Cellular neurometabolism: a tentative to connect cell biology and metabolism in neurology

Metabolism involves thousands of proteins, mostly enzymes, receptors and transporters, the deficit of which causes IEM. Deficits can affect small or complex molecules. The number of IMD (inborn metabolic diseases) obviously depends on the definition of an IMD, and in the -omics era, this is changing quickly (Van Karnebeek et al. 2018). According to Morava, the “classification of a disorder as an IMD requires only that impairment of specific enzymes or biochemical pathways is intrinsic to the pathomechanism,” (Morava et al. 2015).

Using this extended definition a recent tentative nosology of IEM encompasses more than 1100 IEM currently identified and provisionally classified into 130 groups (Ferreira et al. 2018). This list is of little help to clinicians and has little in common with neurologists’ clinical approach. A simplified and updated classification of IEM mixes elements from the practical diagnostic approach with pathophysiological considerations into three large categories based on the size of molecules (“small and simple” or “large and complex”) and their implication in energy metabolism (Saudubray and García-Cazorla 2018).

Whatever their size, metabolites involved in IEM may behave in the brain as signalling molecules, structural components and fuels, and many metabolites have more than one role (see supplemental material Fig. S1).

Almost all these IEM have metabolic marker(s). Their diagnosis relies on plasma, urine, and CSF investigations. Many of them can be detected by neonatal metabolic screening. There are two subcategories in small molecule disorders defined by whether the phenotype primarily results from an accumulation or a deficiency.

Historically, the disorders in this group are the most typical IEM and are characterised by signs and symptoms resulting primarily from the abnormal accumulation of the compound(s) proximal to the block and potentially reverse as soon as the accumulation is removed. They share some characteristics:

In the brain, molecules that accumulate in intoxication disorders can behave as neurotransmitters (Kölker 2018) in the case of amino acids or stimulate biological pathways related to impaired autophagy and nerve growth factors. Synaptic plasticity and excitability are almost constantly impaired and executive functions are especially vulnerable. Therefore, and in spite of proper metabolic control, most of these patients display behavioural, emotional and learning difficulties.

Symptoms result primarily from the defective synthesis of compounds distal to the block or from the defective transport of an essential molecule through intestinal epithelium, blood-brain barrier (Table 1 and Fig. S2 in supplemental material), and cytoplasmic or organelle membranes. Unlike those defects belonging to the intoxication group most of these defects interfere with embryofoetal development causing a neurodevelopmental disruption, have a congenital presentation and share many characteristics with disorders in the complex molecules group (see later). Molecular mechanisms of IEM linked to essential compounds are different from those linked to non-essential ones.

Essential compounds come from diet and must be transported through cellular membranes. Accordingly, IEM are linked to carrier defects (essential amino acids like SLC 7A5 for BCAA (Tărlungeanu et al. 2016), or essential fatty acids (FA) such as MFSD2A for DHA (Guemez-Gamboa et al. 2015). BCDH kinase deficiency overactivates, causing irreversible BCAA oxidation in very low levels of BCAA as in SLC 7A5 mutations and presents a similarly devastating neurological syndrome with neurodevelopmental disruption (Novarino et al. 2013).

Non-essential compounds can be synthesised inside cells. Their availability depends on the integrity of the synthesis pathway. IEM are linked to enzyme defects (such as serine, glutamine, and asparagine synthetase deficiency) (Häberle et al. 2005; Acuna-Hidalgo et al. 2015; Ruzzo et al. 2013). In addition to AA and FA synthesis and transport defects, this group also encompasses the IEM causing metal deficiency (Boycott et al. 2015; Park et al. 2015) as well as neurotransmitter metabolism and transport defects (Tristán-Noguero and García-Cazorla 2018, this issue). Some vitamin-related disorders and purine and pyrimidine defects also belong to this category.

Major neurodevelopmental disruptions lead to severe global encephalopathies where almost all neurological functions are chronically altered. In early onset presentations, patients display severe psychomotor delays affecting both motor and cognitive milestones. Microcephaly and hypomyelination are very common as epilepsy and movement disorders. These defects mimic early “non-metabolic” genetic encephalopathies that affect crucial neurodevelopmental functions such as neuronal precursor proliferation, migration, pruning and dendrite development. This is because these small molecules contribute to antenatal brain “construction” in terms of signalling, cytoskeleton guidance, synapse formation and later on in experience-dependent synapse remodelling.

These consist of IEM with symptoms due, at least in part, to a deficiency in energy production or utilisation within the liver, myocardium, muscle, brain and other tissues.

The brain accounts for 20% of an adult’s energy expenditure at rest and more than 50% in a child (Sokoloff 1960). Neurons expend 70–80% of total energy (the remaining portion used by glia) and the great majority (80%) is utilised to fuel neuronal channels. We could then hypothesise that energy defects in the brain tend to behave as “synaptopathies”, thereby encompassing those symptoms within the spectrum of synaptic disorders (Tristán-Noguero and García-Cazorla 2018, this issue).

Energy compartmentalisation between neurons (lactate use, pentose pathway, oxidative phosphorylation) and glia (mostly glycolytic), and even inside different compartments of a single neuron, has been now largely described (Oyarzábal and Marín-Valencia n.d., this issue; Camandola and Mattson 2017). Fuel molecules such as ATP and lactate also have signalling roles promoting synaptic plasticity. Energy use for axonal transport and nerve conduction is also compartmentalised. Electrically active axons not only rely on glycolysis but also need the supply of lactate provided by oligodendrocytes (that surround axons covering them with myelin) (Trevisiol et al. 2017). Glucose is the obligatory fuel for adult brain, but lactate produced from glucose by astrocytes within brain during activation has been proposed to serve as neuronal fuel. This simple and seductive hypothesis is far from being proven (Dienel 2017). Whatever the source of ATP may be, ATP consumption is greatly increased in demyelinated axons due to the redistribution of ion channels along the length of the denuded axon (Alizadeh et al. 2015; Salzer 2015) and decreasing mitochondrial ATP production results in free radical production leading to lipid peroxidation and severe physico-chemical modifications of cellular and organelle membranes (Dobretsov et al. 1977).

Given the vulnerability of energy homeostasis in the brain, most neurological disorders, and in particular, neurodegenerative diseases are necessarily linked to disturbances in energy metabolism.

This expanding group encompasses diseases that disturb the metabolism of complex molecules that are not or poorly water soluble or diffusible. In general, these defects have no easily identified metabolic markers and diagnosis is primarily based on molecular techniques (NGS, WES). The main chemical categories of such complex molecules encompass the glycogen, the sphingolipids, the phospholipids, the cholesterol and bile acids, the glycosaminoglycans, oligosaccharides and glycolipids and the nucleic acids. Like the small molecules, there are two subcategories in complex molecule disorders defined by whether the phenotype primarily results from an accumulation or a deficiency.

These complex metabolic processes involve all organelles (mitochondria, lysosomes, peroxisomes, endoplasmic reticulum, Golgi apparatus, lipid droplets and the synaptic vesicle) and are highly regulated. Cellular membranes are formed from a chemically diverse set of lipids present in various amounts and proportions. Membrane lipids fulfil many functions including membrane structural components, energy and heat sources, signalling molecules, protein recruitment platforms and substrates for post-translational protein–lipid modification (Harayama and Riezman 2018). There are two subcategories in complex molecule disorders defined by whether the phenotype primarily results from an accumulation or a deficiency.

All defects affecting systems involved in intracellular vesiculation, trafficking, processing of complex molecules and quality control processes (like protein folding and autophagy) recently discovered using the NGS technique belong to this category (Sprecher et al. 2005) (Hirst et al. 2015) (Stockler et al. 2014). This is also the case with the synaptic vesicle (SV) cycle (Cortès-Saladelafont et al. 2018, this issue) and congenital disorders of autophagy like SENDA or Vici syndrome (Ebrahimi-Fakhari et al. 2016).

The neurological manifestations are diverse, complex, with a predominance of motor symptoms (such as spastic paraparesis, ataxia and movement disorders), are often progressive and may be associated with extra-neurological signs. Trafficking, autophagy and quality control processes are major issues in regard to neurochemical mechanisms. These diseases are leading the way to a new neurology that connects metabolism, neuroscience and clinical neurology.

The twenty-first century, with only 18 years of history, has already changed neuroscience thanks to significant new discoveries and techniques. On a global level, the study of the brain has moved to a system-neuroscience/system biology approach (Südhof 2017). The European initiative “Human Brain Project” (http://​www.​humanbrainprojec​t.​eu/​en/​) strives to accelerate neuroscience and brain-related medicine through the strategic alignment of scientific research programmes. It is precisely here, where metabolism and biochemical pathways, so under-represented in neuroscience thus far, can be of precious help. In clinical neurology, disorders are studied in function of the main sign or neurological syndrome, which is the traditional approach in medicine. Combining this symptom-based classification with the pathophysiological mechanisms (Table 1S, Table 2) is not always easy and rarely integrated in the practical procedure of neurologists.

The concept of “molecular machines” was introduced by Alberts about two decades ago to describe large assemblies of biomolecules that are specialised to perform particular cellular functions (Alberts 1998). Despite metabolism strongly orchestrates the regulation of neuronal functions, the “metabolic” approach of brain disorders has been mostly neglected so far. For instance, energy metabolism participates in the intricate organisation of postsynaptic density which is a complex signalling structure involving many proteins related to neuronal plasticity and learning, and is an important niche in monogenic causes of intellectual disability (Camandola and Mattson 2017; Bayès n.d., this issue). Other examples such as vesicular glycolysis, which is used for fast axonal transport (Zala et al. 2013), P4-ATPases which uses the energy from ATP hydrolysis to transport specific lipids across membranes (Martín-Hernández et al. 2016), and complex networks of interactions between SNARE proteins and lipids (mostly phospholipids) that regulate synaptic vesicle trafficking, docking and therefore neurotransmitter release (Darios et al. 2009) are processes that rely on pure biochemistry but are linked to classical “neurogenetic” disorders.

Neurons and glia (astrocytes, oligodendrocytes and microglia) are the two major cell types. Neurons have many phenotypes that reflect different structures, functions, and metabolic signatures but there is still no consensus on how neuron types should be defined. The way that a neuron communicates determines its identity, which in turn relies on functions such as synaptic plasticity and neuronal excitability, the basis of learning and behaviour. These neurobiological functions are prone to be damaged in almost any kind of neurological disease. Not only metabolites, the “drivers” of neurobiological functions, behave mainly as structural molecules, signalling molecules and fuels but the same metabolite may also have more than one function which can also change over time. For example, dopamine is a neurotrophic factor in early development and a neurotransmitter later on. Additionally, metabolites in the brain are highly compartmentalised which is defined as the presence of different neuronal compartments (such as soma, axon, synaptic terminal, each with specific metabolic properties) and more than one distinct pool of a given metabolite. These separate pools of a metabolite are not in rapid equilibrium with each other but maintain their own integrity and turnover rates. At the same time, a particular molecule may have a function in only one anatomic structure or cell type but not elsewhere. For instance, lactate is important for the establishment of long-term memories acting in the neurons of the hippocampus (Suzuki et al. 2017).

Neurons have polarised morphologies. They are very well compartmentalised into pre- and postsynaptic regions, synapses, dendrites, somas and axons, which are essential for their functions. They are the largest known cells, with process lengths ranging from millimetres to centimetres and metres (axons). This combination of specialised morphology with extreme length imposes a need for signals to be actively transferred between neuronal compartments (Wiegert et al. 2007; Terenzio et al. 2017). To integrate information within and between sub-neuronal compartments, neurons need sophisticated transfer mechanisms that ensure necessary molecules are available or generated in the right place at the right time. Metabolites participate in all these processes in a compartmentalised way. Categories of IEM as initially classified could also be related (at least partially) to this “neuronal microanatomic and functional” point of view. The following are models of compartmentalised signalling and metabolism (Fig. 1):

In order to protect the brain from fluctuations in the blood composition, the blood–brain barrier (BBB) and the blood–cerebrospinal fluid barrier (BCSFB) regulate the exchanges of molecules between blood and cerebral fluids. These barriers, made of a variety of complex lipids, limit the free diffusion of metabolites between blood and brain fluids, and selectively transport essential nutrients (AA, lipids), ions and signalling molecules. Only six IEMs related to BBB-specific defects involving the transport of glucose, monocarboxylates, AA and lipids have been described (Fig. S2, Table 1) and only one in the BCSFB, the FOLR1 defect, (Steinfeld et al. 2009). However, secondary disturbances of essential AA transport is commonly observed in case of an elevation of one specific neutral AA using the L-type AA carrier like phenylalanine or leucine in PKU and MSUD, respectively. This observation has been used to treat PKU (Matalon et al. 2006).

Brain barrier defects belong to the category of small molecule defects (both essential and non-essential molecules) and are potentially treatable although they can present different degrees of clinical response.

Organelles, proteins and RNAs are actively transported to synaptic terminals for the remodelling of pre-existing neuronal connections and formation of new ones (Wojnacki and Galli 2016). Mechanisms of intracellular communication can be divided into the following different categories:

Membrane traffic is a highly regulated process that ensures the communication between membrane-bound compartments while maintaining their specific protein and lipid composition (De Matteis and Luini 2011). Although the various membrane compartments (endoplasmic reticulum (ER), Golgi, endosomes, vesicles, organelles…) use a different repertoire of proteins and complex lipids, the basic principles are similar and divided into four main steps that travel from cargo sorting to fusion with the acceptor compartment (Fig. 2). Additionally there is an endosome-free route through mRNA binding proteins for local translation (Boulay et al. 2017; Sakers et al. 2017). Membrane and endosome-related trafficking can lead to a great variety of neurological diseases (Table S2). As a rule, spastic paraparesis is mostly related to defects localised at ER, endosomes and lysosomes(Table 2). Peripheral neuropathies and in particular, diverse subtypes of Charcot-Marie-Tooth, as well as muscle dystrophies, are more likely to have their localisations at the plasma membrane. Trafficking defects between the ER and the Golgi are related to complex neurological and extra-neurological syndromes (Table S1). Finally, and localised at the presynaptic terminal, synaptic vesicle disorders are in great majority trafficking diseases that produce synaptopathies (spectrum epilepsy, intellectual disability, neuropsychiatric signs, movement disorders) (Cortès-Saladelafont et al. 2016).

Disorders of trafficking belong to the category of complex molecule defects. Several glycosylation defects (COG1, 7, 8) are related to trafficking alterations between the ER and the Golgi complex. Lysosome biogenesis defects (Hermansky-Pudlak syndromes; type 7 and 8) and several phosphoinositide-phosphatase-producing myopathies, Lowe syndrome and CMT types 4B1 and 4B2 also belong to this category.

These functions are related to transport via motor proteins in different compartments: soma, dendrites and axons. They are also linked to nerve impulse propagation thanks to the collaboration between the axon and myelin. Proteins of the cytoskeleton play a central role in the creation and maintenance of cell shapes. They provide structural organisation helping to establish metabolic compartments and serve as tracks for intracellular transport (Fig. 3). Motor proteins, including the myosin, dynein and kinesin families of proteins, are responsible for the anterograde and retrograde transport of cargo. In general, myosins are actin-dependent motors, whereas dyneins and kinesins are microtubule-dependent motors (Namba et al. 2015). There are different mechanisms by which cargo (including organelles, mRNA and receptors) are selectively transported to the appropriate subcellular region by motor proteins including protein kinase regulation (Gibbs et al. 2015). Specific diseases are linked to deficits in these mechanisms and motor proteins themselves (Fig. 3, Table S1). Most of them are motor diseases (Tables 1 and S1) (motor neuropathies, CMT2, HSP, ALS, SMA, CFEOM: congenital fibrosis of extraocular movements) and cortical migration abnormalities (Table S1). Organelles (mitochondria, lysosomes, the synaptic vesicle, lipid droplets), autophagosomes, neurotrophin receptors and mRNA, are also transported through the axon by a set of motor proteins and their regulators that are different in every organelle and structure and also have their own diseases (Fig. 3). Myelin from oligodendrocytes covers almost the entire axon in order to provide a fast, low energy-consuming conduction. Myelin composition and related disorders are very numerous and go beyond the scope of this article. Some important metabolic aspects of cytoskeleton, axon and myelin are highlighted below. Disorders of mitochondrial transport exhibit abnormal mitochondrial shape and motility and in some cases decreased oxidative phosphorylation function. Moreover impaired mitochondrial transport has an impact at the presynaptic terminal due to the energy requirements for neurotransmission.

Myelinating glia have been found to be critical for transferring energy metabolites (i.e. lactate) to axons through monocarboxylate transporters, which are present on the inner membrane of the glial sheath and on the axon. However, the role of lactate as a main source of energy to neurons remains severely debated (Dienel 2017) and MCT1 deficiency does not give rise to demyelination (van Hasselt et al. 2014). While myelination markedly reduces the energy requirements for action potential propagation, energy is still required for axonal transport, metabolism and action potential regeneration at nodes (Salzer 2015).

Alterations in membrane potential and synaptic homeostasis lead to severe epilepsies and other progressive neurological features. This is the case of KIF5A deficiency (Rydzanicz et al. 2017) and TRAK1 mutations (Barel et al. 2017) (Table S1). On the other hand, myelin metabolic diseases are strongly related to complex lipid synthesis and remodelling defects but also particularly to those diseases where the enzyme/transporter/metabolic defect is primarily localised at the oligodendrocytes (metabolic leukodystrophies).

The balance between synthesis and degradation in neurons is of utmost importance because there are long distances from the neuronal cell body, where most proteins are synthesised to their place of action. The slow rates of axonal transport are not ideal for replacing misfolded or dysfunctional synaptic proteins on demand. Due to the sustained nature of neurotransmission, the synaptic proteome is prone to protein misfolding. To maintain their specialised structure and function, neurons possess dedicated synapse-specific proteostasis machinery localised to pre- and postsynaptic compartments (Gorenberg and Chandra 2017). Alterations in this network, in particular, disturbances in ER functions trigger abnormal protein aggregation. In turn, ER stress triggers a signalling reaction known as UPR (unfolded protein response) that is also related to synaptic function (Hetz and Saxena 2017). Alterations in the proteostasis network have been related to a great diversity of neurodegenerative diseases (Table S1).

Disorders of quality control mechanisms involve both, small and complex molecules. Protein-folding diseases are related to chaperone defects such as DNAJC12 and DNAJC6, involved in neurotransmitter disturbances and a spectrum of symptoms including early Parkinsonism (and hyperphenylalaninemia in the case of DNAJC12 (Anikster et al. 2017)). A recent study shows that ER stress is involved in the pathophysiology of homocystinuria (Martínez-Pizarro et al. 2016) leading to the hypothesis that other intoxication defects could potentially be related to ER stress and consequently, UPR.

Autophagy defects also belong to this category beside monogenic mitophagy diseases (DJ-1, PINK1 and Parkin) related to Parkinson’s disease. Interestingly, a neurotransmitter defect, SSADH, triggers impaired mitophagy at the postsynaptic level through the mTOR pathway due to the high concentration of GABA at the synaptic cleft (Vogel et al. 2016). Finally, metabolism defects in metabolic repair are exemplified by NAXE deficiency. NAXE catalyses the epimerization of NAD(P)HX, thereby avoiding the accumulation of toxic metabolites (Kremer et al. 2016).