Longitudinal serum proteomics analyses identify unique and overlapping host response pathways in Lyme disease and West Nile virus infection

Abstract

Advancement in proteomics methods for interrogating biological samples has helped identify disease biomarkers for early diagnostics and unravel underlying molecular mechanisms of disease. Herein, we examined the serum proteomes of 23 study participants presenting with one of two common arthropod-borne infections: Lyme disease (LD), an extracellular bacterial infection or West Nile virus infection (WNV), an intracellular viral infection. The LC/MS based serum proteomes of samples collected at the time of diagnosis and during convalescence were assessed using a depletion-based high-throughput shotgun proteomics (dHSP) pipeline as well as a non-depleting blotting-based low-throughput platform (MStern). The LC/MS integrated analyses identified host proteome responses in the acute and recovery phases shared by LD and WNV infections, as well as differentially abundant proteins that were unique to each infection. Notably, we also detected proteins that distinguished localized from disseminated LD and asymptomatic from symptomatic WNV infection. The proteins detected in both diseases with the dHSP pipeline identified unique and overlapping proteins detected with the non-depleting MStern platform, supporting the utility of both detection methods. Machine learning confirmed the use of the serum proteome to distinguish the infection from healthy control sera but could not develop discriminatory models between LD and WNV at current sample numbers. Our study is the first to compare the serum proteomes in two arthropod-borne infections and highlights the similarities in host responses even though the pathogens and the vectors themselves are different.

Keywords: Lyme disease; West Nile virus; acute-phase response; asymptomatic infection; immune system; localized and disseminated stage; longitudinal analysis; serum proteomics.

Figure 1

Protein Expression across Lyme Disease Timepoints. Serum samples from LD patients were assessed by the dHSP platform. The limma package was used to identify differentially expressed proteins in a combined dataset of healthy controls and patients with statistically significant proteins defined with FDR ¾= 0.05. (A) Normalized heatmap displays the progression of abundant proteins (n=46) in both disseminated and localized infections at month 0, 1, and 3 (M0, M1, M3) compared to healthy controls. (B) Normalized heatmap displays the progression of abundant proteins (n = 66) in localized infections at month 0, 1, and 3 (M0, M1, M3). (C) Normalized heatmap displays the progression of abundant proteins (n = 22) in disseminated infections at month 0, 1, and 3 (M0, M1, M3).

Figure 2

Strip plots of differentially expressed proteins between primary acute infection and recovery in Lyme disease. Serum samples from LD patients were assessed by the dHSP platform. Differentially expressed proteins were identified in a combined dataset of acute infection and convalescent time points. Patients with statistically significant proteins were defined with p value ¾= 0.05. Localized Lyme disease infections were statistically significant across time points for AMBP and C9. CD44 was statistically significant across time points in disseminated Lyme Disease infections, however, CFD was statistically significant by month 1.

Figure 3

LEGEND - Schematic representation of the overall study. (A) Diseases studied- Lyme disease (tick-borne) and West Nile Virus infection (mosquito-borne).(B) Blood collection by venipuncture and serum separation (C) Serum protein processing platforms– MStern spotting of one µl serum on PVDF microtiter membrane without depleting high abundant protein followed by protein denaturation, reduction and alkylation of cysteine residues, rapid protein digestion, LCMS (top panel) and dHSP with depletion of high abundance blood proteins using spin columns followed by protein denaturation, reduction and alkylation of cysteine residues, rapid protein digestion and LCMS shotgun proteomics (bottom panel) (D) Data analysis and data interrogation for marker discovery and underlying molecular mechanisms. Venn Diagram of the analyzed proteins from both the dHSP and MStern platforms in LD (E) and WNV (F). The optimized depletion-based method is an unbiased approach that was based on shotgun proteomics method that required 50 μl serum for depletion step and 1μg of tryptic peptides for LC/MS analysis. The MStern platform starts with 1μl serum on a 96 well-plate-based method with no depletion step required in WNV.

Figure 4

(A) Protein Expression across WNV Timepoints. Serum samples from WNV patients were assessed by the dHSP platform. The limma package was used to identify differentially expressed proteins in a combined dataset of healthy controls and patients with statistically significant proteins defined with FDR ¾= 0.05. A Normalized heatmap displays the progression of abundant proteins (n=38) in both symptomatic and asymptomatic infections at month 0, 1, and 3 (M0, M1, M3) compared to healthy controls. (B) Strip plots of differentially expressed proteins between primary acute infection and recovery in West Nile Virus. Serum samples from WNV patients were assessed by the dHSP platform. Differentially expressed proteins were identified in a combined dataset of acute infection and convalescent time points. Patients with statistically significant proteins were defined with p value ¾= 0.05. Both asymptomatic and infections were statistically significant across time points for C4B and CPB2.