Introduction

Amyotrophic lateral sclerosis (ALS) is an adult-onset, rapidly progressive neurodegenerative disease mainly caused by the loss of motor neurons. Although most ALS cases are sporadic (SALS), ~10 % of them are familial (FALS), usually transmitted as an autosomal dominant trait. Pathogenic mutations are found in ~50 % of FALS and ~5 % of SALS cases, most commonly in SOD1, C9ORF72, TARDBP, FUS, but also in ANG and OPTN genes [8, 36, 40].

Several mechanisms have been proposed to explain the pathogenesis of ALS, including neuroinflammatory processes [33]. Although results from routine cerebrospinal fluid (CSF) analysis are usually unremarkable, several studies have shown an increase in total protein levels, and an altered CSF/serum albumin ratio (QAlb) in the CSF of ALS patients, suggesting an altered blood–brain barrier (BBB) permeability. Moreover, CSF oligoclonal bands (OCBs), indicating intrathecal synthesis of IgG, can be detected in 0.5–2 % of all ALS cases [2, 20, 37].

Although there is no evidence so far that the ALS-associated genes encode for proteins directly involved in maintaining BBB integrity, it is possible that disease-causing mutations may lead to BBB disruption and neuroinflammation. For instance, transgenic mice expressing mutant human SOD1G93A display an early BBB dysfunction [14, 31], while VEGF is one of the main modulators of the BBB integrity [24, 38]. Lastly, TDP-43, FUS, and OPTN immunoreactive inclusions have been observed in motor neurons as well as in astrocytic cytoplasmic processes [3, 19, 23], possibly altering the glial–vascular interface.

The aim of this study was to evaluate the occurrence of OCBs in the CSF of ALS individuals genetically characterized for ALS-associated genes.

Methods

Patients and controls

Our cohort included 259 ALS patients of Italian descent. All patients received a diagnosis of probable or definite ALS according to the El Escorial revised criteria at a tertiary care ALS Center. A subset of 13 patients had probable or definite familial ALS (FALS), according to the recently proposed criteria for FALS classification [4]. The demographic and clinical characteristics of our cohort are summarized in supplemental table 1. A panel of 40 control individuals without neurodegenerative or inflammatory diseases was used for comparison of CSF parameters. Specifically, the control panel included individuals with psychiatric disorders (16), vascular encephalopathy (15), cervical spondylotic myelopathy (6), diabetic neuropathy (2), and hereditary neuropathy with liability to pressure palsies (1).

Standard protocol approvals and patient consent

We received approval from the ethical standards committee on human experimentation of the IRCCS Istituto Auxologico Italiano. Written informed consent was obtained from all patients and healthy subjects participating in the study (consent for research). The study has been performed in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki.

CSF analysis

Lumbar puncture was performed in 259 ALS patients as part of the routine diagnostic procedures after they had given a written informed consent. The following CSF parameters were measured according to standard procedures and compared to serum levels: glucose, total proteins, total IgG, albumin, cell count. QAlb was calculated using the formula albuminCSF/albuminserum. Since CSF albumin completely derives from serum albumin and there is no intrathecal synthesis of the protein, QAlb represents the most useful parameter to assess the permeability of the BBB. Link index, an indirect parameter to evaluate intrathecal synthesis of IgG, was calculated using the formula (IgGCSF × albuminserum)/(IgGserum × albuminCSF). Normal ranges for CSF parameters, including QAlb and Link index, were determined according to international standards. Detection of OCBs was performed by isoelectric focusing on agarose gel and subsequent immunoblotting (IEF) with an anti-IgG antibody. By comparing OCBs in CSF and serum, a specific IEF pattern type was determined for each patient according to consensus criteria [12]. IEF is the current “gold standard” for detection of intrathecal IgG synthesis over other quantitative methods such as the Link index. In addition to biochemical tests, CSF samples were assayed for the presence of genetic material of the most common neurotropic viruses.

Hematological analysis

Complete blood cell count, erythrocyte sedimentation rate, C-reactive protein, serum protein electrophoresis, autoantibody screening, VDRL testing were performed in all patients. When the clinical presentation was suggestive, dosage of antibodies against Borrelia Burgdoferi and Human Immunodeficiency Virus 1 was also performed.

Genetic analysis

Genomic DNA was extracted from peripheral blood according to standard procedures. The coding regions of SOD1, TARDBP, FUS, ANG and OPTN genes were amplified by PCR and directly sequenced using BigDyeTerminator v 3.1 cycle sequencing kit on an automated 3100 ABI Prism Genetic Analyzer (Applied Biosystems, Foster City, CA, USA) as previously described [7, 9, 15, 16, 39]. The identified sequence variations were confirmed by sequencing an independent PCR product. The presence of the GGGGCC hexanucleotide repeat expansion in the first intron of C9ORF72 was also assessed through repeat-primed PCR. For the patient carrying the homozygous TARDBP p.A382T mutation we also performed a high-resolution array-Comparative Genomic Hybridization (CGH) analysis using the 244 K Human Genome CGH Microarray Kit (Agilent Technologies, Santa Clara, CA, USA) and a whole-genome single nucleotide polymorphism (SNP) genotyping using the Human660 W-Quad BeadChip (Illumina, San Diego, CA, USA). SNP genotype data were subsequently analyzed for Loss of Heterozigosity (LOH) and copy number variations (CNV) using the Genome Studio software (Illumina).

Results

Cerebrospinal fluid (CSF) analysis in a cohort of 259 ALS patients (246 sporadic and 13 familial cases) revealed an average age-adjusted QAlb value of 5.2 ± 0.8, not dissimilar from that observed in a panel of 40 controls (4.6 ± 1.3; p > 0.05). Similarly, the average Link index, which quantifies the intrathecal IgG production, was not different among patients and controls (0.48 vs 0.47; p > 0.05). Interestingly, age-adjusted QAlb was elevated in 15/259 (5.8 %) ALS cases, while it was normal in all control individuals. OCBs were observed in the CSF of 9/259 (3.5 %) ALS patients and in no control samples. All patients with OCBs had apparently sporadic ALS (SALS). Five patients displayed on IEF a pattern type 2, indicating the presence of specific bands in the CSF only, but not in the serum. The remaining four presented OCBs in the CSF and additional identical bands both in the CSF and in the serum (pattern type 3, also indicative of intrathecal IgG synthesis). In two patients intrathecal IgG synthesis could be explained either by concurrent active neurosyphilis, or by a previous fetal olfactory ensheathing cells transplant in the frontal lobes. The remaining seven patients had no evidence of infectious or inflammatory diseases of the central nervous system. Paraproteinemia, lymphoma or any other systemic inflammatory or infectious conditions were also excluded. Other CSF parameters, including glucose levels and cell count, were normal in all individuals studied, and the screening for neurotrophic virus genomes was negative in all patients. The clinical features of the patients carrying OCBs are reported in Table 1.

Table 1 Clinical characteristics of patients harboring ALS-associated mutations (regular font), OCBs in the CSF (italics), or both (bold)
Full size table

The genetic analysis revealed eight different ALS-associated mutations in 12 patients (4 %). In particular, we identified five individuals carrying mutations in SOD1 (p.A4V, p.F45C, p.G93D, p.I113T, and p.L144F), 4 in TARDBP (one p.G287S, and three p.A382T), and one each in FUS (p.R521C), OPTN (c.1401 + 4A > G) and ANG (p.P-4S). All mutations were observed in SALS cases, with the exception of p.A4V and p.L144F in the SOD1gene, which were found in two FALS patients. The clinical features of the mutated patients in the analyzed cohort are also summarized in Table 1. Several of them have already been described in other papers [7, 9, 15, 16, 39], and the mutations have been deposited in the ALSoD database (http://alsod.iop.kcl.ac.uk/). The nine ALS patients with OCBs in their CSF were screened for the presence of C9ORF72 hexanucleotide repeat expansion, and none of them was found to carry this mutation.

Among the group presenting with OCBs in the CSF, three patients (33.3 %) had ALS-associated mutations, namely the TARDBP p.A382T variant in two cases and the p.P-4S variant in the signal peptide of ANG in one case (Table 2). Interestingly, the TARDBP p.A382T mutation was found in homozygous state in one ALS individual. By performing high-resolution array-CGH we excluded deletions involving TARDBP genomic region (chr 1p36.22) or gene copy loss (supplemental fig. 1). Genotype data also confirmed the absence of CNV, consistent with the array-CGH results. LOH analysis demonstrated a 2.8 Mb region of extended homozygosity (chr1:10,891,317-13,713,882) encompassing TARDBP locus, suggesting that both alleles are identical by descent (supplemental fig. 2).

Table 2 Biochemical parameters of the cerebrospinal fluid in our ALS cohort
Full size table

Average age-adjusted QAlb and Link index were similar both in the individuals positive for OCBs and in those harboring ALS-associated mutations, compared with the remaining cohort. Only patient A46, which carried the ANG p.P-4S mutation, had a mildly elevated QAlb, suggestive of a BBB leakage. The patient had a spinal onset, classic ALS phenotype (Table 1).

Conversely, both patients having OCBs in their CSF and the TARDBP p.A382T mutation displayed an atypical clinical phenotype. Patient A399, carrying the p.A382T variant in homozygous state, developed a bulbar-onset, rapidly progressive motor neuron disease at age 44, requiring invasive ventilation and a percutaneous endoscopic gastrostomy 17 months after onset of first symptoms. The clinical phenotype was mainly characterized by the presence of upper motor neuron signs. Brain MRI showed the presence of multiple small periventricular and subcortical white matter lesions (WMLs) (Fig. 1), without gadolinium contrast enhancement. This finding, together with the presence of OCBs in the CSF, and the absence of systemic inflammatory, infectious or autoimmune diseases, and of cardiovascular risk factors including thrombophilic conditions, was suggestive of a coexisting inflammatory demyelinating disease of the CNS.

Fig. 1
figure 1

Brain MRI acquired on a 1-Tesla scanner of patient A399, homozygous for the TARDBP p.A382T mutation, axial proton density scans. White arrowheads indicate subcortical white matter lesions; the arrow indicates a diffuse, faint signal hyperintensity of the white matter of both semioval centers

Full size image

The other patient (A338) developed a spinal onset motor neuron disease at age 59. The disease progression was markedly slow, and ten years after symptom onset the patient was still able to walk independently and had only mild bulbar signs. The neurological examination, however, showed uncharacteristic signs suggestive of an extramotoneuronal involvement. Ocular movement examination demonstrated saccadation of slow pursuit eye movements, slow saccades, and left gaze-evoked nystagmus. A postural as well as intention tremor of the upper limbs was present, together with dysmetria, bilateral dysdiadochokinesia and Holmes’ sign. Gait evaluation showed a spastic and ataxic paraparesis, with evident reduction of right-arm swinging. Postural reflexes were impaired with marked retropulsion at the pull test. Collectively, these findings suggested that the disease also affected the cerebellar and, possibly, the extrapyramidal system. Brain MRI was normal.

Noteworthy, the other two patients A10 and A349, carrying mutations in TARDBP gene, but without OCBs in the CSF, displayed a classic ALS phenotype, in particular the A349 individual with the same p.A382T variant.

Discussion

Several observations suggest that inflammatory processes contribute to ALS pathogenesis. Increased levels of albumin and immunoglobulin G (IgG) have been found in the CSF [25], while T-lymphocyte infiltration and local activation of immune cells have been observed in the spinal cord of ALS patients [11, 18]. Cerebral microglial activation is also a prominent feature of early-stage ALS, as demonstrated by neuroradiological studies [41]. Additionally, BBB dysfunction with subsequent neuroinflammation has been described as an early feature in ALS animal models [13].

Intrathecal synthesis of OCBs is suggestive of a local humoral immune response, and is a common feature of multiple sclerosis and other chronic inflammatory or infectious diseases of the CNS. OCBs, however, are rarely detected in neurodegenerative diseases such as ALS, being present in only 0.5–2 % of patients with motor neuron disease [20, 37]. This particular finding has usually been explained either by a former infection of the CNS or by an ongoing immune process in a subset of patients. In particular, patients with a concurrent paraproteinemia have a higher probability of presenting OCBs [43].

In our cohort we observed the presence of CSF OCBs in 3.5 % of all analyzed ALS cases, a frequency almost double than previously reported. This finding suggests that immune processes may play a role in a subset of ALS patients, either by contributing directly to the pathogenesis of the disease, or by representing a secondary event in response to neuronal damage. In either case, understanding why some individuals display inflammatory features may help comprehend the pathomechanisms of ALS.

In the majority of the patients positive for OCBs all the remaining CSF parameters analyzed were normal. In particular, with the exception of two individuals, QAlb was within normal limits, indicating a normal BBB permeability at the time the lumbar puncture was performed. It must be observed, however, that the evidence of intrathecal synthesis of IgG is nonetheless strongly suggestive of a qualitative BBB dysfunction leading to exposure of brain antigens to the systemic circulation, penetration of immunocompetent cells into the CNS, and ultimately, to a partial loss of the immune privilege of the CNS itself.

Since an evident cause for OCBs in the CSF could not be found for most of our patients, we aimed to evaluate if genetic factors might be associated to our observation. Interestingly, we found that the occurrence of OCBs was significantly higher in patients carrying ALS-associated mutations than in the remaining cohort (3/12, 25 % vs 6/247, 2.4 %; p < 0.01; Fisher’s exact test).

In particular, this association was stronger for the TARDBP p.A382T mutation, which was found in 2/9 (22.2 %) of patients with OCBs compared to 1/250 (0.4 %) in the remaining cohort (p < 0.01). The p.A382T variant is the most common TARDBP mutation identified in ALS patients, especially in countries of the Mediterranean basin [7, 21]. The mutation has a very low penetrance, having been identified in an exceeding number of apparently sporadic patients and in about one-third of all ALS patients of Sardinian descent [5, 32]. It has been reported that individuals carrying this particular variant may display parkinsonian features or dementia, and up to 2.5 % of patients with idiopathic Parkinson’s disease from Sardinia also harbor this mutation [35]. Our description of an ALS patient presenting with prominent cerebellar involvement thus broadens the spectrum of the extramotoneuronal involvement associated with the TARDBP p.A382T mutation. Interestingly, a recent neuropathological report has shown the presence of cerebellar p62-positive inclusions in a proportion (26 %) of patients affected by TDP-43 proteinopathies (ALS, frontotemporal lobar degeneration or both) [22], although the possible correlation between such inclusions with clinical manifestations remains to be determined.

Although our results seem to suggest an association between TARDBP variants and intrathecal synthesis of IgG, it is not clear whether and how such mutations are responsible for the BBB damage and local immune response, since TDP-43 is not known to be directly involved in either of these processes. A recent study, however, showed the presence of TDP43-immunoreactive aggregates adjacent to or within the capillary basal lamina of small brain vessels in patients with frontotemporal lobar degeneration [28]. It is thus possible that intrathecal IgG synthesis in our patients may be consequent to the deposition of mutant TDP43 protein in astrocytic end-feet, loss of BBB integrity, and penetration of immunocompetent cells into the CNS. We speculate that a mutant gene dose effect may be responsible for the development of subcortical WMLs and the more severe phenotype in the individual carrying the homozygous p.A382T mutation compared to the other patient with the same mutation in heterozygous state.

It must be observed that we could not prove with certainty that Patient A399 had a concurrent demyelinating disease of the CNS, since temporal dissemination of inflammatory lesions was not demonstrated, and the observed WMLs could be nonspecific, being limited in number and without gadolinium contrast enhancement. Several observations, however, suggest a possible demyelinating nature these lesions. Firstly, age-associated nonspecific WMLs are rare in individuals younger than 50 years [27]. Moreover, Patient A399 did not present any feature commonly associated with WMLs, such as cardiovascular risk factors, systemic vasculitis and/or migraine. Lastly, WMLs were observed more frequently in individuals with OCBs compared to the remaining ALS patients (66.7 vs 15.2 %), indicating a strong association with intrathecal IgG syntesis. The concurrence of ALS and a demyelinating disease of the CNS of multiple sclerosis-type in the same patient is in fact an extremely rare event, having been previously reported in only four cases, none of which were tested for TARDBP mutations [6, 10, 17, 26].

Conversely to TARDBP, several genes involved in angiogenesis have been implicated in maintaining the integrity of the BBB. Increased expression of VEGF in response to hypoxia contributes to the disruption of the BBB and vascular leakage in the ischemic brain [44]. Moreover, the VEGF-mediated breakdown of the BBB was shown to result in the contact of sequestered CNS antigens with systemic immunocompetent cells, leading to neuroinflammatory processes [34]. Angiogenin, the principal downstream effector of VEGF, may also play a role in maintaining the BBB integrity. A recent report suggested a CNS-specific dysregulation of ANG expression in ALS [29], and a lack of up-regulation of both ANG and VEGF in hypoxaemic conditions has been observed in the CSF of ALS patients compared to controls [30]. The ANG p.P-4S variant, observed in Patient A46, has been found in individuals with ALS, idiopathic Parkinson’s disease and controls, albeit with a reduced frequency in the latter group [42]. The mutation leads to the substitution of a partially conserved proline with a serine in the signal peptide, possibly altering the correct maturation, subcellular localization, and/or secretion of the mature protein. Altered levels of angiogenin in patient A46’s CNS may thus be responsible of the damage to the BBB, as expressed by the increased QAlb, leading in turn to neuroinflammation and intrathecal IgG synthesis.

In conclusion, our results provide evidence that genetic defects are frequent in ALS patients with OCBs and that genetic determinants may play a role in immune processes in these patients. Recent reports suggest that the distinction between familial and sporadic ALS may be artificial and in fact, disease-associated mutations are often observed in apparently sporadic individuals [1]. Since genetic screening of every SALS patient is not usually feasible in a clinical setting, the identification of phenotypically distinct subgroups with a higher mutational frequency may assist clinicians when asking for specific genetic testing and during counseling. Our data suggest that the incidence of mutations of TARDBP and ANG gene is ten times higher among individuals with OCBs in the CSF compared to the remaining ALS patients. As such, we believe that disease-associated mutations in these two genes should be actively looked for in all patients with evidence of BBB dysfunction and/or intrathecal IgG synthesis. The possible role of mutant TDP-43 and angiogenin proteins in these processes warrants further investigation.