Spontaneous locomotion activity was evaluated using the open field test. Mice were placed in the center of a white polyvinyl chloride open field apparatus (50 cm long × 50 cm wide × 45 cm high) and allowed to move freely. Virtual areas, a central square (30 cm long), and a peripheral zone were determined with videotracking software (Viewpoint, Lyon, France). For 15 min, total distance traveled and time spent in the central area were recorded with the videotracking system.

Motor coordination was determined using a standard mouse rotarod (TSE, Bad Homburg Germany). We used the standard operating procedures described by the EUMORPHIA group program (31). Briefly, after a training phase to identify mice able to stay on the rod at 4 rpm for 60 s, mice underwent the test phase. This consisted of four trials with a 15 min intertrial interval. In each trial (T1–T4), mice were placed on the rod rotating at 4 rpm, the timer was started, and the rod was accelerated from 4 to 40 rpm in 300 s. The latency to fall off the rod was determined automatically. However, the timer was manually stopped if a mouse held onto the rod in a full rotation (“passive rotation”).

Ataxic behaviors or fine motor coordination were evaluated using the grid test adapted from Belknap (32). The grid test apparatus consisted of two 50 cm long × 50 cm wide × 50 cm high white acrylic boxes placed in the center of a videotracking system (Phenorack, Viewpoint, Lyon, France). Two cameras were used, one mounted on the roof (top view of mice) and another on the wall (side view of mice). One side of the white acrylic boxes was made of transparent acrylic to allow the side recording of mice by the second camera. Mice were placed on a 50 cm long × 50 cm wide wire grid floor (wire diameter 1.5 mm, grid mesh 1 × 1 cm) 30 cm above an infrared floor. Locomotor activity was recorded by the first camera. The experiment was stopped when the mice had traveled a distance of 300 cm. The number of times that mice slipped on the wire mesh floor was counted by an experimenter from recordings made by the second camera.

Muscular strength was measured for forelimbs or both hind and forelimbs with a grip strength meter (Bioseb, Chaville, France). We used the standard operating procedure described by the EUMORPHIA group program (31). Mice were lifted and held by their tails so that their limbs could grasp a wire grid. The mice were then gently pulled backward by their tails with their posture parallel to the surface of the bench until they released the grid. The grip strength meter digitally displayed the maximum force applied as the peak tension (in newtons) when grasp was released. Each mouse performed five consecutive trials, first with forelimbs and then with both hind and forelimbs. The mean of the five trials was taken as an index of limb grip strength.

Mechanical sensitivity was evaluated using the von Frey test by the method described by Chaplan et al. (33) and modified by Dixon (34). Mice were placed in individual plastic boxes (3.5 cm long × 8 cm wide × 14 cm high) on an elevated mesh platform, allowing full access to the paws. Mice were habituated to the apparatus for 45–60 min before paw withdrawal threshold measurement (PWT). Stimulation was applied using the up-and-down method. Calibrated von Frey filaments in the range 0.02–1.4 g (Bioseb Aesthesio®, Chaville, France) were applied perpendicularly to the right hindpaw with sufficient force to cause a slight buckling against the paw for 3–5 s. A positive response corresponds to a paw withdrawal, flinching, or licking. The PWT was determined as previously described (33).

Heat hyperalgesia was assessed using the hotplate test (35). The reaction threshold to a high-intensity heat stimulus was measured as an index of peripheral pain response. Mice were placed on a square metal surface heated to a temperature of 52 or 56°C (model-DS 37, Ugo Basile, Gemonio, Italy), and latencies to the first nocifensive response is characterized by the following signs: licking, shaking hindpaws, or jumping. As soon as the nocifensive response was observed, timer was stopped and the mouse was immediately removed from the hot plate. Data validation requisites the getting of two stable latencies (<1 s of difference) with a maximum of four trials performed per mouse. Data that did not fulfill this validation criterion were exclude. Cut-off latencies of 30 s (for 52°C) and 15 s (for 56°C) were used to prevent paw injury.

Cold allodynia was evaluated using the acetone evaporation test adapted from Chen et al. (36). Mice were placed in individual plastic boxes (3.5 cm long × 8 cm wide × 14 cm high) on an elevated mesh platform, allowing full access to the paws. Before stimulation, mice were habituated to the box for 30–45 min. A drop (20 μl) of acetone (Laboratoires Humeau, La Chapelle-sur-Erdre, France) was laid on the plantar surface of the hindpaw without touching the skin with the dropper tip, and the response was observed for 60 s after acetone application. Responses to acetone were scored as 0: no response, 1: quick withdrawal, flick, or stamp of the paw, 2: prolonged withdrawal or repeated stamping or flicking of the paw, 3: licking of the paw, or 4: jumping. The nocifensive score was the sum of all the responses evoked in the 60 s after acetone application. Acetone was alternately applied three times to each hindpaw. Because we did not manipulate one particular side of the CNS, an average of the six nocifensive scores was calculated for each mouse.

Given that some patients with MS experience trigeminal neuralgia [3.8%, according to Foley et al. (5)], we investigated orofacial sensitivity in the QuilA-EAE model.

Dynamic mechanical sensitivity was assessed using the paintbrush test adapted from Thibault et al. (37). Mice were allowed to move freely in individual boxes (3.5 cm long × 7.5 cm wide × 8 cm high) made of plexiglass and wire mesh for the front side allowing full access to the face and thus reducing any restraint effect. Prior to stimulation, the mice were habituated to the box for 30–45 min. Two paintbrushes were used to rub whisker pads, namely, a very smooth one made of marten hairs and a rough one made of pig bristles. The stimulus was applied five times to each whisker pad. The number of positive responses (quick head withdrawal, escape, or attempt to attack the paintbrush) was noted for each side. Because we did not manipulate one particular side of the CNS, scores for the left and right whisker pad were averaged for each mouse.

Static mechanical sensitivity was determined using the air puff test using a protocol adapted from Thorburn et al. (38). Mice were allowed to move freely in individual boxes (3.5 cm long × 7.5 cm wide × 8 cm high) made of plexiglass and wire mesh for the front side, allowing full access to the face and reducing any effects of restraint. Prior to stimulation, the mice were habituated to the box for 30–45 min. Three different air puff stimuli were applied to the whisker pad 1 cm away from the skin to avoid physical contact. Each air puff stimulus applied to the face was characterized according to its duration and maximal intensity perceived at the stimulation site [Table 2, for comparison stimulation of 10 psi described in Thorburn et al. (38) elicited 25.82 mpsi in our quantification system of maximal intensity perceived]. For each air puff stimulus, three stimulations were applied to the right and the left mouse whisker pad and nociceptive responses were scored. Between each air puff, there was a resting time of 30 min. Scoring was as follows: 0, no response; 0.25: brisk withdrawal of the head from the stimulus probe or an attempt to attack the probe; 1, single unilateral or bilateral forepaw swipe down the snout; and 1.5, continuous unilateral or bilateral forepaw swipe down the snout (three or more). Because we did not manipulate one particular side of the CNS, a total nociceptive response was calculated by averaging the scores for the left and right whisker pad for each mouse. The reproducibility of air puffs for each type of stimulation was determined using a force transducer MLTF500/ST device (ADInstruments Ltd, Paris, France) after 60 consecutive stimulations (Supplementary Figure S1).

Sensitivity to cold was evaluated using the facial acetone evaporation test adapted from Constandil et al. (39). Mice were acclimated for 15 min in a glass chamber with three mirrored sides (back and sides) and one transparent side (30 cm long × 30 cm wide × 30 cm high). Mice were gently restrained to allow full access to the face. A drop (20 μl) of acetone was laid on the left side of the face without touching the skin with the dropper and the mice were immediately returned to the glass chamber. The duration of facial rubbing or scratching behavior evoked by vaporization of acetone was counted for 1 min after acetone application.

To achieve this behavioral characterization, we evaluated cognitive deficit given that 20–25% of patients with RR-MS at disease onset suffer from cognitive impairment, which most frequently affects processing speed and memory (22, 23).

Sociability and social novelty preference were evaluated using an adapted protocol of the three-chamber sociability and social novelty test (40). The test comprised three sessions. Between each session, the mouse was returned to its home cage for 5 min. In the habituation session, the mouse was placed in the middle of a white polyvinyl chloride open field (50 cm long × 50 cm wide × 45 cm high) containing two empty wire cups (an inverted stainless steel wire pencil cup; Galaxy, Kitchen Plus) on both sides. A heavy cup was placed on the top of each inverted wire pencil cup to prevent the subject from climbing on top or moving it. In a first session, mice were allowed to explore freely for 5 min. Exploration times and movements of the mice were recorded using a videotracking system (Viewpoint, Lyon, France). Virtual zones were defined to delimitate the three chambers and the area of exploration of the wire cups. In the T1 session (5 min duration), an unfamiliar adult WT female was placed in one of the wire cups. This session allowed the evaluation of the sociability index (exploration time of unfamiliar mice—exploration time of empty cup / total exploration time of both cups). In the T2 session (duration 5 min), while the first intruder was still posted in its wire cup, another unfamiliar adult WT female was placed in the second cup. A novelty percentage was then calculated [(exploration time of unfamiliar mice/total exploration time of both mice) × 100].

Episodic memory was assessed using the novel object recognition test (41). The test comprised two sessions. During the first session T1, the mouse was placed in a white polyvinyl chloride cage (36 cm long × 14 cm wide × 21 cm high) on a layer of sawdust and was presented with two identical objects placed at the two opposite sides of the cage. The session was stopped after 20 s of object exploration. A cut-off time was fixed at 5 min and mice that had not performed this exploration were excluded from the study. Between T1 and T2, the mouse was moved back to its home cage for 10 min. In the second session T2, one object was replaced by a new object, and for 5 min, the exploration time for each object was measured. A recognition percentage was then calculated [(exploration time of new object/total exploration time of both objects) × 100].

Working memory was evaluated using the Y-maze test according to the standard operating procedure described by the EUMORPHIA group program (31). Spontaneous alternation behavior and exploratory activity were recorded. The apparatus was made of three equal arms (40 cm long × 10 cm wide × 16 cm high) made of black polyvinyl chloride radiating at 120° from each other. Mice were placed at the end of one arm, facing the end wall of the arm, and allowed to move freely through the maze for a 10 min session. Latency to leave the first arm and total number and sequence of entries into each arm were scored for each mouse. An arm entry was counted when the mouse had all four paws inside the arm. If mice completed fewer than eight arm entries within 10 min, they were excluded from further analysis. Spontaneous alternation was defined as entries into all three arms on three consecutive choices. The alternation score (%) represents an index of working memory and was calculated as follows: % alternation = 100 × [number of alternations/(total arm entries – 2)].

Anxiety was assessed using the elevated plus maze (EPM), which assesses the natural conflict between the tendency of mice to explore a novel environment and their tendency to avoid a brightly lit elevated open area (42). The EPM comprised of two open arms and two closed arms (37 cm long × 6 cm wide) that extended from a central platform (6 cm × 6 cm). The apparatus was constructed of Plexiglas (black floor and walls) and elevated at 50 cm above floor level. The mouse was placed on the central platform facing one of the enclosed arms and was allowed to explore the maze for 10 min. The number of open arm entries, the time spent in different parts of the maze (open and closed arms, central platform), and the total number of open and closed arm entries were recorded with a videotracking system (Viewpoint, Lyon, France). Mice completing fewer than eight arm entries within 10 min were excluded from further analysis.

Anxiety was also evaluated using the marble burying test adapted with minor modifications from Millan et al. (43). Mice were individually placed in polyvinyl chloride cages (36 cm long × 21 cm wide × 14 cm high) containing a 4 cm layer of sawdust and 20 glass marbles (diameter 1.6 cm) that were evenly spaced throughout the cage (five rows of four marbles). Mice explored the cage freely for 30 min. The number of marbles buried more than two thirds into the sawdust was counted at 30 min.

In Cohorts 1 and 2, the two-way ANOVA of body weight showed a significant main effect of DPI [F_(40,880) = 265.4, p < 0.0001 and F_(40,880) = 221.5, p < 0.0001 in Cohorts 1 and 2, respectively] but not group [F_(1,22) =0.4261, p > 0.05 and F_(1,22) = 2.718, p > 0.05 in Cohorts 1 and 2, respectively] or DPI × group interaction [F_(40,880) = 0.9247, p > 0.05 and F_(40,880) = 0.910, p > 0.05 in Cohorts 1 and 2, respectively]. For the entire duration of the study, no body weight difference was shown by post-hoc analysis comparing groups (Sidak test, p > 0.05). Post-hoc comparing DPI showed an increase in body weight with time (Sidak test, p < 0.05) (Figure 1A).

For EAE clinical score, two-way ANOVA in Cohorts 1 and 2 showed a significant main effect of DPI [F_(40,880) = 27.15, p < 0.0001 and F_(40,880) = 30.00, p < 0.0001 in Cohorts 1 and 2, respectively], group [F_(1,22) = 201.3, p < 0.001 and group F_(1,22) = 291.9, p < 0.0001 in Cohorts 1 and 2, respectively] and DPI × group interaction [F_(40,880) = 11.53, p <0.0001 and F_(40,880) = 18.34, p < 0.001 in Cohorts 1 and 2, respectively]. Post-hoc analysis showed a significant increase in EAE score between mice immunized with MOG₃₅₋₅₅ (EAE) and their controls (CTL) from D10 to D40 in Cohort 1 (Sidak test, D10–D14 p < 0.05; D15–D40 p < 0.001) and from D12 to D40 in Cohort 2 (Sidak test, D12–D40 p < 0.001). In both Cohorts 1 and 2, modeled disease onset started at D16 when EAE mice reached a score ≥1. Behavioral evaluation was assessed at D17 (beginning of disease) and D30 (disease fully installed). For each cohort, a mean EAE clinical score higher than 2 (limp tail with unilateral partial hindlimb paralysis) was never observed. Overall, the time course of the EAE score reproduced the clinical presentation of the QuilA-EAE model with alternating phases of relapse and remission (Figure 1B).

Pregabalin effect was evaluated on mechanical hyperalgesia at D30 in EAE mice after hypersensitivity was fully developed. Over a 2 h follow-up, the effect of vehicle (NaCl) or pregabalin (10 and 30 mg/kg, i.p. injection) was evaluated by the percentage of paw withdrawal evoked by a von Frey hair filament (1.4 g) in QuilA-EAE and CTL mice. The one-way ANOVA of the area under the curves obtained for each group and treatment showed a significant main effect of treatments [F_(2,24) = 10.75, p < 0.001 and F_(2,24) = 11.42, p < 0.001 in CTL and QuilA-EAE groups, respectively]. Post-hoc analysis showed a significant decrease in the percentage of paw withdrawal for CTL mice treated with pregabalin 10 and 30 mg/kg (Sidak test, p < 0.01 and p < 0.001 for Preg10 and Preg30 compared to vehicle, respectively) (Figure 6A, left). A significant decrease in the percentage of paw withdrawal was observed only in QuilA-EAE mice treated with pregabalin 30 mg/kg (Tukey's test, p < 0.001 vs. vehicle) (Figure 6A, right). Similar data were analyzed for a.4 g von Frey hair filament (Table 4). The one-way ANOVA analysis of AUC showed a significant main effect of treatment [F_(2,24) = 4.779, p < 0.05 and F_(2,24) = 4.010, p < 0.05 in CTL and QuilA-EAE groups, respectively]. Post-hoc analysis showed a significant decrease in the percentage of paw withdrawal for CTL and QuilA-EAE mice treated with pregabalin 30 mg/kg (Tukey's test, p < 0.05 vs. vehicle). These results show an analgesic effect of pregabalin on mechanical sensitivity in CTL and QuilA-EAE mice.

Due to the important effect of pregabalin observed in CTL mice, we decide to run a control experiment to eliminate a possible locomotor confounding effect of the drug. Pregabalin effect was evaluated on spontaneous locomotion using the open-field test in wild-type female mice. Over a 2 h follow-up, the effect of vehicle (NaCl) or pregabalin (10 and 30 mg/kg, i.p. injection) was evaluated by the total distance traveled. The one-way ANOVA of the area under the curves obtained for each treatment showed no main effect of treatments [F_(2,25) = 1.202, p > 0.05] (Supplementary Figure S2).

To complete this pharmacological characterization, the potential analgesic effect of pregabalin on heat hyperalgesia was assessed. Over a 2 h follow-up, the effect of vehicle (NaCl) or pregabalin (10 and 30 mg/kg, i.p. injection) was evaluated using the hotplate at 56°C. The one-way ANOVA of latency of first nocifensive response showed a main effect of treatment in the QuilA-EAE group [F_(2,27) = 6.777, p < 0.01] but not the CTL group [F_(2,27) = 0.7424, p > 0.05]. Post-hoc analysis showed a significant increase in latency in QuilA-EAE mice treated with pregabalin at 30 mg/kg (Tukey's test, p < 0.01 vs. vehicle) (Figure 6B, right side). These results show an analgesic effect of the higher dose of pregabalin on heat hyperalgesia in QuilA-EAE but not CTL mice.

Accurately defining motor defects from slight to major dysfunction is a key difficulty not only in MS research but also in other neurologic diseases with different levels of motor dysfunctions according to the disease stage, such as stroke or Parkinson's disease (52, 53). The most common tests used to assess motor function in mice are spontaneous locomotor activity in the open field and coordination in the rotarod test. However, both tests often lack the sensitivity needed to detect subtle alterations. In this variant of the EAE model in which motor dysfunctions are voluntary restricted, it is furthermore important to be able to detect these subtle alterations. As described by Khan et al. (15), we confirmed the absence of severe motor dysfunction with a mean EAE clinical score never exceeding 2 (limp tail with unilateral partial hindlimb paralysis), strengthened by results obtained in more dedicated tests: the open field, rotarod, and grip strength tests. Strikingly, our results obtained in the open field and the rotarod tests were in agreement with data obtained by Dalenogare et al. (20). Finally, weak impairments of fine motor coordination were observed only using the grid test, which enabled us to demonstrate and quantify this symptom in QuilA-EAE mice. The grid test is known to highlight subtle ataxic behaviors in mice, often related to cerebellar dysfunctions (54).

Anxiety disorders are found in 37.5% of MS patients (55). Until now, anxiety-like behaviors were only described in the CFA EAE mouse model. These experimentations showed a reduced exploratory behavior using the light/dark box test (56, 57), the elevated plus maze (58), and open field test (56, 58). In all these studies, behavioral experiments were performed during the pre-symptomatic phase of the disease before the appearance of motor defects. To our knowledge, we are the first to evaluate these behaviors in an QuilA-EAE model. Our data using the elevated plus maze, marble burying, and the open field test showed no anxiety-related behaviors in QuilA-EAE mice. The discrepancy between our results and those already described in the literature might be due to the difference between CFA and QuilA-EAE models or the difference in the timing of the behavioral evaluation (presymptomatic vs. symptomatic phase of the disease). Further experiments are now needed to conclude.

Some 45–60% of MS patients show cognitive impairment and several domains are affected, such as long-term and working memory, executive function, attention, and speed of information processing (22, 23). Social impairment and social cognition deficits have also been observed in MS (59).

Data on cognitive function in EAE mice are scant. The first studies included that of Olechowski, showing defects in working memory in a CFA EAE model using the novel object recognition test (60). De Bruin et al. showed a decrease in social recognition in the QuilA-EAE model induced in SJL mice using the social preference and social novelty test (61). Recently, a deficit of hippocampal memory was demonstrated in a CFA EAE model using a Barnes maze or a dedicated contextual fear conditioning procedure (62, 63) or in a chronic-relapsing EAE model using the hole-board test (64). In line with the literature, the data obtained in our study confirmed that if cognition defects do occur in QuilA-EAE mice, then they are slight. In fact, in only one cohort did we find a decrease in social recognition using the social preference and social novelty test. None of the other cognitive functions evaluated (working and episodic memories, anxiety-related behaviors) were affected.

The very limited occurrence of anxiety-related behaviors or cognitive dysfunctions in our study can be seen as a limitation of the QuilA-EAE model. However, before drawing conclusions, some aspects have to be taken into account. First, in our study, interfering symptoms were evaluated only at D30. It is thus still possible that some symptoms could arise earlier or later in the disease time course. Secondly, we did not evaluate depression-like behaviors in our model. Thirdly, regarding cognitive symptoms, in patients, information processing speed reduction is usually the first cognitive alteration and concerns only 20–25% of the patients at disease onset (65). It is thus probably difficult to explore such minor alterations in animal models, although they may be present. Finally, as in our experimental set up, QuilA-EAE and controls mice are bred together, a possible explanation of the moderate behavioral phenotype observed is a possible “fecal transplants” through coprophagia. Given that the vast majority of fecal transfer experiments realized in EAE mice described in literature use germ-free mice (66) or treatment with antibiotic cocktails during few days to weeks (67, 68) to ensure extensive microbiota changes, we assume that the probability of possible 'fecal transplants' through coprophagia is very low but still possible.