Evolution of glutamatergic signaling and synapses

Abstract

Glutamate (Glu) is the primary excitatory transmitter in the mammalian brain. But, we know little about the evolutionary history of this adaptation, including the selection of l-glutamate as a signaling molecule in the first place. Here, we used comparative metabolomics and genomic data to reconstruct the genealogy of glutamatergic signaling. The origin of Glu-mediated communications might be traced to primordial nitrogen and carbon metabolic pathways. The versatile chemistry of L-Glu placed this molecule at the crossroad of cellular biochemistry as one of the most abundant metabolites. From there, innovations multiplied. Many stress factors or injuries could increase extracellular glutamate concentration, which led to the development of modular molecular systems for its rapid sensing in bacteria and archaea. More than 20 evolutionarily distinct families of ionotropic glutamate receptors (iGluRs) have been identified in eukaryotes. The domain compositions of iGluRs correlate with the origins of multicellularity in eukaryotes. Although L-Glu was recruited as a neuro-muscular transmitter in the early-branching metazoans, it was predominantly a non-neuronal messenger, with a possibility that glutamatergic synapses evolved more than once. Furthermore, the molecular secretory complexity of glutamatergic synapses in invertebrates (e.g., Aplysia) can exceed their vertebrate counterparts. Comparative genomics also revealed 15+ subfamilies of iGluRs across Metazoa. However, most of this ancestral diversity had been lost in the vertebrate lineage, preserving AMPA, Kainate, Delta, and NMDA receptors. The widespread expansion of glutamate synapses in the cortical areas might be associated with the enhanced metabolic demands of the complex brain and compartmentalization of Glu signaling within modular neuronal ensembles.

Keywords: Algae; Aplysia; Aspartate; Cnidaria; Ctenophores; Eukaryotes; GABA; Glutamate receptors; Glutamine; Nervous system evolution; Neurotransmitters; Placozoa; Stress; Synapse; Trichoplax; Vesicular glutamate transporters; scRNA-seq.

Fig. 1.

A brief history of glutamatergic signaling in mammals. The discovery of glutamate (Glu) in 1866 was followed by identifying monosodium glutamate as a chemosensory agent in the human taste (Ikeda, 1909). However, the identification of these taste receptors was about 100 years later, when metabotropic mGlu4, mGlu1, and 7TM receptors (T1R1+T1R3) were cloned (Kurihara, 2015). As sensors for Glu, together with other co-activator/s (5′ nucleotides), they provided the molecular bases for the distinct umami sensation in humans. Synaptic functions of Glu had been suggested by Hayashi (Hayashi, 1954). This hypothesis was based on his observation that Glu induced convulsions after the central injections and earlier studies about the involvement of Glu in cognitive functions (Weil-Malherbe, 1950). The concept of chemical transmission in the brain had been accepted by the end of the 50s. What is the central transmitter in the CNS? Watkins and colleagues in the Eccles laboratory showed that L-Glu had excitatory action (Curtis et al., 1959; Curtis and Watkins, 1961). Nevertheless, the acceptance of this innovative idea took over 20 years. The significant objections were related to the fact that Glu is a crucial amino acid for dozens of metabolic pathways across all domains of life (see Fig. 2 and text for details). How is the universality of Glu functions in virtually every cell transformed into precise synaptic communications in the brain? After the development of specific agonists and antagonists, both ‘fast’ ionotropic (iGluRs - (Hollmann et al., 1989)) and ‘slower’ metabotropic receptors (mGluR - (Houamed et al., 1991; Masu et al., 1991)) were cloned (Hollmann and Heinemann, 1994). And their structural organizations had been revealed using crystallography (Armstrong and Gouaux, 2000; Armstrong et al., 1998; Chen et al., 1999; Kunishima et al., 2000; Mayer, 2006; Mayer et al., 2001; Sun et al., 2002). In invertebrates, glutamate-mediated signaling is more diverse and perhaps is more complex than in vertebrates. Deep comparative and evolutionary roots of glutamate-mediated transmission are also poorly explored. But the very origin and parallel evolution of glutamatergic signaling and synapses are inherently linked to a crucial role of glutamate in metabolism and bioenergetics.

Fig. 2.

The Krebs cycle and glutamate metabolism as evolutionary sources of transmitter signaling. The schematic diagram of universal biochemical pathways emphasizes l-glutamate (Glu) as a critical metabolite in bioenergetics, nitrogen assimilation, and amino acid synthesis. Both Glu and glutamine are nitrogen donors for amino acids (including ubiquitous transamination mechanisms). All enzymes of these pathways are highly conserved across species. 3-D organizations of three key enzymes are our reconstructions from Trichoplax adhaerens. This nerveless placozoan has the simplest-known animal organization but complex behaviors (see text). The co-factor/metabolite controls of glutamate dehydrogenase, glutaminase, and glutamine synthetase are more complicated than the regulation of most neuronal components. GABA, Aspartate, N-acetyl-l-aspartate, Glycine, and Nitric Oxide (NO) are also known as universal extracellular low molecular weight transmitters (bold blue) across all domains of life. Their syntheses are inherently linked to glutamate metabolisms and cellular bioenergetics. Thus, these transmitters might evolve as signal molecules from the intermediate metabolism. And these are essential components of nearly every prokaryotic and eukaryotic cell. Intracellular metabolite concentrations are unpredictable from these biochemical pathways. However, these concentrations can be determined experimentally using metabolomic tools, which gave surprising results, including the highest intracellular glutamate concentration (see Fig. 3 and text).

Fig. 3.

Glutamate as the most abundant metabolite across the tree of life: Intracellular metabolomics from bacteria to mammals. Comparing the metabolomes between a prokaryotic cell (E. coli) and a mammalian cell iBMK from mouse (Park et al., 2016). In both cases, glutamate is the major intracellular metabolite with the highest concentration compared to other molecules. Absolute intracellular concentrations are shown together with % of a particular molecule in the entire metabolome for both cell types. 50 % of the whole metabolome are evolutionarily conserved pan-signal molecules.

Fig. 4.

A comparing the hierarchy of intracellular metabolites based on their absolute concentrations for bacteria (E. coli) and iBMK cells (M. musculus);. The tanglegram shows six clusters with concentrations for 10^–2 to 10^–7 M: red (10^–2 M), orange (10^–3 M), green (10^–4 M), cyan4 (10^–5 M), dodger blue (10^–6 M), darkorchid (10^–7 M). Lines for the same metabolite connect similar sub-trees in each species. Glutamate is a noticeable outgroup for both E. coli and mouse metabolomes with an absolute concentration of about 96 mM. Scale bar: E. coli - 96 mM, iBMK cells - 63.8 mM. Although concentrations of metabolites between prokaryotes and eukaryotes vary, Glu is the most abundant metabolite followed by aspartate, ATP, pyruvate, glutamine. The tanglegram was constructed using ‘dendextend’ packages (Galili, 2015; Park et al., 2016)).

Fig. 5.

The structures of glutamate (Glu) and aspartate (Asp). Amino acids Asp and Glu have 4-carbon and 5-carbon skeletons, respectively. However, their 3D organization and the 3D distribution of polar groups differ substantially; it is reflected in their distinct recruitments in metabolic pathways and signaling, including pharmacological properties. Two right images are carbon skeletons of Glu and Asp. The central and left images show van der Waals surfaces and molecular electrostatic potential (range –0.1 −0.1). The atomic structural data were obtained from the Cambridge Crystallographic Data Center. https://www.ccdc.cam.ac.uk/structures/?ccdc-check=881ded3bf3ea70f6813dada69c8a8eee.

Fig. 6.

Modular organization of iGluRs in prokaryotes and eukaryotes as sensors of the environment. A. The same domain organization is predicted from sequence information both in prokaryotes and unicellular eukaryotes. Two domains of iGluR are recognized in prokaryotes and some unicellular eukaryotes: LBD - ligand-binding domain, TM – transmembrane domain. A nearly identical organization exists between bacteria and cyanobacteria. For the first time, we identify a putative iGluR in Archaea (see text for details). TM4 domain is the eukaryotic innovation, as shown in Aureococcus (red sequence in C-terminus). The organizations of iGluRs are shown for uncultured methanotrophic archaeon (WP_010870874.1), bacterial Legionella micdadei (CEG60559), basic GluR0 from the cyanobacteria Synechocystis sp. (BAA17851 sequences See supplements S1, S2 for sequences). Color codes: N-terminus (blue), C-terminus (red). In Archaea, Bacteria and Cyanobacteria N-terminal starts in LBD, and C-terminal is also located in LBD. In eukaryotes, C-terminal is in TM domain (TM4). B. ‘Putative prototypes’ section shows predicted ancestral protein structures critical for the evolution of iGluRs. Potassium transporter (Trk) and voltage-gated potassium channel (center, (AFV23945)) from Archaea share similar protein topology and orientation with transmembrane domains of iGluRs. The right image shows the pore structure of the predicted archaeon potassium channel (AFV23945) with the position of the potassium ion. Of note, the archaeon potassium channel pore orientation is the same as in iGluRs of both prokaryotes and eukaryotes. The predicted potassium channel sequence (CBH37362.1) was obtained and reconstructed from an uncultured archaeon (methanotrophic ANME-1 group, environment samples (Meyerdierks et al., 2012)). The potassium transporter (AFV23945.1) is from another archaeon Methanolobus psychrophilus R15 (Chen et al., 2016)).

Fig. 7.

Diversity of ligand-gated ionotropic glutamate receptors (iGluRs) in prokaryotes and eukaryotes. Mammals have two extracellular domains. Representative organization of ionotropic glutamate receptors in prokaryotes and eukaryotes: the cyanobacterium, Synechocystis sp., the pelagophyte unicellular algae, Aureococcus anophagefferens, the comb jelly, Mnemiopsis leydyi, and the rodent, Rattus norvegicus (imaging by PyMol software with N- terminus domain – blue, C-terminus domain -red). – Transmembrane domains (TMs) anchor receptors in the cell membrane. LBD – extracellular domains with the ligand-binding site. NTD – amino-terminal domains. iGluRs also contain an intracellular cytoplasmic loop. Some receptors have auxiliary subunits (Ramos-Vicente et al., 2021). iGluRs have a tetrameric structure as illustrated for Synechocystis sp. (PDB:5weo), Aureococcus (PDB:6ruq), Mnemiopsis (PDB:5kuf), and Rattus (PDB:6njm) with the ‘Fab’ domain as an additional part of the native AMPA receptor complex (Zhao et al., 2019). Abbreviations: TMs – transmembrane domains, LBD – ligand-binding domain, NTD – N-terminus domain. Color: blue – N-terminus, red – C-terminus.

Fig. 8.

Early diversification of iGluRs: 3-domain organization of iGluRs correlates with independent origins of multicellularity in eukaryotes (Knoll, 2011). The diagram shows phylogenetic relationships among the major eukaryotic lineages (kingdoms) and the presence of iGluRs (asterisks). Black asterisks – unicellular eukaryotes, red asterisks – iGluRs in multicellular eukaryotes. Inserts illustrate the organization of the iGluRs in the unicellular alga (Vitrella brassicaformis, Alveolata) and the multicellular filamentous brown alga (Ectocarpus siliculosus, Phaeophyta). Four acquisitions of the NTD domains occurred in different eukaryotic lineages (+NTD). Two domain iGluRs (TM + LBD) were present in the last common eukaryotic ancestor, and they could be lost in Euglenozoa, Fungi, and Amoebozoa (-iGluRs, red cross marks). Abbreviations: NTD – N-terminal domain, LBD – ligand-binding domain, TMs – transmembrane domains; red stars – the presence of iGluRs, the golden chain on the model – chain A. All sequences are in the Supplementary Tables S1 and S2.

Fig. 9.. Twenty-two distinct phyletic lineages of iGluRs were identified in eukaryotes.

Phylogenetic tree of iGluRs across representatives of the main eukaryotic groups (Maximum Likelihood, see Supplementary tables for 121 sequences). Black dots in branches denote ultrafast bootstrap approximation (UFBoot) support values above 70 %. Lineage-specific duplications are shown as triangles. Full trees with uncollapsed nodes and numerical bootstrap support values are in Supplemental Fig. 1 and Supplementary Tables S1 and S2. NTD domains are present in four separate clades, and their acquisition occurred in the lineages leading to multicellular and/or colonial organisms: such as animals, plants, colonial choanoflagellates, multicellular red and brown algae. The classes of iGluRs such as Lambda, NMDA, Epsilon, Akdf and Delta correspond to the classification proposed elsewhere (Ramos-Vicente et al., 2018).

Fig. 10.

Domain organization of iGluRs. A. Alignments of the transmembrane domains and cytoplasmic loop (cysteine [cys]-rich loop) part of iGluRs. Cys-rich loop presents between M1 and M2 in ctenophore iGluRs with 4–5 cysteines (Alberstein et al., 2015). Humans, Porifera, Placozoa, choanoflagellates, and Amoebozoa have no cysteines in similar cytoplasmic loop regions but contain many polar charged amino acids (Arg, Lys, His, Asp, Glu). B. 3D organization of iGluR from the ctenophore Pleurobrachia bachei (PbGR3 in the alignment A; one monomer is shown at different rotation angles) and the placozoan Trichoplax sp. (H2 haplotype, RDD38339 in the alignment; tetrameric structure). NTD - N-terminal domain (NTD, I), LBD – ligand-binding domain with the ligand-binding site (LBS) and α-COOH group (αD) (II), TM – transmembrane domain (III), and cytoplasmic loop of iGluRs (IV). C. Trichoplax sp. (RDD38339). A region inside the pore with the Q/R site for potential RNA editing (it is serine Ser633; green asterisk in A, green dots in B and C). Of note, two polar charged amino acids are located nearby (Arg638 - blue asterisk in A, and aspartate Asp637, red asterisk). D. Cytoplasmic loop in Trichoplax sp. iGluR (RDD38339) has 5 basic (blue) and 5 acidic (red) amino acids (with length variations in different species).

Fig. 11.

Basal metazoan clades are highly divergent in their iGluR family composition. Ctenophora, Porifera, Placozoa, and Cnidaria are metazoans that diverged before bilaterians. This analysis includes four medusas, two ctenophores, and three placozoans. Phylogenetic tree of iGluRs across Metazoa consists of 25 species, 10 phyla, 258 sequences (Maximum Likelihood). Black dots indicate support values above 70 %. Lineage-specific duplications are shown as colored triangles. Supplemental Fig. 2 and Tables S1 and S2 show a whole tree with uncollapsed nodes and numerical bootstrap support values. The classes of iGluRs such as Lambda, NMDA, Epsilon, Akdf, and Delta correspond to the classification proposed elsewhere (Ramos-Vicente et al., 2018). Ctenophores have only the Epsilon family, placozoans – Epsilon and AKDF, anthozoans – Epsilon, AKDF and NMDA-Cnid, other cnidarians – AKDF, and highly expanded cnidarian-specific NMDA-Cnid. Demosponge Amphimedon has no iGluR at all, while calcareous and homoscleromorph sponges have Lambda and AKDF family iGluRs. See also Supplementary Fig. S2.

Fig. 12.

Parallel evolution of iGluRs animals with emphasis on basal metazoans. A. Schematic phylogenetic relationships among representative animal clades with gain and loss of different classes of iGluRs. The iGluR classes were named according to (Ramos-Vicente et al., 2018), and reconstruction of gene gain/loss was derived from the phylogenetic trees (Figs. 9, 11, 15 and 16, Supplementary Figs. S2–S5). B. The proposed genealogy and diversification of classes and subclasses of iGluRs in Metazoa. The overall phylogenetic classification of iGluRs will continue to be developed with growing comparative information. See text for detail.

Fig. 13.

Phylogenetic relationships among different classes of anionic vesicular transporters: Vesicular glutamate transporters (VGluT) vs. Sialins. The figure shows Maximum Likelihood phylogenetic tree of SLC17 family transporters (vGluTs, sialins, phosphate, and nucleoside transporters) across Metazoa (13 species, 8 phyla, 71 sequences, Supplemental Table 2). VGluTs were found in Placozoa, Cnidaria, and Bilateria but not in sponges or ctenophores. Two clades contain well-characterized vGluTs and sialins from Homo and Aplysia. Several proteins from Nematostella and placozoans occupy basal positions in these clades and are putative orthologs of VGluTs and sialins in basal metazoans. Ctenophore and sponge SLC17 proteins form their lineage-specific clades without orthology to bilaterian SLC17 family transporters. Nematostella and placozoans also have some unique SLC17 transporters without orthologs in Bilateria. Sialin-like protein from the unicellular choanoflagellate Monosiga is located between metazoan sialin and vGluT clades, suggesting earlier divergence of these transporter clades.

Fig. 14.

Single-cell deciphering of glutamate signaling in Hydra: The majority of glutamatergic cells are non-neuronal (dark blue dots). We used VGluT as a marker of glutamatergic cells. We reanalyzed the original whole body scRNA-seq data from Siebert et al. (Supplementary Methods, Supplement 1), using the reported markers for different cell types (Siebert et al., 2019). There are about 45 cell clusters. In the freshwater polyp, Hydra neurons were present both in endodermal (blue ‘n’ and arrows) and endodermal (red ‘n’ and arrows) layers. Stem cells (ISC) and precursors of neurons (np) and nematoblasts are marked green. Note that the vesicular glutamate transporter was expressed in developmental nematoblasts, and only a few individual cells were clustered with ectodermal neurons. Nematoblasts accumulate poly-gamma-glutamate, and the observed high differential expression of VGluT likely contributes to Glu uptake required to synthesize this polymer. Nematoblasts and precursors of neurons might derive from the same population of ISC. See text for details.

Fig. 15.

Parallel evolution of glutamate receptors with an emphasis on Ecdysozoa: Specification and hybrid immune-neuro receptors. A. Maximum Likelihood phylogenetic tree of iGluRs (28 species, 11 phyla, 361 sequences, see Supplementary Tables S1 and S2) for sequences). Black dots show support values above 70 %. A whole tree with uncollapsed nodes and numerical bootstrap support values is in Supplemental Fig. 3. Various ecdysozoan lineages show highly divergent duplications of iGluRs. The classes of iGluRs such as Lambda, NMDA, Epsilon, Akdf, and Delta correspond to the classification proposed elsewhere (Ramos-Vicente et al., 2018). Spiders and shrimps have an extremely high diversity of iGluRs due to extensive duplications of AKDF subfamily. On the other hand, nematodes have relatively smaller numbers of iGluRs per their genomes. See text for details. B. The family of pentameric cys-loop Glu receptors also show remarkable lineage-specific diversification in Arthropods, including a novel class of chimeric “ligand-pattern recognition” receptors recently identified in the American lobster (Polinski et al., 2021) and the shrimp Litopenaeus. See also Supplementary Fig. S3.

Fig. 16.

Parallel Evolution of iGluRs in animals with emphasis on Spiralia. Maximum Likelihood phylogenetic tree of iGluRs (27 species, 18 phyla, 382 sequences, see Supplementary Tables S1 and S2). This tree includes seven spiralian phyla and Xenoacoelomorpha (the acoel, Hofstenia, and Xenoturbella). The diversity of iGluRs does not correlate with the presence of complex nervous systems. One of the largest complements of iGluR genes were found in the nemertine Notospermus (44), the oyster Crassostrea (29), the flatworm Macrostomum (28), and the acoel Hofstenia (25). Among molluscs, Octopus, with the most complex nervous system in invertebrates, has the fewest number iGluRs (19). Nemertine Notospermus is the only member of Spiralia, which retained Epsilon subfamily iGluRs and completely lost the NMDA subfamily. Black dots show support values above 70 %. (see Supplemental Fig. 4 for the tree with uncollapsed nodes and numerical bootstrap support values). See also Supplementary Fig. S4.

Fig. 17.

Diversity of glutamatergic systems across Metazoa with three eukaryotic outgroups. The diagram shows the predicted number of vesicular Glu transporters and iGluRs in the sequenced genomes of representative species. Two left columns briefly summarized the functions and position of glutamatergic systems in neural circuits (where such data are available). Of note, every lineage shows multiple duplication events with a potential sub-functional specification of iGluR regardless of apparent neuronal morphological complexity.