In Vivo Fiber Optic Raman Spectroscopy of Muscle in Preclinical Models of Amyotrophic Lateral Sclerosis and Duchenne Muscular Dystrophy

Abstract

Neuromuscular diseases result in muscle weakness, disability, and, in many instances, death. Preclinical models form the bedrock of research into these disorders, and the development of in vivo and potentially translational biomarkers for the accurate identification of disease is crucial. Spontaneous Raman spectroscopy can provide a rapid, label-free, and highly specific molecular fingerprint of tissue, making it an attractive potential biomarker. In this study, we have developed and tested an in vivo intramuscular fiber optic Raman technique in two mouse models of devastating human neuromuscular diseases, amyotrophic lateral sclerosis, and Duchenne muscular dystrophy (SOD1^G93A and mdx, respectively). The method identified diseased and healthy muscle with high classification accuracies (area under the receiver operating characteristic curves (AUROC): 0.76-0.92). In addition, changes in diseased muscle over time were also identified (AUROCs 0.89-0.97). Key spectral changes related to proteins and the loss of α-helix protein structure. Importantly, in vivo recording did not cause functional motor impairment and only a limited, resolving tissue injury was seen on high-resolution magnetic resonance imaging. Lastly, we demonstrate that ex vivo muscle from human patients with these conditions produced similar spectra to those observed in mice. We conclude that spontaneous Raman spectroscopy of muscle shows promise as a translational research tool.

Keywords: Amyotrophic lateral sclerosis; Duchenne muscular dystrophy; Raman spectroscopy; biomarker; muscle.

Figure 1

Intramuscular, in vivo, fiber optic Raman spectroscopy assessment. (a) A flowchart of the experiments. (b) A schematic of the fiber optic Raman system. (c) A mouse undergoing the procedure; the laser light can be appreciated within the hindleg as the camera used does not filter out the light. (d) Raman spectra obtained from muscle, blood, and bone. See Supporting Information Table 1 for tentative peak assignments and references.

Figure 2

Intramuscular, in vivo Raman spectra and muscle histology. (a, b) Average spectra are shown for mdx and SOD1^G93A, together with the relevant WT/NTg control for 30 and 90 days of age. Prominent peaks across the different groups are labeled (a and b). (c) Prominent Raman peaks in baseline subtracted spectra (present also in non-baseline subtracted spectra; Supporting Information Figure 2) and tentative peak assignments. See Supporting Information Table 1 for peak references. (d) Histological assessment at 30 days in mdx reveals necrotic fibers with inflammatory cells (arrow) and evidence of early regeneration (small myofibers with central nuclei, arrow heads). (e) The 90 day mdx muscle is characterized by regenerated myofibers, larger cells with central nuclei (arrow heads). (f) The 30 day SOD1^G93A muscle appears normal. The myofibers have peripheral nuclei and a regular shape. (g, h) The 90 day SOD1^G93A muscle displays evidence of denervation in the form of grouped atrophy (g, double arrow), small angular fibers (g, chevrons), and hypertrophic fibers (h, arrows) and centrally placed nuclei (g, h, arrowhead). (i) Normal myofibers from a 90 day WT mouse. Magnification for all images = x40, all scale bars = 100 μm. WT = wild type, NTg = nontransgenic.

Figure 3

Linear discriminant function histograms and loadings plots. (a) Linear discriminant function (LDF) histogram and associated loadings plot for the comparison between 90 day SOD1^G93A and NTg mice. (b) Tentative peak assignments for the 90 day SOD1^G93A vs NTg comparison. (c) LDF histogram and associated loadings plot for the comparison between 30-day mdx and WT mice. (d) LDF histogram and associated loadings plot for the comparison between 90 day mdx and WT mice. (e) Tentative peak assignments for the mdx vs WT comparisons. Peak references can be found in Supporting Information Table 1. Wn = wavenumber.

Figure 4

Post-in vivo Raman spectroscopy motor performance. Rotarod performance for 30 day SOD1^G93A mice for both Raman (a) and sham (b) experiments. At 90 days of age, a significant decline in performance was seen two weeks postprocedure (c). However, there was no difference in rotarod performance at this final age (104 days) between mice undergoing Raman, sham, and no procedure (d), suggesting a disease-related decline. No change in rotarod performance was seen at either 30 days (e, f) or 90 days (g, h) in the mdx mice. ** = P < 0.01. d = day; wks = weeks.

Figure 5

7T MRI evaluation of muscle that has undergone in vivo Raman spectroscopy. (a) Axial MRI (ex vivo) from NTg (healthy) mice. Muscles were studied at three different time points post-Raman measurement. Each panel represents a different mouse. The yellow arrows denote high T2 signal which may be due to postprocedure edema. (b) Axial MRI (ex vivo) from two time points following a sham procedure. Each panel represents a different mouse.

Figure 6

Human muscle spectra (ex vivo) and corresponding in vivo mouse model spectra. (a) Average ex vivo muscle spectra from patients with MND (orange) and in vivo spectra from 90 day-old SOD1^G93A mice (blue). (b) Average ex vivo muscle spectra from patients with DMD (orange) and in vivo spectra from 90 day-old mdx mice (blue).