In kissmyjazz/ketamineCoxPaper: Code for the paper on ketamine-COX experiment

library("papaja")
library("ggplot2")
library("tidyverse")
library("scales")
library("gridExtra")
library("mice")
library("here")
library("kableExtra")
library("ggraph")
library("grid")
library("igraph")

# Seed for random number generation
set.seed(42)
knitr::opts_chunk$set(
  cache = TRUE,
  cache.comments = FALSE,
  collapse = TRUE,
  warning = FALSE,
  message = FALSE,
  error = FALSE,
  echo = FALSE,
  strip.white = TRUE,
  comment = "#>",
  fig.path = "../figures/",
  results = "asis",
  tidy = "styler",
  dev = c('pdf', 'tiff'),
  dev.args = list(pdf = list(colormodel = 'cmyk'), 
                  tiff = list(compression = 'lzw')),
  dpi = 600,
  fig.width = 7,
  fig.height = 7,
  concordance = TRUE,
  global.par = TRUE
)

par(font.main = 1, cex.main = 1.05)
theme_set(theme_apa(base_size = 15) + theme(legend.position = "bottom"))

Introduction

Although the lifetime prevalence of schizophrenia is comparatively low (0.3-0.7%) [@americanpsychiatricassociationDiagnosticStatisticalManual2013], the direct and indirect costs are high owing to early onset and a disabling chronic course [@chongGlobalEconomicBurden2016], making finding an effective treatment important to both patients and society. Schizophrenia symptoms are divided into three categories: positive (hallucinations, delusions), negative (anhedonia, blunted affect, social withdrawal), and cognitive (impaired executive function and memory). Traditionally, aetiology of schizophrenia has been linked to hyperactivity in dopaminergic signal transduction. However, early research on the behavioural effects of glutamate N-methyl-D-aspartate receptor (NMDAR) noncompetitive antagonists phencyclidine (PCP) and ketamine and more recent advances in neurobiology and genetics have indicated that glutamate signal transduction also plays an important role [@javittGlutamatergicTheoriesSchizophrenia2010, @schizophreniaworkinggroupofthepsychiatricgenomicsconsortiumBiologicalInsights1082014].

Glutamate is the most ubiquitous neurotransmitter in the mammalian brain: about 60% of neurons contain glutamate, and virtually all neurons have some type of glutamate receptor. Early pharmacological observations of dissociative anaesthetics ketamine and PCP have lead to the hypothesis that glutamate plays a causal role in schizophrenia. These drugs inactivate signal transduction by blocking the ion channel within the NMDAR, a major type of ionotropic glutamate receptor. Modulation of the postsynaptic NMDARs is crucial for activity-dependent synaptic plasticity and memory. By analogy, a hypofunction of NMDAR-mediated glutamatergic signalling was implicated in the pathophysiology of schizophrenia [@coyleGlutamateSchizophreniaDopamine2006]. Evidence for this hypothesis partly relies on the observation that when normal adults are administered an NMDAR antagonist, such as ketamine, they develop negative, cognitive and positive symptoms like those seen in schizophrenia [@adlerComparisonKetamineinducedThought1999; @krystalSubanestheticEffectsNoncompetitive1994; @newcomerKetamineinducedNMDAReceptor1999]. Administration of these compounds to someone with schizophrenia exacerbates their symptoms [@lahtiEffectsKetamineNormal2001]. Furthermore, stimulating NMDAR-mediated signalling using agonists of the glycine modulatory site has been effective in alleviating some of the symptoms of schizophrenia in clinical trials [@heresco-levyPlacebocontrolledTrialDcycloserine2002; @tsaiDserineAddedAntipsychotics1998]. Despite the early enthusiasm, the therapeutic efficacy of the first antipsychotic drugs developed to augment glutamatergic neurotransmission via allosteric modulation of NMDAR was found to be rather limited [@buchananCognitiveNegativeSymptoms2007; @iwataEffectsGlutamatePositive2015]. Glutamatergic neurotransmission is very complex, as it involves different receptor types, molecular adaptation mechanisms, and tight coupling between oxidative metabolism and the glutamate-glutamine neurotransmitter cycle [@napolitanoNeurometabolicProfilingKetamine2016]. Therefore the nature of glutamatergic deficits in schizophrenia and the best ways to counter them remain very active areas of research [@baluNMDAReceptorSchizophrenia2016].

Subchronic administration of NMDAR antagonists such as ketamine, PCP, or dizocilpine (MK-801) readily replicates cognitive and negative aspects of schizophrenia in animal models. A daily treatment of 7 to 10 days is sufficient to impair declarative and working memory, as well as executive functions in rodents [@badoEffectsLowdoseDserine2011; @liPersistingCognitiveDeficits2011; @neillAnimalModelsCognitive2010]. Negative symptoms are also replicated by a reduction of non-aggressive behaviour in the social interaction test [@beckerKetamineinducedChangesRat2004].

Cytochrome c oxidase (COX; EC 1.9.3.1) is a key enzyme in complex IV of the mitochondrial electron transport chain, where its activity mirrors the generation of ATP. COX activity is tightly coupled to neuronal activity at a molecular level and primarily reflects long-term post-synaptic energy expenditure [@wong-rileyBigenomicRegulationCytochrome2012]. Nuclear respiratory factor 1 (NRF-1) binds to COX subunit genes and functionally regulates neuronal metabolism. NRF-1 also co‐regulates AMPA glutamate receptor subunit 2 and NMDA receptor subunits 1 (NR1) and 2b (NR2b) genes, therefore mitochondrial energy generation and glutamate cell transduction are controlled by overlapping transcriptional mechanisms [@dharCouplingEnergyMetabolism2009; @dharNuclearRespiratoryFactor2009]. Histological processing of brain tissue allows to measure COX levels at high resolution and in a large number of brain regions. It provides a snapshot of mainly excitatory neuronal activity of the entire brain and is an excellent method to look for novel biochemical targets.

COX activity has been found to be increased in post-mortem brain samples of schizophrenia patients, and in correlation with their intellectual and emotional impairment [@princeMitochondrialFunctionDifferentially1999; @princePutamenMitochondrialEnergy2000]. Human patients usually have a history of chronic antipsychotic drug treatment and changes of COX activity also reflect the effects of medication. Accordingly, COX activity in a number of brain regions was enhanced by the administration of several antipsychotic drugs in rats [@princeHistochemicalDemonstrationAltered1998]. In this regard, animal models allow more precision in disentangling the brain localisations of dysfunctions and the therapeutic effect of different drugs. The data on brain metabolic effects of subanaesthetic treatment with NMDAR antagonists is relatively sparse. One study assessed the effect of PCP after chronic administration of 28 days in rats and reported reduced COX activity in several brain regions, especially in basal ganglia and septum [@princeNormalizationCytochromecOxidase1997]. In another study employing the enzymatic assay of COX activity in brain tissue homogenates after ketamine had been administered at 25 mg/kg for 7 days, increased enzyme activity was observed in striatum and hippocampus 1 to 6 h after the last dose of ketamine [@deoliveiraBehavioralChangesMitochondrial2011].

The aim of this study was to identify brain regions with a) persistently different oxidative energy metabolism levels by comprehensive mapping and b) to reveal potentially differential regional activity co-variance after subchronic ketamine administration.

Materials and Methods

Results

Discussion

As measured by cytochrome oxidase histochemistry, oxidative metabolic activity was increased by subchronic ketamine treatment in nine brain regions: 2 regions represented sensory thalamus, 2 basal ganglia, 3 cortical areas, 1 each hippocampus and superior colliculi. No reductions in COX levels after subchronic ketamine treatment were identified. Overall, COX levels differed in less than 5% of the brain regions that were studied. These results reflect the targeted effect of ketamine on the central nervous system, but also probably under-represent the full impact of ketamine treatment because of the modest statistical power of the current study. Because the experimental design had a 24 h washout period between the last ketamine injection and the animal sacrifice, the observed changes in COX activity likely reflect persistent neurochemical and behavioural adaptations. Positive effects of subchronic ketamine regimen on oxidative metabolism are in good agreement with previous studies that show such effects by different methods in both primates and rodents [@deoliveiraBehavioralChangesMitochondrial2011; @langsjoEffectsSubanestheticKetamine2004; @maltbieKetaminePharmacologicalImaging2017]. For example, in humans, subanaesthetic infusion of ketamine during PET scan increased regional glucose metabolic rate in most brain regions, including the anterior and posterior cingulate cortex, basal ganglia, and thalamus. No regions with decreased glucose metabolism were found [@langsjoEffectsSubanestheticKetamine2004]. Similarly, studies of brain haemodynamic activity in rats, monkeys, and humans all have shown quite consistent pattern of regional increases in BOLD signal indicative of neuronal excitation [reviewed in @maltbieKetaminePharmacologicalImaging2017]. Peak BOLD response occurs 3–5 minutes after the start of ketamine infusion and reflects ketamine blood levels [@deakinGlutamateNeuralBasis2008].

Rapid changes in metabolic processes are likely controlled via the release of neurotransmitters. Besides NMDAR, ketamine binds to dopamine D~2~ receptors in high affinity state, serotonin 5-HT~2A~, sigma 1, and opioid receptors [@frohlichReviewingKetamineModel2014; @kapurNMDAReceptorAntagonists2002]. The immediate result of ketamine administration is the inhibition of the NMDA receptors. There is some experimental evidence that GABA-ergic interneurons may be readily susceptible to NMDAR antagonistic properties of ketamine [@homayounNMDAReceptorHypofunction2007]. Inhibition of interneurons would release tonic inhibition from their downstream targets and explain numerous findings of increased neuronal activity after the delivery of subanaesthetic doses of ketamine. As ketamine binds to several other types of receptors with lower affinity, its effects are complex and reflected by changes in a number of neurotransmitter pools [reviewed in @napolitanoNeurometabolicProfilingKetamine2016]. Region-specific increases in neurotransmitter turnover have been recorded in glutamate-glutamine-GABA [@amitaiRepeatedPhencyclidineAdministration2012; @chatterjeeNeurochemicalMolecularCharacterization2012; @moghaddamActivationGlutamatergicNeurotransmission1997], serotonergic, dopaminergic, and acetylcholinergic systems among others [@chatterjeeNeurochemicalMolecularCharacterization2012]. Organismic adaptations to subchronic ketamine administration also include changes in the expression of receptors. Of note here is the finding of the elevated cortical gene-expression of two NMDAR subunits NR1 and NR2b [@chatterjeeNeurochemicalMolecularCharacterization2012]. The expression of these genes and cytochrome c oxidase is controlled by an overlapping transcriptional machinery [@dharCouplingEnergyMetabolism2009; @dharNuclearRespiratoryFactor2009]. The results of the current study therefore likely reflect concurrent adaptations involving several neurotransmitter circuitries. Differential correlation analysis is complimentary to the comparison of the mean COX levels. It accounts for the shape of the distribution of the results and allows to compare pairwise regional interactions between conditions. Four layers of the superior colliculi and dorsomedial subdivision of the periaqueductal gray matter that is located directly underneath the superior colliculus showed reduced positive correlations over all brain regions after subchronic ketamine treatment. Several layers of the superior colliculi were also prominently different between the two rat groups in regional pairwise correlations of their COX levels. Their correlations with cortical and hippocampal brain regions were strongly positive in the control condition and tended to become negative after subchronic ketamine administration. Dorsomedial periaqueductal gray, hippocampal CA3, and prerubral field were three other brain regions with 3-4 pairwise significant reductions in positive correlations in the latter condition.

The opposite pattern of changes in the pairwise correlation coefficients between two conditions was also observed, but in fewer brain regions. None of these brain regions showed increased positive pairwise correlations after ketamine administration with more than 2 other brain regions. There was also little overlap between brain regions that exhibited more positive and more negative pairwise correlations in the ketamine group rats. The reduction of the positive pairwise correlations after subchronic ketamine administration was a more common pattern and various layers of the superior colliculi were prominent nodes in this process. Lateral posterior thalamic nucleus, lateral geniculate nucleus, and ectorhinal cortex had a similar but attenuated profile to the superior colliculi. Lateral posterior and lateral geniculate thalamic nuclei and superficial layers of superior colliculi receive direct retinal projections and are important for visually guided behaviours and multisensory integration [@allenVisualInputMouse2016; @dragerTopographyVisualSomatosensory1976]. Ectorhinal cortex is part of the ventral visual stream in rodents [@nishioHigherVisualResponses2018]. Some neurons in the deeper layers of the superior colliculus are also responsive to auditory stimulation. In cats, the acute infusion of ketamine facilitated spontaneous activity and auditory responses of neurons in the intermediate layers of the superior colliculus [@populinAnestheticsChangeExcitation2005]. Curiously, while several primarily visual regions were less functionally correlated to other brain areas in ketamine-treated rats, the opposite pattern was exhibited by some auditory areas such as medial geniculate and auditory cortex. The ventral part of the medial geniculate is the primary auditory relay nucleus of the thalamus that projects to the primary auditory cortex [@ryugoDifferentialTelencephalicProjections1974]. COX protein levels in MGV were also elevated. It is noteworthy that acute doses of ketamine produce visual and auditory deficits in healthy humans and rodents [e.g. @hillhouseEffectsNoncompetitiveNmethylDaspartate2015; @umbrichtKetamineinducedDeficitsAuditory2000]. Auditory, more so than visual, hallucinations are also a prominent symptom of schizophrenic psychosis. As behaviour was not studied in the current study, we can only speculate here, but it is possible that metabolic changes in visual and auditory system found in our study are to some degree similar to changes in the balance between the two sensory systems observed during acute subanaesthetic ketamine administration as well as psychotic episodes [@lahtiEffectsKetamineNormal2001; @powersKetamineInducedHallucinations2015].

Other sensory systems more crucial to rodents were also affected by ketamine administration. The prerubral field belongs to the H fields of Forel: a meeting place of several fibre bundles forming cortico-striato-thalamo-cortical loops where the convergence is observed of sensorimotor, associative, and limbic pathways [@neudorferNeuroanatomicalBackgroundFunctional2018]. In rat, the prerubral field receives axonal projections from whisker-sensitive region of the spinal trigeminal nucleus [@veinanteThalamicProjectionsWhiskersensitive2000]. The olfactory circuitry was also represented via an increased COX levels in the piriform cortex [@alkoborssyModulationOlfactorydrivenBehavior2019].

Hippocampal regions CA2, CA3, and dentate gyrus exhibited negative pairwise correlations with mostly visual brain areas in ketamine-treated rats. The dentate gyrus also had higher COX levels. Deficits in visual memory are common among schizophrenia patients [@saykinNeuropsychologicalFunctionSchizophrenia1991] and their hippocampi show reduced volumes as well as atypical BOLD activity profiles [@heckersNeuroimagingStudiesHippocampus2001]. Likewise, increased activity in the hippocampus and visual memory deficits were also found in ketamine models of schizophrenia [@chatterjeeNeurochemicalMolecularCharacterization2012; @duncanMetabolicMappingRat1998; @neillAnimalModelsCognitive2010]. Current results largely replicate previous findings.

Ketamine treatment was associated with increased COX activity in both the anterior and posterior parts of the cingulate cortex. Whereas correlational patterns differentiated between the anterior and posterior parts: anterior cingulate cortex exhibited a pattern of decreased pairwise correlations in ketamine-treated rats and retrosplenial cortex showed the opposite tendency. Reduced neuronal density has been reported in both superficial and deep layers of the anterior cingulate cortex (ACC) [@benesAnalysisArrangementNeurons1987; @benesDeficitsSmallInterneurons1991; @benesDensityPyramidalNonpyramidal2001], whereas an acute subanaesthetic dose of ketamine increased blood flow in ACC of schizophrenia patients [@lahtiKetamineActivatesPsychosis1995]. Hyper-responsiveness of ACC in patients with major depression apparently predicts subsequent antidepressant efficacy of ketamine [@salvadoreIncreasedAnteriorCingulate2009]. Furthermore, in rodents, anaesthetic doses of ketamine increase c-fos expression in the anterior cingulate and show some toxicity primarily in the posterior/retrosplenial division of the cingulate cortex [@nagataPropofolInhibitsKetamineinduced1998; @sharpMK801KetamineInduce1991]. We confirmed the activation of the cingulate cortex by ketamine and observed a functional decoupling between anterior and posterior divisions cingulate of the cingulate cortex.

Finally, it is worth mentioning the locus coeruleus. This noradrenergic brain region had borderline positive increase in median correlation in ketamine-treated rats. Its pairwise correlations were elevated primarily with limbic areas. The small size of LC makes it difficult to study its activity in neuroimaging studies. Several post-mortem studies have found increased noradrenaline levels in basal ganglia and cerebrospinal fluid of schizophrenia patients [@crowMonoamineMechanismsChronic1979; @farleyNorepinephrineChronicParanoid1978; @lakeSchizophreniaElevatedCerebrospinal1980]. However, a post-mortem study was not able to identify morphological or volumetric changes of LC in schizophrenia [@lohrLocusCeruleusMorphometry1988]. In the ketamine animal model the brain levels of noradrenaline were also elevated, especially in ketamine withdrawal condition [@chatterjeeNeurochemicalMolecularCharacterization2012]. Changes in noradrenergic neurotransmission may be secondary to the enhanced dopaminergic neurotransmission.

The limitation of our study was a relatively low number of the animals, therefore some significant effects of ketamine administration are likely to be masked by the presence of 1 or 2 outliers or missing observations. Most of the significant results were obtained from subcortical areas, similarly to other recent COX studies on different animal models from our lab [@kanarikSociabilityTraitRegional2018; @matrovCerebralOxidativeMetabolism2019]. Histology remains one of the best ways to measure neuronal activity in smaller subcortical brain regions. This study to our knowledge is the first attempt to evaluate the effect of ketamine administration on cytochrome c oxidase levels by means of histochemistry.

Conclusion

The functioning of several sensory neurocircuits, especially visual and auditory, was affected by ketamine administration. Superior colliculus was the prominent nexus of changes in functional connectivity in ketamine-treated rats. The important role of hippocampi, cingulate cortex, and basal ganglia in a ketamine model of schizophrenia was confirmed as well.

Abbreviations

Brain regions

Brain regions are abbreviated according to the reference brain atlas [@paxinosRatBrainStereotaxic2007]. For each brain region the distance to bregma in millimetres according to the reference brain section is provided in parentheses.

APT (-4.8), Anterior pretectal nucleus; AuD (-4.8), Auditory cortex, secondary, dorsal; AuV (-4.8), Auditory cortex, secondary, ventral; B (-1.3), Nucleus basalis; CA1 (-4.8), Hippocampal CA1(Cornu Ammonis area 1); CA2 (-4.8), Hippocampal CA2; CA3 (-4.8), Hippocampal CA3; Cg1,2 (-0.3), Cingulate cortex, areas 1 & 2; CPu (-1.3), Caudate putamen; CPuDM (+0.2), Caudate putamen, dorsomedial; CPuV (+0.2), Caudate putamen, ventral; DG (-3.8), Hippocampal dentate gyrus; DG (-4.8), Hippocampal dentate gyrus; DLG (-4.8), Dorsal lateral geniculate; DMPAG (-5.8), Dorsomedial periaqueductal gray; DP (+2.2), Dorsal peduncular cortex; Dp (-5.8), Superior colliculi, deep gray/white layer; DRD (-8.0), Dorsal raphe, dorsal; Ect (-4.8), Ectorhinal cortex; GP (-1.3), Globus pallidus; InG (-5.8), Superior colliculi, intermediate gray layer; InWh (-5.8), Superior colliculi, intermediate white layer; LC (-10.04), Locus coeruleus; LP (-4.8), Lateral posterior nucleus; LPMR (-3.8), Lateral posterior thalamus, mediorostral; LSD (-0.3), Lateral septal nucleus, dorsal; M2 (-0.3), Motor cortex, secondary; MCPO (-0.3), Magnocellular preoptic nucleus; MGV (-5.8), Medial geniculate, ventral; MM (-4.8), Medial mamillary, medial; Op (-5.8), Superior colliculi, optic nerve layer; Pir (-1.3), Piriform cortex; PR (-4.8), Prerubral field; PtA (-4.8), Parietal association cortex; RSA (-5.8), Retrosplenial cortex, agranular; RSG (-4.8), Retrosplenial cortex, granular; RSGa (-5.8), Retrosplenial cortex, granular; RSGb (-5.8), Retrosplenial cortex, granular; RTg (-8.0), Reticulotegmental nucleus; Shi (+0.2), Septohippocampal nucleus; SPTg (-8.0), Subpeduncular tegmental nucleus; SubD (-5.8), Subiculum, dorsal; SuG (-5.8), Superior colliculi, superficial gray layer; V1M (-8.0), Primary visual cortex, monocular; V2 (-4.8), Visual cortex, secondary.

Other

AMPA, $\alpha$-amino-3-hydroxy-5-methylisoxazole-4-propionate; BOLD, blood oxygen level dependent; COX, cytochrome c oxidase; DCA, differential correlation analysis; GABA, $\gamma$-aminobutyric acid; NMDAR, N-methyl-D-aspartate receptor; NRF-1, Nuclear respiratory factor 1; PCP, phencyclidine; PET, positron emission tomography.

Acknowledgements

This research has been supported by the Tallinn University Research Fund project TF1515 (RS), the Tallinn University ASTRA project “TU TEE – Tallinn University as a promoter of intelligent lifestyle” financed by the European Union European Regional Development Fund 2014-2020.4.01.16-0033 (SI, MS) and by the Estonian Ministry of Education and Science project IUT20-40 (JH).

\pagebreak