Increased excitability and reduced GABAergic levels in somatosensory cortex under chronic spinal cord injury

The complete or partial damage of ascending somatosensory pathways produced by a spinal cord injury triggers changes in the somatosensory cortex consisting in a functional expansion of activity from intact cortical regions towards deafferented ones, a process known as cortical reorganization. However, it is still unclear whether cortical reorganization depends on the severity of the spinal cord damage or if a spinal cord injury always leads to a similar cortical reorganization process in the somatosensory cortex. To answer these open questions in the field, we obtained longitudinal somatosensory evoked responses from bilateral hindlimb and forelimb cortex from animals with chronic full-transection or contusive spinal cord injury at thoracic level (T9-T10) to induce sensory deprivation of hindlimb cortex while preserving intact the forelimb cortex. Electrophysiological recordings from the four locations were obtained before lesion and weekly for up to 4 weeks. Our results show that cortical reorganization depends on the type of spinal cord injury, which tends to be more bilateral in full transection while is more unilateral in the model of contusive spinal cord injury. Moreover, in full transection of spinal cord, the deafferented and intact cortex exhibited similar increments of somatosensory evoked responses in both models of spinal cord injury - a feature observed in about 80% of subjects. The other 20% were unaffected by the injury indicating that cortical reorganization does not undergo in all subjects. In addition, we demonstrated an increased probability of triggered up-states in animals with spinal cord injury. This data indicates increased cortical excitability that could be proposed as a new feature of cortical reorganization. Finally, decreased levels of GABA marker GAD 67 across cortical layers were only found in those animals with increased somatosensory evoked responses, but not in the unaffected population. In conclusion, cortical reorganization depends on the types of spinal cord injuries, and suggest that the phenomenon is strongly determined by cortical circuits. Moreover, changes in GABAergic transmission at the deprived cortex may be considered one of the mechanisms underlying the process of cortical reorganization and increased excitability.


Introduction
Sensory information from the external world ascends from the peripheral receptors distributed along the entire body surface to the spinal cord and then to the primary somatosensory cortex (SSCx), the main hub for sensory information processing in the brain.Injuries to the spinal cord (SCI) abruptly interrupt the ascending pathways leading to a sensory deprivation at the level of SSCx regions corresponding to body areas located below the injury level.This lack of sensory inputs creates a physiological imbalance between intact (with preserved sensory inputs) and deprived cortical regions that in the long-term leads to the so-called phenomena of cortical reorganization (CoRe) (Endo et al., 2009), defined as an expansion of the cortical activation in response to peripheral stimulation from the intact towards the sensory-deprived regions (Endo et al., 2007;Ghosh et al., 2010;Humanes-Valera et al., 2013;Humanes-Valera et al., 2017;Jutzeler et al., 2019).
Experimental data obtained from animal models indicate that acute (i.e.few minutes to hours) functional changes after sensory deprivation take place at the cortical level in different sensory systems (Aguilar et al., 2010;Hengen et al., 2013;Lemieux et al., 2014;Griffen et al., 2017).These acute alterations in cortical excitability are thought to set the base for the initiation and establishment of CoRe when the sensory deprivation is irreversible, as in the case of a SCI (Yague et al., 2011;Yagüe et al., 2014).After the acute changes, a second phase evolving from days to months is initiated leading to a chronic condition in which cortical excitability, thereby CoRe, is stabilized (Jain et al., 1997;Endo et al., 2007;Jutzeler et al., 2019).In this context, the consensus is that increased magnitude of cortical responses to peripheral stimulation is the main functional feature of CoRe under chronic SCI, which has been confirmed with different experimental approaches as fMRI (Endo et al., 2007), voltage sensitive dye (Ghosh et al., 2010), and in vivo electrophysiological recordings (Humanes-Valera et al., 2017;Jutzeler et al., 2019).However, these experiments also show that CoRe represents an heterogeneous phenomena depending on the time elapsed after the injury, the injury type (contusion, transection, hemisection) and the severity (mild vs severe) (Jain et al., 1997;Ghosh et al., 2010;Endo et al., 2007;Vipin et al., 2016).In this line, a deep characterization and comparison of longitudinal cortical functional changes induced by distinct types of SCI is needed to finally understand CoRe, as well as to identify the physiological features that will dictate those individuals developing or not CoRe.
Most of studies about CoRe after SCI have been made in anesthetized animals, in which the cortical spontaneous oscillation is dominated by slow-wave activity (SWA) characterized by intercalated silent periods or down-states with active periods or up-states (Steriade et al., 1993;Shu et al., 2003;Crunelli et al., 2015).Since up-states events represent the synchronized activity of a neuronal population, distinct parameters of up-states (e.g.frequency, amplitude, slope) are commonly used to study the excitability of cortical network during spontaneous activity in control conditions (Rigas and Castro-Alamancos, 2007;Favero and Castro-Alamancos, 2013;Crunelli et al., 2015;Civillico and Contreras, 2012;Wester and Contreras, 2012) and in animals with SCI (Fernández-López et al., 2019).In addition, the use of anesthetic-induced SWA also allows to determine the ability of cortical networks to respond to peripheral stimulation (Jain et al., 1997;Ghosh et al., 2010;Endo et al., 2007) using either the short-latency responses (i.e.sensory-evoked potentials, SEPs, 5-50 ms) or a long-latency response (i.e.triggered up-states, TUPs, 50-200 ms) (Steriade et al., 1993;Rigas andCastro-Alamancos, 2007, 2009).Interestingly, TUPs have been used to study cortical excitability in animals without pathologies (Steriade et al., 1993;Rigas andCastro-Alamancos, 2007, 2009;Civillico and Contreras, 2012), therefore TUPs is a cortical feature that could be useful to study if a SCI alters cortical excitability.
In the present work, we addressed some open questions in the field related to cortical changes after SCI.First, we investigated whether the temporal features of CoRe, measured by the short-latency sensory evoked responses at the somatosensory cortex, depended on the type of SCI.Second, we determined whether the long-latency triggered cortical UP-states could be used as a functional biomarker of changes in cortical excitability thereby being part of the CoRe phenomenon.Finally, we determined the possible mechanism underlying the increased cortical excitability after SCI by measuring changes in GABAergic markers at the somatosensory cortex.Our overall results indicate that CoRe is a plastic phenomenon that occurs in different regions of the somatosensory cortex a different types of SCI induce differences in the percentage of cortical locations affected.In addition, changes in GABAergic transmission appear to be crucial to the establishment of CoRe after SCI.

Materials and methods
Experiments were performed in accordance with the European Union guidelines (Directive 2010/63/EU), all processes were approved by the Ethical Committee for Animal Research at the Hospital Nacional de Parapléjicos (reference number 152CEEA/2016).A total of 41 male Wistar rats (300-400 g) were used to complete three experimental groups: 1) control group with sham lesion (n = 11); 2) full transected SCI (Trx-SCI) at thoracic level T9-T10 (n = 20); 3) contusion group (Cntx-SCI) in which contusive SCI was performed at thoracic level T9-T10 (n = 10).All animals (2 per cage) were housed in standardized cages in a non-enriched environment, with food and water ad libitum, and kept at 23 • C on a 12 h light/dark cycle.
Electroencephalographic activity (EEG) was recorded using stainless cranial screws implanted bilaterally in both stereotaxic coordinates of the hindlimb primary somatosensory cortex (HLCx) and forelimb primary somatosensory cortex (FLCx) (Fig. 1A, right).General anesthesia was applied by inhalation of a mixture of isoflurane in oxygen (1.5-2%), that was adjusted for each individual to obtain a stable brain state of SWA in the continuously recorded EEG.Electrical stimulation applied in contralateral forelimb and hindlimb (Fig. 1A, left) were longitudinally obtained in pre-lesion condition and weekly after SCI for up to four weeks.After spinal cord transection or contusion, all animals were housed in pairs in a non-enriched environment and were handled for manual voiding of the bladder until normal voiding responses return.All efforts were made to minimize the number of animals used and their suffering and to apply the 3Rs regulation.

Cranial screws implant
In each animal, cranial screws for electrophysiological recordings were implanted in SSCx coordinates, which guaranteed that somatosensory evoked potentials were recorded from the same cranial location during the entire experimental protocol.Animals were anesthetized with pentobarbital (50 mg/kg i.p.) and administered xylazine (10 mg/kg i.p), the body temperature was kept constant (36.5 • C) using an automatically controlled heating pad (Cibertec SL, Madrid, Spain) and the eyes were protected with ophthalmic gel (Lubrithal, Dechra, Northwich, UK).For each animal, a deep anesthesia level was confirmed when spinal and corneal reflexes were lost, and then the animal was placed in a stereotaxic frame (SR-6 Narishige Scientific Instruments, Tokio, Japan).After shaving the head, lidocaine (2%) was applied subcutaneously to locally anesthetized skin and muscles over the skull.A scalpel was used to perform a longitudinal incision in the midline of the skull and all tissues above the cranial bone were removed.Once the cranial surface was exposed and carefully dried, small holes were drilled bilaterally over HLCx (AP -1.5 mm; ML 2.5 mm) and FLCx (AP 0.5 mm; ML 4 mm) according to coordinates from Paxinos and Watson (2007).An additional hole was performed above the cerebellum.Stainless steel screws (Precision Technologies Supplies, M1.2 × 4 DIN 84 A2; 1.2 mm diameter; 4 mm length) were carefully placed in every location using a mini screwdriver avoiding any compressing of the brain surface.After correct screw placement, dental cement (Duralay, Reliance Dental Manufacturing LLC, IL, USA) was applied while keeping the head of screws exposed to facilitate electrophysiological recordings.Screws over the HLCx and FLCx were used as extracranial electrodes to obtain EEG recordings, while the screw over the cerebellum was used as reference.After that, animals were treated immediately and up to 4 days with enrofloxacin (5 mg/kg s.c.) and meloxicam (1 mg/kg s.c.).

Surgery for chronic SCI and post-operative care
General process for all three animal groups was as follows: animals were anesthetized with pentobarbital (50 mg/kg i.p.) and administered xylazine (10 mg/kg i.p.) as a muscle relaxant and analgesic.The body temperature was kept constant (36.5 • C) using an automatically controlled heating pad and their eyes were protected with ophthalmic gel.Once rats were anesthetized, the skin of the surgical area was shaved, swabbed with ethanol 70% and disinfected with Povidone iodine strips.A midline incision was made in the dorsal thoracic region over vertebral T8-T11 level.Then, layers of muscles were carefully retracted keeping both intact the musculature and main blood vessels in E. Alonso-Calviño et al. the surgical area.Finally, once vertebral bones were exposed a laminectomy was performed at thoracic level T9-T10 which allowed direct access to the spinal cord.
1. -Full transected SCI group (Trx-SCI): the duramater was removed and the spinal cord was then completely transected using surgical scissors.Complete transection was confirmed visually under the surgical microscope by the total separation of the borders.Once SCI was confirmed, muscular layers were orderly sutured and finally skin was carefully sutured to close the incision.Before the animals were awakened, the post-surgical treatment was applied using marbofloxacin (5 mg/kg s.c.) and meloxicam (1 mg/kg s.c.).The first day after the surgery the animal's bladder was emptied and they were administered marbofloxacin, meloxicam and saline (same dose previously described).Same pharmacological post-surgery care was applied for 5 days, and the helping to bladder voidance was practiced until normal voiding responses return.
2. -Contused SCI group (Cntx-SCI): after laminectomy, the animals were placed on the experimental table of an Infinite Horizon Impactor Device (Precision System and Instrumentation, Lexington, NY, USA), the vertebrae located rostral and caudal to the laminectomy were stabilized with especial Adson forceps, then the exposed spinal cord was subjected to a severe impact of 250 kdynes (255.4 ± 14.7; mean ± SEM) with a mean tissue displacement of 1255.5 μm ± 116.3 SEM.After contusion, muscular layers were orderly sutured and finally skin was carefully sutured to close the incision.The same post-surgical treatment previously described for animals with spinal transection was applied.
3. -Sham group (sham surgery): this group underwent all aspects of the surgery, including laminectomy, with exception of transection or contusion of spinal cord at thoracic level.Same post-surgical treatment was applied with exception of helping with bladder voidance.

Electrophysiological recordings
For electrophysiological recordings, animals were anesthetized with a mixture of isoflurane in oxygen (1.5-2%, 2 L/min) applied by a mask (Medical Supplies & Services, Int. Ltd., Keighley, UK).The body temperature was kept constant (36.5 • C) using an automatically controlled heating pad (Cibertec SL, Madrid, Spain).The percentage of anesthetic was adjusted to produce an EEG showing cortical SWA that is characterized by alternating silent and active cortical periods with main frequency < 1 Hz (Fig. 1B, left).The same anesthetic level was kept constant in each animal during the following weeks.Continuous EEG from both hemispheres were recorded using an amplifier CyberAmp 380 (Molecular Devices LLC, CA, USA).Signal was amplified (x100), band pass filtered (1-300 Hz), online digitized using a Digidata 1440a (Molecular Devices LLC, CA USA), and finally signals were stored in a PC for posterior offline analysis.
Electrical stimulation was applied using bipolar needle electrodes placed subcutaneously on each side of the forelimb and hindlimb wrist.Note that stimulation protocol started only when the EEG was stable for >10 min under SWA.A Grass S88X stimulator (Grass Astro-Nova, Inc. RI USA) was used to produce electrical pulses with following parameters: pulse duration = 0.5 ms; frequency = 0.5 Hz; two different intensities were used: low intensity (5 V) and high intensity (50 V).The protocol of peripheral stimulation consisted of 50 stimuli applied at low and high intensities to each of four body extremities.The rationale for the stimulation at different intensities is that different somatosensory modalities can be recruited by these two intensities (Lilja et al., 2006).However, in our experimental approach the use of these two different intensities allowed us to compare the activation of the somatosensory cortex by the stimulation of minimal and maximal amounts of peripheral inputs.
Recordings of spontaneous activity and somatosensory evoked cortical potentials were obtained in control conditions (pre-lesion, one week before SCI), and weekly after SCI for up to 4 weeks (Fig. 1A-B).A) Schematic representation of the experimental approach: the animal with thoracic SCI, the inset shows screws electrodes placed on cranial coordinates of the representation of forelimb (FLCx) and hindlimb (HLCx) in the primary somatosensory cortex, red arrowheads indicate peripheral electrical stimulation placed bilaterally in the hindlimbs (HL stim) and forelimbs (FL stim).B) Electrophysiological recordings obtained simultaneously from FLCx (black) and HLCx (gray): on the left, during spontaneous slow-wave activity (SWA); on the right, the somatosensory evoked potentials (SEP) recorded in FLCx and HLCx (red circles) in response to forelimb stimulation (note the higher magnitude specifically in FLCx).C) Examples of SEPs recorded from FLCx (upper traces) and HLCx (lower traces) in response to contralateral stimulation of FL (black traces) or HL (gray traces) at low intensity.In both cases, the higher response magnitude match the stimulation of corresponding somatotopic limb.

EEG data analysis
All animals were recorded for the first time one week after cranial screws were implanted.These recordings were considered as control conditions (pre-lesion) before spinal cord injury or sham lesion.Moreover, the recordings under control conditions were used to confirm the correct cranial locations of screws and the quality of somatosensory evoked responses obtained.In this regard, only cranial screw's locations showing undoubtedly electrophysiological responses in the first recording session were included in the analysis, while a small percentage of implanted cranial screws (11%: 11 out of 101) showed no responses to peripheral stimulation or showed an electrophysiological artifact instead a physiological signal, therefore they were excluded of our study.As result of this initial selection, data were obtained from 29 cranial locations in sham group (14 from HLCx and 15 from FLCx; 11 animals); 53 cranial locations in full transected SCI group (31 from HLCx and 22 from FLCx; 20 animals); 19 cranial locations in contused SCI group (all obtained from HLCx; 10 animals).
First step was intended to confirm that cranial screws were perfectly located over cortical coordinates of hindlimb somatosensory cortex (HLCx) and forelimb somatosensory cortex (FLCx).For that purpose, hindlimb and forelimb were stimulated using single electrical pulses at a frequency of 0.5 Hz, and two different intensities were applied (5 V and 50 V).

Evoked responses
SEPs were obtained by averaging the raw signals recorded from the cranial screws across stimuli.Only stimuli that happen during cortical silent periods of SWA were selected and averaged for posterior analysis (minimum of 25 stimuli per session).The magnitude of the SEPs responses was quantified as the amplitude (mV) of the first signal deflection to the maximum of negative peak in a time window of 5-25 ms after stimuli.

Quantification of Stimulus-triggered Up-States
A quantification of cortical stimulus-triggered up-states (TUPs) by peripheral stimulation was obtained in all time points for all the studied populations in the HLCx and FLCx from the SEP recordings.A stimulus was considered to have triggered an up-state (TUP) if it satisfied the following criteria: (1) the stimulus occurred in a downstate (i.e., no cortical activity was observed in the 300 ms before the stimulus); (2) after the initial short-latency response, a long-latency activation (corresponding to an up-state) was observed within 200 ms of the stimulus (Civillico and Contreras, 2012;Bermudez-Contreras et al., 2022).TUPs were then quantified in terms of percentage, as the ratio of the number of stimuli that triggered an up-state and the number of stimuli that occurred in a down-state.This percentage was calculated only at low intensity stimulation because high stimulation evoked almost 100% of TUPs during SWA.

Histological procedures 2.5.1. Immunofluorescence
Rats were perfused transcardially with a 0.1% heparin in PBS solution, and then with 4% paraformaldehyde.Brains were excised and postfixed in 4% PFA for 24 h at 4 • C. The following day, the tissue was immersed in 10%, 20% and 30% sucrose in PB 0.1 M. Coronal sections of 30 μm thick were obtained using a freezing microtome (Microm HM 450,Microm International GmbH,Dreieich,Germany).
Flotation immunohistochemistry was performed.Sections were rinsed with PBS three times (10 min each) and then preincubated for 1 h at room temperature in 5% normal donkey serum (SouthernBiotech) and 0.2% Triton X-100 (Sigma-Aldrich) diluted in PBS (blocking solution).Immunohistochemistry was performed by incubating the sections overnight at 4 • C with the primary mouse monoclonal GAD 67 (Merck) (1:1000 in PBS containing 5% normal donkey serum and 0.2% Triton X-100).The following day, sections were washed with PBS three times (10 min each) and incubated for 2 h at room temperature with the donkey anti mouse secondary antibody Alexa 488 (ThermoFisher) (1:1000 in PBS containing 5% normal donkey serum and 0.2% Triton X-100).Slices were washed three times (10 min each) with PBS and then incubated with Hoecht (1:1000) for 10 min.Finally, slices were washed three times with PBS (10 min each).
Slices were then mounted on Superfrost slides with Fluoromount-G medium (SouthernBiotech).

Image acquisition and analysis
One coronal S1 slice containing HL and FL cortices (Bregma: − 0.1 to − 1.2) was analyzed from each animal (sham n = 7, full transected n = 11, contused n = 3).Images were acquired using epifluorescence microscopy (Olympus IX83) with a 10× objective controlled by CellSens Dimension software to obtain mosaics of images.Scan images from each channel were collected with a resolution of 0.6467 μm/pixel.Green channel containing GAD 67 signal was then visually inspected to identify and to delimit HL and FL cortices using a Brain Atlas (Paxinos and Watson, 2007).Note that the termination of the motor cortex and the beginning of HL cortex can be easily identified by the emergence of layer 4 in the DAPI channel.Then, we drew a straight line (209 μm width) in each cortical location from the pia surface to the end of layer 6 and plotted the profile of the GAD 67 intensity using the Plot Profile tool in ImageJ.Intensity profiles were normalized to the cortical depth (0-100%) and fluorescence variability (baseline subtracted) averaged using IgorPro (WaveMetrics Inc., OR, USA).Cortical layers were defined as in Fiáth et al., 2016, Narayanan et al., 2017and Zaforas et al., 2021 and manually verified using the DAPI channel in some control experiments.

Statistical analysis
All the statistical analyses were performed using Statistica.Ink (Statsoft Ibérica, Lisboa, Portugal).All values reported are mean ± SEM (main text tables and supplemental tables 1-7. The response amplitude of SEPs obtained from every temporal point were normalized to the averaged magnitude of pre-lesion recordings of each experimental group.SEPs evoked in each cortical location/hemisphere and experimental group were analyzed using repeated measures analyses of variance (ANOVA) with "time" (pre-lesion, 1st week, 2nd week, 3rd week, 4th week) as within-subject variable and the betweensubjects factor "group" (sham, full transected and contused).Note that for FLCx locations, only full transected SCI was performed and the between-subjects factor (group) has only two levels (sham and full transected).Next, we determine a threshold value as reference to identify when the SEP obtained from each cortical location in SCI groups should be considered as increased (underwent cortical reorganization, CoRe) or not increased (unaffected).To this end, we used the confidence interval (mean ± 1.96*SD) from the ratio values (4th week/pre-lesion week) obtained from all cortical locations in the sham group.The obtained threshold was further used to identify the cortical locations within full transected SCI and contused SCI groups that underwent CoRe (ratio values > threshold) from those non-affected by a SCI (ratio values ≤ threshold).A repeated measures ANOVA was performed for each type of SCI separately (full transected, contused) to evaluate SEPs across time (same within-subject variable) but using as between-subjects factor SCI effect over cortical locations instead of type of injury (sham, unaffected and CoRe).
Pre-lesion TUPs were compared using a one-way ANOVA with "group" (sham, full transected, contused) as a between effect.For the TUPs comparison along time, a repeated measures ANOVA was performed, with "time" (pre-lesion, 1st week, 2nd week, 3rd week, 4th week) as within-subject variable and the between-subjects factor "group" (sham, full transected and contused).
GAD 67 fluorescence obtained from each cortical layer was compared E. Alonso-Calviño et al.
using a one-way ANOVA with "group" (sham, unaffected and CoRe) as a between effect.A Tukey Honest Significant Difference Test was applied for all post hoc comparisons.All results were considered significant at P < 0.05.Graphs and figures were made using IgorPro and Adobe Illustrator.

Results
In this study, we used bilateral electroencephalogram (EEG) recordings from the hindlimb and forelimb somatosensory cortex (HLCx and FLCx, respectively) to measure sensory-evoked potentials (SEP) in response to peripheral electrical stimulation (Fig. 1B) in control and following a SCI.Under control conditions, contralateral forelimb stimulation (5 V) induced a strong SEP in the FLCx when compared to contralateral hindlimb stimulation (Fig. 1C, top).Similar response pattern was observed from SEPs recorded from HLCx in response to stimulation of the contralateral hindlimb compared to forelimb (Fig. 1C, bottom).These features of SEP responses were used to confirm the correct somatotopic location of the cranial screws at the HLCx or FLCx coordinates.

Full-transection and contusive SCI increase sensory-evoked responses in the deprived somatosensory cortex
We first determined the cortical neurophysiological changes at the deprived hindlimb somatosensory cortex (HLCx) produced by two models of thoracic (T9) SCI: a full transection (Trx-SCI) that completely interrupts all the ascending sensory inputs, and a contusive injury (Cntx-SCI) that damages most of the dorsal ascending spinal tracts (Fig. 2A).All results were compared to the sham group in which laminectomy was performed without injuring the spinal cord.After a first session of EEG recordings to measure SEPs in the HLCx under control conditions, rats were randomly assigned to sham, Trx-SCI or Cntx-SCI groups.Then, we used the same animals to study the temporal evolution of FL-HL cortical responsiveness as an estimate of CoRe.For that, we obtained weekly measurements of SEPs in response to low and high intensity forelimb stimulation for up to one month in HLCx hemispheres or locations (Fig. 2B-C).These responses were then normalized to the averaged magnitude of population recorded in pre-lesion conditions to reduce inter-subject variability.Comparisons obtained for the low intensity stimulation protocol showed that SCI induced increased SEP responses in a time and group-dependent manner (Fig. 2D, left).The repeated measures ANOVA revealed a significant effect of time (p < 0.001), group (p < 0.001), and the interaction between time and group (F (8,244) = 4.083, p < 0.001).Further post hoc Tukey analysis (see Table 1.1 and supplemental table 1) confirmed that SEP responses in the sham group were not affected over time.Curiously, a similar profile was observed for intra-group analysis of the Trx-SCI group that showed increased responses in the 1st week, while 2nd, 3rd and 4th weeks were no different from the pre-lesion week.On the other hand, Cntx-SCI group showed increased SEP responses during the 4 weeks after injury.This indicates that SEP magnitudes over time are differentially affected in each experimental model, Trx-SCI and Cntx-SCI.These differences were confirmed by the comparison between groups showing that sham and Cntx-SCI groups are different across weeks after SCI (Table 1.1).Finally, no differences were observed between sham and Trx-SCI group or between Trx-SCI and Cntx-SCI, except for the 1st week.
The same comparison performed for high intensity stimulation protocol (Fig. 2B,D right; repeated measures ANOVA) showed significant effects for time (p < 0.001), group (p = 0.019) and interaction between time and group (F (8,244) = 3.284, p = 0.001).The post hoc Tukey analysis confirmed that the sham group was not affected over time, while differences in the temporal profile were observed after intra-group analysis.Thus the Trx-SCI group showed increased SEPs at week 1 and 4 after SCI, but the Cntx-SCI group was only affected at week 1 and 2 after SCI (see Table 1.1 and supplemental table 1).Therefore, high intensity stimulation showed that SEP magnitudes over time were differentially affected in each experimental model.Importantly, both Trx-SCI and Cntx-SCI differed from the sham group at week 4 after SCI.
Together, our results consistently showed that SCI (full transection and contusion) increases SEP in the sensory deprived HLCx in response to forelimb stimulation.However, the longitudinal study of both SCI groups showed that cortical changes follow an irregular profile regarding the time elapsed after injury until stabilization (4th week), which is consistent with data from Endo et al. (2007) and it could be related to individual differences in the temporal evolution of CoRe.

Cortical reorganization is a non-homogeneous process in the population with SCI
Previous findings from our group show that the acute effect of a SCI on the somatosensory cortical activity is not homogeneous among individuals, with some animals showing increased neuronal excitability while others being unaffected (Zaforas et al., 2021).The same results are described in Endo et al. (2007Endo et al. ( , 2009)), showing that SCI-induced cortical changes are nonlinear, as cluster volume obtained using fMRI was decreased at 1-2 weeks after first increase to finally reach a stable increase last week.We have observed a similar pattern of SEP responses over time in both Trx-SCI and Cntx-SCI groups that could be due to a mixed population of affected and unaffected HLCx cortical locations.To test this assumption, it was necessary to determine a threshold value as reference to identify when the SEP obtained from each cortical location in SCI groups should be considered as increased (CoRe) or not increased (unaffected).To this end, we used the confidence interval (mean ± 1.96*SD) from the ratio values (4th week/pre-lesion week) obtained from all cortical locations in the sham group.Once identified the subset of cortical locations with a SEP that could be considered as CoRe (above the threshold) in both experimental groups (Trx-SCI and Cntx-SCI), a new statistical comparison to sham  group was made to confirm the differences between the two subsets of cortical locations considered CoRe and unaffected.The result of statistical analysis for the Trx-SCI group showed differences for both low intensity stimulation protocol (repeated measures ANOVA TIMExGROUP F (8,168) = 7.308, p < 0.001; Table 3.1) and for high intensity stimulation protocol (repeated measures ANOVA TIMExGROUP F (8,168) = 8.692, p < 0.001).The intra-group post hoc Tukey analysis showed that only in the subset CoRe of Trx-SCI group the SEP values over weeks were different from pre-lesion value for both stimulation intensities, while no differences were observed in the sham and unaffected Trx-SCI groups (see Table 3.1).Moreover, post hoc Tukey analysis showed that the subset of CoRe in the Trx-SCI group was different from the sham and unaffected group at all time points for low intensity and the 1st and 4th weeks for high intensity (see Fig. 3C and Table 3.1).
Regarding the analysis of the two subsets identified in the Cntx-SCI (the CoRe and unaffected), statistical comparison to sham group was significant for both low intensity stimulation protocol (repeated   measures ANOVA TIMExGROUP F (8,120) = 8.455, p < 0.001; Table 3.2), and for high intensity stimulation protocol (repeated measures ANOV-A TIMExGROUP F (8,120) = 5.579, p < 0.001).The intra-group post hoc Tukey analysis showed that unaffected subset of Cntx-SCI group increased responses only during 1st and 2nd weeks to finally recover at 3rd and 4th weeks (Fig. 3.C; Table 3.2).Moreover, the post hoc Tukey analysis intragroup for CoRe subset from Cntx-SCI showed differences across weeks with only exception of 3rd week (Fig. 3C; Table 3.2).Interestingly, the post hoc Tukey analysis for between groups comparison showed that the CoRe subset was different from the sham group during the four weeks after SCI at low intensity, and only 1st, 2nd and 4th weeks for high intensity stimulation protocol (Fig. 3C; Table 3.2).On the other hand, it was confirmed that the unaffected subset of Cntx-SCI and the sham group were not different (Fig. 3C; Table 3.2).
Considering animals from both experimental models in which HLCx locations were recorded from both hemispheres at both stimulation intensities (n = 22), we found that animals with Trx-SCI (n = 13) were more bilaterally affected (9 out of 13, 69.2% of animals), while animals with Ctx-SCI (n = 9) showed a more lateralized (unilaterally affected) reorganization (4 out of 9, 44.4% of animals) (Fig. 3B).It is possible that the Ctx-SCI exhibits a more lateralized effect due to the non homogeneous damage of the spinal cord, which is not the case for the Trx-SCI animals (2 out of 13, 15.3%).Finally, the number of animals showing both unaffected hemispheres was similar for full transected and contusion (2 and 1 animals, respectively).
Taken together, these results provide a broad perspective of the complexity of changes in the somatosensory cortex after SCI.More in detail, our results show that a SCI can 1) increase SEP in both deafferented hemispheres, 2) only in one hemisphere, or 3) induce limited changes in cortical excitability in both deafferented hemispheres.The three possibilities were identified in both models of SCI with different percentages of lateralization.

Chronic SCI alters somatosensory responses in the intact somatosensory cortex
For a complete perspective of CoRe after SCI, we next studied the functional changes in the non-deprived forepaw cortices (FLCx) in a subset of sham and Trx-SCI animals (n = 11) in which cranial screws were implanted in the cortical coordinates of FLCx.As shown in Fig. 1C, electrical stimulation of forelimbs or hindlimbs in control conditions (pre-lesion) induced a marked SEP in the FLCx coordinate, showing higher amplitude and smaller latency when stimulation was applied at the forelimb.After the SCI, the experimental protocol applied was exactly the same as for the study of HLCx CoRe.However, it is worth noting that a SCI at thoracic level does not directly affect the pathway from forelimb peripheral afferents to FLCx (Fig. 4A).
The analysis of recordings in the population of FLCx hemispheres, from both sham vs Trx-SCI groups, shows that cortical responses in the intact FLCx were unaffected after SCI for both low and high stimulation intensity (Fig. 4C).A repeated measure ANOVA for sham vs Trx-SCI in response to low stimulation intensity showed no effect for time (p = 0.644), nor group (p = 0.094) or time x group interaction (F (4,140) = 2.035, p = 0.093).In the same way, comparison between groups for high intensity protocol showed no effect for time (p = 0.508), nor group (p = 0.133) or time x group interaction (F (4,140) = 2.202, p = 0.072) (Table 1.2).Next, we used the same method to set a threshold based on the confidence interval (mean ± 1.96 SD) of sham SEPs values Table 3.2 Repeated measures ANOVA for comparisons in S1HL location of contused group vs sham.
(SEP-4th week/SEP-pre-lesion).The threshold values were 1.56 and 1.28 for low intensity and high intensity protocols, respectively.The result obtained for low intensity protocol was that 12 out of 22 (54.5%)showed values above the threshold, while for high intensity stimulation protocol 13 out of 22 (59.1%)showed values above the threshold value (Table 2.1, Fig. 4B, left).Therefore, a new classification for changes in FLCx hemispheres was made: 1) hemispheres for FLCx showing cortical reorganization after SCI (Full transected CoRe); 2) hemispheres for FLCx which showed no changes in cortical responses after SCI (Full transected unaffected).As a final step to validate this classification, statistical comparisons were made between sham group, full transected unaffected and full transected CoRe for low intensity stimulation (repeated measures ANOVA TIMExGROUP (F (8,136) = 5.504, p = 0.000).Post hoc Tukey analysis (Table 3.3) confirmed that no changes were observed in sham and full transected unaffected groups, while FLCx hemispheres in the full transected CoRe group increased SEP at 1st, 3 er and 4th weeks after SCI (Fig. 4D).The same comparison was performed for high intensity protocol (repeated measures ANOVA TIMExGROUP (F (8,136) = 2.609, p = 0.011).In this case, the post hoc Tukey analysis showed that differences were only observed at 1st and 4th week after injury in the SCI-CoRe group against its pre-lesion week, and against 4th week of sham group.
Considering the total number of Trx-SCI animals recorded from both hemispheres in FLCx location at two stimulation intensities (n = 11), a bilateral effect was predominant (72.7%, 8 out of 11 animals; Fig. 4B, right).Similar to results obtained for HLCx, we found that 3 out of 11 animals (27.3%) were unaffected, and interestingly with absence of unilateral effects.These results open new perspectives to decipher complexity of cortical changes after SCI by extending the effects to the non-deafferented somatosensory cortex, which show similar effects to those obtained for HLCx.

Animals with chronic SCI show increased stimulus-triggered Upstates probability
During spontaneous slow-wave activity, the activity of cortical neuronal population is characterized by the alternation of transient active states (up-states) and silent periods (or down-states) with dominant frequency of <1 Hz (Steriade et al., 1993;Crunelli et al., 2015).Moreover, up-states can be triggered (TUP, triggered up-states) by peripheral and thalamic stimulation with different probability depending on the stimulation intensity and the level of cortical excitability (Fig. 5A) (MacLean et al., 2005;Rigas and Castro-Alamancos, 2007;Civillico and Contreras, 2012;Bermudez-Contreras et al., 2022).In this line, low intensity stimulation has been described to produce a low rate of TUPs, therefore, increased rate of the TUPs would indicate an increased cortical excitability and could be used to define temporal changes in cortical excitability after SCI.Here, we determined the percentage of TUPs induced by low-intensity stimulation for the three experimental groups (sham, Trx-SCI, Cntx-SCI).Data obtained during pre-lesion week showed that all three groups have similar percentages of TUPs (one-way ANOVA F (2.67) = 0.430, p = 0.652).The weekly average for the percentage of TUPs in the sham group showed similar values over time: 22.5% pre-lesion, 22.3% (1 st W), 18.6% (2 nd W), 18.2% (3 rd W), and 16.6% (4 th W).On the other hand, Trx-SCI group increased the percentage of TUPs from 25.1% pre-lesion to 66.7% (1 st W), 53.9% (2 nd W), 54.6% (3 rd W), and 56.5% (4 th W).In the same way, the contused-SCI group showed a 19.9% pre-lesion, and the percentage of TUPs  ).The statistical comparison between them confirmed a clear difference between sham group and both Trx-SCI and Cntx-SCI (Fig. 5B, repeated measures ANOVA TIMExGROUP F (8,268) = 5.035, p = 0.000).
While sham maintained a similar percentage of TUPs over time, Trx-SCI and Cntx-SCI significantly increased for each time point after SCI (Table 4, Fig. 5B).Importantly, no differences were found between Trx-SCI and Cntx-SCI (see Table 4).Thus, our results confirmed that TUPs increased in the same manner in different types of SCI (transection and contusion) from the first week after lesion and remained increased for at least 4 weeks after SCI.Based on these results, we propose that TUPs as a new physiological feature that indicate cortical increased excitability (hyperexcitability) after SCI.

Decreased levels of GABAergic staining in animals with increased responses after SCI
Our results show increased probability of TUPs in the deafferented HLCx, which indicate a process of local cortical hyperexcitability that could be linked to reduced GABAergic activity and unmasking of weak synapses (Jacobs and Donoghue, 1991).At the same time, our results show that individuals from both populations with SCI, full-transect and contusion, likely develop 80% CoRe (increased SEP) while a 20% will be unaffected.However, it is not clear if GABA levels are contributing to observed changes in the SEP.Therefore, we quantified the presence of GAD 67 , an enzyme that metabolizes glutamate into GABA in neuronal soma and terminals, as a hallmark of changes in the inhibitory tone in  3 and suppl table 3 and 4).Values are expressed in means ± SEM both the intact SSCx (FLCx) and sensory-deprived SSCx (HLCx) after chronic SCI.We performed this experiment from brain tissue obtained from the same animals used in the electrophysiological dataset, including CoRe, unaffected SCI group and sham group (Fig. 6A-B).GAD 67 quantification was performed by measuring the immunofluorescence intensity along the dorsoventral axis (i.e.cortical depth) of the somatosensory cortex from 21 animals (sham n = 7 and SCI n = 14: 11 full-transected and contusion n = 3) (Fig. 6C).Analysis of the fluorescence area in each layer of the deprived HLCx showed decreased GAD 67 content in layer 4 of CoRe SCI group compared to unaffected group, and in the layers 5 and layer 6 of CoRe SCI compared to sham animals (ANOVA GROUP F (2,31) = 4.345, 4.036 and 4.104 and p = 0.022, 0.028 and 0.026 for layers 4, 5 and 6, respectively; Fig. 6C, right: Table 5).On the other hand, non-deprived, intact FLCx did not show differences in the GAD67 cortical profile neither in the CoRe group nor in the unaffected SCI group when compared to sham (Fig. 6C, left).Fig. 5. Triggered up-state (TUP) are increased in animals with SCI.On the left, a schematic illustration of the experimental protocol recording in the hindlimb cortex HLCx.On the right, two examples on cortical responses in HLCx coordinates to low intensity stimulation in FL: in upper trace, a cortical response with two components is shown: short latency response (< 50 ms) which is the SEP (red circle), and long-latency activation which is a triggered up-state (TUP) (50 ms < x < 200 ms) (black arrow); in lower trace, cortical response showing only the short latency response corresponding to the SEP (red circle).B) Temporal profile of populational averaged percentages of triggered up-states (TUPs) produced in HLCx from sham group (black), the full-transected SCI group (red), and in the contused-SCI group (blue), in response to FL stimulation at low intensity.The sham group was unaffected over the weeks, while full-transected SCI and contused-SCI groups showed a statistic significant increase in the percentage of TUPs every week after injury.Asterisks indicate statistic differences (see Table 4 and supp table 5).Values are expressed in means ± SEM

Table 4
Repeated measures ANOVA for comparisons between groups for percentage of TUPs.Therefore, these results are very consistent with the possibility of reduced GABAergic activity in CoRe after SCI.Moreover, this plastic process occurs specifically in the deprived HLCx of SCI-CoRe group of animals allowing activity from FLCx to expand towards the deprived area.Altogether, data show a decreased intensity GAD 67 in granular and infragranular layers of the sensory-deprived cortex, which could be related to increased evoked responses.

Discussion
In this study, we used two different models of thoracic T9-T10 SCI (i.e., full transection and contusion) to longitudinally determine the changes in excitability at the somatosensory cortex related to the phenomenon of cortical reorganization.By using a stimulus-response paradigm to produce cortical responses as an estimate of CoRe, we show that a SCI triggers functional changes at cortical level that consist in an increased excitability of the deafferented (hindlimb region) and intact cortex (forelimb region) to peripheral stimulation of body regions above the lesion level (forelimb).A threshold-based method obtained from the sham population, allowed us to determine which cortical locations in the SCI population showed increased SEP across weeks in two different models of SCI.However, different temporal profile and lateralization of increased SEP were detected between both experimental models (full transection and contusion).Changes in cortical locations were used to determine with higher precision individuals in which at least one cortical location was affected after SCI, therefore those individuals were considered as suffering cortical reorganization.We found that the percentage of subjects with CoRe was close to 80% of the population and interestingly 20% of subjects were considered unaffected.
In addition, our results also demonstrated an increased percentage of triggered up-states associated with peripheral stimulation in all injured animals, revealing that the somatosensory cortex develops hyperexcitability in a chronic SCI condition.Finally, histological results showed a reduction of immunofluorescence intensity for GAD 67 staining in the deafferented sensory cortex only in animals with CoRe, while no changes were observed in animals without CoRe.Taken together our results show that cortical reorganization after a SCI does not depend on the type of SCI, but how cortical circuits react to a permanent loss of sensory inputs.Moreover, our results point to an increased cortical excitability as the main feature of CoRe after SCI, which could be explained by a lower GABAergic tone.

Somatosensory cortex is differently affected by two models of spinal cord injury
The phenomenon of CoRe after SCI has been described in humans and animal models based on increased magnitude of cortical activity in the deafferented cortex in response to peripheral stimulation of body regions above the lesion level.Therefore, it is considered as a functional expansion of the cortical activity from intact towards the deafferented cortex (Endo et al., 2007;Wrigley et al., 2009;Endo et al., 2009;Jain et al., 2008;Jutzeler et al., 2019).Based on the neuroanatomy of ascending sensory pathways, different types of spinal cord injuries (i.e.full transection, hemisection or contusion) should produce different sensory deafferentation of the somatosensory cortex thereby inducing different level/strength of CoRe.However, patients with different types of SCI exhibit similar neurological evolution during the first 6 months after injury (Scivoletto et al., 2020), indicating that higher-order brain areas sense the distinct types of sensory deprivation as an alike process.In this study, our first analysis taking the entire population showed that both full transection and contusive SCI increased cortical excitability when compared to sham animals, although this effect was not different among SCI models.(Siddall et al., 2003;Finnerup et al., 2003Finnerup et al., , 2007;;Moxon et al., 2014;Vipin et al., 2016;Jutzeler et al., 2019;All and Al-Nashash, 2021).
Intra group comparisons also showed variations in the increased SEP responses over time in both experimental models, indicating variability in the temporal evolution of CoRe.This variability was intensity dependent with contusive SCI showing changes in response to low intensity stimuli and full transection SCI to high intensity stimuli.These non-homogeneous intra-group results drove us to consider that it could be produced by the high intra-group variability of changes across cortical locations recorded.Therefore, we used a statistical threshold based on the longitudinal response variability of the sham group to classify the cortical locations as CoRE (SEP magnitudes above threshold values), or unaffected (SEP magnitudes equal or smaller than threshold).
In this regard we found that increased cortical responses at the hindlimb cortex after both contusive and full transection SCI were only detected in a subset of hemispheres/individuals as previously reported by our group (Zaforas et al., 2021) and other authors (Vipin et al., 2016).In both SCI models, cortical locations undergoing CoRe increased their SEP magnitude in the first week and remained highly excitable for up to one month (Figs. 3 and 4).A different time course was observed in the unaffected cortical locations of contusion-SCI in which SEP in response to low intensity increased only at the first two weeks and then their response magnitude returned to pre-lesion levels, while no differences in

Table 5
Data analysis of GAD67 fluorescence in HLCx and FLCx. the SEP magnitude were found in response to high intensity.The fact that the first week after SCI shows a consistent increased SEP for most of the studied groups indicates a temporal point in which cortical circuits are strongly overexcitable and any incoming sensory input can initiate a synchronized neuronal activation of corticocortical connections.After this first week, distinct cellular and molecular processes may take place to dictate the evolution of CoRe in both groups (Endo et al., 2007).Moreover, the finding that SEP was increased in the contused unaffected group only at the first weeks may indicate different levels of damage in the spinal cord leading to differences in cortical plasticity (Nandakumar et al., 2023).The more plausible origin of these differences may lay on the inflammatory response in the acute phase of the injury (2-13 days) that is enhanced in the contusive model of SCI compared to a full transection model (Popovic et al., 1997), and which directly affect the strength of the ascending sensory inputs.We found that the different pattern of cortical changes observed for full transection and contusion could be of special interest in order to understand spinal cord damage influences on cortical activity.CoRe after SCI is driven not only by changes in excitability of the deprived cortex but also in the intact, adjacent cortex; this has been shown in animal and human models (Curt et al., 2002;Bazley et al., 2014).By using bilateral forelimb recordings in a set of animals with full thoracic transection of spinal cord, we observed that SEPs were not affected by SCI, corroborating previous work from All and Al-Nashash (2021) using a similar experimental approach (chronic cranial screws for recordings) during 3 weeks after thoracic SCI.Nonetheless, motivated by our analysis in HLCx showing distinct groups of CoRe, we used the same thresholding method based on the variability of SEP magnitudes obtained from FLCx of the sham group.Remarkably, after applying the threshold process the results were highly alike to those obtained for SEPs in HLCx, which allowed us to identify FLCx hemispheres with values above threshold, therefore considered as a full transected CoRe group.On the contrary, FLCx hemispheres showing values below threshold for both intensities were not different from the sham, therefore considered as unaffected full transected group.Therefore, similar patterns of CoRe are observed in both the intact and the deafferented somatosensory cortex.

Cortical reorganization takes place in individuals
The first step of our work consisted in the description of how different locations in the somatosensory cortex from the same individual could be affected by a SCI.This approach allowed us to optimize the identification of animals that developed CoRe after SCI in different SCI models (full transection vs contusion), which is one of the main objectives of this work.
The classification of animals undergoing CoRe was made if at least one cortical location showed increased SEP after SCI.Nonetheless recordings were obtained from four cortical locations in sham and full transected and two cortical locations in contused (as described in methods).Therefore, we wonder if all cortical locations from the same animal were equally affected by SCI, or if different models of SCI could induce different patterns of cortical changes.To answer this, we studied the SEP obtained in all cortical locations of each animal in response to the application of two different intensities of peripheral electrical stimulation.The use of two different intensities was intended to uncover if different SCI models could affect in different manner to sensory modalities.In this way, the low intensity stimulation mainly activates the low threshold peripheral fibers (tactile and proprioceptive modalities), while high intensity stimulation includes activation of all types of peripheral fibers (tactile, proprioceptive, thermal and nociceptive).Our results show that in the full transected group a total of 16 out of 20 (80%) animals showed at least one cortical location with increased SEP, while 5 out of 20 (20%) animals did notn't show cortical locations affected by SCI.On the other hand, the contused-SCI group showed that 9 out of 10 (90%) animals showed at least one cortical location with increased magnitude of SEP after SCI, while only 1 animal was not affected.Therefore, the percentage of animals showing increased SEP after SCI is similar between both models of SCI; however, the time course of its effects are slightly different (see Fig. 3.1, 3.2).Taking together, from both populations of animals with SCI (full-transect and contused), a total of 25 out of 30 subjects (83%) showed increased SEP after SCI at least in one cortical location.It is remarkable that 5 out of 30 (17%) animals didn't show increased SEP in any of the cortical locations recorded, which confirm the robustness of our method to identify CoRe in animals with SCI.These data indicate that increased SEP is developed consistently in a population of animals with different types of SCI.

Cortical hyperexcitability: a new physiological feature of CoRE
Evoked cortical responses exhibit two components (Civillico and Contreras, 2012;Bermudez-Contreras et al., 2022).The first/early component (< 50 ms) known as the SEP is associated with direct thalamic inputs, and it is always present in cortical evoked responses.While the second/late component (identified between 50 ms and 200 ms) known as triggered up-state is originated by vertical activity in the cortical column and cortico-cortical horizontal propagation (Wester and Contreras, 2012;Bermudez-Contreras et al., 2022) and depends on the stimulation properties (mainly intensity and frequency) and brain states (Rigas andCastro-Alamancos, 2007, 2009;Civillico and Contreras, 2012;Favero and Castro-Alamancos, 2013).
Our data show that animals with SCI exhibit an increased percentage of TUPs in response to stimulation at low intensity, which can be induced by by reduced inhibitory tone (Sanchez-Vives et al., 2010;Barbero-Castillo et al., 2021;Castro-Alamancos, 2000;Pagès et al., 2021) that in turn increases columnar network connections or increased the strength of cortico-cortical connections from adjacent areas (Endo et al., 2007;Ghosh et al., 2009Ghosh et al., , 2010;;Sydekum et al., 2014;Humanes-Valera et al., 2017).Another alternative is increased TUPs is partially contributed by an increased strength in thalamocortical connectivity after SCI (Crunelli et al., 2015;Alonso-Calviño et al., 2016;Pagès et al., 2021) or also by changes in cortical excitability showing longer up-states during spontaneous activity as described in animals with chronic SCI (Fernández-López et al., 2019;Tran et al., 2004;Jensen et al., 2013).
We want to remark that the increased percentage of TUPs was observed even in animals classified as "unaffected" or non-affected by SCI.Therefore, a higher probability to produce TUPs could be considered as a contribution in the complexity of physiological components underlying CoRE after SCI.Nowadays, it is well accepted that functional invasion underlying the process of CoRe does not involve a change in the somatotopic map (Kikkert et al., 2021) but changes in synaptic properties of cortico-cortical connectivity (Henderson et al., 2011;Ghosh et al., 2012;Zhang et al., 2015).The increased presence of triggered upstates can cause the increased strength of cortico-cortical connectivity, including synaptic plasticity (Ghosh et al., 2012;Humanes-Valera et al., 2017).At the same time, this phenomenon can participate in increased physiological signals as BOLD in fMRI observed in animal models with SCI (Endo et al., 2007;Ghosh et al., 2009Ghosh et al., , 2010)).Taking together our results, we consider the increase of TUPs as a physiological component of CoRe, which indicates increased local cortical excitability (hyperexcitability) that contribute to a better understanding of the physiology of CoRe.

Reduced GABA in the deafferented cortex of animals with SCI
The phenomenon of CoRe has been linked to altered local cortical network of deafferented cortex, as changes in dendritic spines distribution (Ghosh et al., 2010(Ghosh et al., , 2012;;Zhang et al., 2015), as well as to a decrease in GABAergic activity after central and peripheral deafferentation (Garraghty et al., 1991(Garraghty et al., , 2006;;Levy et al., 2002;Keck et al., 2011).Our results show that animals with SCI undergoing CoRe have lower GABAergic labeling in the deafferented cortex (HLCx) compared to sham and unaffected animals.The reduced GABAergic labeling was evident in the granular and infragranular layers, where most of thalamic and cortico-cortical inputs take place, indicating lower inhibitory tone that could explain the increased cortical excitability of animals with SCI and CoRe.The first component of SEPs is produced by synaptic inputs from adjacent cortical regions or thalamocortical inputs, and the triggered up-state is produced by synaptic properties and intrinsic excitability of cortical neuronal network.It has been described that both components are directly and strongly regulated by GABAergic activity (Shu et al., 2003;Wilent and Contreras, 2004;Rigas and Castro-Alamancos, 2007;Mann et al., 2009) and a decreased GABAergic activity could unmask thalamic and cortico-cortical excitatory inputs mediating an increased SEP magnitude as well as the increased probability of TUPs (Rigas and Castro-Alamancos, 2007;Sanchez-Vives et al., 2010;Pagès et al., 2021) and the generation of spontaneous up-states (Mann et al., 2009).
Under natural conditions, higher GABA levels in the somatosensory cortex are associated with enhanced tactile discrimination function (Kolasinski et al., 2017), hence losing fine grade of tactile definition could produce a functional reorganization by increased efficiency of weak adjacent cortical inputs.This idea could explain why the cortical somatotopic organization is preserved in different types of deafferentation but shows increased responses to peripheral stimulation of intact body regions (Garraghty et al., 2006;Kolasinski et al., 2017).However, data obtained from the triggered up-state showed increased probability in almost the entire population of animals with SCI (fulltransection and contusion) including animals classified as unaffected.In this sense, up-states are strongly modulated by cortical inhibitory activity (Shu et al., 2003;Mann et al., 2009), but also by excitatory synaptic recurrent activity (Lemieux et al., 2014).Therefore, although we cannot exclude that the reduction of GABA tone induces an increase in triggered up-states, other cellular and synaptic factors may contribute to increased triggered up-states in the deafferented cortex in our results.In this sense, we have previously observed that triggered up-states propagate from FLCx to HLCx with higher probability in animals with chronic SCI (Humanes-Valera et al., 2017).In addition, different properties of up-states during spontaneous activity are modified in animals with chronic SCI, which can be explained by homeostatic plasticity to recover basal cortical excitability (Fernández-López et al., 2019).Both, increased probability of triggered up-states and changes of spontaneous up-states in animals with SCI indicates changes in the intrinsic excitability of local neuronal network of deafferented cortex due to changes in excitation/inhibition balance, direct thalamic inputs, strength of cortico-cortical connectivity, and level of different neuromodulators (Castro-Alamancos, 2004;Rigas and Castro-Alamancos, 2007).
Interestingly, the FLCx showed no decreased GAD 67 marker, but some locations of FLCx have increased responses and can be the origin of triggered up-states.Therefore, our results indicate that GABA levels in the deafferented cortex could be an important factor to produce local hyperexcitability as part of CoRe phenomenon.But, more complex processes including subcortical changes (Jain et al., 2000) should be involved in CoRe of deafferented and intact cortical regions.

A translational perspective
In this work, we have found that a subset of animals showed consistent increased SEPs in both cortical hemispheres in the full transection model, which can be considered a bilateral development of CoRe.However, in the contusion model there is a predominance of effects over one hemisphere, which can be considered unilateral development of CoRe.This could be a predictive factor to translate to human studies with different types of SCI.Interestingly, based on our results, increased SEP that could be considered as indicative of CoRe can be absent in animals with a SCI.Therefore, the wide range of effects depending on the type of SCI, including bilateral, unilateral and unaffected hemispheres, indicate that CoRe is not a homogeneous phenomenon in our studied population of cortical locations, and more precisely among individuals.
Interestingly, a high level of variability has been described in the study of CoRe in patients with SCI.In this sense, CoRe has been usually linked to development of different associated pathologies referred to as phantom limb sensation and neuropathic pain (Moore et al., 2000;Wrigley et al., 2009;Jutzeler et al., 2015;2016).A detailed view of these works reveals that symmetrical (more bilateral) and asymmetrical (more unilateral) effects have been described in patients with similar SCI severity (ASIA classification) and same anatomical level (Siddall et al., 2003;Wrigley et al., 2009;Henderson et al., 2011;Jutzeler et al., 2015;2016).At the same time, it has been described that no cortical changes take place in some patients with SCI (Freund et al., 2011), as we have found in a subset of animals.
The research works in CoRe in patients with SCI is mainly intended to obtain a more precise way to predict the evolution and prognosis of each patient to apply therapies in order to promote the best sensorimotor functional recovery.In this context, the correct interpretation of a wide range of effects observed in a population is key to make correlations and predictions.However, the research work in animal models is mainly intended to obtain a better understanding of the mechanisms behind the origin of SCI-associated phenomena/pathologies such as CoRe, neuropathic pain, etc.While it has been assumed that CoRe is a general effect over a population of individuals with similar or different SCI, our present results propose the importance to properly identify cortical locations of individuals in which the phenomenon of CoRe takes place and in which individuals not.Posterior steps should focus on the identification of cellular and molecular mechanisms behind animals showing CoRe and in animals without CoRE.Finally, considering that a full transection produces the same deafferentation in both hemispheres but in some individuals only one hemisphere shows physiological changes, then, CoRe must be strongly dictated by cortical circuits.
In conclusion, our methodological approach using multisite recordings in animals with chronic SCI provides a broad perspective about the process of cortical reorganization after SCI, including a better spatial definition and temporal evolution of the phenomenon.Moreover, our results show that increased SEP, as part of CoRe development, depends on the type of SCI (full transection vs. severe contusion).Considering the temporal pattern of increased SEP across weeks, this can help to identify temporal windows in which changes could take place in order to develop CoRe or be unaffected by SCI according to SEPs.In the same manner this approach could be indicative of CoRe evolution in humans with SCI.Finally, decreased GABAergic presence across cortical layers could contribute to the increased excitability related to CoRe after SCI.It this point could be confirmed in humans, it will help to find new therapies for collaterals pathologies to SCI.PID2021-126609NA-I00, co-funded by "ERDF A way of making Europe", and Grant RYC2019-026870-I, co-funded "ESF Investing in your future").

Declaration of Competing Interest
None.

Fig. 1 .
Fig. 1.Schematic experimental model and representation of somatosensory evoked potentials (SEPs) obtained in forelimb and hindlimb cortices.A) Schematic representation of the experimental approach: the animal with thoracic SCI, the inset shows screws electrodes placed on cranial coordinates of the representation of forelimb (FLCx) and hindlimb (HLCx) in the primary somatosensory cortex, red arrowheads indicate peripheral electrical stimulation placed bilaterally in the hindlimbs (HL stim) and forelimbs (FL stim).B) Electrophysiological recordings obtained simultaneously from FLCx (black) and HLCx (gray): on the left, during spontaneous slow-wave activity (SWA); on the right, the somatosensory evoked potentials (SEP) recorded in FLCx and HLCx (red circles) in response to forelimb stimulation (note the higher magnitude specifically in FLCx).C) Examples of SEPs recorded from FLCx (upper traces) and HLCx (lower traces) in response to contralateral stimulation of FL (black traces) or HL (gray traces) at low intensity.In both cases, the higher response magnitude match the stimulation of corresponding somatotopic limb.

Fig. 2 .
Fig. 2. Individual and population data of SEPs recorded in hindlimb cortex from the three experimental groups.A) Schemes of the three experimental SCI models: sham lesion, full transected spinal cord, and contused spinal cord.All lesions performed at thoracic level (T9-T10).B) Upper panel: a single recording session (detailed in the grey rectangles): SWA= recordings of spontaneous activity during slow wave activity; SEP obtained in response to stimulation of HL and FL applied sequentially at low (LI) and high (HI)intensities.Lower panel: Timeline of recording sessions from cranial implant week and pre-lesion recording week to 4th weeks after lesion.C) Examples of recorded SEPs in HLCx in from a representative individual of each experimental group: sham lesion (upper traces/black), full transection SCI (middle traces/red) and contusion SCI (lower traces /blue) across weeks.D) Normalized population data from the three experimental groups across the recording sessions: low intensity stimulation (left plot), and high intensity stimulation (right plot).Asterisks indicate statistic differences (see table 1 and suppl table1).Values are expressed in means ± SEM.

Fig. 3 .
Fig. 3. Identification of CoRe in deafferented Hindlimb cortex after SCI.A) Schematic illustration of the experimental protocol recording in the hindlimb cortex HLCx.B) Pie charts showing the percentage of animals which were affected in one hemisphere (purple color) in both hemispheres (blue color) and unaffected in both hemispheres (gray color), for full-transection and contusion models.C) Temporal evolution profiles of SEPs recorded in HLCx coordinates from the population of hemispheres recorded from: sham group (black line) and full-transected SCI.Data from the full-transected SCI population were divided in two subsets obtained after thresholding process: unaffected hemispheres population (gray line) and hemispheres classified as CoRe from SCI population (red line).D) Temporal evolution profiles of SEPs recorded in HLCx coordinates from the population of hemispheres recorded from sham group (black line) and contused-SCI group.Data from the contused-SCI population were divided in two subsets obtained after thresholding process: unaffected hemispheres population (gray line) and hemispheres classified as CoRe from SCI population (red line).Asterisks indicate statistic differences (see Table2 and supp table 2).Values are expressed in means ± SEM

Fig. 4 .
Fig. 4. Identification of CoRe in intact Forelimb somatosensory cortex after SCI.A) Schematic illustration of the experimental protocol recording in the forelimb cortex FLCx.B) Pie chart showing the percentage of animals which were affected in one hemisphere FLCx location (purple color) in both hemispheres FLCx locations (blue color) and unaffected in both hemispheres (gray color).C) Temporal evolution profiles of SEPs averaged from the hemispheres population recorded in FLCx locations from sham group (black trace) and full transected-SCI group (red traces).D) Temporal evolution profiles of SEPs recorded in FLCx coordinates from the population of hemispheres recorded from: sham group (black line) and full-transected SCI, from which population were divided in two subsets obtained after thresholding process: unaffected hemispheres population (gray line) and hemispheres classified as CoRe from SCI population (red line).Asterisks indicate statistic differences (see Table3 and suppl table 3 and 4).Values are expressed in means ± SEM

Fig. 6 .
Fig. 6.Decreased GABA precursor immunostaining in animals with CoRe after SCI.A) At left, representative image of immunofluorescence for double labeling in the somatosensory cortex (SSCx): GAD67 (green) and DAPI (blue).Dotted lines localize the HLCx and FLCx regions.At right, the GAD67 laminar profile of fluorescence in a representative animal belonging to the CoRe SCI group for HLCx (gray) and FLCx (black).Both cortical locations have similar intensity across layers, showing a double peak of intensity at L4 lower limit and middle L5.B-C) Representative examples (animals) of the laminar profile of immunofluorescence for DAPI (at left in blue) and GAD67 in the three groups (sham, unaffected SCI and CoRe SCI) for HLCx (B) and FLCx (C) locations.Averaged laminar profiles (line graphs) in the middle showing a qualitative comparison of intensity for immunofluorescence of GAD67 in the two cortical locations HLCx (B) and FLCx (C).Profiles were obtained from an averaged signal of 7 animals with sham lesion (black, n = 13 locations for HLCx and 13 for FLCx), 10 animals showing increased SEP (CoRe, n = 14 locations for HLCx and 15 for FLCx) after SCI (red), 4 animals unaffected in SEP (unaffected, n = 7 locations for HLCx and 6 locations for FLCx) after SCI (dotted line black).At right, bar graphs showing quantification and analysis of GAD 67 fluorescence in each layer of cortical locations in HLCx (B) and FLCx (C).One-way ANOVA with group factor (3 levels: sham, unaffected and CoRe SCI) was performed.When ANOVA p < 0.05, post hoc Tukey test was applied (* p < 0.05).Only averaged signals from CoRe animals showed decreased intensity for GAD67 immunofluorescence (A.U.F.= arbitrary units of fluorescence).Related data and statistical details are in supplementary table 6-7 and Table 5, respectively.(For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.)

Table 1 .1
Repetead measures ANOVA for comparisons in S1HL location.Related to Figure 2.
Table 2 and supp table 2).Values are expressed in means ± SEM

Table 2 .2
Cortical reorganization (%) depending on CoRe threshold, intensity and type of lesion, related to cortical locations.

Table 3 .1
Repeated measures ANOVA for comparisons in S1HL location of full transected group vs sham.

Table 3 .3
Repeated measures ANOVA for comparisons in S1FL.

Table 5 .
1 One-way ANOVA of GAD 67 fluorescence in the HLCx locations for comparisons between groups (Sham vs. CoRe SCI vs. Unaffected SCI).

Table 5 .
2 One-way ANOVA of GAD 67 fluorescence in the FLCx locations for comparisons between groups (Sham vs. CoRe SCI vs. Unaffected SCI).