Cerebrospinal fluid, brain, and spinal cord levels of L‐aspartate signal excitatory neurotransmission abnormalities in multiple sclerosis patients and experimental autoimmune encephalomyelitis mouse model

The neuroinflammatory process characterizing multiple sclerosis (MS) is associated with changes in excitatory synaptic transmission and altered central concentrations of the primary excitatory amino acid, L‐glutamate (L‐Glu). Recent findings report that cerebrospinal fluid (CSF) levels of L‐Glu positively correlate with pro‐inflammatory cytokines in MS patients. However, to date, there is no evidence about the relationship between the other primary excitatory amino acid, L‐aspartate (L‐Asp), its derivative D‐enantiomer, D‐aspartate, and the levels of pro‐inflammatory and anti‐inflammatory cytokines in the CSF of MS. In the present study, we measured by HPLC the levels of these amino acids in the cortex, hippocampus, cerebellum, and spinal cord of mice affected by experimental autoimmune encephalomyelitis (EAE). Interestingly, in support of glutamatergic neurotransmission abnormalities in neuroinflammatory conditions, we showed reduced L‐Asp levels in the cortex and spinal cord of EAE mice and increased D‐aspartate/total aspartate ratio within the cerebellum and spinal cord of these animals. Additionally, we found significantly decreased CSF levels of L‐Asp in both relapsing–remitting (n = 157) MS (RR‐MS) and secondary progressive/primary progressive (n = 22) (SP/PP‐MS) patients, compared to control subjects with other neurological diseases (n = 40). Importantly, in RR‐MS patients, L‐Asp levels were correlated with the CSF concentrations of the inflammatory biomarkers G‐CSF, IL‐1ra, MIP‐1β, and Eotaxin, indicating that the central content of this excitatory amino acid, as previously reported for L‐Glu, reflects a neuroinflammatory environment in MS. In keeping with this, we revealed that CSF L‐Asp levels were positively correlated with those of L‐Glu, highlighting the convergent variation of these two excitatory amino acids under inflammatory synaptopathy occurring in MS.


| INTRODUC TI ON
Multiple sclerosis (MS) is an inflammatory and degenerative disease of the central nervous system (CNS) . Although its etiology is still unclear, a large bulk of evidence supports a pathogenic role for diffuse immune-mediated synaptic damage contributing to disease progression and disability accumulation Di Filippo et al., 2018;Mandolesi, Gentile, Musella, Fresegna, et al., 2015).
Consistent with this, in experimental autoimmune encephalomyelitis (EAE) animals, the most used preclinical model of MS, increased intrathecal production of pro-inflammatory cytokines, such as interleukin-1β (IL-1β), interleukin-17 (IL-17), and tumor necrosis factor (TNF), was associated with remarkable abnormalities in excitatory synaptic transmission and synaptic plasticity in different brain areas (Centonze et al., 2009;Di Filippo et al., 2013. Furthermore, clinical studies also showed that abnormally greater cerebrospinal fluid (CSF) levels of pro-inflammatory cytokines are associated with severe alterations in excitatory synaptic transmission and plasticity and with a higher risk of disease progression in MS patients (Rossi et al., 2012(Rossi et al., , 2014Stampanoni Bassi et al., 2019. In line with this, independent findings unveiled the existence of a direct correlation between pro-inflammatory cytokines and Lglutamate (L-Glu) concentrations in the CSF of MS patients (Kostic et al., 2014;Stampanoni Bassi et al., 2021), further suggesting that excitatory synaptic transmission and neuroinflammatory processes are intimately linked in a complex dynamic interaction that contributes to MS clinical phenotypes and disease progression (Levite, 2017).
L-Glu and L-aspartate (L-Asp) are the most abundant excitatory amino acids in the mammalian CNS, responsible for a wide range of metabolic and physiological functions, including brain development, synaptic transmission/plasticity, and cognition (Egbenya et al., 2021;Curtis & Watkins, 1960;Zhou & Danbolt, 2014;Andersen et al., 2021).
In addition to these L-amino acids, also the right-handed derivative of L-Asp, D-Asp, is known to affect glutamatergic neurotransmission by activating N-Methyl-D-Aspartate receptors (NMDARs) at their glutamate binding site (Grasselli et al., 2013;Ota et al., 2012;Usiello et al., 2020). However, unlike L-Glu, little evidence is available on the relationship linking both Asp enantiomers with MS disability progression and central inflammation.
Here, we investigated the potential intrathecal changes of L-Asp and D-Asp during neuroinflammatory processes by assessing their levels through HPLC in the CSF of a large cohort of MS patients and different CNS regions of mice affected by EAE. Specifically, we determined the concentrations of both Asp enantiomers in the CSF of MS patients with different clinical phenotypes, compared to control subjects with non-inflammatory/non-degenerative other neurological disorders (OND). Moreover, we explored the possible correlations between the CSF Asp enantiomer levels and clinical and radiological characteristics (neurological disability, inflammatory disease activity, disease duration, and disease progression) or other biochemical CSF features (L-Glu and various pro-and anti-inflammatory cytokines levels) of MS patients.
Our results indicate that intrathecal L-Asp levels positively correlate with inflammatory cytokines, suggesting the involvement of this amino acid in the modulation of inflammatory synaptopathy occurring in MS.

| Patients' enrolment and cerebrospinal fluid collection
In the present retrospective study, clinical data and CSF samples col-   (Kurtzke, 1983).
MRI examination consisted of 3 Tesla dual-echo proton density, fluid-attenuated inversion recovery (FLAIR), T2-weighted spin-echo images, and pre-contrast and post-contrast (gadolinium, Gd, 0.2 mL/ Kg e.v.) T1-weighted spin-echo images. All images were acquired in the axial orientation with 3 mm-thick contiguous slices. Radiological activity was defined as the presence of Gd-enhancing lesions evaluated by a neuroradiologist who was unaware of the patients' clinical details. CSF samples were collected according to international guidelines (del Campo et al., 2012;Teunissen et al., 2009;Vanderstichele et al., 2011). Lumbar puncture (LP) was performed from 8:00 to 10:00, after an overnight fasting. CSF was immediately collected in sterile polypropylene tubes (Sarstedt® tubes, codes: 62.610.210) and gently mixed to avoid possible gradient effects. Two mL of CSF were used for total cell count. All samples were centrifuged at 2000xg for 10 min at room temperature and then aliquoted in 0.5 mL aliquots in sterile polypropylene tubes (Sarstedt® tubes, codes: 72.730.007). Aliquots were frozen at −80°C pending analysis, avoiding freeze/thaw cycles. Blood-contaminated samples were excluded from the analysis (cut-off of 50 red blood cells/μL). Internal quality controls were assayed in each run. Operators blinded to the diagnosis performed the measurements.

| EAE protocol
EAE was induced in six to eight-week-old wild-type C57Bl/6 male mice (Cat# 027C57BL/6, Charles River, Italy) through the subcutaneous inoculation of 200 μg MOG 35-55 emulsified in Freund's incomplete adjuvant, supplemented with 8 mg/mL M. tuberculosis H37Ra at day 0 . Mice were also intra-peritoneally injected with 500 ng of Pertussis toxin at day 0 and day 2. In the EAE group, animals were monitored and weighed daily, from day 10 post-inoculation (p.i.) onwards, to assess the development of relapsing-remitting paralysis. Clinical signs of the experimental disease were scored as follows: 0 = normal; 1 = fully flaccid tail; 2 = impaired righting reflex; 3 = hind limb paresis; 4 = complete hindlimb paresis; 5 = moribund/death . Mice affected by the first episode of neurological deficit, suggestive of CNS inflammation (total mice subjected to EAE induction: n = 10; n = 2 animals did not develop neurological deficits and were thus excluded from the study; n = 8 mice developed EAE and were included in the study, mean clinical neurological score 2, day 20 p.i.) have been sacrificed by cervical dislocation for HPLC analysis, with the surgical isolation of the cortex, hippocampus, cerebellum and spinal cord, that have been snap frozen in dry-ice and stored at −80°C pending analysis. Six to eight-week-old male wild-type C57Bl/6 mice (Cat# 027C57BL/6, Charles River, Italy), subjected to the same feeding and housing conditions for the same period as EAE mice, were utilized as controls (total control mice: n = 6; no animal was excluded). The study involved a total number of 16 animals. Sample size calculation was not performed; the number of mice was based on a previous study of a similar nature (Musgrave et al., 2011). We performed a post hoc power analysis for a posteriori validation of our study, which showed an effect size of 0.8 (level of significance, p < 0.05). Animals were arbitrarily assigned by the researchers to the experimental groups at the beginning of the study. Animals were housed at the University of Perugia with standard housing conditions and free access to food and water (Center Authorization N: 08/2018-UT; 24/07/2018).
Animals were monitored on a daily basis, assessing specific and non-specific EAE disease parameters (i.e., neurological score, body weight, fur status) with humane end-points aimed at minimizing animal suffering. All procedures involving animals were performed in conformity with the European Directive 2010/63/EU, in accordance with protocols approved by the Animal Care and Use Committee at the University of Perugia.

| HPLC analysis
Brain and spinal cord tissues and CSF samples deriving from EAE mice and MS patients, respectively, were analyzed as previously reported (Nuzzo et al., 2019Palese et al., 2020;Stampanoni Bassi et al., 2021). Mice brain and spinal cord samples were homogenized in 1:10 (w/v) 0.2 M trichloroacetic acid (TCA), sonicated (3 cycles, 10 s each), and centrifuged at 13 000xg for 20 min. All the precipitated protein pellets from brain and spinal cord samples were stored at −80°C for protein quantification.
CSF samples (100 μL) were mixed in a 1:10 dilution with HPLCgrade methanol (900 μL) and centrifuged at 13 000xg for 10 min; supernatants were dried and then suspended in 0.2 M TCA. TCA supernatants were then neutralized with NaOH and subjected to pre-column derivatization with o-phthaldialdehyde (OPA)/N-acetyl-L-cysteine (NAC). Diastereoisomer derivatives were resolved on a UHPLC Nexera system (Shimadzu, Kyoto, Japan) using a Shim-pack GIST C18 3μm reversed-phase column (Shimadzu, 4.0 × 160 mm) under isocratic conditions (0.1 M sodium acetate buffer, pH 6.2, 1% tetrahydrofuran, and 1 mL/min flow rate). A washing step in 0.1 M sodium acetate buffer, 3% tetrahydrofuran, and 47% acetonitrile was performed after every single run. Identification and quantification of amino acids were based on retention times and peak areas, compared with those associated with external standards. The identity of the D-Asp peak was also evaluated by selective degradation catalyzed by a recombinant human D-Aspartate oxidase (hDDO) (Katane et al., 2017(Katane et al., , 2018. Briefly, samples were incubated with hDDO enzyme (12.5 μg) at 30°C for 3 h, and subsequently derivatized. Total protein content of brain and spinal cord tissue homogenates was determined by Bradford assay method, after solubilization of the TCA-precipitated protein pellets in 1% SDS solution. The detected amino acids concentration in tissue homogenates was normalized by the total protein content and expressed as nmol/mg protein; amino acids level in the CSF was expressed as μM. HPLC data analyses were carried out by an experimenter that was unaware of the study groups.

| Statistical analysis
Since it was a retrospective study, no sample size evaluations were analyzed. Normality distribution was tested using the Kolmogorov-Smirnov test. Data were shown as mean (± standard deviation, SD) or, as median (± interquartile range, IQR) if not normally distributed.
Categorical variables were presented as number (n). Differences in continuous variables among the two groups were evaluated by parametric t-test or, if necessary, non-parametric Mann-Whitney test. A p-value <0.05 was considered as statistically significant. Spearman's non-parametric correlation was used to test possible associations between non-parametric variables. When exploring correlations between L-Asp and various inflammatory CSF molecules, Benjamini-Hockberg (B-H) procedure was used to decrease the false discovery rate and avoid type I errors (false positives). To explore associations between L-Asp and different variables, after adjustment for possible confounding factors [i.e., age at LP, disease duration, sex, body mass index (BMI), oligoclonal band (OCB) presence, radiological disease activity], linear regression models were used. Box plot was used to depict statistically significant differences between groups. All analyses were performed using IBM SPSS Statistics for Windows (IBM Corp., Armonk, NY, USA).     Figure 1p,s,v). Overall, these data indicate the occurrence of pronounced alterations in the CNS levels of excitatory amino acids and L-Gln/L-Glu ratio in EAE condition, with a potential influence on glutamatergic neurotransmission under neuroinflammatory and neurodegenerative insults.

| Altered L-aspartate content in the CSF of MS patients
We recently showed a mild reduction of CSF L-Glu levels in MS patients, compared with OND subjects (Stampanoni Bassi et al., 2021).
Here we sought to evaluate the levels of the excitatory amino acids L-Asp and D-Asp by performing HPLC analysis in the same CSF co-  contrast to other investigations (Sarchielli et al., 2003), we did not find any significant correlation of L-Asp amount with any of these parameters (Table 2). In particular, differently to L-Glu (Stampanoni Bassi et al., 2021), we found that L-Asp amount was not correlated with disability progression (measured as EDSS at 1-year follow-up) in both RR-MS and SP/PP-MS patients (Table 2).  For the first time, the present data indicate that CSF levels of L-Asp are significantly associated with soluble inflammatory mediators in RR-MS, as previously described for L-Glu (Kostic et al., 2014;Stampanoni Bassi et al., 2021), supporting the hypothesis that the presence of a neuroinflammatory process might directly influence the metabolism of these excitatory amino acids.

| DISCUSS ION
Although there is still some controversy regarding the status of L-Asp as a neurotransmitter (Herring et al., 2015), this molecule is considered as the secondary excitatory amino acid in the CNS, with some findings suggesting that L-Asp and L-Glu may be coreleased (Andersen et al., 2021;Docherty et al., 1987;Fleck et al., 1993). In particular, neuropharmacological studies indicate a role for L-Asp as a selective endogenous agonist of NMDARs, since its application in rodent brain slices elicits inward currents blocked selectively by NMDAR antagonists (Balazs et al., 2012;Patneau & Mayer, 1990). Hence, alterations of neuronal or glial L-Asp metabolism may contribute to the pathogenesis of glutamatergic synaptic dysfunction and, ultimately, excitotoxicity in MS. In agreement with this possibility, previous work has demonstrated that the enzyme aspartate aminotransferase (AST) and the malate-aspartate shuttle system, which regulate L-Asp metabolism (Satrustegui & Bak, 2015), play a critical role in the microglial pro-inflammatory activation during neuroinflammatory processes (Zhou et al., 2021). Additionally, it has been demonstrated that neuroinflammatory condition occurring in EAE mouse brain triggers reduced expression and functioning of glutamate aspartate transporter 1 (GLAST-1)/excitatory amino acid transporter 1 (EAAT1) in glial cells (Mandolesi, Gentile, Musella, Fresegna, et al., 2015), potentially altering not only L-Glu but also L-Asp and TA B L E 2 Correlation analysis between CSF L-aspartate content and demographic and clinical parameters of MS and OND patients. D-Asp reuptake, as EAAT system recognizes both Asp enantiomers (Palacin et al., 1998). In keeping with this, during CNS inflammatory processes, HPLC measurements showed significantly decreased L-Asp levels in the spinal cord and brainstem of EAE mice, compared to controls (Musgrave et al., 2011). Similarly, NMR spectroscopy experiments confirmed L-Asp downregulation in the spinal cord (Battini et al., 2018) and brain (Brenner et al., 1993) of animal models affected by the neuroinflammatory disorder.
In line with previous data, our HPLC results confirmed a significant reduction of L-Asp levels in the cortex and spinal cord of EAE mice, compared to controls. Noteworthy, we showed that also L-Glu levels were concomitantly reduced in both CNS regions of EAE mice.
This neurochemical evidence suggests that neuroinflammatory con- glutamate binding site of NMDARs . Based on its neuropharmacological features, previous works have demonstrated that D-Asp treatment enhances glutamatergic cortical transmission and slightly mitigates fatigue symptoms in progressive MS patients , while in animal models of MS, it delays the onset of motor symptoms, attenuates their severity, promotes myelin recovery and decreases IL-6 serum levels (Afraei et al., 2017;de Rosa et al., 2019). On the contrary, exaggerated high levels of D-Asp in a genetic mouse model lacking the DDO enzyme that selectively catabolizes this NMDAR agonist, exacerbate EAE symptoms (Grasselli et al., 2013). Interestingly, our HPLC analysis unveiled for the first time a significant reduction of D-Asp levels within the cortex of EAE mice, suggesting a brain-specific alteration in the metabolism of this neuroactive amino acid. This evidence may explain the beneficial effect of its supplementation previously reported in EAE condition (Afraei et al., 2017;de Rosa et al., 2019). Differently from the brain tissue, CSF D-Asp levels were below the HPLC detection limit in the whole cohort of MS patients tested. Therefore, the present data indicate that HPLC might not be sensitive enough to quan-  (Garseth et al., 2001;Qureshi & Baig, 1988).
However, it remains still puzzling whether CSF concentrations of these excitatory amino acids represent per se a reliable biochemical signature of MS given that other previous works revealed unaltered or increased L-Asp and/or L-Glu levels in the CSF of MS patients (Klivenyi et al., 1997;Launes et al., 1998;Sarchielli et al., 2003;Stover et al., 1997). However, differences in age, gender, disease duration, sample sizes and criteria of patients' enrollment, and/or the substantial variability in L-Asp and L-Glu levels measured in our and others' neurological control groups may account for such neurochemical discrepancies. Also, CSF sampling site, diet, and circadian rhythms may contribute to the reported differences among studies (Arciero et al., 2008;Janssens et al., 2019); therefore, in the present work, to avoid possible confounding influences on metabolite levels from these factors, the LP procedures in MS patients and OND were performed from 8:00 to 10:00 am, after overnight fasting.
Our results also showed a mild Immunoinflammatory processes in MS are intimately associated with the activation of the glutamatergic system in the CNS (Levite, 2017). Consistent with this knowledge, a significant positive correlation between L-Glu levels and IL-17-mediated inflammatory and excitotoxic events was reported in the CSF of RR-MS patients (Kostic et al., 2014). In line with this, we have previously reported a strong negative correlation between CSF L-Glu and

CO N FLI C T O F I NTER E S T S TATEM ENT
The authors declare the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article:

DATA AVA I L A B I L I T Y S TAT E M E N T
The datasets used and analyzed in the current study are available from the corresponding authors upon reasonable request.