Volume 10, Issue 6 pp. 1221-1234
Free Access

Mechanisms and consequences of persistence of intracellular pathogens: leishmaniasis as an example

Christian Bogdan

Corresponding Author

Christian Bogdan

*E-mail [email protected]; Tel. (+49) 9131 852 2551; Fax (+49) 9131 852 2573.Search for more papers by this author
First published: 18 March 2008
Citations: 109

Summary

Lifelong persistence after clinical cure of the primary infection is a characteristic feature of many intracellular pathogens, including viruses, bacteria and protozoa. The underlying mechanisms are complex and range from the passive protection against toxic effector molecules of the host and the remodelling of intracellular compartments as safe niches to the active modulation of the immune response at multiple levels. Parasites of the genus Leishmania have been particular helpful in unravelling some of the basic processes and form therefore the centre of the discussion.

Introduction

From a pathomechanistic and immunological point of view infectious diseases are caused by either extracellular or intracellular pathogens. Typical extracellular pathogens (e.g. staphylococci, streptococci and pneumococci) are capable of replication outside of host cells. They are largely controlled by non-opsonic and opsonic phagocytosis (Ofek et al., 1995; Kadioglu and Andrew, 2004), i.e. by neutrophilic granulocytes and other phagocytes that especially in the presence of antibodies or complement factors efficiently endocytose and kill the bacteria by constitutively expressed or rapidly synthesized effector molecules such as antimicrobial peptides, proteases and reactive oxygen intermediates (ROI) generated by the NADPH oxidase (phox) (Bogdan, 2004). T helper lymphocytes support the function of neutrophils and macrophages by stimulating B cells for the production of antibodies against certain antigens such as capsular polysaccharides and by promoting the formation of abscesses (Wang et al., 2006). Intracellular pathogens, in contrast, usually live and replicate within endosomal compartments or the cytosol of diverse host cells such as neutrophils, macrophages, dendritic cells (DC), fibroblasts or epithelial cells. Examples not only include all viruses, but many different bacteria (e.g. Listeria monocytogenes, Mycobacterium tuberculosis, Brucella spp., Salmonella spp., Shigella spp., Coxiella burnetii, Anaplasma phagocytophilum, Ehrlichia chaffeensis) as well as certain protozoa (e.g. Leishmania spp., Toxoplasma gondii, Trypanosoma cruzi) and fungi (e.g. Histoplasma capsulatum). In the case of viruses the host defence depends on neutralizing antibodies, a cytolytic CD8+ T cell response, and/or the induction of antiviral mechanisms by type I interferons (IFN-α/β) (Price et al., 1999; Takeuchi and Akira, 2007). The control of intracellular bacteria and protozoa usually requires CD4+T cell-, IFN-γ- and/or tumor necrosis factor (TNF)-dependent activation of macrophages. This leads to a (post)transcriptional upregulation of antimicrobial effector mechanisms, including the acidification of the phagolysosomes and the expression of inducible nitric oxide synthase (iNOS, NOS2; Bogdan, 2001; 2004; Fang, 2004), indoleamine-2,3-dioxygenase (IDO; Däubener and MacKenzie, 1999; Morrison, 2003), and of interferon-inducible GTPases (Martens and Howard, 2006). Although antibodies are frequently regarded as being irrelevant for the control of intracellular bacteria and protozoa, more recent studies demonstrate that they can both contribute to the development of the disease as well as to its control (Casadevall and Pirofski, 2006). Antibody-mediated aggravation of infections with intracellular pathogens might be due to Fc-receptor-mediated facilitation of the entry of the pathogen into the host cell or to macrophage deactivation conveyed by inhibitory Fc receptors (Kima et al., 2000; Buxbaum and Scott, 2005; Padigel and Farrell, 2005). Conversely, antibody-dependent control of intracellular microbes can result from antibody binding to the pathogen during intermittent extracellular phases (leading to opsonization and classical complement activation) (Bitsaktsis et al., 2007) or from promoting its uptake by DCs, which helps to elicit a protective T cell response (Woelbing et al., 2006).

One of the hallmarks of many intracellular pathogens (e.g. Herpes viruses, M. tuberculosis, Leishmania spp., T. gondii) is their ability to persist lifelong in the host organism after spontaneous or drug-induced resolution of the acute infection. The phenomenon of pathogen persistence without apparent symptoms and signs of infection raises a number of important questions: (i) What are the mechanisms that usually allow for continuous, lifelong control of the persistent pathogen and prevent the occurrence of spontaneous relapses? (ii) How does the pathogen manage to escape its elimination? (iii) Does pathogen persistence bear any benefit for the host organism as compared with a complete eradication of the pathogen? Specifically, is pathogen persistence required for maintaining a specific T-cell immunity against the respective infectious agent? (iv) What are the clinical implications of pathogen persistence?

The mouse models of cutaneous and visceral leishmaniasis have been particularly helpful in addressing these issues and therefore will form the focus of the following discussion. However, recent examples from other intracellular infections such as tuberculosis are also included where appropriate to enrich the discussion of the theme. For in-depth presentations of the mechanisms of persistence of viruses and bacterial pathogens such as mycobacteria, the reader is referred to previous detailed reviews (Honer zu Bentrup and Russell, 2001; Stewart et al., 2003; Monack et al., 2004; Finlay and McFadden, 2006; Roy and Mocarski, 2007).

Mechanisms of long-term pathogen control – a brief recap

In murine cutaneous leishmaniasis elicited by L. major promastigotes the resolution of the skin lesions requires IFN-γ-producing CD4+ T helper cells, which only develop in the presence of IL-12 and the signal transducer and activator of transcription 4 (STAT4) (Bogdan et al., 1993; Sacks and Noben-Trauth, 2002). CD8+T cells, which were dispensable for the ultimate control of high-dose footpad infections (reviewed in Bogdan et al., 1993; Ruiz and Becker, 2007), were later reported to be crucial in low-dose ear infection models (Belkaid et al., 2002a; Uzonna et al., 2004). B cells and antibodies aggravate or ameliorate the course of L. major infection depending on the genetic background of the mouse strains (Padigel and Farrell, 2005; Woelbing et al., 2006). IFN-γ and iNOS, which converts arginine into citrulline and leishmanicidal NO, turned out to be the key effector pathway of macrophages against the intracellular parasite stage (amastigotes) during the acute phase of infection in all models of L. major infection investigated (Belkaid et al., 2000; Bogdan et al., 2000a; Bogdan, 2007). However, it is important to note that the expression of iNOS alone is not necessarily sufficient to control L. major as has been seen in TNF-deficient mice (Wilhelm et al., 2001) or after infection with certain L. major strains (Anderson et al., 2005).

After clinical resolution of the skin lesions, small quantities of L. major parasites persist in the dermis and draining lymph nodes (Aebischer et al., 1993; Stenger et al., 1996), which continue to express iNOS mRNA and protein in a CD4+T cell-dependent manner (Stenger et al., 1996). Inhibition of iNOS led to prompt reactivation of the disease (Stenger et al., 1996). Although NADPH oxidase activity (phox) was redundant for the clinical cure of the acute skin lesions, phox was necessary for the containment of L. major in the spleen as well as for the long-term control of the parasites in the skin (Blos et al., 2003).

With respect to other Leishmania species that cause cutaneous leishmaniasis in mice and man (e.g. L. braziliensis, L. mexicana, L. amazonensis) highly relevant differences in the immune responses and courses of infection have been described (reviewed in McMahon-Pratt and Alexander, 2004), but the role of iNOS as an essential antileishmanial effector mechanism has been confirmed for most of them (Buxbaum et al., 2002; Qadoumi et al., 2002; Soares Rocha et al., 2007). In mice infected with L. donovani (visceral leishmaniasis) iNOS, but not phox, was essential for the clearance of the parasites in the liver. Unexpectedly, after chemotherapy with amphotericin, the standard treatment of visceral leishmaniasis, the residual parasites in the liver did not cause spontaneous post-treatment recrudescences even in the absence of iNOS and phox. Thus, an iNOS/phox-independent antileishmanial effector mechanism is likely to emerge in hepatic macrophages after treatment (Murray, 2005; 2006).

Mice infected with M. tuberculosis developed latent disease after tuberculostatic chemotherapy. Rapid and severe relapses occurred after inhibition of iNOS, as seen in cutaneous leishmaniasis, and also in the absence of TNF (MacMicking et al., 1997; Botha and Ryffel, 2002; 2003). Although T. gondii and T. cruzi are susceptible to iNOS-derived NO during the acute phase of infection, inhibition of iNOS during the chronic phase did not lead to reactivation of the disease (Schlüter et al., 1999; Saeftel et al., 2001). This further corroborates that iNOS-independent control of latent tissue parasites is possible.

Mechanisms of evasion and remodelling of the immune system

Entry into safe target cells

From an operational point of view survival of intracellular infectious pathogens in the host organism is best achieved when the pathogen resides in a cell type that per se is not able to exert antiviral or antimicrobial activities (Table 1). In Leishmania infections a number of cell types were postulated to function as ‘safe targets’ during long-term persistence. These include immature myeloid precursor cells and monocytes (L. major; Mirkovich et al., 1986), sialoadhesin-positive stromal macrophages of the bone marrow (L. infantum and L. donovani; Leclercq et al., 1996; Cotterell et al., 2000), hepatocytes (L. donovani; Gangneux et al., 2005) and fibroblasts (L. major; Bogdan et al., 2000b). For fibroblasts it was shown that they are readily infected by L. major promastigotes and amastigotes in vitro and in vivo, but are ineffective in killing these parasites (Bogdan et al., 2000b; Fig. 1). Whether L. major-infected fibroblasts serve to stimulate CD8+T cells as it has been seen with virally transfected fibroblasts (Kundig et al., 1995) has not yet been investigated.

Table 1. Overview of some of the mechanisms that facilitate long-term persistence of intracellular pathogens in vivo.
Pathogen entry into ‘safe’ target cells (reduced antimicrobial activity)
 • Fibroblasts*
 • Neurons
Pathogen resistance to host cell effector mechanisms
 • Synthesis of antioxidants (e.g. superoxide dismutase, catalase, peroxiredoxins, ROI/RNI-scavenging molecules of the cell surface)*
 • Upregulation of proteasomal activity (degradation of damaged proteins)
 • Metabolic salvage pathways (adaptation to nutrient restriction)
Suppression or avoidance of host cell effector mechanisms by the pathogen
 • Downregulation of the expression of NADPH oxidase* or iNOS*
 • Blockade of the phagosomal recruitment of NADPH oxidase* or iNOS
 • Inhibition of phagosome–lysosome fusion*
 • Exit into the cytosol
Pathogen-mediated immune deviation
 • Inhibition of antigen presentation*
 • Downregulation of costimulatory cell surface molecules*
 • Induction of IL-10*
 • Induction of regulatory/suppressive T cell populations (e.g. CD4+CD25+FoxP3+ Treg cells, CD4+CD25-FoxP3-IL-10+ Th1 cells, CD4+FoxP3-IL-10+IFN-γ-T cells)*
  • For details see text. Mechanisms that also apply for Leishmania parasites based on published data are marked with an asterisk.
Details are in the caption following the image

Hypothetical model for the persistence of Leishmania major during the chronic, clinically silent phase of infection in C57BL/6 mice. During latency live L. major amastigotes (red filled) reside in dendritic cells (DC), macrophages (MΦ) and fibroblasts. Whereas macrophages and, to some extent (not depicted), also dendritic cells express inducible NO synthase (iNOS) protein (presumably in response to IFN-γ+ Th1 cells), fibroblasts remain iNOS-negative (Bogdan et al., 2000b). The iNOS-derived NO kills some of the amastigotes in the macrophages (red stippled) and in neighbouring fibrobasts leading to a stable balance between parasite killing and evasion. Partial parasite survival results from the activity of IL-10 that is produced by macrophages and/or different populations of CD4+ T cells (Belkaid et al., 2002b; Anderson et al., 2007; Nagase et al., 2007). IL-10 inhibits the expression of iNOS (as well as TNF and NADPH oxidase; not depicted) and also induces arginase 1 activity (Arg1), which provides the polyamine precursor ornithin (Orn) to the parasites as a growth factor.

Another cell type that might function as immune privileged niche for pathogens are neurons. T. gondii successfully infected murine cerebellar neurons. Stimulation with IFN-γ and/or TNF did not enable these cells to control parasite invasion and replication (Schluter et al., 2001). Herpesviruses are known to establish latent infections in sensory ganglia in vivo (van Lint et al., 2005; Gupta et al., 2006; Hood et al., 2006). They do so by inhibiting apoptosis in neurons, e.g. by expression of a microRNA that downregulates transforming growth factor-β signalling (Gupta et al., 2006). There is evidence that Herpes simplex virus type 1-infected neuronal cells provide antigenic stimulation to CD8+T cells in sensory ganglia, which helps to suppress reactivation of the virus (van Lint et al., 2005).

Resistance to effector mechanisms

There are at least three different categories of mechanisms how an intracellular pathogen might become phenotypically resistant to effector pathways of host phagocytes: (i) synthesis of antioxidants that directly detoxify (i.e. neutralize or degrade) the effector molecules; (ii) the rapid degradation of microbial proteins that have been modified by the host cell-derived effector molecules and thereby became non-functional or toxic to the pathogen; and (iii) the adaptation of the microbial metabolism that helps the pathogen to bypass the effect of killer molecules or nutrient restrictions in the host cell. In the case of Leishmania parasites partial resistance to ROI or reactive nitrogen intermediates (RNI) has been described for the extracellular (promastigote) and for the intracellular (amastigote) stage of certain Leishmania species (Miller et al., 2000; Holzmuller et al., 2005; 2006; Giudice et al., 2007; Mukbel et al., 2007). Reported underlyingmechanisms are the putative ROI-scavenging function of phosphoglycans (Chan et al., 1989; Späth et al., 2003a,b) and the expression of superoxide dismutase, catalase and/or peroxiredoxins (Barr and Gedamu, 2003; Ghosh et al., 2003; Plewes et al., 2003; Adak and Datta, 2005; Harder et al., 2006), most of which were directly shown to support the survival of Leishmania within macrophages. A mutant of L. major deficient for all phosphoglycans (lpg2-) failed to replicate and survive in wild-type macrophages, but nevertheless retained the ability to persist lifelong in wild-type mice (Späth et al., 2003b). This finding indicates that it is not a single mechanism, but rather a combination of parasite virulence factors that allows Leishmania to establish persistent infections in mammalian hosts (Table 1).

Mycobacterium tuberculosis exhibits numerous mechanisms to defend itself against the oxidative and nitrosative stress of high concentrations of ROI and RNI, which are frequently paradigmatic for other intracellular pathogens. These include superoxide dismutase, catalase and peroxidase, truncated hemoglobins (which metabolizes NO to nitrate) and peroxiredoxins (which metabolize peroxides and peroxynitrite) (Manca et al., 1999; Piddington et al., 2001; Bryk et al., 2002; Pathania et al., 2002). Furthermore, the proteasome of M. tuberculosis was shown to be essential for resistance of the bacteria against RNI in vitro, presumably because it eliminates irreversibly oxidized, nitrated or nitrosated proteins (Darwin et al., 2003). More importantly, silencing of the M. tuberculosis proteasome in vivo using a tetracycline responsive repressor-promoter system led to a 2 log-reduction of the bacterial load in chronically infected mice (Gandotra et al., 2007). In Leishmania parasites, inhibition of the proteasome activity was shown to impair the growth of extracellular promastigotes and the survival of intracellular amastigotes (Robertson, 1999; Silva-Jardim et al., 2004), but whether the proteasome contributes to the defence against ROI or RNI is currently unknown.

Exposure of M. tuberculosis to low concentrations of NO or culture under hypoxic conditions reversibly inhibited aerobic respiration and bacterial growth and induced a specific transcriptional response (Voskuil et al., 2003) that is thought to provide the metabolic changes necessary for long-term survival of M. tuberculosis, such as the metabolism of fatty acids (Munoz-Elias and McKinney, 2005). A related observation was made with Salmonella enterica serovar Typhimurium, where NO inhibited the electron transport chain and the ensuing accumulation of NADH provided increased resistance of the Salmonella against peroxynitrite, hydrogen peroxide and hydroxyl radicals (Husain et al., 2008). Thus, host cell-derived effector molecules such as NO can exert paradoxic protective effects on intracellular pathogens, because they trigger additional antioxidant mechanisms and latency programs. Interestingly, some microbes such as Bacillus anthracis synthesize their own NO (via a bacterial NO synthase), which activates catalase and suppresses the formation of hydroxyl radicals and anions (Fenton reaction) inside the bacteria, thereby protecting them against ROI (H2O2) of macrophages (Shatalin et al., 2008).

An impressive example for the third category of resistance against host cell effector mechanisms was observed in urogenital serovars (D-K) of Chlamydia trachomatis. Human urogenital strains, but not ocular isolates of C. trachomatis, express a functional tryptophan synthase. This enzyme utilizes indole, a frequent product of the vaginal microbial flora, as a substrate for tryptophan synthesis. It thereby functions as a rescue mechanism to by-pass the depletion of the essential amino acid tryptophan that occurs in infected human epithelial cells owing to the IFN-γ-mediated induction of the tryptophan-depleting enzyme IDO (Caldwell et al., 2003).

Suppression or avoidance of host cell effector functions

Pathogens that end up in professional phagocytes and at the same time lack the expression of antioxidants or salvage pathways can still reach the stage of persistence, provided they manage to thwart the transcriptional induction or post-transcriptional activation of antimicrobial effector mechanisms, escape the phagolysosome or remodel their intracellular habitat into a safe compartment (Table 1).

In macrophages intact Leishmania parasites of various species or purified leishmanial compounds (e.g. lipophosphoglycan of the cell wall) were shown to interfere with a broad range of signal transduction events, including the activation of protein kinase C, the IFN-γ/JAK2/STAT1 cascade, and the MAP kinase-NF-κB pathway in response to phorbol esters, IFN-γ or lipopolysaccharide respectively. One possible molecular mechanism is the rapid upregulation of the macrophage phosphotyrosine phosphatase (PTP) activity, notably of the PTP SHP-1, after contact with Leishmania promastigotes (reviewed in Olivier et al., 2005). The increased PTP activity is sufficient to explain the impaired JAK2 and MAP kinase signalling and the reduced expression of iNOS that was seen in L. donovani-infected macrophages (Blanchette et al., 2007).

With a few possible exceptions (Haidaris and Bonventre, 1982; Murray and Cartelli, 1983; Grantt et al., 2001; Sousa-Franco et al., 2006), Leishmania promastigotes and amastigotes do not trigger the production of leishmanicidal amounts of superoxide or hydrogen peroxide by polymorphonuclear or mononuclear phagocytes. In fact, they inhibit the oxidative burst of these cells elicited by secondary stimuli such as bacterial peptides (e.g. N-formyl-methionyl-leucyl-phenylalanine), fungal cell wall components (zymosan), or phorbol myristate acetate (reviewed in Bogdan, 2007). Recent studies suggest several molecular mechanisms. In macrophages infected with L. pifanoi amastigotes the parasitophorous vacuole contained only the 65 kDa premature form of the gp91phox polypeptide. gp91phox together with p22phox and two haem groups forms the membrane-bound flavocytochrome b558 subunit of the NADPH oxidase. Superoxide production was bearly detectable in L. pifanoi-infected cells, unless they were treated with inihibitors of haem oxygenase 1, suggesting that the amastigotes induce an increase in haem degradation and thereby block the production of superoxide (Pham et al., 2005). In mouse macrophages L. donovani promastigotes led to the retention of the Rho-family GTPases Rac1 and Cdc42 on the phagosome membrane with the subsequent accumulation of a protective periphagosomal coat of F-actin that prevented the maturation (i.e. the endosomal–lysosomal fusion) of the phagosome (Lodge and Descoteaux, 2005; Lerm et al., 2006). Although the phosphorylation of the cytosolic NADPH oxidase component p47phoxand its association with p67phox remained intact in macrophages infected with L. donovani promastigotes, neither p47phox nor p67phox was recruited to the phagosomal membrane to assemble an active NADPH oxidase complex (Lodge et al., 2006). Similar observations were made after infection of macrophages with L. donovani amastigotes: the phosphorylation of p47phox and the phagosomal recruitment of p47phox and p67phox were defective and the phagosomal lipid raft integrity was disrupted (Lodge and Descoteaux, 2006).

In primary macrophages or in macrophage cell lines iNOS is efficiently induced by IFN-γ plus TNF or IFN-γ plus LPS and is then localized in the cytosol, in endosomal vesicles (nitroxomes) (Vodovotz et al., 1995), in the subcortical plasma membrane (Webb et al., 2001) and around phagosomes containing complement-opsonized or IgG-coated latex beads (Miller et al., 2004). The recruitment of iNOS to phagosomes was dependent on the scaffolding adaptor ezrin/radizin/moesin-binding phosphoprotein 50 (EBP50), which binds to iNOS and ezrin, a linker between the plasma membrane and the actin cytoskeleton. Phagosomes harbouring live M. bovis BCG or M. tuberculosis H37Rv showed a reduced capacity to retain EBP50 and a lower recruitment of iNOS to the phagosomes (Miller et al., 2004; Davis et al., 2007). Similar observations were also made with virulent S. enterica serovar Typhimurium strains, which induced iNOS in macrophages (owing to the presence of LPS), but blocked its translocation towards the Salmonella containing vacuoles in a Salmonella pathogenicity island 2-dependent manner (Chakravortty et al., 2002). Thus, virulent bacteria escape the antimicrobial activity of NO by establishing themselves in iNOS-negative compartments. Whether Leishmania parasites express virulence factors that prevent the translocation of iNOS-positive vesicles to parasitophorous vacuoles remains to be explored.

Considering the degradative potency of phagolysosomes a putatively successful strategy of Leishmania to survive inside macrophages would be their exit into the host cell cytosol. Indeed, there are anecdotal reports on cytosolic localization of Leishmania amastigotes observed by electron microscopy (reviewed in Rittig and Bogdan, 2000). Live videomicroscopy allowed us to witness newly endocytosed L. major or L. donovani promastigotes moving freely in the cytosol of mouse macrophages (Rittig et al., 1998). There is evidence for the expression of a pore-forming cytolysin (leishporin), especially in L. major and L. amazonensis promastigotes and amastigotes (Almeida-Campos and Horta, 2000; Noronha et al., 2000).

Whereas the leishporin still awaits its molecular and full functional characterization, the listeriolysin O (LLO)-dependent escape of the intracellular bacterium L. monocytogenes from its phagosome into the cytosol has been studied in great detail (Schnupf and Portnoy, 2007). Recently, an unexpected second LLO-dependent mechanism of survival of L. monocytogenes in macrophages was reported. It occurs when the expression of LLO is limited and is characterized by the formation of a novel compartment (spacious Listeria-containing phagosome) that requires autophagy, carries the late endosomal/early lysosomal marker lysosome-associated membrane protein-1 (LAMP-1) and is non-degradative owing to the LLO-mediated inhibition of phagosome acidification and phagosome–lysosome fusion (Birmingham et al., 2008). A possibly related LAMP-1-positive, cathepsin D-negative endosomal compartment in superficial or transitional epithelium cells of the bladder was found to harbour quiescent, non-replicating uropathogenic Escherichia coli, which for a long time were believed to be strictly extracellular bacteria (Mysorekar and Hultgren, 2006).

Inhibition of antigen presentation

Mouse Langerhans' cells, immature DCs and macrophages have all been described to phagocytose Leishmania promastigotes and to present leishmanial antigens to CD4+ T cells (Prina et al., 1996; Moll, 2000). However, macrophages infected with Leishmania parasites of different species (i) have a reduced capacity to present exogenous non-leishmanial peptide antigens and to elicit T cell receptor (TCR)-mediated reorientation of the microtubule-organizing centre (Meier et al., 2003; Chakraborty et al., 2005); (ii) develop a progressive inability to present endogenously synthesized parasite antigens to CD4+T cells during the differentiation of promastigotes into amastigotes. Reported underlying mechanisms are the intracellular sequestration of processed parasite antigens (Kima et al., 1996), the internalization and degradation of MHC class II molecules (Courret et al., 2001) and the parasite-driven increase of cell membrane fluidity with subsequent disruption of the lipid rafts of macrophages (Chakraborty et al., 2005). In human blood monocyte-derived DCs, infection with L. donovani caused inhibition of CD1a, CD1b and CD1c expression and prevented activation of CD1-restricted T cells (Amprey et al., 2004), whereas infection by L. amazonensis most prominently impaired the expression of CD80 and CD1a and diminished the production of IFN-γ in DC/T cell cultures (Favali et al., 2007). Together, these data suggest that the persistence of Leishmania amastigotes in the mammalian host is partly facilitated by inhibitory effects of the parasite on antigen-presenting cells.

Induction of interleukin-10 and T cells with regulatory potential

IL-10 was originally described as a cytokine produced by type 2 T helper cells (Th2) that limits the production of IFN-γ by type 1 T helper cells (Fiorentino et al., 1989). Later it was discovered that IL-10 strongly inhibits the production of TNF, ROI, IL-12 and, to a lesser extent, of RNI by macrophages and/or dendritic cells (Bogdan et al., 1991; 1992; Moore et al., 2001) and thereby strongly impairs their antimicrobial activity. Important sources of IL-10 in addition to Th2 cells are macrophages (de Waal Malefyt et al., 1991; Carrera et al., 1996), myeloid DC, keratinocytes, B cells, NK cells (Moore et al., 2001), naturally occurring regulatory CD4+T cells carrying the surface marker CD25 (IL-2 receptor α-chain) and expressing the transcription factor forkhead box P3 (CD4+CD25+ FoxP3+ Treg; Belkaid et al., 2002b), and various inducible T cell subsets with immunoregulatory/immunosuppressive functions. Per definition the latter arise from conventional CD25-FoxP3-T cells that were stimulated via their TCR, by antigen-presenting cells, cytokines such as IL-10 or TGF-β, or inorganic mediators (e.g. NO). They include CD4+ FoxP3- T regulatory 1 (Tr1; also termed inducible or adaptive Treg) that produce IL-10 and/or TGF-β (McGuirk et al., 2002); converted FoxP3+ Treg (Chen et al., 2003); IL-10-secreting CD8+ T cells (Noble et al., 2006); CD4+T-bet+CD25-FoxP3- Th1 cells that express IFN-γ and transiently also IL-10 (Anderson et al., 2007; Jankovic et al., 2007); and CD4+CD25+FoxP3-NO-Treg that develop from CD4+CD25- T cells after induction by NO (Niedbala et al., 2007). T cells with regulatory or immunosuppressive potential are now recognized as important players during the immune response against numerous pathogens, as they limit immunopathology and at the same facilitate pathogen survival (reviewed in Mills, 2004; Belkaid, 2007).

IL-10 and Leishmania persistence. A series of studies demonstrated that IL-10 is a master cytokine in cutaneous and visceral leishmaniasis that is critical for the initial survival and long-term persistence of Leishmania parasites in mice and man. First, Leishmania promastigotes and amastigotes activated macrophages for the expression of IL-10 in vitro and in vivo, especially if they were opsonized with antibodies or complement and thereby cross-linked deactivating Fcγ or complement receptors (Carrera et al., 1996; Mosser and Brittingham, 1997; Gerber and Mosser, 2001; Guizani-Tabbane et al., 2004; Miles et al., 2005; Padigel and Farrell, 2005; Thomas and Buxbaum, 2008). Second, IL-10 inhibited the production of NO and the antileishmanial activity of IFN-γ/TNF-stimulated mouse or human macrophages (Vieth et al., 1994; Vouldoukis et al., 1997). Third, mice deficient for IL-10 or treated with an anti-IL-10 receptor antibody showed a strongly improved control of L. major (Kane and Mosser, 2001; Noben-Trauth et al., 2003), L. donovani (Murphy et al., 2001; Murray et al., 2002; 2003), L. mexicana (Buxbaum and Scott, 2005) and L. amazonensis (Jones et al., 2002; Ji et al., 2003) during the acute phase of infection. Fourth, in genetically self-healing C57BL/6 mice, the long-term persistence of small numbers of L. major in the skin and draining lymph node was largely dependent on the expression of IL-10, because the vast majority of IL-10−/− mice or of mice treated with anti-IL-10 receptor antibodies during the chronic phase of infection were negative for parasites 9 weeks after infection (Belkaid et al., 2001). Finally, in patients with visceral L. donovani infection increased levels of IL-10 in the serum, in blood cells, the bone marrow or in splenic aspirates correlated with active disease (Nylen and Sacks, 2007). All these data point to a crucial role of IL-10 for parasite persistence in vivo. However, it is important to emphasize that infections with Leishmania species have been described in which deletion of IL-10 did not lead to sterile cure (e.g. L. amazonensis), indicating that IL-10-independent mechanisms of parasite persistence exist (Jones et al., 2002).

IL-10 and regulatory T cell subsets in chronic leishmaniasis. The data summarized above raise two questions: (i) What type of cell is the primary source of IL-10 during Leishmania infections? (ii) What is the functional role of the different regulatory T cell subsets? Both issues have been investigated in diverse experimental settings with partially discrepant results: (i) In Rag2−/− mice that were infected with L. major strain Friedlin and were reconstituted with CD4+CD25- T cells alone or in combination with CD4+CD25+ Treg from wild-type or IL-10−/− mice, IL-10 produced by Treg was essential for long-term parasite persistence, but not for the suppression of IFN-γ production of effector T cells (Belkaid et al., 2002b). (ii) In C57BL/6 mice, depletion of CD25+ T cells was less efficient in eliminating persistent L. major than the blockade of the IL-10 receptor (Belkaid et al., 2001; Mendez et al., 2004; Suffia et al., 2006), indicating additional sources of IL-10. (iii) In C57BL/6 mice infected with the L. major strain NIH/Seidman, which causes non-healing skin lesions with high parasite loads, CD4+CD25-FoxP3- Th1 cells that coexpressed IFN-γ and IL-10 rather than CD4+CD25+FoxP3+ Treg were found to account for the T cell-derived IL-10-dependent immune suppression in a Rag2−/− cell transfer model (Anderson et al., 2007). Consistent with this notion, blockade of the IL-10 receptor caused stronger decrease of the parasite tissue burden than the depletion of CD25+ T cells (Anderson et al., 2005). (iv) In BALB/c IL-4Rα−/− mice infected with the L. major strain LV39, CD4+CD25+FoxP3+ Treg accumulated in the ear skin lesions, but the majority of IL-10 was secreted by CD4+FoxP3- T cells that did not express IFN-γ and were therefore different from CD4+CD25-FoxP3- Th1 cells. Anti-CD25 treatment caused a significant decrease of the parasite load, whereas anti-IL-10 receptor therapy virtually eliminated the parasites in the skin (Nagase et al., 2007). (v) In C57BL/6 mice infected with L. mexicana depletion of CD25+ T cells did not alter the course of chronic infection, whereas IL-10−/− or Fcγ receptor III−/− mice resolved the lesions. In Fcγ receptor III−/− mice the expression of IL-10 mRNA in skin lesions was greatly diminished, whereas the production of IL-10 by CD25+CD4+ T cells was the same as in wild-type controls. As macrophages from Fcγ receptor III−/− mice were defective in the production of IL-10 in response to serum-opsonized amastigotes, the results suggest that macrophages are the main source of IL-10 during L. mexicana infection (Thomas and Buxbaum, 2008). (vi) In the spleen of L. donovani-infected C57BL/6 mice the percentage of IL-10-expressing splenocytes increased and the frequency of CD4+CD25+FoxP3+ Treg decreased as the infection progressed. At day 28 after infection almost all IL-10+ CD4+ T cells were negative for CD25 and FoxP3 and 40% of the IL-10+CD25-Foxp3-CD4+ T cells coexpressed IFN-γ (Stäger et al., 2006). (vii) In the spleen of patients with visceral leishmaniasis (L. donovani), CD4+CD25+FoxP3+ Treg neither accumulated nor mediated the suppression of antigen-specific IFN-γ responses. Instead, IL-10+CD4+CD25-FoxP3- T cells expressed the highest level of IL-10 and formed the predominant lymphocyte population (Nylen et al., 2007).

Together, these data indicate that depending on the parasite species, the experimental model and the time point of infection CD4+CD25+FoxP3+ Treg, CD4+CD25-FoxP3- Th1 cells, CD4+CD25-FoxP3-IFN-γ- T cells or macrophages are the primary producers of IL-10 during leishmaniasis and mediate the suppression of host immunity with subsequent persistence of parasites (Fig. 1, Table 1). Reported mechanisms of action of regulatory T cells are the inhibition of Th1 effector cell proliferation, the suppression of Th1 IFN-γ production, the deactivation of macrophages and the alternative activation of macrophages (Belkaid et al., 2002b; Anderson et al., 2007; Bopp et al., 2007; Nagase et al., 2007; Nylen et al., 2007; Tiemessen et al., 2007; Thomas and Buxbaum, 2008). Macrophages that were alternatively activated by IL-10 in combination with IL-4 express arginase 1, which converts arginine into urea and ornithine (Munder et al., 1999). Arginase 1 limits the availability of arginine for iNOS and can thereby inhibit the production of NO and the synthesis of iNOS protein (El-Gayar et al., 2003). Furthermore, it provides the precursor (ornithine) for the polyamine synthesis in eukaryotic parasites such as Leishmania (reviewed in Naderer and McConville, 2008). As arginase 1 activity was previously shown to correlate with progressive L. major infections in genetically non-healing mice (Iniesta et al., 2005; Kropf et al., 2005), it is conceivable that it also contributes to the long-term persistence of L. major in clinically healed mice (Fig. 1).

Signals for the induction and accumulation of regulatory T cells in leishmaniasis. In the L. major low-dose ear infection model, natural CD4+CD25+FoxP3+ Treg accumulated at the site of infection (Belkaid et al., 2002b; Mendez et al., 2004). The influx of these cells was dependent on their expression of the CCR5 chemokine receptor (which binds MIP-1α, MIP-1β and RANTES; Yurchenko et al., 2006) and of the αEβ7integrin (which interacts with the epithelial adhesion molecule E-cadherin that is highly expressed on keratinocytes and Langerhans' cells in the skin; Suffia et al., 2005). Notably, the majority of the natural Tregin the dermal infection site were specific for L. major antigens (Suffia et al., 2006). In BALB/c IL-4Rα−/− IL-10−/− mice infected with the L. major strain LV39 the accumulation of natural Treg (CD4+CD25+FoxP3+) in the ear dermis was drastically decreased compared to BALB/c IL-4Rα−/− or BALB/c wild-type mice, suggesting that IL-10 itself regulates the expansion and retention of natural Treg at the infection site (Nagase et al., 2007). Considering that NO can also serves as differentiation signal for regulatory T cells (Niedbala et al., 2007), one might speculate that the high-level expression of iNOS in healed skin lesions and draining lymph nodes (Stenger et al., 1996) contributes in two ways to the host–pathogen coexistence: it helps to contain Leishmania parasites and at the same time prevents its eradication by inducing regulatory T cells with macrophage-deactivating properties.

Pathogen persistence and T cell memory

The persistence of intracellular pathogens raises the question whether, from an evolutionary point of view, it has any benefit for the host organism. From an immunologist's perspective, there are at least two possible consequences. First, the clinically silent persistence of a defined pathogen leads to a continuous antigenic stimulation of the innate and adaptive immune system, which confers protection against unrelated pathogens. Indeed, such antigen non-specific protection has been observed in mice that were previously infected with a gamma herpesvirus, L. monocytogenes, or T. gondii and thereby developed increased resistance against L. monocytogenes or Yersinia pestis (Barton et al., 2007), fungal infections (Cryptococcus neoformans; Gentry and Remington, 1971), or infections with L. monocytogenes or S. enterica serovar Typhimurium (Ruskin and Remington, 1968) respectively. Second, a primary infection is expected to protect against the same pathogen (or different species of the same genus). Such antigen-specific, CD4+- and CD8+-T cell-mediated protection has been convincingly demonstrated for various Leishmania species in mice and man (reviewed in Bogdan et al., 1993; Handman, 2001; Khamesipour et al., 2005). Whether persistence of live Leishmania is required for maintaining a pool of protective memory T cells is still a matter of debate (Scott, 2005). IL-10−/− mice or anti-IL-10 receptor-treated mice, which developed sterile cure after primary infection with L. major, did not show clinical resistance to reinfection with L. major (Belkaid et al., 2002b). The presence of memory T cells in these mice was not investigated. In another study, BALB/c mice that had become asymptomatic, parasite-negative and seronegative within 9–12 months after a primary low-dose infection with L. major (i.e. no skin lesion, no palpable draining lymph node, no detectable parasites and no anti-Leishmania antibodies) also lost their parasite-specific effector T cells and their resistance to reinfection (Uzonna et al., 2001). Both studies therefore suggest that parasite persistence is required for maintaining T cell memory and immunity. A third study used a dihydrofolate reductase-thymidylate synthase mutant of L. major that is unable to cause pathology or to persist in C57BL/6 mice. Even 10 weeks after elimination of the parasites, when Leishmania-specific effector T cells were lost, lymph node-homing central memory T cells were still present in these mice and conferred significant protection against reinfection (Zaph et al., 2004). Thus, long-term and protective T cell memory is possible in the absence of live parasites.

Clinical implications and conclusions

In the case of cutaneous and visceral leishmaniasis, Leishmania parasites were shown to persist in the blood, in the unaffected skin, in scars or the lymph nodes of asymptomatic or clinically cured patients for decades after primary infection (Schubach et al., 1998; Costa et al., 2002; Dereure et al., 2003; Mendonca et al., 2004; Vergel et al., 2006). Loss of control of persistent pathogens by the immune system can cause fulminant reactivation of the infection. This has not only been documented for leishmaniasis, but also for infections with cytomegalovirus, M. tuberculosis, T. gondii or H. capsulatum. Reactivation is frequently seen when persistently infected patients undergo chemotherapy or bone marrow transplantation for haematologic disorders or neoplasms, receive immunosuppressive treatments for chronic lung disease or allergies (e.g. glucocorticosteroid therapy for asthma), autoimmune diseases (e.g. glucocorticosteroid or anti-TNF treatment of patients with Crohn's disease or rheumatoid arthritis) or for the prevention of host-versus-graft disease following organ transplantation, or acquire immunodeficiency owing to an infection with the human immunodeficiency virus. In all these cases, the T cell-, macrophage- and/or TNF-dependent containment of small quantities of tissue-resident microbes that have survived since the primary infection in the host organism will collapse and allow for the rapid replication and expansion of the pathogens (de La Rosa et al., 2002; Ljungman, 2002; Montoya and Liesenfeld, 2004; Basset et al., 2005; Young and McGwire, 2005; Winthrop, 2006). Quiescent intracellular reservoirs of uropathogenic E. coli that persist in the bladder epithelium are thought to be a source for recurrent urinary tract infections in humans (Mysorekar and Hultgren, 2006). In this case, it is not yet clear which component of specific defence mechanisms needs to break down to trigger the reactivation of the bacteria.

From the data summarized and discussed it can be concluded that persistence of intracellular microbes results from (i) the combination of pathogen-specific resistance factors that directly or indirectly protect against the toxic molecules of the host and (ii) from the deviation of the innate and adaptive immune response that generates safe compartments and niches in cells and organs for the survival of the pathogen. As long as the immune control by T cells and macrophages remains intact, there is little or no indication that pathogen persistence is detrimental to the host organism.

Acknowledgements

The preparation of this review and conduct of some of the studies reviewed was supported by grants to the author from the German Research Foundation (Bo996/3–1, 3–2 and 3–3; SFB620, A9).The author is grateful to Dr Ulrike Schleicher for reviewing the manuscript.

      The full text of this article hosted at iucr.org is unavailable due to technical difficulties.