Fig 1.
Blood serum samples from subjects that had physician ordered RA panels were used (n = 50). Based on clinical diagnosis from medical professionals, the samples were separated into an RA group (n = 27) and a non-RA group (n = 23). DSC was used to obtain the HDCs, and a characteristic shift was seen between groups. LC/MS-MS experiments were performed to determine the mechanism behind this shift. Quantification and surface amino acid reactivity analyses were performed to determine significant differences in serum proteins between RA and non-RA groups.
Fig 2.
Two possible models explaining HDC shifts in RA subjects.
The decreased intensity of the lower temperature peak in RA samples could be explained by A) a relative HSA concentration decrease, reducing HDC signal intensity, or B) a shift in thermal stability for a portion of the HSA population, shifting the HSA peak to the right, decreasing the first peak’s intensity, and increasing that of the second peak. This thermal stability could be the result of altered binding partners. The structural changes can be seen through biased detection of surface modifications on HSA. When cargo is unbound (top right, light green protein), binding sites are surface accessible for modification, when cargo is bound (top right, dark green protein), these sites are occupied, reducing surface accessibility and ability of these sites to be modified. These structural and functional differences at the molecular level could be the explanation for the observed shift in HDC.
Fig 3.
This study focuses on the two peaks observed between 55 and 75˚C of the heat denaturation curve. (A) The average normalized HDC curve for non-RA and RA samples, with the difference between the two shown in black. The first peak from HSA is consistently found around 63°C (low temp peak) and the second Ig peak is always around 71°C (high temp peak). Inset for A shows the distribution of peak ratios from the HDC of RA and non-RA subjects. The difference in peak ratio between the non-RA and RA groups is statistically significant (p = 0.007) (B) The distribution of peak ratio of all samples, with a peak ratio threshold of 1.00 as the cut-off between the HPR and LPR groups.
Fig 4.
Protein concentration differences in RA.
Volcano plots indicating the fold change and p-value for all 421 detected proteins, comparing A) RA/Non-RA samples and (B) LPR/HPR samples. The top eight most abundant proteins are indicated in green (all insignificant), and the statistically significant proteins (-1 > log2(fold change) > 1, p-value < 0.05) are indicated in red. C-reactive protein is the only significant protein in both plots. The fold change is calculated in each comparison by dividing RA abundance by non-RA abundance and LPR abundance by HPR abundance for all proteins that were detected in all samples.
Fig 5.
(A) The chemical reaction of the AEBSF modification on Serine. The reaction is similar for other nucleophilic amino acids. (B) The heatmap generated with PNNL Inferno showing the intensity differences of AEBSF modification at different HSA sites between different samples. The AEBSF modification amino acid number for HSA is listed on the y-axis, and the serum sample number on the x-axis. The samples are separated into 4 groups according to the hierarchy branch of serum samples, from left to right (S5 Data). Group L1 and L2’s AEBSF modifications are less intense than group H1 and H2 (L stands for lower intensity and H stands for higher intensity). The three clusters, C1 (green), C2 (purple), and C3 (black), are the most intense AEBSF modification clusters and are examined to characterize the modification further. (C) The bar graph shows the number of RA/non-RA and HPR/LPR samples expected in each AEBSF modification group (L1, H1, L2, H2). The percentage of RA samples in L1, H1, L2, H2, is 73%, 25%, 69%, and 31% respectively. For LPR, it is 67%, 89%, 73%, and 50%, respectively.
Fig 6.
Characterizing AEBSF modification site of the 3 clusters (C1, C2, C3) on HSA.
(A) A representative HSA crystal structure (PDB ID: 1N5U) with the 3 AEBSF modification sites clusters colored. The three clusters, C1, C2, and C3 are colored in blue, red, and green, respectively. Individual C1 and C2 sites that are significantly less accessible in RA HSA are labeled. The red oval indicates a C3-rich region in domain I that could be a plausible binding site for RA-specific interactions that most likely to increase HSA stability. (B) The HSA structure is colored by its 3 domains (I, II, III), and subdomains (A, B). The number of AEBSF modification sites regardless of cluster designation in each subdomain is listed in parentheses. The 9 known cargo binding pockets are shown in the gray circles, and the two drug binding sites, Sudlow I & II, are shown by an arrow. The four bottom-right panels show what percentage of the AEBSF sites in each cluster (C) is in each HSA domain, (D) is on each amino acid residue, (E) has each secondary structure (Helix (H) or Loop (L)), and (F) the average SASA score of each cluster. Only SASA scores between C1 and C2 are statistically different (p = 0.019).
Table 1.
Each of the AAmod sites for groups C1, C2, and C3.
Table 2.
A summary of non-RA and RA subjects versus literature for healthy subjects.