We studied the tsetse-trypanosome system to evaluate the molecular and physiological aspects of host-parasite interactions and the consequences of parasite infections on host fecundity. Two trypanosome strains that differentially activate host immunity were employed to assess the cost of the ability to clear parasite infections (resistance) and the cost of midgut parasite infections in susceptible flies. In contrast to the dogma that expression of parasite resistance traits incurs a fitness cost to the insect host, we observed no significant fecundity cost associated with trypanosome resistance in laboratory reared tsetse. However, activation of tsetse immune responses by infecting flies with immunogenic trypanosomes reduced host reproductive output while infections with non-immunogenic trypanosomes did not.
It is unusual that we did not detect a fecundity cost associated with parasite resistance in tsetse. This observation, that differs from what is seen in mosquitoes, may reflect the viviparous reproductive biology of tsetse where investments in reproductive output begin significantly later and continue throughout the lifetime of adult female flies. Female tsetse develop a single oocyte per gonotrophic cycle. Following ovulation and fertilization, the embryo develops within the uterus and hatches into a first instar larva, which molts through two more instars before being deposited as a fully developed larva which quickly pupates in the soil. From this point on, there is a continuous investment in nurturing the single larva produced one at a time, approximately every ten days, through the remainder of the female lifespan. This differs dramatically from mosquitoes, which have a high reproductive output as young adults and hence may be more sensitive to perturbations early in life. We cannot rule out however that our in vitro experiments conducted under uniform environmental conditions and ample nutritional supply may have skewed any potential fecundity cost that flies may experience upon expression of parasite resistance in the wild.
Reduced host fecundity has been observed in various systems with parasitized insects (reviewed in Hurd, 2003). Our result is the first demonstration of a significant loss of fecundity in parasitized tsetse but only when infected with an obviously immunogenic trypanosome line. Loss of fecundity in tsetse may arise from the cost of prolonged immune activation since a similar fitness cost was not observed in flies infected with the non-immunogenic parasites. Our results suggest that the decreased expression of milk gland protein GmmMGP
, the most abundant protein product in the “milk” secretion of modified accessory glands, may cause the observed delayed larvagenesis process 
. We recently noted that in addition to GmmMGP
, which is also expressed in the female milk gland organ and is transported into the developing larva is down regulated by YTat1.1WT
. A similar reduction in the abundance of vitellogenin (Vg) mRNA and the titer of circulating Vg in the hemolymph has been reported in plasmodium infected mosquitoes during early oocyst development 
. Later in the infection process, changes in the ovarian follicular epithelium have also been associated with a decrease in Vg uptake by the ovary and may also result in reduced hormone ecdysone production, which is needed for the transcriptional regulation of Vg
. Our results suggest several avenues for future research. It remains to be seen whether the regulation of host gene expression could result from an adaptive strategy that has evolved in response to parasitism, or that it reflects the manipulation of the host insect by the parasites. The regulation of milk-gland protein and transferrin transcription remains to be described, but may similarly be subject to hormonal regulation, which in turn may be influenced by parasite infections.
One potential difference we identified between the trypanosome strains used to infect tsetse was in their EP and GPEET procyclin compositions. Under our culture conditions, the immunogenic YTATWT
line expressed both major forms of procyclin, while the YTATEP
line only expressed EP procyclins. Procyclin epitopes analyzed from parasites obtained from midgut infections indicated that both lines expressed the EP isoforms later in the infection process, similar to previous reports 
. It is possible that the highly phosphorylated GPEET molecules present on YTATWT
procyclic culture form parasites, upon acquisition in the blood meal, are recognized by the host's immune system early in the infection process and elicit the immune activation we observed in our study. Supporting this idea, when beads carrying different surface charges were introduced into mosquitoes, different physiological networks were induced resulting in varied vector cost outcomes 
. It is possible that as a consequence of host-parasite co-evolutionary dynamics, trypanosomes have evolved to down regulate the expression of GPEET in order to avoid inducing host resistance mechanisms. It is entirely possible however, that variations in parasite molecules other than procyclins may be responsible for the observed host immune outcome. The expression of GPEET alone in the early infection process cannot explain the sustained induction of host immune effectors in the case of infections with the immunogenic line since this line also switches to express predominantly the EP procyclins in late midgut infections. As molecular data on tsetse immunity are accumulating, it would be of interest to conduct a global expression analysis of both RNA and protein using the two parasite lines to gain a broader appreciation of the host-parasite interactions in the tsetse system 
Multiple vector immune components are likely to play a role in parasite transmission and resistance phenotype. Although our previous results have identified an important role for antimicrobial peptides in parasite establishment in tsetse, lack of AMP expression does not result in greater parasite infection prevalence for YTATEP
. It is possible that other trypanocidal effectors, such as lectins 
and antioxidant activity in the midgut milieu at the time of parasite acquisition that are necessary for resistance. Alternatively, cascades in immunity pathways could result in the synthesis of different trypanolytic immune effectors, which are uncharacterized at the present time. The similar midgut infection intensities observed with both trypanosome lines may also suggest general regulatory processes whereby homeostasis of parasite density is maintained to prevent excessive harm on host physiology. Our mixed infection experiments in vivo
however showed that the immunogenic parasites outcompeted the non-immunogenic parasites in midguts. It is plausible that tsetse immune products, in particular antimicrobial peptides, may have varying trypanolytic effect on parasites with initial differences in surface protein composition. It is then possible that infections with the immunogenic parasites, invoking host immune responses, could prevent the transmission of other strains that may be more sensitive to the host immune products.
The prevalence of tsetse infections reported with T. brucei
sspp. in the wild is low, typically in the order of 1–3% 
, suggesting the potential for only a small impact on the whole tsetse population. However, the prevalence of infections with other trypanosome species such as T. congolense
and T. vivax
, can be significantly higher, for example 28% reported in Burkina Faso 
and 24% in Tanzania 
. Furthermore infections with mixed trypanosome species are common, comprising over 30% of infections reported in natural populations 
. Our modeling results suggest that these higher infection rates could have substantial effect on the tsetse population dynamics if the presence of these parasite species were to cause similar reproductive delays. Such reproductive delays caused by immunogenic trypanosome strains could lead to a reduction in tsetse population size and consequently to less transmission of parasites to vertebrate hosts. In turn, the resulting reduced trypanosome prevalence in vertebrates would lead to a lower prevalence in tsetse and a corresponding rebound of the tsetse population until an equilibrium of trypanosome prevalence in tsetse and vertebrate hosts is reached at lower levels than would occur with an avirulent strain. The mechanisms of tsetse population regulation are not well understood 
, so the magnitude of the impact of trypanosome-induced fecundity delays on trypanosome infections in vertebrates is difficult to predict.
Assuming that both parasite lines we investigated have equal probability for maturation in the salivary glands for mammalian transmission, our co-infection dynamics would favor the spread of the immunogenic phenotype. Reducing its tsetse host's fitness is evolutionarily disadvantageous to the trypanosome, but only weakly compared to the possible benefits of increased virulence. Interestingly, YTat1.1WT
is known to be highly virulent in the mammalian host, but we do not have a virulence phenotype for YTat1.1EP
since we cannot obtain salivary gland mammalian transmissible infections with this line. Trade-offs between virulence and parasite transmissibility can maintain virulence, as can competition within a host between multiple strains of a parasite 
. Co-evolution between pathogen and host could lead to more complicated dynamics, from the emergence of mutualism to an escalating cycle of virulence and host resistance. In fact, a field study demonstrated that the resistance of Anopheline
mosquitoes to Plasmodium
parasite development varied considerably between different combinations of parasite and vector isolates, suggesting that there are specific compatibilities between the insect and parasite genotypes 
. If there is no cost of resistance against virulent trypanosomes, we would expect resistance to increase until the “virulent” trypanosome is driven extinct. In contrast, the “avirulent” trypanosome would not impose selection on its tsetse host, and would persist. If there is a cost of resistance however, there would be an equilibrium proportion of resistance genotype at which the cost of resistance is balanced by the cost and prevalence of virulent infection. In this case, if the avirulent trypanosome strain without any fecundity cost emerged and spread, genetically susceptible tsetse would be expected to completely replace resistant tsetse. Simultaneously, disease prevalence in both tsetse and mammalian hosts would rise. Future experiments will need to be conducted with flies in the field to understand the impact of natural parasite infections on host immunity, and the potential burden of host immune activation on fecundity in order to relate to disease epidemiology. Our laboratory findings coupled with our modeling studies now provide a framework to investigate the status of co-infections, host immune activation processes, fecundity outcomes, transmission dynamics and host virulence phenotypes in natural tsetse-trypanosome populations.