Journal of Affective Disorders 120 (2010) 245–248

Contents lists available at ScienceDirect

Journal of Affective Disorders

j ourna l homepage: www.e lsev ie r.com/ locate / j ad

Brief report

Regionally-specific changes in levels of tumour necrosis factor in thedorsolateral prefrontal cortex obtained postmortem from subjects withmajor depressive disorder

Brian Dean a,b,e,⁎, Nahed Tawadros a,c, Elizabeth Scarr a,d, Andrew S. Gibbons a,b

a The Rebecca L. Cooper Research Laboratories, The Mental Health Research Institute, Parkville, Australiab The University of Melbourne Department of Psychiatry, Australiac The University of Melbourne Department of Pathology, Australiad The University of Melbourne Department the Centre for Neuroscience, Australiae The Department of Psychological Medicine, Clayton, Victoria, Australia

a r t i c l e i n f o

⁎ Corresponding author. NHMRC Senior Research FHead, The Rebecca L. Cooper Research LaboratoriesResearch Institute, Locked Bag 11, Parkville, VictoTel.: +61 613 9389 2940; fax: +61 613 9387 5061.

E-mail address: anddali@unimelb.edu.au (B. Dean)

0165-0327/$ – see front matter © 2009 Elsevier B.V.doi:10.1016/j.jad.2009.04.027

a b s t r a c t

Article history:Received 8 April 2009Received in revised form 22 April 2009Accepted 22 April 2009Available online 14 May 2009

Background: From studies in the periphery, changed levels of tumour necrosis factor (TNF)have been implicated in the pathophysiology of major depressive disorders (MDD). Thereforewe decided to determinewhether TNFwas altered in the frontal cortex (Brodmann's areas (BA)24 and 46) from 10 subjects with MDD and 10 control subjects.Methods: Tissue homogenates were prepared from the left hemisphere and levels of TNF trans-membrane (tmTNF) and TNF soluble (sTNF) forms measured by Western blots.Results: tmTNF was significantly increased in BA 46 (mean±SEM: 7.70±0.92 vs. 3.18±0.87Ratio Internal Control, pb0.001), but not BA 24, from subjects with MDD, there was no changein levels of sTNF in either CNS region.Limitations: As the report of tmTNF in postmortem CNS from subjects with MDD, our findingsneed to be replicated in another group of cases.Conclusions: Our data supports the hypothesis that changes in pro-inflammatory pathways maybe involved in the pathophysiology of MDD. Targeting these pathways may be a new approachto treating the disorder.

© 2009 Elsevier B.V. All rights reserved.

Keywords:Major depressive disorderDorsolateral prefrontal cortexTumour necrosis factorPostmortem CNS

1. Introduction

Cytokines are implicated in the pathophysiology of de-pression because subjects with the disorder have an ex vivoreduction in natural killer cell activity, decreased T and B cellproliferation and increased activation of cell-mediated im-mune responses, in vivo (for review see Schiepers et al.,2005). Moreover, blood levels of tumour necrosis factor(TNF), interleukin 1β andmonocyte chemotactic protein 1 arereported as increased in untreated major depressive disorder

ellow and Laboratory, The Mental Healthria, 3052, Australia.

.

All rights reserved.

(MDD) (Piletz et al., 2008) and reduced following antide-pressant treatment (Kenis and Maes, 2002). Thus, high levelsof pro-inflammatory cytokines could be involved in the path-ophysiology of MDD.

TNF-related pathways have been studied extensively be-cause they offer opportunities for therapeutic intervention ininflammatory disorders (Tracey et al., 2008). TNF is a 233amino acid, 26 kDa protein present in cell membranes as TNFtrans-membrane form (tmTNF) that is cleaved to TNF solubleform (sTNF: 17 kDa) by the ADAM metallopeptidase domain17. Given the potential role for TNF in MDD we measuredlevels of sTNF and tmTNF in postmortem CNS from subjectswith MDD and age / sex matched subjects with no history ofpsychiatric disease (controls). Levels of TNFweremeasured inthe dorsolateral prefrontal cortex (DLPFC: Brodmann's area(BA) 46) and anterior cingulate cortex (ACC: BA 24) because

246 B. Dean et al. / Journal of Affective Disorders 120 (2010) 245–248

neuroimaging studies suggest both regions are affected by thepathophysiology of the disorder (Baxter et al., 1989; Pizzagalliet al., 2001).

2. Methods

Following approval from the Ethics Committee of theVictorian Institute of Forensic Medicine and the NorthWesternMental Health Program Behavioural and Psychiatric Researchand Ethics Committee, tissue was collected from Brodmann'sareas (BA) 46 and 24 from 10 MDD subjects and 10 controls.MDD was diagnosed according to DSMIV criteria followingcase history reviews using the Diagnostic Instrument for BrainStudies (Roberts et al., 1998). Duration of illness (DOI) wascalculated as the time from first contact with a psychiatricservice to death and postmortem interval (PMI) was calculatedas the time between death and autopsy. Where death was notwitnessed, tissuewas taken from individuals seenalive up to 5hbefore being found dead and PMI was calculated as the intervalhalf way between the donor being found dead and last beingseen alive. All cadavers were refrigerated within 5 h of beingfound; tissue was rapidly frozen to −70 °C within 30 min of

Table 1Demographic, CNS collection and treatment data on subjects with MDD and age / s

ID no. Age(year)

Sex pH PMI(h)

Diagnoses Cause of death Antidepressants

1 37 M 6.84 57.75 MDD Hanging Sertraline2 51 M 6.71 41 MDD Hanging Clomipramine

Desmethylclomi3 50 F 6.85 50.5 MDD Mixed drug

toxicity

4 77 F 6.49 16.7 MDD Doxepintoxicity

Doxepin

5 69 M 6.45 44.5 MDD Drowning6 79 M 6.32 24 MDD Asphyxia Nortriptyline7 55 M 6.6 47.75 MDD Hanging Venlafaxine8 68 M 6.65 60.75 MDD Hanging Citalopram

Mianaserin9 87 F 6.44 24.5 MDD Chest infection10 51 F 6.49 23.5 MDD Toxicity to

quetiapineMean 62 6M / 4F 6.60 39SEM 5.0 0.06 5.01 70 M 6.11 59 Control RV Rupture following pericardial2 75 F 6.01 53 Control Multi organ failure. septicaemia.

3 75 M 6.19 69.4 Control Cardiogenic shock4 55 M 6.69 30.5 Control Coronary artery atherosclerosis5 52 M 6.52 33.75 Control Cardiomegaly and IHD, CAD6 63 M 6.55 50.25 Control Coronary artery atherosclerosis7 66 M 6.47 71.75 Control Coronary artery atheroma8 42 M 6.45 30.5 Control Ishaemic heart disease9 80 F 6.28 55 Control Ishaemic heart disease10 47 F 5.89 24 Control Pulmonary embolismMean 62 7M / 3F 6.31 48SEM 4.1 0.08 5.4

autopsy. The CNS tissue pHwasmeasured as a good indicator oftissue integrity (Torrey et al., 2000).

Levels of sTNF and tmTNF were measured by Westernblot analyses essentially as described extensively for otherproteins (Dean et al., 2007; Scarr et al., 2006), but optimisedfor the measurement of sTNF and tmTNF. Thus, the proteinloaded was 42 µg / well, the resolving gel was 15% PAGE Gel,primary antibody was diluted 1:1000 (TTBS, incubated o/n at4 °C) and the secondary antibody diluted 1:2000 (TTBS,incubated 45min r.t.). Importantly, the person completing theanalyses was blind to diagnoses but the coding of the samplesensured that a case and their matched control were on thesame gel. Two images were captured (light mode for theMW standards and chemiluminescent for the immunogenicbands) and merged to allow the estimation of the molecularweight of each antigenic band. The densities of the immuno-genic bands were measured on the chemiluminescent imageand expressed as a ratio of the corresponding immunogenicband in the internal control to accommodate gel-to-gel var-iation. As we have shown previously (Dean et al., 2007; Scarret al., 2006), this approach also allows comparisons of levelsof proteins both within and across CNS regions.

ex matched controls from which cortex was obtained for this study.

Dose(mg/L)

Benzodiazepines Dose(mg/L)

Other Dose(mg/L)

~0.2 Ethanol Detected0.3

pramine DetectedDiazepam ~0.8 Codeine 0.1Nordiazepam ~1.5 Paracetamol ~220

Propoxyphene 7.4Norpropoxyphene 27Metoclopramide ~53Verapamil ~3.5Zolpidem ~0.1Tramadol ~15.2

5.4

Diphenhydramine ~4~0.7 Nordiazepam ~0.1 Paracetamol ~1000~0.2Trace Temazepam ~0.2~0.1

Diazepam 100 Quetiapine ~0.3Nordiazepam ~0.4

drainageIHD Midazolam Trace Morphine 0.2

Lignocaine 0.1Paracetamol Trace

247B. Dean et al. / Journal of Affective Disorders 120 (2010) 245–248

For experimental data, outliers were identified with aGrubb's test; if necessary datawas analysed with and withoutthe outlier(s). Each data set was then analysed with theD'Agostino and Pearson normality test, which is best suitedfor analyses of distribution in small data sets. Student t-testswere used to compare age, PMI and CNS pH betweendiagnostic cohorts whilst gender frequency between cohortswas compared with Fisher's exact test. Two-way ANOVA fol-lowed by a post hoc Bonferroni test was used to identifyvariances in sTNF and tmTNF with diagnoses and across CNSregions. Finally, as TNF levels may vary with gender (O'Brienet al., 2007) and a number of subjects with MDD died bysuicide, an ANCOVA with sex and suicide as covariants withCNS region and diagnoses as variables was completed.Relationships between demographic and CNS collection datawere assessed using Pearson product moment correlationsassuming a straight line as best fit.

3. Result

One outlier for tmTNF in BA 24 from the control subjectswas detected, all data was binomially distributed whether ornot this data point was included. There was no significantdifference in the mean age or PMI between the subjects withMDD and controls (mean±SEM; Age+outlier: 62.4±5.0 vs.62.5±4.1 year, p=0.99; Control Age−outlier: 63.7±4.4 year, p=0.85. PMI+outlier 39.1±5.0 vs. 47.7±5.4 h,p=0.26; Control PMI−outlier 49.3±5.8 h, p=0.20). CNSpH was higher in MDD (pH+outlier: 6.58±0.06 vs. 6.29±0.09; pb0.05; Control−outlier: 6.31±0.08, pb0.05). Thenumber of males and females were closely matched acrosscohorts, therefore there was no significant difference infrequency of gender between diagnostic cohorts (+outlierp=1.00, −outlier p=1.00; Table 1).

sTNF did not vary with diagnoses (F=0.5, df=1,1,36,p=0.53) but did vary between CNS regions (F=9.4,df=1,1,36, pb0.005: Fig. 1); there was no significant interac-tion between the variables (F=2.7, df=1,1,36, p=0.11).The regional variation in sTNF was due to higher levels ofthe protein in BA 46 compared to BA 24 in control subjects(1.68±0.10 vs. 1.12±0.10 ratio IC, pb0.01). Neither gender(p=0.93) nor suicide (0.80) were significant covariants.

Fig. 1. Levels of sTNF and tmTNF in BA 24 and BA 46 from subjects with MDD and an edetected for tmTNF in BA 24 from the control subjects (inset) and hence results are

tmTNF varied with diagnoses and CNS region whether(diagnosis: F=15.1, df=1,1,36, pb0.0005, CNS region: F=19.3 df=1,1,36, pb0.0001) or not (diagnosis: F=17.0, df=1,1,35, pb0.0005, CNS region F=21.4, df=1,1,36, pb0.0001)the outlier was included. In both scenarios there was asignificant interaction between variables (+outlier: F=6.29,df=1,1,36, pb0.05; −outlier: F=5.18, df=1,1,35, pb0.05).The variation with diagnoses was due to increased tmTNF inBA 46 from subjects with MDD (7.70±0.92 vs. 3.18±0.87Ratio IC, pb0.001: Fig. 1). Variation with region was due tohigher levels of tmTNF in BA 46 compared to BA 24 fromsubjects with MDD (7.70±0.92 vs. 2.82±0.53 ratio IC,pb0.001). Neither gender (p=0.84) nor suicide (p=0.48)were significant covariants.

When comparing experimental variables to demographicor CNS collection data, the slope of the regression line did notdeviate significantly from zero.

4. Discussion

We found increased (+142%) levels of tmTNF in BA 46,but not BA 24, from subjects with MDD compared to controls;sTNF did not vary with diagnoses. These data suggest thereare not generalised changes in tmTNF throughout the CNS ofsubjects with MDD and further analyses will be required tounderstand the full extent of changes in that protein in theCNS from subjects with the disorder. Moreover, the DFPLC is aregion of the cortex recognised as being important in main-taining cognition (Duncan and Owen, 2000) whereas the ACCis suggested to be involved in controllingmood (Drevets et al.,2008). Hence, our finding that changes in tmTNF are presentin the DFPLC, but not the ACC, may be a preliminary indicatorof a role for changes in that protein in the genesis of thecognitive deficits associated with MDD (Gohier et al., 2009).sTNF is generated by enzymic cleavage of tmTNF (Tracey et al.,2008) and therefore our current data would suggest that thecontrol of this process is not perturbed in MDD despite thepresence of extra substrate in the DFPLC from subjects withthe disorder.

One limitation of our study is small cohort sizes but aretrospective power analyses (PS Power and Sample SizeCalculations Software) shows that cohort sizes of 10 are

quivalent number of age / sexed matched controls. An outlying data point wasreported with and without that point.

248 B. Dean et al. / Journal of Affective Disorders 120 (2010) 245–248

sufficient to confidently delineate differences in levels oftmTNF of the order of magnitude we report between MD andcontrols. An additional confound is that subjects with MDDhad received antidepressant drugs before death but it isdifficult to postulate how antidepressant treatment couldincrease tmTNF levels in the regionally-selective manner wereport.

The mechanisms by which tmTNF may contribute to thepathophysiology of MDD are unclear. Studies in transgenicanimals suggest roles for sTNF and tmTNF in ataxia, seizures,paresis, CNS inflammation and demyelination (Probert et al.,1997), none of which are strongly associated with the path-ophysiology of MDD. It has been suggested that the pro-inflammatory cytokines can affect serotonergic function(Schiepers et al., 2005) whilst TNF can affect glutamatergicneurotransmission (Pickering et al., 2005). Significantly, bothneurotransmitter systems are affected in MDD (Pittenger et al.,2007; Trivedi, 2006).

TNF transgenic mice show abnormal expression of cholineacetyltransferase (Aloe et al., 1999) whilst muscarinic recep-tor antagonists reduce levels of CNS TNF in rats (Pavlov andTracey, 2006). In addition, we have recently reported changesin the levels of muscarinic M2 receptors in BA 46 fromsubjects with MDD (Gibbons et al., 2009) and it has beenshown that stimulating HEL 299 cells with TNF andinterleukin-1β decreases the expression of muscarinic M2receptors (Haddad et al., 1996). These data all support a longstanding hypothesis that altered cholinergic function in theCNS contributes to MDD (Janowsky et al., 1972). Thereforedata from this study added to our earlier postmortem CNSstudy (Gibbons et al., 2009), plus in vitro (Haddad et al., 1996)and knockout mouse (Pavlov and Tracey, 2006) data showinginteractions between TNF and cholinergic markers, may in-dicate changes in cholinergic function are associated withchanges in levels of tmTNF.

Whatever the role of TNF in MDD, our data is significant inthat it is the first to showchanges in a cytokine in the DFPLC ofsubjects with MDD, supporting the psychoneuroimmunologyhypotheses of depression (Leonard and Myint, 2009).

Role of funding sourceFunding for this study was provided by NHMRC (Grants 400016, 192399

and 509333), Australian Rotary, the Preston and Loui Geduld Foundation, theRebecca L. Cooper Medical Research Foundation, TheWoods Family Trust andthe Victorian State Government. None of these organisations had any otherrole in any aspect of the study.

Conflict of interestNo conflict declared.

Acknowledgements

Brian Dean is a Senior NHMRC Research Fellow (Level B)(400016). Elizabeth Scarr is the Australian Rotary RoyceAbbey Postdoctoral Fellow (Mental Illness). NT is therecipient of scholarships from the Preston and Loui Geduldand the Rebecca L. Cooper Medical Research Foundation. Thisresearch was funded in part by NHMRC Project Grants #192399 and 509333, The Woods Family Trust and theOperational Infrastructure Support (OIS) from the VictorianState Government.

References

Aloe, L., Fiore, M., Probert, L., Turrini, P., Tirassa, P., 1999. Overexpression oftumour necrosis factor alpha in the brain of transgenic mice differentiallyalters nerve growth factor levels and choline acetyltransferase activity.Cytokine 11, 45–54.

Baxter Jr., L.R., Schwartz, J.M., Phelps, M.E., Mazziotta, J.C., Guze, B.H., Selin,C.E., Gerner, R.H., Sumida, R.M., 1989. Reduction of prefrontal cortexglucose metabolism common to three types of depression. Archives ofGeneral Psychiatry 46, 243–250.

Dean, B., Boer, S.A., Mackinnon, A., Berk, M., 2007. CNS 14-3-3zeta: changeswith sex but not psychiatric diagnoses or psychotropic drug treatment.Schizophrenia Research 93, 51–57.

Drevets, W.C., Savitz, J., Trimble, M., 2008. The subgenual anterior cingulatecortex in mood disorders. Central Nervous Sysem Spectrum 13, 663–681.

Duncan, J., Owen, A.M., 2000. Common regions of the human frontal loberecruited by diverse cognitive demands. Trends in Neurosciences 23,475–483.

Gibbons, A.S., Scarr, E., McLean, C., Sundram, S., Dean, B., 2009. Decreasedmuscarinic receptor binding in the frontal cortex of bipolar disorder andmajor depressive disorder subjects. Journal of Affective Disorders 116 (3),184–191.

Gohier, B., Ferracci, L., Surguladze, S.A., Lawrence, E., El, H.W., Kefi, M.Z.,Allain, P., Garre, J.B., Le, G.D., 2009. Cognitive inhibition and workingmemory in unipolar depression. Journal of Affective Disorders 116 (1–2),100–105.

Haddad, E.B., Rousell, J., Lindsay, M.A., Barnes, P.J., 1996. Synergy betweentumor necrosis factor alpha and interleukin 1beta in inducing transcrip-tional down-regulation of muscarinic M2 receptor gene expression.Involvement of protein kinase A and ceramide pathways. Journal ofBiological Chemistry 271, 32586–32592.

Janowsky, D.S., el-Yousef, M.K., Davis, J.M., Sekerke, H.J., 1972. A cholinergic–adrenergic hypothesis of mania and depression. Lancet 2, 632–635.

Kenis, G., Maes, M., 2002. Effects of antidepressants on the production ofcytokines. International Journal of Neuropsychopharmacology 5, 401–412.

Leonard, B.E., Myint, A., 2009. The psychoneuroimmunology of depression.Human Psychopharmacology 24, 165–175.

O'Brien, S.M., Fitzgerald, P., Scully, P., Landers, A., Scott, L.V., Dinan, T.G., 2007.Impact of gender and menstrual cycle phase on plasma cytokineconcentrations. Neuroimmunomodulation 14, 84–90.

Pavlov, V.A., Tracey, K.J., 2006. Controlling inflammation: the cholinergic anti-inflammatory pathway. Biochemical Society Transactions 34, 1037–1040.

Pickering, M., Cumiskey, D., O'Connor, J.J., 2005. Actions of TNF-alpha onglutamatergic synaptic transmission in the central nervous system.Experimental Physiology 90, 663–670.

Piletz, J.E., Halaris, A., Iqbal, O., Hoppensteadt, D., Fareed, J., Zhu, H., Sinacore,J., Lindsay, D.C., 2008. Pro-inflammatory biomakers in depression:treatment with venlafaxine. World Journal of Biological Psychiatry 9,1–11.

Pittenger, C., Sanacora, G., Krystal, J.H., 2007. The NMDA receptor as atherapeutic target in major depressive disorder. Central Nervous SystemNeurological Disorder Drug Targets 6, 101–115.

Pizzagalli, D., Pascual-Marqui, R.D., Nitschke, J.B., Oakes, T.R., Larson, C.L.,Abercrombie, H.C., Schaefer, S.M., Koger, J.V., Benca, R.M., Davidson, R.J.,2001. Anterior cingulate activity as a predictor of degree of treatmentresponse in major depression: evidence from brain electrical tomogra-phy analysis. American Journal of Psychiatry 158, 405–415.

Probert, L., Akassoglou, K., Kassiotis, G., Pasparakis,M., Alexopoulou, L., Kollias,G.,1997. TNF-alpha transgenic and knockoutmodels of CNS inflammationand degeneration. Journal of Neuroimmunology 72, 137–141.

Roberts, S.B., Hill, C.A., Dean, B., Keks, N.A., Opeskin, K., Copolov, D.L., 1998.Confirmation of the diagnosis of schizophrenia after death using DSM-IV:a Victorian experience. Australian and New Zealand Journal of Psychiatry32, 73–76.

Scarr, E., Keriakous, D., Crossland, N., Dean, B., 2006. No change in corticalmuscarinic M2, M3 receptors or [35S]GTPgammaS binding in schizo-phrenia. Life Sciences 78, 1231–1237.

Schiepers, O.J., Wichers, M.C., Maes, M., 2005. Cytokines and majordepression. Progress in Neuro-psychopharmacology & Biological Psy-chiatry 29, 201–217.

Torrey, E.F., Webster, M., Knable, M., Johnston, N., Yolken, R.H., 2000. Thestanley foundation brain collection and neuropathology consortium.Schizophrenia Research 44, 151–155.

Tracey, D., Klareskog, L., Sasso, E.H., Salfeld, J.G., Tak, P.P., 2008. Tumornecrosis factor antagonist mechanisms of action: a comprehensivereview. Pharmacology and Therapeutics 117, 244–279.

Trivedi, M.H., 2006. Major depressive disorder: remission of associatedsymptoms. Journal of Clinical Psychiatry 67 (Suppl 6), 27–32.