The human nervous system is the organ of consciousness, cognition, ethics, and behavior; as such, it is the most intricate structure known to exist. More than one-third of the 23,000 genes encoded in the human genome are expressed in the nervous system. Each mature brain is composed of 100 billion neurons, several million miles of axons and dendrites, and >1015 synapses. Neurons exist within a dense parenchyma of multifunctional glial cells that synthesize myelin, preserve homeostasis, and regulate immune responses. Measured against this background of complexity, the achievements of molecular neuroscience have been extraordinary. Advances have occurred in parallel with the development of new enabling technologies—in bioengineering and computational sciences, imaging, and cell, molecular and chemical biology—and moving forward it is likely that the pace of new discoveries will only increase. This chapter reviews a number of the most dynamic areas in neuroscience, specifically highlighting advances in immunology and inflammation, neurodegeneration, and stem cell biology. In each of these areas, recent discoveries are providing context for an understanding of the triggers and mechanisms of disease, and offering new hope for prevention, treatment, and repair of nervous system injuries. Discussions of the neurogenetics of behavior, advances in addiction science, and diseases caused by network dysfunction can be found in Chap. 443 (Biology of Psychiatric Disorders); and new approaches to rehabilitation via harnessing of neuroplasticity, neurostimulation, and computer-brain interfaces are presented in Chap. 477 (Emerging Neurotherapeutic Technologies).
NEUROIMMUNOLOGY AND NEUROINFLAMMATION
OLIGODENDROCYTES AND MYELIN
Myelin is the multilayered insulating substance that surrounds axons and speeds impulse conduction by permitting action potentials to jump between naked regions of axons (nodes of Ranvier) and across myelinated segments. Molecular interactions between the myelin membrane and axon are required to maintain the stability, function, and normal life span of both structures. A single oligodendrocyte usually ensheaths multiple axons in the central nervous system (CNS), whereas in the peripheral nervous system (PNS), each Schwann cell typically myelinates a single axon. Myelin is a lipid-rich material formed by a spiraling process of the membrane of the myelinating cell around the axon, creating multiple membrane bilayers that are tightly apposed (compact myelin) by charged protein interactions. Several inhibitors of axon growth are expressed on the innermost (periaxonal) lamellae of the myelin membrane (see below). A number of clinically important neurologic disorders are caused by inherited mutations in myelin proteins of the CNS or PNS (Chap. 438), and constituents of myelin also have a propensity to be targeted as autoantigens in autoimmune demyelinating disorders (Chap. 436).
Premyelinating oligodendrocyte precursor cells (OPCs) are highly motile cells that migrate extensively during development and in the adult brain following injuries to the myelin sheath. OPCs migrate along the inner (or abluminal) surface of endothelial cells, a process regulated by Wnt pathway signaling and upregulation of the chemokine receptor Cxcr4 that drives their attachment and retention to the vasculature. Initial specification to OPCs is transcriptionally regulated by the Olig 2 and Yin Yang 1 genes, whereas the later stage of myelination mediated by postmitotic oligodendrocytes depends on a different transcription factor, myelin gene regulatory factor (MRF). In the normal adult brain, large numbers of OPCs (expressing PDGFR-α and NG2) are widely distributed but do not myelinate axons, even in demyelinating environments such as in lesions of multiple sclerosis (MS). In addition to Wnt, several families of molecules have been identified that regulate oligodendrocyte differentiation and myelination, including LINGO-1, PSA-NCAM, hyaluronan, Nogo-A, the Wnt pathway, notch signaling (and its receptor Jagged), and the M1 muscarinic receptor Chrm1, all of which are inhibitory, and the retinoic acid receptor RXRγ, which is excitatory. All are also potential targets for myelin repair therapies. In an experimental allergic encephalomyelitis (EAE) model, oligodendrocyte-specific knockout of Chrm1 improved remyelination, protected axons and restored function, directly demonstrating that remyelination can be neuroprotective following injury. A recently reported pivotal trial of a monoclonal antibody against LINGO-1 failed to promote remyelination, a disappointing result given that the antibody appeared to have promising clinical effects in an earlier phase 2 trial.
A series of observations has called into question the traditional concept that axon-derived cues are always required for myelination to occur. Fixed (i.e., dead) axons could be efficiently myelinated by oligodendrocytes in vitro, as could artificial polystyrene nanowires of a similar diameter. This led to development of new high-throughput screening assays based on myelination of polystyrene nanowires to identify compounds that could promote myelination and in a preliminary human trial a molecule that emerged from this assay, the antihistamine clemastine, had clear efficacy as a remyelinating agent in patients with chronic optic neuropathy due to MS. Remarkably, the drug appears to work via binding to the Chrm1 muscarinic receptor.
MACROPHAGES AND MICROGLIA
These represent the major cell types in the nervous system responsible for antigen presentation and innate immunity. Brain microglia migrate from the yolk sac early in embryogenesis before the blood-brain barrier is formed, and are believed to maintain their cell numbers through cell division within the nervous system and not via repopulation from the circulation. Depletion of microglia in adult mice by administration of a selective inhibitor of colony-stimulating factor receptor 1 (CSFR1) was followed by their rapid repopulation, suggesting that a pool of resident microglial precursor cells exist throughout the CNS. Additional roles for brain microglia are known to exist in neurogenesis, through secretion of brain derived neurotrophic factor (BDNF) and other molecules, as well as in the development and regulation of neural circuits through pruning of excitatory synapses and control of dendritic spine densities (Fig. 417-1). Mice depleted of microglia during development exhibit a variety of cognitive, learning and behavioral deficits, including abnormal social behaviors; these processes are dependent on the classical complement pathway molecules and the chemokine receptor CX3CR1. A challenge to the field has been that tools to definitively separate brain microglia from perivascular macrophages do not currently exist. A recent advance that could provide a possible solution utilizes adult skin-derived pluripotent stem cells (iPSCs), and the development of methods to generate microglia-like cells from iPSCs using media containing IL-34 and colony-stimulating factor 1.
The multifunctional microglial cell. Microglia have diverse functions that can support healthy development and maintain homeostasis, or contribute to tissue damage in pathologic conditions. Homeostatic functions include promotion of learning and memory through secretion of soluble proteins such as brain derived neurotrophic factor (BDNF); participation in normal synaptic pruning; and clearing cellular debris and protein aggregates via phagocytosis. However, in pathologic states activated microglia also contribute to tissue damage, by targeting normal healthy neurons and synapses; by promoting formation of β-amyloid or other misfolded proteins deposited in neurodegenerative diseases; and secreting cytokines (such as IL-1α, TNF, and the complement component C1q) incriminated in induction of neurotoxic A1 astrocytes. In addition, microglia have diverse functions in adaptive immunity, including roles in antigen presentation and immune regulation (Fig. 417-2). (Figure adapted from J Herz et al: Myeloid cells in the central nervous system. Immunity 46:943, 2017, Fig. 2.)
Microglia are located throughout the brain parenchyma, whereas brain macrophages occur primarily in perivascular regions, including the meninges and choroid plexus. In contrast to microglia, brain macrophages are derived from monocytes that enter the nervous system at low levels from the bloodstream on a continuing basis and in higher numbers during pathologic states. A possible exception to this rule are the meningeal macrophages, located primarily in the subdural space, that appear to enter the brain at an early developmental stage and remain throughout the life of the individual. In a murine model of autoimmune demyelination, EAE (Fig. 417-2), macrophages derived from bone marrow monocytes, but not microglia, were the critical population that initiated inflammatory demyelination at paraxonal regions near nodes of Ranvier. Brain macrophages have been found to have multiple pro-inflammatory functions, including promoting adhesion, attraction and activation of B and T lymphocytes; providing antigen-specific activation of T cells via antigen presentation of specific immunogenic peptides, including autoantigens, complexed to surface class II major histocompatibility complex (MHC II) molecules; and contributing to cell injury through generation of oxidative stress and cytotoxicity. By contrast, microglia have been traditionally thought to downregulate inflammatory responses and promote tissue repair. This model of M1 (pro-inflammatory) and M2 (regulatory/repair) macrophage/microglial functions, derived primarily from experimental models of autoimmunity, is certainly an oversimplification, and more nuanced functions of these cell types can be revealed depending on the specific context and environmental cues.
A model for experimental allergic encephalomyelitis (EAE). Crucial steps for disease initiation and progression include peripheral activation of preexisting autoreactive T cells; homing to the central nervous system (CNS) and extravasation across the blood-brain barrier; reactivation of T cells by exposed autoantigens; secretion of cytokines; activation of microglia and astrocytes and recruitment of a secondary inflammatory wave; and immune-mediated myelin destruction. ICAM, intercellular adhesion molecule; IFN, interferon; IL, interleukin; LFA-1, leukocyte function-associated antigen-1; TNF, tumor necrosis factor; VCAM, vascular cell adhesion molecule.
Evidence also supports a primary role for brain macrophages and microglia in neurodegenerative diseases, in contrast to earlier views in which their role was seen as largely secondary and involving phagocytosis of cell debris. In different situations, macrophages and microglia can be either protective or pathogenic. In mice, macrophages contribute to spatial memory when activated in the presence of the cytokine interleukin (IL)-4 produced by invading lymphocytes, and microglia through secretion of BDNF support learning and memory through promoting synaptic plasticity. Experimentally, microglia and brain macrophages also participate in clearance of pathogenic β-amyloid aggregates in Alzheimer’s disease (AD) mice, and disruption of brain macrophages by knockout of CCR2, a chemokine required for entry of bloodstream monocytes into the CNS, exacerbated AD pathology. On the other hand, data indicate that disease exacerbating effects of microglia and macrophages may predominate in other situations. A direct role for microglia in human AD was suggested by genetic evidence implicating the phagocytosis-associated gene TREM2, and other genes belonging to the complement system, in AD susceptibility. Activation of the classical complement cascade is also assuming an increased role in concepts of pathogenesis, as follows; synapses targeted for elimination express the complement proteins C1q and C3, the levels of which increase in the presence of excess β-amyloid; C3-bearing synapses are then targeted for elimination by microglia that express the complement 3 receptor (CR3); and knockout of C3 can rescue the clinical and pathologic abnormalities associated with neurodegeneration in AD-prone mice. In familial frontotemporal degeneration (FTD) due to mutations of progranulin (pgrn), a prominent immune pathology has also been identified, including the presence of activated microglia expressing high levels of pro-inflammatory cytokines. In pgrn-/- mice, an age-dependent microglial activation is associated with upregulation of genes associated with innate immunity including complement proteins, and with enhanced pruning of inhibitory synapses in key regions of the CNS, leading to behavioral disorders reminiscent of human FTD. Moreover, inhibition of complement activation rescued all of these deficits. Taken together, these data indicate a primary role for microglial activation in pgrn associated FTD, likely occurring through enhanced lysosomal trafficking and increased production of cleavage products of the C3 complement component, and leading to enhanced and deleterious synaptic pruning in regions of the brain affected in FTD. Although it is likely that the specific mechanisms of complement dependent neurodegeneration will differ in distinct neurodegenerative conditions, these data provide hope that complement pathway interventions could represent a possible approach to control of neurodegenerative pathologies mediated at least in part through the innate immune system.
Astrocytes represent half or more of all cells in the CNS. Traditionally thought to function as simple interstitial supporting cells that provide scaffolds for neuronal migration and contribute to homeostasis, emerging data indicate far more pleiotropic functions for this cell type. Astrocytes exert profound roles in the life of synapses by secreting factors (such as apolipoprotein E, thrombospondins, and glypicans) that regulate development, maintenance, and pruning of presynaptic and postsynaptic structures. Influenced by local neuronal activity, astrocytes actively phagocytose synapses. Astrocytes also participate in dynamic regulation of vascular tone, in part through astrocyte-astrocyte communication mediated through gap junctions and calcium waves modulated by neuronal activity; support blood brain barrier and glymphatic (see below) integrity through extension of foot process to the vascular structures and expression of aquaporin-4 water channels; and carry out additional metabolic functions essential for the maintenance of neuronal health.
One characteristic of the response to many types of brain injury is reactive astrocytosis, or the formation of a glial scar. Recent work has identified two fundamentally different types of reactive astrocytes that appear to have countervailing functions; the terms A1 and A2 astrocytes have been proposed, by analogy to brain macrophage/microglia M1 and M2 designations, described above. A2 astrocytes are induced by ischemia and may serve beneficial functions, including a contribution to tissue repair after injury. By contrast, A1 astrocytes are induced in diverse inflammatory and degenerative states, and appear to actively participate in the injury process. Interestingly, secreted products of activated microglia, specifically IL-1a, TNF, and C1q, induce astrocytes to transform to the A1 type. Functionally, A1 astrocytes lose the capacity to phagocytose synapses and myelin debris, and are strikingly toxic in vitro to various populations of neurons and to mature oligodendrocytes, potentially at least in part via complement mediated damage. Interestingly, OPCs, abundant in active lesions of multiple sclerosis (MS; Chap. 436) despite the inflammatory milieu, are resistant to A1 mediated death. The nature of the toxic factor is unknown. There is speculation that products of A1 astrocytes could promote damage in conditions as varied as MS, AD (Chap. 423), Parkinson’s disease (PD) (Chap. 427), and amyotrophic lateral sclerosis (ALS) (Chap. 429), despite their distinct etiologies and pathologies.
Two newly identified lymphatic structures of the CNS are the glymphatic and deep dural lymphoid systems, responsible for clearance of debris in the CNS, and likely also serving a role in immune surveillance. The brain has traditionally been considered to lack a classical lymphatic system, and immune responses against antigens are less effectively generated in the CNS than in other organ systems, a concept termed “immune privilege.” However, there is abundant evidence that the immune privilege status of the brain is only relative and not absolute. Furthermore, given the high metabolic demands of the brain some mechanism for efficient removal of solute and debris must be present. One well-established pathway involves the passive flow of solutes from the brain parenchyma into the cerebrospinal fluid (CSF), and their exit via the arachnoid granulations, as well as along cranial and spinal nerve roots to a series of lymphoid structures located in the cribriform plate and nasal mucosa and elsewhere.
The glymphatic system derives its name from a distinctive architecture involving lymphoid-like structures and astroglial cells. CSF synthesized in the arachnoid villi circulates through the ventricles and subarachnoid space surrounding the convexities of the brain and spinal cord, and exits through conduits surrounding arterioles penetrating into the brain parenchyma. These spaces are lined by endothelial cells internally, and by astrocyte foot processes that form the external walls. Aided by arterial propulsion, CSF moves out of these specialized conduits and into astrocytes via foot processes rich in aquaporin-4 water channels, and then in the interstitium of brain parenchyma picks up solutes and particulate debris that are then carried to perivenous spaces where they passage to exit the brain and drain into the lymphatic system. In mice, knockout of aquaporin-4 markedly reduced the flow of interstitial fluids in the brain, underscoring the critical role of astrocyte uptake of CSF in this process. Interstitial flow in the CNS is also impaired with aging, possibly related to changes in astrocytic aquaporin-4 expression. Another fascinating aspect of the glymphatic system is that the transport of fluids and solutes accelerates with sleep, arguing for a critical role for sleep in promoting clearance of debris needed to meet the high metabolic demands of the nervous system. Furthermore, in disease models, aggregated proteins associated with neurodegenerative disease, such as β-amyloid associated with AD (Chap. 423), were also more efficiently cleared during sleep. Indeed, in mice genetically engineered to produce excess β-amyloid and develop Alzheimer’s-like cognitive decline, sleep deprivation increased accumulation of amyloid plaques. Glymphatic pathways are also likely to represent an important egress pathway for lymphocytes in the CNS and a route for lymphocyte encounter with CNS antigens in cervical lymph nodes. In this regard, recent data indicate that deep cervical lymph nodes might be a site for antigen-specific stimulation of B-cells in MS (Chap. 436).
A second recently identified pathway consists of a plexus of small lymphatic-like vessels located on the external surface of meningeal arteries and deep dural sinuses (including the sagittal and transverse sinuses), structures that exit the brain along the surface of veins and arteries and drain to the deep cervical lymph nodes. These conduits are comprised of cells that express a transcriptome indicating that they are components of a lymphoid drainage system distinct from vascular endothelium. These sinus-associated lymphoid structures may be most important in clearing solutes from the CSF, in contrast to the glymphatic system that likely functions to remove waste products from the brain interstitium; however, the exact functions of these two systems and their interrelationships are only beginning to be understood.
MICROBIOTA AND NEUROLOGIC DISEASE
The human microbiome (Chap. 459) represents the collective set of genes from the 1014 organisms living in our gut, skin, mucosa, and other sites. Different microbial communities are associated with different ethnicities, diets, and environments. In any individual, the predominant gut microbiota can be remarkably stable over decades, but also can be altered by exposure to certain microbial species, for example by ingestion of probiotics.
There is compelling evidence that gut microbes can shape immune responses through the interaction of their metabolism with that of humans. These gut-brain interactions are likely to be important in understanding the pathogenesis of many autoimmune neurologic diseases. For example, mice treated with broad-spectrum antibiotics are resistant to EAE, an effect associated with decreases in production of proinflammatory cytokines, and conversely more production of the immunosuppressive cytokines IL-10 and IL-13 and an increase in regulatory T and B lymphocytes. Oral administration of polysaccharide A (PSA) from Bacillus fragilis also protects mice from EAE, via increases in IL-10. Intestinal microbiota from patients with MS were found to promote EAE when transferred to germ free mice, possibly due to imbalances between bacterial species that promote inflammation (such as Akkermansia muciniphila and Acinetobacter calcoaceticus) and those that induce regulatory immune responses (such as Parabacteroides distasonis).
In addition to nonspecific effects on immune homeostasis mediated by cytokines and regulatory cells, some microbial proteins can trigger, in susceptible individuals, a cross-reactive immune response against a homologous protein in the nervous system, a mechanism termed molecular mimicry. Examples include cross-reactivity between the astrocyte water channel aquaporin-4 and an ABC transporter permease from Clostridia perfringens in neuromyelitis optica (Chap. 437); the neural ganglioside Gm1 and similar sialic acid–containing structures from Campylobacter jejuni in Guillain-Barré syndrome (Chap. 439); and the sleep-promoting protein hypocretin and hemagglutinin from H1N1 influenza virus in narcolepsy (Chap. 27).
Recently, a number of tantalizing observations have incriminated the microbial environment in the pathogenesis of a much wider spectrum of neurologic conditions and behaviors, extending well beyond the traditional boundaries of immune-mediated pathologies. It has long been known that gut bacteria can influence brain function, based mostly on classic studies demonstrating that products of gut microbes can worsen hepatic encephalopathy, forming the basis of treatment with antibiotics for this condition.
Mice that developed in a completely germ-free environment displayed less anxiety, lower responses to stressful situations, more exploratory locomotive behaviors, and impaired memory formation compared with non-germ-free counterparts. These behaviors were related to changes in gene expression in pathways related to neural signaling, synaptic function, and modulation of neurotransmitters. Moreover, this behavior could be reversed when the germ-free mice were co-housed with non-germ-free mice. In other experiments, intestinal microbiota were also found to be required for the normal development and function of brain microglia, potentially linking these behavioral effects to specific cellular targets in the CNS.
The enteric autonomic nervous system in humans provides a bidirectional neural connection between the brain and gut. The vagus nerve, which innervates the upper gut and proximal colon, has been implicated in anxiety- and depression-like behaviors in mice. Ingestion of Lactobacillus rhamnosus induced changes in expression of the inhibitory neurotransmitter GABA1b in neurons of the limbic cortex, hippocampus, and amygdala, associated with reduced levels of corticosteroids and reduced anxiety- and depression-like behaviors. Remarkably, these changes could be blocked by vagotomy.
Another area of emerging interest is in a possible contribution of the gut microbiome to autism and related disorders. Children with autistic spectrum disorders have long been known to have gastrointestinal disturbances, and it has been claimed that the severity of dysbiosis correlates with the severity of autism. A murine model of autism was recently induced in offspring after injecting the pregnant mother with the viral RNA mimic polyinosinic:polycytidylic acid (poly I:C). Remarkably, oral treatment of offspring with B. fragilis corrected a range of autistic behaviors in these mice and also improved gut permeability.
PATHOLOGIC PROTEINS, PRIONS, AND NEURODEGENERATION
The term “protein aggregation” has become widely used to describe easily recognizable hallmarks of neurodegeneration (Fig. 417-3). While such neuropathologic hallmarks including plaques, neurofibrillary tangles, and inclusion bodies are often thought to cause neurologic dysfunction, numerous new discoveries over the past several decades have rendered this view increasingly unlikely. Instead, protein aggregates represent accumulations of toxic proteins that become less harmful when they are sequestered into plaques, tangles, and inclusion bodies.
Neurodegeneration caused by prions. A. In sporadic neurodegenerative diseases (NDs), wild-type (Wt) prions multiply through self-propagating cycles of posttranslational modification, during which the precursor protein (green circle) is converted into the prion form (red square), which generally is high in β-sheet content. Pathogenic prions are most toxic as oligomers and less toxic after polymerization into amyloid fibrils. The small polygons (blue) represent proteolytic cleavage products of the prion. Depending on the protein, the fibrils coalesce into Aβ amyloid plaques in AD, neurofibrillary tangles in AD and Pick’s disease, or Lewy bodies in PD and Lewy body dementia. Drug targets for the development of therapeutics include: (1) lowering the precursor protein, (2) inhibiting prion formation, and (3) enhancing prion clearance. B. Late-onset heritable neurodegeneration argues for two discrete events: The (i) first event is the synthesis of mutant precursor protein (green circle), and the (ii) second event is the age-dependent formation of mutant prions (red square). The highlighted yellow bar in the DNA structure represents mutation of a base pair within an exon, and the small yellow circles signify the corresponding mutant amino acid substitution. Green arrows represent a normal process; red arrows, a pathogenic process; and blue arrows, a process that is known to occur but unknown whether it is normal or pathogenic. (Micrographs prepared by Stephen J. DeArmond. Reprinted with permission from SB Prusiner: Biology and genetics of prions causing neurodegeneration. Annu Rev Genet 47:601, 2013.)
Deposition of β-amyloid is strongly implicated in the pathogenesis of AD. Genetic mutations in familial AD cause increased production of β-amyloid with 42 amino acids, which has an increased propensity to aggregate, as compared to β-amyloid with 40 amino acids. Furthermore, mutations in the amyloid precursor protein (APP) which reduce the production of β-amyloid protect against the development of AD and are associated with preserved cognition in the elderly. Mutations in genes encoding MAPT lead to altered splicing of tau and the production of neurofibrillary tangles in frontotemporal dementia and progressive supranuclear palsy. Familial PD is associated with mutations in leucine-rich repeat kinase 2 (LRRK2), α-synuclein, parkin, PINK1, and DJ-1. PINK1 is a mitochondrial kinase (see below), and DJ-1 is a protein involved in protection from oxidative stress. Parkin, which causes autosomal recessive early-onset PD, is a ubiquitin ligase. The characteristic histopathologic feature of PD is the Lewy body, an eosinophilic cytoplasmic inclusion that contains both neurofilaments and α-synuclein. Huntington’s disease (HD) and cerebellar degenerations are associated with expansions of polyglutamine repeats in proteins, which aggregate to produce neuronal intranuclear inclusions. Familial ALS is associated with superoxide dismutase mutations and cytoplasmic inclusions containing superoxide dismutase. An important finding was the discovery that ubiquinated inclusions observed in most cases of ALS and the most common form of frontotemporal dementia are composed of TAR DNA binding protein 43 (TDP-43). Subsequently, mutations in the TDP-43 gene, and in the fused in sarcoma gene (FUS), were found in familial ALS. Both of these proteins are involved in transcription regulation as well as RNA metabolism. In autosomal dominant neurohypophyseal diabetes insipidus, mutations in vasopressin result in abnormal protein processing, accumulation in the endoplasmic reticulum, and cell death.
Another key mechanism linked to cell death is mitochondrial dynamics, which refers to the processes involved in movement of mitochondria, as well as in mitochondrial fission and fusion, which play a critical role in mitochondrial turnover and in replenishment of damaged mitochondria. Mitochondrial dysfunction is strongly linked to the pathogenesis of a number of neurodegenerative diseases such as Friedreich’s ataxia, which is caused by mutations in an iron-binding protein that plays an important role in transferring iron to iron-sulfur clusters in aconitase and complex I and II of the electron transport chain. Mitochondrial fission is dependent on the dynamin-related proteins (Drp1), which bind to its receptor Fis, whereas mitofuscins 1 and 2 (MF 1/2) and optic atrophy protein 1 (OPA1) are responsible for fusion of the outer and inner mitochondrial membrane, respectively. Mutations in MFN2 cause Charcot-Marie-Tooth neuropathy type 2A, and mutations in OPA1 cause autosomal dominant optic atrophy. Both β-amyloid and mutant huntingtin protein induce mitochondrial fragmentation and neuronal cell death associated with increased activity of Drp1. In addition, mutations in genes causing autosomal recessive PD, parkin and PINK1, cause abnormal mitochondrial morphology and result in impairment of the ability of the cell to remove damaged mitochondria by autophagy.
One major scientific question is whether protein aggregates directly contribute to neuronal death or whether they are merely secondary bystanders. A current focus in all the neurodegenerative diseases is on small protein aggregates termed oligomers. These may be the toxic species of β-amyloid, α-synuclein, and proteins with expanded polyglutamines such as are associated with HD. Protein aggregates are usually ubiquinated, which targets them for degradation by the 26S component of the proteasome. An inability to degrade protein aggregates could lead to cellular dysfunction, impaired axonal transport, and cell death by apoptotic mechanisms.
Autophagy is the degradation of cystolic components in lysosomes. There is increasing evidence that autophagy plays an important role in degradation of protein aggregates in the neurodegenerative diseases, and it is impaired in AD, PD, and HD. Autophagy is particularly important to the health of neurons, and failure of autophagy contributes to cell death. In HD, a failure of cargo recognition occurs, contributing to protein aggregates and cell death. Rapamycin, which induces autophagy, exerts beneficial therapeutic effects in transgenic mouse models of AD, PD, and HD.
There is other evidence for lysosomal dysfunction and impaired autophagy in PD. Mutations in glucocerebrosidase are associated with 5% of all PD cases as well as 8–9% of patients with dementia with Lewy bodies. Therefore, this is the most important genetic cause of both disorders thus far identified. There appear to be reciprocal interactions between glucocerebrosidase and α-synuclein. It has been shown that glucocerebrosidase concentrations and enzymatic activity are reduced in the substantia nigra of sporadic PD patients. Furthermore, α-synuclein is degraded by chaperone-mediated and macro autophagy. The degradation of α-synuclein has been shown to be impaired in transgenic mice deficient in glucocerebrosidase as well as in mice in which the enzyme has been inhibited. Finally, it is known that α-synuclein inhibits the activity of glucocerebrosidase. Therefore, there is bidirectional feedback between α-synuclein and glucocerebrosidase. An attractive therapeutic intervention could be to use protein chaperones to increase the activity and duration of action of glucocerebrosidase. This would also reduce α-synuclein levels and block the degeneration of dopaminergic neurons.
The retromer complex is a conserved membrane-associated protein complex that functions in the endosome-to-Golgi complex. The retromer complex contains a cargo selective complex consisting of VPS35, VPS26, and VPS29, along with a sorting nexin dimer. Recently, mutations in VPS35 were shown to be a cause of late-onset autosomal dominant PD. The retromer also traffics APP away from endosomes, where it is cleaved to generate β-amyloid. Deficiencies of VPS35 and VPS26 were also identified in hippocampal brain tissue from AD. A new therapeutic approach to these diseases might therefore be to use chaperones to stabilize the retromer and reduce the generation of β-amyloid and α-synuclein.
The LRRK2 mutations were shown to have effects on clearance of Golgi-derived vesicles through the autophagy-lysosome system both in vitro and in vivo. LRRK2 mutations also are linked to elevated protein synthesis mediated by ribosomal protein s15 phosphorylation. Blocking this phosphorylation reduces LRRK2-mediated neurite loss and cell death in human dopamine and cortical neurons.
Interestingly, in experimental models of HD and cerebellar degeneration, protein aggregates are not well correlated with neuronal death and may be protective. A substantial body of evidence suggests that the mutant proteins with polyglutamine expansions in these diseases bind to transcription factors and that this contributes to disease pathogenesis. In HD, there is dysfunction of the transcriptional co-regulator, PGC-1α, a key regulator of mitochondrial biogenesis. There is evidence that impaired function of PGC-1α is also important in both PD and AD, making it an attractive target for treatments. Agents that upregulate gene transcription are neuroprotective in animal models of these diseases. A number of compounds have been developed to block β-amyloid production and/or aggregation, and these agents are being studied in early clinical trials in humans. Another approach under investigation is immunotherapy with antibodies that bind β-amyloid, tau, or α-synuclein. These studies have shown efficacy in preventing the spread of amyloid, tau, and α-synuclein in animal studies, raising hopes that this could lead to effective therapies by blocking neuron-to-neuron propagation. Two large clinical trials of β-amyloid immunotherapy, however, did not show efficacy, although this therapeutic strategy is still being studied.
PRIONS AND NEURODEGENERATIVE DISEASES
As we have learned more about the etiology and pathogenesis of the neurodegenerative diseases, it has become clear that the histologic abnormalities that were once curiosities, in fact, are likely to reflect the etiologies. For example, the amyloid plaques in kuru and Creutzfeldt-Jakob disease (CJD) are filled with the PrPSc prions that have assembled into fibrils. The past three decades have witnessed an explosion of new knowledge about prions. For many years, kuru, CJD, and scrapie of sheep were thought to be caused by slow-acting viruses, but a large body of experimental evidence argues that the infectious pathogens causing these diseases are devoid of nucleic acid. Such pathogens are called prions, which are composed of host-encoded proteins that adopt alternative conformations that undergo self-propagation (Chap. 430). Prions impose their conformations on the normal, precursor proteins, which in turn become self-templating resulting in faithful copies; most prions are enriched for β-sheet and can assemble into amyloid fibrils.
Similar to the plaques in kuru and CJD that are composed of PrP prions, the amyloid plaques in AD are filled with Aβ prions that have polymerized into fibrils. This relationship between the neuropathologic findings and the etiologic prion was strengthened by the genetic linkage between familial CJD and mutations in the PrP gene, as well as (as noted above) between familial AD and mutations in the APP gene. Moreover, a mutation in the APP gene that prevents Aβ peptide formation was correlated with a decreased incidence of AD in Iceland.
The heritable neurodegenerative diseases offer an important insight into the pathogenesis of the more common, sporadic ones. Although the mutant proteins that cause these disorders are expressed in the brains of people early in life, the diseases do not occur for many decades. Many explanations for the late onset of familial neurodegenerative diseases have been offered, but none are supported by substantial experimental evidence. The late onset might be due to a second event in which a mutant protein, after its conversion into a prion, begins to accumulate at some rather advanced age. Such a formulation is also consistent with data showing that the protein quality-control mechanisms diminish in efficiency with age. Thus, the prion forms of both wild-type and mutant proteins are likely to be efficiently degraded in younger people but are less well handled in older individuals. This explanation is consistent with the view that neurodegenerative diseases are disorders of the aging nervous system.
A new classification for neurodegenerative diseases can be proposed based on not only the traditional phenotypic presentation and neuropathology, but also the prion etiology (Table 417-1). Over the past decade, an expanding body of experimental data has accumulated implicating prions in each of these illnesses. In addition to kuru and CJD, Gerstmann-Sträussler-Scheinker disease (GSS) and fatal insomnia in humans are caused by PrPSc prions. In animals, PrPSc prions cause scrapie of sheep and goats, bovine spongiform encephalopathy (BSE), chronic wasting disease (CWD) of deer and elk, feline spongiform encephalopathy, and transmissible mink encephalopathy (TME). Similar to PrP, Aβ, tau, α-synuclein, superoxide dismutase 1 (SOD1), and possibly huntingtin all adopt alternative conformations that become self-propagating, and thus, each protein can become a prion and be transferred to synaptically connected neurons. Moreover, each of these prions causes a distinct constellation of neurodegenerative diseases.
TABLE 417-1A Prion-Based Classification of Neurodegenerative Diseases ||Download (.pdf) TABLE 417-1 A Prion-Based Classification of Neurodegenerative Diseases
|Neurodegenerative Diseases ||Causative Prion Proteins |
Creutzfeldt-Jakob disease (CJD)
Bovine spongiform encephalopathy (BSE)
Chronic wasting disease (CWD)
Feline spongiform encephalopathy
Transmissible mink encephalopathy
|Alzheimer’s disease (AD) ||Aβ → tau |
Multiple system atrophy
Frontotemporal dementias (FTDs)
Posttraumatic FTD, called chronic traumatic encephalopathy
|Tau, TDP43, FUS (C9orf72, progranulin) |
|Amyotrophic lateral sclerosis ||SOD1, TDP43, FUS (C9orf72) |
|Huntington’s disease ||Huntingtin |
Evidence for a prion etiology of AD comes from a series of transmission experiments initially performed in marmosets and more recently in transgenic (Tg) mice inoculated with a synthetic Aβ peptide folded into a prion. Studies with the tau protein have shown that it not only features in the pathogenesis of AD, but also causes such illnesses as the frontotemporal dementias including chronic traumatic encephalopathy, which has been reported in both contact sport athletes and military personnel who have suffered traumatic brain injuries. A series of incisive studies using cultured cells and Tg mice has demonstrated that tau can become a prion and multiply in the brain. In contrast to the Aβ and tau prions, a strain of α-synuclein prions found in the brains of patients who died of multiple system atrophy (MSA) killed the Tg mouse host ~90 days after intracerebral inoculation, whereas mutant α-synuclein (A53T) prions formed spontaneously in Tg mouse brains killed recipient mice in ~200 days.
For many years, the most frequently cited argument against prions was the existence of strains that produced distinct clinical presentations and different patterns of neuropathologic lesions. Some investigators argued that the biologic information carried in different prion strains could only be encoded within a nucleic acid. Subsequently, many studies demonstrated that strain-specified variation is enciphered in the conformation of PrPSc, but the molecular mechanisms responsible for the storage of biologic information remains enigmatic. The neuroanatomical patterns of prion deposition have been shown to be dependent on the particular strain of prion. Convincing evidence in support of this proposition has been accumulated for PrP, Aβ, tau, and α-synuclein prions.
Although the number of prions identified in mammals and in fungi continues to expand, the existence of prions in other phylogeny remains undetermined. Some mammalian prions perform vital functions and do not cause disease; such nonpathogenic prions include the cytoplasmic polyadenylation element binding (CPEB) protein, the mitochondrial antiviral-signaling (MAVS) protein, and T cell–restricted intracellular antigen 1 (TIA-1).
All mammalian prion proteins adopt a β-sheet-rich conformation and appear to readily oligomerize as this process becomes self-propagating. Control of the self-propagating state of benign mammalian prions is not well understood but is critical for the well-being of the host. In contrast, pathogenic mammalian prions appear to multiply exponentially, but the mechanisms by which they cause disease are poorly defined. We do not know if prions multiply as monomers or as oligomers; notably, the ionizing radiation target size of PrPSc prions seems to suggest it is a trimer. The oligomeric states of pathogenic mammalian prions are thought to be the toxic forms, and assembly into larger polymers, such as amyloid fibrils, seems to be a mechanism for minimizing toxicity.
To date, there is no medication that halts or even slows a human neurodegenerative disease. The development of drugs designed to inhibit the conversion of the normal precursor proteins into prions or to enhance the degradation of prions focuses on the initial step in prion accumulation. Although a dozen drugs that cross the blood-brain barrier have been identified that prolong the lives of mice infected with scrapie prions, none have been identified that extend the lives of Tg mice that replicate human CJD prions. Despite doubling or tripling the length of incubation times in mice inoculated with scrapie prions, all of the mice eventually succumb to illness. Because all of the treated mice develop neurologic dysfunction at the same time, the mutation rate as judged by drug resistance is likely to approach 100%, which is much higher than mutation rates recorded for bacteria and viruses. Mutations in prions seem likely to represent conformational variants that are selected for in mammals where survival becomes limited by the fastest-replicating prions. The results of these studies make it likely that cocktails of drugs that attack a variety of prion conformers will be required for the development of effective therapeutics.
Normal and genetically modified (“transgenic”) mice are the most widely used model systems to study features of human nervous system diseases. However, modeling genetic diseases in rodents is limited to the relatively small number of monogenic human diseases where the specific gene mutations are known, and is further limited by species differences. The latter can be particularly important in brain regions such as the cerebral cortex that have undergone significant evolutionary expansion in humans. These shortcomings, that likely contribute to the low probability that therapeutic efficacy translates from animal models to humans, can potentially be overcome through stem cell models that enable the use of human cells and tissues to model human diseases. The advent of new stem cell technologies is transforming our understanding of the pathobiology of human neurologic diseases. Stem cell platforms are being used to screen for therapeutic agents, to uncover adverse drug effects, and to discover novel therapeutic targets.
Among the most exciting recent advances in stem cell technology is the ability to convert somatic cells, either skin fibroblasts or blood cells, into pluripotent stem cells known as induced pluripotent stem cells (iPSCs). This technology has introduced an entirely new and powerful approach to study the pathobiology of heritable diseases. Pluripotent stem cells can be easily obtained through minimally invasive procedures such as a skin biopsy or blood sample, and converted to pluripotency through application of a cocktail of reprogramming factors to create iPSCs. Initially, a set of four programming factors, Oct3/4, Klf-4, Sox2, and c-Myc, were delivered to cells using lentiviruses which stably integrated the reprogramming factor genes into the iPSC genome, potentially altering disease phenotypes and also abrogating expression of native genes at the DNA sites where the factors integrated. Newer techniques have been developed that use non-integrating approaches such as through the use of Sendai virus, messenger RNA (mRNA), or episomal vectors that circumvent these problems. Once created, iPSC lines can be expanded indefinitely to produce a limitless supply of stem cells. These cells are the starting material for the derivation of specific cell types based on protocols that use small molecules, proteins, or direct gene induction to recapitulate developmental programs. Most current protocols derive neuronal progenitors through “dual-SMAD inhibition,” a step that involves the use of small molecule inhibitors to block endoderm and mesodermal cell fates, thereby creating neural cells by default. Multiple protocols have been developed over the last decade for creating large numbers of human neuron progenitor cell types and directing them toward specific nervous system cell fates, including neuron subtypes from multiple regions of brain and spinal cord as well as retinal cells, glial cells including astrocytes and oligodendrocytes, immune cells, and PNS cells.
The primary medical benefit of iPSC technology is that it enables the creation of patient-specific cells or tissues that are genetically matched to individual patients. This approach not only enables the study of monogenetic disorders, but also sporadic forms of disease, and complex polygenic disorders including those with unidentified risk loci. Furthermore, by deriving iPSC cell lines from multiple patients it would be possible to explore how disease phenotypes may vary according to genetic background. Another approach that has been used to generate specific neuron and glial cell types from somatic cells such as fibroblasts is through direct reprogramming. This approach relies on a cocktail of specific transcription factors to directly convert somatic cells into the alternate desired cell type. This approach bypasses the epigenetic reset that accompanies cells as they are reprogrammed to a pluripotent state. The advantage of this approach is that age-related epigenetic signatures are not erased, so that derived neurons may more readily reflect diseases that manifest in older cells.
Despite the advantages of using in vitro models of nervous system diseases derived from patient-specific iPSCs, several potential roadblocks remain. There are no standard reprogramming or derivation protocols, and the different methods can result in considerable variability in the disease phenotypes reported by different laboratories. Confidence in the specificity of a particular phenotype is therefore increased if it has been validated across multiple laboratories. There is also the problem of inherent variability between patient lines that may result from their different genetic backgrounds. One solution, available only in the case of monogenic disorders, is to use isogenic controls generated using gene editing, such as with CRISPR-Cas9 technology, to create disease and control lines on an identical genetic background. However, because differences in genetic background can influence the penetrance of a particular trait, it will still be necessary to compare disease lines from multiple patients to discern a true disease phenotype. For polygenic disorders where the causative mutations are unknown it will not be possible to create isogenic controls, and in these situations the best strategy for improving reliability and sensitivity is to compare lines from multiple patients.
Most nervous system disorders, including autism spectrum disorder, schizophrenia, PD, AD, and ALS are complex disorders, resulting from an unknown combination of gene mutations and manifest not only in specific cell types, but also in alterations of the local tissue environment. These disorders are difficult to model in animals, but they are approachable using three-dimensional human iPSC stem cell models, often referred to as “organoids.” Organoids are derived from pluripotent stem cells that are directed along a tissue-specific lineage through the timed application of growth factors, genes, or small molecule activators or inhibitors, and allowed to aggregate into three-dimensional structures. With time, cell intrinsic programs are spontaneously engaged and the cellular aggregates begin to self-organize and develop into structures that recapitulate the complex topographical and cellular diversity of normal organ development. In this way it has been possible to create, at least in part, in vitro brain-like organoids that resemble the human forebrain at early stages of development. These structures, when allowed to develop from an anterior neural tube stage, can become heterogeneous containing regions with forebrain, midbrain, and/or hindbrain identity and can often include retina-like structures. The high degree of variability in such “cerebral organoids” can be a liability for controlled studies, and can be reduced by the use of more directed protocols that restrict outcomes to more defined brain regions, such as forebrain, cortex, or ganglionic eminence. A variety of protocols have now been developed to generate organoids with specific regional identity, and fusing organoids of different regional identity with each other has been used to reproduce cellular interactions such as neuronal migration across regions. Many protocols are focused on modeling cortical development, and they can reproduce developmental features including a diversity of progenitor and neuronal cell types topographically distributed within ventricular and subventricular progenitor regions and rudimentary cortical layers. However, the organoids follow a human developmental timetable and still remain at stages roughly comparable to late fetal development after 6–9 months. Moreover they lack key cell types such as endothelial cells, pericytes, microglia, and have few if any astrocytes or oligodendrocytes. Nonetheless, while still only reflecting rudimentary organizational and compositional features, organoids have become attractive models to study human brain development and the pathophysiology of human nervous system diseases in the context of an organized brain-like structure.
Brain Development and Developmental Disorders: Microcephaly and Lissencephaly
Transcriptional analysis has suggested that the neurons produced by most stem cell protocols resemble early to mid-gestational stages of human brain development. The immaturity of stem cell-derived human neurons may limit their utility for modeling adult diseases, but makes them ideally suited for the study of brain development and the pathophysiology of neurodevelopmental disorders.
Primary autosomal recessive microcephaly (MCPH) is a rare neurodevelopmental disorder producing severe microcephaly with simplified cortical gyration and intellectual disability. MCPH was one of the first disorders to be studied using cerebral organoids. Mutations in genes encoding microtubule spindle components and spindle-associated proteins are the most frequent causes of congenital microcephaly. Among them is cyclin-dependent kinase 5 related activator protein 2 (CDK5RAP2). Skin fibroblasts derived from a single microcephalic patient carrying a mutation in CDK5RAP2 were used to generate four iPSC lines. Cerebral organoids grown from these cell lines contained fewer proliferating progenitor cells and showed premature neural differentiation compared to wild type controls. Introducing functional CDK5RAP2 by electroporation partially rescued the disease phenotype, supporting the notion that failure of the founder population of neural progenitors to properly expand underlies the smaller brain. This study demonstrated that brain organoids derived from patients with microcephaly can be used to reproduce features of the disease, but did not reveal new insights or disease features of CDK5RAP2 microcephaly that had not already been described in mouse models.
In a study using cortical organoids to model Miller-Dieker syndrome (MDS), a severe congenital form of lissencephaly or “smooth-brain,” features of the human disease were observed that had not been noted in murine models. Classical lissencephaly is a genetic neurological disorder associated with mental retardation and intractable epilepsy, and MDS is a severe form of the disorder. Cortical folding in humans begins toward the end of the second trimester, a stage of development that has not yet been modeled in organoids, but gyrencephaly depends upon earlier events such as neural progenitor cell proliferation and neuronal migration that can be modeled in organoids. The human organoid model of MDS exhibited several neural progenitor cell phenotypes that had already been reported in mouse models, including altered mitotic spindle orientation and neuronal migration defects. But the organoids also displayed a mitotic defect in a specific neural stem cell subtype, the outer radial glia cell (oRG) that had not been observed in mice. oRG cells are enriched in the outer subventricular zone, a proliferative region that is large in primates and not present in rodents. These cells are particularly numerous in the developing human cortex and are thought to underlie the developmental and evolutionary expansion of the human cortex. oRG cells from MDS patients behaved abnormally and had arrested or delayed mitoses. MDS organoids also identified non-cell autonomous defects in WNT signaling as an underlying mechanism. These insights into mechanistic and cell type specific features of human disease highlight how organoid technology can provide new and valuable perspectives on the pathophysiology of disorders of in utero development.
Acquired Neurodevelopmental Disorders: Zika
The recent outbreak of Zika virus (ZIKV) and associated microcephaly cases in the Americas provided a test case for the utility of brain organoids to model acquired human microcephaly. Despite a correlation between Zika infection rates and the incidence of congenital microcephaly, compelling evidence that ZIKV caused microcephaly was lacking in the early phases of the epidemic. The causal link between ZIKV and congenital microcephaly was buttressed by two studies in 2016 that used human iPSC-derived neural progenitor cells and organoids to demonstrate ZIKV tropism for human neural progenitor cells. Neural progenitor cells (radial glia) were readily infected in vitro with subsequent progenitor cell death and involution of organoid size. Forebrain organoids were further used to highlight the role of the flavivirus entry factor, AXL, in determining viral tropism, and were also used to explore the disease mechanism by demonstrating upregulation of the innate immune receptor toll-like receptor 3 (TLR) in response to ZIKV infection. Stem cell-derived models of human brain development have also demonstrated centrosomal abnormalities in radial glia and alteration in the cleavage plane of mitotic radial glia associated with premature neural differentiation. Mouse models are also being used to study the pathophysiology of congenital ZIKV syndrome, but the availability of unlimited numbers of human neural cells produced using stem cell technology has enabled high-throughput screening assays to test libraries of clinically approved compounds for potential therapeutic agents. This strategy has already highlighted several compounds that could potentially help protect against ZIKV microcephaly.
Neurodevelopmental Disorders: Autism and Schizophrenia
Autism spectrum disorders (ASD) are complex and heterogeneous neurodevelopmental disorders usually manifesting in childhood with difficulties in social interaction, verbal and nonverbal communication and repetitive behaviors. The cellular and molecular mechanisms underlying ASD are thought to arise at stages of fetal brain development, making them well-suited for exploration using human iPSC-derived disease models. The pathophysiology of disorders associated with ASD that are caused by monogenic mutations have been studied using iPSC-derived neurons; these include Fragile X, Rett, and Timothy syndromes.
Fragile X is the most common heritable cause of intellectual disability, affecting 1 in 4000 males and 1 in 8000 females, and is a leading genetic cause of ASD. Patients also have speech delay, growth and motor abnormalities, hyperactivity, and anxiety. The causative mutation lies in the FMR1 gene and produces a CGG triplet repeat expansion from a normal number of 5–20 to >200, leading to epigenetic silencing of the FMR1 gene and loss of the Fragile X mental retardation protein. The epigenetic mechanism means that unlike a simple gene deletion that would lead to ubiquitous loss of expression, the FMR1 locus becomes hypermethylated and epigenetically silenced during differentiation, thus FMR1 protein is expressed by the early embryo and becomes absent only around the beginning of the second trimester. Interestingly, this expression pattern is recapitulated during cellular differentiation in stem cell models. Pluripotent Fragile X stem cell lines have been derived from embryos identified through pre-implantation genetic diagnosis and by reprogramming skin fibroblasts from Fragile X patients to create iPSC lines. In both cases, FMR1 was expressed by the pluripotent stem cells, but underwent transcriptional silencing following differentiation. Fragile X stem cell lines can therefore be used to study the mechanism of FMR1 silencing, an effort that is ongoing. Neurons generated from Fragile X iPSC cells reproduce features observed in neurons from transgenic FMR1 mouse models and patients, including stunted neurites with decreased branching, increasing confidence in the iPSC model. In addition to providing a model that can be used to study disease pathogenesis, Fragile X iPSC-derived neurons could be used to screen for potential therapeutic agents or gene editing strategies that could be able to remove the repressive epigenetic marks induced by the mutation and rescue the phenotype.
Rett syndrome is an X-linked neurodevelopmental disorder with dominant inheritance caused by a mutation in the MECP2 gene. Because males carrying one copy of the defect gene usually die in infancy, most patients are girls. Random inactivation of the X chromosome in girls results in mosaic cellular expression of the mutation that circumvents fatality and produces a variable phenotype. The symptoms are present in early childhood and include microcephaly associated with developmental delay, autistic-like behaviors and cognitive dysfunction, seizures, and repetitive motor actions; these then progress to include difficulties with gait, swallowing, and breathing before usually stabilizing with patients surviving to adulthood. The pathophysiology of RETT syndrome is presumed to involve abnormal epigenetic regulation leading to decreased transcriptional repression of genes whose overexpression produces the disease phenotype, although this concept has been contested. In one of the first studies to use iPSC modeling to study RETT syndrome, it was discovered that when fibroblasts from patients were reprogrammed to pluripotent stem cells, X inactivation was erased. In apparent recapitulation of endogenous events, X chromosome inactivation re-occurred during neuronal differentiation, producing a mosaic of cells carrying the mutant gene intermingled with normal cells. RETT neurons had fewer dendritic spines and synapses, smaller cell bodies, and reduced network activity. Another iPSC model of RETT syndrome highlighted the potential role of altered inhibitory function. RETT neurons were found to have a deficit of a potassium/chloride cotransporter (KCC2) that is developmentally regulated and normally leads to a switch in GABA signaling from excitatory at embryonic ages to inhibitory by birth. In RETT neurons KCC2 expression level was low, and the functional switch in GABA effects was delayed, contributing to some of the disease features and possibly accounting for the developmental onset of the disease. One curious feature of some iPSC RETT lines was that despite the mosaic expression of the mutation, disease phenotypes were observed in all cells. Possibly, this could reflect a non-cell autonomous effect, but as in all iPSC disease models, confidence in disease-specific features will be increased when similar phenotypes are seen across multiple independent studies.
Timothy syndrome, another severe neurodevelopmental disease associated with ASD has been modeled using iPSC-derived organoids. Timothy syndrome is caused by a mutation in the CACNA1C gene coding for a voltage-gated calcium channel, and neuron defects in Timothy syndrome organoids were rescued by selectively altering calcium channel activity. In one study two separate organoids were produced with different regional identity, one represented neocortex and one a more ventral structure known as the medial ganglionic eminence, which is the source of most cortical interneurons. The two organoids were then fused together to allow the interneurons to migrate into the cortex, mimicking their endogenous behavior. The ability to model interneuron migration led to the discovery of a cell-autonomous migration defect in the disease-carrying neurons.
The majority of nervous system diseases, including ASD, are multigenic and cannot be modeled in animals but can be modeled using patient-derived iPSCs. For example, a subset of patients with ASD have large head size, and a cohort of patients with this phenotype were used to generate iPSCs which were converted to neural progenitor cells and forebrain neurons. The progenitors had an accelerated cell cycle and produced an excess of inhibitory interneurons and had exuberant cellular overgrowth of neurites and synapses. This last feature is in contrast to the decrease in spines and synapses observed in other iPSC models of ASD such as Fragile X and Rett syndrome and underscores the need for replication and validation of purported disease phenotypes given the high variability based on differences between stem cell lines, protocols, patient genetic background, and other factors. Moreover, the clinical features of most neuropsychiatric diseases reflect disorders in processes such as circuit formation and refinement that occur after birth and may be difficult to capture at the fetal stage of development reflected in stem cell models.
Patient stem cells have also been used by multiple groups to study the pathophysiology of schizophrenia, producing a variety of diverse and sometimes contradictory results. Reports claim obvious phenotypes such as disruptions in the adherens junctions of forebrain radial glia or aberrant neuronal migration, although such gross abnormalities observed at the equivalent of in utero stages of development seem very unlikely to underlie a disease that usually manifests at adolescence or young adulthood. Other studies report abnormalities related to abnormal microRNA expression, disordered cyclic AMP and Wnt signaling, abnormal stress responses, diminished neuronal connectivity, fewer neuronal processes, problems with neuronal differentiation, and mitochondrial abnormalities among others. While the pathophysiology of as complex a neurodevelopmental disorder as schizophrenia may be multidimensional, it is unclear which, if any, of the reported findings in iPSC models reflect the true pathology of schizophrenia. Progress will likely depend on the adoption of more standard and reproducible protocols, more rigorous identification of cell types, markers of regional identity, and indicators of maturity.
As noted above, the leading concept of AD pathogenesis, the amyloid hypothesis, suggests that an imbalance between production and clearance of β-amyloid leads to excessive accumulation of β-amyloid peptide and the formation of neurofibrillary tangles within neurons, composed of aggregated hyperphosphorylated tau proteins. Additionally, aggregates of amyloid fibrils are deposited outside neurons in the form of neuritic plaques. Recent failures of anti-β-amyloid therapies, which were highly effective in mouse models, have led to a search for alternative models that might be more predictive of therapeutic effectiveness in humans. Among the causes of familial AD are mutations in genes involved in β-amyloid production, including APP and presenilin 1 and 2. Shortly after the introduction of iPSC technology, human stem cell-derived neurons were generated from patients carrying mutations in AD causative genes as well as from sporadic AD cases. The disease neurons developed hallmarks of AD including intracellular accumulation of β-amyloid and phosphorylated tau, as well as secretion of APP cleavage products, features that could be reduced by adding β- or γ-secretase inhibitors or β-amyloid specific antibodies. The neurons also demonstrated other disease features observed in postmortem AD tissues. However, extracellular β-amyloid aggregation and neurofibrillary tangles were not robustly modeled in these two-dimensional systems, presumably because secreted factors were able to readily diffuse away. The use of three-dimensional organoids to model AD overcame this limitation, presumably by recreating a more faithful extracellular matrix. Organoid models promoted the aggregation of β-amyloid, and more readily recapitulated the pathologic features of AD, including the formation of neurofibrillary tangles and neuritic plaques.
It is hoped that the new stem cell models, particularly organoid models, will accelerate our understanding of AD by enabling the study of human disease-carrying cells in a quasi in situ setting. These new models may lead to discovery of novel druggable targets and new diagnostic and prognostic biomarkers. One concern is that the pathogenic features of AD usually appear in the sixth or seventh decade of life and progress slowly over years, while most protocols for the derivation of human cortical neurons generate cells over weeks or months and most remain comparable to immature neurons at fetal stages of development. Nonetheless, these young cells have been used to model neurodegenerative diseases such as AD and HD that strike patients in mid to late adulthood. Possibly the onset of disease phenotype is accelerated in stem cell models due to increased cellular stress, or disease features may actually have a subtle onset at earlier stages than generally suspected. Indeed, 3-year-old children at genetic risk of developing early-onset AD appear to have smaller hippocampal size and lower scores on memory tests than children in a non-risk group. The phenotypes of adult neurodegenerative diseases that are visible at fetal stages may or may not correspond to those manifest at later, adult stages, but they may offer the possibility of devising preventative strategies effective at very early stages of the disease.
Cell Type Disorders: ALS and HD
In diseases such as ALS, PD, and HD, that mostly target specific neuron subtypes, stem cells provide an ideal means to study the vulnerable human cell populations. By enabling the production of unlimited numbers of normal and diseased human midbrain dopaminergic neurons for the study of PD, medium spiny striatal neurons for HD, and spinal and cortical motor neurons for ALS, iPSC approaches have the potential to transform our understanding and management of these diseases. Stem cell-derived neurons serve as platforms to explore mechanisms of cell vulnerability, to screen drugs for neural protection, and potentially to derive neurons for replacement therapy.
Amyotrophic Lateral Sclerosis
One of the first protocols for producing neurons of a specific subtype from embryonic stem cells recapitulated normal developmental programs to generate mouse spinal motor neurons. Pluripotent mouse stem cells underwent neural induction and adopted a caudal identity through the application of retinoic acid, and subsequently adopted motor neuron fate through the action of sonic hedgehog, a ventralizing factor. Generating human motor neurons proved more complex, requiring additional steps, such as early exposure to the growth factor, FGF2. The first application of stem cell-derived motor neurons to study ALS involved the use of mouse motor neurons generated from transgenic mice expressing a mutation in the SOD1 gene, the most common mutation responsible for familial ALS. Only 5–10% of ALS cases are familial, but the known mutations provide a useful entry point to tease apart the causative pathophysiology. Mutations in SOD1 produce ALS through a toxic gain of function for which the mechanism remains unclear, despite the use of multiple transgenic animal and iPSC models. The use of mouse ESC-derived motor neurons, however, demonstrated that toxic factors secreted by SOD1 astrocytes contribute to the death of motor neurons. Interestingly, stem cell-derived interneurons were spared, indicating a specific vulnerability of motor neurons. These findings helped establish the notion that a non-cell autonomous toxic mechanism contributes to ALS pathogenesis and may ultimately lead to novel treatment strategies. These findings also highlight that modeling the full pathophysiology of ALS may require the reproduction of a complex environment including motor neurons, astrocytes, and possibly additional cell types such as microglia. A variety of approaches including co-culture of specific cell types, three-dimensional spinal cord organoids, and microfluidic organ-on-chip models are being explored to achieve a more complete facscimile of spinal cord organization. Similar to other neurologic disorders where a clearly defined phenotype has been observed in human stem cell-derived models, there is hope that drug screening using human disease-expressing cells will identify a potential therapeutic compound.
HD is caused by an expansion in CAG triplet repeats in the huntingtin gene which leads to an expanded polyglutamine tract in the huntingtin protein. HD is dominantly inherited, with symptoms of cognitive decline and uncontrollable gait and limb motions beginning in the third to fifth decade of life with progression to dementia and death ~20 years later. Mutant huntingtin causes a toxic gain-of-function, with the degree of effect related to the CAG repeat length. For example, a CAG length of 40–60 repeats produces adult onset HD, while repeats of ≥60 produce juvenile onset disease. Although it has been 25 years since the discovery of this causative mutation, the disease mechanism remains poorly understood. Excess huntingtin protein and protein fragments accumulate in specific subtypes of neurons where they misfold and form aggregates that are visible as cellular inclusions. Affected cells eventually die, possibly as a result of metabolic toxicity. The medium spiny neurons of the striatum are the most vulnerable neurons, spurring ongoing attempts to produce replacement cells derived from stem cells, but neuron loss is widespread including in the cortex, complicating a cell replacement approach for this disease. HD iPSCs have been generated from patients with various CAG repeat lengths, but those from juvenile onset disease with the longest repeat lengths have been favored as being most likely to express robust disease phenotypes at an early stage. This is particularly important given the immature stage of maturation of stem cell-derived human neurons. This approach has been able to produce disease phenotypes observed in patients including huntingtin protein aggregation, decreased metabolic capacity, increased oxidative stress with mitochondrial fragmentation, and apoptosis enhanced by withdrawal of growth factor support. However, many of these phenotypes were observed in pluripotent cells prior to neural differentiation and in neural progenitors and a broad array of CNS neurons in contrast to the cell type specific features of the disease. Nonetheless, neurons that assumed striatal fate appear to be more vulnerable to stress and apoptosis than other cell types. As with other iPSC models of nervous system diseases, there have so far been few efforts to validate results in multiple iPSC lines having different genetic backgrounds but with similar CAG repeat lengths. An HD consortium has been formed to address this problem by generating a series of iPSC lines from multiple patients. An alternative strategy to validate disease phenotypes has been to use gene editing to create isogenic iPSC lines that are corrected to produce wild type control and HD iPSC lines against the same genetic background.
Despite early successes, it may prove difficult to reconstitute neurodegenerative disease conditions in human cells in vitro over a short time-course because the pathogenic changes of degenerative diseases progress slowly and commence in the later stages of life. The differentiation and maturation of human neurons from stem cell lines occurs over a span of months, which may not be long enough to establish the aged brain conditions under which patients develop robust neurodegenerative pathology. Possible manipulation through gene editing or by application of aging-associated stresses, such as DNA damaging agents or proteasome inhibitors, may accelerate the expression of degenerative phenotypes in human iPSC-derived cellular models. Stem cell-derived organoid models are also ideal platforms to apply methods for cellular level visualization such as clarity and multi-electrode recording techniques to better evaluate three-dimensional organoid structures and explore early-forming circuits. These applications are only just beginning.
Two-dimensional cell cultures are ideal for production and evaluation of large numbers of specific cells of a particular identity, but may not provide the complex extracellular environment necessary to model certain disease processes, such as extracellular protein aggregation. These features can be best modeled using three-dimensional organoids, but current methods do not reproduce all the relevant features of brain tissue. Optimization will be needed to better reproduce the cellular composition of brain, including endothelial cells, astrocytes, microglia, and oligodendrocytes. It may also be necessary to combine different brain regions generated separately, possibly by fusion of tissues such as dorsal cortex, subpallium, thalamus, retina, and others. However, currently there is a limited ability to recreate tissues or neurons with regional brain identity, such as hippocampus, thalamus, or cerebellum. More faithful organoid models could also emerge through the application of bioengineered scaffolds, matrices, or perfusion systems that might allow the growth of larger structures. Of course, not all aspects of mature brain architecture and function will be modeled by these tissue structures, particularly as they represent fetal stages of development, but perhaps the most precocious events in disease etiology can be captured and investigated and these may share mechanistic pathways with disease features that manifest at later stages.
The current excitement surrounding human stem cells has more to do with their promise to improve on animal models of disease than their potential as a source for cell based therapies. Even without new insights into disease pathogenesis, there is promise that iPSC models such as brain organoids will act as drug screening platforms for discovery of novel therapeutics and for detection of off-target and toxic effects. The failure of many neurotherapeutic approaches to translate from animal models to clinical practice underscores the need for better predictive models, and stem cell models and brain organoids based on human cells may be ideally suited to bridge this divide.
A Current Perspective on Neural Stem Cells in the Clinic
The prospect of stem cell therapies to treat diseases or injuries of the nervous system has captured the attention of researchers, clinicians, and the public. The pace of research is usually slow and deliberate, but in the stem cell arena there has been enormous pressure to accelerate the pace of progress in order to bring cell-based therapies to the clinic. Expectations have been raised, and clinics have already begun offering unproven or dangerous treatments to a public that is ill-informed and vulnerable to exploitation. Nonetheless, there is cautious optimism that stem cells will eventually realize the promise of regenerative therapy for at least some currently untreatable or incurable nervous system diseases.
Pursuit of a cell-based therapy for PD has been ongoing for many decades. Following anecdotal success in a handful of patients who appeared to improve following striatal grafts of fetal midbrain dopaminergic cells, two NIH-funded double-blind control studies were launched in the 1990s. However, only a small number of younger patients showed some benefit, and several patients developed spontaneous dyskinetic movements related to the therapy. These efforts constituted a failed trial as the treated patients who did not experience side effects failed to improve significantly. The dyskinesias that curtailed the trials were eventually ascribed to an abundance of serotonergic neurons that were inadvertently included in some of the cell grafts. Protocols for deriving dopaminergic neurons from stem cells could potentially avoid this complication by providing a more purified cell population, and several groups in the United States and Europe have been aggressively pursuing a stem cell-based approach and are nearing clinical trials. Meanwhile, techniques to extract dopaminergic cells from fetal tissue have been improved, and on the basis of encouraging results in individual transplanted patients, some of whom have managed to go off their Parkinson’s medication, a new trial of fetal cell transplantation for PD has started in Europe. This is a very consequential trial, as a poor clinical outcome could dampen enthusiasm for the planned follow-on stem cell trials in PD and possibly in other disorders as well.
One of the first cell-based clinical trials for a neurological disease targeted patients suffering from an untreatable childhood disorder, Batten disease. Batten disease is an autosomal recessive metabolic disorder resulting from an inability to synthesize a lysosomal enzyme critical to brain function. The Phase 1 trial involved six patients with infantile and late infantile forms of the disease who received neural stem cells rather than any specific postmitotic cell type. Neural stem cells derived from donated fetal tissue were expanded in vitro prior to surgical grafting into the brain. This approach was not without risk, as the neural stem cells were proliferating and could potentially form an abnormal growth. The rationale was that the cells would be capable of synthesizing and secreting the missing lysosomal enzyme and would therefore serve as a delivery device. Animal studies using a transgenic mouse model of Batten disease demonstrated rescue, and this promising result led to a small Phase 1 trial. The Phase 1 study was considered a success as no adverse events were reported and the cells appeared to be safe, though there was no clinical improvement and no clear evidence of whether the cells had dispersed, transformed into neurons or glia, or indeed survived at all. Despite clearing the Phase 1 trial, the company did not pursue further trials for Batten disease, but instead initiated clinical trials using the same cell product for several other indications, including an inherited fatal dysmyelination syndrome known as Pelizaeus-Merzbacher disease (PMD). The human neural stem cells have both neurogenic and gliogenic potential, and when delivered to white matter regions in experimental animals most persisting cells had become oligodendrocytes. This supported use of the cells to promote myelin formation in conditions such as PMD. The company also initiated trials in spinal cord injury. However, the spinal cord trial failed to achieve sufficient benefit in Phase II and the company ceased its work on stem cell therapies.
Spinal cord injury is an attractive target for novel therapies since there are no effective treatment options currently. A series of stem cell trials designed to treat subacute spinal cord injury are underway in the United States and Europe. The first to enter clinical trials in the United States was based on a protocol designed to generate oligodendrocytes from pluripotent embryonic stem cells. Evidence of efficacy was obtained in animal models following surgical grafting of cells to sites of spinal cord injury. However, evidence of myelination of host axons was minimal, and other mechanisms were invoked for improvement in gait, including trophic support and immune modulation. Regulatory permission for a Phase 1 trial for subacute mid-thoracic injury was initially stalled by concern over abnormal growths at sites of cell deposit in some animals, but this was satisfactorily addressed and patient trials commenced. However, following a change in leadership, the stem cell program was terminated. The program was acquired by another company that has resumed the spinal cord injury trial and received regulatory approval to advance to include cervical level injuries.
The possibility of treating ALS by replacing dying motor neurons with stem cell–derived substitutes has excited interest but this prospect seems very remote. Even if new neurons are able to integrate into spinal cord circuits and become properly innervated, they would have to grow long axons that would take many months to years to project to appropriate targets and attract myelinating Schwann cells. Furthermore, cells would need to be grafted at multiple spinal cord and brainstem levels, and the upper motor neuron deficit would need to be treated by replacing projecting neurons in the motor cortex. An additional complication is the recent finding that spinal motor neurons have unique segmental identity, and replacement cells might need to be generated with a range of molecular identities in order to integrate at multiple spinal levels. This would still leave unaddressed the toxic effects recently shown to be produced in ALS by diseased astrocytes and microglia that could attack the replacement cells. A more tractable near-term solution would be to graft support cells that could rescue or protect endogenous motor neurons from damage. This approach was tried in a mouse model of ALS. Human stem cell–derived neural progenitor cells engineered to express GDNF, a growth factor known to provide trophic support for neurons, were grafted to the spinal cord of young ALS mice. The cells dispersed and were able to rescue motor neurons, a very promising result, but disappointingly, the animals became weak and died at the same rate as untreated control animals. However, ALS is a deadly disease with no known treatment. In the hope that patients will respond differently than mice, a clinical trial based on this approach has been approved by the U.S. Food and Drug Administration (FDA) and will begin soon.
Following Shinya Yamanaka’s discovery of iPSCs, the Japanese government has invested in bringing iPSC-derived cell therapy to the clinic. Banks of iPSC lines selected to capture the diversity of HLA haplotypes found in the Japanese population are being produced in the hope that these will allow cell therapies to be matched to individual patient haplotypes in order to avoid immune rejection. While these stem cell banks were still being produced, the first Japanese study to use stem cells was approved in August, 2013, and involved patients who were to receive customized therapy using cells derived from their own skin fibroblasts. The targeted disease was age-related macular degeneration, a common cause of blindness in the elderly that results from loss of retinal pigment epithelial (RPE) cells. RPE cells are relatively easy to generate from pluripotent stem cells, making replacement therapy an attractive target in this condition. A challenge is to coax the replacement cells to recreate an epithelium in the subretinal space. The Japanese approach involves surgical insertion of a biofilm seeded with RPE cells into the retina. One patient was treated with his/her own stem cell-derived RPE cells, but prior to treating a second patient, the genome of the RPE cell line was sequenced, and a mutation was discovered in a known oncogene. The trial was halted and a decision made to discontinue the effort for customized cell therapy in favor of using RPE cells derived from the national repository of banked iPSC lines which undergo extensive gene sequencing and quality controls. This outcome serves as a caution for the challenges involved in bringing a customized cell therapy to the clinic.
By far the largest number of human trials have been performed using mesenchymal stem cells (MSCs) sourced from a variety of sites including bone marrow, peripheral blood, adipose tissue, umbilical cord, etc. Interest in the potential utility of MSCs for regenerative therapy began with the optimistic report that bone marrow stem cells were pluripotent and capable of generating nerve and heart muscle as well as blood cells. The possibility that easily obtainable MSCs could be used to regenerate injured or diseased cells or organs to treat diseases ranging from stroke, neurodegenerative disease, myocardial infarct, and even diabetes, generated enormous enthusiasm. The enthusiasm proved irresistible to many, and even after the initial reports were discredited—MSCs turned out not to be pluripotent stem cells as initially thought—a veritable flood of papers began to appear claiming disease-modifying activity of MSCs in mouse models of almost every degenerative disease and injury model. But when it became clear that the MSCs were not transforming into or generating new neurons or cardiac myocytes, alternative mechanisms of action were invoked, including the release of trophic factors, cytokines, or inflammatory modulators that were credited with producing their remarkable restorative effects. The relative ease with which blood or adipose tissue can be harvested from patients or donors and MSCs extracted has led to a rapidly expanding number of clinical trials for conditions ranging from stroke and MS to AD and PD. Furthermore, a loophole in the regulatory framework of the FDA allows autologous cell therapy to escape regulation provided that the cells have not been significantly processed. This lax regulation has spawned a veritable industry of stem cell clinics making unsubstantiated claims of success in treating nervous system diseases. Patients have died from treatments in unregulated clinics operating in countries around the world and three patients became blind in a well-publicized incident following stem cell treatments delivered by a Florida clinic. The “stem cells” were derived from the patients’ own fat tissue and blood. These activities represent the dark side of the stem cell revolution perpetrated by practitioners who exploit the desperation of patients and their families. Legitimate and effective stem cell therapies will emerge over time, but given the prevalence and abundance of misleading information available on the internet and elsewhere, a trusted and well-informed physician can play a key role in helping patients navigate the current cell therapy minefield.
et al: Do glia drive synaptic and cognitive impairment in disease? Nat Neurosci 18:1539, 2015.
AR: The use of brain organoids to investigate neural development and disease. Nat Rev Neurosci 18:573, 2017.
et al: Interactions between the microbiota, immune and nervous systems in health and disease. Nat Neurosci 20:145, 2017.
et al: Myeloid cells in the central nervous system. Immunity 46:943, 2017.
et al: Complement and microglia mediate early synapse loss in Alzheimer mouse models. Science 352:712, 2016.
et al: Neurotoxic reactive astrocytes are induced by activated microglia. Nature 481:541, 2017.
et al: Evidence for α-synuclein prions causing multiple system atrophy in humans with parkinsonism. Proc Natl Acad Sci USA 112:E5308, 2015.
P: Selling stem cells in the USA: Assessing the direct-to-consumer industry. Cell Stem Cell 19:154, 2016.
SK: The Enteric Network: Interactions between the immune and nervous systems of the gut. Immunity 46:910, 2016.