Skeletal muscle biomarkers of amyotrophic lateral sclerosis: a large-scale, multi-cohort proteomic study

Abstract

Background: Biomarkers with clear contexts-of-use are important tools for ALS therapy development. Understanding their longitudinal trajectory in the untreated state is key to their use as potential markers of pharmacodynamic response. To this end, we undertook a large-scale proteomic study in well-phenotyped cohorts to identify biomarker candidates of ALS disease state and disease progression.

Methods: Clinical phenotypic data and biofluid samples, collected from patients with ALS and healthy controls through multiple longitudinal natural history studies, were used to identify biomarker candidates. SOMAmer (Slow Off-rate Modified Aptamer)-based relatively quantitative measurement of ~7,000 proteins was performed in plasma and CSF, with immunoassay validation of candidates of interest.

Results: We identified 329 plasma proteins significantly differentially regulated between ALS and controls (adjusted p-value <0.05), with 25 showing >40% relative abundance. PDLIM3, TNNT2, and MYL11 had the greatest log-fold elevation, while ANTXR2 and ART3 had the greatest log-fold reduction. A similar set of plasma proteins was found to increase (e.g. PDLIM3, TNNT2, MYL11) or decrease (e.g. ANTXR2, ART3, MSTN) with disease progression. CSF proteins with the greatest log-fold elevation included NEFL, NEFH, CHIT1, CA3, MYL11 and GPNMB. These results were confirmed in an independent replication cohort. Moreover, tissue-specific signature enrichment suggests a significant contribution of muscle as a source of these biomarkers. Immunoassays provided orthogonal validation of plasma TNNT2 and CSF GPNMB.

Conclusion: We identified an array of novel biomarkers with the potential to serve as response biomarkers to aid therapy development, as well as to shed light on the underlying biology of disease.

Keywords: context of use; disease progression biomarkers; disease state biomarkers; monitoring biomarkers; pharmacodynamic response biomarkers.

Figure 1.. Plasma and CSF markers differentially regulated in ALS vs. healthy controls

(A-D) Plasma markers. (A) Volcano plot of ALS compared to healthy controls. SOMAmers in red have intensities that are statistically different. (B-C) Boxplots showing normalized age-, sex- and plate-adjusted SOMAmer intensities across controls and ALS for ART3 and PDLIM3. (D) Boxplot showing normalized age-, sex-, and plate-adjusted SOMAmer intensities for PDLIM3, in controls and ALS subgroups based on estimated disease progression rate. (E-I) CSF markers. (E) Volcano plot of ALS compared to healthy controls. SOMAmers in red have intensities that are statistically different. (F-I). Boxplots showing normalized age-, sex-, and plate-adjusted SOMAmer intensities for NEFL, CHIT1, MYL11 and PDLIM3, in controls and in ALS subgroups based on estimated progression rate. Adjusted P-value <0.001 denoted by ***; 0.001–0.01 denoted by **; 0.01–0.05 denoted by *; and 0.05–0.1 by “x”. RFU = relative fluorescence units. Seq.number, annotations of SOMAmers

Figure 2.. Monitoring biomarkers in plasma (A-C) Disease progression biomarkers.

(A) Volcano plot showing adjusted p-values vs. change of log2 SOMAmer intensity per 1-point change in ALSFRS-R, estimated by LME. Negative values signify negative correlation, with protein (e.g. TNNT2, PTN, PDLIM3) levels increasing as ALSFRS-R declines (i.e. disease progresses). Positive values signify positive correlation, with protein (e.g. ANTXR2, CLEC3B, CRTAC1) levels decreasing as ALSFRS-R declines. SOMAmers marked in red are significantly associated with ALSFRS-R, with an adjusted p-value of <0.05. (B-C) Boxplots (comparing ALS to controls) and spaghetti plots (SOMAmer intensity vs. ALSFRS-R) of ANTXR2 and TNNT2, as two illustrative examples. (D-F) Temporally stable biomarkers. (D) Heatmap showing regression coefficients for top 10 examples of proteins, whose matching SOMAmers meet the dual criteria of having the greatest log-fold change in SOMAmer intensity in ALS compared to controls AND remaining relatively stable (i.e. no evidence for significant change) over time and across disease progression. The left panel shows log2FoldChange indicating difference of mean protein level in plasma of ALS vs control, while the right panel shows regression coefficients (Estimate) for change of protein level on log2 scale per 1 month disease duration since baseline and the same change per 1 unit of ALSFRS-R (2^nd and 3^rd columns, respectively). (E-F) Boxplots (comparing ALS to controls) and spaghetti plots (SOMAmer intensity vs. ALSFRS-R) of MYOM2 and ACTN2, as two illustrative examples. RFU = relative fluorescence units. padj = adjusted p-values.

Figure 3.. Plasma biomarkers are encoded by muscle-enriched genes

Left and right panels show results of GTEX- and Human Protein Atlas (HPA)-based enrichment analysis. (A-C) Heatmaps showing scaled mean transcript per million (TPM) values for each organ for genes encoded by plasma biomarkers selected by lowest adjusted p-value. Sidebars on the right indicate effect size based on log2FC in SOMAmer intensity, comparing ALS vs. controls (A, C) and per unit change in ALSFRS-R (B). (D-F) Heatmaps showing scaled normalized TPM values for plasma biomarkers selected by lowest adjusted p-value in the selected cell types. Sidebars on the right indicate effect size based on log2FC in SOMAmer intensity, comparing ALS vs. controls (D, F) and per unit change in ALSFRS-R (E). The log2FC value for MSTN corresponds to MSTN|GDF11 SOMAmer seq.2765.4 (E).

Figure 4.. Gene ontology-based enrichment

Skeletal muscle signature in plasma is among the top enriched gene sets differentiating ALS vs. controls, while immune/inflammatory features are among the top signatures associated with disease progression. Enrichment plots show top 5 upregulated and top 5 downregulated Gene Ontology (GO) terms and cell type-specific signatures in plasma SOMAmers that are differentially regulated in ALS vs. controls (A, B), and associated with ALSFRS-R (C, D). NES = normalized enrichment score. Dot size represents the number of proteins overlapping with a gene signature. Note that the sign for NES in A and B coincides with the directional change of protein level in ALS vs. controls, with a positive NES indicating upregulation of a given gene signature, and a negative NES indicating depression of a given gene signature. By contrast, the sign for NES in C and D coincides with the ALSFRS-R, with a negative NES reflecting genes associated with a lower ALSFRS-R (i.e. more advanced disease).