ABSTRACT
Locusts have been reported to elevate metabolic rate in response to high carbohydrate diets; this conclusion was based on metabolic rates calculated from CO2 production, a common practice for insects. However, respiratory exchange ratio (RER, CO2 production divided by O2 consumption) can rise above 1 as a result of de novo lipid synthesis, providing an alternative possible explanation of the prior findings. We studied the relationship between macronutrient ingestion, RER and lipid synthesis using South American locusts (Schistocerca cancellata) reared on artificial diets varying in protein:carbohydrate (p:c) ratio. RER increased and rose above 1 as dietary p:c ratio decreased. Lipid accumulation rates were strongly positively correlated with dietary carbohydrate content and ingestion. RERs above 1 were only observed for animals without food in the respirometry chamber, suggesting that hormonal changes after a meal may drive lipid synthesis. Schistocerca cancellata does not elevate metabolic rate on low p:c diets; in fact, the opposite trend was observed.
INTRODUCTION
Diets in nature vary tremendously in protein:carbohydrate (p:c) ratio, but we still lack a firm understanding of how animals cope physiologically with this variation. During the last 30 years, locusts and grasshoppers have become one of the most important models for testing behavioral and physiological responses to dietary variation (Behmer, 2009; Simpson and Raubenhimer, 2012). It has been well demonstrated that locusts feeding on diets low in p:c ratio synthesize more lipid (Simpson and Raubenheimer, 2001; Zanotto et al., 1993). Additionally, locusts fed on artificial diets with lower p:c ratio have been reported to increase their CO2 production rate (V̇CO2), leading to the suggestion that locusts exhibit ‘wastage respiration’ on low p:c diets as a way to get rid of excess ingested carbohydrates (Zanotto et al., 1993, 1997). However, in these prior studies, metabolic rate was inferred from V̇CO2, leaving open the possibility of the alternative hypothesis that the elevated CO2 emission rate observed for locusts consuming low p:c food occurred because of an increase in the respiratory exchange ratio [V̇CO2 and O2 consumption rate (V̇O2)], RER.
RER is a dynamic parameter, widely used in organismal and clinical biology to indicate fuel usage and calculate metabolic rate (Gluck et al., 2011; Högberg et al., 2006; Longo et al., 2010). For starved animals, RER values of 1 or 0.7 occur when metabolism is completely fueled by carbohydrates or lipids, respectively. Catabolism of proteins yields a RER of 0.8–0.85 (Elia and Livesey, 1988; Kleiber, 1961; Livesey and Elia, 1988). While RER is thought to go no higher than 1 in starved animals, when food is accessible, RER can exceed 1. For example, hummingbirds showed diurnal fluctuation of RER between 0.7 and 1.3 (night and day, respectively), reaching a maximum during the day when they had access to ∼20% sugar water (Powers, 1991). When commercial pigs are fattened, RER values up to 1.34 have been measured (Jakobsen and Thorbekt, 1993). One-week-old Ross chickens (broilers) had RER values of ∼1.2 when feeding on commercial food (Geelissen et al., 2006), and force-fed geese had post-feeding RER values as high as 1.4 (Benedict and Lee, 1937). The primary hypothesis to explain high RER values is de novo lipid synthesis, and this hypothesis is supported by multiple theoretical models (Elia and Livesey, 1988; Ferrannini, 1988; Hellerstein et al., 1996; Hellerstein, 1999; Livesey and Elia, 1988). In rodents, RER routinely goes above 1 in the feeding nocturnal phase, and this has been shown to be associated with de novo lipid synthesis (Ono-Moore et al., 2020; Wahlig et al., 2012). An alternative hypothesis to explain a RER above 1 is the use of the pentose phosphate pathway to generate antioxidants (Levin et al., 2017).
Here, we tested whether high carbohydrate diets induce elevated metabolic rate in locusts, whether dietary p:c ratio affects RER, and whether lipid accumulation explains variation in RER in locusts. To do this, we examined the effect of dietary p:c ratio on V̇CO2, V̇O2, RER, carbohydrate consumption and lipid accumulation. We utilized the South American locust Schistocerca cancellata, as their large size facilitates such measurements, and a contemporary outbreak of this species meant that they were highly available in the field.
MATERIALS AND METHODS
South American locusts
Schistocerca cancellata (Serville 1838) is a South American locust species which is usually limited to a narrow breeding zone in Argentina. However, during massive outbreaks, they can cover up to four-million square kilometers, including 6 countries (Medina et al., 2017). The last outbreak started at the beginning of 2015, and was still continuing during 2020. They are highly polyphagous, but earlier research suggests field populations of 5th and 6th instar nymphs prefer carbohydrate-biased diets (Talal et al., 2020).
Ethics statement
No special collecting permit or animal care protocol was required for this work. Field efforts in Paraguay were supported by the SENAVE, Paraguay. USDA permit to import locusts: P526P-19-03892 (permit holder: Dr Arianne Cease).
Animals and experimental design: field-based experiments
We collected S. cancellata 5th instar nymphs as they were marching in Gran Chaco, Paraguay, during April 2019. Details on collection, handling and lab rearing are provided in Talal et al., 2020; a brief description is provided here. Animals were brought to a SENAVE field lab and reared in group cages on locally collected grass (Paspalum sp.) for 3 days. Locusts were then provided with one of five different isocaloric artificial foods, which varied in protein and digestible carbohydrates (Raubenheimer and Simpson, 1993; Simpson and Abisgold, 1985): 7p:35c, 14p:28c, 21p:21c, 28p:14c and 35p:7c (% protein and % digestible carbohydrates, by dry mass), or were provided with both of the two most biased diets, 35p:7c and 7p:35c, to allow them to select their preferred p:c mixture. All the diets contained 54% cellulose and 4% vitamins and salts; proteins were provided as a mix of 3:1:1 casein:peptone:albumen, and carbohydrates were provided as a 1:1 mix of sucrose and dextrin. For 3–5 days before respirometry, nymphs were kept in groups of 15–20 individuals, in cages (20.3×20.3×20.3 cm) containing adlibitum amounts of their treatment diet and water tubes. During the respirometry measurements, locusts were not provided with food. Following respirometry, each individual was weighed using a portable scale (SLF103, Fisher Science Education, Waltham, MA, USA). In the field lab, the light:dark cycle was approximately 12 h:12 h, the temperature in the room averaged 32.2±1.9°C and relative humidity averaged 58.7±4.5% (means±s.d.), though humidity may have been higher in the locust chambers.
Animals and experimental design: lab experiments
In order to confirm the effect of different diets on energy metabolism, and to test whether metabolic responses differed during versus after food ingestion, we carried out respirometry under lab conditions with 6th (terminal) instar nymphs from a lab-reared population, with two food accessibility treatments (with or without food during respirometry). The lab experiments were conducted at Arizona State University (ASU) using 6th instar S. cancellata nymphs (two days after molt) from a population reared for nine generations in the lab from locusts collected from two locations in Argentina (Rio Cuarto, Córdoba, and Casa de Piedras, Catamarca). Lab rearing conditions were 30.0±0.5% relative humidity, 34.0±0.5°C during the day and 25.0±0.5°C during the night, under a 14 h:10 h light:dark photoperiod (supplementary radiant heat was supplied during the daytime by incandescent 40 W electric bulbs). During standard rearing, locusts were fed daily with wheat shoots, fresh lettuce leaves and wheat bran ad libitum. During the 5 days prior to respirometry, locusts were provided with one of three artificial diet treatments: 35p:7c, 21p:21c and 7p:35c. During this time, as for the field-based experiments, nymphs were reared in cages containing 15–20 individuals with adlibitum treatment diet and water. During the respirometry measurements, some locusts were provided with 0.3 g of their treatment diet and others had no food. Following respirometry, each individual was weighed.
Lipid accumulation and carbohydrate consumption
We measured lipid accumulation from the change in body lipid content over the course of the field experiment, using a different set of individuals. We used freshly caught 5th instar marching nymphs to measure initial lipid content. The nymphs were reared for 8 days on the two complementary diets (choice), or 6 days on one of the five no-choice diet treatments described above, and then killed by freezing and dried to a constant mass. We measured the macronutrient consumption by measuring the change in dry mass of the provided diets.
We used a chloroform extraction technique to measure lipid content (Loveridge, 1973), and calculated lipid accumulation (g) and accumulation rate (g day−1) from the change in body lipid mass on each diet divided by number of days on the diet treatment. We measured carbohydrate consumption from the change in mass of dishes containing chemically defined artificial diets during the experiment.
Respirometry
We performed constant volume respirometry using a FoxBox field respirometry system (Sable Systems International, Las Vegas, NV, USA). The span of the oxygen analyzer was calibrated several times a day by flushing the system with dry, CO2-free air for at least 20 min. The calibration of the CO2 analyzer was carried out at the ASU lab, using pure nitrogen and two certified calibration tanks (252±1 and 1010±1 ppm of CO2 balanced in nitrogen, factory certified). The respirometry chambers were 20 and 60 ml syringes (closed with a three-way valve) for the field studies and lab studies, respectively, because of the almost 3-fold difference in body mass between field-collected (∼0.5 g) and lab-reared (∼1.3 g) animals.
After inserting the nymph into the metabolic chamber, the chamber was flushed with dry, CO2-free air for 1 min at a flow rate of 500 ml min−1. The syringe was then sealed and placed at the rearing temperature for 50–70 min, after which 18 ml/45 ml (depending on the syringe size) of air was injected into a stream of dry, CO2-free air, at a flow rate of 500 ml min−1, which passed through a magnesium perchlorate column, CO2 analyzer (FoxBox), an Ascarite®/silica gel column and then an oxygen analyzer (FoxBox). We corrected the metabolic chamber volume by subtracting the animal volume from it, which was calculated from animal mass assuming a density of 1. Baselining was repeated between individual measurements by passing dry CO2-free air directly through the analyzers. Data collection and analysis were carried out using a UI-3 data acquisition interface and Expedata software (Sable Systems International).
Statistics
Statistical analyses were performed using SPSS 19.0 statistical software (IBM, Armonk, NY, USA). Prior to using any parametric analysis, data normality and homoscedasticity were confirmed. log10 transformation of RER yielded data that satisfied assumptions of parametric analysis. In order to compare the effect of different diets on V̇CO2 and V̇O2, we used one-way ANCOVA with mass as a covariate. We used one-way ANOVA to test diet effects on log10RER. Because we predicted a positive linear effect of dietary carbohydrate on gas exchange, we also used a general linear model (GLM) to test the effect of body mass and dietary carbohydrate content (%) on V̇O2 and V̇CO2. For the lab-reared locusts, we used a GLM, testing the effects of dietary carbohydrate content (%), body mass and food availability (scoring ‘0’ when food was not available and ‘1’ for available food) on V̇CO2 and V̇O2. We used a two-way ANOVA to test the effect of different diets and food availability on log10RER measured on the lab-reared locusts.
RESULTS AND DISCUSSION
For field-collected S. cancellata, diet treatments did not significantly affect V̇O2 (F5,86=1.41, P=0.23), and V̇CO2 (F5,86=0.88, P=0.50) when analyzed by ANCOVA (Fig. 1A). However, with a GLM, both the effects of body mass and dietary carbohydrate content (%) were highly significant for V̇O2 (Fig. 1B: GLM corrected model: r2=0.55, F2,73=47.45, P<0.001; mass: F1,73=82.49, P<0.001; dietary carbohydrate content: F1,73=5.31, P=0.024); V̇CO2 was only affected significantly by mass (Fig. 1B: GLM corrected model: r2=0.51; F2,73=40.66, P<0.001; mass: F1,73=81.23, P<0.001; dietary carbohydrate content: F1,73=2.17, P=0.145). The measured RER increased strongly as dietary p:c ratio decreased, with RER rising from a mean of 0.88 on the most protein-biased diets to a mean of 1.15 on the most carbohydrate-biased diet (one-way ANOVA: F5,86=22.55, P<0.001; Fig. 1C). Individuals given a choice of two diets had RER values of ∼1.05, significantly higher than those of the 28p:14c and 35p:7c treatment groups, but not significantly different from the other three treatment groups (Bonferroni post hoc tests; Fig. 1C). The corrected RER values (see Materials and Methods) were below 1 for all diet treatments (Fig. 1C).
For the lab-reared S. cancellata, V̇O2 was affected by mass, food availability and the percentage of dietary carbohydrates (GLM corrected model: r2=0.51, F3,89=30.94, P<0.001; mass: F1,89=66.29, P<0.001; food availability: F1,89=4.38, P=0.039; dietary carbohydrate content: F1,89=18.32, P<0.001; Fig. 2A). As in the field, the V̇CO2 of lab-reared locusts was affected only by mass (GLM corrected model: r2=0.335, F3,89=14.96, P<0.001; mass: F1,89=40.73, P<0.001; food availability: F1,89=1.67, P=0.200; dietary carbohydrate content: F1,89=1.75, P<0.189) (Fig. 2B). The RER was significantly affected by food availability (F1,86=11.22, P=0.001) and diet treatment (F2,86=26.33, P<0.001), as well as by the interactions of these two factors (F2,86=36.07, P<0.001).
Mean lipid accumulation rate was highly correlated with the mean carbohydrate consumption rate of each diet treatment group (Fig. 3A). The slope, which is the efficiency of conversion of dietary carbohydrates to lipid stores, was ∼17%. In addition, lipid accumulation rate was highly correlated with RER across diet treatment groups (Fig. 3B).
Our data do not support the ‘wastage respiration hypothesis’ for high carbohydrate diets of Zanotto et al. (1997). In contrast, we show that, for S. cancellata, V̇O2 falls as the p:c ratio drops (Figs 1 and 2). Our results are consistent with other studies that have found higher metabolic rate (elevated V̇O2) on diets higher in protein content, including research on shrimps (Taboada et al., 1998), birds (MacLeod and Dabutha, 1997), fish (Jobling and Davies, 1980) and humans (Johnston et al., 2002), likely due to the energetic costs of increased gluconeogenesis from amino acids when dietary p:c ratio is high (Veldhorst et al., 2009; Williamson et al., 1971). Additionally, having food available during respirometry elevated V̇O2 (Fig. 2A). This could be the effect of increased metabolism by digestion, absorption and assimilation processes (McCue, 2006).
Consumption of low p:c diets is associated with de novo synthesis of lipid from carbohydrate, driving RER above 1 in S. cancellata. The effects of diet on RER were similar for the unfed field-caught and lab-reared locusts (Figs 1 and 2), though we only measured lipid accumulation in the field-caught locusts. The elevated RER during lipid synthesis from carbohydrate occurs as a result of CO2 release as pyruvate is converted to acetyl COA, as well as through activation of the pentose phosphate pathway (Schulz, 1978; Vagelos, 1971). The pentose phosphate pathway is activated during lipid synthesis to produce NADPH, which is required for fatty acid elongation (Elia and Livesey, 1988). Our results do not refute Levin et al.’s (2017) hypothesis that RER above 1 can be caused by activation of the pentose phosphate pathway in order to generate antioxidants. However, our results suggest that any study documenting RER above 1 should test for lipid synthesis as an explanatory factor.
When the quantity of lipids synthesized from carbohydrate can be estimated (as in this study), it is possible to calculate a RER corrected for lipid synthesis, that should represent the RER associated with catabolism, using Eqn 1. The lipid synthesis-corrected RER was ∼0.96 on the highest carbohydrate diet, and decreased to near 0.83 on the highest protein diets (Fig. 1C), reasonable values that suggest that locusts catabolize primarily carbohydrate when eating carbohydrate-rich diets and protein when eating protein-rich diets.
RER values only rose above 1 when food was not available, suggesting hormonal stimulation of lipid synthesis occurs after meal cessation. Our results showing that RER rose when food was not available contrast somewhat with results for rodents, in which RER falls when food is not consumed (Hou et al., 2019; Wahlig et al., 2012). However, the time frame studied and the temporal pattern of feeding differ in locusts and rodents. Locusts can consume food throughout the day and night if conditions are right, alternating multiple minutes of consumption and ‘rest’ during which food is processed (Simpson, 1990). Regardless of diet, locusts with artificial food in the respirometer (all of which ingested food for at least part of the meal) had a RER near 1, with RER only rising strongly above 1 on the high carbohydrate food when no food was available (Fig. 2). Perhaps de novo lipid synthesis is not appreciably activated until after cessation of feeding; alternatively, during feeding, much of the metabolic rate may be due to carbohydrate oxidation by the muscles required for feeding. The hormonal control of lipid synthesis and its timing relative to ingestion remain poorly known. In mammals, high lipid synthesis is triggered by post-feeding insulin secretion (Moustaïd et al., 1996; Smith and Kahn, 2016). While insulin/insulin like growth factor (IGF) occur in insects, there is insufficient evidence to indicate whether insulin-related peptides have similar effects in insects (Badisco et al., 2013; Nässel and Vanden Broeck, 2016). In Rhodnius prolixus, hemolymph lipids and carbohydrates rise following RNAi injection against IGF transcript, suggesting inhibition of lipid and glycogen synthesis in the fat body (Defferrari et al., 2016). However, injection of insulin-like peptide into Bombyx mori larvae did not affect the lipid stores in hemolymph or fat body (Kawabe et al., 2019).
RER values are routinely used to calculate metabolic rate (Lighton, 2008); however, our study suggests that current conversions may lead to considerable errors for feeding/growing animals and for insects catabolizing substantial protein. Many studies of insect respiration report only V̇CO2 and calculate metabolic rate based on assumed RER. It is also relatively common for insect physiologists to measure metabolic rate for non-starved individuals, given that starvation is not a normal situation for a herbivorous insect such as a locust. Here, we demonstrated that the RER value is highly variable and is affected by dietary macronutrient ratio, and that metabolic rate calculated only from V̇CO2 may be misleading. Also, the standard conversion from oxygen to joules or calories assumes no protein catabolism. For vertebrates, it is relatively common to calculate metabolic rate based on non-protein RER, which is usually estimated from measurements of nitrogenous waste (Elia and Livesey, 1988; Ferrannini, 1988; Livesey and Elia, 1988; Simonson and DeFronzo, 1990). The equations that exist to do this are based on mammalian models that assume that the major nitrogenous end-product is urea (Gessaman and Nagy, 1988). However, insects excrete urates, allantoin and ammonia, the ratio of which can depend on hydration state and diet (Harrison, 1995; Harrison and Kennedy, 1994; Zanotto et al., 1993). No calculations exist in the literature to estimate metabolic rate from non-protein RER for animals with such diverse excretory products. In mammals, it has been estimated that the error in metabolic rate from not including protein catabolism is only 1–2% if protein metabolism is ∼12% of the total metabolism (Ferrannini, 1988; Gessaman and Nagy, 1988; Kaiyala et al., 2019; Simonson and DeFronzo, 1990). However, it is plausible that the contribution of protein catabolism to total metabolism is much greater on very high p:c diets, such as may occur with predaceous insects, which could lead to much higher error.
Importantly, in order to use these common equations for metabolic rate calculation, two assumptions must not be violated, which, surprisingly, are not familiar to many physiologists. First, because indirect calorimetry estimates heat production based on nutrient combustion and respirometry gases, it assumes that V̇CO2 and V̇O2 values are produced/consumed only by oxidation. However, CO2 is also produced when de novo lipid synthesis occurs (Eqn 1). Second, all excretory nitrogen products are assumed to be produced by protein/amino acid deamination for oxidation (Simonson and DeFronzo, 1990). However, on high p:c diets, nitrogenous waste may be produced to support gluconeogenesis from amino acids, which could be relatively high when animals are fed on high protein/low carbohydrate diets (Veldhorst et al., 2009; Williamson et al., 1971). Clearly, further research is required to develop appropriate conversion factors to calculate metabolic rate from gas exchange for growing animals, and for animals such as insects that utilize a diversity of nitrogenous waste products. Careful measurement of RER under various conditions, combined with assessment of metabolic pathway flux and/or direct calorimetry will be required.
Acknowledgements
We want to thank Jacob Youngblood for his essential help in the field and field animal maintenance, as well as technical assistance during our field trip to Paraguay. Many thanks to Kelly O'Meara and Rick Overson for organizing and coordinating our research trip to Paraguay. We want to thank SENAVE for providing us with accommodation in Chaco, Paraguay, as well as field lab conditions for nutritional experiments. Many thanks to Marco Antonio Sosa Rolon who provided any help that we needed throughout our research in Paraguay. We also want to thank Eduardo Trumper and Fernando Copa for providing their essential knowledge of this locust species. Many thanks to Fredy Colque, Ricardo Oyols, Cesar Espinoza Vilca and Milton Cortez, our SENASAG-Bolivian collaborators for helping us locate locust bands near the Paraguay–Bolivia border and who took part in field experiments and plant and locust collections.
Footnotes
Author contributions
Conceptualization: S.T., A.C., H.E.M., J.R., J.H.; Software: S.T.; Validation: S.T., A.C., R.F., J.H.; Formal analysis: S.T., R.F.; Investigation: S.T., A.C., R.F., H.E.M., J.R., J.H.; Writing - original draft: S.T., A.C., J.H.; Project administration: A.C., J.H.
Funding
This work was supported by National Science Foundation (NSF) IOS-1826848 and United States - Israel Binational Agricultural Research and Development Fund (BARD) FI-575-2018 grants.
References
Competing interests
The authors declare no competing or financial interests.