Binding of Leishmania infantum Lipophosphoglycan to the Midgut Is Not Sufficient To Define Vector Competence in Lutzomyia longipalpis Sand Flies

It is well established that the presence of LPG is sufficient to define the vector competence of restrictive sand fly vectors with respect to Leishmania parasites. However, the permissiveness of other sand flies with respect to multiple Leishmania species suggests that other factors might define vector competence for these vectors. In this study, we investigated the underpinnings of Leishmania infantum survival and development in its natural vector, Lutzomyia longipalpis. We found that LPG-mediated midgut binding persists in late-stage parasites. This observation is of relevance for the understanding of vector-parasite molecular interactions and suggests that only a subset of infective metacyclic-stage parasites (metacyclics) lose their ability to attach to the midgut, with implications for parasite transmission dynamics. However, our data also demonstrate that LPG is not a determining factor in Leishmania infantum retention in the midgut of Lutzomyia longipalpis, a permissive vector. Rather, LPG appears to be more important in protecting some parasite strains from the toxic environment generated during blood meal digestion in the insect gut. Thus, the relevance of LPG in parasite development in permissive vectors appears to be a complex issue and should be investigated on a strain-specific basis.

developing in P. papatasi (12), nor was wild-type (WT) Le. major fed along with anti-PpGalec antibodies (11). In both cases, the infections were mostly lost once the digested blood was passed. Le. major lpg1 knockout (KO) parasites also failed to develop in another natural vector of Le. major, Phlebotomus duboscqi (25,26).
Although the factors defining vector competence of the sand fly P. papatasi for Le. major are well known (6), they may not govern interactions of other sand fly-Leishmania pairs (12,(24)(25)(26)(27)(28)(29). In fact, purified LPGs from multiple Leishmania species bind to the midgut of the sand fly P. argentipes (12), a permissive vector, despite exhibiting divergent carbohydrate side chains (16). Interestingly, Le. major and Le. tropica outcompete Le. infantum binding to midguts of its natural vector Lu. longipalpis when simultaneously in contact with the epithelium (29). In addition, the Le. major lpg1 knockout line can successfully develop in both permissive Lu. longipalpis (25) and permissive P. perniciosus (26) sand flies. Even though an LPG-independent mechanism based on a potential Leishmania lectin attaching to the microvilli glycocalyx has been proposed for Leishmania development in permissive vectors (24,28,30) and a candidate midgut mucin of 45 kDa that binds to Le. major was identified previously (30), confirmation of its function as a midgut receptor is yet to be demonstrated. Further, there is also a possibility that this phenomenon is restricted to Le. major development in permissive vectors and is not extendable to other Leishmania species naturally transmitted by such vectors. Such species-specific features have been demonstrated for the FLAG1/SMP1 flagellar protein that mediates midgut attachment of Le. major to P. papatasi but not Le. infantum to Lu. longipalpis (31,32), and the data are further supported by apparent survival of Le. infantum line Δlpg1 in P. perniciosus (27) and of Le. mexicana line Δlpg1 in Lu. longipalpis (33).
The nature of the mechanisms defining vector competence in permissive sand fly vectors is still an open issue. The importance of LPG for midgut binding and parasite survival needs to be further analyzed, particularly as the lack of Δlpg1 ϩ LPG1 add-back lines in a previous study (27) precluded a definitive conclusion about the importance of LPG in vector competence of permissive vectors. Here, we assessed midgut binding and survival and development of Le. infantum in the sand fly Lu. longipalpis using multiple wild-type strains as well as two Δlpg1 mutants and Δlpg1 ϩ LPG1 add-back lines to answer the following questions. (i) Does Le. infantum bind to the midgut of Lu. longipalpis? (ii) If it does, is the Leishmania binding to the midgut stage specific? (iii) Is Le. infantum LPG necessary for parasite binding to the midgut? (iv) Is the Le. infantum LPG sufficient to define vector competence in the natural permissive vector? (v) Do components of the blood bolus affect Le. infantum survival?
Le. infantum binding to the midgut epithelium of Lu. longipalpis is stage dependent. To validate and extend our observations, we used a different strain of Le. infantum, MCAN/BR/09/52 (LAB), and assessed its binding to sand fly midguts. Parasites harvested on day 3 attached to the midgut epithelium of Lu. longipalpis at a median of 5,250 parasites/midgut ( Fig. 2A). As the LAB parasite cultures aged, the number of parasites binding the sand fly midgut increased proportionally, peaking with day 6 parasites at a median of 8,500 parasites/midgut ( Fig. 2A). A greater proportion of parasite binding to the midgut epithelium as the culture aged was also observed for the Le. infantum BH46 strain (see Fig. S1A to D in the supplemental material), ranging from a median of 4,200 parasites/midgut for 4-day-old culture parasites (Fig. S1A) to a median of 26,000 parasites/midgut with 6-day-old culture parasites (Fig. S1C). As the culture got older ( Fig. 2B; see also Fig. S2A), the parasites began to differentiate to the infective form, the metacyclic promastigotes, increasing from 10% to 30% on day 4 to about 70% to 80% on day 7 of culture ( Fig. 2C; see also Fig. S2B). As we had observed that many, if not most, of the bound parasites were at late stages of differentiation (leptomonads and metacyclic), we used a Ficoll gradient to separate early-stage parasites (ESP) from late-stage parasites (LSP) and tested if they could bind to sand fly midgut epithelium. Surprisingly, we observed that significantly more LSP than ESP bound to the sand fly midgut epithelium, for both Le. infantum LAB ( Fig. 2D and E) and BH46 ( Fig. 2F and G) strains, harvested on both day 4 and day 5.
In order to assess the abundance of LPG on the surface of BH46 WT parasites, we stained LSP and ESP Ficoll-purified parasites with a LPG backbone-specific monoclonal  antibody (CA7AE). Using flow cytometry, we found that LSPs exhibited 2-fold-higher and 20-fold-higher levels of antibody binding than ESP samples for 4-day-old and 5-day-old cultures, respectively ( Fig. 2H and I), indicative of the increased abundance of LPG in the former. The increase in fluorescence intensity as the parasites aged ( Fig. 2H and I) correlated with a greater number of parasites binding the sand fly midgut from 4-day-old and 5-day-old cultures ( Fig. 2F and G). In order to further evaluate such observations, we measured LPG abundance in these parasites by confocal microscopy ( Fig. 3A to H). The LPG staining that we observed was more intense in LSP ( Fig. 3A to D) than in ESP ( Fig. 3E to H). In comparison, 3-day-old parasite cultures stained poorly with the CA7AE antibody ( Fig. 3I to L).
Le. infantum binding to the midgut epithelium of Lu. longipalpis is LPG dependent. In order to assess whether or not the major surface glycoconjugate of Le. infantum (LPG) was the parasite ligand attaching to the midgut epithelium of Lu. longipalpis, we carried out midgut binding assays with the Le. infantum BH46 wild-type strain (BH46 WT) as well as with both the LPG-deficient (Δlpg1) and add-back (Δlpg1 ϩ LPG1) lines. Similarly to what was observed for unfed midguts (Fig. S1), aging parasites from the BH46 WT line bound to the epithelium of midguts dissected 5 days after blood feeding with increasing efficiency, whereas parasite binding was limited in unopened midguts (Fig. 4). For both fed and unfed Lu. longipalpis midguts, binding of the BH46 Δlpg1 ϩ LPG1 line was intermediate and the level of binding was significantly higher than that seen with the BH46 Δlpg1 mutant, which failed to bind, exhibiting less than 1,500 parasite/midgut regardless of the age of the harvested parasites or the feeding status of the midguts ( Fig. 4; see also Fig. S1). Similar results were obtained with the Le. infantum BA262 Δlpg1 mutant, which also failed to bind to unfed midguts (Fig. S3). Comparatively, the BA262 WT bound with increased efficiency as the parasites aged ( Fig. S3 and S4). Despite the low level of midgut binding of the BA262 Δlpg1 ϩ LPG1 line, such parasites bound to midguts in significantly greater proportions than the BA262 Δlpg1 mutant (Fig. S3). Le. infantum BH46 ⌬lpg1 mutants grow in the midguts of Lu. longipalpis. As the presence of Le. infantum LPG is sufficient to mediate midgut epithelium attachment in vitro but binding was substantially increased in LSP compared to ESP, we tested whether this ligand is necessary for in vivo parasite development in the midgut of the sand fly Lu. longipalpis. We infected Lu. longipalpis sand flies with the BH46 WT, BH46 Δlpg1, and BH46 Δlpg1 ϩ LPG1 lines and followed both parasite load and infection prevalence at five time points after infection. When seeded at 5 million parasites/ml, the BH46 Δlpg1 mutant exhibited a significantly lower parasite load and a lower infection prevalence than the BH46 WT strain or the BH46 Δlpg1 ϩ LPG1 strain on day 3 postinfection ( Fig. 5A and B), but it recovered on subsequent days, exhibiting a parasite load and infection prevalence similar to those seen with either the BH46 WT or BH46 Δlpg1 ϩ LPG1 line (Fig. 5C to J). When the infection was started with 2 million parasites/ ml, the BH46 Δlpg1 mutant displayed a lower, and yet not statistically significant, parasite load on day 3 (Fig. S5A) but higher loads than either the WT strain or the Δlpg1 ϩ LPG1 line from day 6 onward (Fig. S5B to E).
Le. infantum BA262 ⌬lpg1 mutants fail to grow in the midguts of Lu. longipalpis. As the LPG side chain decorations are polymorphic among Le. infantum strains (34), we also assessed the ability of the BA262 WT strain, along with the BA262 Δlpg1 and BA262 Δlpg1 ϩ LPG1 lines (35), to develop in Lu. longipalpis (Fig. 6). Differing from the BH46 strain, which exhibits side chains with 1 to 3 ␤-glucose residues, the LPG of BA262 is devoid of side chains, like most of the Le. infantum strains (34). Seeded at 5 million parasites/ml, the BA262 Δlpg1 mutant displayed lower parasite loads on days 2 and 3 (median 0 to 500 parasites/midgut) than the WT (median 3,000 to 5,000 parasites/ midgut) and the Δlpg1 ϩ LPG1 line (median 1,600 to 13,750 parasites/midgut) lines and a lower infection prevalence before the blood was passed (Fig. 6A to D). After the blood meal was passed, the BA262 Δlpg1 mutant was lost in the majority of sand flies, persisting in only a few specimens that displayed a reduced number of parasites and a decreased infection prevalence (Fig. 6E to L). The BA262 WT and Δlpg1 ϩ LPG1 lines, on the other hand, developed well in the midguts, reaching medians of 27,000 and 35,000 parasites per midgut, respectively, and an 80% to 85% infection prevalence on day 15 postinfection (Fig. 6K and L).
Le. infantum BH46 and BA262 ⌬lpg1 mutants display different susceptibilities to components of the blood meal. To investigate the differences in sand fly survival rates of the BH46 and BA262 Δlpg1 mutants, we tested if resistance to by-products of blood meal digestion could be a factor that explains these differences. For this, we incubated in vitro parasites in the exponential phase of growth with extracts of midguts, collected at 24 and 48 h after blood feeding, for 4 h at 26°C, and compared them to the WT and Δlpg1 ϩ LPG1 lines (Fig. 7). BH46 Δlpg1 mutants were not affected by the components of the digested sand fly blood meal (Fig. 7A). In contrast, the BA262 Δlpg1 mutants were severely affected by incubation with midguts collected 24 h and 48 h post-blood feeding (Fig. 7B). The WT and Δlpg1 ϩ LPG1 lines of both strains were not affected by incubation with midgut extracts (Fig. 7).

DISCUSSION
Many studies have demonstrated that binding of Le. major to the midguts of the restrictive vectors is mediated by LPG (6,11,12,(24)(25)(26)36). Incubation of P. papatasi midguts with purified PGs from procyclic Le. major parasites prevented the binding of procyclic Le. major parasites (6), and Le. major Δlpg1 parasites cannot develop in P. papatasi or P. duboscqi (25,26) sand flies. On the other hand, the Le. major Δlpg1 mutant binds to midguts and develops well in permissive vectors, such as Lu. longipalpis (24,25), as well as Phlebotomus arabicus (24), P. argentipes (26), and P. perniciosus (26). Based on such findings, an LPG-independent mechanism for midgut binding was proposed for the permissive vectors (28). Differences in midgut glycosylation between restrictive and permissive vectors were observed that could possibly account for nonspecific binding of parasites in the latter. It was hypothesized that a lectin on the Leishmania surface might bind to the O-linked glycans of the midgut microvilli of permissive vectors, thus allowing parasite binding (26,28,30). Nonetheless, such studies were carried out using Le. major and unnatural permissive vectors (24)(25)(26)28); thus, the LPG-independent mechanism might be restricted to Le. major development in unnatural permissive vectors. Therefore, we decided to revisit the role of LPG in parasite development in natural permissive vectors, focusing on two Le. infantum strains bearing intraspecies polymorphisms in their LPGs (25). The LPG of one strain, BA262, is devoid of side chains (type I) whereas that of the BH46 strain has ␤-glucose sugars branching off the repeat unit backbone (type III [20]). Here, we demonstrate that Le. infantum (BH46) wild-type and Δlpg1 ϩ LPG1 lines bound to both unfed and day-5post-feeding midgut epithelia of Lu. longipalpis in vitro. In contrast, the LPG-deficient Δlpg1 mutant failed to bind to Lu. longipalpis unfed and fed midguts. Similar results were reported for Δlpg1 mutants of Le. mexicana (MPRO/BR/72/M1845) and Le. infantum (MHOM/BR/76/M4192), which exhibited poor binding to midguts of Lu. longipalpis and P. perniciosus (27), respectively. Altogether, these data indicate that LPG mediates Leishmania binding to midguts of natural permissive vectors and suggest that the previously described LPG-independent midgut binding mechanism may be limited to Le. major binding to midguts of permissive vectors.
Similarly to the conclusions regarding Le. major binding to the midgut of its natural restrictive vector, previous reports claimed that Le. infantum and Le. donovani binding to the midgut of a naturally permissive vector was also restricted to early-stage parasites (20,21). Nonetheless, the use of peanut agglutinin (PNA) to sort procyclic and metacyclic Le. infantum (21) and Le. donovani (20) parasites may have confounded interpretation of results. PNA has been used to purify Le. major metacyclics as the LPGs of such parasite stages are decorated with arabinose, which replaces numerous ␤-galactose sugars on side chains of early-stage parasites (37). In contrast, Le. infantum (21) and Le. donovani (20) LPGs display only a single ␤-galactose residue in the cap. Whereas it is not clear whether or not the same residue is absent in the Le. infantum metacyclic stage LPG cap (21), the Le. donovani cap bears ␤-galactose or mannose residues at the same proportions as procylic parasites (20), precluding accurate purification of procyclics and metacyclics by PNA. Additionally, experiments showing the developmental differences of Le. infantum LPG (21) and Le. donovani LPG (20) found that PNA-purified metacyclic LPG was representative of only 10% of the parasites in stationary-phase cultures (20,21). It is possible that PNA-purified metacyclic LPGs are representative of a small metacyclic subpopulation that does not bind to PNA, as the whole metacyclic population usually accounts for about 80% of late-phase cultures. Importantly, this small population of nonbinding/free-swimming Leishmania parasites likely comprises the metacyclics that are transmitted, as the parasites inoculated by sand flies represented only 0.02% to 14% of the population in a mature Leishmania infection in all sand fly species investigated to date (38)(39)(40). The observed increase in the intensity of LPG labeling as Le. infantum aged in culture, which correlated positively with the number of parasites bound to midguts, supports this hypothesis. Knowing whether stronger LPG labeling is related to an increase in the number and/or length of LPG molecules during metacyclogenesis will shed light on the mechanism of binding of the Le. infantum LPG to the receptor on the midgut microvilli.
Electron microscopy images of Le. infantum developing in Lu. longipalpis show that the parasites, previously described as long and short nectomonads (41) and now termed nectomonads and leptomonads (42), respectively, were observed attached to the posterior and anterior midguts throughout the parasite life cycle in the sand fly (41). In our experiments, early-stage parasites were not able to attach to midguts in vitro, which might be explained by a lack of bona fide nectomonad parasites in cultures expressing LPG on their surface. On the other hand, we observed strong Lu. longipalpis midgut binding by late-stage Le. infantum parasites, enriched in leptomonads and metacyclics, confirming early observations by electron microscopy that Le. infantum late-stage parasites also bind to the Lu. longipalpis midgut epithelium (41). Whether midgut binding by late-stage parasites is necessary for parasite migration to the anterior midgut (42), for genetic exchange (43,44), and/or for preventing the pushing of metacyclic parasites to posterior midgut upon sequential blood meals (45) has yet to be determined.
Our proposed model of LPG-mediated Le. infantum attachment to the midgut of a permissive vector complements and expands upon previous findings that demonstrated that changes in LPG during metacyclogenesis mediated parasite detachment from the midgut epithelium to be transmitted. Here, we propose that metacyclics encompass two subpopulations: one that binds to the midgut epithelium, as was observed in this study and also earlier (41), and a free-swimming one that is transmitted by a sand fly bite (20,21). The existence of such subpopulations is also supported by polymorphisms in the LPG cap of metacyclic Le. infantum (21) and Le. donovani (20) parasites. It was previously shown that lack of LPG in Le. infantum seems to affect the ability of this parasite to develop in the midgut of the permissive natural vector P. perniciosus (27). A closer look at the results, nonetheless, suggests that the Le. infantum Δlpg1 infection showed some recovery on day 8 postinfection, after the digested blood was passed, compared to day 4, a time point when the blood meal is still being digested  LPG1) lines. Low, 500 to 5,000 parasites/midgut; Moderate, 5,000 to 10,000 parasites/midgut; Heavy, Ͼ10,000 parasites/midgut. n ϭ 2. *, statistically significant at P Ͻ 0.05. , and add-back (Δlpg1 ϩ LPG1) lines were exposed in vitro to either a PBS control (CTR) or extracts of engorged midguts dissected at 24 h or 48 h post-blood meal. (B) BA262 WT (wild-type), Ϫ/Ϫ (Δlpg1), and add-back (Δlpg1 ϩ LPG1) parasites were exposed in vitro to either a PBS control (CTR) or extracts of engorged midguts dissected at 24 h or 48 h post-blood meal. n ϭ 2. *, statistically significant at P Ͻ 0.05. (27), as the prevalence of sand flies infected with heavy (Ͼ10,000 parasites/midgut) and moderate (5,000 to 10,000 parasites/midgut) populations increased. In order to further investigate the importance of LPG for Le. infantum development in Lu. longipalpis, we assessed survival and development of the Le. infantum BH46 and BA262 WT, Δlpg1, and Δlpg1 ϩ LPG1 lines in the naturally permissive vector Lu. longipalpis for 15 days. In accordance with a previous report (27), the Le. infantum BH46 Δlpg1 mutant struggled to survive at the time of blood digestion on day 3, and yet it thrived after the blood was passed, reaching parasite loads similar to those seen with the WT and add-back lines at later time points. In contrast, the BA262 Δlpg1 mutant not only struggled to survive up to day 3 but succumbed afterward, revealing important interstrain differences in Le. infantum in vivo development. These data indicate that LPG-mediated epithelium binding is not a determinant of parasite survival in the midgut for this parasite-vector pair. However, the presence of LPG does confer protection to some Le. infantum strains such as BA262.
Toxicity of blood components can affect the survival of Leishmania parasites in the sand fly midgut during digestion (9). In the present study, incubation of Le. infantum strains BH46 and BA262, harvested from 3-day-old cultures with extracts of engorged midguts, produced different outcomes. Whereas the BH46 and BA262 WT and Δlpg1 ϩ LPG1 lines, as well as the BH46 Δlpg1 mutant, survived exposure to the toxic components of the digested blood, the BA262 Δlpg1 mutant was highly affected by midgut extracts. These results correlated well with the survival of the BH46 Δlpg1 mutant and the lack of further development of the BA262 lpg1 KO line in sand flies. Together, these results indicate that neither LPG type I nor LPG type III is needed to prevent Le. infantum from being eliminated along with the digested blood bolus in a permissive vector, but it is likely important for early survival during the digestive period for some but not all parasite strains. The survival of the BH46 Δlpg1 mutant after exposure to engorged midgut extracts is intriguing; the nature of the glycosylation of the galactose-mannose backbone of other surface glycoconjugates may correlate with protection for this strain and needs to be further explored. This possibility is supported by early findings showing that Le. major lacking LPG survived within the bloodmeal (24,26) whereas Le. major lacking LPG and proteophosphoglycans succumbed within the blood bolus (25,26).
As stated previously by Sacks and Kamhawi (16), for permissive species such as Lu. longipalpis, persistence of parasites in the midgut after blood is passed is possibly due to factors other than attachment, such as a slower peristaltic movement. These observations are supported by our findings which indicate that LPG docking of parasites does occur but appears not to define vector competence for Le. infantum in Lu. longipalpis. Rather, vector competence in permissive vectors seems to be more complex, affected by strain-specific differences and involving multiple stages of parasite development in the midgut.