Identification of a Compensatory Mutant (lpg2-REV) of Leishmania major Able To Survive as Amastigotes within Macrophages without LPG2-Dependent Glycoconjugates and Its Significance to Virulence and Immunization Strategies

Different Leishmania species rely to different extents on abundantglycoconjugates, such as lipophosphoglycan (LPG) and relatedmolecules, in mammalian infections. Previously, we showed thatLeishmania major deletion mutants lacking the Golgi GDP-mannosetransporter LPG2, which is required for assembly of the dominantphosphoglycan (PG) repeats of LPG, were unable to survive inmacrophages. These lpg2^– mutants, however, retained theability to generate asymptomatic, persistent infections in mice.In contrast, Ilg and colleagues showed that Leishmania mexicanaLPG2 mutants retained virulence for mice. Here we identifieda partial revertant population of the L. major lpg2^– mutants(designated lpg2^–REV) that had regained the ability toreplicate in macrophages and induce disease pathology througha compensatory change. Like the lpg2 parent, the lpg2^–REVrevertant was unable to synthesize LPG2-dependent PGs in thepromastigote stage and thus remained highly attenuated in theability to induce infection. However, after considerable delaylpg2^–REV revertant-infected mice exhibited lesions, andamastigotes isolated from these lesions were able to replicatewithin macrophages despite the fact that they were unable tosynthesize PGs. Thus, in some respects, the lpg2^–REV amastigotesresemble L. mexicana amastigotes. Future studies of the gene(s)responsible may shed light on the mechanisms employed by L.major to survive in the absence of LPG2-dependent glycoconjugatesand may also improve the potential of the lpg2^– L. majorline to serve as a live parasite vaccine by overcoming its tendencyto revert toward virulence.

INTRODUCTION

Top

Introduction

Infections by the protozoan parasite Leishmania can cause severediseases in humans and animals, whose manifestations range frommild cutaneous to fatal visceral pathology. Leishmaniae aretransmitted in the form of highly virulent metacyclic parasites,which develop inside the midgut of phlebotomine sand flies andare inoculated into the mammalian host by biting. Inside theacidified phagolysosome of host macrophages, leishmaniae differentiateinto the aflagellate amastigote stage, which resists leishmanicidalactivities and inhibits macrophage activation, processes thatultimately contribute to amastigote survival and replicationand disease pathology (8). In most immunocompetent hosts, parasitegrowth and pathology are eventually controlled by adaptive immunemechanisms (2, 20), and the infection enters a chronic phaseof parasite persistence in the absence of significant pathology.There are a number of host factors which are known to modulateparasite persistence and establish a delicate balance betweengamma interferon-induced leishmanicidal NO and immunesuppressiveinterleukin-10 produced by CD25⁺ regulatory T cells (1, 2, 9,25). In contrast, our knowledge concerning the identities androles of parasite factors which function differentially in mediatingpersistence and in the acute pathogenic phase of infection ismore limited.

Glycosylphosphatidyl inositol-anchored surface glycoconjugates,including lipophosphoglycan (LPG), glycoinositol phospholipids(GIPLs), and proteins, such as proteophosphoglycan (PPG) orgp63, form a dense glycocalyx in Leishmania. Many of these compoundsare important determinants of parasite virulence, as shown bystudies of purified molecules, as well as mutants defectivein the synthesis of one or more of the glycoconjugates (11,12, 14, 22-24, 28). In Leishmania major, LPG plays an importantrole in the establishment of infections in macrophages followingmetacyclic invasion, while both GIPLs and other alkylacyl phosphoglycerolipidmoeities appear to have small roles in macrophage survival despitetheir abundance (12, 33). In contrast, parasite molecules thatare dependent on the activity of the LPG2-encoded Golgi GDP-mannosetransporter, such as LPG and other phosphoglycans (PGs) bearingthe signature Gal-Man-P-based repeating unit, are required forboth initial establishment and subsequent survival and replicationof amastigotes within macrophages (24). It should be emphasizedthat the findings described above pertain to L. major and probablyLeishmania donovani as well (M. Wilson and S. M. Beverley, unpublisheddata). Remarkably, studies of Leishmania mexicana have suggestedthat this species requires neither LPG, PGs, nor GIPLs for establishmentor replication within macrophages (7, 10). This difference inreliance on highly abundant, structurally conserved moleculesamong Leishmania species was unanticipated and is not well understood(4, 29).

Significantly, the L. major lpg2^– parasite maintainedthe ability to persist at the infection site in the absenceof overt pathology for more than 2 years (24). This ruled outthe possibility that LPG2-dependent factors (such as PGs) arerequired in the persistent, asymptomatic phase of the disease.To explain the fact that the lpg2^– parasite is unableto survive in macrophages yet is able to survive indefinitelyin animals, we suggested a model in which persistent parasitesresided in another, presumably more hospitable compartment withinthe host, a compartment from which the parasite would be unableto induce disease through perturbation of macrophage function.A corollary of this safe haven model is that parasites thathad recovered the ability to survive within macrophages wouldsimilarly recover the ability to induce disease pathology. Herewe present evidence which supports this prediction and discussits relevance to Leishmania virulence and the use of live parasite-basedimmunization strategies.

MATERIALS AND METHODS

Top

Materials and Methods

Parasite strains and culture. All parasites were derivatives of L. major strain LV39 clone5 (Rho/SU/59/P) and were cultivated in M199 medium (13). Thelpg2 line obtained by homologous gene replacement ( ${Delta}$ LPG2::HYG/ ${Delta}$ LPG2::HYG)has been described previously (24). P+ and P– derivativesof this line (Fig. 1A) were transfected with the construct pSNBR-LPG2(B4831) and plated on semisolid M199 medium containing 20 µgof G418 per ml.

View larger version (32K):
[in this window]
[in a new window]
FIG. 1. Emergence of the lpg2^–REV compensatory partial revertant during mouse infection. (A) Lesion formation. A total of 10⁶ WT or lpg2^– metacyclic parasites were inoculated into groups of susceptible BALB/c mice, and the infections were monitored by comparing the thickness of the injected footpad to the thickness of the uninjected footpad with a vernier caliper. Inoculation of WT parasites resulted in infections that appeared within 25 days ( ${circ}$ ), and, as shown previously, five of eight mice infected with lpg2^– parasites showed no lesion formation (P–) ( ${diamond}$ ). In contrast, three of eight mice infected with lpg2^– parasites in this experiment showed lesion formation beginning around day 125 (P+) ( ${square}$ ); the parasites in these mice were designated lpg2^–REV parasites. The bars indicate the standard deviations for WT parasites (n = 4), P– lpg2^– parasites (n = 5), and P+ lpg2^– parasites (n = 3). (B) Limiting dilution assay. Representative P– or P+ mice from the experiment shown in panel A were sacrificed on day 346 postinfection, and the numbers of parasites in the infected footpads were determined by limiting dilution. The lesion sizes (upper panel) and the parasite burdens (lower panel) of individual animals are indicated.

Monoclonal antibodies, lectins, immunofluorescence, and flow cytometry. Monoclonal antibody WIC79.3 recognizes L. major Gal-substitutedGal-Man-P repeating units (6), and anti-trypanosomal tubulinantibody was provided by D. Russell (Cornell University, Ithaca,N.Y.). Flow cytometry with fluorescein-conjugated ricin agglutinin(Sigma, St. Louis, Mo.) was performed with live Leishmania.Parasites in the logarithmic growth phase (10⁵ parasites/ml)were washed twice in cold phosphate-buffered saline and incubatedwith 5 µg fluorescein-conjugated ricin agglutinin perml in phosphate-buffered saline for 10 min at 4°C. The analysiswas performed with a Becton Dickinson FACS-Calibur system (excitationwavelength, 488 nm; detection wavelength, 520 nm). Cells werefixed with 4% paraformaldehyde for 5 min at room temperature,permeabilized by incubation with ice-cold 100% ethanol for 15min at 4°C, and sequentially incubated with monoclonal anti-PGantibody WIC79.3 (27) and fluorescein isothiocyanate-conjugatedanti-mouse immunoglobulin G. Parasites were counterstained withpolyclonal anti-tubulin and Texas Red-conjugated anti-rabbitimmunoglobulin G.

Mouse infection and limiting dilution assay. Virulence was assessed by using 6- to 8-week-old female BALB/cmice (Charles River Laboratories Inc., Wilmington, Mass.) followingsubcutaneous inoculation of 10⁶ metacyclic parasites into thehind footpads (26). The metacyclic parasites were isolated bydensity centrifugation as described previously (21). Infectionswere monitored by comparing the thickness of the injected footpadto the thickness of the uninjected footpad with a vernier caliper.Parasites in the infected tissue were enumerated by a limitingdilution assay (26).

Macrophage infection. Starch-elicited peritoneal exudate macrophages (PEM) from BALB/cmice were infected as described previously (22). Metacyclicparasites were opsonized by 30 min of incubation in Dulbeccomodified Eagle medium containing 4% C5-deficient mouse serum(16) and were allowed to invade macrophages in Dulbecco modifiedEagle medium containing 0.7% bovine serum albumin for 2 h at33°C at multiplicities of infection of 10 and 3 parasitesper PEM for wild-type (WT) and lpg2 parasites, respectively.Amastigotes were isolated from progressive lesions by tissuehomogenization and differential centrifugation, and PEM wereinfected for 2 h at 33°C at a ratio of three parasites perPEM. Intracellular growth was determined by nuclear stainingand fluorescence microscopy as described previously (22).

RESULTS

Top

Results

Emergence of an lpg2^– parasite population showing late pathology. In previous studies performed in three different laboratories,it was found that in most infections of susceptible BALB/c micewith the lpg2^– null mutant, parasites were able to persistindefinitely without pathology (24, 30). In these experiments,footpad inoculation was used to infect groups of female BALB/cmice with 10⁶ WT or lpg2^– metacyclic parasites that wereisolated by Ficoll density centrifugation (21). WT metacyclicparasites elicited lesions that appeared within 25 days (Fig.1A), while in most experiments lpg2^– metacyclic parasitesfailed to induce pathology, in agreement with previous findings(24). However, in two experiments involving inoculation withhigher numbers of parasites, we observed that a few mice developedlesions, which developed late (after 150 to 200 days) but progressedrapidly thereafter. For example, in the experiment shown inFig. 1A, three of eight of the lpg2^– parasite-infectedmice showed this same late-lesion profile (P+ mice), while theremaining mice showed no pathology, as observed previously (24).The numbers of parasites were determined by a limiting dilutionassay (26) on day 346 postinfection, and the results showedthat, as expected, infected mice that did not exhibit overtpathology (mice 3-3, 4-3, 6-1, and 6-5) contained a low levelof persistent parasites ( $~$ 10³ parasites) (Fig. 1B). In contrast,in mice with the late-lesion pathology (mice 4-4, 6-3, and 6-4),the numbers of parasites were nearly 10,000-fold higher (Fig.1B).

The timing of lesion appearance differed from the timing observedpreviously for LPG-deficient lpg1^– L. major, which requiresLPG for establishment of a macrophage infection but not forreplication as amastigotes (22). Prompted by the suppositionthat this finding might reflect a new genetic difference betweenthe late-lesion lpg2^– parasites and the canonical persistentlpg2^– parasites, we recovered parasites separately frommice which never exhibited lesion formation (designated P–)and mice which had lesions (designated P+) (Fig. 1A).

Late-pathology lpg2^– P+ parasites do not make PGs in the promastigote form. Following recovery, the two populations of parasites were allowedto differentiate into promastigotes in vitro, and PG expressionwas analyzed by examining reactivity with anti-PG antibody WIC79.3in a Western blot analysis (Fig. 2A) and by flow cytometry withthe ß-Gal-binding lectin ricin agglutinin (Fig. 2B).WT promastigotes expressed an abundance of LPG and PPG, as shownby Western blotting, while the lpg1^– mutant (defectivein LPG core synthesis) showed the expected synthesis of PPGbut did not express LPG (Fig. 2A). Importantly, the lpg2^–P– and P+ populations maintained the total PG-deficientphenotype expected for the lpg2^– promastigotes, showingthat they were neither WT contaminants nor second-site revertantsthat exhibited restored promastigote PG synthesis. It shouldbe emphasized that the LPG2 coding region had been removed completelyfrom the lpg2^– line by homologous gene replacement (24),and thus reconstitution of LPG2 itself was impossible. Interestingly,inspection of the available L. major genome sequence data revealedthe presence of several members of the nucleotide sugar transporterfamily whose functions are unknown, in addition to the anticipatedUDP-Gal transporters (A. Capul and S. M. Beverley, unpublisheddata). However, activation of an alternative GDP-Man transportactivity in P+ lpg2^– promastigotes was ruled out by thesedata.

View larger version (36K):
[in this window]
[in a new window]
FIG. 2. Characterization of P+ and P– lpg2^– promastigotes. (A) Western blot analysis. Promastigotes from WT, lpg1^–, and lpg2^– P+ and P– populations were subjected to Western blot analysis with the anti-PG antibody WIC79.3 (6). The positions of WT PPG and LPG are shown. (B) PG flow cytometry. A total of 10⁵ promastigotes from a logarithmic culture were incubated with ricin-fluorescein isothiocyanate (FITC) at 4°C prior to analysis. (Upper panel) WT and lpg2^– P– parasites; (lower panel) lpg2^– P+ parasites. The parental lpg2^– population was indistinguishable from the P– and P+ populations (data not shown).

P– and P+ lpg2^– parasites breed true. When 10⁶ metacyclic parasites were inoculated into mice, theP– lpg2^– parasites yielded persistent infectionssimilar to those seen previously; as in previous studies, noinfected mouse went on to develop lesions or pathology overa 1-year period (Fig. 3A and data not shown). Notably, restorationof LPG2 expression restored PG synthesis to WT levels (Fig.3C) and lesion formation to control levels (Fig. 3A). Theseresults demonstrated that the P– lpg2^– parasiteswere not generally and irreversibly avirulent, a potential concernsince loss of virulence occurs sporadically during long-termculture or genetic manipulation of Leishmania in vitro (5).Thus, the P– lpg2^– parasites appear to be identicalto the original lpg2^– population described previously(24).

View larger version (43K):
[in this window]
[in a new window]
FIG. 3. P+ lpg2^– population breeds true. (A and B) Mouse infections. A total of 10⁶ purified WT ( ${blacklozenge}$ and ${blacksquare}$ ), lpg2^– ( ${diamond}$ ), or lpg2^– P+ ( ${square}$ ) metacyclic parasites were inoculated into the footpads of BALB/c mice, and lesion formation was monitored. Each point indicates the average for three mice inoculated with WT parasites or five mice inoculated with lpg2^– or lpg2^– P+ parasites; the error bars indicate standard deviations. (C and D) Restoration of LPG2 expression fully restores surface LPG expression in P+ and P– lpg2^– lines. LPG expression in promastigotes of WT parasites (solid profile) and the lpg2^– (P–)/+ LPG2 (panel C, open profile) or lpg2^– (P+)/+ LPG2 addback lines was assessed by flow cytometry with fluorescent ricin as described in Materials and Methods. As shown previously, the dominant PG detected by this assay is LPG (22). FITC, fluorescein isothiocyanate.

In contrast, the P+ lpg2^– parasites produced lesions inevery infected mouse, again with a delay of $~$ 125 days (Fig. 3B).This showed that the late-lesion or P+ phenotype was a stabletrait, and we designated this population lpg2^–REV. Asin the lpg2^– parental line, reexpression of LPG2 in thelpg2^–REV population restored LPG and PG synthesis (Fig.3D).

lpg2^–REV parasites survive in macrophages in the amastigote form but not in the promastigote form. Infection of mouse macrophages with lpg2^– (P–) orlpg2^–REV (P+) metacyclic parasites showed that neithersurvived, as observed previously (24; data not shown). Whilelpg2^– (P–) amastigotes could not be obtained, wewere able to recover amastigotes from mouse lesions obtainedwith lpg2^–REV parasites (obtained around day 150). Thelpg2^–REV amastigotes were capable of surviving and replicatingwell in macrophages, although the growth rate was about twofoldlower than that of WT amastigotes (Fig. 4B). The size of thelpg2^–REV parasitophorous vacuole was similar to that seenin other L. major infections, and the vacuole was not extendedlike that seen in L. mexicana infections (Fig. 4A).

View larger version (27K):
[in this window]
[in a new window]
FIG. 4. lpg2^–REV parasites replicate in macrophages but do not synthesize PGs. (A) Immunofluorescence analysis of amastigote PGs. PEM were infected with WT and lpg2^–REV amastigote parasites and stained 2 days postinfection for reactivity with tubulin or PGs as described in Materials and Methods. Bar = 10 µm. (B) Macrophage replication. Murine macrophages were infected with WT ( ${circ}$ ) or lpg2^–REV ( ${square}$ ) amastigotes as described in Materials and Methods, and the numbers of parasites were determined and normalized to the initial infection obtained at 2 h (day zero). The values are averages and standard deviations for a representative experiment performed in triplicate.

lpg2^–REV parasites do not make PGs. As described above for promastigotes, it was possible that thelpg2^–REV amastigotes had recovered the ability to synthesizePGs through activation of an amastigote-specific pathway thatrestored Golgi GDP-Man uptake. In WT amastigotes, PGs predominantlylocalized to the parasite cytoplasm and accumulated in a vesicularcompartment, most likely the flagellar pocket, as expected forsecreted PGs (Fig. 4A, left panels). However, immunofluorescenceanalysis of the infected macrophages with anti-PG antisera showedthat the lpg2^–REV amastigotes failed to synthesize detectablelevels of PGs, and only background fluorescence was evident(Fig. 4A, right panels). These data eliminated the possibilitythat there was activation of an alternative amastigote GolgiGDP-Man transport activity in this line.

DISCUSSION

Top

Discussion

In previous studies it was shown that LPG was important forthe ability of L. major to establish infections in macrophages,while LPG2-dependent molecules, such as LPG and related PGs,were required for replication of parasites within macrophagesin the amastigote form and subsequent pathology (24). The lpg2^–parasites used in these studies failed to induce any pathologyin mice, although they were able to persist indefinitely inthe absence of overt lesion formation. In this study we characterizeda variant of the lpg2 line designated the lpg2^–REV line,which is able to reproducibly produce pathology in mice, albeitafter a considerable delay (Fig. 1A). Notably, neither promastigotenor amastigote forms of the lpg2^–REV parasite regainedthe ability to synthesize PGs, and thus this mutant did notarise through activation of an alternative Golgi GDP-Man transporteractivity.

In our experience the delayed-lesion phenotype manifested bythe lpg2^–REV parasite is characteristic of mutants withmutations that affect the ability of Leishmania to establishinfections in macrophages (e.g., to enter, survive, and differentiate)but not the ability to replicate in the amastigote form. Thismay be due to defects in macrophage survival (as observed previouslywith LPG-deficient LPG1 null mutants or ether phospholipid-deficientADS1 null mutants [22, 33]) or to differentiation to form theinfectious metacyclic stage (as observed in sphingolipid syntheticmutants [32; unpublished data]). Consistent with this notion,lpg2^–REV parasites were poorly infective in macrophagesin the promastigote form, reflecting the absence of PGs, suchas LPG. However, in striking contrast to the amastigote formof the lpg2^– parent, lpg2^–REV amastigotes showedgood survival and replication in macrophages (Fig. 4B). Thus,these parasites in some manner acquired a compensatory mutationthat enables amastigote replication in the absence of PGs orperhaps other molecules dependent on the activity of LPG2.

The following hypothetical scenario accounts for the emergenceand properties of the lpg2^–REV parasites (Fig. 5A) (24).As shown previously, in a normal infection with metacyclic promastigotesmost lpg2^– parasites were phagocytosed by macrophagesand destroyed, while a small proportion of the parasites escaped,possibly by residing in a cell type other than macrophages thatlacked leishmanicidal potential (3, 22). In this presumptivesafe haven the lpg2^– parasites were able to persist indefinitely(>700 days) at low levels ( $~$ 1,000 parasites). Little is knownabout the parasites during this stage, including their differentiationstate or whether they replicate. One possibility is that thepersistent parasites slowly turn over and constantly leave theirsafe haven, where they are then challenged by host macrophages.This possibility is consistent with the role of negative hostfactors, such as NO, which mediate parasite killing by macrophages,thereby keeping persistent Leishmania infections in check (25).For the lpg2^– parasites, entry into macrophages wouldlead to destruction until the emergence of variant cells thathad acquired the ability to survive and replicate in macrophages,such as lpg2^–REV cells, which ultimately would lead todisease pathology. Since the lpg2^–REV parasites did notregain the ability to synthesize LPG or PGs, they remained attenuatedin the ability to establish macrophage infections initiallyin the promastigote form, as observed here. This model makesseveral testable predictions regarding the nature of the hostcell phenotype, its leishmanicidal potential, and the interactionwith the host immune system. However, validating these predictionsrequires identification of the persistent lpg2^– parasitepopulation, a challenging task which is currently under way.

View larger version (38K):
[in this window]
[in a new window]
FIG. 5. Summary of comparisons of lpg2^–REV L. major. In both panels, the known role of LPG2 is in transport of GDP-mannose into the parasite Golgi apparatus, where it is utilized for the synthesis of PGs of various forms. (A) Proposed working model for pathology and persistence of WT parasites (upper panel) and lpg2^– or lpg2^–REV L. major parasites (lower panel) in mouse infections. See Discussion for more information. This figure is modified from that presented by Spáth et al. (24). (B) L. major lpg2^–^–REV parasites have a PG dependence phenotype intermediate between those of WT L. mexicana and L. major. See Discussion for more information. M ${phi}$ , macrophage.

The emergence of the lpg2^–REV phenotype is relevant tothe development of live vaccines in which the lpg2^– parasiteis used (30). To date, numerous protocols in which recombinantleishmanial antigens or heat-killed Leishmania have been usedhave failed to induce long-lasting immunity (15, 18, 19). Incontrast, long-term protection is achieved following recoveryfrom a natural infection, suggesting that parasite persistenceis a prerequisite for sterile immunity (2, 31). lpg2^–parasites fulfill the major requirements for a live vaccinecandidate (i.e., persistent infection in the absence of disease),and vaccination studies with BALB/c mice and lpg2^– parasitesindeed induced dramatic protection against virulent challengein the absence of a strong Th1 response (30). An interestingquestion for future study is the contribution of the host immuneresponse to the delayed emergence of lpg2^–REV parasiteinfections and whether this may provide some further perspectiveon the nature of the immune response to lpg2^– parasites.Regardless, it is very clear the lpg2^–REV phenotype compromisesthe potential utility of the lpg2^– line as a vaccine candidate,and thus preventing its occurrence is important.

In contrast to L. major, L. mexicana does not require PGs forvirulence or amastigote replication, implying that there arePG-independent mechanisms of macrophage survival (7, 10). Ina recent review the possibility that different Leishmania speciesplaced different levels of emphasis during evolution on theuse of surface glycoconjugates for intracellular survival wasdiscussed (29). The differences among the species are summarizedin Fig. 5B, which shows that L. major requires LPG2-dependentglycoconjugates, such as LPG and/or PGs, for establishment ofmacrophage infections by promastigotes, as well as for survivaland induction of pathology by amastigotes, while L. mexicanadoes not. Interestingly, the L. major lpg2^–REV line exhibitsintermediate behavior; it is deficient in the ability to establishmacrophage infections without PGs, like L. major, but it retainsthe ability to survive and replicate in the amastigote formwithout PGs, like L. mexicana. It is tempting to speculate thatthe gene(s) or pathway(s) altered in the L. major lpg2^–REVline is related to the pathways that enable L. mexicana to replicatein macrophages in the absence of PGs. Some interesting glycoconjugatesrecently implicated in amastigote virulence in L. mexicana arethe ß1-2 mannans, which are also found in L. major(17; unpublished data). However, the cytoplasmic location ofmannans makes it unlikely that their synthesis is LPG2 dependent(17).

It may be possible to exploit the differences between the L.major lpg2^– and lpg2^–REV lines identified in thisstudy in genetic strategies to identify the relevant genes responsible.Thus, the lpg2^–REV line provides a unique tool that mayfacilitate studies to gain important insight into mechanismsof vaccination and virulence.

ACKNOWLEDGMENTS

We thank A. Capul and K. Zhang for reading the manuscript andP. Scott and J. Uzonna for thoughtful discussions about thenature of the immune response to the lpg2^– and lpg2^–REVparasites.

This work was supported by NIH grant AI31078 (to S.J.T. andS.M.B.), by the Deutscher Akademischer Austauschdienst (DAAD),and by the Human Frontiers Science Program (G.F.S.).

FOOTNOTES

* Corresponding author. Mailing address: Department of Molecular Microbiology, Washington University Medical School, Campus Box 8230, 660 S. Euclid Ave., St. Louis, MO 63110. Phone: (314) 747-2630. Fax: (314) 747-2634. E-mail: beverley{at}borcim.wustl.edu.

Editor: W. A. Petri, Jr.

Present address: Department of Medical and Molecular Parasitology,New York University School of Medicine, New York, NY 10010.

REFERENCES

Top