System A amino acid transporters regulate glutamine uptake and attenuate antibody‐mediated arthritis

Summary

Proliferation of rapidly dividing bone marrow‐derived cells is strongly dependent on the availability of free glutamine, whose uptake is mediated through different amino acid transporters. The sodium‐coupled neutral amino acid transporter (SNAT) family was previously reported to be associated with the development of collagen‐induced arthritis in mice. Here, we tested the hypothesis whether impairment of SNAT proteins influences immune cell function and in turn alters arthritis development. The 2‐(methylamino)isobutyric acid (MeAIB), a SNAT‐specific substrate, was used to modulate the function of SNAT proteins. We demonstrate that glutamine uptake by murine naive lymphocytes, and consequent cell proliferation, is strongly associated with system A transporters. Physiological impairment of SNAT proteins reduced the antibody‐initiated effector phase of arthritis, mainly by affecting the levels of circulating monocytes and neutrophils. MeAIB was also shown to affect the proliferation of immortalized cells, through trans‐inhibition of SNAT proteins. Based on our observations, we conclude that SNAT proteins regulate the initial stages of lymphocyte activation by regulating glutamine uptake, and that the effector phase of arthritis can be affected by non‐metabolized SNAT substrates. Most probably, metabolically active cells within both the adaptive and the innate immune systems are regulated by SNAT proteins and play a role in modifying arthritis development.

Abreviations

CAIA

collagen antibody‐induced arthritis

CIA

collagen‐induced arthritis

CII

type II collagen

DMEM

Dulbecco's modified Eagle's medium

ERK

extracellular signal‐regulated kinase

FBS

fetal bovine serum

IFN‐γ

interferon‐γ

IL‐2

interleukin‐2

LPS

lipopolysaccharide

mAb

monoclonal antibody

MeAIB

2‐(methylamino)isobutyric acid

mTOR

mammalian target of rapamycin

P/S

penicillin/streptomycin

SNAT

sodium‐coupled neutral amino acid transporter

TCR

T‐cell receptor

Introduction

Quickly dividing cells, such as bone‐marrow‐derived cells and tumour cells, have high energetic demands. These metabolic needs are strongly associated with high levels of glucose consumption and ATP production, and can be affected by the availability of amino acids, particularly glutamine. During the 1980s, the dependency of the immune system on glutamine and glutaminase activity was demonstrated,1, 2, 3 which was later linked to late T‐cell activation events.1, 2, 3, 4 Glutamine is the most abundant amino acid in the blood and different transmembrane proteins mediate its transport into the cell. At the time of the experiments by Ardawi and Newsholme,1, 2, 3 the enhanced activity of certain families of amino acid transporters in the thymus and secondary lymphoid organs had already been described.5, 6, 7, 8, 9, 10 Among them, the system A family of amino acid transporters (or sodium‐coupled neutral amino acid transporters – SNAT), has glutamine as one of the main substrates (see ref. 11 for review), and its transcription up‐regulation has recently been implicated during T‐cell activation processes.12, 13, 14

The SNAT proteins co‐transport a sodium ion together with a neutral amino acid in a unidirectional process using substrate gradients from the outer environment to the intracellular compartment. The study of these transporters can be specifically achieved by using a non‐metabolized synthetic amino acid – 2‐(methylamino)isobutyric acid (MeAIB15) – which competes with the protein's natural substrates. The members of the system A family of amino acid transporters, SNAT1, SNAT2 and SNAT4, are genetically clustered in mice, rats and humans. We have identified a quantitative trait locus of 800 kb harbouring these genes (Cia36 locus) and regulating the development of collagen‐induced arthritis (CIA)16 in mice. Here, we carried out both in vitro and in vivo experiments to demonstrate the importance of the system A family of amino acid transporters as mediators of immune cell function, arthritis development and homeostasis of immortalized cell lines. We demonstrate that glutamine uptake by immune cells is primarily carried out by SNAT proteins. The activity of SNAT proteins was shown to influence the effector functions of granulocytes, and the proliferation and homeostasis of immortalized cell lines. In addition, blocking of SNAT proteins in vivo suppressed the development of antibody‐induced arthritis.

Materials and methods

Animals

Male C57BL/10.Q mice (hereafter referred to as BQ) were bred in our mouse facility under specific‐pathogen‐free conditions and used for experiments at 10–14 weeks of age. T‐cell receptor‐β‐deficient (TCR‐β ^−/−) mice originated from The Jackson Laboratory (Bar Harbor, ME) were backcrossed to the BQ background and maintained as individual colonies in our facility. All animals used were fed a standard rodent chow and given water ad libitum. Different experimental groups were housed together to minimize experimental bias.

Ethics statement

The local ethics committee approved all animal experiments (Stockholms Norra Djurförsöksetiska Nämnd, Stockholm, Sweden). All in vivo arthritis experiments as well as in vitro experiments using samples from laboratory mice were covered by the ethics number N490/12. Anaesthesia of animals was accomplished by isoflurane inhalation, whereas killing was performed with CO₂.

Collagen antibody‐induced arthritis

Generation of collagen type II (CII) ‐specific B‐cell clones and antibody purification have been described previously.17, 18 On day 0, mice were intravenously injected with 4 mg of monoclonal antibody (mAb) cocktail containing the following mAbs: M2139 (IgG2b) binding the J1 epitope of CII; CIIC1 (IgG2a) binding the C1 epitope; CIIC2 binding the D3 epitope; and UL1 binding the U1 epitope. On day 7, mice were boosted with 25 μg of lipopolysaccharide (LPS) from Escherichia coli (administered intraperitoneally) to enhance the severity and incidence of arthritis. After mAb injection, clinical scoring of mice was performed daily based on the number of inflamed (reddish and swollen) joints. Briefly, each red and/or swollen finger or knuckle represents one point, whereas an inflamed wrist or ankle represents 5 points, resulting in a possible maximum of 15 points per limb, and 60 points per animal. MeAIB was administered daily (3 mg in PBS intraperitoneally) from 7 days before the antibody transfer. Control mice were given an identical volume of the vehicle alone.

Collagen‐induced arthritis

Collagen type II was isolated from the rat Swarm chondrosarcoma and prepared by limited pepsin digestion. The CII was maintained at 5 mg/ml in 0·1 m acetic acid. Mice were injected with 100 μg/100 μl of rat CII in complete Freund's adjuvant containing a final concentration of 25 μg of Mycobacterium smegmatis (complete Freund's adjuvant; Difco, Detroit, MI) intradermally at the base of the tail on day 0. Mice received a booster dose on day 35 of 50 μg/50 μl emulsion of rat CII in incomplete Freund's adjuvant (Difco). After the first signs of arthritis, mice were blindly scored three times per week as described above, and also monitored for weight changes.

Cell culture

Freshly isolated splenocytes from naive BQ mice were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with 5% heat inactivated fetal bovine serum (FBS), 10 mm HEPES buffer, 50 U/ml penicillin and 50 μg/ml streptomycin (P/S), in a sterile 96‐well U‐bottom plate (NUNC, Roskilde, Denmark) at 10⁶ cells/well. When assaying different glutamine concentrations, glutamine‐free DMEM (Sigma‐Aldrich, St Louis, MO) was supplemented with 5% dialysed FBS (Gibco, Invitrogen, Carlsbad, CA), 10 mm HEPES and P/S. The dipeptide l‐alanyl‐l‐glutamine (Sigma‐Aldrich) was used because of its higher stability and identical properties to glutamine alone.19 Cell stimulation was achieved with 1 μg/ml of anti‐CD3 and anti‐CD28 mAb (anti‐CD3/28; Pharmingen, BD Biosciences, San Diego, CA). All cells were incubated for the described length of time at 37°C in a 5% CO₂ incubator. When applicable, MeAIB or d‐mannitol (Sigma‐Aldrich) was added to the cultures at the indicated concentrations. T‐cell proliferation was assessed by [³H]thymidine incorporation using a liquid scintillation counter (1450 Microbeta Plus; Wallac Sverige AB, Upplands Väsby, Sweden), and by measuring cytokine levels using ELISA. Interleukin‐2 (IL‐2) production was determined in 48‐hr culture supernatants. Anti‐IL‐2 antibodies from clones JES6‐IA12 and 5H4‐biotin (homemade) were used for coating and detection, respectively. The read‐out was performed using streptavidin‐Eu (Delfia, Wallac OY, Turku, Finland). For interferon‐γ (IFN‐γ) and IL‐17A detection, 72‐hr cell culture supernatants were used. Antibodies from clones R46‐A2 and AN18‐biotin (homemade), and TC11‐15H10.1 and TC11‐8H4‐biotin (BioLegend, San Diego, CA) were used for detecting IFN‐γ and IL‐17A, respectively.

Human embryonic kidney cells (HEK293T) cells and murine CII‐specific HCQ3 T‐cell hybridomas20 maintained in DMEM containing 10% heat‐inactivated FBS (Gibco, Invitrogen) and P/S were used for assessing the effect of SNAT activity on proliferation and homeostasis. For T‐cell hybridoma assays, 5 × 10⁴ HCQ3 cells were cultured together with 5 × 10⁵ naive BQ splenocytes and 100 μg/ml of rat CII in DMEM containing 5% heat‐inactivated FBS, 10 mm HEPES, and P/S in a 96‐well plate. Interleukin‐2 production was measured using ELISA upon 48‐hr stimulation, as described earlier.

Amino acid uptake assay

Splenocytes, 10⁶/well, were stimulated with 3 μg/ml of concanavalin A (Sigma‐Aldrich) for 24 hr at 37°C in glutamine‐free medium. After 24 hr, cells were washed in glutamine‐free medium and re‐suspended in Krebs–Ringer solution (119 mm NaCl, 2·5 mm KCl, 1 mm NaH₂PO₄, 2·5 mm CaCl₂.2H₂O, 1·3 mm MgCl₂.6H₂O, 20 mm HEPES, 11 mm d‐glucose) containing 0·2 mm [³H]glutamine (PerkinElmer Sverige AB, Upplands Väsby, Sweden) with a radioactivity of 0·5 μCi. For the establishment of Na⁺‐free conditions, NaCl and NaH₂PO₄ were substituted in the Krebs–Ringer buffer by the respective choline reagents. Cells were incubated for 60 min at 37°C and the amino acid uptake was measured at different time‐points. At each time‐point, cells were collected, washed five times with the respective ice‐cold Krebs–Ringer buffer and then harvested. The amino acid incorporation was read using a liquid scintillation counter (1450 Microbeta Plus; Wallac Sverige AB).

Flow cytometry

At the peak of disease in collagen antibody‐induced arthritis (CAIA) (day 11), blood samples were collected and cell populations were evaluated by flow cytometry. Anti‐mouse CD3ε (PerCp‐Cy5.5, 145‐2C11; BioLegend), CD4 (Qdot605, RM4‐5; Life Technologies, Grand Island, NY), CD8α (Brilliant Violet 650, 53‐6.7; BD Biosciences), CD11c (PE‐Cy7, HL3; BD Biosciences), CD11b (APC, M1/70; eBioscience, San Diego, CA), Ly6C (FITC, AL‐21; BD Biosciences), Ly6G (PE, IA8; BD Biosciences) and LIVE/DEAD near‐IR dead cell stain kit (Molecular Probes, Eugene, OR, Life Technologies) were used. Mice given LPS without mAb transfer were used as controls. Flow cytometry acquisition was performed in an LSR‐II (BD) and analysed using flowjo software (TreeStar, Ashland, OR). Gating strategy is presented in the Supplementary material (Fig. S1).

Nitric oxide determination

Peritoneal macrophages from naive mice were collected by peritoneal lavage using ice‐cold PBS. Cells were re‐suspended in DMEM supplemented with 5% heat‐inactivated FBS, 10 mm HEPES and P/S, plated at 10⁶/well in a 96‐well plate and stimulated with 1 μg/ml of LPS in the presence or absence of MeAIB. After 20 hr of culture, 50 μl of Griess reagent (Sigma‐Aldrich) was added to an equal volume of culture supernatant and nitric oxide (NO) production was measured at 520 nm. A sodium nitrite standard curve was used for quantification of NO.

Quantitative real‐time PCR

Primers for the genes Slc38a1, Slc38a2, Slc38a4, Mapk3, Mapk1, Mtor and Actb were designed for species‐specific mRNA sequences, placed in exon–exon junctions and aligned for gene specificity (UCSC Genome Bioinformatics, mouse and human assembly NCBI m38 and h38, respectively). Total RNA isolation was performed using PureLink RNA Mini Kit (Ambion, Thermo Fisher Scientific, Waltham, MA, Life Technologies, Inc., Foster City, CA), and converted to cDNA with a high‐capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA). Quantitative real‐time PCR (qRT‐PCR) was performed in a 7900HT Fast Real‐Time PCR system (Applied Biosystems) using SYBR Green (Applied Biosystems) as the fluorescence source. Gene expression C _t values were all normalized to the housekeeping gene (Actb) expression C _t value (∆C _t), and calibrated to a referred sample ∆C _t value (∆∆C _t), for comparative analysis. Expression fold is presented as $2^{-\mathrm{\Delta }\mathrm{\Delta }{C}_{\mathrm{t}}}$.

Statistical analysis

All mice and mouse‐derived samples used in the experiments were included in the statistical calculations. Arthritis severity was analysed by two‐way analysis of variance with repeated measures followed by post‐test, whereas a Fisher's exact test was used for incidence. For all in vitro assays a Man–Whitney U‐test was used for comparison of significance between two different groups, and Kruskal–Wallis test was used for analysis of the whole set of results. In all figures showing in vitro results, the statistical representations are with regard to Kruskal–Wallis analysis. All statistical calculations were performed using graphpad prism 5.0d (GraphPad Software, San Diego, CA). Significance was considered when P < 0·05 for a 95% confidence interval.

Results

SNAT1 and SNAT2 proteins are up‐regulated after T‐cell activation

We have previously identified a quantitative trait locus in the mouse (Cia36) harbouring the genes coding for the system A family of amino acid transporters (Slc38a1, Slc38a2 and Slc38a4, corresponding to SNAT1, SNAT2 and SNAT4 proteins) as a regulator of arthritis (CIA16). Such Na⁺‐dependent amino acid transporters had earlier been shown to affect lymphocyte proliferation.9, 10, 21 Recently, SNAT proteins were implicated by others in the mediation of metabolic changes during T‐cell activation.12, 13, 14 Here, we evaluated the effect of these amino acid transporters on immune functions and inflammatory responses.

We started by evaluating the gene expression of system A proteins in different lymphoid organs and in stimulated lymphocytes. In naive animals, expression of SNAT1 was homogeneous between different immunological compartments (Fig. 1a). In contrast, SNAT2 expression was significantly higher in the thymus, an organ where constant T‐cell activity takes place and high amounts of glutamine are consumed.22 We could not detect the expression of SNAT4 in any of the investigated lymphoid tissues. Nevertheless, the qualitative analysis of the PCR was positively assessed in liver‐derived RNA samples, where SNAT4 is known to be preferentially expressed23 (data not shown). Upon T‐cell activation by anti‐CD3/28 stimulation, both SNAT1 and SNAT2 genes were up‐regulated over 100‐fold within the first 2 hr, in control conditions (DMEM; Fig. 1b,c). This up‐regulation coincided with a 30‐fold increase in mammalian target of rapamycin (mTOR) expression (Fig. 1d), a major regulator of cell growth (see ref. 24 for review). The effect was similarly observed for extracellular signal‐regulated kinase 1 (ERK1) expression (Fig. 1e), supporting previous observations that glutamine uptake during lymphocyte activation induces expression of ERK proteins.12 After 5 and 24 hr of stimulation the expression of these genes was found to be decreased to basal levels. In the presence of MeAIB, the expression of mTOR and ERK1 was unaffected when comparing each of the time‐points to baseline or between themselves, hence demonstrating the non‐metabolizing property of this synthetic amino acid. Interestingly, in the presence of MeAIB, the expression of SNAT genes was maintained significantly higher than control conditions at 5 and 24 hr of stimulation, which may result from a constant, although residual, Na⁺ influx induced by MeAIB. Taken together, the gene expression patterns observed in the presence of MeAIB suggest that system A transporters are relevant for T‐cell activation. In a scenario where the presence of MeAIB reduces the activity of SNAT proteins and consequently all cellular processes downstream, the energetic demands between naive and MeAIB‐treated cells will differ. Hence, under normal conditions, but not in the presence of MeAIB, the initial stimulatory signals (2 hr) will induce a rapid de novo synthesis of SNAT proteins to supply the cell's energy demands. As these are being poorly compensated by the presence of MeAIB, both SNAT1 and SNAT2 genes are maintained and up‐regulated with time, whereas their expression subsides to basal levels in normal conditions after 5 hr of stimulation.

Figure 1.

T‐cell activation up‐regulates the expression of sodium‐coupled neutral amino acid transporter 1 (SNAT1) and SNAT2. (a) Expression of Slc38a1 and Slc38a2 genes in different immunological organs. (b–e) Expression of Slc38a1, Slc38a2, Mtor and Mapk3 genes after anti‐CD3/28 stimulation of naive splenocytes in the absence (Dulbecco's modified Eagle's medium; DMEM) or presence of 10 mm 2‐(methylamino)isobutyric acid (MeAIB) (n = 5). Data were normalized to the expression levels of non‐stimulated cells. Data shows mean ± SEM and are representative of three independent experiments, with a total of 18 mice used per group. Kruskal–Wallis test was used and data were considered significant when P < 0·05 for a 95% confidence interval. *P < 0·05.

Glutamine uptake by immune cells occurs via SNAT proteins in the presence of Na⁺

In the absence of extracellular glutamine, T‐cell proliferation is restricted to the intracellular pool of available nutrients, and therefore extremely reduced (Fig. 2a).25, 26, 27 We then assessed how this dependency on glutamine involves SNAT proteins by quantifying the uptake of glutamine in stimulated cells in the presence or absence of MeAIB. The data clearly demonstrate that glutamine uptake by T cells is essentially dependent on system A transporters (Fig. 2b). When Na⁺ ions were no longer present in the medium, similar levels of glutamine uptake were observed (Fig. 2c). This indicates that immune cells have alternative ways of supplementing the amino acid needs and can adapt the usage of membrane transporters in accordance with their environment. Nevertheless, under normal physiological conditions, the essential glutamine uptake by activated lymphocytes occurs preferentially via SNAT proteins. Similarly, we evaluated the effect of MeAIB on T‐cell activation. In a dose‐dependent manner, MeAIB significantly reduced T‐cell proliferation as indicated by the levels of [³H]thymidine incorporation (Fig. 2d). Moreover, when used at a 20 mm concentration, MeAIB reduced the levels of IL‐2, IFN‐γ and IL‐17A production by 90%, 83% and 50%, respectively (Fig. 2e). Because MeAIB can affect the osmolarity when used at relatively high concentrations and consequently alter the pattern of pro‐inflammatory cytokine production28 (unpublished observations), we determined the effect of d‐mannitol in the culture of stimulated spleen cells. Contrary to our observations with MeAIB, the presence of d‐mannitol did not reduce lymphocyte proliferation at any of the tested concentrations (see Supplementary material, Fig. S2), suggesting that change in osmolarity per se may not be sufficient to alter lymphocyte functions.

Figure 2.

Glutamine‐dependent T‐cell proliferation is associated with system A family of amino acid transporters. (a) Cytokine secretion by splenocytes at different concentrations of glutamine after anti‐CD3/28 stimulation (n = 5). Glutamine uptake in the presence (b) or absence (c) of Na⁺ after T‐cell stimulation with or without 2‐(methylamino)isobutyric acid (MeAIB) in the culture medium (n = 4). [³H]Thymidine incorporation (d) and cytokine secretion (e) by stimulated lymphocytes in the presence of MeAIB (n = 6). Data shows mean ± SEM and are representative of four independent experiments, with a total of 20 to 22 mice used. Kruskal–Wallis (a, d and e) and two‐way analysis of variance (b and c) tests were used and data were considered significant when P < 0·05 for a 95% confidence interval. *P < 0·05; **P < 0·01; ***P < 0·005.

Reduced function of SNAT proteins lowers arthritis symptoms

To evaluate the effect of system A transporters on inflammatory responses, we induced arthritis using collagen type II‐specific mAbs (CAIA). We administered either MeAIB or PBS to different groups of animals. The systemic presence of MeAIB throughout the disease course resulted in a significant reduction of both arthritis severity and prevalence (Fig. 3a,b). At the peak of the disease (day 11), a lower level of circulating neutrophils (CD11b⁺ Ly6G⁺) but higher levels of CD11b⁺ Ly6C^lo monocytes (Fig. 3c) were observed in the MeAIB‐administered group. No differences in T‐cell (defined as CD3⁺ CD4⁺ or CD3⁺ CD8⁺), natural killer T cell (CD3⁺ NK1.1⁺) or CD11b⁺ Ly6C^hi monocyte levels were observed. To relate the granulocyte levels to the observed disease phenotypes, we evaluated the capacity of NO production by peritoneal macrophages in the presence of MeAIB. In this situation, the competition for natural substrates induced by the presence of MeAIB reduced the capacity of the macrophage cells to produce NO (Fig. 3d). As observed for T cells (Fig. 1), the data suggest a similar requirement of SNAT proteins by different inflammatory and metabolically active cells to achieve their full activation status. Nevertheless, using the same MeAIB treatment in CIA (see Supplementary material, Fig. S3) and experimental autoimmune encephalomyelitis (data not shown), which are mostly T‐cell‐dependent diseases, no significant effects were seen. In both cases, the systemic administration of MeAIB resulted solely in a mild delay of the disease onset. Nevertheless, a contributing role of T cells in the CAIA model has been suggested, as the development of CAIA in TCR‐β ^−/− mice shows reduced disease symptoms. In this scenario, the full beneficial therapeutic effects of MeAIB can no longer be observed (see Supplementary material, Fig. S4).

Figure 3.

Administration of 2‐(methylamino)isobutyric acid (MeAIB) attenuates collagen antibody‐induced arthritis (CAIA) by affecting circulating polymorphonuclear cells. Mice were given 3 mg/150 μl of MeAIB intraperitoneally daily from day –7 until day 18. At day 7, 25 μg of lipopolysaccharide (LPS) was given intraperitoneally to all mice. Development of arthritis was evaluated by visual scoring of the inflamed joints (a) and prevalence (b). (c) Circulating levels of neutrophils (CD11b⁺ Ly6G⁺) and monocytes (CD11b⁺ Ly6C^lo) in the peripheral blood of mice at day 11 after anti‐collagen II (CII) antibody transfer. The data shown are representative of two independent experiments using a total of 26 PBS‐treated mice and 25 MeAIB‐treated mice. Gating strategies are presented in the Supplementary material (Fig. S1). (d) Effect of MeAIB on NO production by in vitro stimulation of naive peritoneal macrophages with LPS. Data shows mean ± SEM. Kruskal–Wallis (c and d), two‐way analysis of variance (a) and Fisher's exact test (b) were used and data were considered significant when P < 0·05 for a 95% confidence interval. *P < 0·05; **P < 0·01; ***P < 0·005; n.s., not significant.

System A transporters affect proliferation of immortalized cell lines

In normal cells, SNAT proteins can be trans‐inhibited by their own substrates and repressed by certain metabolites, such as glutamate. Apart from immune cells, tumour cells are also major consumers of glutamine (see refs 29 and 30 for review). However, an interesting particularity of tumour cells and glutamine uptake via SNAT proteins is the fact that these cells are insensitive to repression, and are therefore described as presenting a de‐repressed system A.31 Hence, we speculated that a way of affecting immortalized cell metabolism and growth would be through trans‐inhibition of SNAT proteins. To assess this we seeded HEK293T cells with different concentrations of MeAIB and evaluated the cells' proliferation capacity as well as expression of SNAT proteins (Fig. 4a–d) after 72 hr. At a concentration of 50 mm, MeAIB was able to reduce the total cell numbers by 32%. To understand how this affected SNAT proteins, we determined their genetic expression at different MeAIB concentrations. Slc38a1 was only affected by MeAIB concentrations above 20 mm (Fig. 4b), whereas Slc38a2 expression was not affected by the presence of MeAIB. Using a different system, we stimulated HCQ3 T‐cell hybridomas, recognizing the T‐cell epitope of CII (amino acids 260–27020) with its antigen using different concentrations of MeAIB. Comparable to what was observed in freshly isolated splenocytes (Fig. 1), the activation of HCQ3 T‐cell hybridomas was compromised by the presence of MeAIB in a concentration‐dependent manner (Fig. 4e).

Figure 4.

System A substrates affect proliferation of immortalized cell lines: 2·5 × 10⁵ HEK293T cells were seeded per 25‐cm² culture flask in Dulbecco's modified Eagle's medium (DMEM) supplemented with different concentrations of 2‐(methylamino)isobutyric acid (MeAIB). After 72 hr, the total number of cells was assessed (a) and total RNA samples were collected for quantitative expression of Slc38a1 and Slc38a2 genes (b, c; n = 4). (d) Visualization of HEK293T cells under normal culture conditions or after addition of 50 mm MeAIB. (e) Interleukin‐2 (IL‐2) production by HCQ3 T‐cell hybridomas stimulated with rat CII for 48 hr in different concentrations of MeAIB (n = 5). The data shows mean ± SEM and are representative of three independent experiments. Kruskal–Wallis test was used and data were considered significant when P < 0·05 for a 95% confidence interval. *P < 0·05; **P < 0·01.

Discussion

The identification of the Cia36 locus as a regulator of arthritis in mice16 indicated that solute carrier proteins transporting neutral amino acids regulate cellular metabolism and thereby affect immune cell functions and inflammatory outcomes. Here, we show that blocking of SNAT proteins in vivo suppressed the development of antibody‐induced arthritis, most likely through effects on neutrophil numbers and macrophage functions.

It is well known that immune responses cannot be mounted properly in the absence of extracellular glutamine.1, 2, 3 SNAT proteins, encoded in the Cia36 locus, and system ASC represent the main protein systems by which glutamine uptake can occur in cells.31

Even though multiple studies have shown the association of Na⁺‐dependent amino acid transport with lymphocyte activation,5, 6, 7, 8, 9, 10, 21 the relevance of SNAT proteins in these processes has not been fully described. More recent studies have demonstrated that T‐cell activation leads to up‐regulation of SNAT proteins, in particular SNAT1, but they failed to identify their relevance in T‐cell activation.12, 13, 14 Our data clearly demonstrate that, in the presence of Na⁺, the glutamine uptake by stimulated lymphocytes is strongly dependent on the system A transporters. However, in the absence of Na⁺ the rate of glutamine uptake is not compromised. This observation indicates that lymphocytes have the capacity to adapt to their environmental conditions, and are able to use Na⁺‐independent amino acid transporter systems to satisfy their metabolic needs, such as system L.13 As it was not our goal to identify which other families of amino acid transporters could replace SNAT proteins in the import of glutamine, we cannot assure which of those systems have a predominant role in glutamine uptake. Nevertheless, Borghetti et al.21 have previously shown that systems A and ASC are functional at different stages of lymphocyte activation. More recently, just like system ASC, the Na⁺‐independent system L has been found to be up‐regulated in T cells upon TCR engagement, and consequently coordinates the uptake of neutral amino acids.13 Although systems ASC and L can replace and complement SNAT proteins, we clearly demonstrate that system A proteins are directly associated with the capacity of naive lymphocytes to produce and secrete the cytokines IL‐2 and IFN‐γ, with a minor effect on IL‐17A. These data are in line with recent observations demonstrating that system ASC was crucial for the development of Th1 and Th17, whereas IL‐2 and Treg cell levels were not affected.32 As IL‐2 is required for T‐cell proliferation, the association of SNAT proteins with lymphocyte stimulation may involve rather initial stages of activation and/or Treg cell activation, whereas system ASC would then affect later stages of activation and lineage differentiation of pro‐inflammatory T cells.

When challenged with an inflammatory trigger (anti‐CII mAbs followed by LPS), mice treated with MeAIB developed significantly less arthritis. This observation suggests that MeAIB can trigger trans‐inhibition of system A proteins in vivo and so affect the competence of immune cells. The reduced arthritis phenotype was associated with lower levels of circulating neutrophils and higher Ly6C^lo monocytes in the blood. The observed correlation between circulating innate cells and the disease phenotype is supported by previous reports demonstrating that both CAIA33 and K/BxN serum transfer‐induced arthritis34, 35 are to a large extent dependent on neutrophil infiltration into the synovium. Depletion of neutrophils using anti‐Gr1 antibody abolishes the severity of antibody‐mediated arthritis in mice.33, 36 Recognition of immune complexes by FcγRs present on the surface of neutrophils leading to secretion of pro‐inflammatory cytokines and proteases might be the plausible mechanism for the involvement of neutrophils during joint inflammation at the effector phase of arthritis. Neutrophils have previously been identified as an essential component in the initiation and progression of arthritis34 as well as a source of IL‐17 in the K/BxN serum transfer model.37 Contrary to Ly6C^hi, Ly6C^lo monocytes are known to express FcγRIV, which is essential for IgG2a‐ and IgG2b‐mediated effector functions.38 The observed higher levels of these cells in the blood of MeAIB‐treated mice may result in more efficient clearance of the administered pathogenic anti‐CII mAb and subsequently in a lower arthritis severity. Moreover, we show here that the presence of MeAIB reduces the LPS‐stimulation capacity of peritoneal cells in vitro.

Although we observed a clear reduction in IL‐2 and IFN‐γ levels by stimulated T cells when blocking SNAT proteins (Fig. 2), we did not observe any major phenotypic alteration during CIA (see Supplementary material, Fig. S3) when mice were systemically administered with MeAIB. Several mechanisms could explain these observations. First, the relevance of SNAT proteins in vivo may be restricted to cell populations that are mostly affecting the onset of the disease (naive T cells). Once a few cells trigger the disease, continued administration of MeAIB will not affect effector T cells or memory T cells and consequently CIA will develop equally in both groups. Along this line, in vitro experiments comparing activation of effector/memory T cells to that of naive T cells in the presence of MeAIB and other specific amino acid transporter inhibitors would be informative to disclose the role of SNAT proteins in experienced T cells. As it is, we cannot exclude the possibility that, in contrast to what was clearly observed in Figs 1 and 2 and discussed by others,4, 5, 6, 7, 8, 9, 10 the absence of a therapeutic effect from MeAIB in the CIA model may be a result of the non‐involvement of SNAT proteins in the regulation of T‐cell function in this model. Alternatively, the later disease onset of CIA compared with CAIA (average onset at day 32 and 2, respectively) means that a longer exposure to SNAT blockers is required in the former model. This scenario of substrate deprivation via SNAT proteins may induce the activation of other families of amino acid transporters and consequently the transport of glutamine, essential for T‐cell activation. This alternative supply of glutamine would activate T cells at a later stage with consequent delay of onset as observed in the Supplementary material (Fig. S3), but with identical disease severity.

Surprisingly, treatment of CAIA with MeAIB using TCR‐β ^−/− mice (see Supplementary material, Fig. S4) did not significantly reduce the arthritis phenotype, as in wild‐type animals. We cannot exclude the possibility that the absence of the therapeutic effect induced by MeAIB in these settings is a result of a general reduction of arthritis in TCR‐β ^−/− mice. Nevertheless, as the development of CAIA has been reported to be partially affected by T cells,39, 40, 41 our data suggest that the reduction of cytokine levels by blockage of SNAT proteins (namely IL‐2) is likely to preferentially affect certain T‐cell populations and so contribute to the observed results. Besides arthritis, the involvement of system A transporters has been implicated in diabetes42 as well as in patients with systemic lupus erythematosus.43

Generally referred to as a non‐essential amino acid, the cellular demands of glutamine during periods of rapid cell growth or illness makes it a conditionally essential amino acid. The phenomenon is not only restricted to lymphocytes or fibroblasts, but is also extendable to cancer cells. In culture, immortalized cell lines are eager glutamine consumers29 and its uptake is primarily mediated by systems A and ASC.31 Several studies indicate that over‐expression of these systems is associated with a poorer diagnosis of different types of cancer.44, 45, 46, 47, 48, 49, 50, 51 Here, we show that de novo synthesis of SNAT1 proteins, but not SNAT2, occurs upon effective blocking of system A proteins, in HEK293T cells. The unaltered expression of SNAT2 suggests that, although used at relatively high concentrations (20–50 mm), MeAIB did not induce osmotic stress in these cells. This conclusion is based on previous studies demonstrating that de novo synthesis of SNAT2 associates with the regulation of cellular osmotic stress (see refs 52, 53 and 54 for review).

The reduced proliferation capacity of HEK293T and reduced activation of HCQ3 hybridoma cells in the presence of MeAIB brings back the idea of using metabolic blockers as a cancer therapy. A couple of glutamine analogues have indeed been used in the clinic as anti‐cancer drugs (reviewed in ref. 55). However, both the inconclusive benefits and the degree of secondary effects observed in treated patients suggest that new molecules need to be developed. Although highly specific for the system A proteins, MeAIB is rapidly taken up by the kidney,56, 57 and is therefore far from an optimal solution. Moreover, our in vitro data indicate that the effect of MeAIB is limited and that the metabolic needs of cells may be compensated by concomitant expression of different families of amino acid transporters. Nevertheless, the expression of SNAT proteins in different types of tumour cells has led to the use of MeAIB as a cell tracer in certain cancer patients.58, 59 Similarly, the use of MeAIB or other glutamine analogues affecting SNAT proteins and cell metabolism could be considered in therapeutic approaches of autoimmunity. However, the results would probably be similar to those previously observed in cancer therapy. The effect of presently used therapies in rheumatoid arthritis, such as methotrexate,60 or other broad therapeutic approaches used in CIA (e.g. oral tolerance with CII61), on the activity of SNAT proteins is largely unknown. It is likely that either of these strategies can be associated with a reduced activity of SNAT proteins, because both therapeutic interventions affect cell proliferation. Nevertheless, a therapeutic approach that could solely affect the metabolism of immune cells, particularly the initial stages of activation involving SNAT proteins, is still to be developed.

To conclude, we demonstrate that naive lymphocytes preferentially use SNAT proteins for the uptake of extracellular glutamine. Trans‐inhibition of SNAT proteins in immortalized cell lines resulted in a reduction of cell proliferation, although to a lesser extent than fresh lymphocytes. Also, in vivo administration of the SNAT inhibitor MeAIB significantly diminished the severity of antibody‐mediated arthritis, most likely through its effect on metabolically active inflammatory cells.

Author contributions

BR, DV, EA, KSN and RH designed the study and wrote the paper. BR, DV and KSN performed the experiments and analysed the results.

Disclosures

The authors declare that they have no competing interests.

Supporting information

Figure S1. Gating strategy for analysis of peripheral blood cell populations presented in Fig. 2(c).

Figure S2. Small increase in osmolarity does not affect cytokine production. 10⁶ naive spleen cells were stimulated in vitro with anti‐CD3/28 monoclonal antibody in the presence of increasing concentrations of d‐mannitol. Interleukin‐2 (IL‐2) (48 hr), interferon‐γ (IFN‐γ) and IL‐17A (72 hr) production was determined in the culture supernatants by ELISA (n = 5). The data shows mean ± SEM and are representative of two independent experiments. Kruskal–Wallis test was used and data were considered significant when P < 0·05 for a 95% confidence interval.

Figure S3. Systemic administration of 2‐(methylamino)isobutyric acid (MeAIB) does not affect the course of collagen‐induced arthritis. Mice were given 3 mg/150 μl MeAIB daily by intraperitoneal route with or the same volume of saline solution during the whole course of disease (day –7 to 60). (a) Mean arthritis score of sick animals (only animals that developed arthritis at any time‐point during the experiment) and disease prevalence. (b) Anti‐collagen type II antibody titres at day 14 and 50 post‐immunization. The data shows mean ± SEM. Kruskal–Wallis (b), two‐way analysis of variance and Fisher's exact test (a) were used and data were considered significant when P < 0·05 for a 95% confidence interval.

Figure S4. Systemic administration of 2‐(methylamino)isobutyric acid (MeAIB) does not significantly affect the course of antibody‐mediated arthritis in the absence of T cells. T‐cell receptor‐β‐deficient (TCR‐β ^−/−) mice were given 3 mg/150 μl MeAIB or saline solution daily during the whole course of disease. Mean arthritis score and disease prevalence are shown. The data show mean ± SEM. Two‐way analysis of variance and Fisher's exact test were used and data were considered significant when P < 0·05 for a 95% confidence interval.