Parasitic gut infection in Libellula pulchella causes functional and molecular resemblance of dragonfly flight muscle to skeletal muscle of obese vertebrates

Obesity is a common metabolic disease in humans and can be induced in laboratory animals by feeding with a high-fat diet (Birse et al., 2010; Fenton and Dowling, 1953; Ingle, 1949). Interestingly, a recent meta-analysis demonstrated that during the past five decades, obesity has also become common in laboratory mammals, and feral (rural and urban) rodents and domestic pets that are not exposed to high-fat diets and lack of exercise (Klimentidis et al., 2011). This suggests that, in addition to genetic predisposition, there are other factors that can contribute to the development of metabolic disease. Infections by viral, bacterial and protozoan agents are considered such factors, and may affect host metabolic health in various ways, including endocrine modulation, enhanced expression of genes controlling adipogenic and lipogenic pathways (Dhurandhar, 2011; McAllister et al., 2009; Pasarica and Dhurandhar, 2007), or possibly, through changes to the composition of the gut microbiome and/or induction of a systemic inflammatory state (Sonnenburg et al., 2006; Steinberg, 2007; Turnbaugh et al., 2006). This potential for infectious origins of metabolic disease is interesting because it opens a path to examine their mechanistic bases in non-model, vertebrate and insect hosts faced with infection risks in nature. Given the evolutionarily conserved nature of cellular pathways controlling energy storage, catabolism, and aspects of immunity in vertebrates and insects (Sacktor 1975; Schlegel and Stainier, 2007; Arrese and Soulages, 2010; Kimbrell and Beutler, 2001; Hoffmann et al., 1999; Palmer and Jiggins, 2015; Martinelli and Reichhart, 2005; Beutler, 2004), there are likely interesting parallels between the physiological underpinnings of infection-associated metabolic impairments in these systems.

Insect species whose fitness crucially depends on flight make for potentially useful subjects to study such parallels. Because of the high energetic cost of this locomotor mode (e.g. Sacktor 1975 in Candy and Kilby, 1975; Gilmour 1965), flight motor designs that are optimized for high mechanical performance and energy catabolism, will need equally fine-tuned, homeostatic interactions with digestive, energy anabolic and nutrient mobilization physiology. In a fashion similar to how a high-performance race car differs from a mini-van (Marden et al., 1999; Mead et al., 2017), flying insects may have lower built-in safety factors that protect against disturbances of these homeostatic interactions than non-flying insects, and show more pronounced effects of infections. Studies of the effects of parasitic infections on insect flight performance traits have focused mainly on mosquitos and honeybees (e.g. Rowland and Boersma, 1988; Yee and Anderson, 1995; Wolf et al., 2014; Dosselli et al., 2016; Newman, Anderson, and Goldberg, 2016; Dussaubat et al., 2013; but see Altizer et al., 2015), but mechanistic details of infection effects are rarely studied (but see Harrison et al., 2001).

We previously reported that infection of Libellula pulchella dragonflies with protozoan intestinal parasites (Apicomplexa: Eugregarinorida, Hoplorrhynchus sp., also known as ‘gregarines’) causes a metabolic disease syndrome involving a doubling of hemolymph total carbohydrate concentration, a significant increase (26%) in mean thoracic lipid content, chronic stress/inflammation pathway activation and a reduced ability to oxidize fatty acids by flight muscles (Schilder and Marden, 2006). These symptoms are very similar to mammalian metabolic disease phenotypes such as obesity and type II diabetes (Goodpaster and Wolf, 2004; Kelley and Goodpaster, 2001; Khovidhunkit et al., 2004; Perseghin, 2005; Ritov et al., 2005). In addition to these metabolic impairments, we demonstrated that gregarine infection causes reduced mass-specific flight muscle performance in L. pulchella dragonflies (Schilder and Marden, 2006), which is similar to the decline in skeletal muscle performance generally observed in vertebrate obesity (Bollinger, 2017; Lafortuna et al., 2005; Maffiuletti et al., 2007; Seebacher et al., 2017; Tallis et al., 2017; Tallis et al., 2018; Tomlinson et al., 2016).

The mechanism(s) underlying reduction of mass-specific muscle performance during metabolic disease are poorly understood, but likely affect muscle sarcomere function at a molecular level. In vertebrates, a role for obesity-associated shifts in the composition of muscle fiber types, as a result of changes in expression of myosin heavy chain genes (Myhc), has been indicated. However, no single satisfactory explanatory principle has been uncovered, as the effects of obesity on muscle fiber type can vary from slow-to-fast shifts (i.e. MyHC type I→MyHC type II) to fast-to-slow shifts, or can be absent altogether, depending on the species and muscle examined (DeNies et al., 2014; Schönke et al., 2018; Tallis et al., 2017; Tanner et al., 2002).

Whether MyHC isoform composition has any role in infection-associated decline of muscle function in L. pulchella is unclear. Insect flight muscle is generally not believed to express multiple fiber types that differ in MyHC isoform composition (Ayme-Southgate et al., 2015; Hooper and Thuma, 2005) and contractile properties. However, two recent studies suggest that multiple Myhc mRNAs can be expressed in Manduca sexta and Locusta migratoria flight muscle tissue (Ayme-Southgate et al., 2015; Li et al., 2016). Although it is not known whether these mRNAs are translated into functional MyHC protein isoforms for M. sexta, Ziegler et al. (1994) reported the presence of two distinct MyHC protein isoforms in flight muscle of L. migratoria (and the fly Phormia terraenovae). However, if and how these different MyHC protein isoforms affect flight muscle function is unknown. Effects of transgenic insertion of different Myhc genes on muscle contractile properties have been detailed in Drosophila (reviewed in Maughan and Swank, 2013), but naturally occurring shifts in MyHC protein isoforms with functional consequences have, to our knowledge, not been reported for adult insect flight muscle.

In contrast, isoform expression of another key regulatory sarcomere protein, troponin T (Tnt), varies significantly in flight muscles of male L. pulchella obtained from wild populations and does affect flight muscle function (Fitzhugh and Marden, 1997; Marden et al., 1999; Marden et al., 2001). Similarly to insect Myhc, variation in Tnt isoform expression is generated by alternative pre-mRNA splicing. Importantly, the relative abundance of one of two highly abundant Tnt mRNA splice variants (i.e. Lp_Tnt261; see Table 1) in L. pulchella flight muscle was shown to correlate positively with maximum mass-specific flight muscle force and power output, wingbeat frequency use in free flight, and muscle fiber calcium sensitivity (Marden et al., 2001). These findings are consistent with a rich vertebrate literature describing effects of variation in Tnt protein isoforms on muscle calcium sensitivity (Greaser et al., 1988; Nassar et al., 1991; Reiser et al., 1990; Schachat et al., 1987), and the general idea that varying muscle calcium sensitivity affects acto-myosin cross-bridge recruitment, and thus the rate and amount of force development, and power output (Marden et al., 2001; Moss et al., 1995; Perry, 1998; Solaro and Rarick, 1998; Sweeney et al., 1993).

In rat load-bearing muscle (i.e. gastrocnemius), we recently demonstrated that obesity causes a mismatch between body mass and Tnt alternative pre-mRNA splicing (Black et al., 2017; Schilder et al., 2011). Thus, rat skeletal muscle Tnt expression of >300 g obese rats resembles that of much younger (<150 g), non-obese specimens. In contrast, experiments in non-obese, healthy rats, flies (Drosophila melanogaster) and moths (Spodoptera frugiperda), demonstrated that the relative abundance of Tnt mRNAs splice variants in load-bearing muscle is adjusted in proportion to body weight increase resulting from natural growth or experimental body weight manipulations (Marden et al., 2008; Schilder and Raynor, 2017; Schilder et al., 2011). Obesity therefore appears to cause an impairment of a sensing mechanism that relates body weight to skeletal muscle molecular composition. Importantly, the Tnt expression pattern in obese rats is expected to lower muscle calcium sensitivity and force output (Briggs and Schachat, 1996; Schachat et al., 1987; Schilder et al., 2011). This, in turn, is predicted to reduce muscle mass-specific performance, as demonstrated in aging humans (Coble et al., 2015) and, as mentioned earlier, in obese mammals (Tallis et al., 2017) and gregarine-infected dragonflies (Schilder and Marden, 2006). However, it is currently unknown if the metabolic syndrome associated with gregarine infection of L. pulchella affects relationships between Tnt expression, body mass and flight muscle performance in an similar fashion to that observed in rats.

More potential overlap between the obesity phenotypes in vertebrates and dragonflies was indicated by a recent study on zebrafish (Seebacher et al., 2017), which showed that mass-specific power output by locomotor muscle of obese zebrafish declined specifically at higher contraction frequencies. Territorial behavior of non-infected L. pulchella males involves the use of high-frequency wing beats (Dudley, 2002; May, 1991; Rüppell, 1989), powered by equally high flight muscle contraction frequencies. In contrast, gregarine-infected male L. pulchella are relatively inactive and avoid territorial disputes (Marden and Cobb, 2004). Interestingly, this behavior resembles that of recently eclosed, physiologically immature ‘teneral’ male L. pulchella, which do not engage in territorial disputes, but instead focus on foraging. Pertinent to the current study, teneral L. pulchella employ significantly lower wing beat frequencies, and display significantly lower flight muscle fiber calcium sensitivity compared with mature males (Marden et al., 1998). Thus, gregarine infection may prevent L. pulchella males from participating in territorial behavior by affecting their ability to operate competitively at higher contractile frequencies.

Combined, these observations and findings suggest there are similarities at the level of behavior, muscle contractile performance and molecular composition between locomotor muscles of obese vertebrates and infected L. pulchella males, and also between infected mature and non-infected teneral L. pulchella males. Here, we first tested the hypothesis that gregarine infection causes a mismatch between body mass, flight muscle Tnt alternative splicing pattern and maximum flight muscle performance in this species. In addition, in a similar fashion to the Seebacher et al. (2017) zebrafish study, we tested the hypothesis that gregarine infection causes a left shift for the optimum of flight muscle power–cycle frequency curves (i.e. they are predicted to lose power output capacity at higher cycle frequencies), and compared these power–cycle frequency curves with those produced by teneral L. pulchella flight muscles.

Mature adult male Libellula pulchella (Drury 1770) dragonflies were caught at several ponds located within a 7–8 mile radius around the Pennsylvania State University campus (University Park, PA, USA). Teneral (recently emerged, immature) adult males were caught at the periphery of these ponds, and identified by overall coloration (i.e. lack of abdominal purple pruinescence seen in mature males), flight behavior and the presence of a rainbow-colored sheen on their wings. After capture, dragonflies were transported to the laboratory in cooled plastic containers containing a moistened paper towel. All animals were used in experiments within 2–6 h of capture. After experiments and determination of infection status and body composition values (see below), basalar muscles (muscle dvm3; Simmons, 1977) were placed in a cryo-vial and flash frozen in a liquid nitrogen Dewar, after which the tubes were stored under liquid nitrogen or at −80°C until used. The remaining carcasses were similarly stored, but separate from the basalar muscles.

It is important to note that we were blind to an individual's gregarine (Hoplorrhynchus sp.) infection status until the end of experiments, when parasite load was assessed by surgically opening the midgut post mortem. Our previous work in this system demonstrated that flight muscle performance of male dragonflies is affected significantly only when a mature infection is present compared with when an immature infection is present (Schilder and Marden, 2006). As in that study, an infection was therefore classified as immature if small gregarine trophozoites were observed, but no fully grown (sexually mature) gregarine trophozoites (the feeding stage of gregarines; Smith and Clopton, 2003) were present. An infection was classified as mature if one or more fully grown gregarine trophozoites were present, regardless of whether any additional smaller trophozoites were present. Body mass, thoracic mass, abdominal mass, and flight (basalar) muscle mass were measured to the nearest 0.1 mg by means of a Mettler AE100 scale.

In situ flight muscle performance was determined using the work loop technique described previously, applied to the basalar muscle (Marden et al., 2001; Schilder and Marden, 2004). Briefly, dragonflies (N=65 mature, N=11 teneral) were decapitated and wings and legs were clipped off, after which the basalar muscle was mechanically isolated from the rest of the thorax via several small incisions around the humoral plate. A fine suture was then tied around the apodeme and glued to a modified insect pin suspended from the lever arm of a lever system (Cambridge Technology 300B, Cambridge, MA, USA). Imposed strain was set to 0.52 mm, which is 10% of mean basalar muscle length for this species. The preparation was electrically stimulated with 0.25 ms, 0.5 V pulses, initiated at 44% of lengthening phase (Marden et al., 2001). Net work produced by the basalar muscle during strain cycles was calculated using the work loop area produced during the fourth of a total of five length cycles imposed at frequencies varying between 20 and 45 Hz. Muscle power output at a given cycle frequency was calculated as net muscle work divided by period (i.e. frequency⁻¹), and standardized by dividing by muscle mass (i.e. mass-specific power output; W kg⁻¹ muscle). Thoracic temperature was monitored using a fine-gauge thermocouple that was inserted into the metathorax and connected to a TC-1000 thermometer (Sable Systems, Las Vegas, NV, USA). Thoracic temperature during these assays was regulated between 32 and 34°C using a water circulator that pumped water through an aluminium stage to which the preparation was affixed. Data acquisition and hardware timing was controlled through custom scripts in Igor Pro v6.3 software, via a National Instruments Multifunction I/O board (NI-6011E). For an independent set of dragonflies (N=38), we used the same protocol but determined maximum mass-specific power output at any contractile frequency, which was subsequently correlated with body mass, and relative abundance of Lp_Tnt261 and Lp_Tnt258, the two most abundant Tnt mRNA splice variants (Table 1).

Tnt splice variant expression was analyzed in similar fashion to previous insect studies on this gene (Marden et al., 1999; Schilder and Raynor, 2017). Briefly, total RNA was extracted from frozen basalar muscle using Trizol reagent (Invitrogen, Carlsbad, CA, USA), precipitated in isopropanol and washed in ethanol, according to the manufacturer's instructions. Total poly-A mRNA was reverse transcribed using a Superscript III reverse transcriptase kit (Invitrogen) and a custom primer (ATTCTAGAGGCCGAGGCGGCCGACATGdT₃₀-VN). L. pulchella flight muscle Tnt splice variants, previously identified by Marden et al., 1999, were amplified from cDNA samples using a fluorescein (FAM)-labeled forward primer 5′-6FAM-CGCTTCTTTCACTCGTTGTTCAAAC-3′ and reverse primer 5′-CCTTCGCTTGGCTTGCTTC-3′. 6FAM-labeled PCR products obtained in this fashion were diluted 1:25 and separated by capillary electrophoresis (ABI DNA Analyzer, Applied Biosystems), after which the abundance of individual Tnt splice variants was quantified based on the amount of fluorescence detected. Relative abundance of individual Tnt splice variant amplicons was calculated by dividing out total Tnt splice variant fluorescence. Tnt splice variant amplicon size was calculated based on the GeneScan™ 500 LIZ™ dye internal size standard and GeneScan™ fragment analysis software (Applied Biosystems). Tnt splice variant amplicons were designated Tnt_Lp2xx based on amplicon size determined in this fashion (Table 1).

The electrophoretic analyses of teneral basalar muscle samples (see Discussion) were performed as described previously (Schilder and Marden, 2007). Briefly, four muscle protein samples were prepared and separated using two-dimensional PAGE. Sample preparation was performed according to instructions provided with a MK-1 tissue preparation kit (Kendrick Labs, Madison, WI, USA). Basalar muscles were first tested for contractile performance (see above) before being dissected out, homogenized in osmotic lysis buffer and heated in a SDS boiling buffer prior to being shipped to Kendrick Labs for 2D PAGE separation (see Schilder and Marden, 2007 for details). Gels were stained with Coomassie Brilliant Blue R-250 to allow protein identification and quantification. Protein spot profiles were analyzed from scanned 2D PAGE gel images using PD_Quest analysis software (Bio-Rad, Hercules, CA, USA). Quantities of the protein spot identified as a ∼155 kDa proteolytic degradation product of the full-length Myhc gene product were calculated as the percentage of total gel spot intensity and designated %MyHC155 (Schilder and Marden, 2007) and log₁₀-transformed prior to statistical analyses. The location of five Tnt protein spots on scanned 2D PAGE gels were identified based on previous work that used a combination of gel electrophoresis and western blotting with an antibody against Tnt proteins (Marden et al., 1999). Quantities of individual Tnt protein spots were normalized to total quantities for all five Tnt protein spots. We then performed a principal component analysis of all Tnt protein spot data and used the first principal component (PC1 Tnt relative abundance) in our analyses that included previously published Tnt protein spot data for non-infected and infected mature adults (Schilder and Marden, 2007).

We based our sampling on availability of field-collected specimens and previous experience with the L. pulchella system, instead of a formal a priori statistical power analysis. Statistical analyses were performed in JMP v12 (SAS Institute) or R v3.4.3 (R Foundation for Statistical Computing) software. Body and tissue mass comparisons were analyzed using one-way ANOVA, followed by either Student's t-tests or Tukey's HSD tests of group means where appropriate. Tnt splice variant relative abundances were arcsine transformed prior to statistical analyses, however, presented in figures and tables are actual, non-transformed relative abundances. Relationships between flight muscle power output, Tnt splice variant relative abundance and body mass were examined by linear regression and ANOVA. In addition, effects of infection on these relationships were examined using two-way ANOVA. The relationship between muscle power output and cycle frequency was analyzed with a mixed effect, two-way ANOVA using ‘infection status’, ‘cycle frequency’ and their interaction term as fixed effects, and ‘frequency’ nested within ‘individual ID’ as a random effect to account for the repeated-measures nature of this experiment (i.e. flight muscle from each individual underwent 25 length–stimulus cycles varying between 20 and 45 Hz). We used post hoc least-squares means contrast tests to identify significant differences between groups for power output at a given cycle frequency, with Bonferroni-adjusted critical α values. The 2D-PAGE protein expression data were analyzed via one-way ANOVA followed by Tukey's HSD tests of group means. A second-order polynomial fit was applied to the data presented. All reported ANOVA F-statistics are in the format F_x,xx, where x=k−1, and xx=number of observations−k, with k=number of groups. Effect sizes were calculated as η² or Cohen's d where appropriate (i.e. for model effects and post hoc pairwise mean comparisons, respectively). Graphs were generated in Igor Pro v8 (Wavemetrics, Inc; Portland, OR, USA).

Gregarine infection had no effects on mature adult male L. pulchella body and tissue masses examined, as mean body, thorax, basalar muscle and abdomen mass, as well as the thorax:body mass ratio was not significantly different between non-infected (NI) L. pulchella and those with immature (II) or mature (MI) infections (Table 2).

However, when body mass variation was taken into account using a two-way ANOVA, infection was associated with a significantly higher basalar muscle mass (F_2,35=4.05; P=0.0264), and this effect was specific to MI individuals (post hoc Tukey HSD on least-square means: P=0.0283) as II individuals showed no significant differences from NI individuals (post hoc Tukey HSD on least-square means: P=0.961; Fig. 1A). Abdomen mass, thorax mass and the thorax:body mass ratio were not significantly different between groups when body mass variation was taken into account (F_2,35=0.015, P=0.985; F_2,35=1.270, P=0.293; F_2,35=1.265, P=0.295, respectively).

The relative abundance of Tnt_Lp261, the second most abundant Tnt splice variant in L. pulchella flight muscle (see Table 1) was positively correlated with maximum mass-specific power output (F_0,37=10.91, P=0.0022) among all individuals examined (Fig. 2A). A two-way ANOVA indicated that infection did not affect this relationship (i.e. the interaction term ‘Tnt_Lp261*infection status’ was not a significant contributor to the model; F_2,35=0.691, P=0.508), even though infection significantly reduced mean flight muscle power output (F_2,35=4.41, P=0.020). When infection status was taken into account in this fashion, relative abundance of Tnt_Lp261 was no longer a significant predictor of maximum mass-specific power output (F_2,35=3.02, P=0.092).

There was no significant relationship between body mass and maximum flight muscle mass-specific power output (F_0,37=1.74, P=0.195) among all individuals examined. However, a statistically significant positive correlation was observed when non-infected individuals were considered separately (F_0,23=8.79, P=0.007; solid line in Fig. 2B) from individuals with immature (F_0,5=5.136, P=0.086) and mature (F_0,7=0.706, P=0.433) infections. Two-way ANOVA indicated that while body mass itself was not a significant predictor of power output (F_2,35=2.504, P=0.123), the interaction between body mass and infection status was highly significant (F_2,35=7.120, P=0.003, η²=0.212), suggesting that infection affected the relationship between flight muscle power output and body mass.

Similarly, the relative abundance of Tnt_Lp261 was positively correlated with body mass in non-infected individuals (F_0,23=7.10, P=0.0142; solid line in Fig. 2C), but not in individuals with immature (F_0,5=0.011, P=0.922) and mature (F_0,7=1.106, P=0.334) infections. Two-way ANOVA indicated that infection significantly reduced mean Tnt_Lp261 relative abundance (F_2,35=4.18, P=0.024, η²=0.177), while body mass, and the interaction between body mass and infection status were not significant predictors of Tnt_Lp261 relative abundance (F_2,35=0.012, P=0.913, η²=0.0003; F_2,35=2.074, P=0.142, η²=0.088, respectively).

In contrast to Tnt_Lp261, there was no significant relationship between Tnt_Lp258 (the most abundant Tnt splice variant) relative abundance and flight muscle power output for either non-infected, or individuals with immature and mature infections (F_0,23=0.670, P=0.422, F_0,5=0.333; P=0.594, F_0,7=5.757, P=0.053, respectively; Fig. 3A). Two-way ANOVA indicated that Tnt_Lp258 relative abundance and its interaction with infection status, were non-significant predictors of flight muscle power output (F_2,35=3.324, P=0.078, F_2,35=0.243, P=0.786, respectively), while as before, infection status was a significant predictor of flight muscle power output (F_2,35=7.839, P=0.0017). Similarly, no significant correlation between body mass and Tnt_Lp258 relative abundance was found for non-infected individuals (F_0,23=0.315, P=0.581) and individuals with immature or mature infections (F_0,5=0.029, P=0.873; F_0,7=0.076, P=0.792, respectively; Fig. 3B). Two-way ANOVA indicated that body mass, infection status or their interaction were not significant predictors of Lp_Tnt258 relative abundance (F_2,35=0.031, P=0.862; F_2,35=0.216, P=0.807; F_2,35=0.095, P=0.909, respectively).

Male L. pulchella flight muscle power–cycle frequency curves, i.e. the performance behavior across the entire range of contractile frequencies (20–45 Hz) that we imposed on 54 (13 MI, 10 II, 31 NI) flight muscle preparations, were significantly affected by gregarine infection. Infection status (F_2,1401=65.20, P<0.0001), cycle frequency (F_2,1401=67.71, P<0.0001) and the infection status–cycle frequency interaction term (F_2,1401=2.04, P<0.0001) were all significant predictors of mean power output. The statistical significance of the interaction term suggested that the effect of infection was specific to certain cycle frequencies. Indeed, post hoc least-squares means contrast tests showed that muscle power output of NI individuals is significantly higher than that of MI individuals at higher cycle frequencies (38–45 Hz, P≤0.00012, α_adj=0.00064; Fig. 4A). II individuals showed an intermediate power–cycle frequency relationship that was not statistically different from NI individuals, but was significantly different from MI individuals at 45 Hz (P=0.00023, α_adj=0.00064; Fig. 4A).

The mean power–cycle frequency curve observed for MI individuals also resembled that of flight muscle preparations of teneral individuals. Addition of this fourth group (i.e. in addition to NI, II and MI) did not significantly change the outcome of the two-way, repeated-measures ANOVA as ‘status’ (F_3,1686=55.04, P<0.0001), cycle frequency (F_3,1686=27.82, P<0.0001) and the interaction term of cycle frequency and ‘status’ (F_3,1686=3.68, P<0.0001) were significant predictors of mean power output. Post hoc least-squares means contrast tests (of teneral versus NI, II, MI, for all cycle frequencies) showed that mean power output by tenerals was not significantly different from MI individuals at any cycle frequency, but was significantly lower than NI individuals at cycle frequencies of 36–45 Hz (P≤0.00006, α_adj=0.00032; also see Fig. 4A) and lower than II individuals at cycle frequencies of 40–45 Hz (P≤0.00023, α_adj=0.00032; also see Fig. 4A), respectively.

Thus, infection caused a significant reduction in the cycle frequency at which maximum performance was achieved (Fig. 4B). That is, a one-way ANOVA (F_3,63=10.26, P<0.0001, η²=0.332) followed by post hoc LSmeans Tukey HSD test indicated that the mean cycle frequency at which maximum power output was achieved (i.e. optimal cycle frequency) was not different between NI and II individuals (P=0.955, Cohen's d=0.174). Mature infection caused a significant reduction in mean optimal cycle frequency (P=0.0061 and P=0.0089 for MI–NI and MI–II comparisons, respectively; Cohen's d=1.232 and 1.406, respectively), so that it was no different from mean optimal cycle frequency for teneral flight muscle preparations (P=0.883, Cohen's d=0.321).

This set of studies aimed to further our understanding of the effects of a non-lethal intestinal parasitic infection on flight performance in the dragonfly L. pulchella through a comparison of several skeletal muscle phenotypes known to be affected in obese vertebrates. We first demonstrated that infection is associated with an increase in mean relative basalar flight muscle mass (Fig. 2) in MI individuals. This was a somewhat unexpected finding as for both relative (i.e. normalized to total body mass) thoracic and abdominal mass there was no such effect of infection, suggesting that this effect is specific to the basalar muscle. While it is tempting to speculate that this result may reflect compensatory muscle growth to offset mass-specific performance loss during infection, this appears to not be the case and maximum absolute (i.e. non-normalized) power output is also significantly lower in infected individuals (data not shown). Muscle hypertrophy in insects is still poorly understood mechanistically, and typically described in the context of maturational or developmental muscle growth (Marden, 2000; Piccirillo et al., 2014). Even less well-studied are muscle mass and function dynamics in mature adult insects in response to natural aging and abiotic or biotic stressors such as infections. Our current findings in the L. pulchella system suggest that this may be a fruitful research avenue, and given that most insects suffer such stressors in nature, there may be many other invertebrate systems in which to examine the existence and relevance of locomotor muscle mass dynamics across their lifespan. Such studies could be relevant to biomedical research fields, because, for example, selective hypertrophy of load-bearing skeletal muscle occurs in obese vertebrates (Tomlinson et al., 2016) – even though skeletal muscle performance per unit mass is reduced (Bollinger, 2017; Lafortuna et al., 2005; Maffiuletti et al., 2007) – in highly similar fashion to results obtained in the current study.

Data presented in Fig. 2 indicate that infection impairs normal relationships between flight muscle performance, Tnt alternative splicing and body mass. Interestingly, this impairment is similar to that observed in rats in whom obesity develops as a result of genetically induced hyperphagy (i.e. Zucker rats; Schilder et al., 2011). In healthy rats (Marden et al., 2008; Schilder and Raynor, 2017), Tnt alternative splicing is adjusted in a quantitative fashion in response to body weight changes due to natural growth or experimental body weight manipulations. Our current study suggests that similar, body weight-dependent adjustments in Tnt alternative splicing occur in L. pulchella flight muscle, as in non-infected individuals there was a statistically significant, positive correlation between body mass and the relative abundance of Tnt_Lp261 (Fig. 2C), one of two highly abundant Tnt splice variants in this tissue (see Table 1). In addition to signals reflecting quantitative variation in body weight, Tnt alternative splicing in mammals and insects is sensitive to dietary quantity and quality (Marden et al., 2008; Black et al., 2017; Portman et al., 2015). Moreover, in non-infected L. pulchella males, flight muscle power output is correlated with total body fat content, but this correlation is absent in infected males (Marden and Cobb, 2004).

Combined, these observations from different vertebrate and insect systems suggest that in healthy organisms, muscle performance can be adjusted by an unknown mechanism that integrates the input of sensors receiving information about body weight and energetic status. It appears that in insects and vertebrates alike, Tnt alternative splicing is a useful molecular trait to study this mechanism, although it likely represents a much larger suite of sarcomere and non-sarcomere genes whose expression is controlled by this mechanism (Black et al., 2017).

Mature gregarine infections also caused a change in the shape of L. pulchella flight muscle power–cycle frequency curves (Fig. 4A), as mean optimal cycle frequency in MI individuals was significantly reduced compared with mean values in NI and II (Fig. 4B). This finding is supported by the observed effects of infection on flight muscle Tnt_Lp261 relative abundance (Fig. 2), which are expected to lower flight muscle calcium sensitivity, and mass-specific performance at higher cycle frequencies (Marden et al., 1998; Marden et al., 2001). Interestingly, this left-shifted power-cycle frequency curve of MI individuals was not significantly different from that produced by flight muscles of teneral L. pulchella males, and is also similar to the results recently obtained for obese zebrafish (Seebacher et al., 2017). Seebacher et al. (2017) suggested that this loss of performance at higher contractile frequencies may be explained by the overall reduction in both slow and fast MyHC content (i.e. reduced overall myofibrillar density) that they observed in obese zebrafish individuals. That study also noted that this defect may be due to impairment of signaling pathways that mediate expression of multiple muscle sarcomere proteins (Bassel-Duby and Olson, 2006; Watts et al., 2013, in: Seebacher et al., 2017). Our results in Fig. 2 showing a disruption of normal relationships between Tnt alternative splicing, body mass and flight muscle power output appear to be in agreement with this notion, even if they do not address MyHC protein expression.

As discussed in the Introduction, it remains unclear whether functionally distinct MyHC protein isoforms are expressed in insect flight muscle as in mammalian skeletal muscles, but, related to the zebrafish findings above, we did previously demonstrate that gregarine gut infections are associated with a significant reduction of L. pulchella flight muscle quantities of a (putative) MyHC proteolytic degradation product (MyHC155; Schilder and Marden, 2007). Using the same methodology as in that paper (see Materials and Methods), we here show that the quantity of teneral flight muscle MyHC155 on 2D PAGE gels (Fig. 5A) and its relationship with flight muscle power output (Fig. 5B) are similar to that of flight muscles of MI individuals (P=0.999) but significantly different from NI and II individuals (P<0.0001, P=0.0372, respectively; Fig. 5A,B).

How might this reduction in flight muscle quantities of MyHC155 relate to flight muscle performance? Interestingly, MI individuals have significantly increased flight muscle mass (Fig. 1) but with lower maximum mass-specific power output (Schilder and Marden, 2006; Fig. 4A). It is therefore possible that relatively low levels of MyHC155 (i.e. MyHC degradation) reflect relatively low overall flight muscle MyHC protein production in MI individuals, which may contribute to the reduction of muscle mass-specific performance at higher contractile frequencies, as was suggested for obese zebrafish muscle (see above; Seebacher et al., 2017). Alternatively, MyHC degradation could be lower in MI individuals, and similar to that in teneral flight muscle, because of infection-induced reduction of flight behavior intensity (Marden and Cobb, 2004; Schilder and Marden, 2006) and, associated with this, reduced intensity of overall sarcomere protein use (for further discussion, see Schilder and Marden, 2007). Unfortunately we do not have data on infection effects on myofibrillar protein density (i.e. sarcomere protein content per unit muscle mass), so this has to remain a speculative set of ideas.

While in this study we did not quantify Tnt mRNA splice variant abundance for teneral flight muscles, the same 2D PAGE analyses indicated that overall teneral flight muscles Tnt protein expression (i.e. PC1 Tnt relative abundance) is significantly different from mature adults, irrespective of their infection status (P<0.0001 for all comparisons; Fig. 5C). Therefore, teneral flight muscle appears to be similar to that of MI individuals with respect to MyHC degradation dynamics, but different at the level of Tnt protein expression, suggesting that different combinations of sarcomere traits may yield similar physiological phenotypes (e.g. power–cycle frequency curves) in these two groups.

Regardless of the exact molecular traits involved, the similarity of power–cycle frequency curves between MI and teneral individuals does raise the interesting question as to whether the parasite is directly manipulating the dragonfly host physiology and behavior for its benefit or if infected mature male L. pulchella may be adopting a low-intensity more energy efficient behavioral strategy to cope with metabolic impairments caused by infection. Gregarine parasites potentially use excretory chemicals to manipulate their dragonfly host (Schilder and Marden, 2006), but currently available data on these chemical interactions do not yet allow us to distinguish between these two scenarios definitively.

While it is evident that the non-lethal gregarine infection has significant effects on multiple aspects of L. pulchella flight physiology, we remain somewhat cautious with our comparisons to mammalian obesity, as currently available data limit us to identification of associations of these effects with infection, rather than exact mechanisms through which these effects are accomplished. Here, we have demonstrated that relationships between Tnt alternative splicing, body mass and flight muscle contractile performance are affected in L. pulchella males in similar ways to those described for obese mammals and fish. However, as appears to be the case with these vertebrate systems, our understanding of molecular mechanisms that lead to these impairments is still incomplete. Nonetheless, what should be evident from this study is that metabolic disease syndromes with a high degree of similarity to mammalian obesity and diabetes exist in nature, and that detailed study of additional natural systems may provide pertinent information regarding environmental factors that can contribute (in parallel, or interactively with effects of diet and exercise) to the development of such syndromes. As mentioned earlier, it is possible that flying insect systems such as dragonflies are particularly vulnerable to homeostatic disturbance caused by such environmental factors (e.g. parasitic infections) due to their ‘high-octane’ lifestyle, but given that gregarine infections are common in arthropods (Smith and Clopton, 2003), this possibility can be readily explored through the study of other species.

The authors would like to thank two anonymous reviewers, whose comments greatly improved the manuscript.