Paradoxical Expression of Ionotropic Glutamate Receptors in Leucocytes.

Introduction
In the Central Nervous System (CNS), receptors for amino acids regulate neuronal excitability, synaptic transmission and the plasticity underlying behaviour. A major excitatory, depolarising compound is glutamate, which acts on receptor sub-populations characterised by the selective agonists α-amino-isoxazole-propionic acid (AMPA), N-methyl-D-aspartate (NMDA) and kainic acid; NMDA and AMPA receptors are a type of ionotropic glutamate receptor (GluR). In the CNS, NMDA receptors are unique as they require both ligand binding and membrane depolarization to activate. This dual requirement makes NMDA receptors critical for synaptic plasticity, particularly in processes like long-term potentiation, which is essential for learning and memory. NMDA receptors are permeable to calcium (Ca2⁺), sodium (Na⁺), and potassium (K⁺) ions. The influx of Ca2⁺ through NMDA receptors triggers various intracellular signalling pathways that lead to synaptic strengthening. Additionally, NMDA receptors are involved in neurodevelopment and neuroprotection, but excessive activation can lead to excitotoxicity, contributing to neurodegenerative diseases. AMPA receptors mediate fast synaptic transmission in the central nervous system. They are primarily permeable to Na⁺ and K⁺ ions. AMPA receptors are responsible for the rapid depolarization of the postsynaptic membrane, leading to the generation of excitatory postsynaptic potentials. These receptors are crucial for the initial phase of synaptic transmission and play a role in synaptic plasticity by modulating their number and function at the synapse. In the CNS, these receptors are necessary for learning and memory, which makes them therapeutics targets for neurological disorders. The role of these receptors in the pathophysiology of brain disorders is well-recognized and current research focuses on new therapeutic targeting approaches to ameliorate excitotoxicity [1], neurodegenerative disease [2] and neurodevelopmental disorders [3] as well as treatment for pain [4] and drug dependence [5].
AMPA receptor-targeting antagonist therapies are currently limited to Perampanel, although at one time Talampanel was considered a promising therapy [6]. Tezampanel (which is also an antagonist of kainate receptors) may also be further developed due to its neuroprotective properties [7]. A wider range of therapies exists for targeting NMDA receptors and includes ketamine, memantine and amantadine [8, 9]. The earliest identified AMPA antagonists, such as CNQX, are unsuitable treatments for neurological conditions due to poor blood brain barrier penetration. However, they may be useful for targeting AMPA receptors in non-neurological cells for, while GluRs are widely distributed on neurons around the CNS, they have also been reported on non-neuronal cells (astrocytes/microglia) and this has prompted studies to determine whether they are also expressed and functionally active on other types of immune system cells. Some of the results summarised in Table 1 lead to the conclusion that at least some of these ionotropic GluR subunits seem to be expressed at intermediate levels in leucocytes, as they are readily detectable at the protein level. As other studies have demonstrated a functional response to ionotropic GluR targeting in leucocytes, the potential to target leucocytes with agonists or antagonists for these receptors that do not cross the blood brain barrier may prove useful for treating disease. The AMPA receptor has been demonstrated to interact with the protein tyrosine kinase Lyn, which is also expressed in leucocytes, and by signalling through Lyn can modulate MAP kinase signalling [29]. Additionally, depolarisation of the membrane in lymphocytes is associated with a range of T cell phenotypes [30].
It is apparent from these studies that there is variation in the expression of receptors, which may depend on the cell type (lymphoid cells [10, 12, 18], myeloid cells [11, 26]), the species (human [10], mouse [21, 26] or rat [11, 12]) or tissue of origin (blood [10], thymus [20], bone marrow [27]), their metabolic or activation state (steady state vs disease [25]), the nature of the response examined (e.g. the response to glutamate [16]), and the sensitivity of the quantification or analytical methodology employed (e.g. PCR [10, 12] or flow cytometry [25]); see Supplementary Table 1 for further particulars. Similar findings in several laboratories indicates a degree of reproducibility, although laboratory-based factors may play a part in the variation of the results. Such factors include the limits of sensitivity using different techniques, technical factors such as the specificity of reagents (e.g. antibodies, oligonucleotides), or even the limits of cell isolation which do not preclude contamination at some stage—a recognised difficulty when using amplifying techniques such as PCR on bulk cell isolations.
To clarify how GluR expression, including AMPAR, NMDAR, kainate, and metabotropic receptors, differs in CD4+ cells between disease states and steady state, we aggregated sequencing data from multiple clinical studies and several human leucocyte subsets with differing states of activation, reasoning that these data would better demonstrate which conditions and cell types were most amenable to therapeutic targeting via each receptor class. Surprisingly, there is a substantial mismatch between the results in Table 1 and online repositories of gene expression data including The Human Protein Atlas [31] and Gene Expression Omnibus [32], which indicate that lymphocytes have virtually no expression of ionotropic glutamate receptor subunit genes.
As a common feature of gene expression measurement from Table 1 (see also Supplementary Table 1) is the use of endpoint PCR with gel electrophoresis or RT-PCR with an intercalating dye (e.g. SYBR™ Green), which can lack specificity and sensitivity, we measured the expression of the genes for individual subunits of each GluR using TaqMan qPCR assays. Most of the GluR subunit genes were not widely detected, leading to the conclusion that gene expression for most GluR subunits is generally absent from circulating leucocytes, consistent with information in the online repositories. Nevertheless, our analysis has measured low expression of three GluR subunit genes (GRIN2D, GRIN3B, GRIA3) in the pro-monocytic THP-1 cell line and one AMPA receptor subunit GluA3 (gene GRIA3) in plasmacytoid cells (pDCs).

Methods and Materials

Analysis of Clinical Studies
Relevant studies archived at Gene Expression Omnibus[32] were collated and the relative gene expression (normalised transcripts per million, TPM) in leucocytes were compiled for 4 classes of GluRs – AMPA and NMDA ionotropic receptors as well as kainate receptors and metabotropic receptors. Additionally, 3 genes that are not typically expressed in lymphocytes (IRAK3, IDO1, CSF1R), THP-1 cells (DNTT, CD3E, LAG3) or pDC (IRAK3, TDO2, DNTT) were also included as an indicator of background or non-specific expression.

Human Leucocytes
Aphaeresis cones from healthy donors were obtained from the NHS Blood Service in accordance with approval from the Fulham Research Ethics Committee (11/H0711/7). Aphaeresis material was diluted with PBS to ~ 90 mL and layered over Lympholyte®-H (Human) Cell Separation Media (Cedarlane) and separated by centrifugation at 20 °C at 2000 rpm for 20 min, without brake. The PBMC were isolated by aspiration and washed twice with PBS by centrifuging at 1500 rpm at 4 °C for 5 min. CD4+ lymphocytes were isolated from PBMC by firstly depleting CD14+ monocytes by positive selection (CD14 microbeads, Miltenyi 130–050–201) and then positively selecting CD4+ cells (CD4 microbeads, Miltenyi 130–045–101) from the CD14– cells. Blood dendritic cells (pDC, cDC1, cDC2) were isolated from PBMC firstly by negative selection of non-DC (Pan-DC Enrichment Kit, Miltenyi 130–100–777) then staining the isolated cells for FACS with antibodies for CD19, CD1c, CD141, CD303 and viability (Zombie Violet™ Fixable Viability kit); all antibodies and viability dye were from BioLegend (302215, 331505, 344105, 354207 and 423113, respectively). From the enriched starting population, DC subsets were flow sorted as singlet, live, CD19– cells and then further divided into pDC (CD1c–CD303+), cDC1 (CD141+CD303–) or cDC2 (CD1c+CD303–); blood DC subsets were washed with PBS and snap frozen for RNA extraction. Additionally, the THP-1 pro-monocytic and HL-60 pro-myelocytic cell lines were also used for a transcriptomic analysis of ionotropic GluR expression; the DBTRG glial cell line was used for comparison.

Human Monocyte-Derived Macrophages (MDM)
Human MDM were derived as previously described[33]. Briefly, monocytes were isolated from PBMC using immunomagnetic CD14 positive selection beads and columns (CD14 microbeads, Miltenyi 130–050–201). Macrophages were differentiated from monocytes for 5 days in 10% FBS 1% Penicillin/Streptomycin RPMI (10^7 cells/10 mL/10 cm dish) supplemented with 50 ng/mL M-CSF (Peprotech, 300–25); MDM were detached and re-plated into 6-well plates, rested overnight, then stimulated with M1-polarising (LPS (10 ng/mL, Merck L6529) and IFNγ (10 ng/mL, Peprotech 300–02)) or M2-polarising (IL-4 (50 ng/mL, Peprotech 200–04)) stimuli; prior to stimulation, some macrophages were collected as a pre-treatment condition. After 20 h of stimulation, cells were washed with PBS then harvested for RNA extraction.

RNA Extraction and Reverse Transcription
RNA extraction was performed with a Total RNA miniprep kit (New England Biolabs, T2010S) including the DNase I on-column digestion, according to the manufacturer’s protocol. Reverse transcription of 500 ng of RNA was performed with the High Capacity cDNA Reverse Transcription Kit (ThermoFisher Scientific, 4368814) in a 40 μL reaction, which was then diluted to a total of 120 μL with RNAse-free water.

PCR
For THP-1 pretreated with AMPA then stimulated with TNF, gene expression was measured with TaqMan™ Array Human Immune Panel microfluidics cards (Fisher Scientific) on a Viia7 real-time PCR system. Seven potential housekeeping genes were measured for each sample (HPRT1, PGK1, GUSB, TFRC, ACTB, GAPDH and 18S). As the variation and range for the Ct of HPRT1 was less than a third of the alternatives, it was chosen as the housekeeping gene used for analyses using the δδCT approximation method. For GluR subunit gene expression, TaqMan quantitative PCR (qPCR) was performed in a reaction of 6 µL containing 2.4 µL of template (equivalent to 10 ng of RNA). TaqMan assays were from ThermoFisher Scientific and as follows – GRIN1 (Hs00609557_m1), GRIN2A (Hs00168219_m1), GRIN2B (Hs01002012_m1), GRIN2C (Hs01016628_m1), GRIN2D (Hs00181352_m1), GRIN3A (Hs01077968_m1), GRIN3B (Hs00879911_g1), GRIA1 (Hs00181348_m1), GRIA2 (Hs00181331_m1), GRIA3 (Hs01557466_m1), GRIA4 (Hs00898778_m1), KCNA3 (Hs00704943_s1) and HPRT1 (Hs02800695_m1). Expression was measured relative to housekeeper gene expression (HPRT1) using the δδCT approximation method.
For the measurement of flip/flop isoforms of GRIA3 in pDC and THP-1 cells, the template of each reaction was 8.32 and 4.16 ng, respectively, in a total reaction volume of 10.5 µL consisting of 5.25µL of MangoMix™ (Meridian Bioscience, BIO-25033), water and primers; primers for GRIA3 isoforms were Flip_F GGAATGTGGAGCCAAGGACT, Flop_F ATGAGCAAGGCCTCTTGGAC and Flip + Flop_R TTGAAGCAGCCACGTTTTCG. For each isoform, a PCR positive control template plasmid was created by traditional cloning methods; water alone was used as negative control. PCR reactions were run on a TAE gel containing 1.5% agarose (Meridian Bioscience, BIO-41025) alongside a DNA ladder (Hyper Ladder™ 1 kb, Meridian Bioscience, BIO-33053). Standard curve qPCR was performed as previously described [34, 35].

Transcriptomic Response to AMPA
THP-1 cells were cultured in 10% FBS 1% P/S RPMI in 6 well plates (2 × 106/well). Cells were pretreated with vehicle or AMPA (Abcam (ab120130)) for 1 h then stimulated with TNF (5 ng/mL, Peprotech, 300-01A) for 6 h. Cells were washed with PBS and snap frozen for RNA extraction.

Statistics
Results were analysed with MS Excel (Microsoft) or Graphpad Prism (GraphPad Software). Heatmaps and dendrograms were drawn with Multi-Experiment Viewer [36] or RAWGraphs [37]. Statistical tests are outlined in Figure Legends.

Results
Over the previous two decades, many researchers have reported that lymphocytes, particularly CD4+ T cells, express intermediate levels of ionotropic GluRs (Table 1, Supplementary Table 1) and also, using single cell analyses (FACS, ICC/IHC), indicate that a significant proportion of T cells have detectable expression at the protein level [10, 22, 23, 27]. Additionally, several researchers have measured functional responses in cells to ligands of these receptors. However, RNA sequencing data from several studies indicates that the level of gene expression in these cells, is negligible (Fig. 1, Supplementary Table 2); this is apparent in either healthy controls or patients with a myriad of health conditions (autoimmune arthritis, lupus, multiple sclerosis, cancer, viral infection and schizophrenia), and in naïve or stimulated conditions; additionally, there was also negligible expression on mast cells [38, 39], which have the capacity to express CD4 [40, 41]. A comparison of naïve and activated lymphocytes (CD4+/CD8+) revealed no substantial levels or difference in expression [42]. In many instances, the level of expression of genes that are not typically expressed in CD4+ lymphocytes, or at least the great majority, (IRAK3, IDO1, CSF1R) are greater or equivalent to the very low levels seen for most GluRs. Across all samples, GRIN3B expression was consistently higher than other genes, but present at very low levels.
To validate this observation, we measured the expression of the genes for AMPA and NMDA receptor subunits (Fig. 2) in a range of leucocytes, by TaqMan qPCR. As there is substantial literature supporting the expression of these molecules in unstimulated, circulating CD4+ T cells, we measured expression in these cells, as well as PBMC that were freshly isolated or stimulated with anti-CD3 or LPS for either 6 h or 20 h (Fig. 2a); the expression of KCNA3 was also measured as a non-ionotropic receptor, the activity of which is modulated by glutamate. The Ct values of AMPA/NMDA receptor subunit qPCRs were very high (typically more than 35 cycles) indicating very low expression, and for some subunits (e.g. GRIA1, GRIN2C) no amplification was observed. A similar pattern of gene expression was observed for myeloid cells (Fig. 2b), however for some subunits (GRIN2D, GRIN3B, GRIA3), expression was sufficient to perform δδCT analysis. For the three genes expressed, the pro-monocytic THP-1 cell line had low but reproducible expression that was higher than HL-60 cells or human MDM (Fig. 2c).
Given the low expression in CD4+ lymphocytes, it was hypothesized that another, less-prevalent cell type may be contributing to the gene expression observed in several studies where less stringent isolation techniques were used. GRIA3 expression was undetectable in circulating CD14hiCD16– classical human monocytes and negligible (Ct≈35) in circulating CD34+ haematopoietic precursors (data not shown, n = 5 healthy donors for both). To further explore the possibility that non-lymphoid leucocytes may have contaminated previous studies of PBMC-derived cells, blood DC were isolated and gene expression for GluR was measured. GRIN1, GRIN3B, GRIA1-4 expression was measured in human circulating DC subsets (pDC, cDC1, cDC2). In pDC, GRIA3 was the only detectable AMPA receptor detected (Fig. 2d and e); we did not detect GRIA1, −2 or −4 in pDC (data not shown).
AMPA subunits have isoforms, known as flip or flop isoforms, which form receptors that have differing trafficking and depolarisation kinetics in nerve cells [43–45]. When measured, the THP-1 cells and pDC expressed the flip but not the flop isoforms of GRIA3 (Fig. 2f and g). The expression of GluRs was investigated in 9 studies of THP-1 cells (Fig. 3) and 7 studies of pDC, (Fig. 4) and found to be broadly in line with our results; the 2 studies that included primary human macrophage-lineage cells [46, 47] had lower levels of GluR expression compared to THP-1 cells. As the THP-1 cells appeared to express at least one type of AMPA subunit, we measured the transcriptomic response of AMPA on TNF stimulation in these cells (Fig. 5, Supplementary Fig. 1); the inflammatory response to TNF was upregulated by AMPA stimulation.

Discussion

The Paradox of Expression
There are several potential explanations for the differences between measurements of GluR expression. Firstly, for the measurement of gene expression in isolated cells, the methods of isolation may result in different purities of cells or the composition of subsets within the total population may change. Several studies have made use of the propensity of monocytes to adhere to tissue culture plastic to deplete them from PBMC [10, 13, 15, 16, 18] in combination with other methods of isolation. Other studies relied upon density separation to isolate lymphocytes [12, 17–19, 22, 24]. It is possible that isolation methods with a higher level of stringency, while more costly and time consuming, would provide greater certainty of gene expression in the cells intended to be isolated for PCR analysis. This is evident in the data from clinical studies where density separation alone [48] or CD4 positive selection without prior depletion of CD14+ monocytes (which express low levels of CD4) [49] have higher expression of non-lymphoid genes (IRAK3, IDO1, CSF1R) compared to studies using negative isolation or flow sorting [50, 51].
Secondly, the amount of template used for PCR detection of genes has some bearing on the capacity to detect and measure the gene. In the current study, we used the equivalent of 10 ng of RNA per reaction and measured very little or no expression for some genes; in some of the literature, the template required for detection is significantly higher with 100 ng, 400 ng or 2 µg of template used per reaction [10, 13, 17]. To some degree, the use of gel electrophoresis to measure the PCR may necessitate an increase in the amount of PCR template, though this method has the drawback of being less quantitative than qPCR as it measures the PCR end product. We have previously used linearised plasmids containing sequences specific for standard curve qPCR using TaqMan assays [34, 35] and determined that a Ct value of ~ 30 is equivalent to 100 copies of template per reaction, or 1 copy per 10 cells using the estimation of 10 pg per cell [52] and 10 ng equivalent of RNA per qPCR reaction (see Supplementary Fig. 2). Although the amount of RNA per cell will vary according to cell type and activation status, our data is in broad agreement with the sequencing data. Regardless of the amount of template used to detect GluR gene expression, the relatively higher expression of ‘non-canonical’ genes indicates GluR gene expression is very low.
Finally, the mismatch between nucleic acid or protein expression and apparent functional responses in leucocytes may possibly arise when there are sub-optimal tools (such as antibodies), analytical methodology, control reagents or the use of different cell types, sources, and storage. It is, therefore, important to consider whether there could be another explanation. For example, the reported functional effects of ligands might be mediated by molecular targets which are distinct from the those on neurons. In addition, many genes may be present at extremely low levels normally, but which could be induced to produce a disproportionate amount of protein [53]. One unlikely possibility is the substantial potential for ‘molecular crosstalk’ generated by the movement and exchange of nucleic acids and proteins between cells. Molecular transfers could involve exocytosis and endocytosis, extracellular vesicles or nanotubes, in addition to the uptake of molecules resulting from cell damage or disruption. Such intercellular exchanges might explain the apparent presence of ‘foreign’ proteins in leucocytes lacking the necessary intrinsic transcriptional expression. This concept of macromolecule secretion and uptake has been specifically proposed as an important factor in the frequently observed mismatch between gene and protein expression [53].
A further point of doubt is that the concentration of glutamate in the blood is 10–100 µM [54, 55], whereas in the extracellular fluid of the brain it is 0.5–2.5 µM [56] and less than 0.4 µM is reported for cerebrospinal fluid [57]. Given that the reported proportion of leucocytes expressing GluRs was measured by surface staining flow cytometry from as low as 5% [13] to ~ 20% [22, 25], ~ 35% [23] or ~ 75% (vs a control stain of ~ 13% [10]), some thought must be given to understanding which mechanism prevents the potential continuous transfer of ions via the GluR channels due to the canonical ligand being present at a concentration of up to 2 orders of magnitude greater than in the brain; the presence of 20 mg/mL (136 µM) of glutamate in Roswell Park Memorial Institute 1640 medium (RPMI), which is commonly used for leucocyte cell culture, might also trigger significant ion transfer.
Although our analysis of GluR in different cell types and disease conditions did not reveal disease associations but rather a near absence of expression for most GluR subunit gene expression, comparing datasets from different clinical studies provides a framework for elucidating pathological mechanisms and identifying disease-specific biomarkers. This strategy can help to validate findings across independent populations but also enhance reproducibility, enabling the discovery of clinically relevant biomarkers and mechanistic insights that single-study analyses might miss.

The Relevance of pDCs to Neuroimmunology
The presence of cDCs and pDCs in the CNS has been recognised for some years[58–61]. Although there are few DCs present in the normal, uninjured CNS parenchyma, migrations occur if there is local damage from an injury or stroke, or during peripheral inflammation, infections or exposure to stress. One consequence of the increased blood levels of cytokines and chemokines in these cases is an increased permeability of the blood–brain barrier, allowing access of leucocytes to the CNS including cDCs and pDCs, with increased entry also of myeloid cells.
In the example of cerebral ischaemia, activated microglia attract the movement of DCs into the brain parenchyma, including cDCs and pDCs [62]. The pDCs are especially important in combatting viral infections since their differentiation is promoted by viral contact. As a result, MHC class I and class II proteins are induced, along with surface membrane molecules involved in T cell activation such as CD80 and CCR7. Their persistence for long periods after inflammation [63–65] reflects their neuroprotective activity. Indeed, the entry of pDCs into the CNS has been associated with reducing local inflammation, thus improving the progression and outcome from disorders such as multiple sclerosis (MS) and its rodent equivalent (experimental autoimmune encephalitis, EAE) [66, 67]. Subpopulations of cDCs are required for the generation of T cells protective against progressive inflammation in the CNS [68, 69]. The role of pDCs may then become pivotal since the IFN-β produced by them will contribute to the further influx of cDCs [70] and their promotion of Treg differentiation. The trafficking of pDCs is regulated partly by microRNA species [71] and chemokines such as CCL17 [72], with recruitment to the CNS enhanced in the early stages of EAE [73]; cytoskeletal changes are also involved [74].
GluR activation on pDCs may exert dual effects on immune regulation depending on signalling pathways. Activation can enhance type I interferon production and promote antitumor immunity, while chronic signalling might induce tolerogenic phenotypes that facilitate tumour escape or exacerbate autoimmune inflammation [75]. These context-dependent outcomes require a refined understanding of receptor subtype-specific roles and the influences of the inflammatory milieu before considering GluRs as therapeutic targets [76, 77].

GluA3 Subunits
In addition to their expression in neurons and glia in the CNS, individual AMPAR GluA3 subunits (GRIA3) are present in some non-neuronal sites such as cardiac, pancreatic, endothelial and tumour cells[78–82]. GluA3 is one of the few GluR subunits whose expression is increased by anti-inflammatory steroids, which is consistent with an immune-related activity [83]. Expression is, however, normally associated with the expression of other AMPAR subunits, especially GluA2 (GRIA2), where GluA3 has a significant influence on synaptic speed, promoting vesicle release and plasticity [84]. Increased levels of cyclic AMP increase the conductance of GluA3-operated channels, leading to a potentiation of transmission [85].
It has been demonstrated that all four AMPAR subunits (GluA1-4) can assemble into homomeric or heteromeric receptors when experimentally generated in appropriate cells [86–99]. However, there is preferential incorporation of GluA2 into the tetramer [99–101] and this subunit differs from the others as its incorporation renders the receptor Ca2+ impermeable, more stable and better able to traffic to the neuronal synapse [102]; essentially, the structural flexibility of GluA3 predisposes it to interact preferentially with other AMPAR subunits [101, 103] and, as a result, GluA3 homomers can only form in significant numbers in the absence of other subunits [94, 104, 105]. To our knowledge this is the first observation of GluA3 subunits occurring naturally in the absence of GluA2 subunits in pDC. Their expression and functional activity in pDCs may have substantial clinical implications for immunological disorders.
It is possible that glutamate itself is the major endogenous ligand for homomeric GluA3 receptors in pDCs, as it is for homomeric GluA1 combinations [106, 107]. Glutamate is released by lymphocytes such as Th17 cells in response to activation by β1-integrin and potassium channel activation, and by DCs in the process of T cell activation [108]. Similarly, glutamate activates Kv1.3 potassium channel opening, leading to increased T cell activity [109]. Since the primary role of glutamate in the CNS is the regulation of excitability, this may also contribute to its effects on leucocytes. Indeed, regulation of DC polarisation seems to represent an important form of intercellular communication since the membrane potential of cDCs is influenced by the activation of sensory receptors such as nociceptors [110, 111].

The Flip-flop Balance
An additional novel property displayed by the GluA3 subunits observed here is that they are predominantly in the ‘flip’ conformation. In excitable cells generally, especially neurons, the flip and flop phenotypes regulate channel open time, with the flop form exhibiting greater flexibility, reducing channel open time and having fast desensitisation kinetics [112–114]. In neurons, changes in the flip-flop ratio are dependent on neuronal activity, being a recognised mechanism of feedback control of excitability and plasticity in neurons and in other cell types including HEK293 [115] or Xenopus oocytes [90, 116]. Expression is also dependent on neuronal type, exemplified by the presence of flop forms primarily in small interneurons, whereas larger pyramidal projection neurons express mainly the flip variant [45, 117–123]. A shift from largely flip to flop forms occurs early in development [43, 118, 124]. This change may occur together with the phosphorylation of Lyn, a component of the transduction pathways activated by AMPAR activation, which leads to the activation of MAPK and the expression of Brain Derived Neurotrophic Factor (BDNF), a key factor in early neuronal development [29].

Conclusions
The subunit which stands out in our analysis is the GluA3 protein (gene GRIA3), a subunit of the AMPA receptor group which we have now examined in more detail. Our data suggest that this subunit is functionally active on THP-1 cells and has clinical implications for several immunological disorders and, as THP-1 cells are a leukaemia cell-line, the potential for further translational research in related disorders. GluR involvement in acute myeloid leukaemia [125] and chronic myelogenous leukaemia [126, 127] has been observed, which may in part be due to pro-survival signals associated with this class of receptors [128]. We have used comparative analyses to demonstrate that the level of gene expression is far below what would be expected from the levels of protein measured in numerous reports, and have demonstrated that the level of gene expression is often very low or undetectable by qPCR or, as seen in many datasets, if detectable then below the level of expression of genes that would not normally be expressed in the cell type analysed. The paradox of negligible gene expression of AMPA and NMDA receptors versus the intermediate protein expression reported, and the implications for this level of expression in the circulation where the concentration of glutamate is relatively high, will require a better understanding of the role of amino acid receptors in leucocytes.

Supplementary Information
Below is the link to the electronic supplementary material.