MicroRNA-206: a potential circulating biomarker candidate for amyotrophic lateral sclerosis.

Amyotrophic lateral sclerosis (ALS) is a lethal motor neuron disease that progressively debilitates neuronal cells that control voluntary muscle activity. Biomarkers are urgently needed to facilitate ALS diagnosis and prognosis, and as indicators of therapeutic response in clinical trials. microRNAs (miRNAs), small posttranscriptional modifiers of gene expression, are frequently altered in disease conditions. Besides their important regulatory role in variety of biological processes, miRNAs can also be released into the circulation by pathologically affected tissues and display remarkable stability in body fluids. In a mouse model of ALS that expresses mutated human superoxide dismutase 1 (SOD1-G93A) skeletal muscle is one of the tissues affected early by mutant SOD1 toxicity. To find biomarkers for ALS, we studied miRNA alterations from skeletal muscle and plasma of SOD1-G93A mice, and subsequently tested the levels of the affected miRNAs in the serum from human ALS patients. Fast-twitch and slow-twitch muscles from symptomatic SOD1-G93A mice (age 90 days) and their control littermates were first studied using miRNA microarrays and then evaluated with quantitative PCR from five age groups from neonatal to the terminal disease stage (10-120 days). Among those miRNA changed in various age/gender/muscle groups (miR-206, -1, -133a, -133b, -145, -21, -24), miR-206 was the only one consistently altered during the course of the disease pathology. In both sexes, mature miR-206 was increased in fast-twitch muscles preferably affected in the SOD1-G93A model, with highest expression towards the most severely affected animals. Importantly, miR-206 was also increased in the circulation of symptomatic animals and in a group of 12 definite ALS patients tested. We conclude that miR-206 is elevated in the circulation of symptomatic SOD1-G93A mice and possibly in human ALS patients. Although larger scale studies on ALS patients are warranted, miR-206 is a promising candidate biomarker for this motor neuron disease.

Introduction

Amyotrophic lateral sclerosis (ALS) is a motor neuron disease that affects 4–6 individuals per 100 000 worldwide, a lifetime risk for developing the condition being approximately one in 400 [1], [2]. ALS leads to degeneration of upper and lower motor neurons. The clinical disease is characterized by progressive skeletal muscle weakness/atrophy and in majority of patients culminates in fatal respiratory paralysis within 3–5 years of the onset of symptoms. There is no cure for ALS and the only approved disease-modifying drug, Riluzole, slows the progression to death only modestly [3]. The identification of ALS is complicated and time consuming as no test can currently establish the disease. The diagnostic process relies on series of clinical examinations combined with electrodiagnostics and other tests to rule out alternative disorders with a similar presentation. Indeed, a significant overlap with some clinically alike conditions at the early stage of ALS considerably delays the progress to the definite diagnosis [4], [5]. While the etiology of ALS at large remains unknown, both environmental and genetic factors are known to be involved. Most cases are sporadic (SALS) and approximately 5% have family history (FALS) [6]. However, the two forms of the disease are clinically indistinguishable. Causes of FALS have been mapped to more than 20 genes [7] and mutations in superoxide dismutase 1 (SOD1), account for approximately one fifth of FALS and a small percentage of apparently sporadic cases.

A transgenic mouse strain overexpressing one of the pathogenic human SOD1 alleles, SOD1-G93A, is a frequently used transgenic model of ALS [8], [9]. Most ALS-linked SOD1 mutations are thought to exert their molecular effects through toxic gain of function(s) by the mutant enzyme [10]. Although the exact nature of the novel molecular properties gained remains elusive, mutant SOD1 (mSOD1) models mimic the main features of ALS and have offered valuable insights to the basic pathology of the disease. One of these is that ALS is not simply a neuronal disease. Neuron-restricted expression of the mSOD1 is not always sufficient to cause the disease [11], [12], although this may depend on neuronal mSOD1 load [13]. Contribution by non-neural cells such as neuroglia, however, is evident [14] and altered function of inhibitory interneurons that monitor and control motor neuron activity may be involved [15]. Furthermore, skeletal muscle-autonomous mSOD1 toxicity strongly contributes to the disease progression in these models; muscle-restricted expression of the mSOD1 is sufficient to trigger ALS symptoms and muscle atrophy [16] as well as motor neuron degeneration [17]. Muscle atrophy in rodent models and ALS patients is also associated with similar perturbations in molecular networks controlling muscle organization and function including autophagy, mitochondrial homeostasis, hypermetabolism and myosatellite cell function [16], [18]–[24]. Therefore, structural and metabolic alterations in cells that connect with neurons, including skeletal muscle, may synergistically exacerbate the disease outcome.

At early stages of ALS pathology, motor neuron cell bodies in the central nervous system remain mostly intact. Motor neuron degeneration initiates by distal withdrawal of nerve terminals from the neuromuscular junction (NMJ) and proceeds retrogradely along the motor axon. This “dying back” pathology is observed in both SOD1-G93A mice and post-mortem samples from human ALS patients [25], [26]. Motor units, functional units of the motor system, consist of a single motor neuron and a group or muscle fibers with similar properties. The denervation in ALS preferentially affects large, fast-conducting motor neurons and, as a result, fast-twitch motor units are more vulnerable to the pathology than slow-twitch units [26]–[29]. The symptomatic stage characterized by tremors and hind-leg spasticity commences in SOD1-G93A mice approximately at postnatal day 90 [8]. However, muscle girth measurements, medical imaging and histological findings all indicate loss of hind-limb muscle volume and decreased fast muscle fiber diameter before the age of 60 days [30], [31]. This suggests early subclinical pathology of skeletal muscle in mSOD1 models.

microRNAs (miRNAs) are evolutionarily conserved non-coding RNA molecules involved in post-transcriptional regulation of gene expression. This regulation is achieved by pairing of miRNA with complementary sequences found in its target mRNAs [32]. The usual consequence of the miRNA:mRNA binding is a downregulation of the corresponding mRNA and/or protein levels due to mRNA destabilization or translational inhibition. Recent evidence gained through simultaneous mRNA and proteomic/ribosomal profiling suggests that the former is dominant in mammalian cells [33]–[35]. miRNAs are frequently altered in conditions such as cancer, cardiovascular and metabolic diseases, and in disorders of central nervous system [36]–[39]. miRNAs are known to be secreted by various cell types and, unlike most mRNAs, are remarkably stable in circulating body fluids due to proteic or vesicular protection from ribonucleases [40]–[42]. Because of these properties miRNAs have recently gained attention for their potential as minimally invasive and cost-effective disease biomarkers.

Here, we first studied by microarrays how miRNA expression is affected in vulnerable (fast) and resistant (slow) muscle types of symptomatic SOD1-G93A mice. The rationale for using skeletal muscle was based on its contribution to the ALS pathology in mice [16], [17] and on the fact that it represents the largest tissue in the human body, possessing a great potential to release miRNAs to the circulation. The selected candidate miRNAs from the microarray data were then investigated individually with quantitative PCR (qPCR) from neonatal, presymptomatic, symptomatic and terminal stage mice in the two muscle types and in the circulation of same model. The relative abundance of the implicated miRNAs was finally measured in serum samples from ALS patients and control individuals. The data indicates that muscle-enriched miR-206 may serve as a non-invasive circulating biomarker for ALS, and warrants larger scale studies on SALS and FALS patients.

Results

Muscle miRNA Profile in Symptomatic SOD1-G93A Mice

To gain first insight into the miRNA expression in the skeletal muscle tissue of SOD1-G93A animals, RNA derived from fast extensor digitorum longus (EDL) and slow soleus (SOL) muscles of mutant (denoted +) and control (denoted -) mice at postnatal day P90 (P+days of age) was analysed with miRNA microarrays. It was first assessed how miRNA profile is altered in each muscle type of the mutant compared to the wild type littermates (EDL+ vs. EDL−, and SOL+ vs. SOL−). Secondly, we analysed which miRNA species are significantly more expressed in one of the muscle types in the normal context (EDL− vs. SOL−) and if this enrichment is affected in the mutants (EDL+ vs. SOL+). For main subtypes of probesets found in the arrays were mature miRNA (miRNA), passenger RNA (miRNA*) and stem-loop (pre-miRNA). Some probesets corresponding to non-miRNA species (fragments of tRNA, rRNA, mtDNA etc.) are listed on the Tables S1, S3 and S4 but were not considered as real miRNA changes.

Without multiple corrections (see Materials and methods), comparison of SOD1-G93A with controls revealed 179 significant (p<0.05) changes in EDL+ (108 up, 71 down), and 42 changes in SOL+ (19 up, 23 down). Only 12 probesets were common among those affected in EDL+ and SOL+ muscles, leaving 167 mutant EDL-specific changes and 30 changes specific to mutant SOL muscle (Figure 1, see also clustering of the affected probesets in Figure S1). The fold changes associated with SOD1-G93A expression in the EDL varied from 11.9 fold upregulation to −4.2 fold downregulation and in SOL from 2.7 fold upregulation to −2.6 fold downregulation. The most upregulated mature miRNA in the mutant EDL were on average more highly expressed (7.8±0.6 fold) than most upregulated miRNA* (3.6±0.6 fold, p<0.001) or pre-miRNA (1.9±0.3 fold, p<0.001) species, whereas most downregulated mature miRNA did not significantly differ from that of miRNA* or pre-miRNA (Figure S2). In EDL, 56 probesets were upregulated and 10 probesets downregulated more than two fold, most (79%) of those being mature miRNAs (Figure S3). In SOL, only one probeset was up and one down more than two fold (both mature miRNAs). The complete lists of significant changes in SOD1-G93A EDL and SOL compared to the respective wild type muscles are shown in Tables S1 and S2, respectively.

Figure 1

miRNA alterations in the skeletal muscle of symptomatic ALS mice.

Venn diagram showing the significant (p<0.05) changes observed in EDL (green, total 179 changes) and SOL (orange, total 42 changes) muscles of SOD1-G93A mutants (+) compared with wild type (−) littermates. Arrows pointing up and down indicate upregulation and downregulation, respectively. An overlap with the two muscle types was observed in 12 probe sets, only three of which were changed in the same direction (brown square). Total 1412 Mus musculus miRNA probe sets were analysed using Affymetrix GeneChip miRNA 2.0 Arrays. Significantly affected non-miRNA species (n = 8), such as fragments of tRNA, rRNA and the mitochondrial genome, are not included in the data but are listed in Tables S1 and S2.

Comparison of the fast and slow muscle type within wild type animals (EDL− vs. SOL−) revealed total 118 probesets with preferential expression in one of the muscle types (Table S3). Of these, approximately one half were preferentially expressed in EDL and the other half in SOL. Of those 31 probesets that were enriched more than 2 fold in one of the muscles, 12 was primarily expressed in EDL and 19 in SOL. Similar comparison with the two muscle types in SOD1-G93A mutants (EDL+ vs. SOL+, Table S4) revealed total 77 significant changes, 43 of which were preferentially EDL expressed and 34 preferentially SOL expressed. Eight probesets were enriched more than 2 fold in one of the muscle types (4 in EDL and 2 in SOL) in the mutants.

One quarter (25%) of the total probesets up-regulated in the mutant EDL muscle (EDL+ vs. EDL−) were coincident with those preferentially expressed in wild type SOL (EDL− vs. SOL−), most (84%) of these being SOL enriched for more than 2 fold. Respectively, one third (32%) of the total downregulated miRNA in the mutant EDL muscle (EDL+ vs. EDL−), were preferably expressed in EDL of the controls (EDL− vs. SOL−). As expected, multiple corrections by false discovery rate (FDR), which indicates the fraction of false positives within a list of genes exceeding a given statistical cutoff (see materials and methods fo details) markedly reduced the number of significant changes in the miRNA list. In fact only two miRNA from the EDL+ vs. EDL− comparison remained significant (q-value<0.05) after FDR correction, and none from SOL+ vs. SOL− comparison. The most promising miRNA candidates were then shortlisted for individual assays considering following criteria: 1) published evidence supporting that the target is real miRNA, 2) preference for mature miRNAs, 3) preference for miRNA affected in EDL, 4) preference for high/moderate array signal intensities to low ones, 5) preference for higher fold change and most significant p-values and, 6) those miRNA implicated in muscle biology. Total 24 miRNA were tested first for sufficient expression level in EDL, SOL and plasma from wild type and mutant animals. A limit where miRNA is considered expressed in sufficient level for reliable analysis was set to 32 PCR cycles at least in one of the muscles and plasma (Figure S4). Sufficiently high levels of expression was found only in 10 miRNAs: miR-133a, miR-206, miR-1, miR-145, miR-24, miR-19b, miR-17, miR-106b, miR-20a and miR-21. Additionally, miR-133b that was not altered in the arrays was selected for its known function in muscle biology. Some of the miRNA were found to be expressed in very low amounts in muscle and not detectable in plasma. Such included some of the most upregulated miRNA in the array data, such as miR-379, miR-671-5p and miR-708 (Figure S4).

Disease-stage Affects miRNA Alterations in SOD1-G93A Skeletal Muscle and Blood Plasma

The selected 11 miRNAs were examined in various postnatal stages that in SOD1-G93A mice correspond to neonatal (P10), early presymptomatic (P40), late presymptomatic (P60), symptomatic (P90) and terminal phase (P120). Individual miRNAs were measured from EDL and plasma of male mice from all age groups, and from EDL and plasma of females from P60, P90 and P120. As soleus is known to be affected later in SOD1-G93A model, SOL was tested only at two oldest age groups for both sexes. Briefly, the only miRNA whose expression was increased consistently at any stage/tissue/sex was miR-206. The other significantly altered miRNAs (miR-1, -133a, -133b, -145, -21 and miR-24) were always found to be diminished, although modestly (≤50%) and varying between the age and muscle/plasma/sex groups. All observed miRNA alterations in the two muscles and plasma of SOD1-G93A animals are listed in Table 1.

Table 1

Relative* miRNA alterations verified by qPCR in SOD1-G93A animals.

Tissue	miRNA	P60	p (P60)	P90	p (P90)	P120	p (P120)
EDL	miR-206	1.7 (M)	0.1	7.7 (M)/7.7 (F)	0.0001/0.001	11.7 (M)/4.2 (F)	0.0001/0.03
	miR-1	–	–	0.5 (F)	0.1	0.4 (F)	0.05
	miR-133a	–	–	0.6 (F)	0.004	–	–
	miR-133b	–	–	–	–	0.7 (F)	0.04
	miR-106b	–	–	2.3 (M)	0.11	–	–
Plasma	miR-206	3.8 (M)	0.1	4.5 (M)/3.3 (F)	0.0004/0.09	0.6 (M)/4.2 (F)	0.04/0.02
	miR-1	–	–	–	–	0.2 (M)	0.001
	miR-133a	–	–	–	–	0.3 (M)/2.6 (F)	0.005/0.1
	miR-133b	–	–	–	–	0.3 (M)	0.01
	miR-145	–	–	1.7 (M)	0.12	0.5 (M)	0.04
	miR-21	–	–	–	–	0.5 (M)	0.05
	miR-24	–	–	–	–	0.5 (M)	0.03
	miR-106b	2.3 (M)	0.1	–	–	–	–
SOL	miR-206	N/A	N/A	–	–	–	–
	miR-1	N/A	N/A	0.7 (M)	0.1	0.5 (M)/0.6 (F)	0.002/0.06
	miR-133a	N/A	N/A	0.8 (M)	0.01	–	–
	miR-133b	N/A	N/A	0.7 (M)	0.1	–	–
	miR-145	N/A	N/A	0.7 (M)/0.6 (F)	0.02/0.1	0.7 (M)/0.6 (F)	0.08/0.01

Significant results (p<0.05) shown in bold, those close to significance (0.05

0.15) in normal font.

*Relative numbers >1 refer to upregulation, and those <1 to downregulation. Abbreviations: M, males; F, females; p(PX), p-value at indicated postnatal day X.

Significant results (p<0.05) shown in bold, those close to significance (0.05