Neuroinflammation Induced by Intracerebroventricular Injection of Microbial Neuraminidase

In the present paper, we describe the facts that took place in the rat brain after a single injection of the enzyme neuraminidase from Clostridium perfringens into the right lateral ventricle. After injection, it diffused through the cerebrospinal fluid of the ipsilateral ventricle and the third ventricle, and about 400 μm into the periventricular brain parenchyma. The expression of ICAM1 in the endothelial cells of the periventricular vessels, IBA1 in microglia, and GFAP in astrocytes notably increased in the regions reached by the injected neuraminidase. The subependymal microglia and the ventricular macrophages begun to express IL1β and some appeared to cross the ependymal layer. After about 4 h of the injection, leukocytes migrated from large venules of the affected choroid plexus, the meninges and the local subependyma, and infiltrated the brain. The invading cells arrived orderly: first neutrophils, then macrophage-monocytes, and last CD8α-positive T-lymphocytes and B-lymphocytes. Leukocytes in the ventricles and the perivascular zones penetrated the brain parenchyma passing through the ependyma and the glia limitans. Thus, it is likely that a great part of the damage produced by microorganism invading the brain may be due to their neuraminidase content.


INTRODUCTION
In agreement to Graeber et al. (1) the term neuroinflammation implies not only microglial activation but also the invasion of the nervous tissue and/or the cerebrospinal fluid (CSF) by bloodcirculating leukocytes. Inflammation of the brain can be triggered by the presence in the ventricle, the meninges, or the nervous parenchyma of biological agents such as bacteria or viruses, or by a plethora of chemical or mechanical stimuli (2)(3)(4)(5). In any case, the clinical relevance of brain inflammation lies in part on the detrimental or even lethal effects of the invading defensive cells or proteins, which eventually can give rise to chronic neurodegenerative diseases (6,7).
Among the many agents that trigger neuroinflammation some are located in the microbial envelope, such as lipopolysaccharides (8,9) or the enzyme neuraminidase (NA) (10). Indeed, the abundance of NA in the outer surface of certain viruses and bacteria has been related to their pathogenicity (2,(11)(12)(13)(14). So much so that a single mutation in the hemagglutinin-NA (HN) surface complex of the mump virus is enough as to greatly decrease its pathogenicity, an interesting feature in the design of vaccines (15). Moreover, inhibitors of the NA activity are among the most important antiviral compounds [i.e., oseltamivir; (16)]. Despite all evidences supporting a pre-eminent role of NA in neuroinflammation, little is known about its exact mechanism of action and the events that are triggered after its administration. Moreover, the molecular differences among the neuraminidases that can be found in the microbial covers, probably lead to differences in the pathogenesis that they provoke (17).
Clostridium perfringens is a ubiquitous bacterium that possesses NA at its surface (18). NA from C. perfringens is easily available, so we used it before to desialylate some molecules present in the ventricular CSF (19). We reported that a single intracerebroventricular (ICV) injection of a high dose of NA (about 10 µg) produced the massive loss of the multiciliated ependyma followed by acute inflammation of meninges and ventricles and the development of hydrocephalus (10). By decreasing the amount of injected NA, only a partial detachment of the ependymal layer occurred in the injected lateral ventricle, but not in the contralateral ventricle or the third ventricle. This model turned also useful to investigate the importance of the ependyma on postnatal neurogenesis (20). We investigated the time course of inflammatory events after ICV administration of low doses of NA from C. perfringens, what might prove to be a useful model to study neuroinflammation. The facts suggest the importance of NA in degenerative events known to occur after microbial invasion of the central nervous system.

ANIMALS
Eighty-four adult male Wistar rats (body weight, bw, 250-300 g) were used in this study (Charles River Laboratories, Barcelona, www.frontiersin.org Spain). All animals were housed under a 12 h light/dark cycle with food and water available ad libitum. Animal procedures were done according to the European Union (86/609/EEC) and Spanish (RD 1201/2005) legislations.

HISTOLOGICAL PROCEDURES
Animals were anesthetized with 2,2,2-tribromoethanol and transcardially perfused with saline followed by Bouin's fixative. Brains were removed and immersed in the same fixative for 24 h at room temperature (RT), and later embedded in paraffin wax. Sevenmicrometer sections were obtained from each brain, and series along the region of interest (from Bregma −0.3 to −1.2 mm approximately) were mounted in gelatine treated slides.

LECTIN BINDING
The Limax flavus agglutinin (LFA, Calbiochem, San Diego, CA, USA) was chosen for its exclusive affinity for sialic acid. Deparaffinized thin sections were incubated with 4 µg/ml LFA in lectin buffer (1 mM Cl 2 Ca, 1 mM Cl 2 Mg, 0.6% TRIS, pH 7.6) for 1 h at RT. The lectin bound to the tissue was later identified using a rabbit anti-LFA antibody [(22); see Immunohistochemistry]. The peanut (Arachis hypogea) agglutinin (PNA, Vector, Burlingame, CA, USA. l-1070) has affinity for terminal residues of galactose, which is the subterminal sugar residue in complex-type glycoproteins.
After the removal of terminal sialic acid, galactose is exposed as terminal residue, and the PNA binding sites arise. To check for the removal of sialic acid by the action of NA, sections were incubated in 3 µg/ml peroxidase-labeled PNA in 0.1 M phosphate buffered saline (PBS) for 1 h at RT. The peroxidase activity was detected with 0.1 M PBS containing 0.05% diaminobenzidine (DAB; Sigma) and 0.03% hydrogen peroxide (Merck).

IMMUNOHISTOCHEMISTRY
Immunohistochemistry was carried out on deparaffinized tissue sections using either immunoperoxidase or immunofluorescence techniques. The primary and secondary antibodies used are listed in Tables 1 and 2, respectively. All of them were diluted with PBT buffer (0.3% bovine serum albumin, 0.3% Triton X-100 in PBS pH 7.3). Primary antibodies were incubated overnight at RT. Negative controls for the immunostaining consisted in equivalent sections subjected to the same protocol but omitting the primary antibody.
Sections for immunoperoxidase procedures were incubated in 3% hydrogen peroxide and 10% methanol in PB 0.1 M to quench endogenous peroxidase activity, prior to overnight incubation with the primary antibody. Biotinylated secondary antibodies were used ( Table 2; 60 min incubation), along with the avidin-biotin-complex (ABC) amplification method. The ABC reagent was prepared according to manufacturer's instructions (Thermo Fisher Scientific, Waltham, MA, USA) and incubated during 30 min. Peroxidase activity was revealed with 0.05% DAB and 0.03% hydrogen peroxide. All the incubations were performed in a moist chamber at RT.
To study the co-localization of particular markers, double immunofluorescence was performed using the following pairs of antibodies: ICAM1 and IBA1, IBA1, and IL1β, myeloperoxidase (MPO) and IL1β, and CD3ε and CD8α. The primary antibodies were mixed and incubated simultaneously (overnight at RT). Alexa fluor secondary antibodies ( Table 2) were used. Coverslides were mounted with buffered glycerol containing 10% of the anti-fading agent Mowiol 4-88 (Calbiochem/EMD Chemicals, San Diego, CA, USA). The sections were then observed under a confocal microscope (Leica, SP5 II). Some antibodies required specific antigen retrieval pretreatment prior to immunostaining. Thus, the sections incubated with anti-CD3ε and anti-CD8α were pretreated with  10 mM TRIS-EDTA buffer pH 9.0, and the sections incubated with anti-PAX5 were pretreated with 10 mM citrate buffer pH 6.0. In both cases, pretreatment was done at 95°C in microwave for 8 months.

QUANTITATIVE STUDIES
Two different strategies were used depending on the staining properties of the cells or structures to be quantified. When the quantified resident cells displayed processes or irregular shapes (microglia, astrocytes, and endothelial cells) the immunostained area was measured. If the cells had well-defined profiles easily distinguishable from each other (infiltrating leukocytes), a direct count of the number of cells was preferred.

Quantification of ICAM1, IBA1, and GFAP immunostaining
To study the response of the resident cells to NA-injection, we quantified the area immunostained with ICAM1, IBA1, and GFAP at different times (0, 4, 24 h, 7, and 15 days) after the inoculation of NA. These times were determined after the qualitative observation of the differences between the injected and the contralateral sides over time.
A previous detailed study of immunostained serial sections allowed the selection of the levels to be studied. Thus, two sections from each animal (Bregma −0.3 and −0.92 mm approximately) were chosen for the quantification. In any case, the selected sections were close to the foramen of Monro but far enough from the site of injection as to avoid areas of surgically damaged tissue. In each section studied, micrographs were taken from the striatal ventricular wall, and from the septum-fimbria wall, both in the injected side and in the contralateral one (red squares in Figure 2A show the approximate location of the micrographs). Pictures were then processed using the image analysis software Visilog 6.3 (Noesis, France). To introduce a correction related to the basal level of expression of a particular marker in each animal, the immunostained area measured in the contralateral side was subtracted to the value obtained in the same location in the injected side. This correction was made to subtract the basal expression of a marker in each animal, which was found to vary from rat to rat. Afterwards, the mean value was obtained from the two locations and the two sections studied from each animal, which was considered the total stained area of that animal. A total of five animals were analyzed in each time post-injection studied.

Quantification of infiltrating leukocytes
Cell infiltration was evaluated by immunostaining with the following antibodies: anti-MPO for neutrophils, anti-IL1β for monocyte-macrophages, anti-CD3ε for T-lymphocytes, anti-CD8α for cytotoxic T-lymphocytes, and anti-PAX5 for B-lymphocytes. We realize that these substances could also be synthesized by other cell types. However, presently and in agreement to many researchers, these markers are used to identify these different kind of immunologic blood-borne cells with a high degree of confidence. However, a low possibility exists that some cell does not belong to the corresponding group.
Immunoreactive cells were counted in three different locations: (1)

Statistical analysis
Differences between times post-injection of NA were evaluated by ANOVA (significance level of 0.05), followed by a Tukey post hoc test with a level of significance of 0.05. The values measured at time 0 were considered the baseline. The software SPSS® Statistics 20 (IBM®) was used for this analysis. In the histograms, the existence of significant differences between groups is indicated with different letters. Thus, two bars labeled with the same letter means that there is no significant difference between the two groups. Conversely, two bars labeled with different letters indicate that there is a significant difference between them.

RESULTS
Animals were sacrificed immediately after the injection of NA (T: 0 h), LFA lectin exhibited a negative halo (~400 µm from the ventricle surface) around the injected lateral ventricle and the rostral third ventricle (Figures 1A,C). The ependymal surface of the injected lateral ventricle showed no staining at www.frontiersin.org all ( Figure 1G). Conversely, the contralateral ventricle showed no halo (Figures 1A,D) and a conspicuous staining of the subependyma and the surface of the ependymal cells ( Figure 1H). On the contrary, when using PNA lectin (terminal galactose affinity) a positive label was found in the subependyma of the injected ventricle (Figures 1B,E) that was particularly intense in the surface of the ependymal cells ( Figure 1I). The walls of the contralateral ventricle showed virtually no binding to PNA (Figures 1B,F), and the surface of the ependymal cells was only feebly positive ( Figure 1J). After about a week, the ependymal cells that remained in the ventricular lining of the injected ventricle recovered the sialic acid and thus the LFA binding.
Control rats, which were injected with the vehicle, did never present any sign of brain inflammation. Only the brain cortex at the level of the needle path showed a few signs of local inflammatory reaction that spontaneously healed as observed in brains examined at different times after the injection. No evidences of inflammation were observed in the brain ventricles, the meninges or the periventricular regions.
The elimination of sialic acid triggers a series of events resulting in inflammation. Some responses depend on changes in resident brain cells while others on infiltrating blood leukocytes. Therefore, both cell populations were quantified afterwards.

NA INDUCES UPREGULATION OF IBA1, GFAP, AND ICAM1 IN RESIDENT CELLS
In NA-injected animals, the immunoreactivity to IBA1 antiserum increased notably in the affected periventricular zones (compare Figures 2E,F). We quantified the differences in two selected regions of the striatum and the fimbria-fornix at different post-injection times. IBA1 was upregulated from early times after NA injection, and the basal levels were recovered after about 2 weeks (Figure 2B). A notable increase in IBA1 immunoreactivity was also evident in the injected side in places located far from the site of injection, as is the optic tract ( Figure 2K).
GFAP staining in astrocytes of control rats (non-injected or saline-injected) was as expected, with no staining detected in the ependyma or the choroid plexus. As for IBA1, quantification of the GFAP stained area was carried out in selected regions of the striatum and the fimbria-fornix area of the injected ventricle relative to the contralateral one. GFAP staining showed a higher expression 24 h after NA-injection, which further increased one week later. It remained overexpressed even 2 weeks post-injection ( Figure 2C). The increase in immunoreactivity was clearly observed also in other region such as the stria medullaris (compare Figures 2G,H). Regions located farther from the site of injection also presented increased GFAP expression in the injected side with respect of Frontiers in Medicine | Pathology the contralateral one, a fact particularly evident in the optic tract ( Figure 2L).
In control rats, the wall of the large vessels of the meninges, the choroid plexus, the subependyma around the foramen of Monro, and the surface of the ependymal cells (especially of the choroid plexus) were immunoreactive to ICAM1. Parenchymal capillaries were virtually negative. Conversely, in the injected rats the wall of the parenchymal vessels located in the subependyma of the affected ventricles (up to~400 µm from the ventricle surface) became ICAM1-immunoreactive (Figure 2A; compare Figures 2I,J). As for IBA1 and GFAP, the changes in immunoreactivity to ICAM1 were quantified in the same selected zones of the lateral ventricle wall. There was a progressive increase in immunoreactivity that was evident 4 h after the injection and peaked at 24 h ( Figure 2D). ICAM1 expression decreased afterwards to reach control levels after about 2 weeks.
In control rats, IBA1 stained epiplexus cells in the choroid plexus and very few subependymal microglial cells (Figure 3A). In the experimental animals, a quite striking observation was that some IBA1-positive cells beneath the ependyma, seemed to penetrate between the ependymal cells even before the onset of blood cell infiltration (Figure 3B and inset). In similar locations, IL1βpositive cells could be found at early times after NA administration (Figures 3C,D). Moreover, double immunostaining demonstrated the existence of cells that were positive to both IBA1 and IL1β (Figures 3E,F).

NA INDUCED EXTRAVASATION OF BLOOD LEUKOCYTES
The most noticeable observation after the ICV inoculation of NA was the extravasation of leukocytes in the injected, but not the contralateral side. From about 4 h after the injection, cellular infiltration took place in large blood vessels located in the meninges, choroid plexus, and subependymal parenchyma. Extravasation in smaller vessels was infrequent. The extravasated leukocytes penetrated the brain parenchyma directly from the vessels passing through the blood-brain barrier, the pial barrier, or the ependyma of the choroid plexus ( Figure 4A) and of the ventricle walls (Figures 4B,C,E). To identify the nature of infiltrating cells, we used antibodies against MPO for neutrophils (Figures 4C,D), IL1β for monocyte-macrophages (Figures 3F and  4B,D), CD3ε and CD8α for T-lymphocytes (Figures 5A-G), and PAX5 for B-lymphocytes (Figures 5H-J). Some infiltrating cells, seemingly of the macrophagic-monocyte lineage, also stained for IBA1 (Figures 3E and 4A) and ICAM1 (Figures 4E-G). Double immunostaining with anti-MPO and anti-IL1β ( Figure 4D) revealed no co-localization of these two markers, indicating that the antibodies discriminate quite well between neutrophils and macrophages. Moreover, no polymorphonucleated cell was stained using anti-IL1β, IBA1 or ICAM1 at any time post-injection ( Figures 4A,B,F). It is likely that the monocyte-macrophage cells may be recognized by the IBA1 antiserum. Moreover, some of them (i.e., those that were activated) were also recognized by IL1β and ICAM1 antibodies (Figures 4B,D,E-G).
Lymphocytes infiltrated from the same vessels that neutrophils and macrophages (Figure 5), crossed the choroid plexus ependyma (Figures 5C,I), and invaded the ventricle (Figures 5A,I), the meningeal spaces (Figures 5D,J), and the brain parenchyma (Figures 5B,H). As revealed by double immunofluoresce using anti-CD3ε and anti-CD8α antibodies, most of them were cytotoxic T-lymphocytes (Figures 5E-G). B-lymphocytes  (PAX5-positive) were also present, but they were much less abundant (about one-tenth) than T-lymphocytes (Figures 5H-J).
Infiltrating cells were counted at different post-injection times in three selected locations (Figure 6): (i) the choroid plexus of the injected ventricle (Choroid plexus in Figure 6); the contralateral plexus showed virtually no activation; (ii) the zone around the ipsilateral foramen of Monro, where large venules are found (Vessels in Figure 6); (iii) the region of the meninges close to the ipsilateral side of the optic chiasm (Optic chiasm in Figure 6).
Neutrophils were by far the most numerous infiltrating cells, followed by T-lymphocytes and monocyte-macrophages, while B-lymphocytes were the least abundant. The choroid plexus was the richest source of blood cells, followed by the large vessels in the meninges and those around the foramen of Monro. The meningeal vessels, although situated far from the site of injection (i.e., the optic chiasm) were quite reactive.
Regarding the time course of infiltration, the earliest events were the invasion of neutrophils (MPO-positive; Figures 4C,D) and macrophages (IL1β-positive; Figures 4B,D), which reached a maximum only 4 h after the inoculation of NA; at 24 h these cells were already clearing (Figure 6). Lymphocytes (CD3ε-positive; Figures 5A-D) appeared later, and peaked at about 24 h (Figure 6). www.frontiersin.org  In the choroid plexus, these cells were more persistent and remained for up to 2 weeks.

DISCUSSION
The present work describes a novel model of neuroinflammation based on the ICV administration of the enzyme NA, which specifically removes terminal sialic acid from carbohydrate chains. This model intends to demonstrate that the NA activity from brain-targeting pathogens might result in pleiotropic inflammatory changes. The events that follow NA administration closely resemble those of an inflammatory process, and could be divided into phases. The phase 1 comprises the first 4 h after injection; the events that take place during this period are (a) the cleavage of sialic acid from the ependymal surface and the periventricular parenchyma, (b) the detachment and death of a fraction of the ependymal cells; and (c) the response of resident cells such as ventricular macrophages, microglia, astrocytes and endothelial cells of the vessels. The phase 2 starts about 4 h after NA injection. Extravasation of leukocytes occurs from certain brain vessels, resulting in the invasion of the affected zones in an ordered sequence of neutrophils, macrophages, and lymphocytes. These cells invade mostly the ventricles, the meninges, and can even penetrate the brain parenchyma in the affected regions.

TOXICITY OF NEURAMINIDASE
The enzyme NA from C. perfringens used in the present investigation contains <0.1% of unspecific contaminating protease activity. Thus, we think that the effects described may be mostly www.frontiersin.org attributable to the specific cleavage of sialic acid. Moreover, we did also inject NA from other sources, named the bacteria Arthrobacter ureafaciens, the Newcastel virus, and the influenza virus and obtained similar results. The lectin study demonstrated that the NA from C. perfringens cleaves sialic acid and exposes galactose residues from the ependyma and the subependymal parenchyma to a depth of about 400 µm. The desialyzation affects the injected ventricle and (at least) the rostral part of the third ventricle but not the contralateral ventricle. In previous studies using a higher dose of NA (10), desialyzation extended up to the level of the foramina in the fourth ventricle. In this location, NA seems to escape from the ventricle toward the outer CSF compartments. By this route, the NA could exert its effects in the meninges and in places located farther from the injection site such as the optic tract.
Many microbes that provoke ependymal detachment, inflammation, and hydrocephalus posses NA in their envelopes (14,23,24). Kelly and Greiff (11,12) demonstrated in mice the high toxic effects of bacterial and viral NA when injected directly into the brain. Also, in humans, NA activity seems to be associated with adverse prognosis in pneumococcal meningitis (13). So much so that mump viruses with low neurovirulence due to point mutations in hemagglutinin-neuraminidase capside proteins are being used to develop new harmless vaccines (15). Thus, it seems that the sole presence of NA in the CSF may trigger inflammation that gives rise to important brain damages.

ACTIVATION OF RESIDENT CELLS
The activation of microglia, astrocytes, and endothelial vascular cells after the administration of NA was studied. IBA1 and GFAP, used here to identify microglia and astrocytes, respectively, are known to increase its expression after brain insults (25)(26)(27). The increase in IBA1 expression could be related to a higher mobility of microglia (28,29). In this sense, we observed many IBA1-positive cells crossing the ependymal barrier after NA treatment and before the invasion of the brain by blood-borne cells. Interestingly, many of these cells also expressed IL1β, as observed by double immunostaining. Furthermore, we did not see any IL1β-positive cell in control specimens. This suggests that some IBA1-positive resident subependymal cells might have acquired IL1β positivity after the injection of NA, and that they move toward the lumen of the ventricle through the ependymal lining. However, since the ventricular macrophages (i.e., epiplexus cells in the choroid plexus) also expressed IBA1 and, after NA insults, begun to express IL1β, we do not have the definitive evidence to know whether the cells crossing the ependyma are indeed subependymal microglia or ventricular macrophages. IL1β is a known inflammatory mediator of the monocyte-macrophagic cell lineage (30). The capability of microglia to release IL1β was reported before in a variety of brain insults (31)(32)(33). It is likely that the increased expression of IBA1 and IL1β is related to the posterior pleocytosis, and reflects a tight relationship between the subependymal microglia and the resident ventricular macrophages. In this sense, both cell types share a common ontogenetic origin (34,35) and may serve as sentinels of the ventricular system, probably acting as an important firstline defensive barrier to a variety of brain insults. It has been recently shown that may exists at least two subsets of microglial cells capable of increase the expression of IBA1 after experimental insults (36), one in the acute phase and the other in the chronic phase. We agree that different populations of microglial cells, as for macrophages, may play different roles in neuroinflammation after traumatic injuries. This subject deserves future investigation in our model too.
Regarding the increased expression of GFAP, the activation of astrocytes in those places more affected by the NA could be associated to the formation of a glial scar that takes place after ependymal denudation (20,37). However, in structures located relatively far from the site of the injection, such as the stria medullaris or the optic chiasm, the activation of astrocytes and microglia is most probably related to the neuroinflammation process. It is quite evident that microglia and astrocytes play active roles in neuroinflammation but the exact mechanism of action is still controversial (38,39).
ICAM1 is a key player in the recruitment of leukocytes during inflammation (40,41). Its expression is increased in endothelial cells of vessels located in brain regions affected by an inflammatory insult (42)(43)(44)(45)(46)(47). We also found a significant increase in the expression of ICAM1 in the cerebral blood vessels located within the regions affected by the injection of NA. In agreement to Alvarez and Teale (48), the vessels displaying a more prominent extravasation were the venules of the leptomeninges, the choroid plexus and the subependyma around the injected lateral ventricle and the foramen interventricularis. These same vessels showed the higher expression of ICAM1, which was detectable even in control noninjected specimens. Interestingly they also presented the highest expression of receptors for IL1β (49) and for other inflammatory mediators (44). This suggests that inflammatory mediators, such as IL1β, might preferentially elicited extravasation in these vessels. The functional relevance of the different permeability of parenchymal vessels versus large meningeal, choroidal, and subependymal vessels is still an open question (48,50). On the other hand, ICAM1 was also expressed in the surface of the ependymal cells (especially in the choroid plexus) and of some ventricular cells. This could speak in favor of a role of ICAM1 in the penetration of some invading cells into the brain parenchyma (51)(52)(53).

INFILTRATION OF BLOOD LEUKOCYTES
In the model here described a profuse invasion of white blood cells takes place from about 4 h after the NA-injection. The time course of such neuroinflammation was similar to that of other models in rodents (23,(54)(55)(56). The first extravasated cells were neutrophils, followed by macrophages and then lymphocytes. In agreement to Ma et al. (57), most lymphocytes were T-lymphocytes (CD3εpositive) and of these most were CD8α-positive, as occurred in other cases of brain inflammation (58). The leukocytes crossed both the blood-brain barrier and the blood-CSF barrier, and invaded the perivascular spaces and the ventricular or meningeal CSF (59). They also reached the brain parenchyma, either directly from vessels located in the parenchyma or from the ventricle after crossing through the ventricular ependyma. The production by the invading or the resident cells of metalloproteinases that degrade specific junctional proteins seems to be essential for the invasion process (60). The passage of cells (resident or infiltrative) through the ependymal lining is still an unresolved question (61,62) and it should also require the production of metalloproteinases (63)(64)(65).
The nature of the signals involved in this trafficking of cells, the fate of the amoeboid cells, and the consequences in the periventricular tissues are still not well understood and deserve future investigations.
In summary, we describe here a novel model, compatible with life, to study the time course of aseptic meningitis and ventriculitis in rodents. The pathology, which is induced by the ICV injection of the enzyme NA, is highly reversible, and rats cure almost completely after some weeks. Apparently, only minor long lasting consequences remain in certain zones, mostly nearby the site of injection. However, these minor consequences could make the NA-treated animals more prone to develop neurological disorders, a possibility that deserves further research.

AUTHOR CONTRIBUTIONS
PD made the main laboratory works and hence contributed greatly to the acquisition of the information of the present paper. He drafted the work and approved the final form. They were in agreement with the contributions made by other authors. MR, JG, MM, MC, and MA supported technically all methods used in the present investigation; they also read critically the final form of the manuscript and approved it in agreement. JG also initiated the research. MA made a substantial contribution in the design and corrections of the final approved form of the manuscript, ensuring that any part of the work was appropriately investigated. Finally, FF and PL designed the research and conducted the group, making substantial contributions during all processes.

ACKNOWLEDGMENTS
This work was carried out with fundings from: Ministerio de Economía y Competitividad (SAF2010-19087) and Junta de Andalucía Consejería de Innovación Ciencia y Empleo (P-11-CVI-07637), Consejería de Salud SAS 08-0029. The Leica confocal microscope (SP5 II) was acquired with FEDER funds of the European Union. The authors are grateful to D. Navas Fernández for his help with confocal microscopy, the technical help of J. E. Casares-Mira and the skillfull immunohistochemistry support of E. Gil Lara from Centro Nacional de Investigaciones Oncológicas.