Differential Requirements for l-Citrulline and l-Arginine during Antimycobacterial Macrophage Activity

Leukocytes are readily equipped to defend against invading pathogens. One such defense mechanism, NO, is produced by macrophages (MΦs) in high concentrations in response to pathogen associated molecular patterns in combination with IFN-γ produced by innate and/or adaptive effector lymphocytes. This defensive mechanism is crucial in mice infected with mycobacterial species. A loss of function mutation in Nos2, which encodes inducible NO synthase (iNOS), results in increased lethality to Mycobacterium tuberculosis and M. bovis bacillus Calmette–Guérin (BCG) (1, 2). The contribution of iNOS in human immunity remains less defined than in mice. For instance, many studies showed that human MΦs make significantly less NO than do mouse MΦs in similar in vitro conditions (3). However, in vivo evidence suggests some protective role for iNOS in human tuberculosis (TB). Genome-wide association studies link NOS2 with an increased incidence of TB (4–8), although the expression and/or activity of iNOS resulting from these polymorphisms are unknown. Still other investigators reported that iNOS and NO production in TB patients correlates with less severe disease, and NO production by human alveolar MΦs ex vivo inhibits mycobacterial growth (9–15).

iNOS catalyzes the generation of NO in MΦs by converting the amino acid l-arginine to l-citrulline. This process can be interrupted by arginase activity, which hydrolyzes l-arginine to l-ornithine and urea, thereby competing with iNOS for bioavailable l-arginine. Similar to iNOS, type 1 arginase (Arg1) is induced in human MΦs within TB granulomas and in the lungs of mice infected with M. bovis BCG (16–19), suggesting substrate competition for l-arginine in vivo. Indeed, mice lacking Arg1 in MΦs make more NO and clear M. tuberculosis faster than do control mice (20).

As l-arginine becomes limiting, MΦs replenish intracellular stores of l-arginine via l-citrulline metabolism (21). We recently described that mice unable to convert l-citrulline to l-arginine in their hematopoietic compartment produce less NO and are more susceptible to M. tuberculosis and M. bovis BCG infection (22). These data led us to ask whether synthesized l-arginine had similar metabolic consequences as imported l-arginine in mycobacteria-infected MΦs. In this article, we describe that synthesized l-arginine, as compared to imported l-arginine, is a preferred substrate for NO production rather than conversion to l-ornithine. Moreover, we show that the combination of l-arginine and l-citrulline provides MΦs with an increased antimycobacterial potential compared with either amino acid alone.

C57BL/6 and Asl^flox/flox;Tie2-cre mice (and controls) were bred within the Division of Veterinary Services at Cincinnati Children’s Hospital Medical Center. Strains were obtained from The Jackson Laboratory (C57BL/6J, 000664; B6.129S7-Asl^tm1Brle/J, 018830; B6.Cg-Tg(Tek-cre)1Ywa/J, 008863). Procedures were approved by the Institutional Animal Care and Use Committee at Cincinnati Children’s Hospital Medical Center.

Complete DMEM (C-DMEM; 400 μM l-arginine, 10-013-CV, Cellgro; Corning Life Sciences) was prepared by adding bovine calf serum (SH30073.03; Thermo Fisher Scientific) to 10% and by adding penicillin/streptomycin (15140-122; Gibco, Life Technologies) to 1% final concentration. l-arginine–free (R-free) C-DMEM (A14431-01, Life Technologies, Life Technologies) was prepared by adding dialyzed FBS (35-071-CV, Cellgro; Corning Life Sciences) to 10% final concentration. l-arginine and l-citrulline were prepared at a stock concentration of 100 mM in sterile water. l-arginine and l-citrulline stocks were added to R-free C-DMEM to concentrations noted in the text. Cells were cultured in a humidified atmosphere at 37°C plus 5% CO₂.

Mice were administered 1 ml sterile thioglycollate (R064710; Thermo Fischer Scientific) by i.p. injection. After 4 d, peritoneal cells were collected by lavage. Following red cell lysis, cells were plated on tissue culture plastic in C-DMEM. C-DMEM and nonadherent cells were aspirated after 4 h, and fresh R-free C-DMEM was added with l-arginine and/or l-citrulline.

Bone marrow was flushed from femurs and tibias and resuspended in bone marrow C-DMEM (BM-C-DMEM) (10% FBS, 1% penicillin/streptomycin, 40 ng/ml human M-CSF; a gift from P. Murray, St. Jude Children’s Research Hospital, Memphis, TN). Bone marrow was plated on tissue culture plastic (2 × 75 cm² flasks/mouse, 15 ml/flask) for 5–7 d. A total of 5 ml fresh BM–C-DMEM was added every 48 h until harvest. Bone marrow–derived MΦs (BMDMs) were collected by scraping, and cells were plated on tissue culture plastic in BM–C-DMEM. The following morning, medium was aspirated, and fresh R-free C-DMEM was added with l-arginine and/or l-citrulline. In all experiments, initial culture in l-arginine containing C-DMEM did not result in appreciable amounts of extracellular or intracellular l-arginine or l-citrulline after reconstituting with R-free C-DMEM (data not shown), so this method was used to conform to previously published methods (19, 20, 22).

M. bovis BCG Pasteur strain (gift from P. Murray) was cultured in 7H9 (M0178; Sigma-Aldrich) plus OADC (R450605; Thermo Fisher Scientific) containing 0.05% Tween-80 (P4780; Sigma-Aldrich) at 37°C with shaking ∼ 50 rpm. Cultures were used for infection within 5–14 d following thaw. Bacilli were washed twice with sterile PBS and passed through a 40 μm strainer. A₆₀₀ = 0.3 by spectrophotometric analysis consistently resulted in 1 × 10⁷ CFU/100 μl volume. Multiplicity of infection (MOI) ranged from one to five bacilli/MΦ.

A total of 5 × 10⁶ to 6 × 10⁶ cells was washed twice with PBS and then lysed in 1 ml ice-cold methanol. Cell debris was pelleted, and methanol lysates were collected.

Supernatants from the above cells were collected and mixed (1:1, v/v) with ice-cold methanol. Amino acids were measured by liquid chromatography-tandem mass spectrometry (LC-MS/MS) with selected-reaction monitoring and with the use of stable isotopic-labeled amino acids as internal standards. Samples were analyzed with the LC20AD HPLC system (Shimadzu) coupled to the TSQ Quantum Ultra Triple Quadrupole Mass Spectrometer (Thermo Scientific). Chromatographic separation of amino acids was achieved on a 50 × 2.1 mm Atlantis HILIC column (Waters). A gradient mobile phase was used with a binary solvent system, which changed from 90% mobile phase B (95% acetonitrile/5% water/0.1% formic acid/1.5 mM ammonium formate) to 28% mobile phase A (water/0.1% formic acid/1.5 mM ammonium formate) at a flow rate of 0.6 ml/min. The total run time was 10 min, and the injection volume was 10 μl. The optimal signal for the analytes was achieved in positive ion mode with the following instrument settings: spray voltage: 4 kV; sheath gas pressure: 35; auxiliary gas flow: 10; and capillary temperature: 350°C. Argon was used as the collision gas. Data were acquired and processed with Xcalibur 2.2 (Thermo Scientific).

RNA was collected using TRIzol reagent (Invitrogen). cDNA was synthesized using SuperScript II reverse transcriptase (Invitrogen) and analyzed by SYBR Green (Applied Biosystems) quantitative real-time PCR.

Protein RIPA lysates were separated by Tris-HCl buffered 4–15% gradient SDS-PAGE, followed by transfer to Protran membranes. Membranes were analyzed by Ponceau S staining and then blocked in 3% milk in TBS plus 0.05% Tween, followed by immunoblot to detect argininosuccinate lyase (Asl; PA5-22300; Thermo Scientific) and Grb2 (610112; BD Biosciences).

Equal volumes of cell culture supernatant and Griess reagent were mixed in a 96-well plate. Sodium nitrite (237213; Sigma-Aldrich) was used as a standard. Absorbance values were measured at A₄₉₂ using a DTX-880 Multimode plate reader and detection software (Beckman Coulter).

Medium was removed, and cells were washed with sterile PBS, followed by aspiration to remove extracellular M. bovis BCG. Following the wash, MΦs were lysed in water containing 1% IGEPAL (18896; Sigma-Aldrich) for 10 min at 37°C. Serial dilutions were plated on 7H10 agar (262710; BD Biosciences) containing OADC. Agar plates were incubated at 37°C for 14–18 d prior to counting colonies.

LPS from Escherichia coli 0111:B4 (L3012), l-arginine (A8094), and l-citrulline (C7629) were obtained from Sigma-Aldrich; IFN-γ (14-8311-63), IL-4 (14-8041-62), and IL-10 (14-8101-62) were obtained from eBioscience; S-(2-boronoethyl)-l-cysteine (BEC) was obtained from Cayman Chemical; and 1,2,3,4,5-¹³C₅ l-citrulline, 1,2,3,4,5-¹³C₅ l-arginine, ¹³C₅ l-ornithine, ¹³C₆ [¹⁵N]₄ l-arginine, ureido ¹³C l-citrulline, and 1,2-¹³C₂ l-ornithine were obtained from Cambridge Isotope Laboratories. The final concentrations used in cell culture were: LPS, 100 ng/ml; IL-4, 10 ng/ml; IL-10, 10 ng/ml; IFN-γ, 2 ng/ml; and BEC, 250 μM.

The following primers were used: Asl (forward: 5′-ACTCTTGGAGGTGCAGAAGC-3′, reverse: 5′-AGTAGCTCCCGGTCCACAC-3′) and Gapdh (forward: 5′-GGTGCTGAGTATGTCGTGGA-3′, reverse: 5′-CGGAGATGATGACCCTTTTG-3′).

Error bars represent the SD from the mean. Data were analyzed for statistical significance by the Student t test.

We first sought to determine whether differences exist in the metabolism of distinct intracellular l-arginine sources in mycobacteria-infected MΦs. Thioglycollate-elicited peritoneal-derived MΦs (PDMs) were used to address this question because of their increased basal arginase activity compared with other MΦ sources and the subsequent implications that this activity may have on antimycobacterial NO production. PDMs were cultured in media containing equimolar amounts of l-arginine, l-citrulline, or both amino acids, followed by M. bovis BCG infection. Analysis of extra- and intracellular amino acid contents revealed that MΦs cultured in l-citrulline did not produce as much l-ornithine as did MΦs cultured in l-arginine (Fig. 1A, 1B). This effect was most pronounced when infected MΦs were also stimulated with IFN-γ. Similar differences in l-ornithine synthesis were observed in PDMs and BMDMs costimulated with IFN-γ and the TLR agonist LPS (Supplemental Fig. 1A, 1B). l-citrulline was reported to inhibit arginase activity (23–25), yet we did not observe this effect because MΦs cultured in l-arginine alone, or in combination with l-citrulline, produced similar amounts of l-ornithine in all stimulation conditions (Supplemental Fig. 1A, 1B). When amino acids were reduced to 0.2 mM (within physiological ranges), l-arginine–cultured MΦs converted nearly all of the l-arginine to l-ornithine or l-citrulline following infection (Fig. 1C, 1D). As with cells cultured in 1 mM amino acids, l-citrulline resulted in less l-ornithine synthesis than did l-arginine, suggesting that imported l-arginine is the preferred substrate for l-ornithine synthesis compared with l-arginine synthesized intracellularly from l-citrulline. This was further confirmed when infected MΦs were cultured in normal (¹²C₅) l-arginine and heavy-isotope (¹³C₅) l-citrulline to detect the source of l-ornithine (Fig. 2A). More than 90% of all l-ornithine detected within or released from M. bovis BCG–infected MΦs with IFN-γ stimulation contained the ¹²C label, whether cultured in 1 mM amino acids (Fig. 2B, 2C) or 0.2 mM amino acids (Fig. 2D, 2E).

It was possible that the decrease in l-ornithine from synthesized l-arginine was entirely due to insufficient metabolism of l-citrulline to l-arginine, so we next analyzed NO production. We detected less NO from infected MΦs cultured in l-citrulline (Fig. 3A, 3B); however, the deficiency in total l-ornithine production was even larger (Fig. 3C, 3D). When comparing the NO/l-ornithine production ratio, we found that MΦs cultured in l-citrulline were more likely to produce NO versus l-ornithine compared with those cultured in l-arginine (in the 1 mM and 0.2 mM conditions) (Fig. 3E, 3F, Supplemental Fig. 1C–F). Taken together, our data indicate a preferential use of l-arginine synthesized from l-citrulline by iNOS that may not be impacted by arginase-mediated inhibition of NO and NO-mediated antimycobacterial MΦ functions.

Because MΦs cultured in l-citrulline displayed differential NO and l-ornithine production compared with those cultured in l-arginine, we next evaluated whether metabolism of l-citrulline to l-arginine was necessary to elicit these effects. The sole known pathway of l-arginine synthesis from l-citrulline is mediated by the sequential action of argininosuccinate synthase 1 (Ass1) and Asl. Ass1 converts l-citrulline to l-argininosuccinate, and Asl mediates l-arginine synthesis from l-argininosuccinate. To determine whether this metabolism was necessary for l-citrulline–mediated NO production and antimycobacterial MΦ function, mice were generated to facilitate conditional deletion of Asl in hematopoietic cells by crossing Asl^flox/flox mice (26) with Tie2-cre mice (Asl^flox/flox;Tie2-cre) (Fig. 4A). Compared with control mice, PDMs from Asl^flox/flox;Tie2-cre mice displayed markedly diminished Asl mRNA and protein (Fig. 4B, 4C). PDMs were harvested from Asl^flox/flox;Tie2-cre and control mice to characterize the effect of Asl deletion on antimycobacterial functions. MΦs from both groups displayed similar NO production and reduction in live M. bovis BCG CFU when cultured in l-arginine media (Fig. 4D, 4E). NO production and antimycobacterial function of control PDMs cultured in l-citrulline mirrored those of MΦs cultured in l-arginine. However, Asl^flox/flox;Tie2-cre PDMs cultured in l-citrulline did not produce NO or decrease M. bovis BCG burden, similar to our previous report analyzing MΦs lacking functional Ass1 (22). These data confirm that conversion of l-citrulline to l-arginine is required for l-citrulline–mediated antimycobacterial MΦ function.

Thus far, our data have shown a significantly larger NO/l-ornithine ratio from MΦs that use synthesized l-arginine from l-citrulline compared with those using imported l-arginine. This suggests that MΦ polarization prior to infection, altering the balance of iNOS versus arginase activity, may dictate the antimycobacterial potency of l-citrulline. Further, in IL-4 + IL-10–polarized MΦs, the use of l-citrulline as a source of l-arginine may bypass arginase-mediated inhibition of iNOS function and elicit better antimycobacterial immunity. This method of polarization is likely relevant in vivo, because M. tuberculosis–infected patients with elevated type 2 cytokines have increased prevalence of TB (27–29). To test this, PDMs were stimulated with IL-4 + IL-10 for 24 h prior to infection to further induce Arg1 activity. PDMs were washed following prestimulation to remove residual cytokines. Upon infection, PDMs were stimulated with IFN-γ, with or without the arginase inhibitor BEC, in culture containing 1 mM l-arginine, l-citrulline, or both amino acids (Fig. 5A). We found that pretreatment of PDMs with IL-4 + IL-10 did not have a dramatic effect on NO production or recovered M. bovis BCG. Furthermore, PDMs treated with BEC did not drastically increase NO production or result in any significant change in M. bovis BCG CFU (Fig. 5B, 5C). We expected that this phenomenon might be due to supraphysiological concentrations of l-arginine and l-citrulline, so we modified subsequent experiments to contain 0.2 mM l-arginine and l-citrulline. In contrast to PDMs cultured in supraphysiological l-arginine and l-citrulline, those cultured at physiological concentrations produced more NO and exhibited greater mycobacterial killing when cultured with l-arginine compared to those cultured solely in l-citrulline (Fig. 6). Notably, this effect was reversed when MΦs were prestimulated with IL-4 and IL-10. In prestimulated PDMs, l-citrulline provided better NO production and mycobacterial control than did l-arginine alone, and the combination of the two provided the best antimycobacterial effects (Fig. 6B, 6C). Interestingly, this difference was reversed in MΦs treated with BEC. Together, these data suggest that MΦ polarization prior to infection dictates the requirement for l-citrulline–mediated antimycobacterial function, but only when these amino acids approach physiological concentrations.

As an additional method to alter iNOS and arginase-mediated metabolism, we modified the timing of IFN-γ stimulation, resulting in a delay in iNOS expression and, thereby, NO production. We expect that a scenario of delayed IFN-γ stimulation may occur early during primary infection, prior to amplified IFN-γ responses delivered by effector T cells. We infected PDMs cultured in l-arginine, l-citrulline, or both for 24 h prior to IFN-γ addition and determined antimycobacterial activity after an additional 72 h (Fig. 7A). Similar to the data in Fig. 5, the combination of amino acids at supraphysiological concentrations did not enhance antimycobacterial MΦ activity compared with MΦs cultured in l-arginine alone (Fig. 7B, 7C). Even when amino acid concentrations were decreased, the delay in IFN-γ stimulation did not result in statistical differences in NO or the amount of recovered M. bovis BCG from MΦs cultured in l-arginine compared with l-citrulline. However, the combination of l-arginine and l-citrulline provided MΦs the ability to increase NO production and control M. bovis BCG better than MΦs cultured in either amino acid alone (Figs. 7D, 7E, 8).

These data reveal that l-citrulline and l-arginine are essential for antimycobacterial defense. Depending on the activation state of MΦs, one or the other of these amino acids may be the dominant contributor to mycobactericidal effectors brought about by differential metabolism of synthesized versus imported l-arginine (Fig. 8). Importantly, the enhancement of antimycobacterial activity in MΦs cultured in l-citrulline or the combination of l-citrulline and l-arginine compared with l-arginine alone only came to light under physiological culture conditions; it was not observed when using supraphysiological concentrations of l-arginine and l-citrulline. This observation should be stressed for designing cell culture experiments, considering that common cell culture media, DMEM and RPMI 1640, contain 2 to 10 times the concentration of l-arginine found in plasma. Furthermore, these media do not contain l-citrulline, concealing the impact that this amino acid may have on cellular biology. Current investigations are aimed at determining how differential usage of l-arginine occurs depending on its source. One mechanism of interest that was described in other models (26, 30) is the possibility of a protein complex containing Asl, Ass1, and iNOS that aids in compartmentalizing synthesized l-arginine so that it is not available for arginase-mediated metabolism.

One might expect the metabolic consequences for l-arginine and l-citrulline that we observed in MΦs to extend to NO synthases in other tissues. Two additional NO synthase isoforms exist in mammals, neuronal NO synthase (nNOS; encoded by Nos1) and endothelial NO synthase (encoded by Nos3), which may benefit from l-citrulline metabolism. Arginase activity regulates NO output in endothelial cells during hypertension and in aged rodents (31–33). Furthermore, it was suggested that endothelial NO synthase relies on l-arginine synthesis from l-citrulline to mediate optimal NO production (34, 35). Arg1 is expressed in the mouse brain and upregulated under inflammatory conditions in the CNS where nNOS-derived NO serves as a neurotransmitter (36–38). Although evidence of nNOS and Arg1 in neuronal cells is limited, Arg1 is capable of inhibiting NO production in Arg1/nNOS-cotransfected cells (39), suggesting that l-citrulline metabolism could rescue nNOS function if such an environment exists in vivo. Moreover, astrocytes use both nNOS and iNOS for NO production, upregulate Ass1 upon stimulation, and produce NO in l-citrulline–only conditions (40–43), supporting the notion that l-citrulline metabolism provides metabolic advantages to NO synthase isoforms in multiple cell types.

How do our results relate to amino acid metabolism and NO production in human MΦs? As mentioned in the 1Introduction, in vitro human monocyte-derived MΦs produce very little NO. Yet, evidence that human tissue MΦs make iNOS and NO, both in vivo and ex vivo, suggests that in vitro conditions are missing an essential factor(s) that would enable NO production (3). With the recent work by Mattilla et al. (17) demonstrating iNOS and arginase protein expression within human TB granuloma MΦs, we expect that competition for l-arginine, which we attempted to model in this work, applies to in vivo pathologies. The broad implications of these findings include the possibility of enhancing immune function with amino acids in vivo. l-arginine supplementation, aimed at enhancing NO-dependent immunity against TB, was attempted in M. tuberculosis–infected patients in two independent studies. Schön et al. (44) treated TB patients with oral l-arginine and observed subtle benefits, including increased sputum conversion and reduced cough and chest pain in HIV⁻ patients. Yet a more recent study found no clinical benefit in supplementing l-arginine to M. tuberculosis–infected patients (45). The metabolic consequences of oral l-arginine supplementation may be responsible for these results. Following ingestion, the bulk of l-arginine is metabolized within intestinal enterocytes or imported into liver hepatocytes, reducing the concentration available for immune utilization (46). In contrast, l-citrulline bypasses enterocyte metabolism and liver import (46, 47). Taken together with our present data that show that l-citrulline–derived l-arginine bypasses arginase-mediated suppression of mycobactericidal MΦ activity, future studies investigating the benefits of supplementing l-citrulline on mycobacterial host defenses are warranted.

We thank Drs. S. S. Way and D. Haslam, as well as members of their laboratories, for discussions during the preparation of this manuscript.

The authors have no financial conflicts of interest.