VISUALISING MICROGLIA IN NEURODEGENERATIVE DISEASES ERIK NUTMA
VISUALISING MICROGLIA IN NEURODEGENERATIVE DISEASES Erik Nutma
The image on the front cover is a negative of The Bubble Nebula (NGC 7635) captured by the Hubble Space telescope. The Bubble Nebula is located 7,100 light-years away and was discovered by William Herschel in 1787. The seething star forming this nebula is 45 times more massive than our sun. Gas on the star gets so hot that it escapes away into space as a “stellar wind” moving at over 4 million miles per hour. This outflow sweeps up the cold, interstellar gas in front of it, forming the outer edge of the bubble. As the surface of the bubble’s shell expands outward, it slams into dense regions of cold gas on one side of the bubble. This asymmetry makes the star appear dramatically off-center from the bubble, with its location in the 10 o’clock position in the Hubble view. The colours correspond to red for oxygen, pink for hydrogen and blue for nitrogen. I was drawn to this image because of the similarities between astronomy and visualising complex molecular processes in the human brain. The font used for the chapter pages is a personally modified Gilbert Color Bold. The Gilbert font is a tribute font to honor the memory of Gilbert Baker, the creator of the LGBT Rainbow Flag. The Gilbert Color font was later designed to express diversity and inclusion. The work described in this thesis was performed at the Department of Pathology of the Amsterdam UMC - location VUmc (Amsterdam, The Netherlands) under supervision of prof. dr. S. Amor and prof.dr. P. van der Valk and in collaboration with the Department of Brain Sciences of the Imperial College (London, United Kingdom) under supervision of dr. D.R.J. Owen. Financial support for the printing of this thesis was kindly provided by the Stichting MS Research and by the Graduate School Neurosciences Amsterdam Rotterdam (ONWAR). Printing: Ridderprint | www.ridderprint.nl ISBN: 978-94-6458-048-8 Copyright © 2022 E. Nutma. All rights reserved. No part of this publication may be reproduced, stored, or transmitted in any form or by any means without prior permission of the author
VRIJE UNIVERSITEIT Visualising microglia in neurodegenerative diseases ACADEMISCH PROEFSCHRIFT ter verkrijging van de graad Doctor of Philosophy aan de Vrije Universiteit Amsterdam, op gezag van de rector magnificus prof.dr. J.J.G. Geurts, in het openbaar te verdedigen ten overstaan van de promotiecommissie van de Faculteit der Geneeskunde op donderdag 7 Juli 2022 om 11.45 uur in een bijeenkomst van de universiteit, De Boelelaan 1105 Door Erik Nutma Geboren te Leeuwarden
promotoren: prof.dr. S. Amor prof.dr. P. van der Valk copromotor: dr. D.R.J. Owen promotiecommissie: prof.dr. J.M. Rozemuller prof.dr. M. Kipp prof.dr. D. Baker prof.dr. E.M. Hol dr. W. Scheper dr. J.J. Bajramovic dr. M. Bugiani
It has been said that the great events of the world take place in the brain. - Oscar Wilde
Table of contents Chapter 1 General introduction Inflammation in CNS neurodegenerative diseases Immunology 2018; 154(2):204-219 PART I. VISUALISING MICROGLIA IN NEURODEGENERATION Chapter 2 A quantitative neuropathological assessment of translocator protein expression in multiple sclerosis Brain 2019; 142(11):3440-3455 Chapter 3 Activated microglia do not increase 18 kDa translocator protein expression in the multiple sclerosis brain Glia 2021; 69(10):2447-2458 Chapter 4 Translocator protein is a marker of activated microglia in rodent models but not human neurodegenerative diseases Manuscript submitted Chapter 5 Cellular sources of TSPO expression in healthy and diseased brain European Journal of Nuclear Medicine and Molecular Imaging 2021; 49(1):146-143 9 37 61 77 113
PART II. MULTIPLE SCLEROSIS PATHOLOGY Chapter 6 Synaptic Loss in Multiple Sclerosis Spinal Cord Annals of Neurology 2020 88(3):619-625 PART III. GLIA AS INNATE IMMUNE CELLS OF THE CNS Chapter 7 Astrocyte and oligodendrocyte crosstalk in the central nervous system Cells 2020; 9(3):600 Chapter 8 White matter microglia heterogeneity in the CNS Acta Neuropathologica 2022; 143(2):125-141 Chapter 9 General discussion Imaging immunological processes from blood to brain in amyotrophic lateral sclerosis Clinical and Experimental Immunology 2021; 206(3)301-313 APPENDICES Summary Nederlandse samenvatting List of publications Dankwoord About the author 141 155 177 201 215
General introduction Adapted from: Inflammation in CNS neurodegenerative diseases Erik Nutma+, Jodie Stephenson+, Paul van der Valk, Sandra Amor +shared first author Immunology 2018; 154(2):204-219
10 Chapter 1 Neuroinflammation in the central nervous system The prevalence of neuroinflammatory and neurodegenerative diseases highly depends on the country surveyed, yet the most prevalent disease globally is dementia, with an estimated incidence of 9.33% worldwide1. The increase in the incidence of dementia, as with many neurodegenerative diseases, is in part due to the ageing population3, since an ageing brain, or one following peripheral infections or other insults is ‘primed’ to render the central nervous system (CNS) more susceptible to damage4. Although neurodegenerative and neuroinflammatory diseases often have different aetiologies, a common feature is chronic activation of innate immune cells within the CNS, mainly microglia (Table 1), and in other diseases such as multiple sclerosis (MS), also the influx of peripheral immune cells across the blood-brain barrier (BBB). Microglia can be implicated as the primary cause of neuropathology in many CNS diseases and can induce neuronal cell death through both direct and indirect pathways. For example, when microglia are activated by infectious pathogens, aggregated proteins or superoxides, they can cause excitotoxic neuronal death by expressing iNOS, releasing glutamate or proteases such as cathepsins or matrix metalloproteases5. Additionally, activated microglia release tumor necrosis factor (TNF) increasing neuronal apoptosis5. Innate immunity in the CNS, and specifically microglia and astrocytes, are becoming increasingly implicated in neuropathology of CNS diseases due to the exponential increase of knowledge on microglia and astrocyte biology over the past decade. Due to this, there is an increasing need to monitor glial cell behaviour in vivo in patients during disease progression but also in response to therapies. Innate immunity in the brain Innate immunity is the first line of defence in infection, but also plays a key role in tissue repair, clearance of apoptotic cells and cellular debris as well as in response to tumours. While the key innate immune cells in the CNS are microglia and astrocytes, peripheral macrophages as well as oligodendrocytes contribute to innate immune responses in the CNS. Additionally, the immune system in the CNS is unique in that mostly occurs behind several CNS barriers (Figure 1). A B Astrocyte Basal Lamina Pericyte Endothelial cell Interneuron Macrophage T-cell Tight Junction Fenestrated blood vessel Choroid plexus epithelium Ependymal layer CSF Stroma CNS parenchyma Epithelial basal membrane Figure 1. Blood–central nervous system (CNS) barriers. The blood–brain barrier (BBB) (a) and blood–spinal cord barrier (BSCB) limit potential immune cells, antibodies and soluble factors entering the CNS in health. Likewise, while the choroid plexus (CP) also limits cell migration, evidence suggests that regulatory T‐cells enter the brain via the CP (b) during health in order to ensure surveillance of the CNS. CSF, cerebrospinal fluid.
11 General introduction Table 1. Current and predicted incidence of neuroinflammatory diseases Disease (Proposed) Aetiology Innate immune response involvement Adaptive immune response involvement Incidence % or number/ 100000 Predicted change in prevalence Ref MS Autoimmune Viral Microglial and macrophage activation, ↑ROS, complement, ↑innate receptors, ↑cytokines, ↑chemokines ↑HSPs ↑neurotrophins Antibodies / T cells to CNS antigens. 9.64 ↑2.4% per year 6,7 AD, other dementias AD – misfolded and aggregated tau and Abeta Activated and dystrophic microglia, ↑TNF-α, ↑IFN-γ, ↑chemokines, ↑complement, ↑TLRs ↑antibody and T-cell response 9.33% ↑3.3% per year (triple by 2050) 1,3,7 PD Loss of dopaminergic neurons in SN due to α-syn-inclusions ↑TLRs, ↑CD14, activated NK cells, microglial activation, ↑IL-1β, ↑IL-6, ↑TNF-α ↑T-cells ↑antibody response 100-200 double in 25 years 7-9 HD Expansion of CAG (Q) in huntingtin gene induces aberrant toxic protein ↑microglial proliferation, ↑complement, not reported 0.02-9.71 ↑15-20% per decade 7,10 SMA SMN1 gene mutations ↑IL-6, ↑IL-1β not reported 1-2 not reported 7,11 ALS (MND) Aberrant aggregated proteins due to mutations SOD1, TDP; C9orf72 or FUS genes ↑complement, ↑CD14, ↑macrophages, ↑IL-6, ↑TNF-α ↑CD4+ ↑CD8+ T-cells 1.9 ↑69% in 25 years 7,11, 12 Prion diseases Infectious forms of misfolded aggregated forms of prion protein ↑microglial activation, ↑IL-1β, ↑IL-6, ↑complement, ↑ROS, mast cells expressing PrP B cells aid transport of PrP variable not reported 7 Stroke Ischemia or Haemorrhage ↑lymphopenia, ↑NK cells, ↑IL-10 ↑Th2 responses 115 ↑44% in 20 years 13,14 TBI Open and head Injury, Deceleration Injuries, Chemical/Toxic, Hypoxia, Tumors, ↑pro-inflammatory cytokines, ↑TLRs ↑T-cells ↑B-cells 295 not reported 7,15 HIV/AIDS HIV encephalopathy, toxoplasmosis, PML ↓IL-27, ↓IFN-γ, ↓CD4 cells, ↑IL-4 ↓T-cells, immunesenescence 0.8% depending on country 16 Meningitis Bacterial and viral infections ↑IL-6, ↑TNF-α, ↑NK cells, ↑microglial activation ↑T-cells ↑B-cells 0.2-1000 outbreak dependent 17,18 Ageing Natural event ↑proinflammatory cytokines, ↓NK cell function ↓T-cells n/a n/a 19 Epilepsy Unprovoked seizures, febrile events, autoantibodies ↑proinflammatory cytokines, ↑chemokines, ↑TLRs, ↑complement ↑autoantibodies T and B-cell activation 45-81.7 ↑ 20-22
12 Chapter 1 Table 1. Current and predicted incidence of neuroinflammatory diseases (continued) Disease (Proposed) Aetiology Innate immune response involvement Adaptive immune response involvement Incidence % or number/ 100000 Predicted change in prevalence Ref Autism Genetic / Environmental ↑proinflammatory cytokines ↓T-cells 425-760 variable 23-25 Depression Multifactorial e.g. genetics, hormonal Microglial activation, ↑ cytokines, ↑chemokines ↑ T-reg cells 3% ↑ 26-28 Schizophrenia Multifactorial Microglial activation, ↑ROS, ↑proinflammatory cytokines, ↑chemokines, ↑TLRs, ↓NK cells not reported 18.5 not reported 29-31 Bipolar disorder Genetic / Environmental Microglial activation, ↑proinflammatory cytokines, ↑complement, ↑TNF-α ↑T-cell activation 2.4% debated 32-34 Abbreviations: AD, Alzheimer’s disease; ALS, amyotrophic lateral sclerosis; APP, amyloid-β precursor protein; CD, cluster of differentiation; CNS, central nervous system; FUS, Fused in Sarcoma; HD, Huntington’s disease; HIV/AIDS, human immunodeficiency virus/acquired immunodeficiency syndrome; HSP, heat shock protein; IFN, interferon, IL, interleukin; MS, multiple sclerosis; NK, natural killer; PD, Parkinson’s disease; PML, progressive multifocal leukoencephalopathy; ROS, reactive oxygen species; SMA, spinal muscular atrophy; SOD1, superoxide dismutase 1; SMN, Survival of Motor Neuron protein; SN, substantia nigra; TBI, traumatic brain injury; TDP, TAR DNA-binding protein 43 kDa; TLR, toll-like receptor; TNF, tumor necrosis factor. Neuroimmune privilege and CNS barriers The concept of immune privilege originated from Sir Peter Medawar’s studies in the mid20th century showing that tissue grafts in the CNS were not rejected. It also takes into account the presence of the BBB which was revealed by Paul Ehrlich’s studies in the late 1800s showing that solutes and molecules were excluded from the brain. However, it is now clear that entry of compounds into the CNS occurs via capillary venules, while cell migration occurs at the post-capillary venules and is controlled by adhesion molecules, cytokines, and chemokines35. Anatomically, the CNS is separated by three barriers: the BBB/blood spinal cord barrier (BSCB), the blood-cerebrospinal fluid barrier (BCSFB) at the choroid plexus (CP), and the arachnoid barrier. Differences in the structure of the BBB and BSCB, as well as differences in the cranial and spinal meninges, in white and grey matter, and other regional differences may explain the differential susceptibility of anatomical regions to neuroinflammatory events. For example, the BSCB has reduced levels of ZO-1, occludin, VE cadherin and P-gp (p-glycoprotein), and fewer pericytes than the BBB36, indicating that the spinal cord may well be more susceptible to inflammatory insults than the brain. The presence of barriers originally explained why CNS antigens in the brain were ignored by the peripheral immune response. However, this dogma has been challenged recently by the identification of the glymphatic system37 and rediscovery of lymphatic vessels in the dura mater38,39 that are crucial to clear waste products such as Aβ peptides and tissue debris that accumulate during disease. Dysfunction of these barriers is well known to occur in neuroinflammatory disorders including MS, Parkinson’s disease (PD), Alzheimer’s disease (AD), stroke, epilepsy and traumatic brain injury (TBI)40 and is associated with activated endothelial cells that display an altered pheno-
13 General introduction type and a decrease in tight junction proteins. These changes that are also observed during ageing41 may explain the increase in susceptibility to neuroinflammation and neurodegenerative disorders in the elderly. Microglia and macrophages Historically microglia have been described as the phagocytes of the CNS, however over the last decades microglia are increasingly implicated in exerting diverse roles. Microglia comprise roughly 10% of the glial population in the CNS but unlike astrocytes or oligodendrocytes, microglia are derived from the yolk sac from which they migrate and populate the brain during embryonic development42. Microglia are highly motile and versatile cells, they self-renew, through proliferation, and upon damage or injury can increase their population by clonal expansion43. This clonal expansion is directed by cues from the micro-environment and results in microglia numbers that are dependent on the local need43. Activated microglia are often seen as drivers for neuropathology44, however, loss of vital homeostatic functions of microglia has also been linked with leukodystrophies, characterised by progressive myelin damage. Initially, microglia were classified as being either classically activated (M1) or alternatively activated (M2), based on chemokine and cytokine expression in vivo45. More recently the field has switched to account for a plethora of microglial states recognising the microglial heterogeneity that is present within the brain parenchyma in health and disease. Several reports indicate that microglia have different transcriptomic profiles dependent on the region of the brain, the stage of development and age46, neuropathological state47, and the microbiome48 (Figure 2). NEURODEGENERATION INFLAMMATION HOMEOSTASIS AGEING DEVELOPMENT REGION MYELINATION Switching between microglial states is vital for processes such as remyelination and is affected by ageing49. Microglia secrete pro-inflammatory as well as anti-inflammatory factors which can either be beneficial or detrimental in neurodegenerative diseases50, for example microglia depletion in a mouse model of AD reduced neuronal loss without affecting β-amyloid pathology51. Additionally, microglia-derived factors such as blood derived neurotrophic Figure 2. Kaleidoscope of microglia heterogeneity. Microglia are highly heterogeneous and can adapt to any function that is needed for the local environment which is different in health when it is compared to neuroinflammation, remyelination, as well as neurodegeneration. Additionally, microglia are influenced by extrinsic factors such as ageing, the stage of development, or the region where they reside.
14 Chapter 1 factor (BDNF) are important for learning and memory processes, a process that can be affected by maternal inflammation leading to disrupted behaviour and learning in later life52. On the other hand, monocytes are the blood-borne precursors to macrophages and dendritic cells, and play a key role in innate immunity although their distinct roles in CNS disorders are frequently hard to distinguish from microglia. The novel markers TMEM119 and P2Y12 have helped differentiate microglia and macrophages53, indicating that the relative contribution of these cells to neuroinflammatory diseases can be examined. However, microglia have been found to downregulate both TMEM119, and P2RY12 upon activation53, complicating the picture. A more recent study has found the enzyme HexB has high microglial specificity, during both health and disease54, a finding that may better clarify their role in health and disease in the future. Astrocytes While astrocytes were originally viewed as supportive cells for neurons, it is now clear that astrocytes perform a broad array of physiological and immunological functions in the CNS55-59. Similar to the M1/M2 polarization of macrophages and microglia, subpopulations of astrocytes have been reported that produce proinflammatory mediators (A1) and immunoregulatory mediators (A2). The A1 astrocytes that secrete IL-1a, TNF and C1q are considered to be neuroinflammatory, and damage neurons and oligodendrocytes in vitro as well as inducing apoptosis, suppress T helper cell activation, proliferation and function of activated T-cells, while in contrast, A2 astrocytes are neuroprotective, promoting neuronal growth, survival, and synaptic repair60. Astrocytes respond to a plethora of insults and are frequently observed as hypertrophic in many neurodegenerative diseases including stroke, TBI, MS, amyotrophic lateral sclerosis (ALS) and viral infections and other inflammatory conditions60. A1 reactive astrocytes have been suggested as having toxic effects in ALS, AD, MS, Parkinson’s disease (PD), Huntington’s disease (HD), schizophrenia and ageing60,61; whilst synapse-promoting A2 astrocytes may be responsible for unwanted synapses in epilepsy and neuropathic pain62. However, similar to recent advances in the microglial field, astrocytes should be considered to be more heterogeneous than A1 and A2. Multiple subtypes of astrocytes were identified using single cell RNA sequencing; each with their own distinct gene enrichment profile and physiological functions63. For example, different subtypes were found to contribute differentially to synaptogenesis which might be locally regulated by neuronal activity. More recently, different subtypes were identified that during LPS induced inflammation undergo distinct inflammatory transitions with defined transcriptomic profiles64. There is increasing evidence that astrocytes in the brain are heterogeneous in function depending on the context and time of injury and disease, similar to microglia (Figure 2), and that astrocytes interact with microglia and oligodendrocytes to exert immunological functions. Oligodendrocytes As well as the classical innate immune cells, i.e. microglia and astrocytes, oligodendrocytes also contribute to innate immune reactions, expressing receptors and producing immunomodulatory cytokines and chemokines. Originally, oligodendrocytes were regarded as bystanders in immunological responses. However, during CNS insults and disease, oligodendrocytes can aid in protection and regenerative processes, but can also contribute to neurodegeneration through poor production or repair of myelin. Expression of MHC Class I on oligodendrocytes was initially labelled as controversial. However, oligodendrocytes upregulate MHC class I expression after IFN-γ treatment65. Additionally, oligodendrocytes express pattern recognition
15 General introduction receptors (PRRs) such as Toll-like receptors (TLRs), as well as a range of receptors for inflammatory mediators such as IL-4, IL-6, IL-7, IL-10, IL-11, IL-12, and IL-1866-70. As microglia are the primary immune cells of the CNS, cross-talk between oligodendrocytes and microglia is a key area of interest in many CNS diseases71. Indeed, when oligodendrocytes are stressed they may be triggered to produce CXCL10, CCL2, and CCL3 to attract microglia to the area of damage72,73. Stressed oligodendrocytes upregulate HSPB5 (also known as αB-crystallin), a molecule that is reported to activate microglia74, is involved in immunoregulatory functions, and reduces clinical symptoms and tissue damage75. While astrocytes are increasingly implicated in being involved in immune functions the cross talk between oligodendrocytes and astrocytes is a relatively unexplored field. Imaging and monitoring neuroinflammation Over the last decades there have been significant advances in imaging techniques to visualise the brain during health and disease. One of the main challenges with developing current therapies for neurodegenerative and neuroinflammatory diseases is to monitor the efficacy and impact of the drug on the pathological processes in the CNS. Biomarkers of neuroinflammation and innate immune processes are considered essential for monitoring disease diagnosis, progression, and response to therapy, however there is a lack of accurate and reliable biomarkers for many neurological diseases76. Generally, blood and CSF are commonly used to monitor neuroinflammation, however in vivo imaging the CNS during disease has become a more widely accepted approach due to its ability to provide region-specific information as well as being minimally invasive depending on the technique. Such techniques include magnetic resonance imaging (MRI), positron emission tomography (PET), single photon emission computed tomography (SPECT), and optical imaging85. These approaches allow the study of some aspects of neuroinflammation: a) monitoring activation of resident CNS immune cells e.g. microglia activation, b) BBB permeability e.g. upregulation of adhesion molecules, c) CNS infiltration of immune cells and, d) pathology as a result of neuroinflammation e.g. demyelination and cell death (Table 2). In addition, aspects of BBB integrity, regarded as a hallmark of neuroinflammation, is imaged by leakage of gadolinium using MRI, or by nuclear imaging of P-gp and vascular cell adhesion molecule (VCAM-1), which are differentially expressed in MS86, stroke87, AD and vascular dementia88. Indicators of leukocyte function include markers of oxidative stress, such as proinflammatory and oxidative enzymes secreted by activated monocytes and neutrophils. One such product is myeloperoxidase (MPO) that can be detected by gadolinium (MPO-Gd) to track the oxidative activity of MPO non-invasively. Thus MPO has been used as a potential biomarker of neuroinflammation in experimental models of MS, namely experimental autoimmune encephalitis (EAE)89, and experimental stroke90. Cell-labelling approaches include radiolabeled antibodies and radiolabelled cytokines, which are imaged using SPECT, PET or optical imaging. For example, anti-CD3, anti-CD4, IL-1 and IL-2 have all been used to visualise receptors on T-lymphocytes in MS91 and rheumatoid arthritis92. As well as ongoing neuroinflammation, several approaches image the resultant pathology. As an example, PET ligands have been used to visualise myelin damage in MS93,94 while many approaches are used to visualise cell death e.g. neuronal loss, such as annexin-V, caspases and ML-10 85. Imaging of neuro-inflammatory biomarkers is an expanding topic with the potential to expedite diagnosis and improve disease and therapeutic monitoring. While many approaches are examined in preclinical models, fewer are available for studies in humans76.
16 Chapter 1 Table 2. Biomarkers and Imaging of Neuroinflammatory Diseases Target type Target Marker Methods Ref Resident CNS cells TSPO Innate immune activation PET, SPECT 76 Monoamine oxidase-B Reactive astrocytes PET 76 Cyclooxygenase 1 Activated microglia and astrocytes PET 76 MPO Inflammatory mediator in leukocytes MRI, PET 77 Adenosine receptors Cell injury PET 78 A4B2 nicotinic acetylcholine receptors Activated microglia and astrocytes PET 76 Myo-inositol Astrocyte hypertrophy MRS 76 N-acetyl-aspartate Neuronal integrity MRS 76 Iron accumulation Free radical formation, mitochondrial or neuronal dysfunction MRI 79 Myelin Demyelination and loss of myelin integrity in white matter disorders PET 76 BBB integrity VCAM Activation BBB Molecular imaging 76 P-gp Alterations of expression in relation to BBB activity PET, optical imaging 76 Immune markers Cytokines Pro or anti-inflammatory signals CSF 80 Chemokines Pro or anti-inflammatory signals CSF 80 SPIO SPIO-labelled phagocytic cells MRI 81 Antibodies Oligoclonal bands IgG of unknown specificity CSF 82 Anti-aquaporin 4 antibodies Antibodies to aquaporin 4 (water channel protein) Blood 82 Anti-NF antibodies Neuronal damage Blood 82 Free proteins Neurofilaments Neuronal damage CSF 82 microRNAs Circulating microRNAs involved in inflammation Blood 82 β-amyloid Proteins involved in disease pathology Blood 83 Tau Proteins involved in disease pathology Blood 84 Annexin V Apoptosis PET, SPECT, blood 76 Exosomes A potential mechanism by which pathology is spread and/or toxic proteins are transported CSF/blood 84 Abbreviations: BBB, blood brain barrier; CNS, central nervous system, CSF, cerebrospinal fluid; MRI, magnetic resonance imaging; MPO, myeloperoxidase; PET, positron emission tomography; Pgp, P-glycoprotein; SPECT, single-photon emission computed tomography; SPIO, superparamagnetic particles of iron oxide; TSPO, translocator protein; VCAM, vascular cell adhesion molecule 1.
17 General introduction TSPO Oneproteinthathasbeenof interest as the target inPET imagingof ongoingneuroinflammation is the translocator protein (TSPO). TSPO is an outer mitochondrial membrane protein, expressed inmany tissues in the body, whose exact functions are unknown95-99. TSPOPET signal is markedly upregulated in many neurodegenerative and neuroinflammatory diseases95-111. However, one of the caveats of using TSPO PET is that the exact cellular origin of the TSPO PET signal in the brain is unclear. Knowing the origin of TSPO PET is necessary for clinical meaningful decisions, e.g. whether a therapeutic intervention is having the desired in vivo effects on neuroinflammation. Due to the fast pace of research and development of new TSPO radiotracers many conclusions of TSPO PET have been made without proper investigation into the cellular origin100. This has led to the belief that the TSPO PET signal mostly originates in activatedmicroglia99,105,106,109,112-124. The contribution of other cell types such as astrocytes have been largely ignored, even though there have been multiple reports of TSPO in astrocytes for many years125-141. Most of these studies have demonstrated astrocytic TSPO in animal models of CNS diseases125-138, however only a few studies have investigated expression of TSPO in astrocytes in the human CNS, most of which are qualitative rather than quantitative139-142. Additionally, human microglia do not upregulate expression of TSPO after pro-inflammatory stimulation, and human macrophages even downregulate TSPO expression143. This recent finding has sparked interest into whether TSPO is a marker of all microglia or a readout on microglial density rather than activated microglia. The use of TSPO PET also allows us to investigate the efficacy of newly developed drugs on neuroinflammation in vivo in a preclinical setting in experimental animal models. But similar to the human conundrum, we need to know what cells express TSPO (e.g. microglia or astrocytes). Of similar importance is to find out whether the TSPO gene is regulated in a similar manner in animal models compared to humans. In contrast to the human microglia and macrophages, mouse microglia and macrophages upregulate TSPO after pro-inflammatory stimulation up to 10-fold, indicating that there are differences in the regulation of the TSPO gene between humans and mice143147. This questions whether biological processes in animal and experimental models of human diseases truly reflect the pathological processes occurring in humans148-150. Due to the limitations of TSPO, other candidate markers for PET imaging of microglia have been suggested, such as cyclooxygenase-2 (COX-2), cannabinoid receptor type 2 (CB2R), purinergic ion channel receptor 7 (P2X7R). However, the radioligands for these targets each come with their own obstacles in terms of radioligand binding, sensitivity and specificity of microglia. Multiple sclerosis Multiple sclerosis (MS) is a chronic inflammatory demyelinating disease of the CNS, and is the most common disabling disease affecting young adults. MS most often presents itself between 20 and 40 years of age, but may occur in younger and older people. The first report of a disease with characteristics showing similar transient neurological deficits goes back as far as the 14th century, originating in the Netherlands. Despite decades of research into the aetiology of MS, the exact cause of MS is unknown, there is however evidence that the incidence of MS increases based on several environmental risk factors including the EpsteinBarr virus (EBV)151. EBV is present in nearly all people with MS (pwMS, > 99%), however roughly 95% of the general population has had an EBV infection during their lifetime152. Nevertheless, the risk of developing MS increases with EBV antibody titer153, and is linked
18 Chapter 1 to infectious mononucleosis, i.e. EBV infection in childhood and young adulthood154. EBV interacts with complement C2 receptors on memory B cells to infect and subsequently immortalise them155-157. There are reports that EBV can infect specific memory B-cells to produce anti-MOG (myelin oligodendrocyte glycoprotein) auto-antibodies158. These autoantibodies induce upregulation of heat shock protein B5 (HSPB5) in oligodendrocytes, which activates microglia159, but also activates HSPB5-reactive memory T-cells160. When these T-cells are activated they in turn start producing IFNγ, a pro-inflammatory cytokine with detrimental effects in MS159, possibly initiating the development of demyelinating lesions in MS. The latitudinal effect on risk of MS, is mainly attributed to sunlight exposure, as well as vitamin D production161. High vitamin D consumption in Norway, or through supplementation in the US military have protective effects against MS162,163. Additionally, there is a genetic risk to developing MS. Risk of developing MS increases 10-25 times for first-degree relatives, compared to the general population, where monozygotic twins have the highest risk if either one twin develops MS164,165. Risk of developing MS has also been linked to human leukocyte antigen (HLA) and is increased in people expressing HLA-DR*1501 common in northern Europeans. This is also supported by an increased risk people homozygous for HLA-DR1501 compared to heterozygous166. Additionally, regional variation revealed many haplotypes of the HLA gene that have either protective or detrimental effects on the risk of developing MS. These HLA haplotypes differ in magnitude of effect which and can either interact with each other or act on their own167. Clinical symptoms The clinical presentation of MS is heterogeneous depending on the presence of lesions and extent of damage affecting the brain but also the spinal cord and optic nerve. The variety in clinical presentation is a consequence of the transient nature of focal attacks of neuroinflammatory lesions in the CNS. The location and timing of the lesions can result in clinical symptoms ranging from visual problems, if lesions arise in the optic nerve, to motor problems, when lesions arise in the motor areas. The first symptoms of MS often include signs of visual problems such as diplopia and optic neuritis, but also sensory disorders and limb weakness. During the early stages of the disease when neuronal reserve is high, pwMS may fully recover from their clinical attacks, recognised as relapsing-remitting forms of the disease (RRMS). Over time, this reserve and ability to repair the damage in the CNS is lost and pwMS develop a more progressive form of the disease: secondary progressive MS (SPMS). SPMS develops in 90% of people with RRMS over the course of 10-15 years after onset. However, up to 10% of people don’t recover from their clinical attacks and have a primary progressive disease course (PPMS) with increasing disease disability. The diagnosis of MS is made according the McDonald criteria168 (Table 3). The McDonald criteria have been extensively revised due to increasing knowledge of the pathology of the disease, and the incorporation of rapidly advances in development of imaging modalities. However, the key requirement of a diagnosis according to the criteria is that people have experienced damage of the CNS that is disseminated in time and space, meaning that there is evidence of CNS damage at 2 or more time points and different areas of the CNS. These criteria are supported by the presence of oligoclonal bands in the CSF.
19 General introduction Table 3. 2017 McDonald Criteria for diagnosis of Multiple Sclerosis Clinical presentation What additional data is needed for anMS diagnosis? Two or more relapses AND EITHER objective clinical evidence of two or more lesions OR objective clinical evidence of one lesion together with reasonable historical evidence of a previous relapse None Two or more relapses; objective clinical evidence of one lesion (shows DIT) Dissemination in space shown by: One or more MRI detected lesions typical of MS OR A further relapse showing damage to another part of the CNS One relapse; objective clinical evidence of two or more lesions (shows DIS) Dissemination in time shown by: Oligoclonal bands OR MRI evidence of a new lesion since a previous scan OR A further relapse One attack/relapse; objective clinical evidence of one lesion (known as ‘clinically isolated syndrome’) Dissemination in space shown by: One or more MRI detected lesions typical of MS OR A further relapse showing activity in another part of the CNS Dissemination in time shown by: Oligoclonal bands OR MRI showing new lesions since a previous scan OR A further relapse Insidious neurological progression suggestive of multiple sclerosis (typical for primary progressive MS) Continued progression for one year (from previous symptoms or by ongoing observation) plus any two of: One or more MRI detected lesions in the brain typical of MS, Two or more MRI detected lesions in the spinal cord, Oligoclonal bands in the spinal fluid Abbreviations: CNS, central nervous system; DIS, dissemination in space; DIT, dissemination in time; MRI, magnetic resonance imaging; MS, multiple sclerosis. Neuropathology The pathology of MS is primarily based on the presence of microglia and macrophages, the loss of myelin, the degree of astrogliosis and neuronal and synaptic loss (Figure 3). Histopathological, the lesions are classified as preactive, active, chronic active, or inactive. Upon histopathological examination, preactive lesions are classified as clustering microglia in the normal appearing white matter (NAWM). In the NAWM there is no myelin damage, no astrogliosis and only rarely foamy macrophages are observed169. The preactive lesion is a relatively new description in neuropathology and the term suggests it could be the start of the classical active lesions that are commonly described. However, the discrepancy between number of preactive lesions and the relatively lower number of active lesions suggest that a significant amount of preactive lesions do not develop into active lesions. Active lesions are characterised by myelin loss and damage, the presence of foamy macrophages and microglia, periventricular infiltrates filled with cells of the peripheral immune system, mainly lymphocytes169. Chronic active lesions are characterised by a rim of foamy macrophages around the area of demyelination, the centre of chronic active lesions shows signs of astrogliosis and hypertrophy and is hypocellular169. Inactive lesions show signs of severe myelin damage, hypertrophic astrocytes and is characterised by hypocellularity169. In inactive
20 Chapter 1 Inac�ve Ac�ve Preac�ve Schema�c PLP HLA-DR Chronic Ac�ve (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) (k) (l) Figure 3. Characterisation of white matter MS lesions. MS lesions are characterised mainly on the presence of activated microglia/macrophages, marked by HLA-DR staining (i-l), demarcated by the dotted lines, and the preence of myelin, marked by PLP staining (e-h). Pre-active lesions (a,e,i) are microglial clusters (arrow) in the normal appearing white matter (NAWM). Active lesions (b,f,j) are characterised on absence of myelin and the presence of high numbers of activated microglia/macrophages. Chronic active lesions (c,g,k) have a demarcated rim of activated microglia/macrophages around a centre of inactivity marked by loss of myelin. In inactive lesions (d,h,l) activity subsides and the area is left with a plaque with no myelin, and astroglial scarring. lesions there are almost no signs of inflammatory cells. Then there is some evidence that some lesions show signs of remyelination where oligodendrocyte precursor cells remyelinate naked axons, although the new myelin has other characteristics such as thinner sheaths170. In the grey matter of the brain MS lesions are classified based on their anatomical location rather than on their degree of inflammation due to the lack of apparent neuroinflammation that is present in the cortex of the brain171 (Figure 4). Lesions that comprise both the white and grey matter are classified as leukocortical lesions (type I), sometimes the white matter of these lesions show clear signs of inflammation which gradually decreases when crossing to the grey matter172. Lesions that are purely within the grey matter are classified as intracortical lesions (type II)172. Areas of demyelination along the outer border of the cortex area are classified as subpial lesions (type III) and are the most commonly observed172. Lesions that extend from the surface of the brain to the border of the white and grey matter are classified as transcortical (type IV)2,172. I Grey ma�er White ma�er II III IV Figure 4. Grey matter lesions in MS. Lesions are classified by their location and spread in the grey matter. Type I comprises both white and grey matter. Type II lesions are surrounded by grey matter. Type III lesions are subpially located along the grey matter. Type IV lesions comprise the whole width of the grey matter but do not cross into the white matter2.
21 General introduction Neuroinflammation in neurodegenerative diseases Neuroinflammation is an apparent feature of MS, however, neuroinflammation is also present in other CNS diseases (Figure 5), as well as their experimental models, albeit in different forms and gradations. For example, while ALS is primarily a neurodegenerative disease characterised by motor neuron loss, neuroinflammation plays a critical role in the degeneration of neurons in the brainstem, spinal cord and motor cortex173. Recently, it was shown that 1573 out of 2637 of genes related to inflammation were differentially expressed compared to controls in themotor cortex of people with ALS. Themost dysregulated signalling pathways are involved in antigen presentation, the complement system, and reactive oxygen species (ROS) production174. Microglia activation is correlated with disease severity in the spinal cord of ALS patients175. Additionally, one of the genes that has been implicated in the aetiology of ALS, C9orf72, is highly expressed in myeloid cells. Loss of C9orf functions results in lysosomal trafficking defects, and a reduced ability of microglia to clear cellular debris and aggregated proteins and aberrant microglia responses176. Furthermore, astrocytes are also affected in ALS showing increased expression of small heat shock protein to protect the cell against damage by refolding or promoting degradation of misfolded proteins175,177. TSPO PET studies of ALS have indicated that increased signal was caused by activated microglia while ignoring the contribution of other cell types such as astrocytes106,178-180. Over the last years, increasing insights into immunological processes in the CNS have shed light on the role of neuroinflammation in AD. The misfolded and aggregated proteins, characteristic of AD pathology in the form of Aβ plaques and neurofibrillary tangles, bind to PRRs on microglia and astrocytes, activating innate immune system cascades181. Microglia and astrocytes start engulfing Aβ fibrils, which is mostly resistant to degradation resulting in inefficient clearance of Aβ182,183. As a result, astrocytes around plaques are reactive but do not form gliotic scars. It has been suggested that astrocyte activation may even happen before microglia involvement in AD pathogenesis, where astrocyte atrophy leads to aberrant synaptic functioning resulting in cognitive deficits184. One of the most affected areas in AD is the hippocampus and several PET studies have found increased TSPO binding in the hippocampus101,185,186. However, some studies have not found differences in TSPO binding, complicating the use of TSPO PET to identify neuroinflammation in AD187-190. On the protein level, increased expression of TSPO is present in the cortex, but no studies have investigated TSPO expression in the hippocampus or have looked at the microglial states that were responsible for the increased expression191. Experimental animal models of CNS disease MS related pathology such as inflammation, neurodegeneration and demyelination is most commonly studied in EAE. EAE is a T-cell mediated disease induced by immunising mice or rats with CNS antigens. However, the clinical severity of EAE is highly dependent on the antigen, species, age, gender and strain of animals, but also the method of induction. All these factors can influence immunological processes as well as the response to therapies tested in EAE mice192 which unfortunately has resulted in a very low translation of therapeutic compounds for MS. For ALS, themost commonly used experimental model is the SOD1G93A mouse, as it was one of the first mutations discovered to be associated with ALS193. Whilst the SOD1G93A model replicates a moderate disease duration with misfolded SOD1, motor neuron loss, metabolic deficits and gliosis, most therapeutic compounds have failed to translate to humans148.
22 Chapter 1 (a) (b) (f ) (l) WML WM GML GM (c) (g) (m) GM WM (j) (k) (e) (o) WML WM GML GM (h) (i) (d) (n) Figure 5. Immune responses in human and experimental inflammatory neurodegenerative disorders. B‐cells (arrows) are observed in white (a) and grey matter lesions (b) in multiple sclerosis (MS). (c) and (d) depict an MS leucocortical lesion. The white matter (WML) is associated with HLA+ microglia (d, WML) in contrast to the lack of HLA + microglia in the grey matter (d, GML). A similar pattern of HLA+ cells is seen in the white and grey matter in an X‐ALD case (e) and where peripheral macrophages infiltrate the white matter (f). Granulocytes (arrow) in suspected vasculitis cases (g). Ageing influences the activity of microglia in a mouse model of MS: microglia in the CNS of young mice (h; Iba1 staining) are less active than in aged mice (i). In MS cases microglia in NAWM express P2Y12 (j) and TMEM119 (k). In progressive multifocal leucoencephalopathy astrocytes (l, arrow) and activated microglia/ macrophages (m, arrow) are highly reactive in an area of demyelination. The paucity of astrocytic glial fibrillary acidic protein expression (red circle, n) is associated with an area of microglial activation (red circle, o) in acute haemorrhagic leucoencephalitis. This is likely at least partly explained by the low representation (1-2%) of the mutation in people with ALS. Animal models for AD most commonly represent either amyloid pathology by increasing Aβ, or increasing the relative ratio of Aβ42 which is prone to aggregation or tau pathology by overexpressing the microtubule-associated protein tau194. Again, both models have relatively poor translational value when it comes to disease modifying therapies as over the last decades almost no positive clinical results have been yielded. Although experimental animal models are often used to study CNS pathology they rarely reflect the multifactorial complexity of human diseases. Failure of translation of disease modifying therapies may be attributed to varying study designs, imperfect animal models or timing of therapy. Nevertheless, TSPO PET is utilised to monitor ongoing neuroinflammation in vivo in experimental animals. Indeed, increases in TSPO PET signal are found in EAE, SOD1G93A, and APP and TAUP301S experimental animal models141,195-197. However, there is no consensus on whether this increase in TSPO is due to increased microglial density, an increase in TSPO expression, or an increase in ligand binding or whether this is disease, model or contextdependent.
23 General introduction Aim of thesis Advancing techniques and increasing knowledge on cellular processes during neuroinflammation have contributed towards generating many targets for imaging neuroinflammation with PET. The aim of this thesis is to better understand the expression of TSPO in CNS diseases and respective animal models as well as investigating the role of microglia and astrocytes as innate immune cells of the brain in CNS diseases. Hypothesis The hypothesis tested in this thesis is that TSPO is not a marker of activated microglia in the human brain. We hypothesise that TSPO is not increased in activated microglia and there is a substantial contribution of other cell types to the TSPO PET signal. While TSPO PET might be a good marker for innate immune processes, increased TSPO PET signal might originate from multiple cell types in the CNS, and/or reflect cell density rather than phenotype of TSPO expressing cells. Outline Increased TSPO PET signal is widely attributed to activated pathogenic microglia in CNS diseases and the contributions of other cells, as well as the cell phenotype has been overlooked in many neuroinflammatory diseases. On this basis, the identification of the cellular expression of TSPO in MS in microglia, astrocytes and other cell types as well as their activation state was considered key to understanding the data arising from TSPO PET imaging in MS (Chapter 2). Studies of multiple sclerosis (MS) lesions reveal that TSPO is not restricted to pro-inflammatory microglia/macrophages, but also present in homeostatic or reparative microglia. Here, we investigated quantitative relationships between TSPO expression and microglia/macrophage phenotypes in white matter and lesions of brains with MS pathology (Chapter 3). Since TSPO PET is increasingly used as a marker for neuroinflammation in the CNS, characterisation of TSPO for PET in neurodegenerative diseases with inflammatory components, such as ALS and AD, is of importance. Additionally, many studies are done on animal models for CNS diseases but results do not translate well to humans. To identify possible phylogenetic diversity between humans and rodents in the expression and regulation of TSPO we directly compared ALS, and AD with their respective animal models, as well as EAE (Chapter 4). TSPO expression is altered in many neurodegenerative, neuroinflammatory, and neuropsychiatric diseases. In PET studies, the TSPO signal is often viewed as a marker of microglial cell activity. However, there is little evidence in support of a microglia-specific TSPO expression. Therefore, we described the cellular sources and functions of TSPO in animal models of disease and human studies, in health, and in CNS diseases (Chapter 5). Clinical disability of pwMS is considered primarily a result of axonal loss. Given the reported correlation between indices of spinal cord cross-sectional area (CSA) and disability, and earlier evidence suggesting axonal loss as the key driver of chronic disability. We have compared the extent of axonal loss and whether this correlates with local neuronal loss in the spinal cord of pwMS (Chapter 6).
24 Chapter 1 Over the last decade knowledge of the role of astrocytes in CNS neuroinflammatory diseases has changed dramatically. Rather than playing a merely passive role in response to damage it is clear that astrocytes are increasingly implicated in exerting immunological relevant functions within the CNS. Whilst most astrocyte research focuses on modulating neuronal function and synaptic transmission little is known about the cross-talk between astrocytes and oligodendrocytes, the myelinating cells of the CNS. We thus reviewed evidence for the immunological roles astrocytes and oligodendrocytes play and how the two cell types interact (Chapter 7). As primary innate immune cells in the CNS, microglia play critical roles in shaping the brain during development, responding to pathogens, and clearing tissue debris or protein aggregations during ageing, neuroinflammation and neurodegeneration. The last decade has given us many new insights in the heterogeneity of microglia states. However, many of the single cell/nucleus RNA-seq studies have focused on microglia in the grey matter or of the whole brain while ignoring the heterogeneity of microglia in the white matter. Therefore, we provide an update on the current knowledge of microglia heterogeneity in the white matter and how microglia are important for development of the white matter as well as how ageing affects white matter homeostasis (Chapter 8). To conclude this thesis, the results of the previous chapters are summarised. The data in this thesis is discussed and put into context, which enables us to give an overview on the current knowledge of innate immune cells in the CNS, and how to monitor these cells in vivo during neuroinflammation (Chapter 9).
25 General introduction 1. Wimo A, et al. The worldwide costs of dementia 2015 and comparisons with 2010. Alzheimers Dement. Jan 2017;13(1):1-7. doi:10.1016/j. jalz.2016.07.150 2. Bo L, et al. Grey matter pathology in multiple sclerosis. Acta Neurol Scand Suppl. 2006;183:4850. doi:10.1111/j.1600-0404.2006.00615.x 3. Cova I, et al. Worldwide trends in the prevalence of dementia. J Neurol Sci. Aug 15 2017;379:259-260. doi:10.1016/j.jns.2017.06.030 4. Perry VH. The influence of systemic inflammation on inflammation in the brain: implications for chronic neurodegenerative disease. Brain Behav Immun. Sep 2004;18(5):407-13. doi:10.1016/j. bbi.2004.01.004 5. Hickman S, et al. Microglia in neurodegeneration. Nat Neurosci. Oct 2018;21(10):1359-1369. doi:10.1038/s41593-018-0242-x 6. McKenzie IA, et al. Motor skill learning requires active central myelination. Science. Oct 17 2014;346(6207):318-22. doi:10.1126/ science.1254960 7. Amor S, et al. Inflammation in neurodegenerative diseases–an update. Immunology. 2014;142(2):151-166. 8. Myall DJ, et al. Parkinson’s in the oldest old: Impact on estimates of future disease burden. Parkinsonism Relat Disord. Sep 2017;42:78-84. doi:10.1016/j.parkreldis.2017.06.018 9. Tysnes OB, et al. Epidemiology of Parkinson’s disease. J Neural Transm (Vienna). Aug 2017;124(8):901-905. doi:10.1007/s00702-0171686-y 10. Rawlins MD, et al. The Prevalence of Huntington’s Disease. Neuroepidemiology. 2016;46(2):144-53. doi:10.1159/000443738 11. Verhaart IEC, et al. Prevalence, incidence and carrier frequency of 5q-linked spinal muscular atrophy- a literature review. Orphanet J Rare Dis. Jul 4 2017;12(1):124. doi:10.1186/s13023-017-0671-8 12. Arthur KC, et al. Projected increase in amyotrophic lateral sclerosis from 2015 to 2040. Nat Commun. Aug 11 2016;7:12408. doi:10.1038/ncomms12408 13. Mohan KM, et al. Risk and cumulative risk of stroke recurrence: a systematic review and meta-analysis. Stroke. May 2011;42(5):1489-94. doi:10.1161/ STROKEAHA.110.602615 14. Wong CH, et al. Prolonged Activation of Invariant Natural Killer T Cells and TH2-Skewed Immunity in Stroke Patients. Front Neurol. 2017/01/19 2017;8:6. doi:10.3389/fneur.2017.00006 15. Nguyen R, et al. The International Incidence of Traumatic Brain Injury: A Systematic Review and Meta-Analysis. Can J Neurol Sci. Nov 2016;43(6):774-785. doi:10.1017/cjn.2016.290 16. Zheng YH, et al. The Role of IL-27 and its Receptor in the Pathogenesis of HIV/AIDS and Anti-viral Immune Response. Curr HIV Res. 2017/08/25 2017;15(4):279-284. doi:10.2174/157016 2X15666170517130339 17. Harrison LH, et al. Global epidemiology of meningococcal disease. Vaccine. Jun 24 2009;27 Suppl 2:B51-63. doi:10.1016/j.vaccine.2009.04.063 18. Ribes S, et al. The Early Adaptive Immune Response in the Pathophysiological Process of Pneumococcal Meningitis. J Infect Dis. Jan 1 2017;215(1):150-158. doi:10.1093/infdis/jiw517 19. Deeks SG. HIV infection, inflammation, immunosenescence, and aging. Annu Rev Med. 2011/02/18 2011;62(1):141-55. doi:10.1146/ annurev-med-042909-093756 20. Amor S, et al. Imaging immunological processes from blood to brain in amyotrophic lateral sclerosis. Clin Exp Immunol. Dec 2021;206(3):301-313. doi:10.1111/cei.13660 21. Ngugi AK, et al. Incidence of epilepsy: a systematic review and meta-analysis. Neurology. Sep 6 2011;77(10):1005-12. doi:10.1212/ WNL.0b013e31822cfc90 22. Singh A, et al. The Epidemiology of Global Epilepsy. Neurol Clin. Nov 2016;34(4):837-847. doi:10.1016/j.ncl.2016.06.015 23. Baxter AJ, et al. The epidemiology and global burden of autism spectrum disorders. Psychol Med. Feb 2015;45(3):601-13. doi:10.1017/ S003329171400172X 24. Careaga M, et al. Immune Endophenotypes in Children With Autism Spectrum Disorder. Biol Psychiatry. Mar 1 2017;81(5):434-441. doi:10.1016/j.biopsych.2015.08.036 25. Scott JG. Environmental Contributions to Autism: Explaining the Rise in Incidence of Autistic Spectrum Disorders. Journal of Enviromental Immunology and Toxicology. 2013;1(2):75. doi:10.7178/jeit.7 26. Ferrari AJ, et al. Global variation in the prevalence and incidence of major depressive disorder: a systematic review of the epidemiological literature. Psychol Med. Mar 2013;43(3):471-81. doi:10.1017/ S0033291712001511 27. Hidaka BH. Depression as a disease of modernity: explanations for increasing prevalence. J Affect Disord. Nov 2012;140(3):205-14. doi:10.1016/j. jad.2011.12.036 28. Hughes MM, et al. Stress-Related Immune Markers in Depression: Implications for Treatment. Int J Neuropsychopharmacol. Jun 2016;19(6):pyw001. doi:10.1093/ijnp/pyw001 29. Burns JK, et al. Income inequality and schizophrenia: increased schizophrenia incidence in countries with high levels of income inequality. Int J Soc Psychiatry. Mar 2014;60(2):185-96. doi:10.1177/0020764013481426 30. Karpinski P, et al. Reduced number of peripheral natural killer cells in schizophrenia but not in bipolar References
www.ridderprint.nlRkJQdWJsaXNoZXIy MTk4NDMw