MicroRNAs as Biomarkers of Charcot-Marie-Tooth Disease Type 1A

Abstract

Objective

To determine whether microRNAs (miRs) are elevated in the plasma of individuals with the inherited peripheral neuropathy Charcot-Marie-Tooth disease type 1A (CMT1A), miR profiling was employed to compare control and CMT1A plasma.

Methods

We performed a screen of CMT1A and control plasma samples to identify miRs that are elevated in CMT1A using next-generation sequencing, followed by validation of selected miRs by quantitative PCR, and correlation with protein biomarkers and clinical data: Rasch-modified CMT Examination and Neuropathy Scores, ulnar compound muscle action potentials, and motor nerve conduction velocities.

Results

After an initial pilot screen, a broader screen confirmed elevated levels of several muscle-associated miRNAs (miR1, -133a, -133b, and -206, known as myomiRs) along with a set of miRs that are highly expressed in Schwann cells of peripheral nerve. Comparison to other candidate biomarkers for CMT1A (e.g., neurofilament light) measured on the same sample set shows a comparable elevation of several miRs (e.g., miR133a, -206, -223) and ability to discriminate cases from controls. Neurofilament light levels were most highly correlated with miR133a. In addition, the putative Schwann cell miRs (e.g., miR223, -199a, -328, -409, -431) correlate with the recently described transmembrane protease serine 5 (TMPRSS5) protein biomarker that is most highly expressed in Schwann cells and also elevated in CMT1A plasma.

Conclusions

These studies identify a set of miRs that are candidate biomarkers for clinical trials in CMT1A. Some of the miRs may reflect Schwann cell processes that underlie the pathogenesis of the disease.

Classification of Evidence

This study provides Class III evidence that a set of plasma miRs are elevated in patients with CMT1A.

The heritable peripheral neuropathies known as Charcot-Marie-Tooth disease (CMT) are the most common genetic neuromuscular disease, affecting 1:2,500 individuals.^1,2 The most common form is CMT1A, which is caused by a duplication of the PMP22 gene.^3,4 Efforts are ongoing to develop therapeutic approaches for CMT1A^5-8 and to develop appropriate clinical outcome measures and informative biomarkers.^9-11

The progression of CMT1A results in muscular atrophy, and it has been demonstrated that the intramuscular fat accumulation (IMFA) within calf muscles increases ∼1–2% per year in patients with CMT1A.^12,13 To complement IMFA assessments, it will be ideal to utilize patient plasma as an accessible and minimally invasive source of biomarkers. Indeed, plasma neurofilament light (NfL) from degenerating axons is elevated and correlated with disease severity in patients with CMT1A.^9,10 However, neither of these are direct biomarkers of the primary affected Schwann cells (SC) in CMT1A.

Given the muscular atrophy in CMT1A,^12,13 it may be that muscle microRNA (miR) biomarkers^14,15 (termed myomiRs) could be useful for CMT, and it is possible that miRs could be released from CMT1A Schwann cells. To investigate novel tools for clinical trial design, we determined whether miRs from muscle or Schwann cells could serve as pharmacodynamic assays of abnormal Schwann cells and muscular atrophy in CMT1A. Assays of plasma miRs could potentially provide CMT1A biomarker monitoring without invasive nerve or muscle biopsies.

Methods

This study was designed to determine whether miRs from muscle and Schwann cells are elevated in plasma of patients with CMT1A. This study provides Class III evidence that a set of plasma miRs are elevated in patients with CMT1A.

Standard Protocol Approvals, Registrations, and Patient Consents

Patients with CMT1A were identified and evaluated in the Inherited Neuropathy Consortium (INC) clinic in the Department of Neurology at the University of Iowa and the National Hospital for Neurology and Neurosurgery, London, UK. Institutional review board approval was obtained from the University of Iowa. Written informed assent/consent was provided by participants under protocols approved by the ethics board of the NIH Rare Diseases Clinical Research Network (protocol INC6601) and the National Hospital for Neurology and Neurosurgery Research Ethics Committee/Central London REC 3 09/H0716/61. All of the patients are followed in the INC natural history studies and most of the patients were enrolled in the recently published natural history study of CMT1A.¹⁶ The sample cohorts include many samples used in a previous study of protein biomarkers.¹⁰

Patients were diagnosed with CMT1A on the basis of clinical evidence of sensory or motor peripheral neuropathy (including length-dependent sensory loss, weakness and atrophy of the distal musculature, and decreased deep tendon reflexes), nerve conduction studies, and confirmatory genetic testing for the PMP22 duplication in the patient or affected first-degree relatives. Patients and normal controls provided two 6-mL EDTA-containing tubes of blood during their visit for plasma extraction,¹⁷ which was performed within 15 minutes of the blood draw and stored in aliquots at −80 °C. All participants were examined clinically by investigators who were certified by the INC for the proper administration of the CMT Neuropathy Score (CMTNS) version 2, a validated 9-item, 36 composite score based on patients' symptoms (3 items), examination findings (4 items), and electrophysiology (2 items).¹⁸ CMT Examination Score (CMTES) version 2 scores were also calculated, which included the 7 items of patients' symptoms and examination findings in the CMTNSv2 but excluded the physiologic results. Thus, the CMTES has a maximum score of 28 rather than 36 points.¹⁸ These scores were then subjected to Rasch modification to generate CMTNS-R and CMTES-R.¹⁹

The controls were age-matched, with a similar average age (table 1) and similar male/female distribution. In some cases, controls are unaffected family members of patients with CMT1A who have been demonstrated to not have CMT. None of the controls have acquired neuropathy.

Table 1.

Demographics

Nerve Conduction Studies

Ulnar motor nerve conduction velocities (MNCVs) were performed by standard techniques²⁰ with recording over the belly of the abductor digiti minimi with stimulation at the wrist and below the elbow. Maximum compound muscle action potential (CMAP) amplitudes were recorded using baseline to peak measurements in mV.

Expression Profiling of Plasma miRs

Frozen CMT1A patient plasma samples (n = 10) and normal control plasma samples (n = 10) were submitted to Qiagen for miR profiling by Illumina next generation sequencing. QIAseq miRNA library kit (1 × 75 bps) (331505) was used for the library construction, and 10–20 million (unpaired) reads per sample were acquired for the expression analysis. A total of 2,656 miRs were profiled. The level of each miR is reported as normalized counts of unique molecular indices.

MicroRNA Isolation From Plasma

miRs were extracted from plasma samples using miRCURY RNA Isolation Kit–Biofluids (Exiqon; catalog 300112) or by Qiagen purification kit (217204). Thawed samples were centrifuged at 3,000g for 5 minutes; then 200 μL of supernatants were used to isolate RNA using the manufacturer's protocol; final elution volume is 50 μL water.

miRNA Quantitative RT-PCR on Biomark HD Platform

The custom reverse transcription (RT) primer pool and preAmp primer pool were made according to the protocol for creating custom RT and preamplification pools using TaqMan MicroRNA Assays (Life Technologies). Ten microliters of each individual 5× RT primer of the 24 candidate miRNAs were combined, and an RT primer pool was made by adding TE buffer (10 mM Tris, 1 mM EDTA) to 1 mL with each primer at a final concentration of 0.05×. Five microliters of each 20× TaqMan MicroRNA assay for the 24 candidate miRNAs were combined with 380 μL TE buffer, resulting a 500 μL PreAmp primer pool with each assay at a final concentration of 0.2×.

The quantitative PCR (qPCR) procedure was a modification of a Fluidigm protocol (Fluidigm Corporation; MicroRNA Real-Time PCR Using Dynamic Array IFCs, PN 100–1616 C1). The 10-μL RT reaction was composed of the following (μL): RT primer pool (0.05×) 4.0, 100 mM dNTPs (with dTTP) 0.2, multiScribe reverse transcriptase (50 U/μL) 1.5, 10× reverse transcription buffer 1.0, RNase inhibitor (20 U/μL) 0.13, nuclease-free water 0.17, total RNA 3.0. The RT reaction was performed in a thermocycler: 16°C for 2 minutes, 42°C for 1 minute, and 50°C for 1 second, repeating these 3 steps for 40 cycles; then 85°C for 5 minutes; then holding at 4°C. Five microliters of preamplification mixture contained the following (μL): TaqMan PreAmp master mix (2×) 2.5, PreAmp primer pool 1.0, RT product 1.5. The preamplification reaction was performed using the following condition: 95°C for 10 minutes, 55°C for 2 minutes, 72°C for 2 minutes, 18 cycles of (95°C for 15 seconds, 60°C for 4 minutes), 99.9°C for 10 minutes, then hold at 4°C. The preamplification products were diluted 10-fold by adding 45 μL of 0.1 × TE to 5 μL of preamplification product and used for subsequent real-time qPCR run on the Biomark High-Density platform.

The 10× assays were prepared by combining 3 μL of 20× TaqMan miRNA assay (Life Technologies) and 3 μL of 2× assay loading reagent (Fluidigm 100–7,611), resulting in final concentration (at 10×) of primers at 9 μM and probe at 2 μM. Samples were prepared by combining 3 μL of 2× master mix (TaqMan Fast Advanced Master Mix, Life Technologies 4,444,557), 0.3 μL of 20× GE Sample Loading Reagent (Fluidigm 100–7,610), and 2.7 μL of 10× diluted preamplification product. For the Fast TaqMan assays (Biomark High-Density only, 100–6,174 C1), 3 μL of each assay and 3 μL of each sample were loaded into the respective inlets on the 192.24 IFC. RT-qPCR results, which are expressed as raw Ct values, were normalized to 3 reference miRNAs: miR-16, miR-103, and miR-30e. The relative expression (fold change) was calculated using the 2^−ΔCt method.

Bioinformatic Analysis

Potential targets regulated by each candidate CMT1A upregulated plasma miR were extracted from miRTarBase,²¹ which is a curated database of experimentally validated miR–target interactions. Pathway enrichment analysis for those putative miR targets were done using Ingenuity Pathway Analysis (Qiagen Inc., 2020).

Statistical Analysis

Data were analyzed using GraphPad Prism v7.03 using unpaired t tests. Pearson correlation coefficients were calculated for each marker relative to neuropathy score (CMTES-R, CMTES-N, ulnar CMAP, ulnar MNCV). The transmembrane protease serine 5 (TMPRSS5), NfL, and miRNA-206, -133a, and -223-3p receiver operating characteristic (ROC) curves and area under the curve (AUC) calculations were generated using GraphPad Prism v7.03.

Data Availability

All data are available by request to qualified investigators, and are available at Dryad (doi.org/10.5061/dryad.nvx0k6drx).

Results

Identification of Elevated Plasma miRs in CMT1A by miR Profiling

The goal of this study was to identify molecular markers of Schwann cell abnormalities or denervated muscle that could serve as biomarkers for clinical trials. MiRs are small endogenous RNAs that target specific mRNAs for degradation and repression of translation. MiRs are secreted by many cell types and have already been used as biomarkers in several disease states.^14,15

To identify miR biomarkers for CMT1A, we initially performed a next-generation sequencing (NGS)–based broad miRNA profiling on plasma samples obtained from 10 patients with CMT1A and 10 controls (table 1). A total of 35 miRNAs were elevated in CMT1A plasma compared to controls (average >1.8 fold, all p < 0.05; table 2). Eight of these were low abundance. Sixteen miRs were decreased (average <−1.8-fold change, all p < 0.05) but these were generally at low levels even in normal samples.

Table 2.

Next-Generation Sequencing Profiling of Plasma MicroRNAs (miRs)

One myomiR (miR-206)^15,22 showed higher levels in CMT1A samples (1.75-fold, p < 0.05). While myomiRs may be released as a function of muscular atrophy, we also assessed miRs potentially linked to CMT1A pathogenesis, which is driven by Schwann cells and degenerating axons. Eight of the miRs with higher levels in CMT1A plasma (table 2) are highly expressed in expression profiles of peripheral nerve.²³ Schwann cell enriched miRs that were elevated included miR-143, -150, -199a-3p, -223-3p, -328-3p, -342-3p, -409, and -431. The expression levels of genes and miRs in Schwann cells is highly axon-dependent and activated/repressed after peripheral nerve injury.^23,24 The designated Schwann cell miRs were also induced or repressed >2-fold at 3 days or 7 days after injury in mouse nerve.²³

Confirmation of Increased miRs in CMT1A Plasma

We analyzed 21 candidate miRs using an orthogonal, higher throughput Biomark HD qRT-PCR platform in a larger cohort of 109 CMT1A and 52 control plasma samples (table 3), which included all samples from our initial broad profiling. Expression levels of these 21 miRs were normalized to combined levels of 3 reference miRNAs: miR-16, -103, and -30e,²⁵ which were found to be stable in the NGS profiling. The 21 candidate miRNAs included the top ranked 12 (by fold change and p value) and the 4 lower ranked miR-224-5p, -744-5p, -409-3p, and -151a-3p from the initial screen (table 2). The panel included the previous 8 Schwann cell–enriched miRs, and miR338-3p, known also to be enriched in Schwann cells,^23-31,33 as well as the 4 myomiRs (miR-1, -133a, -133b, and -206).^14,15 The data (table 3) demonstrated increased levels of all 21 of the candidate miRNAs (average fold change >1.2; p < 0.05) in this larger patient data set, with the largest fold change of 4.1-fold for miR-206 in CMT1A samples compared to controls. The myomiRs are among the most highly and significantly elevated in CMT1A plasma (miR-206, -133a, -133b, and -1; p < 0.01).

Table 3.

Quantitative PCR Analysis of Plasma MicroRNAs (MiRs)

All 9 miRs expressed in Schwann cells were also significantly elevated in CMT1A (figure 1). MiR-338 showed the highest -fold increase (2.9×) in CMT1A samples, but the p value was less significant compared to other elevated miRs. MiR-338 is a low abundance miR in both normal and disease samples, as it was not detected in a significant proportion of samples. This likely accounts for why it was not identified as significant in the initial profiling. Even if the initial profiling samples are excluded from the analysis, most miRs are still significantly elevated in CMT1A samples, with the exception of 4 more modestly affected miRs that are no longer significantly different: miR-339-5p, -361-3p, -625, and -143.

Figure 1. MicroRNA (MiR) Levels in Charcot-Marie-Tooth Disease Type 1A (CMT1A) vs Control Plasma.

A total of 21 candidate miRs were analyzed using quantitative real-time PCR platform in a cohort of 109 CMT1A and 52 control plasma samples (table 2). Expression levels of these 21 miRs were normalized to combined levels of 3 reference miRNAs: miR-16, -103, and -30e; negative delta Ct is shown as a scatterplot for control and CMT1A samples. The median and interquartile range are shown.

Correlation Analysis of miR Levels

The miR measurements in CMT1A compared to control samples can be assessed for their ability to discriminate patients from controls using an ROC curve that plots sensitivity vs specificity (figure 2). For comparison, we include the plots for the NfL and TMPRSS5 proteins analyzed earlier from the same sample set.¹⁰ The AUC for the most significantly elevated myomiR (miR-206, 0.85) was comparable to that of TMPRSS5 (0.87) and higher than NfL (0.8), demonstrating good specificity and sensitivity for miR-206. For comparison, the AUCs for another myomiR (miR-133a, 0.77) and an SC-associated miRNA (miR-223-3p, 0.74) are also shown. Although muscular atrophy in CMT occurs more slowly than in muscular dystrophy, these data demonstrate that the myomiRs are nonetheless clearly elevated in CMT1A samples.

Figure 2. Receiver Operating Characteristic (ROC) Analysis of Charcot-Marie-Tooth Disease Type 1A (CMT1A) Associated MicroRNAs (MiRs).

ROC plots are shown for selected myomiRs (-206 and -133a) and a Schwann cell–associated miR (-223-3p) in comparison with the transmembrane protease serine 5 (TMPRSS5) and neurofilament light (NfL) levels from the same sample sets. Area under the curve data are shown below the graph.

Correlations of miRs to Age, Neuropathy Score, and Other Biomarkers

We determined the correlation between the miRs to each other, to age, and to markers of disease severity such as clinical outcome assessments (COAs) and neurophysiology. Because almost all of the same samples had been used in a prior investigation,¹⁰ we also determined the correlation with the plasma protein biomarkers, NfL and TMPRSS5 (figure 3).

Figure 3. Correlation Analysis of Plasma MicroRNA (MiR) With Protein Biomarker and Clinical Measurements of Neuropathy.

The 2 panels show Pearson correlation analysis of plasma miRs with age, 2 protein biomarkers (neurofilament light [NfL] and transmembrane protease serine 5 [TMPRSS5]), neuropathy scores (Charcot-Marie-Tooth Examination Score [CMTES] and Charcot-Marie-Tooth Neuropathy Score [CMTNS]), and electrophysiologic measures (nerve conduction velocity [NCV] and compound muscle action potential [CMAP]) in (A) control and (B) Charcot-Marie-Tooth disease type 1A (CMT1A) sample sets. Red and blue shading show those correlations that are positive or negative, respectively. Red numbers indicate those correlations where p < 0.05.

In contrast to the documented positive correlation of NfL with age in controls,¹⁰ there was little change of myomiRs with age for controls except for a modest positive correlation with miR-206. In controls, the myomiRs were not correlated with NfL, except a modest positive correlation for miR-133b. Most miRs were not correlated to TMPRSS5 except modest positive correlations with miR-223-3p, -28-3p, -199a-3p, -152, and -143.

However, in CMT1A samples, there were significant modest positive correlations of all the myomiRs with age. Similar modest positive correlations of some myomiRs, except miR133b, were found with neuropathy scores (CMTES and CMTNS). There was a modest negative correlation for miR-133a and -1 with CMAP amplitudes, but not nerve conduction velocities (NCVs). Although myomiR-206 was the most highly elevated miR in CMT1A, it was not significantly correlated with NfL. Of the myomiRs, miR-133a was the most abundant (figure 1) and its levels correlated to a significant degree with NfL, CMTES, and CMTNS, and negatively correlated with CMAP. Because elevated NfL and reduced CMAP amplitudes are markers of axonal degeneration, this is consistent with a causal relationship between axonal degeneration and muscle atrophy. None of the myomiRs correlated with the NCVs that reflect demyelination. TMPRSS5 and NCVs do not correlate with disease severity in CMT1A.¹⁰ While there are variable correlations of myomiRs with other clinical outcome measures, the myomiRs are highly correlated with each other. We did assess whether there was any correlation of miR levels with sex (not shown), and found only one (miR-150) that was significantly elevated in male compared to female patients with CMT1A (1.41-fold change, p = 0.039), but the remainder were not significantly (p < 0.05) different.

In control samples, the 9 putative Schwann cell miRs did not correlate with age or NfL (except a modest positive correlation for miR-223-3p); miR-223-3p, -199a, and -143 had modest positive correlations with TMPRSS5. In CMT1A, these miRs did not correlate with age (except a modest positive correlation for miR-431), NfL, CMTES, or CMTNS. Only miR-328 and -223-3p had modest negative correlations with CMAP. In contrast, 6 miRs correlated to TMPRSS5, with a higher positive correlation of 0.6 for miR-431, and modest positive correlations to miR-409-3p, -328, -223-3p, 199a-3p, and -143. Therefore, the putative Schwann cell miRs are elevated in CMT1A independent of disease severity, and there is a link between elevated Schwann cell miRNA and TMPRSS5 in CMT1A plasma. Among the 9 putative Schwann cell miRs, there are significant correlations that can be observed in both control and CMT1A sample sets.

Of the 8 novel miRs in figure 3 that are not myomiRs or known Schwann cell miRs, none correlated with age, or NfL, and only miR-28-3p and -151 modestly positively correlated with TMPRSS5 in controls. One of these 8 (miR-223-5p) arises from alternate processing of the pri-miRNA that produces the Schwann cell–associated miR-223-3p mentioned above, but is lower in abundance compared to 3p in plasma (figure 1) and in mouse peripheral nerve.²³ In CMT1A, none of these miRs correlated with age, NfL, or CMTES/CMTNS (except miR-224), but 3 (miR-28-3p, -744, -151) negatively correlated to CMAP. Notably, 5 of these positively correlated to TMPRSS5 (miR-28-3p, -744, -151, -339-5p, -625), and there were strong correlations among them and also with the previously defined Schwann cell miRs in CMT1A samples. While these 5 did not meet our original criteria to be assigned as Schwann cell miRs, the positive correlations with TMPRSS5 and the designated Schwann cell miRs indicate that these miRs may be considered Schwann miRs, although they are generally less abundant in plasma (table 2 and figure 1).

Determining Longitudinal Consistency in Plasma Samples

For several of the CMT1A plasma samples, we had longitudinal follow-up samples that were collected within 1–2 years of the first sample. This allowed us to test whether the levels of specific miRNAs were consistent from one collection to the next, as shown in figure 4A. In general, the delta Ct values varied no more than 1 cycle for most of the miRNAs shown here, and none of the changes was statistically significant. However, miR-338 had an apparently larger variation, although its levels are low, making the measurements somewhat more variable.

Figure 4. Reproducibility and Predicted Targets of Charcot-Marie-Tooth Disease Type 1A (CMT1A)–Associated MicroRNAs (MiRs).

(A) The graph shows miR levels in repeat samples (from 22 CMT1A cases) that had been obtained from patients within 1–3 years from the initial sample collection. The levels of the miRs are shown to compare the miR levels between initial and follow-up samples. (B) Predicted targets of the Schwann cell–associated miRs were analyzed by gene ontology to identify statistically enriched pathways.

Gene Ontology of miR Target Genes

The predicted targets of the Schwann cell microRNA cluster were assessed by gene ontology analysis and the most highly enriched pathways are shown in figure 4B. Interestingly, several of the target gene categories are related to PI3 kinase/AKT signaling, including IGF signaling, HER2 pathway, and PTEN pathways. Predicted targets in these pathways include AKT1, CCND1, CDC42, PIK3CA, and FOXO3. HER2/neuregulin signaling plays an important role in Schwann cell development.²⁶ Of relevance, neuregulin signaling and the balance of PI3 and ERK signaling pathways have emerged as a therapeutic target in preclinical studies of CMT1A and other neuropathy models.^27,28

Discussion

To support clinical trials for CMT, composite scoring systems have been developed based on symptoms and neurologic examination of patients with CMT1A: R-CMTNS and R-CMTES.²⁹ However, although the CMTNS has been used as a primary outcome in clinical trials,^30,31 the annual change is small and several hundred patients would be needed in a double-masked clinical trial to detect significant slowing of disease progression using either of these disease scoring instruments as primary outcome measures,¹⁶ and therefore additional CMT1A biomarkers are being developed.^9-11,32,33

Whereas NfL serves as a plasma biomarker for axon degeneration,^9,10 additional molecular markers are lacking for abnormalities in SC or denervated muscle that could serve as biomarkers for severity/progression of CMT1A and also provide therapeutic efficacy biomarkers. A recent survey of plasma protein biomarkers using Olink's Proximity Extension Assay screen of 460 unique proteins identified only NfL and a novel SC-derived protein, TMPRSS5, to be consistently elevated in independent cohorts of CMT1A samples.¹⁰ In contrast to NfL, TMPRSS5 is elevated early in the disease and does not correlate to disease score. Because it is highly expressed in Schwann cells, its elevation may reflect the early and ongoing disease process that is manifested by early reduction of nerve conduction velocities in CMT1A. Overall, the progression of symptoms in CMT1A most likely reflects axonal degeneration during the course of the disease. In parallel, we have recently developed a novel skin biopsy analysis to precisely assess the PMP22 expression level as a target engagement biomarker for CMT1A clinical trials.³² Nonetheless, there is a need for additional biomarkers to support efficient clinical trial design, as assessments of disease burden/progression (neuropathy scores) are insufficient measures of the biological changes occurring in Schwann cells, neurons, and muscle in patients with CMT1A.

Biomarkers can be used to not only measure target engagement and disease burden (MRI) but also to monitor disease processes in clinical trials.³⁴ Just as nerve conductions are reduced early in CMT1A progression, SC miR and myomiR levels may reflect an ongoing steady state of abnormal myelination and muscle degeneration, respectively. Therefore, if NfL is a treatment-responsive biomarker of axonal degeneration, as shown in preclinical studies of CMT1X,³⁵ we propose that myomiRs and SC miRs could be treatment-responsive biomarkers that would reflect stabilization of myelin and arrest of muscle degeneration. If so, this would make them the earliest sentinels of improvement in clinical trials, although additional studies are required to determine whether they are indeed treatment-responsive. We envision that miR analysis could be used in conjunction with protein biomarkers like NfL and TMPRSS5.

MyomiRs induced by degenerating axons were the most elevated class of miRs in CMT1A compared to control samples. The elevation of myomiRs was to some degree expected since the IMFA within calf muscles has emerged as a sensitive progression biomarker for CMT1A.^12,13 The IMFA of muscle is specifically a measure of a neurogenic or myopathic disease process and is independent of a person's overall level of activity or fitness.³⁶ The ROC curve for myomiR-206 demonstrated a highly significant ability to distinguish patient samples from control samples, at levels that were comparable to those obtained for both TMPRSS5 and NfL. The AUC for miR-133a and the Schwann cell–associated miR-223-3p are significant but less able to discriminate cases from controls.

These miRs were identified from the same samples we had previously used to demonstrate that TMPRSS5 as a Schwann cell protein and NfL as an axonal protein are elevated in CMT1A patient plasma.^9,10 Thus, we were able to compare the miR data for correlations with NfL and TMPRSS5, as well as COA and electrophysiologic assessments such as NCV and CMAP amplitudes. In CMT1A, the NCV are reduced at a very early age in individuals with CMT1A, and remain relatively unchanged throughout the course of the disease.³⁷ In contrast, CMAP amplitudes become progressively reduced and correlate with disease progression since these amplitudes are a measure of reduced axonal innervation of CMT1A muscle. Consistent with these clinical data, levels of NfL released from degenerating axons correlated with COA and CMAP amplitudes.

MyomiR-206 was the most highly elevated, but other myomiRs such as 133a had the strongest correlation with NfL levels. In addition, 2 of the myomiRs (miR-133a and -1) correlate negatively with CMAP and positively with neuropathy score. In addition, the myomiRs generally correlate with age, which is consistent with the observations that age-dependent progression of CMT1A is accompanied by a greater annual change in patients with IMFA with a baseline fat fraction of >10% compared to those with <10%.¹³ It is not clear why miR-133a has the strongest correlations among the myomiRs, although miR-133a is the most abundant of the myomiRs in CMT1A plasma and there may be less inherent error in its measurement.

In contrast to myomiRs, Schwann cell miRs did not generally correlate with age, COA, or physiology, consistent with the results from our prior studies of TMPRSS5.¹⁰ Accordingly, 6 of the SC miRs correlate with the SC marker (TMPRSS5), but not generally with NfL levels, neuropathy score, or CMAP (except for miR-328 and -223-3p). The SC miRs also strongly correlate with each other, with more modest correlations with myomiRs. A number of miRs have been identified as regulators of Pmp22 expression in Schwann cells (mir-149, miR-29a, miR-381),^38-41 but we did not find any change in these in our profiling of CMT1A plasma.

Aside from myomiRs and SC miRs, there were 8 additional ones that were evaluated in the larger set of samples. None correlated with age, NfL, or neuropathy scores, but 5 correlated with TMPRSS5, and 4 had an inverse correlation with CMAP. The 5 correlated with TMPRSS5 also correlated strongly with SC miRs and could be considered members of the SC cluster. However, these 5 are generally less abundant in plasma, although there is some evidence of miR-28 expression in Schwann cells.⁴²

There are several positive correlations between individual miRs, as would be expected since they are significantly elevated in CMT1A samples. One of the stronger correlations was between miR-409 and -431, which both lie in a region of human chr14 containing miR clusters and Meg3/Meg8 genes. This area appears to have arisen from an ancestral chromosomal duplication on chr14, but miR-409 and -431 are not part of the same miR family. Interestingly, both are highly correlated with TMPRSS5, but not with NfL serum levels. Both miR-409 and -431 were found to be elevated in a profile of schwannomas,⁴³ suggesting that they may correlate with Schwann cell proliferation. Previous studies have identified Schwann cell proliferation in CMT1A.⁴⁴

MicroRNAs secreted by muscle or Schwann cells may regulate transcripts that are involved in CMT pathogenesis. Studies of miR-206 in a mouse model of amyotrophic lateral sclerosis have indicated that it delays disease progression and promotes regeneration of neuromuscular junctions.⁴⁵ Therefore, it is interesting to speculate which targets and molecular pathways are modified by the secretion of the myomiRs or Schwann cell miRs. The gene ontology analysis of predicted targets of SC miRs identified enrichment of PI3 kinase and neuregulin signaling pathways, both of which play important roles in Schwann cell development.^26,46 Interestingly, studies in CMT1A models have shown reduced PI3 kinase pathway activation and increased ERK signaling, which can be modulated by soluble neuregulin to improve neuropathy.²⁷

However, it is unclear whether these miRs represent an adaptive response to neuropathy or are just released from atrophic muscle or Schwann cells in neuropathic nerve. In addition, the Schwann cell miRs are highly expressed in peripheral nerve^23,47 and are co-correlated in the CMT1A plasma samples. several of these miRs are also expressed in other tissues, so it has not been established that their elevation in serum is attributed to peripheral nerve. MiRNA profiling has been performed in other types of neuropathy, and a recent study of transthyretin amyloidosis (ATTR) caused by TTR mutation⁴⁸ identified induction of some miRs (miR-150, -223, -328) shared with our set in CMT1A. In addition, miR-199a-3p was also identified in diabetic neuropathy.⁴⁹

We have identified a group of miRs from muscle and Schwann cells that are elevated in plasma of patients from CMT1A. As with NfL and TMPRSS5 protein levels, a major question will be if these normalize with successful clinical trials as well as how levels of these miRs correlate with IMFA and other measures in natural history studies. In this regard, an antisense oligonucleotide study in a mouse model of spinal muscular atrophy showed elevation of miR-206 that was normalized by the treatment.⁵⁰

Glossary

AUC

area under the curve

CMAP

compound muscle action potential

CMT

Charcot-Marie-Tooth disease

CMT1A

Charcot-Marie-Tooth disease type 1A

CMTES

Charcot-Marie-Tooth Examination Score

CMTNS

Charcot-Marie-Tooth Neuropathy Score

COA

clinical outcome assessment

IMFA

intramuscular fat accumulation

INC

Inherited Neuropathy Consortium

miR

microRNA

MNCV

motor nerve conduction velocity

myomiR

miR biomarker

NCV

nerve conduction velocity

NfL

neurofilament light

NGS

next-generation sequencing

qPCR

quantitative PCR

R-CMTES

Rasch-modified Charcot-Marie-Tooth Examination Score

R-CMTNS

Rasch-modified Charcot-Marie-Tooth Neuropathy Score

ROC

receiver operating characteristic

RT

reverse transcriptase

TMPRSS5

transmembrane protease serine 5

Appendix. Authors

Study Funding

This work was supported by a grant from the Charcot-Marie-Tooth Association; by NIH R21 TR003034, U54 NS065712 provided by NINDS/NCATS-ORD; and a core grant to the Waisman Center from NICHD (U54 HD090256).

Disclosure

The authors report no disclosures relevant to the manuscript. Go to Neurology.org/N for full disclosures.