Resolution of Cutaneous Leishmaniasis and Persistence of Leishmania major in the Absence of Arginase 1

Arginase (Arg) is a manganese-dependent enzyme that exists in a cytosolic (Arg1) and a mitochondrial (Arg2) isoform and converts the amino acid l-arginine into l-ornithine and urea (1). Arg1 has long been known for its constitutive expression and central role within the urea cycle of hepatocytes, in which it is an essential enzyme as illustrated by the neonatal hyperammonemia and lethality of mice lacking Arg1 (2). Within the immune system, Arg1 has been extensively studied in rodent macrophages, dendritic cells, and myeloid suppressor cells, in which it is readily inducible by a variety of stimuli (3). These include 1) cytokines that cause alternative (M2) macrophage activation in a STAT6-dependent manner (e.g., IL-4, IL-13) (4–7); 2) cytokines or growth factors that deactivate macrophages (e.g., IL-10, TGF-β) (8) or confer a repair phenotype (e.g., FGF-2, IGF-1) (9); 3) microbial products (e.g., LPS) (10, 11) or pathogens (e.g., Mycobacterium tuberculosis), which drive the production of Arg1-inducing cytokines (e.g., IL-6, IL-10, G-CSF) (12, 13), or 4) micromilieu factors (e.g., hypoxia) (14–16).

The (hyper-)expression of Arg1 by myeloid cells has been linked to a number of functional changes and consequences in innate and adaptive immunity (reviewed in Refs. 17–20). First, consumption of l-arginine by Arg1 impaired the enzymatic activity or even the protein expression of inducible or type 2 NO synthase (NOS2) (5, 6), which generates l-citrulline and NO from l-arginine and molecular oxygen. As a result, Arg1 antagonized the NOS2/NO-dependent antimicrobial effector functions of macrophages (12, 21, 22) but also helped to limit inflammatory effects of NOS2/NO (23, 24). Second, depletion of extracellular l-arginine by myeloid Arg1 inhibited the proliferation of mouse and human T lymphocytes (reviewed in Refs. 25–27). This effect can contribute to the immune escape of microbial pathogens (28, 29) or tumor cells (30, 31); however, prevention of Th1 or Th2 cell–mediated immunopathology by Arg1 has also been observed (32, 33). Reported molecular mechanisms underlying the Arg1-mediated T cell suppression are the downregulation of CD3ζ (34) and the induction of cell cycle arrest that was characterized by a failure to upregulate cyclin D2 and D3 (35, 36), a decline of mTORC1 activity, and a repression of lipid biosynthesis–linked genes (36). Third, Arg1 was also found to deprive endothelial or type 3 NO synthase (NOS3) of l-arginine, which resulted in enhanced immune cell adhesion to vascular endothelium due to the lack of antiadhesive endothelial NO (37, 38). In addition to these immunoregulatory effects, Arg1 expressed by myeloid and endothelial cells has been implicated in promoting wound healing and facilitating tissue repair (33, 39) because its product ornithine is a precursor of polyamines and proline, which are required for cell proliferation and collagen synthesis, respectively (1). However, other publications questioned a principal role of alternatively activated or M2 macrophages and of Arg1 in wound healing and repair of tissue injuries (40–42).

Leishmania major is a flagellated protozoan parasite. Its extracellular promastigote form is transmitted by the bites of sand flies and, in mammalian hosts, transforms into intracellular, amastigote Leishmania. In humans, L. major causes local plaque-like nodular or ulcerative cutaneous lesions that usually heal spontaneously (43). The human course of infection is mimicked by L. major–infected C57BL/6NCrl (C57BL/6) mice, which develop nonulcerative and self-healing skin swellings. Cure of acute cutaneous leishmaniasis in these mice is strictly dependent on an intact IL-12/Th1/IFN-γ/TNF axis and the expression and activity of NOS2 (44–47); whether resolution of the skin lesions also requires Arg1 has never been analyzed.

Clinically healed C57BL/6 mice will still harbor small and stable numbers of parasites at the former skin site of infection as well as in the draining lymph node (dLN) (48, 49). These parasites remain under permanent control of NOS2-derived NO, as demonstrated by rapid recurrence of cutaneous disease and a strong increase of parasite numbers at the skin site and in dLNs after pharmacological inhibition of NOS2 activity (49). Recent elegant in vivo labeling data provided further evidence that persistent L. major parasites are not dormant but are actively replicating within macrophages and dendritic cells (50), which supports the concept of a dynamic process of killing and evasion during latent cutaneous leishmaniasis (49, 51, 52).

Lifelong persistence of L. major in C57BL/6 mice has been linked to 1) its escape into NOS2⁻ ER-TR7⁺ host cells or extracellular matrix, which in two studies using different L. major strains harbored 43% (FEBNI strain) (51) or 10% (LV39c5 strain) (50) of all parasites within chronically infected dLNs; 2) its ability to at least partially survive inside NOS2⁺ macrophages and dendritic cells, which accommodated between 36 (49) and 60% (50) of all persisting parasites; and 3) to the activity of CD4⁺ regulatory T cells and their production of IL-10, which inhibits effector Th1 cells and macrophages (53–55). As IL-10 is also a known (co)inducer of Arg1 (4, 11) and Arg1 is a competitor of NOS2 (1, 19), we previously proposed that expression of Arg1 by (infected) host cells during the latent phase of infection might impede the enzymatic activity of NOS2 and thereby promote lifelong persistence of L. major despite the presence of NOS2 protein (52). In contrast, Mandell and Beverley (50) concluded that Arg1 is unlikely to play a role in L. major persistence, as only a relatively small proportion of parasites were located in or close to Arg1-expressing host cells.

In this manuscript we tested the hypothesis that in L. major–infected C57BL/6 mice Arg1 is required for coordinated tissue repair and resolution of inflammation as well as for long-term parasite survival using a transgenic mouse strain lacking Arg1 in all hematopoietic and endothelial cells. We found that both processes remained unaffected in C57BL/6 mice lacking the expression of Arg1 in hematopoietic and endothelial cells.

Six- to eight-week-old female mice of the inbred strain C57BL/6 were purchased from Charles River (Sulzfeld, Germany). Conditional Arg1-deficient mice (12) on C57BL/6 background were obtained by independent backcrossing of Tie2cre-deleter mice and floxed Arg1 mice with C57BL/6 mice for 12 generations and intercrossing thereafter (22). For infection experiments, female mice at 8–14 wk of age and the respective age-matched wild-type (WT) or littermate controls were used and kept under specific pathogen-free conditions. Animal care and experiments were conducted in accordance with German regulations after local governmental approval (Ansbach and Würzburg, Germany).

The origin and propagation of L. major parasites (strain MHOM/IL/81/FEBNI) were previously published (49). The swelling of skin lesions was measured with a metric caliper (Kroeplin, Schlüchtern, Germany) and related to the footpad thickness before infection (bilateral infection) or to the thickness of the uninfected foot (unilateral infection) as percentage increase. Tissue parasite burden was determined by limiting dilution analysis with serial 2- or 3-fold dilutions of the organ or cell suspensions in modified Schneider Drosophila medium (56) and 12 replicates per dilution step (49). Statistical significance was assumed when 95% confidence intervals did not overlap.

Bone marrow–derived macrophages (BMM) were generated as described (57, 58) and used at day 7–10 of culture. Fibroblast cell lines (CHF, NOBO-1) were isolated from the skin of naive mice (CHF) and from the draining (popliteal) lymph nodes (NOBO-1) of mice that had healed a cutaneous infection with L. major (day >200 postinfection) (51). The lymphatic endothelial cell line mlEND1 [derived from mesenteric lymph nodes of C57BL/6 mice (59)] was kindly provided by Prof. R. Hallmann (University of Münster, Münster, Germany).

Cells were cultured in 6-well (3 × 10⁶ cells per well, 1.5 ml) or 24-well (1 × 10⁶ cells per well, 0.5 ml) tissue culture plates (Nunc) at 37°C and 5% CO₂/95% humidified air under normoxic (20% O₂) or hypoxic conditions (0.5% O₂, Invivo₂ 300/ICO₂N₂IC workstation; Baker Ruskinn, Bridgend, U.K.) for 24 h using RPMI 1640 (Life Technologies) supplemented as described (58) plus 5% FCS (Sigma-Aldrich) and stimulated with cytokines (recombinant murine IL-4, IL-10, IL-13, IFN-γ, TNF, or TGF-β1; all from R&D Systems, Wiesbaden-Nordenstadt, Germany) and/or LPS (Escherichia coli O111:B4; Sigma-Aldrich) as detailed in the figure legends.

At the concentrations used in the cell culture experiments, the LPS content of all reagents as well as of the supplemented media was <10 pg/ml as determined with a colorimetric Limulus amebocyte lysate assay (Whittaker M.A. Bioproducts, Walkersville, MD) (60).

The concentration of nitrite (NO₂⁻; μM) in culture supernatants was determined by the Griess assay as described before (60).

Total RNA was extracted from homogenized tissue or cell culture and reverse transcribed as described previously (22). The following gene-specific assays (TaqMan Gene Expression Assays-on-Demand; Thermo Fisher Scientific) were used for quantitative real-time PCR: Arg1 (Mm00475988_m1), Arg2 (Mm00477592_m1), Hprt-1 (Mm00446968_m1), Ifnγ (Mm00801778_m1), Il4 (Mm00445259_m1), Il10 (Mm00439616_m1), Il13 (Mm00434204_m1), Nos2 (Mm00440485_m1), and Tgfβ (Mm00441724_m1). The mRNA levels were calculated using the following formula:

with C_T denoting the cycle threshold and f = 10⁴ as an arbitrary factor. In some experiments, relative expression was calibrated to controls, as indicated in the figure legends.

Western blot analysis was carried out as described before (22). Briefly, the lysates of cells or organs were separated by 10% SDS-PAGE (loading of 40–80 μg of protein per lane) and transferred to a PVDF membrane (0.45 μm, Millipore Immobilon-P) using the tank blot technique (1 h, constant current of 1 A). After blocking of nonspecific binding sites, the blots were analyzed by ECL-based chemiluminescence (ECL Plus Pierce Western Blotting Substrate) using different primary Abs [goat anti-Arg1, V20, catalog no. sc-18354; rabbit anti-Arg2, H64, catalog no. sc-20151; and rabbit anti-Hsp90, H114, catalog no. sc-7947, all from Santa Cruz Biotechnology; rabbit anti–β-actin, catalog no. A2066 (Sigma-Aldrich); rabbit anti-Grb2, catalog no. 610112 (BD Biosciences); rabbit anti-mouse NOS2, lot 3055E (60)] and the respective HRP-conjugated secondary Abs (HRP-conjugated donkey anti-goat IgG, catalog no. 705-035-147 [Jackson ImmunoResearch]; HRP-conjugated goat anti-rabbit IgG, catalog no. 111-035-144 [Jackson ImmunoResearch]). If necessary, the Western blots were stripped (Restore Western Blot Stripping Buffer; Thermo Scientific) and reprobed. Protein expression was recorded with an Intas ChemiLux Imager. Images were processed only with respect to whole-blot brightness and contrast using Adobe Photoshop CS5 (Adobe Systems, San Jose, CA).

Confocal laser scanning fluorescence microscopy (CLSFM) analyses were performed as described before (22). Briefly, 5 μm of frozen tissue sections were fixed with acetone (−20°C), blocked with PBS/5% (v/v) donkey serum plus 0.01% (w/v) saponin, and stained for Arg1, NOS2, L. major, and cell types for 1 h at room temperature or overnight at 4°C using specific primary Abs [goat anti-Arg1, V20, catalog no. sc-18354 (Santa Cruz Biotechnology); rabbit anti-mouse NOS2, lot 3055E (60); high titer human anti–L. major patient antiserum (61); rat anti–ER-TR7 (BMA Biomedicals, catalog no. T-2109); biotin-conjugated rat anti-CD11b (eBioscience, M1/70/Mac-1, catalog no. 13-0112); rat anti-CD31 (eBioscience, clone 390, catalog no. 14-0311-81); rat anti-MECA32, a gift of Prof. R. Hallmann (62)], fluorochrome- or biotin-conjugated secondary Abs (Alexa Fluor 488–conjugated donkey anti-goat IgG, catalog no. 705-546-147; Alexa Fluor 647–conjugated donkey anti-human IgG, catalog no. 709-606-149; Rhodamine Red-X–conjugated donkey anti-rabbit IgG, catalog no. 711-296-152; Rhodamine Red-X–conjugated donkey anti-rat IgG, catalog no. 711-293-153; biotin-conjugated donkey anti-rabbit IgG, catalog no. 711-066-152 [all from Jackson ImmunoResearch]) and fluorochrome-conjugated streptavidin (Cy3, catalog no. 40-5215, and PerCP/Cy5.5, catalog no. 45-4317; both from BioLegend). Staining of the different Ags (Arg1, NOS2, L. major, cell type) was carried out sequentially (i.e., incubation with the primary Ab against the first Ag was directly followed by the respective secondary reagent, then equivalent staining for the other Ags). Nuclei were stained with DAPI (final concentration: 10 ng/ml), and the samples were covered with Vectashield (Vector Laboratories) and dried overnight. For staining of two biotinylated Abs, the streptavidin/biotin blocking kit (Vector Laboratories) was used to block free streptavidin/biotin-binding sites between the two stainings.

In situ detection of NO was done as described (22). Briefly, unfixed frozen tissue sections were stained with 10 μM 1,2-diaminoanthraquinone (DAQ; Sigma-Aldrich) in PBS with 10% DMSO at 37°C for 45 min, followed by extensive washing steps (63, 64).

For confocal microscopy, a ZEISS LSM 700 was used, and data were recorded using 40×/1.3 immersion oil objective and ZEN software 2012 (ZEISS, Göttingen, Germany). Using a 405-nm laser for excitation, fluorescence of DAPI was detected with a short pass filter (>490 nm). Classical fluorochromes (Alexa Fluor 488) that were excited with a 488-nm laser line were detected with a band-pass (490–555 nm) filter. In combination with a long-pass (>640 nm) filter, other fluorochromes (PerCP/Cy5.5) became detectable because of their far red-shifted emission. Rhodamine Red-X Abs and DAQ were excited with 555 nm and detected with a band-pass (490–635 nm) filter. Alexa Fluor 647–labeled Abs were excited with 639 nm and detected with a long-pass (>640 nm) filter. To cover a representative area of the tissue, 320 μm² was recorded in an optimal resolution of 1024 × 1024 pixels, and, for detailed analysis, single Leishmania parasites were optically zoomed in separately. For experimental setup, the Fluorescence Spectra Analyzer from BioLegend was used (http://www.biolegend.com/spectraanalyzer).

Results are displayed as mean ± SEM and were statistically analyzed using GraphPad Prism v.4, v.6, or v.7, as detailed in the figure legends.

We and others previously observed that 1) NOS2 expressed by macrophages and dendritic cells was critical for the control of L. major (49, 65–68), 2) fibroblasts served as host cells for L. major during latent cutaneous leishmaniasis (51), and 3) NOS3 restricted the inflammatory immune response to L. major during acute cutaneous leishmaniasis (69). Considering that each of these processes can be affected by Arg1, we first compared the expression of Arg1 in C57BL/6 macrophages, fibroblasts, and lymphoid endothelial cells in vitro.

In BMM, IL-4, IL-4 plus IL-10, IL-13, LPS, or hypoxia (0.5% O₂) all strongly induced Arg1 mRNA (Fig. 1) and protein (Fig. 2), confirming earlier data (4–7, 10, 11, 14–16), whereas L. major promastigotes and TGF-β1 failed to do so (Figs. 1, 2). Arg2 mRNA was only weakly upregulated by these stimuli, except for hypoxia (Fig. 1); however, Arg2 protein expression remained unaltered and was even undetectable under hypoxia (Fig. 2). In accordance with previous studies (19, 22, 70), IFN-γ or IFN-γ/TNF activated macrophages for expression of NOS2 but not of Arg1 or Arg2 (Figs. 1, 2).

In skin-derived fibroblasts (CHF) or reticular fibroblasts of the lymph node (NOBO-1) both Arg1 and Arg2 mRNA were undetectable under the stimulation conditions tested (Fig. 1). In contrast, NOS2 protein was readily induced by combined stimulation with IFN-γ plus TNF (Fig. 2). C57BL/6 lymphoid endothelial cells (mlEND-1) showed a constitutive expression of Nos3 mRNA and NOS3 protein, which is characteristic for endothelial cells. Arg1 and NOS2 were upregulated by IL-4/IL-10 and IFN-γ/TNF or IFN-γ/LPS, respectively, but not in response to L. major promastigotes (Fig. 3A, 3B and data not shown). Arg2 mRNA was expressed at 10-fold lower levels than Arg1 mRNA, so that Arg2 protein was hardly detectable (Fig. 3 and data not shown).

These data illustrate that all three cell types were capable of synthesizing NOS2, whereas expression of Arg1 was restricted to macrophages and endothelial cells.

Next, we investigated the regulation of Arg1 in vivo. During acute cutaneous leishmaniasis, we found that L. major–infected C57BL/6 mice transiently expressed Il4 and Il13 mRNA and also upregulated Il10 and Tgfβ1 mRNA (Fig. 4A), extending recently published data (22, 71). The appearance of these known Arg1 inducers was much more prominent in the skin lesions than in the dLNs and paralleled by the temporary induction of Arg1 mRNA, which was remarkably stronger than that of Arg2 (Fig. 4B). In line with the mRNA results, Arg1 and Arg2 proteins were readily detected at the cutaneous site of infection, whereas their expression was much less pronounced (Arg1) or even missing (Arg2) in the dLNs (Fig. 5A, upper panels). These Western blot results were confirmed by microscopy. CLSFM revealed that Arg1⁺ cells (white) as well as Arg1/NOS2 double-positive cells (pink) were routinely present in the acutely infected skin lesions (Fig. 5Ba). In contrast, Arg1 was only weakly expressed and Arg1⁺ cells were rare in the dLNs, whereas NOS2⁺ cells were readily found in both tissues (Fig. 5Ba, b).

As soon as the acute cutaneous lesions started to resolve, the levels of Arg1 mRNA continuously declined in the skin and dLNs. During the latent phase of L. major infection [day >100 (49, 51)], Arg1 protein became undetectable in Western blots of whole lysates of skin and dLN samples, whereas low levels of Arg2 protein were still found at the skin site but not in the dLN of chronically infected mice (Fig. 5A, upper panels). Contrary to our hypothesis, Arg1 was also no longer visible on a single cell level in tissue sections that were analyzed by CLSFM (Fig. 5Bc, d). In contrast, NOS2 mRNA and protein remained persistently expressed in long-term infected dLNs (Figs. 4B, 5A, 5Bc, d), as seen before (49, 51).

From these data, we conclude that Arg1 expression is characteristic for the acute but not for the latent phase of L. major infection.

In a previous analysis of L. major–infected C57BL/6 mice, we found that the cells expressing Arg1 in acute skin lesions had the phenotype of inflammatory macrophages/dendritic cells (CD11b⁺CD11c⁺MHCII⁺Ly6C⁺CD64⁺F4/80⁺CCR2⁺CD206⁺PDL2⁺CD207⁻CCR7⁻). Flow cytometry did not reveal Arg1⁺ cells within the CD45⁻ population (22). However, in an independent project on L. major–infected BALB/c mice, we noticed low levels of Arg1 mRNA in blood endothelial cells (CD45⁻ CD31⁺ podoplanin⁻) and lymphatic endothelial cells (CD45⁻ CD31⁺ podoplanin⁺) purified from infected dLNs, which were not seen in naive dLNs (A. Debus, U. Schleicher, and C. Bogdan, unpublished observations). This observation raised the possibility that endothelial cells might still contribute to the Arg1 signal in infected dLNs. We therefore decided to use Tie2Cre^+/−Arg1^fl/fl C57BL/6 mice, which lack Arg1 expression in all hematopoietic and endothelial cells (12, 22, 72), and the respective WT littermates (Tie2Cre^−/−Arg1^fl/fl C57BL/6 mice) for our functional analyses (see below). Importantly, when we compared the two mouse strains for expression of Arg1, the Tie2Cre-mediated gene deletion no longer allowed for the detection of Arg1 protein in the skin lesions and dLNs throughout the course of L. major infection (Fig. 5A, lower panels). There was no compensatory upregulation of Arg2; in fact, at the skin site, L. major–infected Tie2Cre^+/−Arg1^fl/fl C57BL/6 mice showed even a reduced expression of Arg2 protein compared with WT controls, and in the dLNs Arg2 protein was undetectable (Fig. 5A).

From these results, we conclude that host cells other than hematopoietic and endothelial cells are not a relevant source of Arg1 during L. major infection of C57BL/6 mice and that Tie2Cre-mediated deletion of Arg1 does not cause upregulation of Arg2.

As Arg1 was clearly expressed at least during the acute phase of L. major infection of C57BL/6, we next tested whether Arg1 competes with NOS2 for its substrate and thereby impedes the generation of NO. To this end, we comparatively evaluated dLN sections from L. major–infected Tie2Cre^−/−Arg1^fl/fl (WT) and Tie2Cre^+/−Arg1^fl/fl (Arg1-deficient) mice for the in situ production of NO using the NO-reactive compound DAQ (Ref. 22 and references therein). During both the acute and the latent phase of infection, the number and mean fluorescence intensity (MFI) of DAQ⁺ cells, which in serial sections largely colocalized with NOS2⁺ myeloid cells (CD11b⁺) (Fig. 6Aa, b, e, and f versus c, d, g, and h), remained unaltered in the absence of Arg1 (Fig. 6B).

Arg1⁺ host cells can support the survival of Leishmania also in an NOS2-independent manner by providing ornithine and subsequently polyamines to the parasites (73), which led us to consider these cells as potential niches for persisting L. major. We therefore envisaged the possibility that the deletion of Arg1 might cause a shift in the host cell tropism of L. major, which we assessed by detailed CLSFM analyses of dLN sections. During the acute phase of infection, the majority of L. major parasites in the dLNs were associated with CD11b⁺ myeloid cells (53.9%) (Figs. 7, 8A), which were mostly positive for NOS2 (90.4%) (Fig. 8B). Approximately two fifths (42.1%) of all parasites colocalized with ER-TR7⁺ reticular fibroblasts or extracellular matrix, with less than one half (46.6%) being positive for NOS2, whereas only very few parasites (4%) were attached to cells carrying the endothelial marker MECA32 or CD31 (Figs. 7, 8). Overall, 53.7, 37.1, and 9.2% of the L. major parasites were associated with NOS2⁺ cells, NOS2⁻ cells, or necrotic areas of tissues, respectively. Most importantly, there was no difference in the host cell distribution of L. major between Tie2Cre^−/−Arg1^fl/fl (WT) and Tie2Cre^+/−Arg1^fl/fl (Arg1-deficient) C57BL/6 mice, based on the evaluation of 2741 or 2657 parasites, respectively (Fig. 8A, 8B). When we analyzed the latent phase of infection, 54.8, 42.1, and 3.0% of the persisting parasites colocalized with ER-TR7⁺, CD11b⁺, or MECA32⁺/CD31⁺ host cells, of which 9.5, 23.7, or 0% were positive for NOS2, respectively. Overall, only 15.2% of the persisting parasites were located in NOS2⁺ cells. Again, we did not observe differences in the host cell associations of L. major parasites persisting in WT mice (2452 parasites assessed) or Arg1-deficient mice (2635 parasites assessed).

Together, these data demonstrate that in L. major–infected C57BL/6 mice, Arg1 expression had no impact on the generation of NO or on the host cell tropism of the parasite.

Finally, we tested whether a Tie2Cre-mediated deletion of Arg1 had any consequences for the course of infection or for the long-term persistence of L. major in clinically healed mice. As depicted in Fig. 9, development and clinical healing of the nonulcerative skin lesions (Fig. 9A), as well as the parasite burden at the primary site of infection, in the dLNs, and in the spleen (Fig. 9B–D), remained unaffected by the deletion of Arg1 in hematopoietic and endothelial cells. Most notably, L. major persisted to the same degree in the presence and absence of Arg1 (Fig. 9B–D, day >100). These data unequivocally demonstrate that Arg1 is not required for clinical resolution of L. major skin lesions or for long-term parasite persistence.

In the current study, we analyzed the expression and function of Arg1 during the acute and chronic phase of a self-healing L. major infection in C57BL/6 mice. In cultures of macrophages and lymph node endothelial cells, Arg1 mRNA and protein were readily induced in response to cytokine stimulation, whereas cultured skin or lymph node fibroblasts remained negative for Arg1 under all conditions tested. In vivo, Arg1 mRNA and protein were prominently expressed in the skin lesions and weakly also in the dLNs of acutely infected mice but became undetectable during the chronic phase of infection. Despite the expression of Arg1 during the acute phase of L. major infection, deletion of Arg1 in hematopoietic and endothelial cells had no impact on the clinical development of the skin lesions, the tissue parasite burden, the in situ production of NO, or on the host cell tropism of L. major parasites. Most notably, long-term parasite survival remained unaffected by the absence of Arg1, excluding the possibility that Arg1 protein levels below the detection thresholds of Western blot and CFLSM analyses might play a role. Thus, Arg1 was dispensable both for the resolution of acute cutaneous leishmaniasis and for parasite persistence.

Previous reports on the expression and function of Arg1 in acute cutaneous leishmaniasis focused on the analysis of L. major–infected BALB/c mice. This mouse strain shows genetically determined impaired macrophage functions (Refs. 65 and 74 and references therein) and, unlike C57BL/6 mice, a persistent Th2 immune response with high levels of IL-4 and IL-13, which leads to progressive, ulcerative skin lesions and to visceral and fatal disease (reviewed in Ref. 45). This exceptional and unique course of L. major infection in BALB/c mice was associated with a hyperexpression of Arg1 in skin lesions (22, 73, 75) and dLNs (22), which was largely but not exclusively dependent on IL-4 and IL-13 (22). Treatment of L. major–infected BALB/c mice with the nonselective Arg1/Arg2 inhibitor N^ω-hydroxy-nor-l-arginine strongly ameliorated the skin lesions, which was attributed to a reduced availability of polyamines to L. major parasites in the absence of host cell Arg activity (73, 75). The use of hematopoietic and endothelial cell–specific, conditional Arg1-knockout mice (Tie2Cre^+/−Arg1^fl/fl) unequivocally proved that hyperexpression of Arg1 accounts for local disease progression, parasite visceralization, and lethality in L. major–infected BALB/c mice because, in the absence of Arg1, BALB/c mice showed an enhanced in situ production of NO, and the ulcerative, progressive skin lesions turned into nonulcerative, stable skin swellings, despite unaltered CD4⁺ (Th1, Th2, Th17, regulatory T) and CD8⁺ T cell cytokine expression and proliferation (22). The finding that in Arg1-deficient Tie2Cre^+/−Arg1^fl/fl BALB/c mice clinically stable (i.e., nonulcerative and nonprogressive) skin swellings persisted (22) raised the possibility that a certain amount of Arg1 is required for complete tissue repair and ultimate resolution of inflammation. However, our current observation that Tie2Cre-mediated deletion of Arg1 in otherwise self-curing C57BL/6 mice did not impair clinical healing of the nonulcerative skin lesions clearly suggests that systemic sources of ornithine and polyamines fully compensate for the absence of Arg1 at the local site of infection. The fact that in C57BL/6 mice Tie2Cre-mediated Arg1-deficiency also did not further accelerate the spontaneous resolution of the skin lesions is likely due to the transient (Figs. 3, 4) and much lower expression of Arg1 in C57BL/6 as compared with BALB/c mice (22, 73, 75).

In mice infected with M. tuberculosis, which, similarly to L. major, resides in macrophages and is controlled by NOS2 (23, 76), Arg1 was found to be critical for preventing exuberant tissue pathology. This effect became apparent when, in NOS2^−/− mice (used to imitate the defective NO production in hypoxic/anoxic mycobacterial granulomas), Arg1 was additionally deleted in hematopoietic and endothelial cells (32). Although L. major skin lesions are also characterized by a reduced oxygen content (partial pressure O₂ ∼ 2.8%), which certainly limits the production of NO (16), the residual activity of NOS2 and NOS3 in L. major–infected tissues appears to be still sufficient to restrict T cell proliferation and vascular adhesion and recruitment of immune cells (including Th1 cells) (18, 19, 69, 77–80). In the absence of Arg1, the amount of locally generated NO, which suppresses T cells and blocks immune cell adhesion, might be even more pronounced. These considerations offer an explanation for why Arg1-deficiency did not cause exacerbation of acute clinical disease in L. major–infected C57BL/6 mice.

Our finding that Arg1 was hardly detectable during the latent phase of L. major FEBNI infection is in line with a previous report using a different strain of L. major (LV39c5), in which only 3 or 12% of the persisting parasites in the clinically healed footpad resided in Arg1⁺ cells or in cells adjacent to Arg1⁺ cells, respectively (50). As our analysis of Tie2Cre^+/−Arg1^fl/fl C57BL/6 mice revealed that Arg1 is nonessential for long-term survival of L. major parasites in skin, dLN, and spleen, alternative mechanisms of parasite persistence need to be discussed. First, a possible role of host cell Arg2 has not yet been formally tested by us but appears unlikely. Arg2 was transiently detectable in chronically infected C57BL/6 mice but only in the skin and not in the dLN (i.e., the site where the vast majority of parasites persist). In addition, Tie2Cre^+/−Arg1^fl/fl (Arg1-knockout) mice showed a reduced expression of Arg2 in the skin compared with Tie2Cre^−/−Arg1^fl/fl (Arg1-competent) (Fig. 5A), but parasite persistence was fully preserved (Fig. 9B–D). Second, the Arg of L. major does not seem to be essential for parasite persistence. A previous analysis of WT and Arg-deficient L. major parasites of the LV39c5 strain revealed that the Arg-mutant caused less severe acute skin lesions in C57BL/6 mice but nevertheless persisted in the skin after clinical healing (day 120) to an extent comparable with WT L. major (81). However, at this stage we cannot formally exclude the possibility that Arg1, Arg2, and L. major Arg might substitute for each other. We will address this possibility by creating Tie2Cre^+/−Arg1^fl/fl×Arg2^−/− mice and an Arg-deficient mutant of the L. major FEBNI strain. This will allow us to definitively test to what extent locally produced ornithine (derived from myeloid/endothelial Arg1, Arg2, and parasite Arg) and polyamines versus systemically generated ornithine and polyamines contribute to parasite persistence. As a third possibility, it has been proposed that persisting L. major parasites become partially resistant or tolerant to NOS2-derived NO (50). In principle, it is possible that Leishmania parasites become resistant to NO. Leishmania promastigotes with naturally acquired reduced susceptibility to NO were repeatedly isolated from patients, and their presence was correlated with a more severe course of infection or reduced response to antimony treatment (82–84). Furthermore, continuous exposure of L. infantum amastigotes to NO donors led to the induction of NO resistance (85). However, it is important to emphasize that L. major amastigotes persisting in C57BL/6 mice are efficiently controlled by NOS2-derived NO, as demonstrated by the immediate recurrence of skin lesions and parasite proliferation following pharmacological inhibition of NOS2 (49, 86). If persistent L. major parasites became truly resistant to NO, spontaneous reactivation of disease should occur, which, however, is not the case.

We therefore favor, as previously proposed (49, 51, 52), a dynamic model of parasite killing and survival in latent leishmaniasis, in which NOS2⁺ cells (∼2/3 myeloid cells, ∼1/3 fibroblast reticular cells/extracellular matrix; this study) and NOS2⁻ cells (∼2/5 myeloid cells, ∼3/5 fibroblast reticular cells/extracellular matrix; this study) harbor ∼15 and 85% of all parasites, respectively.

NO-mediated destruction of L. major or inhibition of its metabolism and proliferation will not be restricted to NOS2⁺ infected host cells but will also occur in adjacent NOS2⁻ infected cells because of the diffusion capacity of NO, whereas parasites in more distant NOS2⁻ cells and necrotic areas presumably survive (51, 87). This process of immune evasion by L. major is dependent on local production of IL-10, which inhibits the generation of IFN-γ, NO, reactive oxygen species, and TNF (53–55), but independent of Arg1 (this study). It is also possible that NOS2-derived NO itself contributes to parasite persistence based on its ability to restrain cytokine and Th1 cell responses (18, 68, 77, 80). The negative feedback regulatory effect of NO helps to resolve inflammation and clinical disease at the expense of incomplete macrophage activation and parasite eradication and might even entail NO produced by reticular fibroblasts (88, 89), which otherwise was found to be insufficient to control L. major parasites (51).

Apart from Leishmania infections, the impact of Arg1 on tissue pathology, parasite control, and host survival has also been studied in other parasitic disease. In murine Toxoplasma gondii infections, LysMcre-mediated deletion of Arg1 in myeloid cells completely prevented wasting disease and lethality, which are characteristic for T. gondii-infected C57BL/6 WT mice, indicating that Arg1 impaired parasite control (12). In infections with the nematode Trichuris muris, mice lacking Arg1 in all hematopoietic and endothelial cells (Tie2Cre^+/−Arg1^fl/fl) showed unaltered intestinal inflammation, cytokine responses, and worm expulsion, demonstrating that in this helminth model Arg1 neither facilitated nor impeded cure (42).

In summary, to our knowledge, the current study demonstrates for the first time that clinical resolution of an infection caused by an intracellular pathogen occurs in the absence of local Arg1 expression. Unlike noninfectious skin injuries (39), the spontaneous healing of L. major–induced cutaneous lesions was clearly independent of Arg1. Contrary to our previous model and hypothesis, Arg1 in the skin and dLN was also dispensable for long-term persistence of L. major in self-healing mice. These results shed new light on the immunological requirements for tissue repair and maintaining latent infections.

We thank Till König, Eugenie Luft, and Katharina Pracht for participating in pilot experiments. We are grateful to Prof. Peter Murray (formerly of St. Jude’s Children’s Hospital, Memphis, TN, now at Max Planck Institute of Biochemistry, Munich, Germany) for supplying initial breeding pairs of Arg1-deficient mice, to Prof. Rupert Hallmann (University of Münster) for providing the mlEND1 cell line and initial batches of MECA32 Ab, to the operators of the Core Facility for Cell-Sorting and Immunomonitoring, to the members of the Optical Imaging Center Erlangen of the Friedrich-Alexander-Universität for advice on confocal image analysis, and to the personnel of the Franz-Penzoldt Animal Center Erlangen of the Friedrich-Alexander-Universität for animal care.

The authors have no financial conflicts of interest.