical structure decreased with increases in temperature for both rGH and the rGH-IONP mixtures. For temperatures higher than 40 °C, the percent α-helical structure was similar between rGH and rGH-IONP mixture. However, the CD spectra at different temperatures showed minor changes (Figure 8).

Figure 8:

Comparison of CD spectra of rGH and rGH-0.42 mM IONP mixtures at selected temperatures A) 10 °C, B) 30 °C, C) 40 °C, D) 57.5 °C, E) 60 °C, F) 62.5 °C

The CD spectra of rGH and the rGH-IONP mixture were different at higher temperatures (Figure 8). These findings probably reflect a role of IONPs in the alteration of the secondary structure of rGH at high temperature. There was not, however, any significant difference in the CD-determined transition temperatures at various IONP concentrations. In the case of proteins having predominantly α-helical structure, 2 bands are usually observed at around 220 and 208 nm.²³ Similar bands were observed for rGH and rGH-IONPs mixtures. In addition, the CD spectra of rGH were similar to the previously published CD spectra acquired from human growth hormone, human pituitary growth hormone and ovine pituitary lactogenic growth hormone.^43,44 In addition, tryptophan fluorescence intensity was used to study effect of IONPs on the tertiary structure of the protein (Supplementary Data S23).⁴⁵ The oxidation and structural changes in the proteins can lead to alteration of their immunogenicity potential.^46,47 For example, oxidation-mediated aggregation elevated the immunogenicity of recombinant human interferon beta in the transgenic mice. In contrast, relatively non-oxidized, but aggregated protein did not show any immunogenicity.⁴⁸ hGH preparations with fewer and smaller aggregates were reported to be less immunogenic in mice⁴⁹, and patients administered heavily-aggregated hGH were reported to express more persistent antibodies compared to those administered less aggregated forms.⁵⁰ Overall; oxidation, deamidation, and aggregate formation are the major determinants of immunogenicity of TPs.⁵¹ In addition, it is important to investigate the impact of oxidation or other chemical degradation of TPs on their structure and vice versa.

CONCLUSIONS

This study found that IONPs (20 nm) can affect the degree of oxidation and secondary structure of rGH. In contrast to other reports of increased oxidation of proteins catalyzed by metal nanoparticles, the IONPs decreased H₂O₂- and AAPH-mediated oxidation. This could indicate that the protein corona that may form around the IONPs or the altered secondary structure of TPs may be protective against some forms of chemically induced oxidation. Furthermore, the relative rates of oxidation for amino acids in different locations changed, which means that IONPs may alter the composition of peptide epitopes derived from degraded TPs. Proteins are generally oxidized in the presence of soluble iron. However, rGH incubated with IONPs, was oxidized to a lesser extent when compared to control incubations without IONPs. Thus, IONPs generated from the manufacturing processes may impact the stability of TPs, as shown by the altered degree of oxidation and secondary structure. In addition, the utility of PCA was demonstrated on the categorization of different protein stress conditions.