Proteomic characterization of particle-protein coronas shows differences between osteoarthritic and contralateral knees in a rat model

Abstract

Objective: When synthetic particles are injected into a biofluid, proteins nonspecifically adsorb onto the particle surface and form a protein corona. Protein coronas are known to alter how particles function in blood; however, little is known about protein corona formation in synovial fluid or how these coronas change with osteoarthritis (OA). In this study, protein coronas were characterized on particles incubated within OA-affected or healthy rat knees.

Design: First, to evaluate particle collection techniques, magnetic polystyrene particles were placed in bovine synovial fluid and separated using either magnetics or centrifugation. In a second experiment, 12 male and 12 female Lewis rats received a simulated medial meniscal injury. At 2, 5, or 8 weeks post-surgery, operated and contralateral limbs were injected with clean magnetic particles (n = 8 per timepoint). After a 4-h incubation, animals were euthanized and particles were magnetically recovered. In both experiments, protein coronas were characterized using an Orbitrap fusion mass spectrometer.

Results: In the first experiment, the particle separation method affected the identified proteins, likely due to centrifugation forces causing some large proteins to spin-down with the particles. In the OA model, 300-500 proteins were identified in the particle-protein coronas with 35, 59, and 13 proteins differing between the OA-affected and contralateral limbs at 2, 5, and 8 weeks, respectively. In particular, plectin, a serine (or cysteine) proteinase inhibitor, and cathepsin B were more prominent in the particle-protein coronas of OA-affected knees.

Conclusions: Synthetic particles nonspecifically adsorb proteins in synovial fluid, and these binding events differ with OA severity.

Keywords: AMPK-ULK1-autophagy axis; Periodontitis; creatine; dental pulp stem cells; osteogenic differentiation.

Fig. 1.. Normalized spectral count visualization for two sample separation methods.

Scatterplots of the normalized spectral counts of samples separated and washed via centrifugation (y-axis) versus the normalized spectral counts of samples separated and washed via magnetic separation (x-axis). Samples were incubated and separated from the same lot of bovine synovial fluid. The dotted line is a parity line, representing proteins equally represented in the centrifuged and magnetically separated samples. The most abundant proteins have been labeled and listed to the right of Panel A. Orange circles in Panel B show nine proteins that were present in all 4 centrifuge-separated samples, but were not present in any magnetically-separated samples.

Fig. 2.. Normalized spectral count scatterplots.

Scatterplots of the normalized spectral counts of the operated limbs (y-axis) versus the normalized spectral counts of the contralateral limbs (x-axis) for each timepoint. Abundant proteins (> 5% of total corona composition) are annotated and labeled. Most proteins made up a small composition of the total protein corona (< 5%). All data shown are raw values. A list of annotated proteins is provided underneath the plots

Fig. 3.. Volcano plots for protein abundance in operated and contralateral limbs.

Volcano plots demonstrate fold changes in each protein between the operated and contralateral limbs at various stages of disease progression. Values above the horizontal dashed line represent p < 0.05 (Student’s t-test). Values on either side of the vertical dashed lines represent a normalized parity of greater than 0.33 and less than −0.33. Note, these parity lines are equivalent to a fold change of greater than 2 or less than −2, respectively (see Supp. Figure 1), but this analysis approach avoids zeros and undefined numbers that commonly occur with a fold change approach. Each protein that was significantly more prevalent or less prevalent is annotated and listed in the table below the graphs.

Fig. 4.. Scatterplots for protein change over time.

For proteins that were identified at more than one timepoint using the volcano plot approach in Figure 3, scatterplots were generated and assessed using an ANOVA approach with post-hoc Tukey’s HSD tests. Despite being identified in the volcano plots, no significant differences were found for anion exchange protein, plectin, or keratin associated protein 3–3. Serine proteinase inhibitor (clade H, member 1, isoform CRA_B) differed between OA-affected and contralateral controls at week 5 (p=0.0008). NADH ubiquinone oxidoreductase 75 kDa subunit differed at week 2 and week 5 (p<0.001). ATP synthase subunit e differed at week 2 and week 5 (p=0.0005). Cathepsin B differed at week 5 (p=0.0002). All significant differences between operated and contralateral limbs within a specific time point are denoted with a *. In addition, three of these proteins were found to change with a main effect of time, where NADH ubiquinone oxidoreductase 75 kDa subunit decreased with time(p=0.002), ATP synthase subunit e (p=0.001) decreased with time, and cathepsin B increased at weeks 5 and 8 compared to week 2; for clarity, the main effects of time are not noted on the graph. All data points are shown as raw values. Finally, while sex-effects were not considered in the statistical model, the data are denoted by different symbols to inform future experiments on this subject.

Fig. 5.. Heat map of Z-scores from PCA.

Each line represents an individual protein from mass spectrometry and is grouped by biological process as determined by DAVID. The heat map described the biological processes captured by each principal component.

Fig. 6.. Histology over time.

Frontal sections of the rat knee compartment for comparison of contralateral and operated joints following surgical induction of OA in males and females. The OARSI scores for all histological sections are provided for the graph below the representative images.