Introduction
From epidemiology and ecology to economy, host–parasite interactions impact a wide variety of fields, making the study of parasites a priority. The fitness of a parasite crucially depends on its transmission success. Understanding how parasites transmit is thus essential for studying the ecology and evolution of host–parasite interactions (Antonovics et al. Reference Antonovics, Wilson, Forbes, Hauffe, Kallio, Leggett, Longdon, Okamura, Sait and Webster2017).
Parasites usually infect their hosts via two (non-exclusive) transmission routes: vertically, from an infected parent to its offspring, and horizontally, from an infected individual to an unrelated, uninfected individual. Each transmission route or mode has its own advantages. For example, vertical transmission can help the parasite survive in low-density host populations, while horizontal transmission with free-living stages can be favourable in unstable environments (Ironside et al. Reference Ironside, Smith, Hatcher and Dunn2011; Ebert, Reference Ebert2013). Transmission modes are also associated with different degrees of virulence. The success of a vertically transmitted parasite is linked to the survival and reproductive success of its host and thus strict vertical transmission is associated with lower virulence (Fine, Reference Fine1975; Lipsitch et al. Reference Lipsitch, Siller and Nowak1996; Haine, Reference Haine2008). Horizontally transmitted parasites have no such constraints and their infections are usually associated with higher virulence in comparison with vertically transmitted parasites (Frank, Reference Frank1996; Lipsitch and Moxon, Reference Lipsitch and Moxon1997; Restif and Kaltz, Reference Restif and Kaltz2006).
In nature, however, many parasites have the ability to infect their host both horizontally and vertically (mixed-mode transmission or MMT), and use both infection routes to some degree. The combination of both vertical and horizontal transmission allows MMT parasites to thrive in a wider variety of environmental conditions, and gives us the opportunity to study the dynamics of both transmission routes within one parasite. While MMT parasites have the benefit of both transmission routes, they face several constraints. In many cases, the conditions which favour horizontal transmission are opposite from those which favour vertical transmission (Koella et al. Reference Koella, Agnew and Michalakis1998; Turner et al. Reference Turner, Cooper and Lenski1998; Stewart et al. Reference Stewart, Logsdon and Kelley2005). Investing in one transmission route can come at the expense of the other, as vertical and horizontal transmissions may be a trade-off against each other (Messenger et al. Reference Messenger, Molineux and Bull1999). However, if a MMT parasite is able to accurately assess whether the environmental conditions favour horizontal or vertical transmission, and choose its transmission route accordingly, the parasite can circumvent these constraints (Kaltz and Koella, Reference Kaltz and Koella2003; Tintjer et al. Reference Tintjer, Leuchtmann and Clay2008). For example, in viruses it is known that certain bacteriophages reproduce via two mutually exclusive cycles known as lysis and lysogeny. These bacteriophages can switch from lysogenic to lytic life cycle in response to cues regarding the state of their host cell (Stewart and Levin, Reference Stewart and Levin1984; Refardt and Rainey, Reference Refardt and Rainey2010). In the bacterial parasite Holospora undulata, environmental cues also play a role in the parasite's investment in one of the two transmission routes, and both transmission forms can be jointly found in the same host (Kaltz and Koella, Reference Kaltz and Koella2003).
MMT is also very common in microsporidia, single-celled obligate intracellular parasites, which are closely related to fungi. Microsporidian parasites utilize different spore types (or morphologies) for vertical and horizontal infection (Dunn et al. Reference Dunn, Terry and Smith2001). For example, in Edhazardia aedis which alternates between horizontal and vertical transmissions, the parasite produces binucleate spores for vertical transmission, and uninucleate spores for horizontal transmission (Agnew and Koella, Reference Agnew and Koella1999). In the case of the microsporidium Ambylospora connecticus, which has an indirect life cycle involving a mosquito and a copepod host, three different spore types are used in vertical and horizontal transmissions of the parasite (Dunn et al. Reference Dunn, Terry and Smith2001).
The microsporidium Hamiltosporidium tvaerminnensis is an obligate parasite of Daphnia magna, with infection prevalence reaching more than 40% in some populations (Carius et al. Reference Carius, Little and Ebert2001). Hamiltosporidium tvaerminnensis can infect its host vertically through transovarial transmission or horizontally through decomposing hosts that release spores into the environment and subsequent ingestion of these spores by other host individuals. Hamiltosporidium tvaerminnensis is known to possess three distinguishable spore morphologies that can be seen under a phase-contrast microscope. The first morphology is more oval, has a thick spore wall and is more refractive when viewed under the microscope, which is suggestive of a thick spore wall. The second morphology is more pear-like and has reduced refractivity, thereby suggestive of a thinner spore wall. The third morphology is elongated, either straight or with a slight curve, has intermediate refractivity, but is much less commonly seen than the other two (Vizoso and Ebert, Reference Vizoso and Ebert2004). These morphologies are similar to those typically found in many microsporidia (Dunn and Smith, Reference Dunn and Smith2001). The prevalence and quantity of the different spore morphologies is known to vary throughout the infection process. At the beginning of infection, the majority of the spores are of the pear-shaped type, but as infection progresses, their frequency decreases and the frequency of oval-shaped spores increases (Vizoso and Ebert, Reference Vizoso and Ebert2004).
In light of these findings, we investigated the role of the two common but distinct spore morphologies (i.e. oval and pear) of H. tvaerminnensis in horizontally infected D. magna. Using two D. magna clones and two H. tvaerminnensis isolates, we individually exposed Daphnia to solutions consisting of either mostly oval- or pear-shaped spores, to determine which spore type drives horizontal transmission. Based on the observations of Vizoso and Ebert (Reference Vizoso and Ebert2004), and given that horizontal transmission occurs only upon host death, we expected the oval-shaped spores to be responsible for horizontal transmission.
Host–parasite system
Daphnia magna Straus is a freshwater crustacean that reproduces via cyclical parthenogenesis. It is commonly found in eutrophic ponds, where it often hosts a wide variety of parasites (Green, Reference Green1974; Decaestecker et al. Reference Decaestecker, Declerck, De Meester and Ebert2005; Goren and Ben-Ami, Reference Goren and Ben-Ami2013). Hamiltosporidium tvaerminnensis is an obligate intracellular microsporidium of D. magna. It can infect its host both horizontally through environmental spores that allow direct transmission among Daphnia, and vertically through parthenogenetic and sexually produced eggs (Vizoso et al. Reference Vizoso, Lass and Ebert2005; Haag et al. Reference Haag, Larsson, Refardt and Ebert2011). We used two laboratory-grown D. magna clones (BM1 and G), each isolated from a single female sampled from natural populations located in the Negev Mountains and central coastal plain in Israel, respectively (Goren and Ben-Ami, Reference Goren and Ben-Ami2013). We further used two H. tvaerminnensis isolates (NZ2 and G3), which were isolated from infected D. magna from the southern and central coastal plains in Israel, respectively (Goren and Ben-Ami, Reference Goren and Ben-Ami2013). Each parasite isolate was maintained separately in the laboratory to avoid contamination.
Materials and methods
Experimental design
We individually exposed females from each D. magna clone to one of the two H. tvaerminnensis isolates, using either 200 000 parasite spores containing the ‘pear solution’ or 200 000 spores containing the ‘oval solution’. This dose was chosen, because it was previously found to maximize the number of infected individuals, while preventing high early mortality rates caused by exposure to high spore doses (H. Urca and F. Ben-Ami, unpublished data). The pear solution was prepared by diluting a solution containing mostly pear-shaped spores until reaching a ratio of one oval-shaped spore for every six pear-shaped spores (ratio of 1:6, i.e. mostly pear spores). The oval solution was prepared by diluting a solution containing mostly oval-shaped spores until reaching a ratio of one pear-shaped spore for every 18 oval-shaped spores (ratio of 18:1, i.e. mostly oval spores). The spore ratio was determined by checking the solutions under a light microscope. Solutions were repeatedly diluted, then examined under the microscope, until the maximum purity possible for each spore solution was achieved. In total, our experiment consisted of 600 Daphnia individuals (two Daphnia clones × two Hamiltosporidium isolates × two spore solutions × 60 individuals per infection treatment + two control groups × 60 individuals).
Prior to the experiment, D. magna mothers from each Daphnia clone (separate maternal lines) were kept in 400 mL jars with 10–12 individuals in each jar. Newborn offspring were separated from their mothers within 24 h of their release, placed in 500 mL jars and fed 1 × 106 algae cells per day per Daphnia. On day 4, the offspring were placed in individual 100 mL jars filled with 30 mL of medium. On day 5, all Daphnia individuals, except those in the control groups, were administered the relevant infection treatment (either the pear or oval solution of one of the parasite isolates). All exposed individuals were infection free from the beginning. Except on the first day of exposure, all animals were fed with 2 × 106 algae cells per day per Daphnia. On day 10 (5 days post-exposure), we placed all individuals in jars filled with 100 mL of fresh medium, and from that point onwards, fresh medium was replaced on a weekly basis. To adjust the food demands of the growing Daphnia, on days 9, 15, 18 and 22, we increased the daily food level for all animals to 3 × 106, 5 × 106, 6 × 106 and 7 × 106 algae cells per day, respectively. The temperature was kept at 21 ± 0.5° C and the light:dark cycle was 16:8 h. All treatments were randomly distributed on the shelves and rearranged often to prevent position effects.
Throughout the experiment, we separated offspring from their mothers and monitored Daphnia survival (every three days). Dead animals were frozen at −20° C in 0.1 mL of medium for spore counting using a haemocytometer (Thoma ruling) under a phase-contrast microscope (400×). Thirty minutes prior to spore counting, Daphnia cadavers were removed from refrigeration and crushed in medium to allow the release of spores. Pear and oval spores (Vizoso and Ebert, Reference Vizoso and Ebert2004) were counted separately. Animals that had died before day 15 were not included in the analysis, because infected Daphnia do not show quantifiable signs of infection before that (i.e. spores are not visible in dissected individuals under the microscope; H. Urca & F. Ben-Ami, unpublished data).
Statistical analysis
All statistical tests were carried out using R, version 3.2.3 (R Core Team, www.R-project.org). Infectivity was analysed using binary logistic regression (proc glm, family = binomial), with host clone, parasite isolate and spore type coded as indicator variables. Cox regression (proc coxph) was used in a similar way to compare parasite-induced host mortality (virulence) among treatments, with time-from-exposure-to-host-death as the dependent variable. The effects of host clone, parasite isolate, spore type and their interactions on parasite spore production were examined using a general linear model (proc glm, family = quasi). In all tests, we initially constructed a full factorial model that included all main effects and their interactions. We then one-by-one removed non-significant interactions (P > 0.1), thereby deriving a reduced model.
Results
General effects
Twenty-two (3.6%) out of the 600 Daphnia died during the first 2 weeks of the experiment for unknown reasons. This unexplained mortality was more pronounced in the Daphnia clone BM1, but was otherwise unrelated to the infection treatments. Daphnia individuals that did not become infected in the infection treatment groups were excluded from the analyses of host survival and parasite spore production. None of the control Daphnia became infected. Host longevity was significantly higher in the control group of each Daphnia clone in comparison with the respective infection treatment groups (BM1: z = 5.74, d.f. = 1, P < 0.001; G: z = 3.87, d.f. = 1, P < 0.001).
Host susceptibility and parasite infectivity
The probability of becoming infected was unaffected by host clone, parasite clone and the spore type to which Daphnia were exposed (Table 1, Fig. 1). However, in three out of the four host–parasite combinations (BM1/NZ2, G/G3 and G/NZ2), the proportion of infected hosts was higher in Daphnia exposed to oval spores, whereas in one combination (BM1/G3) the reverse was true, i.e. the proportion of infected hosts was slightly higher in Daphnia exposed to pear spores (spore type × host clone interaction: z = 3.41, d.f. = 1, P < 0.001). Hosts infected with oval spores of parasite isolate G3 had higher infection rates when compared with hosts infected with pear spores of the same isolate, whereas there was no difference in infection rates between hosts infected with oval and pear spores of isolate NZ2 (spore type × parasite isolate interaction: z = 2.11, d.f. = 1, P = 0.034).
Fig. 1. Proportion of Daphnia magna infected by Hamiltosporidium tvaerminnensis for each combination of spore type, host clone and parasite isolate.
Table 1. GLM analysis of the effects of spore type, host clone, parasite isolate and their interactions on infection status
d.f., degrees of freedom.
Bold typeface indicates significant effects.
Parasite-induced host mortality (virulence)
Parasite isolate, but not spore type and host clone, affected parasite-induced host mortality (i.e. virulence; Table 2, Fig. 2). Moreover, Daphnia exposed to parasite isolate NZ2 lived longer when exposed to pear spores in comparison to oval spores, while the survival of Daphnia exposed to isolate G3 did not differ between spore types (spore type × parasite isolate interaction: z = 7.95, d.f. = 1, P < 0.001). In both host clones, Daphnia died at a faster rate when infected with parasite isolate NZ2 in comparison with isolate G3, but this effect was stronger in host clone BM1 (host clone × parasite isolate interaction: z = 5.27, d.f. = 1, P < 0.001). Three-way interactions also influenced host mortality, i.e. Daphnia from clone G infected with parasite isolate G3 lived longer when exposed to pear compared with oval spores, whereas the difference in survival between the two spore treatments was much less pronounced when Daphnia from clone G were infected with isolate NZ2. The survival of Daphnia from clone BM1 did not noticeably differ by spore type or parasite isolate.
Fig. 2. Time-to-host-death-since-exposure (days) for infected Daphnia in the different host clone–parasite isolate combinations: (a) BM1 × G3, (b) BM1 × NZ2, (c) G × G3, (d) G × NZ2.
Table 2. Cox regression analysis of the effects of spore type, host clone, parasite isolate and their interactions on time-to-host-death-since-exposure
d.f., degrees of freedom.
Bold typeface indicates significant effects. This analysis was done without the controls.
Parasite spore production (parasite fitness)
Spore type, host clone and parasite isolate all affected the production of parasite spores (Table 3, Fig. 3). In three out of the four host–parasite combinations (BM1/G3, G/G3 and G/NZ2), Daphnia infected with oval spores produced more spores than Daphnia infected with pear spores, while in one combination (BM1/NZ2), the opposite occurred. On average, Daphnia from clone BM1 produced more spores than Daphnia from clone G, and Daphnia infected with parasite isolate G3 had a higher spore load than Daphnia infected with isolate NZ2 (Fig. 3). However, the existence of highly significant two- and three-way interactions (spore type × parasite isolate, host clone × parasite isolate and spore type × host clone × parasite isolate) point to the existence of strong genotype-by-genotype (G × G) interactions (Table 3).
Fig. 3. Mean parasite spore production by D aphnia magna infected with Hamiltosporidium tvaerminnensis for each combination of spore type, host clone and parasite isolate. Error bars are standard errors.
Table 3. GLM analysis of the effects of spore type, host clone, parasite isolate and their interactions on parasite spore production
d.f., degrees of freedom.
Bold typeface indicates significant effects.
Discussion
In the present study, we examined the role of two spore morphologies of H. tvaerminnensis during horizontal transmission of the parasite. We found that spore type per se did not affect infection rates and host mortality. Instead, host clone and parasite isolate interacted with spore type in shaping infection outcome and determining host mortality. These results are not consistent with the predictions that different spore types carry different roles in the transmission of H. tvaerminnensis (Vizoso and Ebert, Reference Vizoso and Ebert2004; Vizoso et al. Reference Vizoso, Lass and Ebert2005). We further found that parasite spore production was higher in Daphnia infected with oval spores, thereby suggesting that host death was caused by the secondary effects of infection, rather than by parasite within-host proliferation. Nevertheless, parasite spore production was also influenced by complex interactions among spore type, host clone and parasite isolate.
Our finding that spore type alone does not affect infection rates and host mortality, but rather the existence of complex interactions among spore type, host clone and parasite isolate, may suggest that host and parasite genetics have a contributing role. Although in three out of the four host–parasite combinations, hosts exposed to oval spore types had higher infection rates, which support the hypothesis that the two spore types are functionally different, the relatively high infection rates in both the oval and pear treatments, and the lack of correlation between infection rates and host survival, do not support it. Moreover, even though spore type alone affected parasite spore production, we still found evidence of complex interactions among spore type, host clone and parasite isolate. Future studies of this system should focus on disentangling these seemingly strong G × G interactions.
Several studies have shown that parasite spore production and host survival are negatively correlated with parasite dose (Agnew and Koella, Reference Agnew and Koella1999; Ebert et al. Reference Ebert, Zschokke-Rohringer and Carius2000; Regoes et al. Reference Regoes, Ebert and Bonhoeffer2002). The lack of correlation between the mean number of parasite spores produced per host and either infection rates or host mortality makes it difficult to determine whether higher spore load is indicative of exposure to a higher dose of infective spores. If oval spores are responsible for horizontal transmission, we would expect higher spore loads in hosts exposed to the oval spore solution, yet our results show the opposite effect in one of the four host–parasite combinations. Our results, however, should be interpreted with caution, because the lack of correlation between spore type and host survival makes it difficult to determine the exact relationship between spore type and parasite spore load.
Our technical inability to prepare pure pear and oval spore solutions meant that the pear spore solution contained a relatively small percentage of oval spores, and vice versa. It is well known that infection rates are greatly influenced by the quantity/density of the infectious agent the host is exposed to (Ben-Ami et al. Reference Ben-Ami, Regoes and Ebert2008b; Ben-Ami et al. Reference Ben-Ami, Ebert and Regoes2010; Ben-Ami and Routtu, Reference Ben-Ami and Routtu2013), as well as specific interactions between host and parasite genotypes (Ben-Ami et al. Reference Ben-Ami, Mouton and Ebert2008a; Duneau et al. Reference Duneau, Luijckx, Ben-Ami, Laforsch and Ebert2011; Luijckx et al. Reference Luijckx, Ben-Ami, Mouton, Du Pasquier and Ebert2011; Hall and Ebert, Reference Hall and Ebert2012). The two Daphnia clones used in this study are highly susceptible to the two parasite isolates we used (H. Urca and F. Ben-Ami, unpublished data). It is possible that the amount of oval spores present in the pear spore solutions was sufficient to infect our Daphnia hosts, hence the lack of significant differences in infection rates and host survival between the two spore types. This would not explain, however, the complex interactions we found among spore type, host clone and parasite isolate.
An alternative explanation is that the two spore morphologies do not have different roles in the transmission of H. tvaerminnensis, and that their visible morphological differences are not related to their transmission function. For example, spore morphology may change to facilitate autoinfection (within-host cell-to-cell infections) and thus have little to do with between-host transmission (Keeling and Fast, Reference Keeling and Fast2002; Williams, Reference Williams2009; Vávra and Lukeš, Reference Vávra and Lukeš2013). When examining H. tvaerminnensis spores under a light microscope, a significant number of spores could not be easily categorized and seem to be in an intermediate state between oval and pear morphology. In other words, these intermediate spores had some features that were consistent with one morphology (like an oval shape), and other features that were compatible with the second morphology (like reduced refractivity). Although slight variation and abnormalities in spore morphologies are to be expected, the large quantity of spores possessing intermediate features suggests that these spores might be transitioning from one morphology to another.
The coexistence of oval and pear spores in the host also casts doubt on the different transmission roles hypothesis. In MMT microsporidia, the production of vertical and horizontal spores occurs during spatially distinct sporulation events, or might even include an intermediate host to achieve horizontal transmission (Agnew and Koella, Reference Agnew and Koella1999; Dunn and Smith, Reference Dunn and Smith2001). However, despite being investigated for several decades, there is no evidence that H. tvaerminnensis (previously Octosporea bayeri) uses an intermediate host for horizontal transmission. A recent study on the ultrastructure of H. tvaerminnensis found that almost all spores contained an identical number of coils (i.e. number of times the polar tube is coiled inside the spores). It was also observed that the spores exhibited an identical ultrastructure (Haag et al. Reference Haag, Larsson, Refardt and Ebert2011). Studies describing microsporidia with coexisting spore types emphasize the difference in coil numbers within the two spores (Iwano and Ishihara, Reference Iwano and Ishihara1991; Iwano and Kurtti, Reference Iwano and Kurtti1995). It is also known that in microsporidia spores responsible for horizontal and vertical transmissions have different structures, with a noticeable difference in the thickness of the spore wall. Further studies, which implement techniques allowing a more complete segregation of the two spore types, are therefore required to elucidate the role of the different spore types of H. tvaerminnensis.
To the best of our knowledge, this is the first attempt to test experimentally the role of spore morphology in horizontal transmission of H. tvaerminnensis. Our results do not support this hypothesis, namely that either oval- or pear-shaped spores have a distinct role in horizontal transmission of the parasite. There is no dispute that H. tvaerminnensis can transmit both vertically and horizontally. Therefore, further laboratory and molecular studies are required to understand better the infection dynamics of H. tvaerminnensis and the role of different spore morphologies in the parasite's transmission. Notwithstanding, the existence of strong G × G interactions in this system make it a suitable candidate for studying host–parasite coevolution and negative frequency-dependent selection.
Acknowledgements
We thank Mor Zarkor and Nicolas Zemelman for help in the laboratory. We are grateful to four anonymous reviewers for valuable comments on the manuscript.
Financial support
This research was supported by grant #938/16 from the Israel Science Foundation (ISF) to FBA.
