PAIN�

154 (2013) 2397–2406

w w w . e l s e v i e r . c o m / l o c a t e / p a i n

Prefrontal cortex and mediodorsal thalamus reduced connectivityis associated with spatial working memory impairment in ratswith inflammatory pain

0304-3959/$36.00 � 2013 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.http://dx.doi.org/10.1016/j.pain.2013.07.020

⇑ Corresponding author. Address: Faculdade de Medicina do Porto, Departamentode Biologia Experimental, Alameda Hernâni Monteiro, 4200-319 Porto, Portugal.Tel.: +351 225 513 600; fax: +351 225 513 601.

E-mail address: galhardo@med.up.pt (V. Galhardo).

Helder Cardoso-Cruz a,b, Mafalda Sousa a,b,c, Joana B. Vieira c,d, Deolinda Lima a,b, Vasco Galhardo a,b,⇑a IBMC – Instituto de Biologia Molecular e Celular, Grupo de Morfofisiologia do Sistema Somatosensitivo, Universidade do Porto, 4150–180 Porto, Portugalb Departamento de Biologia Experimental, Faculdade de Medicina, Universidade do Porto, 4200-319 Porto, Portugalc Programa Doutoral em Neurociências, Faculdade de Medicina, Universidade do Porto, 4200-319 Porto, Portugald Laboratório de Neuropsicofisiologia, Faculdade de Psicologia e Ciências da Educação, Universidade do Porto, 4200-392 Porto, Portugal

Sponsorships or competing interests that may be relevant to content are disclosed at the end of this article.

a r t i c l e i n f o a b s t r a c t

Article history:Received 28 March 2013Received in revised form 27 June 2013Accepted 15 July 2013

Keywords:Local field potentialsInflammatory painSpatial working memorymPFCMediodorsal thalamus

The medial prefrontal cortex (mPFC) and the mediodorsal thalamus (MD) form interconnected neural cir-cuits that are important for spatial cognition and memory, but it is not known whether the functionalconnectivity between these areas is affected by the onset of an animal model of inflammatory pain. Toaddress this issue, we implanted 2 multichannel arrays of electrodes in the mPFC and MD of adult ratsand recorded local field potential activity during a food-reinforced spatial working memory task. Record-ings were performed for 3 weeks, before and after the establishment of the pain model. Our results showthat inflammatory pain caused an impairment of spatial working memory performance that is associatedwith changes in the activity of the mPFC–MD circuit; an analysis of partial directed coherence betweenthe areas revealed a global decrease in the connectivity of the circuit. This decrease was observed over awide frequency range in both the frontothalamic and thalamofrontal directions of the circuit, but wasmore evident from MD to mPFC. In addition, spectral analysis revealed significant oscillations of poweracross frequency bands, namely with a strong theta component that oscillated after the onset of the pain-ful condition. Finally, our data revealed that chronic pain induces an increase in theta/gamma phasecoherence and a higher level of mPFC–MD coherence, which is partially conserved across frequencybands. The present results demonstrate that functional disturbances in mPFC–MD connectivity are a rel-evant cause of deficits in pain-related working memory.

� 2013 International Association for the Study of Pain. Published by Elsevier B.V. All rights reserved.

1. Introduction

The most prevalent among the cognitive problems that com-monly affect patients with chronic pain syndromes are complaintsof memory failures. These are already widely described in the sci-entific literature and are recognized today to constitute a problemconfined to the processes of short-term memory that require a sig-nificant degree of cognitive complexity rather than generalizedmemory loss [3,7,41,56]. This cognitive dysfunction in chronic painpatients is both a crucial and paradoxical problem, since the etiol-ogy of these disturbances is unknown.

The most common explanations for the onset of memory distur-bances are (1) the prolonged regimens of medication acting cen-trally at brain level [15] and (2) the possibility that sudden

episodes of pain disrupt memory acquisition, leading to momen-tary lapses of attention [18,20,28]. However, neither of thesehypotheses explains the specificity of disruption to short-termmemories compared with declarative retrospective memories, orthe fact that the majority of the cognitive dysfunctions observedin animal models of chronic pain occur in the absence of prolongedpharmacological habituation. We have previously proposed thatthe sensory overload typical of painful syndromes induces neuro-physiological maladaptations in several areas of the limbic system,and that this imbalance may be at the root of the functionalchanges in cognitive processes dependent on these same areas, inparticular in processes for spatial memory acquisition [11], atten-tion [54], and emotional assessment of risk [30,53,55]. Additionalstudies are necessary to understand the brain areas involved inpain-induced perturbations of cognitive functioning.

Neurophysiological studies in animals have demonstrated thatthe neural activity of the medial prefrontal cortex (mPFC) is corre-lated with working memory (WM) processes [5,22,32,62,77]. Theimportance of the mPFC in memory processes is well known from

2398 H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406

human EEG and imaging studies [23,43,57,61]. Anatomically, themPFC connects to the dorsal hippocampal CA1 field through themediodorsal thalamus (MD) [34,81] and this mPFC–MD–hippo-campus circuit is known to be particularly engaged in spatialWM [1,50]. The pivotal role of the MD in memory processes is sup-ported by the fact that medial thalamic lesions in rats induce spa-tial memory deficits and amnesia [27,45,46], while functional andmorphological changes in the human mPFC–MD circuit are ob-served in pathological memory impairment [42,60,85].

However, no information exists on the contribution of themPFC–MD circuit to the onset of cognitive impairment caused bychronic pain. Our interest was therefore to study how persistentinflammatory conditions affect frontothalamic connectivity. Forthis, we recorded the local field potentials (LFPs) of mPFC–MD neu-ral activity in freely moving rats in a spatial WM task before andafter the induction of an animal model of inflammatory pain: kneejoint injection of complete Freund’s adjuvant (CFA).

2. Materials and methods

2.1. Subjects and ethics statement

Twelve adult male Sprague–Dawley rats (275–325 g) wereused. The rats were maintained on a 12-hour light/dark cycle,and all experimental procedures were run at approximately thesame time each day during the light portion of the cycle. Duringtraining sessions all animals were food deprived to 90–95% of theirad libitum body weight. All procedures and experiments adheredto the guidelines of the Committee for Research and Ethical Issuesof IASP [86], with the Ethical Guidelines for Animal Experimenta-tion of the European Community Directive (2010/63/CE), and wereapproved by local boards. All efforts were taken to minimize thenumber of animals used in this study.

2.2. Apparatus and behavioral task

The behavioral task consisted of a food-reinforced spatial alter-nation task on a figure-eight maze. Fig. 1A presents a schematiclayout of the operant chamber. The total dimension of the arenawas 90 � 60 cm, with opaque Plexiglas corridors 15 cm wide andwalls 30 cm high with intramaze cues in the reward locations.Starting from the center of the maze (C), the rats were trained toalternatively visit 2 reward sites (R) to obtain one chocolate-fla-vored pellet (45 mg) that was automatically delivered by a pelletdispenser (Coulbourn Instruments, Whitehall, PA, USA). After visit-ing 1 of the reward locations, the animal had to continue forwardand cross again the central corridor before visiting the opposite re-ward site; food rewards were not dispensed if the animal failed tocross the central corridor immediately before arriving at the re-ward sites or if the animal made 2 consecutive visits to the samereward site (arrows in Fig. 1A illustrate the pattern of correct mazenavigation). Control of pellet dispensers was fully automated usingOpenControl software adapted to this particular task [2].

The behavioral arena was spatially divided into 3 analysiszones: the ‘‘reward zones’’ were the 15 � 5 cm corner areas wherethe animal received a pellet upon a correct alternation; the ‘‘delayzones’’ comprised the area between the reward zones and the cen-tral corridor; the ‘‘choice zone’’ was the area preceding the rewardzones and immediately following the central corridor. The testingroom was sound attenuated, moderately illuminated, and rich invisual cues. During the training period the rats were given 3 daily10-minute sessions to learn the alternation schedule until theyachieved at least 80% of correct runway alternations (Fig. 1B). Anerror was defined by a consecutive visit to the same reward siteor the return route was the same as the approach route. Errors

were not reinforced. After reaching this performance criterionthe animals were surgically implanted with 2 arrays of electrodes(1 per area) for recording extracellular LFPs.

2.3. Surgical procedures

Surgical procedures were the same as in our previous studies[11,74], with the exception of the location of the intracranial arraysof multielectrodes. After reaching the criterion levels for task per-formance (see Section 2.2) the animals were surgically implantedwith 2 multielectrode arrays. The arrays were oriented rostrocau-dally and driven to the right side mPFC and right side MD using astereotaxic microdrive; advancement of the arrays was performedwhile recording the populational neuronal activity to help the cor-rect placement of the electrodes. Each microelectrode array in-cluded 8 filaments (1 array per area) and isonel-coated tungstenwire (35 lm in diameter) (California Fine Wire Company, GroverBeach, CA, USA) with impedances varying between 0.5 and0.7 MOhms at 1 KHz. The arrays were constructed in 4 � 2 architec-ture, interspaced 250 lm between lines and 450 lm between col-umns. The following coordinates (relative to the bregma [58])were used to center the arrays: mPFC (+2.2 mm rostrocaudal,+1.0 mm mediolateral, �3.2 mm dorsoventral), and MD (�2.3 mmrostrocaudal, +0.8 mm mediolateral, �4.7 mm dorsoventral).

2.4. Inflammatory pain model

After 7 days of recovery from the implantation surgery animalswere retrained and recorded for 5 consecutive days (the controlperiod) while performing the runway alternation task in 2 dailysessions of 10 minutes. The following day a persistent monoar-thritic inflammatory pain model [8] was induced by injecting50 lL of CFA in the knee joint of the left side, contralateral to therecording probe implantation, under xylazine and ketamine anes-thesia (10 and 60 mg/kg mixture, respectively) (n = 6, hereafter re-ferred as CFA animals). CFA was prepared by mixing 60 mgdesiccated Mycobacterium butiricum (Difco Laboratories, Oxford,England) with paraffin oil (6 mL), saline (4 mL), and Tween 80(1 mL). A control group (n = 6, SHAM animals) was performedinjecting the same volume of saline.

Animals were then recorded for a period of 16 days (withrecordings performed at days 4, 5, 6, 7, 11, and 16 after CFA injec-tion). The sensory threshold for noxious stimulation was assessedby placing the animals in cages with a metal mesh floor and touch-ing the plantar surface of the paw with von Frey filaments (Somed-ic, Sweden) until slight buckling was caused, and maintained for 6–8 seconds as previously described [14]. These measurements weredone 1 hour after the end of each neural recording session, beforeand after CFA injection. Each filament was applied 10 times with aninterval of 5 seconds between stimuli. A positive response wasconsidered if the paw was sharply withdrawn or flinching oc-curred. The sensory threshold was considered as a minimum of 5positive responses. Nominal force values are presented in grams.

2.5. Neural recordings

Extracellular LFPs were recorded from the implanted microwireelectrodes and processed by a 16-channel Multi-Neuron Acquisi-tion Processor (16-MAP, Plexon Inc., Dallas, TX, USA). LFP signalsrecorded from electrodes were preamplified (500�), band-pass fil-tered (0.5–200 Hz), and digitized at 500 Hz. An overhead videotracking system (CinePlex, Plexon Inc.) was used to provide infor-mation about the rat’s location in the testing environment and,to synchronize the behavior, video recordings with the acquiredneural data. For simplicity, only the signals obtained from theCFA-injected animals were fully analyzed and are presented in this

Fig. 1. Apparatus and behavioral performance. (A) Schematic diagram of events in a figure-eight spatial alternation working memory (WM) task. Starting from the center ofthe maze (C), the animal had to alternately visit 2 reward sites (R) (feeder A and B) to obtain chocolate-flavored pellets. The animal was required to come back to the centerfrom a given reward site before visiting the other reward site. The arrows indicate the direction of travel when going to the left and right goals. (B) The rats were given 3 daily10-minute sessions to learn the alternation schedule until they reach at least 80% of correct runway alternations during a training period of 10 days. (C) Level of sensitivity tomechanical stimulation evaluated using von Frey filaments. A large decrease was observed in the threshold required to induce a paw response after complete Freund’sadjuvant (CFA) injection. (D) Recording period performance for the spatial WM task. A significant decrease in performance level, total number of performed trials per session(E), and running velocity (F) was observed after CFA injection. In addition, the animals spent more time navigating in the delay zone of the behavioral test after CFA injection(G), and also increased their mean interval between 2 consecutive correct alternations (H). Values are means ± SEM. Comparisons between control (CT) and postsurgeryrecording sessions are based on two-way rmANOVA (experimental groups � recording sessions), followed by a post hoc Bonferroni test. CFA, group injected in the knee jointwith complete Freund’s adjuvant; SHAM, control group injected with the same volume of saline. ⁄P < 0.05, ⁄⁄P < 0.01, and ⁄⁄⁄P < 0.001.

H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406 2399

report. This study is a within-subject design in which the statisticalanalyses are made comparing the neurophysiology signals re-corded after the CFA injection with the baseline activity recordedbefore. After the end of all experiments the rats were deeply anes-thetized and their brains were removed to perform the identifica-tion of recording sites under the microscope. In all animals themedial prefrontal electrodes were located in the most ventral re-gion of the prelimbic cortex (bordering the infralimbic corticalarea) and in all animals the thalamic electrodes were located inthe mediodorsal thalamus pars medialis. No animals were re-moved from the study owing to misplacement of the recordingelectrodes.

2.6. Data analysis

2.6.1. Behavioral dataCustom MatLab (R14, MathWorks, Natick, MA, USA) and Python

(v2.6) scripts were used to sort behavioral intervals based on previ-ous or next trial outcome (correct vs incorrect), maze navigationzone (reward, delay, choice zones), and by experimental session(control vs CFA). These measurements were calculated based ontracking navigation vectors obtained from the behavior video record-ings (see Section 2.5). Several parameters concerning the behavioralperformance and navigation maps were examined across experi-mental sessions: percentage of correct alternations per session; totalnumber of performed trials per session; averaged running speed;

average time of navigation across delay zone; and average intervalbetween 2 consecutive correct alternations per session.

2.6.2. Power spectral densityNeural activity data were processed and validated by offline

analysis using NeuroExplorer 4 (NEX, Plexon Inc.) and exportedto MatLab for complementary computational analysis.

The power spectral density (PSD) of mPFC and MD LFP signalswas calculated between 1 and 50 Hz using the Welch method(MatLab function), with 512-point fast Fourier transform of non-overlapping 1 sec epochs (Hanning window). Data are shown asthe percentage of total PSD within the frequency range considered(1–50 Hz). To characterize the theta/gamma spectral power ratios,the raw mPFC/MD LFPs were filtered in the theta (4–9 Hz) andgamma (30–50 Hz) ranges using a zero-phase forward and reversedigital 4-pole Butterworth band-pass filter. Five frequency bandintervals were considered: 1–4 Hz (delta), 4–9 Hz (theta), 9–15 Hz (alpha), 15–30 Hz (beta), and 30–50 Hz (gamma).

2.6.3. Squared coherenceIn order to determine the spectral coupling among signals from

mPFC and MD recorded regions, we calculated the magnitudesquared coherence (COHxy). The COHxy for 2 signals, x and y, isequal to the average squared cross-power spectrum (Pxy) normal-ized by the averaged power spectra of the two signals:

COHxy ¼ jPxyj2=ðPxxPyyÞ

2400 H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406

Coherence assesses the strength of the linear relationship be-tween 2 signals at every frequency f, and its value lies between 0and 1. It estimates the degree to which phases and amplitudesare dispersed at the frequency of interest. COHxy = 0 means thatphases and amplitudes are randomly dispersed among signal sam-ples. Signals are perfectly coherent (COHxy = 1) at a given fre-quency when they have both a constant phase difference and aconstant amplitude ratio over the time considered. In this case,the phases of signals x and y are identical, and the signals are com-pletely phase-locked at this frequency.

2.6.4. Phase coherenceThe phase coherence (PCOHxy) of simultaneously recorded

mPFC–MD LFPs was evaluated calculating the Hilbert transformof each LFP segment for the theta (4–9 Hz) and gamma (30–50 Hz) frequency bands. The phase angles of each signal segmentwere extracted and wrapped between 0� and 360� and displayedas a rose plot histogram. The whole rose plot (360�) was dividedin 60 bins of resolution (6�per bin), with each bin displaying therespective population percentage. The Rayleigh test of uniformity(P < 0.01) was used to assess the resulting phase distributions fordeviations from the circular uniform distribution. The degree ofphase-locking was determined by calculating the concentrationaround the preferred phase in the circular distribution, with its va-lue lying between 0 and 1. PCOHxy = 0 indicates that phase valuesat a particular frequency range are randomly distributed across thetime interval, while PCOHxy = 1 indicates that phase values are ex-actly the same across the time interval. Phase coherence values areinversely related to the Rayleigh P value, with PCOHxy = 0 standingfor uniform distribution. The CircStat MatLab toolbox [6] was usedfor circular plotting, and circular statistics were calculated accord-ing to Fisher [21].

2.6.5. Partial directed coherenceThe statistical method of partial directed coherence (PDC) was

used to identify and quantify the information flow interactionsamong the mPFC–MD circuits. The PDC method has been describedin detail elsewhere [4,11,12,66]. Briefly, PDC is an alternative rep-resentation of multivariate processes involving Granger causalityto uncover the frequency domains of direct influence. Its value liesbetween 0 and 1, and it estimates the degree of functional connec-tivity from one structure to another at a particular frequencyrange. PDC = 0 can be interpreted as the absence of functional con-nectivity and a high PDC, approaching 1, indicates strong connec-tivity between the structures. This can be interpreted as theexistence of information flow from brain area source to target.

2.7. Statistical analysis

All averaged values are given as the mean ± SEM. Statistical sig-nificance was accepted at the level of 5%. Prism 3.0 software(GraphPad) was used for all statistical analyses. For multiple com-parisons, one-way or two-way repeated measures analysis of var-iance (rmANOVA) was used, and a post hoc analysis was carriedout when appropriate using the Bonferroni test. All measures wereaveraged per recording days, except the control period, which wasrepresented by the average of the total recording sessions withinthis period.

3. Results

3.1. Pain-related behavior in spatial working memory task

Our results show that all animals developed mechanical allo-dynia as indicated by a significant decrease in the mechanical force

needed to evoke paw withdrawal to von Frey filament stimulationin the hindpaw ipsilateral to the CFA injection (two-way rmANO-VA; groups: F(1,70) = 403.00, P < 0.0001; time: F(6,70) = 23.36,P < 0.0001; Fig. 1C).

As shown in Fig. 1D, the induction of chronic pain caused a sig-nificant impairment of the performance level with respect to thecontrol (SHAM) group (two-way rmANOVA; groups:F(1,70) = 53.17, P < 0.0001; time: F(6,70) = 8.99, P < 0.0001). Post hocanalysis revealed a significant decrease in the recording sessionsof days 4 and 5 after CFA injection (P < 0.001, Bonferroni test). Thisobservation was also accompanied by changes in the total numberof performed trials (two-way rmANOVA; groups: F(1,70) = 23.75,P < 0.0001; time: F(6,70) = 7.25, P < 0.0001; Fig. 1E), and by changesin the overall running speed (two-way rmANOVA; groups:F(1,70) = 25.73, P < 0.0001; time: F(6,70) = 9.57, P < 0.0001; Fig. 1F).Post hoc analysis revealed that the CFA group performed fewer tri-als during the recording sessions on days 4 and 5 (P < 0.01). Addi-tionally, there was a significant decrease in the running velocity ondays 4 and 7 after CFA injection (P < 0.05, P < 0.01, P < 0.001, andP < 0.05, respectively).

During WM tasks, performance depends on the maintenance ofretrospective memory of the previous response in order to planthe immediate response. Our data show that after CFA injectionthe animals spent more time navigating in the delay zone of thetesting environment (Fig. 1G). ANOVA revealed significant statisti-cal differences across experimental groups and recording sessions(two-way rmANOVA; groups: F(1,70) = 64.11, P < 0.0001; time:F(6,70) = 9.08, P < 0.0001). An additional post hoc analysis demon-strated an increase in the average time of navigation in the delayzone for the CFA group. Another important measure was the aver-age time interval between 2 consecutive correct alternations (ortrials), giving an idea of the level of improvement in theperformance of the animals (Fig. 1H). ANOVA revealed a signifi-cant effect across groups and recording days (two-way rmANOVA;groups: F(1,70) = 21.93, P < 0.0001; time: F(6,70) = 10.64, P < 0.0001),which was increased particularly during day 4 (Bonferroni,P < 0.001).

3.2. Power oscillatory activity of the mPFC–MD circuit

Fig. 2 illustrates the mPFC (Fig. 2A) and MD (Fig. 2B) LFP powerspectral patterns across recording days during the execution of thespatial WM task. Data were calculated separately for the 3 naviga-tion zones described in section 2.2 previously, and only for theCFA-injected animals.

For mPFC, significant differences were encountered across fre-quency bands (two-way rmANOVA; reward zone: F(4,150) = 320.10,P < 0.0001; delay zone: F(4,150) = 419.60, P < 0.0001; choice zone:F(4,150) = 146.00, P < 0.0001) and recording sessions but only forthe reward zone (F(6,150) = 5.25, P < 0.0001; Fig. 2C). An additionalpost hoc analysis revealed an increase of PSD in the delta (day 11,P < 0.05; day 16, P < 0.001) and theta (day 16, P < 0.001) frequencybands for the reward zone; a decrease in the theta band (days 4,5, and 11, P < 0.001) and an increase in the alpha band (day 11,P < 0.01) for the delay zone; and an increase across the theta bandfor the choice zone (day 5, P < 0.01; and P < 0.001 for the remainingdays).

For MD, significant differences were encountered across fre-quency bands (two-way rmANOVA; reward zone: F(4,150) = 247.20,P < 0.0001; delay zone: F(4,150) = 265.10, P < 0.0001; choice zone:F(4,150) = 57.27, P < 0.0001) as well as across recording sessions forthe reward and choice zones (F(6,150) = 2.32, P = 0.0361; andF(6,150) = 2.64, P = 0.0184, respectively, Fig. 2D). Post hoc analysis re-vealed an increase in theta band power for the reward zone (day 6,P < 0.05; day 7, P < 0.001; day 11, P < 0.05; day 16, P < 0.001); a de-crease in the theta band (day 4, P < 0.001; day 5, P < 0.001; day 11,

Fig. 2. Power oscillatory activity of the medial prefrontal cortex (mPFC) and mediodorsal thalamus (MD). The power spectral density (PSD) of local field potentials (LFPs)normalized by the percentage of total power within the frequency range analyzed (1–50 Hz) for mPFC (A) and MD (B) LFP signals, comparing the control period (CT) and afterCFA injection. Data were presented individually across the 3 considered navigation zones of the behavioral task. The inspection of PSD revealed a strong theta-rangecomponent that was present during execution of the behavioral task, which oscillated across pain condition onset. In terms of frequency bands, significant differences wereencountered after CFA injection, namely across the delta, theta, and alpha bands for mPFC (C) and MD (D). In terms of theta/gamma spectral power ratios, significantdifferences were encountered for navigation in the reward and choice zones but not for the delay zone during the pain period (E). Five frequency band intervals wereconsidered: 1–4 Hz (delta), 4–9 Hz (theta), 9–15 Hz (alpha), 15–30 Hz (beta), and 30–50 Hz (gamma). Values are means ± SEM. Comparisons between control (CT) andpostsurgery recording sessions are based on two-way rmANOVA (frequency bands � recording sessions) and one-way rmANOVA for panel E data, followed by a post hocBonferroni test. ⁄P < 0.05, ⁄⁄P < 0.01, and ⁄⁄⁄P < 0.001.

H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406 2401

2402 H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406

P < 0.001) and an increase in the alpha band (day 11, P < 0.05) forthe delay zone; and an increase across day 5 (P < 0.001) and a de-crease across day 11 (P < 0.05) in the theta band during the naviga-tion across the choice zone after CFA injection.

Fig. 2E shows the results of the theta/gamma power ratio anal-ysis. In the case of mPFC, significant differences were observed forthe reward zone (one-way rmANOVA; F(6,35) = 6.93, P < 0.0001) andthe choice zone (F(6,35) = 5.62, P = 0.0004) but not for the delay zone(F(6,35) = 2.05, P = 0.0845). In addition, post hoc analysis showed anincrease in the theta/gamma power ratio with respect to the con-trol period at day 6 (choice zone, P < 0.001; Bonferroni test) andday 16 after CFA injection (reward zone, P < 0.001). In the samemanner there were differences in MD activity in the reward zone(F(6,35) = 4.91, P = 0.0010) and choice zone (F(6,35) = 4.16,P = 0.0029), while the delay zone activity remained without signif-icant differences with respect to the control period (F(6,35) = 1.58,P = 0.1828). Post hoc analysis revealed an increase of the theta/gamma ratio on day 5 (choice zone, P < 0.01) and day 16 (rewardzone, P < 0.001) after CFA injection.

3.3. Oscillatory interactions between mPFC and MD

The coherence activity of simultaneously recorded mPFC andMD LFP channels in the CFA-injected animals are illustrated inFig. 3A. Statistical differences were observed across frequencybands (two-way rmANOVA; reward zone: F(4,150) = 5.83,P = 0.0019; delay zone: F(4,150) = 13.54, P < 0.0001; choice zone:F(4,150) = 19.46, P < 0.0001) and recording days (reward zone:F(6,150) = 15.71, P < 0.0001; delay zone: F(6,150) = 9.33, P < 0.0001;choice zone: F(6,150) = 2.50, P = 0.0246) (Fig. 3B). In addition, posthoc analysis revealed an increase of coherence activity in the deltafrequency band on day 7 postinjection (reward and delay zones;P < 0.01, Bonferroni test) and a decrease in the alpha and betabands across days 11 (both bands, P < 0.05) and 16 (P < 0.01 andP < 0.001, respectively) for the reward zone (Fig. 3B).

Fig. 3C shows an example of the magnitude of the averagetheta and gamma phase coherence oscillations between mPFCand MD during navigation in the choice zone of the arena(Fig. 3C). Data represent the comparison between the controlperiod (CT) and day 6 after CFA injection. The rose plots presenta narrow range of phase coherence in both measured frequencybands, indicating a high degree of phase synchronization ofmPFC–MD activity (Fig. 3C); this suggests a broader coordinationof theta and gamma circuit activities. Indeed, all phase distribu-tions are significantly nonuniform (circular concentrationcoefficient, U = 0.23/0.68 for the theta band and U = 0.21/0.44for the gamma band, respectively; P < 0.01, Rayleigh test ofuniformity).

A more detailed illustration of the mPFC–MD signal phasecoherence interactions across the recording sessions is presentedin Fig. 3D. In the case of the reward zone, ANOVA revealed no dif-ferences for the theta band and differences for the gamma bandduring recording sessions after CFA injection (one-way rmANOVA;F(6,35) = 3.31, P = 0.0109). With respect to the delay zone, significantdifferences were encountered for both frequency bands acrossrecording sessions (theta: F(6,35) = 2.91, P = 0.041; gamma:F(6,35) = 3.36, P = 0.0101). Post hoc analysis revealed an increase ofphase coherence during day 16 for theta (P < 0.05, Bonferroni test)and a decrease for gamma (P < 0.05; Fig. 3D). In addition we foundsignificant differences for both frequency bands during navigationin the choice zone (theta: F(6,35) = 2.75, P = 0.048; gamma:F(6,35) = 3.09, P = 0.0155; Fig. 3D). For the theta band, post hoc anal-ysis revealed an increase in phase coherence during day 6 after CFAinjection (P < 0.05), and for the gamma band an increase duringdays 5 (P < 0.05), 6 and 16 (P < 0.01; Fig. 3D).

3.4. Pain-induced disruption of connectivity between mPFC and MDduring the spatial working memory task execution

The changes in mPFC–MD circuit information flow during theexecution of the spatial WM task by the CFA-injected animals weredetermined by PDC analysis (Figs. 4A and B). In the case of the MDto mPFC direction, statistical differences were observed across fre-quency bands (two-way rmANOVA; reward zone: F(4,150) = 4.34,P = 0.0084; delay zone: F(4,150) = 6.15, P = 0.0014; choice zone:F(4,150) = 36.61, P < 0.0001) and across recordings days (rewardzone: F(6,150) = 141.50, P < 0.0001; delay zone: F(6,150) = 47.66,P < 0.0001; choice zone: F(6,150) = 162.6, P < 0.0001). Post hoc anal-ysis revealed that the PDC level after CFA injection decreased incomparison to the control period. This was observed in all fre-quency bands for the reward zone (Bonferroni test, P < 0.001; andP < 0.01 for the beta band at day 11); for the delay zone (fromthe theta to the gamma bands across all recorded days and forthe delta band across days 6 and 7, P < 0.001); and for the choicezone (all bands, P < 0.001; Fig. 4C).

In the case of the mPFC to MD direction, statistical differenceswere observed across frequency bands (two-way rmANOVA; re-ward zone: F(4,150) = 9.29, P < 0.0001; delay zone: F(4,150) = 66.66,P < 0.0001; choice zone: F(4,150) = 134.60, P < 0.0001) and acrossrecording days (reward zone: F(6,150) = 22.71, P < 0.0001; delayzone: F(6,150) = 10.37, P < 0.0001; choice zone: F(6,150) = 29.21,P < 0.0001). In addition, post hoc analysis revealed that the PDC le-vel in the case of reward zone after CFA injection increased in thealpha band at day 4 postinjection (Bonferroni test, P < 0.01) and de-creased in all considered frequency bands during day 16 (delta:P < 0.01, theta: P < 0.05, alpha: P < 0.01, beta: P < 0.01, and gamma:P < 0.05). A significant decrease in the PDC level with respect to thecontrol period was observed in the delay zone across the theta andalpha bands. In the choice zone, the PDC levels decreased for thealpha, beta, and gamma bands across all recording sessions; forthe delta band only across day 4; and for the theta band only dur-ing days 11 and 16 after injection (Fig. 4D). Taken together, theoverall decreases in PDC level show the CFA injection disruptedthe information transmitted along the mPFC–MD circuit duringthe execution of the WM task.

4. Discussion

In this study we report how the induction of chronic pain affectsthe patterns of mPFC–MD activity by examining LFP activity whilerats performed a spatial alternation WM task. The behavioral taskused in the present study is a classical spatial WM task, in whichthe animals need to maintain in memory the location of the previousreward in order to execute the next goal-oriented decision. These re-sults expand our previous demonstration of pain-induced disruptionof WM processing and suggest that the medial thalamus plays a piv-otal role in the prefrontal–hippocampal connectivity [11].

Our behavioral data showed an impairment of spatial WM per-formance after the induction of the inflammatory chronic painmodel. These results are in agreement with preclinical animalmodel studies [9,25,36,38,64] and clinical studies [35,39,41,70]that have shown that persistent pain affects WM. A comprehensivereview of the cognitive effect of pain has been published elsewhere[49].

The observed performance impairment was also accompaniedby changes in the oscillatory and connectivity patterns of themPFC–MD circuit. The most relevant finding of the present studyis that the onset of chronic pain caused a global decrease inmPFC–MD circuit connectivity as measured by PDC [4,66], indicat-ing that less information is processed by the circuit. More impor-tantly, this decrease occurs over a wide frequency range and in

Fig. 3. Oscillatory and interaction activity of the medial prefrontal cortex–mediodorsal thalamus (mPFC–MD) circuit. Squared coherence and phase coherence of mPFC andMD simultaneously recorded local field potential (LFP) signals. Data were presented individually across the 3 considered navigation zones of the figure-eight behavioral task.The mPFC–MD LFPs showed higher levels of coherence activity during the navigation zones considered (A), which were partially conserved across frequency bands (B).Examples of theta (red) and gamma (gray) LFP phase distributions (C), comparing the control period (CT) and day 6 after injection of complete Freund’s adjuvant (CFA), for onerat during the navigation across the choice zone of the figure-eight spatial working memory task. The numbers in the bottom-right corner of each rose plot indicate thecircular coefficient (U), and the respective level of significance (Rayleigh uniformity test, P < 0.01). The level of mPFC/MD LFP signal phase coherence across the recordingsessions (D). Note that different phase distributions were observed for the theta and gamma frequency bands in each navigation zone. Values are means ± SEM. Five frequencyband intervals were considered: 1–4 Hz (delta), 4–9 Hz (theta), 9–15 Hz (alpha), 15–30 Hz (beta), and 30–50 Hz (gamma). Comparisons between the control period (CT) andpostsurgery recording sessions are based on two-way rmANOVA (frequency bands � recording sessions) and one-way rmANOVA for panel D data, followed by a post hocBonferroni test. ⁄P < 0.05, ⁄⁄P < 0.01, and ⁄⁄⁄P < 0.001.

H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406 2403

both directions of the circuit, albeit more evidently from MD tomPFC. We note that this study uses a within-subject design inwhich the statistical analyses were made comparing the pain-in-duced neurophysiology with the baseline activity recorded beforethe injection of CFA.

It has previously been reported that alterations in the functionalconnectivity of the mPFC–MD–hippocampal circuit translate toimplications for memory processing [22,45,46,83]. However, no

information exists on the possible role of the mPFC–MD circuit inpain-induced memory deficits. Neural activity in the mPFC isknown to be modulated by pain [17,71], and there is morphologicaland functional reorganization in the mPFC region under neuro-pathic pain conditions [29,44]. Neural activity in the hippocampushas also been shown to be affected by persistent pain [10,33,51].At the circuit level, the mPFC and dorsal hippocampus are indi-rectly connected through the MD and the supraspinal projection

Fig. 4. Connectivity patterns of the medial prefrontal cortex–mediodorsal thalamus (mPFC–MD) circuit. The oscillations of connectivity between the 2 recorded brain regionswere determined by partial directed coherence (PDC) analysis; (A) from MD to mPFC and (B) from mPFC to MD. Data were presented individually across the 3 navigationzones considered in the figure-eight behavioral task. A global decrease in connectivity was observed after the induction of the inflammatory pain model, indicating that lessinformation was processed in the mPFC–MD circuit. The main disturbances occurred from MD to mPFC (C) and over a wide frequency range, whereas in the opposite directionfewer changes in the PDC level were observed (D). Values are means ± SEM. Five frequency band intervals were considered: 1–4 Hz (delta), 4–9 Hz (theta), 9–15 Hz (alpha),15–30 Hz (beta), and 30–50 Hz (gamma). Comparisons between the control period (CT) and postsurgery recording sessions are based on two-way rmANOVA (frequencybands � recording sessions), followed by a post hoc Bonferroni test. ⁄P < 0.05, ⁄⁄P < 0.01, and ⁄⁄⁄P < 0.001.

2404 H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406

of spinothalamic pathways, suggesting that the MD is a relay in themedial pain system and plays an important role in the affective andmotivational aspects of pain processing [37]. It has been reportedthat MD neurons respond to noxious stimulation [19] and displaypain-induced c-fos expression [59,79]. Our data show that the maindisturbances in the information balance occur in the MD to mPFCdirection, suggesting that the information that reaches the mPFCfrom the MD is modified during nociceptive stimulation or that itis insufficiently integrated by the mPFC. Also importantly, theMD shares large projections to the anterior cingulate cortex[13,82], which is a forebrain area particularly involved in process-ing the affective–motivational aspects of pain [63,72,80,84], andstimulation of MD thalamic region evokes unit and field responsesin the anterior cingulate cortex [24].

With regard to the spectral power properties of LFPs, we foundthat the mPFC and MD shared similar patterns. In both cases astrong theta-range component was encountered during the behav-ioral task execution, particularly for the delay and choice zones,which oscillates after the onset of the pain condition. Anotherimportant aspect was the theta/gamma spectral power ratios,which increased in the pain period for both recorded regions dur-ing navigation in the reward and choice zones but not in the delayzone. Additionally, LFP activity showed higher levels of coherencethat were partially conserved across the frequency range studied(1–50 Hz) after induction of the chronic pain condition. With re-spect to theta and gamma phase coherence oscillations, we foundthat the phase oscillation of the signals was partially preservedduring navigation through the reward locations but increased in

H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406 2405

the theta band for the delay and choice zones. An intensification oftheta-range coherence has been described during the execution ofWM tasks using human EEG recordings in frontal brain regions[52] and the hippocampal formation [78]. These theta rhythmsare believed to be necessary to orchestrate spatial memories. Ro-bust hippocampal theta rhythm coordination has been reportedto be crucial for phase-firing patterns of frontohippocampal circuitneurons [26,31,73,75], which are largest when spatial WM demandincreases. Another important point is the interactions of thetaoscillations with higher frequency oscillations in the gamma range.Together, these oscillations allow multiple memories to fire duringeach theta cycle but in a way that preserves the temporal order ofthe memories [40]. This suggests that the timing scheme organizedby theta/gamma oscillations may be a general brain code for stor-ing multiple items in order. In fact, the increase in theta power, orthe decrease in gamma power, have been proposed as markers ofcognitive decline [16,47,48,65,76]. It should be noted that our re-sults confirm an increase of the theta/gamma power ratio, whichis correlated with a reduction in behavioral performance. In fact,this modification could be driven by a pathological imbalance ofthe intrinsic characteristics of the theta/gamma peak that occursin both recorded regions after the onset of the pain model. Forexample, an increase in theta EEG power, an increase in peak oscil-lations, and enhanced coherence in thalamocortical modules haveall been reported in neurogenic pain patients, and these changesare correlated with a disruption of WM performance in the pa-tients [67–69].

In summary, our neurophysiological recordings of mPFC–MDconnectivity obtained from awake behaving rats performing a spa-tial alternation task show that the onset of an inflammatory painmodel caused a global decrease in frontothalamic connectivity thataccompanies the deficit in WM performance. This disturbance inthe mPFC–MD circuit expands our previous report of a similar de-crease in connectivity between the prefrontal cortex and the hip-pocampus [11]. This broad change in neurophysiological activitysuggests that the prefrontal cortex plays a pivotal role in theimpairment of cognitive functioning that is an important compo-nent of the decrease in quality of life experienced by chronic painpatients.

Conflicts of interest statement

The authors do not have any conflicts of interest.

Acknowledgments

This work was funded by FEDER funds through the OperationalCompetitiveness Programme (COMPETE) and by national fundsthrough Fundação para a Ciência e a Tecnologia under the projectFCOMP-01-0124-FEDER-011202 (PTDC/SAU-NEU/100773/2008),PhD grant SFRH/42500/2007 and BIAL Foundation – Project 126/08.

References

[1] Aggleton JP, Brown MW. Episodic memory, amnesia, and the hippocampal–anterior thalamic axis. Behav Brain Sci 1999;22:425–44.

[2] Aguiar P, Mendonça L, Galhardo V. OpenControl: a free opensource softwarefor video tracking and automated control of behavioral mazes. J NeurosciMethods 2007;166:66–72.

[3] Apkarian AV, Hashmi JA, Baliki MN. Pain and the brain: specificity andplasticity of the brain in clinical chronic pain. PAIN� 2011;152:S49–64.

[4] Baccala LA, Sameshima K. Partial directed coherence: a new concept in neuralstructure determination. Biol Cybern 2001;84:463–74.

[5] Baeg EH, Kim YB, Huh K, Mook-Jung I, Kim HT, Jung MW. Dynamics ofpopulation code for working memory in the prefrontal cortex. Neuron2003;40:177–88.

[6] Berens P. CircStat: a MATLAB toolbox for circular statistics. J Statist Software2009;31:1–23.

[7] Buhle J, Wager TD. Performance-dependent inhibition of pain by an executiveworking memory task. PAIN� 2010;149:19–26.

[8] Butler SH, Godefroy F, Besson JM, Weil-Fugazza J. A limited arthritic model forchronic pain studies in the rat. PAIN� 1992;48:73–81.

[9] Cain CK, Francis JM, Plone MA, Emerich DF, Lindner MD. Pain-related disabilityand effects of chronic morphine in the adjuvant-induced arthritis model ofchronic pain. Physiol Behav 1997;62:199–205.

[10] Cardoso-Cruz H, Lima D, Galhardo V. Instability of spatial encoding by CA1hippocampal place cells after peripheral nerve injury. Eur J Neurosci2011;33:2255–64.

[11] Cardoso-Cruz H, Lima D, Galhardo V. Impaired spatial memory performance ina rat model of neuropathic pain is associated with reduced hippocampus–prefrontal cortex connectivity. J Neurosci 2013;33:2465–80.

[12] Cardoso-Cruz H, Sameshima K, Lima D, Galhardo V. Dynamics of circadianthalamocortical flow of information during a peripheral neuropathic paincondition. Front Integr Neurosci 2011;5:43.

[13] Chai SC, Kung JC, Shyu BC. Roles of the anterior cingulate cortex and medialthalamus in short-term and long-term aversive information processing. MolPain 2010;6:42.

[14] Chaplan SR, Bach FW, Pogrel JW, Chung JM, Yaksh TL. Quantitative assessmentof tactile allodynia in the rat paw. J Neurosci Methods 1994;53:55–63.

[15] Choi DS, Choi DY, Whittington RA, Nedeljkovic SS. Sudden amnesia resulting inpain relief: the relationship between memory and pain. PAIN�

2007;132:206–10.[16] Deiber MP, Ibañez V, Missonnier P, Herrmann F, Fazio-Costa L, Gold G,

Giannakopoulos P. Abnormal-induced theta activity supports early directed-attention network deficits in progressive MCI. Neurobiol Aging2009;30:1444–52.

[17] Devoize L, Alvarez P, Monconduit L, Dallel R. Representation of dynamicmechanical allodynia in the ventral medial prefrontal cortex of trigeminalneuropathic rats. Eur J Pain 2011;15:676–82.

[18] Dick B, Eccleston C, Crombez G. Attentional functioning in fibromyalgia,rheumatoid arthritis, and musculoskeletal pain patients. Arthritis Rheum2002;47:639–44.

[19] Dostrovsky JO, Guilbaud G. Nociceptive responses in medial thalamus of thenormal and arthritic rat. PAIN� 1990;40:93–104.

[20] Eccleston C. Chronic pain and distraction: an experimental investigation intothe role of sustained and shifting attention in the processing of chronicpersistent pain. Behav Res Ther 1995;33:391–405.

[21] Fisher NI. Statistical analysis of circular data. Cambridge, UK: CambridgeUniversity Press; 1993.

[22] Floresco SB, Seamans JK, Phillips AG. Selective roles for hippocampal,prefrontal cortical, and ventral striatal circuits in radial-arm maze tasks withor without a delay. J Neurosci 1997;17:1880–90.

[23] Hillary FG, Genova HM, Chiaravalloti ND, Rypma B, DeLuca J. Prefrontalmodulation of working memory performance in brain injury and disease. HumBrain Mapp 2006;27:837–47.

[24] Hsu MM, Shyu BC. Electrophysiological study of the connection betweenmedial thalamus and anterior cingulate cortex in the rat. Neuroreport1997;8:2701–7.

[25] Hu Y, Yang J, Wang Y, Li W. Amitriptyline rather than lornoxicam amelioratesneuropathic pain-induced deficits in abilities of spatial learning and memory.Eur J Anaesthesiol 2010;27:162–8.

[26] Hyman JM, Zilli EA, Paley AM, Hasselmo ME. Medial prefrontal cortex cellsshow dynamic modulation with the hippocampal theta rhythm dependent onbehavior. Hippocampus 2005;15:739–49.

[27] Isseroff A, Rosvold HE, Galkin TW, Goldman-Rakic PS. Spatial memoryimpairments following damage to the mediodorsal nucleus of the thalamusin rhesus monkeys. Brain Res 1982;232:97–113.

[28] Jamison RN, Sbrocco T, Parris WC. The influence of problems withconcentration and memory on emotional distress and daily activities inchronic pain patients. Int J Psychiatry Med 1988;18:183–91.

[29] Ji G, Sun H, Fu Y, Li Z, Galhardo V, Neugebauer V, Pais-Vieira M. Cognitiveimpairment in pain through amygdala-driven prefrontal cortical deactivation.J Neurosci 2010;30:5451–64.

[30] Ji G, Sun H, Fu Y, Li Z, Pais-Vieira M, Galhardo V, Neugebauer V. Cognitiveimpairment in pain through amygdala-driven prefrontal cortical deactivation.J Neurosci 2010;30:5451–64.

[31] Jones MW, Wilson MA. Theta rhythms coordinate hippocampal–prefrontalinteractions in a spatial memory task. PLoS Biol 2005;3:e402.

[32] Jung MW, Qin Y, McNaughton BL, Barnes CA. Firing characteristics of deeplayer neurons in prefrontal cortex in rats performing spatial working memorytasks. Cereb Cortex 1998;8:437–50.

[33] Kodama D, Ono H, Tanabe M. Altered hippocampal long-term potentiationafter peripheral nerve injury in mice. Eur J Pharmacol 2007;574:127–32.

[34] Kuroda M, Yokofujita J, Murakami K. An ultrastructural study of the neuralcircuit between the prefrontal cortex and the mediodorsal nucleus of thethalamus. Prog Neurobiol 1998;54:417–58.

[35] Lee DM, Pendleton N, Tajar A, O’Neill TW, O’Connor DB, Bartfai G, Boonen S,Casanueva FF, Finn JD, Forti G, Giwercman A, Han TS, Huhtaniemi IT, Kula K,Lean ME, Punab M, Silman AJ, Vanderschueren D, Moseley CM, Wu FC, McBethJ, Group ES. Chronic widespread pain is associated with slower cognitiveprocessing speed in middle-aged and older European men. PAIN�

2010;151:30–6.[36] Leite-Almeida H, Almeida-Torres L, Mesquita AR, Pertovaara A, Sousa N,

Cerqueira JJ, Almeida A. The impact of age on emotional and cognitive

2406 H. Cardoso-Cruz et al. / PAIN�

154 (2013) 2397–2406

behaviours triggered by experimental neuropathy in rats. PAIN�

2009;144:57–65.[37] Lima D. Ascending pathways: anatomy and physiology. In: Basbaum A,

Bushnell MC, editors. The science of pain. New York: Academic Press; 2008. p.477–526.

[38] Lindner MD, Plone MA, Francis JM, Cain CK. Chronic morphine reduces pain-related disability in a rodent model of chronic, inflammatory pain. Exp ClinPsychopharmacol 1999;7:187–97.

[39] Ling J, Campbell C, Heffernan TM, Greenough CG. Short-term prospectivememory deficits in chronic back pain patients. Psychosom Med2007;69:144–8.

[40] Lisman JE, Idiart MA. Storage of 7 +/� 2 short-term memories in oscillatorysubcycles. Science 1995;267:1512–5.

[41] Luerding R, Weigand T, Bogdahn U, Schmidt-Wilcke T. Working memoryperformance is correlated with local brain morphology in the medial frontaland anterior cingulate cortex in fibromyalgia patients: structural correlates ofpain–cognition interaction. Brain 2008;131:3222–31.

[42] Marenco S, Stein JL, Savostyanova AA, Sambataro F, Tan HY, Goldman AL,Verchinski BA, Barnett AS, Dickinson D, Apud JA, Callicott JH, Meyer-Lindenberg A, Weinberger DR. Investigation of anatomical thalamo-corticalconnectivity and FMRI activation in schizophrenia.Neuropsychopharmacology 2012;37:499–507.

[43] McNab F, Klingberg T. Prefrontal cortex and basal ganglia control access toworking memory. Nat Neurosci 2008;11:103–7.

[44] Metz AE, Yau HJ, Centeno MV, Apkarian AV, Martina M. Morphological andfunctional reorganization of rat medial prefrontal cortex in neuropathic pain.Proc Natl Acad Sci USA 2009;106:2423–8.

[45] Mitchell AS, Browning PG, Wilson CR, Baxter MG, Gaffan D. Dissociable rolesfor cortical and subcortical structures in memory retrieval and acquisition. JNeurosci 2008;28:8387–96.

[46] Mitchell AS, Dalrymple-Alford JC. Dissociable memory effects after medialthalamus lesions in the rat. Eur J Neurosci 2005;22:973–85.

[47] Moretti DV, Fracassi C, Pievani M, Geroldi C, Binetti G, Zanetti O, Sosta K,Rossini PM, Frisoni GB. Increase of theta/gamma ratio is associated withmemory impairment. Clin Neurophysiol 2009;120:295–303.

[48] Moretti DV, Miniussi C, Frisoni GB, Geroldi C, Zanetti O, Binetti G, Rossini PM.Hippocampal atrophy and EEG markers in subjects with mild cognitiveimpairment. Clin Neurophysiol 2007;118:2716–29.

[49] Moriarty O, McGuire BE, Finn DP. The effect of pain on cognitive function: areview of clinical and preclinical research. Prog Neurobiol 2011;93:385–404.

[50] Moser E, Moser MB, Andersen P. Spatial learning impairment parallels themagnitude of dorsal hippocampal lesions, but is hardly present followingventral lesions. J Neurosci 1993;13:3916–25.

[51] Mutso AA, Radzicki D, Baliki MN, Huang L, Banisadr G, Centeno MV, RadulovicJ, Martina M, Miller RJ, Apkarian AV. Abnormalities in hippocampalfunctioning with persistent pain. J Neurosci 2012;32:5747–56.

[52] Onton J, Delorme A, Makeig S. Frontal midline EEG dynamics during workingmemory. Neuroimage 2005;27:341–56.

[53] Pais-Vieira M, Aguiar P, Lima D, Galhardo V. Inflammatory pain disrupts theorbitofrontal neuronal activity and risk-assessment performance in a rodentdecision-making task. PAIN� 2012;153:1625–35.

[54] Pais-Vieira M, Lima D, Galhardo V. Sustained attention deficits in rats withchronic inflammatory pain. Neurosci Lett 2009;463:98–102.

[55] Pais-Vieira M, Mendes-Pinto MM, Lima D, Galhardo V. Cognitive impairmentof prefrontal-dependent decision-making in rats after the onset of chronicpain. Neuroscience 2009;161:671–9.

[56] Park DC, Glass JM, Minear M, Crofford LJ. Cognitive function in fibromyalgiapatients. Arthritis Rheum 2001;44:2125–33.

[57] Passingham D, Sakai K. The prefrontal cortex and working memory:physiology and brain imaging. Curr Opin Neurobiol 2004;14:163–8.

[58] Paxinos G, Watson C. The rat brain in stereotaxic coordinates. 4th ed. SanDiego: Academic Press; 1998.

[59] Pearse DD, Bushell G, Leah JD. Jun, Fos and Krox in the thalamus after C-fiberstimulation: coincident-input-dependent expression, expression acrosssomatotopic boundaries, and nucleolar translocation. Neuroscience2001;107:143–59.

[60] Pergola G, Gunturkun O, Koch B, Schwarz M, Daum I, Suchan B. Recall deficitsin stroke patients with thalamic lesions covary with damage to theparvocellular mediodorsal nucleus of the thalamus. Neuropsychologia2012;50:2477–91.

[61] Petrides M. The role of the mid-dorsolateral prefrontal cortex in workingmemory. Exp Brain Res 2000;133:44–54.

[62] Rainer G, Rao SC, Miller EK. Prospective coding for objects in primateprefrontal cortex. J Neurosci 1999;19:5493–505.

[63] Rainville P, Duncan GH, Price DD, Carrier B, Bushnell MC. Pain affect encodedin human anterior cingulate but not somatosensory cortex. Science1997;277:968–71.

[64] Ren W-J, Liu Y, Zhou L-J, Li W, Zhong Y, Pang R-P, Xin W-J, Wei X-H, Wang J,Zhu H-Q, Wu C-Y, Qin Z-H, Liu G, Liu X-G. Peripheral nerve injury leads toworking memory deficits and dysfunction of the hippocampus byupregulation of TNF-a in rodents. Neuropsychopharmacology2011;36:979–92.

[65] Rossini PM, Rossi S, Babiloni C, Polich J. Clinical neurophysiology of agingbrain: from normal aging to neurodegeneration. Prog Neurobiol2007;83:375–400.

[66] Sameshima K, Baccalá L. Using partial directed coherence to describe neuronalensemble interactions. J Neurosci Methods 1999;94:93–103.

[67] Sarnthein J, Jeanmonod D. High thalamocortical theta coherence in patientswith neurogenic pain. Neuroimage 2008;39:1910–7.

[68] Sarnthein J, Morel A, von Stein A, Jeanmonod D. Thalamocortical thetacoherence in neurological patients at rest and during a working memory task.Int J Psychophysiol 2005;57:87–96.

[69] Sarnthein J, Stern J, Aufenberg C, Rousson V, Jeanmonod D. Increased EEGpower and slowed dominant frequency in patients with neurogenic pain. Brain2006;129:55–64.

[70] Schmidt-Wilcke T, Luerding R, Weigand T, Jürgens T, Schuierer G, Leinisch E,Bogdahn U. Striatal grey matter increase in patients suffering fromfibromyalgia: a voxel-based morphometry study. PAIN� 2007;132:S109–16.

[71] Seifert F, Bschorer K, De Col R, Filitz J, Peltz E, Koppert W, Maihöfner C. Medialprefrontal cortex activity is predictive for hyperalgesia and pharmacologicalantihyperalgesia. J Neurosci 2009;29:6167–75.

[72] Shyu BC, Sikes RW, Vogt LJ, Vogt BA. Nociceptive processing by anteriorcingulate pyramidal neurons. J Neurophysiol 2010;103:3287–301.

[73] Siapas AG, Lubenov EV, Wilson MA. Prefrontal phase locking to hippocampaltheta oscillations. Neuron 2005;46:141–51.

[74] Silva A, Cardoso-Cruz H, Silva F, Galhardo V, Antunes L. Comparison ofanesthetic depth indexes based on thalamocortical local field potentials inrats. Anesthesiology 2010;112:355–63.

[75] Sirota A, Montgomery S, Fujisawa S, Isomura Y, Zugaro M, Buzsáki G.Entrainment of neocortical neurons and gamma oscillations by thehippocampal theta rhythm. Neuron 2008;60:683–97.

[76] Stam CJ, van der Made Y, Pijnenburg YA, Scheltens P. EEG synchronization inmild cognitive impairment and Alzheimer’s disease. Acta Neurol Scand2003;108:90–6.

[77] Stern CE, Sherman SJ, Kirchhoff BA, Hasselmo ME. Medial temporal andprefrontal contributions to working memory tasks with novel and familiarstimuli. Hippocampus 2001;11:337–46.

[78] Tesche CD, Karhu J. Theta oscillations index human hippocampal activationduring a working memory task. Proc Natl Acad Sci USA 2000;97:919–24.

[79] Traub RJ, Silva E, Gebhart GF, Solodkin A. Noxious colorectal distentioninduced-c-Fos protein in limbic brain structures in the rat. Neurosci Lett1996;215:165–8.

[80] Treede RD, Kenshalo Jr DR, Gracely RH, Jones AKP. The cortical representationof pain. PAIN� 1999;79:105–11.

[81] Vertes RP. Interactions among the medial prefrontal cortex, hippocampus andmidline thalamus in emotional and cognitive processing in the rat.Neuroscience 2006;142:1–20.

[82] Wang CC, Shyu BC. Differential projections from the mediodorsal andcentrolateral thalamic nuclei to the frontal cortex in rats. Brain Res2004;995:226–35.

[83] Wang GW, Cai JX. Disconnection of the hippocampal–prefrontal corticalcircuits impairs spatial working memory performance in rats. Behav Brain Res2006;175:329–36.

[84] Wang J, Luo F, Chang J, Woodward D, Han J. Parallel pain processing in freelymoving rats revealed by distributed neuron recording. Brain Res2003;992:263–71.

[85] Woodward ND, Karbasforoushan H, Heckers S. Thalamocorticaldysconnectivity in schizophrenia. Am J Psychiatry 2012;169:1092–9.

[86] Zimmermann M. Ethical guidelines for investigations of experimental pain inconscious animals. PAIN� 1983;16:109–10.