Protein storage and reproduction increase in grasshoppers on a diet matched to the amino acids of egg yolk protein

Animals' diets often do not match their nutritional needs, and protein consumption is especially important (Simpson et al., 2017). Many agricultural studies have identified dietary proteins that increase animal growth (e.g. Bergen, 2007), but fewer studies have addressed how dietary protein affects animal development from one distinct stage to the next (e.g. molting, clutch laying). Studies on insects have shown that high protein intake is associated with faster juvenile development (e.g. Kemirembe et al., 2012; Simpson and Raubenheimer, 2000; Simpson et al., 2017). While high dietary protein quantity typically accelerates development (reviewed by Behmer, 2009), the effects of dietary protein quality (with known amino acid composition) on development are not well understood. Foods in the natural world can vary greatly in amino acid composition. Protein quality, the proper dietary balance of amino acids for the species, is critical for normal growth and development while maintaining good health (Marinangeli and House, 2017).

Storage of nutrients can help support major developmental events, by decoupling developmental needs from immediate dietary availability. While storage for developmental events is often conceptualized as fat accumulation (e.g. for fertility, Frisch and Revelle, 1970), arthropods can store proteins to support developmental events including molting (Burmester, 2002), egg production (Hatle et al., 2001; Juliano et al., 2004), male reproductive displays (Short et al., 2020) and caste determination (Zhou et al., 2007). Stored protein, as opposed to stored fat or glycogen (both of which lack nitrogen), has the potential to support developmental transitions that demand large amounts of amino acids. For example, in adult female insects, the accumulation of hexameric storage proteins precedes egg production (Hatle et al., 2001). The best evidence of protein storage coming before oocyte growth is from the caterpillar Hyalophora cecropia, in which radiolabeled leucine and methionine carried in hexamerins was injected into pupae and recovered in vitellogenin (Vg) in adults (Pan and Telfer, 1999). Understanding how protein storage accumulates in animals on diets of varying protein quality can provide insight into the mechanisms by which high-quality dietary protein supports development. In particular, high-quality protein diets could be predicted to speed the rate of egg production and increase the quantity of eggs.

To rigorously test for effects of protein quality on development and reproduction, treatment groups fed isonitrogenous diets (i.e. diets with the same quantity of amino acids, but different amino acid compositions) should be compared, and development measured. In one such study, caterpillars (Spodoptera littoralis) fed on the low-quality plant protein zein developed slower than caterpillars fed casein, especially when the diet contained a low quantity of zein (Lee, 2007). Fruit flies (Drosophila melanogaster) on high-quality protein diets can be fed a low protein concentration but nonetheless have high reproductive output (Ma et al., 2020). However, none of these studies have investigated the role of storage. The goal of our study was to test protein storage and development in animals on a high-quality protein diet, in comparison to a low-quality protein diet.

Determining the composition of high-quality protein for an animal requires identifying the species-specific amino acids needs (Akiyama et al., 1997). Examples of methods that previously have been employed to evaluate dietary protein quality include measuring nitrogen intake and excretion, determining the absorption rate of the amino acids, and measuring oxidation of an indicator amino acid (e.g. Jahan-Mihan et al., 2011; Elango et al., 2012).

A new approach to determining the species-specific balance of dietary amino acids that exemplifies high protein quality has recently been identified. Feeding fruit flies a diet prepared using the amino acids coded for by the exome supported development: these flies ate less, developed faster and had a full lifespan, in comparison to fruit flies fed an isonitrogenous diet (Piper et al., 2017). The exome-matched profile of amino acids closely matched that of the composition of egg yolk proteins in fruit flies (Piper et al., 2017). This suggests that a diet that matches the profile of amino acids found in that species' eggs may serve as a high-quality diet for the species.

In most oviparous animals, egg production requires that the diet include the essential amino acids to build the egg yolk protein vitellin. In insects, vitellin typically makes up about half of the dry mass of eggs, with the other half of the egg being mostly lipid. During egg production, insects synthesize Vg (the precursor to egg yolk protein) in the fat body and secrete it into the hemolymph (Chapman, 1998). It is transported into developing oocytes and dehydrated to become vitellin (e.g. Roy et al., 2018). Our experimental approach was based on the demonstration that Drosophila melanogaster yolk proteins are the proteins that best match the fruit fly exome (Piper et al., 2017). Even though dipteran yolk proteins are not homologous to the Vgs conserved across most insects (Tufail and Takeda, 2008; Wu et al., 2013), the amino acids that compose the eggs still may support storage and reproduction for many oviparous animals.

In lubber grasshoppers, Romalea microptera (Palisot de Beauvois 1817) [=R. guttata (Houttuyn); hereafter ‘grasshoppers’] undergoing egg production, hexameric storage proteins and Vg can be quantified in serial hemolymph samples from single individuals throughout the oviposition cycle. Hexameric storage proteins steadily accumulate in the hemolymph until about 2 weeks before oviposition, then storage protein levels fall to an intermediate level at laying of the first clutch. Vg follows the same pattern (Hatle et al., 2001, 2003). Grasshoppers have two major hexamerins and a third with lower levels. The amino acid composition of these proteins is known (Hathaway et al., 2009) and is clearly distinct from the amino acid composition of Vg (see Table 1). The most abundant storage protein is hexamerin-90, which is an arylphorin with high levels of phenylalanine and tyrosine, and makes up about 50% of total hemolymph protein. The second most abundant storage protein is hexamerin-270, which is a methionine-rich storage protein. The third, hexamerin-500, makes up less than 10% of total grasshopper hexamerin.

Here, we hypothesized that diets with an amino acid composition that closely approximates that of egg yolk protein will be a high-quality dietary protein to support storage protein accumulation and oviposition. We designed a diet for grasshoppers based on the amino acid composition of their Vg (after Piper et al., 2017). This diet was compared with an isonitrogenous diet with a dissimilar amino acid profile from Vg. First, these diets were tested for reproductive development and adult somatic growth. Next, in these same individuals, we measured the developmental profiles of the hemolymph storage proteins hexamerin-90 and hexamerin-270. We predicted that the diet with amino acids matched to Vg would allow faster and greater egg production, as well as faster and greater accumulation of hemolymph storage proteins, compared with an isonitrogenous diet with a dissimilar composition of amino acids. Last, we tested how well the profiles of hexamerins explain reproductive parameters such as the time to oviposition.

Lubber grasshoppers were collected in Miami, FL, USA, as hatchlings and shipped to the lab in Jacksonville, FL, USA. They were reared en masse in 0.027 m³ cages at 24–32°C. Juveniles were fed fresh Romaine lettuce daily, occasionally supplemented with oatmeal, wheat germ, green beans, carrot tops or green onions. The day of molt to adult, each female was individually isolated into a 500 cm² ventilated cage and reared at 32°C:24°C on a corresponding 14 h light:10 h dark photoperiod.

For the first 2 days of adulthood, all individuals were fed lettuce ad libitum. Starting at about 1 week after adult molt, all individuals were provided with supplementary water, via a 50 ml plastic tube capped with moist cotton. Starting at day 2, each female was serially assigned to one of four diets. Each treatment group was defined by the solution the individual was force-fed daily and the amount of lettuce it was offered daily (adults require some fresh lettuce daily to thrive). These four groups were: Vg-balanced + 1 g lettuce (n=23), Unbalanced + 1 g lettuce (n=21), Buffer + ad libitum lettuce (n=22) and Buffer + 1 g lettuce (n=18) (Fig. 1; see ‘Preparation of Diets’ below for more details). All groups received ad libitum access to a dry diet with carbohydrates and vitamins, but no protein (i.e. a 0P:35C diet as per Simpson and Abisgold, 1985). The Buffer + ad libitum lettuce group, as the only group allowed free access to a protein source, served as a positive control for reproductive timing and output in well-fed animals. The Buffer + 1 g lettuce group, as the only group on limited lettuce and with no supplemental amino acids, served as a negative control. The Vg-balanced + 1 g lettuce and Unbalanced + 1 g lettuce groups were the isonitrogenous groups that this design was built to compare. That is, they were fed the same total moles of amino acid, but in different ratios of amino acids, to generate diets of different protein quality. This strategy allowed direct statistical comparison of the isonitrogenous groups.

The Vg-balanced (72.8 μl day⁻¹) + 1 g lettuce daily diet was designed to closely approximate the proportions of amino acids in grasshopper Vg (see below). Therefore, to design the composition of the Vg-balanced solution, we subtracted the amino acids in 1 g lettuce from the composition of Vg. This produced the amount of each amino acid that needed to be delivered via the Vg-balanced solution (see Table 1). Because of the low solubility of tyrosine, we substituted extra phenylalanine for tyrosine. The ultimate composition of the Vg-balanced + 1 g lettuce diet approximated the amino acid composition of Vg, especially closely for the essential amino acids. We did not allow any ad libitum feeding on protein sources; because of this, the protein quantities of these diets did not allow reproduction at the rate observed in the positive control group with ad libitum lettuce feeding.

The Unbalanced + 1 g lettuce daily diet provided nearly the same total moles of amino acids as the Vg-balanced + 1 g lettuce diet, but in proportions out of balance with the composition of Vg. Our hypothesis was that the Unbalanced + 1 g lettuce diet would be a low-quality protein for grasshoppers, and this is why the group was named ‘Unbalanced’. To make the Unbalanced solution, we identified the amino acid in highest concentration in the Vg-balanced solution (viz. serine) and assigned it the lowest concentration in the Unbalanced solution. Similarly, the amino acid in second highest concentration in the Vg-balanced solution (viz. arginine) was assigned the second lowest concentration in the Unbalanced solution. A few exceptions to this approach were made because of solubility challenges. As was done for the Vg-balanced solution, we substituted extra phenylalanine for tyrosine because of solubility challenges. The ultimate composition of the Unbalanced + 1 g lettuce diet was quite different from the composition of Vg, by having higher methionine, cysteine and tryptophan, but lower leucine and valine (Table 1).

The diet for the Vg-balanced + 1 g lettuce group was designed to closely approximate the amino acid profile of lubber grasshopper Vg, the precursor to egg yolk protein. To estimate the amino acid sequence of lubber grasshopper Vg, a partial transcriptome of the fat body and brain was obtained by RNAseq at the University of Florida Interdisciplinary Center for Biotechnology Research. Adult female, adult male, and juvenile grasshoppers were used (total n∼6) in an attempt to capture as many different transcripts as possible. RNA was isolated using an Invitrogen Ambion Pure Kit, followed by ribosome depletion. The RNAseq was run on an Illumina HiSeq2000, with 100 cycles per pair-end read. Raw reads were quality checked using FASTQC and cleaned using Trimmomatic. The cleaned reads were normalized and then assembled using Trinity, with default parameters. After omitting contigs shorter than 1000 bp, the partial transcriptome was complete.

A local BLAST of this partial transcriptome was run using Vgs from other Orthoptera as reference sequences (accession number KF171066.3v for Locusta migratoria and MK206970.1 for Schistocerca gregaria), and several contigs were identified. The contigs were confirmed to have similarity with Vgs of other insects using BLASTx. Using EMBOSS Transeq, the nucleotide sequences of these contigs were converted to all possible reading frames. Using NCBI ORF Finder, two sequences in the proper reading frame were identified. One was 750 bp (Vg ORF_8 #41798) and the other was 1537 bp (Vg ORF_30 #37292) (out of ∼9000 bp for the expected Vg mRNA sequence). These were tested for overlap using Clustal Omega and found not to overlap substantially, yet their amino acid compositions were similar. Therefore, the sequences of ORF_8 and ORF_30 were combined to make a single peptide of ∼760 amino acids. This was the partial sequence that was analyzed using EMBOSS Pepstats to estimate the mol% of each amino acid in lubber grasshopper Vg (see Table 1, Vg column). These mol% for essential amino acids in lubber grasshoppers averaged within 20% of those for L. migratoria and S. gregaria.

A quantitative means of comparing protein quality would be more precise than ‘high’ and ‘low’ quality. However, comparing the quality of proteins, each with numerous differences in amino acid content, can be difficult. Akiyama et al. (1997) developed a method for quantifying dietary protein quality for fish aquaculture. Their approach involved calculating the difference (i.e. Euclidian distance) between the amount (in grams) of each amino acid, converting these to absolute values, and then summing these absolute values. This approach inspired us to develop two potential indexes of protein quality.

Our second index used principle components analysis (PCA) to identify which diets are most similar, and via which amino acids. The first principal component was primarily loaded by valine, proline and leucine. The second principal component was primarily loaded by glutamic acid, glutamate and histidine.

These indexes were used to compare the diets and proteins, namely the Vg-balanced + 1 g lettuce diet, the Unbalanced + 1 g lettuce diet, Vg, hexamerin-270, hexamerin-90 and the three hexamerins added together. For the first index, each diet or protein was compared with Vg (see Table S1). These comparisons showed the Unbalanced + 1 g lettuce diet was at least 2-fold more different from Vg than any other diet or protein was from Vg. For the PCA index, to confirm its validity, using the PC1 and PC2 together to calculate Euclidean distance on the Cartesian coordinates, Vg was 6.86 times more similar to the Vg-balanced + 1 g lettuce diet than Vg was to the Unbalanced + 1 g lettuce diet. PC1 (see Fig. S1, x-axis) showed the Unbalanced + 1 g lettuce diet as very different from all other diets or proteins, and from lettuce. For PC2, lettuce separated from the other diets and proteins, which were all similar. Taken together, both indexes identified the Unbalanced + 1 g lettuce diet as most different, and the PCA analysis was better able to identify the amino acids responsible for this difference and to show that difference visually.

Diet solutions were prepared each week. Solutions of 40 ml were prepared in pH 5 PBS. Amino acids (all from Sigma-Aldrich Chemical, St Louis, MO, USA, or Thermo Fisher Scientific, Pittsburgh, PA, USA) were added one at a time, from least soluble to most soluble. Each was dissolved by heating to ∼70°C via stints of <15 s in a microwave oven, up to 45 s for non-polar amino acids. Vortex mixing was used as needed, usually <30 s. When all amino acids except methionine and cysteine had been added and were dissolved, the solution was cooled and then 3.5 ml was aliquoted into 20 ml glass vials. Each aliquot was stored and kept frozen at −20°C until the day of use. On the day of use, an aliquot was thawed, the methionine and cysteine were added, and the solution was heated and mixed as needed to dissolve all the amino acids. Before feeding, solutions were allowed to cool enough so that they were comfortable to touch.

Liquid solutions were force-fed to each grasshopper, both in the morning (four aliquots of 9.1 μl each) and in the afternoon (another four aliquots of 9.1 μl each), using a 10 μl pipette. Force-feeding involved gently pushing the tip of the pipette between the mandibles, and slowly ejecting the solution. Visual observation confirmed that grasshoppers reliably consumed all the solution, consistent with prior experiments with ¹³C-labeled nutrients (e.g. Hatle et al., 2019). Each grasshopper was weighed each morning. The solutions were kept at 4°C between the morning and afternoon feedings. During feeding, solutions were checked regularly for precipitation. If precipitation occurred, the solution was re-heated and mixed as needed to dissolve the amino acids; this was required most frequently for the Unbalanced solution. Lettuce was given in the morning.

An overview of the sample and data collection procedures for each individual is given in Fig. S2. Starting at 30 days of age, each individual was tested 3 times weekly for oviposition, by placing it in a cup nearly filled with moist sand (e.g. Drewry et al., 2011; Judd et al., 2011; Hatle et al., 2013; Tetlak et al., 2015). If an individual probed into the sand to lay, it was left until the next day, and skipped the afternoon feeding. Upon laying, the date of oviposition was recorded, and the number eggs was counted. A subsample of eggs was measured for length using a caliper. Two laid eggs from each individual were stored at −80°C for later analysis of protein and lipid content.

Either 2 or 3 days after oviposition, each individual was dissected to measure organ size. We measured fat body wet mass and approximate hemolymph volume as indexes of storage (Hatle et al., 2013), and ovary mass as an index of the preparation of the subsequent clutch.

Eggs were frozen and then freeze-dried using a FreeZone 2.5 l Benchtop Freeze Dryer (Labconco, Kansas City, MO, USA). Each egg was placed in a gel capsule with a pinprick on each end of the capsule for air and solvent flow. Each capsule was then placed within a 1.5 ml microcentrifuge tube. Tubes were placed in the freeze dryer, which was run at −50°C for 48 h. Dry mass was then recorded using an XP6 Microbalance (Mettler Toledo, Columbus, OH, USA), with each egg being individually weighed without its capsule.

Egg fat was extracted using a gravimetric Soxhlet method. Gel capsules containing eggs were punctured with a small metal wire, forming a string. These strings were then placed in a cellulose thimble that was loaded into the Soxhlet extractor, with petroleum ether as a solvent. Heat was applied to the distillation flask at 80°C, and tap water was run through the condenser at room temperature. The extractor was left to run for 3 days for each load of eggs. Afterward, eggs were removed from the device and placed in a desiccator until petroleum ether had evaporated out of the sample. Once dry, eggs were weighed again on the microbalance. The difference between the initial dry mass and the mass after lipid extraction is the mass of lipid in the egg.

To determine the protein content of each egg, extracted egg samples were bead homogenized using a Precellys Evolution Homogenizer (Berlin Instruments, Montigny-le-Bretonneux, France) with 10 small and 10 large zircon homogenization beads at a speed of 6800 rpm for 20 s in ice-cold PBS with 1% Halt™ Protease Inhibitor Cocktail (Thermo Fisher Scientific, Waltham, MA, USA). Once homogenized, the samples were centrifuged at 13,000 rpm for 30 min at 4°C. After centrifugation, the supernatant from each sample was collected and stored at −80°C. Later, samples were thawed, and water-soluble protein was measured using the Bradford (1976) dye assay. The process of Soxhlet extraction may have led to protein degradation in our samples, because egg protein context was much lower than expected based on previous work with eggs of other insects, so the protein measurements are relative between treatments, rather than absolute.

Hemolymph samples (5 μl) were collected from the oldest half of the individuals in the study. If a full 5 μl sample was not readily collected, the volume collected was recorded. We collected samples twice weekly, from ∼7 days after adult molt until oviposition, which is ∼85% of the oviposition cycle, so there was excellent coverage of the entire developmental period. Hemolymph samples were stored at −80°C until analysis for hexamerin storage proteins.

Hexamerins were analyzed by native polyacrylamide gel electrophoresis as described previously (Hathaway et al., 2009). Samples were thawed once to re-organize and then re-frozen before analysis. Hemolymph was diluted in PBS, which was mixed 1:1 with loading dye. The volumes of diluted hemolymph and loading dye were adjusted as needed so that each sample contained 0.25 μl of hemolymph equivalents in 15 μl. This mixture was loaded onto a 4–20% acrylamide pre-cast gel (Bio-Rad Mini-PROTEAN TGX, Hercules, CA, USA). Gels were run at 4°C for 8 h, then stained with Coomassie Blue. Gels were imaged using a mobile phone camera, and the images were analyzed using NIH ImageJ. Samples were re-frozen and stored for later analysis of Vg.

Vg was analyzed by enzyme-linked immunosorbent assay (ELISA) (Borst et al., 2000; Hatle et al., 2001). Each sample was tested at both 10 nl hemolymph equivalents and 1 nl hemolymph equivalents. If the results for both dilutions for an individual fitted the standard curve, they were averaged. The antibody was produced in the lab of Dr David Borst and is stored in the Hatle lab. It has been validated as a female-specific hemolymph protein by testing male hemolymph in the assay, which produces almost no response at 50 nl. In our assays, this Vg antibody was used at 1:6000 in PBS.

We calculated four parameters from each individual's profile for a specific protein (i.e. hexamerin-270, hexamerin-90 or Vg). For each individual, levels of each protein typically showed a steady increase to a maximum about 2 weeks before oviposition, and then a steady decline in levels to an intermediate level at oviposition, as in previous work (e.g. Hatle et al., 2001) (Fig. S3). The four parameters were: the area under the curve to age 55 days (AUC55), the area under the curve to age 35 days (AUC35), the individual's maximum level of the protein, and age at the individual's maximum level of the protein. Fifty-five days (i.e. AUC55) was used because that is the age by which most individuals in the study had oviposited. For AUC35, 35 days is justified by previous work on developmental profiles of hexamerins and Vg in these grasshoppers (Hatle et al., 2001). Individual grasshoppers attain their maximum level of these proteins ∼15 days before oviposition, regardless of the diet before or after the maximum. The Vg-balanced + 1 g lettuce group oviposited at a mean of 51.5 days. Because of this, after day ∼35, protein levels in Vg-balanced + 1 g lettuce grasshoppers would be expected to fall, while protein levels in Unbalanced + 1 g lettuce grasshoppers would be expected to continue rising. Because of this, comparing protein levels after day 35 obscures differences in early developmental accumulation of protein that might exist between groups.

Data were analyzed by grouping them into related response variables (i.e. oviposition data, somatic size data, egg composition data, hexamerin-270 parameters, hexamerin-90 parameters, Vg parameters). Each set of response variables was analyzed using a one-way MANOVA to test for overall significance, with REGWQ post hoc tests to determine the ranking of the treatments. Then, the Vg-balanced + 1 g lettuce group was compared with the Unbalanced + 1 g lettuce group using a one-tailed t-test with Holm's sequential correction (Holm, 1979; Abdi, 2010). This post hoc test is based on the a priori prediction that the Vg-balanced + 1 g lettuce group would have faster development, greater reproductive output, and more protein storage than would the isonitrogenous Unbalanced + 1 g lettuce group. The Buffer + ad libitum lettuce group was taken as a positive control, and the Buffer + 1 g lettuce group was taken as a negative control.

To examine the extent to which hemolymph hexamerin profiles are associated with variation in reproductive phenotypes, we competed 24 linear mixed models (Burnham and Anderson, 2020). All models included the date of a grasshopper's molt to adult as a random factor to account for possible variation across cohorts. Models with the lowest Akaike information criterion (AIC) values were selected as the best models, and in cases where multiple models were within 2 AIC units of each other, the model with fewer variables (more parsimonious) was selected as the best. In cases when models were within 2 AIC units and were equally parsimonious, models were averaged using the MuMIn package in R (https://CRAN.R-project.org/package=MuMIn). Our explanatory factors were a grasshopper's dietary treatment, time of maximum level of hexamerin, and AUC55 for each of the three hexamerins. We also included the total amino acids accumulated in all three hexamerins (i.e. AUC55 for hexamerin-90+AUC55 for hexamerin-270+AUC55 for hexamerin-500) as a proxy for total stored amino acids. We examined each response factor individually. Our response factors included a grasshopper's age at oviposition, number of eggs laid, egg size (egg dry mass), total egg mass laid (egg number×egg dry mass), total fat oviposited (egg number×egg fat content), and total protein oviposited (egg number×egg protein content). We checked for correlations between numeric explanatory factors using Person's r to verify that correlations between variables were not biasing our models. For post hoc analysis, we compared reproductive response variables by diet using Welch's t-tests. If another explanatory variable was present in the winning model that had an interaction with diet, we performed a Pearson's correlation test for each diet to illuminate the relationship between explanatory variable and reproductive response variable. The dataset and code used to model competition and post hoc analysis can be found in Supplementary Materials and Methods.

Reproductive development was strongly affected by the diet treatments (MANOVA, Pillai's Trace F_9,240=12.07; P<0.001; Fig. 2A,B). The Buffer + ad libitum lettuce group showed much greater reproductive development than any of the other three groups, with the earliest age of oviposition, the largest number of eggs and the largest ovary mass after oviposition (REGWQ post hoc tests); this was the only group allowed to feed ad libitum on a protein source to fully support reproduction. The Buffer + 1 g lettuce group had the oldest age of oviposition, the smallest number of eggs, and was tied with the Unbalanced + 1 g lettuce group for the smallest ovary mass after oviposition. Direct comparison of the isonitrogenous diet groups showed that the Vg-balanced + 1 g lettuce group oviposited at a younger age (P=0.004), laid more eggs (P<0.001) and had larger ovaries after oviposition (an index of preparation of the subsequent clutch; P<0.001) than the Unbalanced + 1 g lettuce group. All these one-tailed t-tests were significant upon Holm's correction (Holm, 1979; Abdi, 2010).

The size of somatic tissues after oviposition was generally greater in the Buffer + ad libitum lettuce group than in the other three groups (Pillai's Trace F_9,240=3.36; P<0.001; Fig. 2C,D). The Buffer + ad libitum lettuce group had greater somatic mass after oviposition than all three other groups (REGWQ post hoc tests). Fat body mass was greater in the Buffer + ad libitum lettuce group than in the Unbalanced + 1 g lettuce group, but all other comparisons were not detectably different. Last, approximate hemolymph volume was not different across any of the four treatment groups. Direct comparison of the isonitrogenous diet groups showed no differences between the Vg-balanced + 1 g lettuce group and the Unbalanced + 1 g lettuce group in any somatic tissues.

Egg composition was strongly affected by the diet treatments (MANOVA, Pillai's Trace F_9,231=11.51; P<0.001; Fig. 3). The Vg-balanced + 1 g lettuce group had larger eggs than the other three groups (REGWQ post hoc tests). Egg lipid was greater in the Unbalanced + 1 g lettuce group than in the Vg-balanced + 1 g lettuce and Buffer + ad libitum lettuce groups. Egg protein was lower in the Buffer + 1 g lettuce group than in the other three groups, while the Buffer + ad libitum lettuce group had higher egg protein than the Vg-balanced + 1 g lettuce group. Direct comparisons of the isonitrogenous diet groups showed that the Vg-balanced + 1 g lettuce group laid eggs that were larger (P<0.001), but with lower lipid (P=0.014) and lower protein (P=0.008) than the Unbalanced + 1 g lettuce group (all significant upon Holm's correction).

Parameters of the hexamerin-270 developmental profiles were strongly affected by diet (MANOVA; Pillai's Trace F_12,90=3.47; P<0.001; Fig. 4), with the significant effects due to AUC35 and the age at maximum hexamerin-270. AUC55 did not differ across the four diet treatments (REGWQ post hoc test). In contrast, the relative ranking of the AUC35 for hexamerin-270 across diets was Buffer + ad libitum lettuce=Vg-balanced + 1 g lettuce>Unbalanced + 1 g lettuce=Buffer + 1 g lettuce. The maximum level of hexamerin-270 for each individual did not differ across the diet groups. The relative ranking of the age at maximum hexamerin-270 level for each individual was Buffer + ad libitum lettuce=Vg-balanced + 1 g lettuce<Unbalanced + 1 g lettuce=Buffer + 1 g lettuce. Direct comparison of the isonitrogenous diet groups showed that the Vg-balanced + 1 g lettuce group had greater AUC35 (P=0.015) and an earlier age of maximum hexamerin-270 (P=0.011), which were both significant upon Holm's correction.

Parameters of the hexamerin-90 developmental profiles were affected by diet (MANOVA; Pillai's Trace F_12,93=2.93; P=0.002; Fig. 5). AUC55 did not differ across the four diet treatments (REGWQ post hoc test). By MANOVA, AUC35 for hexamerin-90 was greater in the Buffer + ad libitum lettuce group than in the Unbalanced + 1 g lettuce group and the Buffer + 1 g lettuce group. Similarly, AUC35 for the Vg-balanced + 1 g lettuce group was greater than that for the Buffer + 1 g lettuce group. All other comparisons of AUC35 were not significantly different. The maximum level of hexamerin-90 did not differ across the diet treatments. Last, the age at maximum hexamerin-90 level was earlier in the Buffer + ad libitum lettuce group than in the other three groups, with these other three groups all statistically similar. Direct comparison showed that the Vg-balanced + 1 g lettuce group had greater AUC35 (P=0.009) than the Unbalanced + 1 g lettuce group (significant upon Holm's correction). The three other parameters did not differ between the two isonitrogenous groups, but all trended toward greater accumulation of hexamerin-90 in the Vg-balanced + 1 g lettuce group (AUC55 P=0.045, maximum level P=0.027, and age at maximum level P=0.182; all non-significant after Holm's correction).

Hexamerin profiles only affected the timing of oviposition, and not other reproductive parameters. To explore the extent to which hemolymph hexamerin profiles are associated with variation in reproductive parameters, we competed a series of linear mixed models. These used the reproductive outputs of each grasshopper as response variables (grasshopper's age at oviposition, number of eggs laid, egg size, total egg mass laid, total fat oviposited and total protein oviposited). For explanatory variable, a combination of diet and hexamerin parameters was used. When explaining variation in ovipositional timing, our best-fit model included diet, AUC55 for hexamerin-270, the interaction of diet and AUC55 for hexamerin-270, total amino acids accumulated in all hexamerins, the interaction of diet and total amino acids accumulated in all hexamerins, and the timing of the hexamerin-90 maximum (Table 2). In post hoc analysis of the relationship between these explanatory variables and oviposition timing, ovipositional timing varied significantly among all four experimental diets upon post hoc analysis (Table 2; Welch's t-test, d.f.>11.6, t<−2.86, P<0.0126 for all comparisons). The timing of the maximum hexamerin-90 level was strongly correlated with ovipositional timing (Pearson's r, d.f.=33, r=0.80, t=7.66, P<0.001).

Because we detected an interaction between diet and AUC55 for hexamerin-270, we also examined the correlations between the AUC55 for hexamerin-270 and timing of oviposition for each diet. Only for grasshoppers fed a Vg-balanced + 1 g lettuce diet did AUC55 for hexamerin-270 significantly correlate with ovipositional timing (Pearson's r, d.f.=6, r=0.80, t=3.22, P=0.018). For all other diets there was no semblance of a correlation between AUC55 for hexamerin-270 and the timing of oviposition (Table 2; Pearson's r, d.f.=[6–8], |r|<0.17, t=[−0.444–0.0424], P>0.67 for all other diets). Similarly, total amino acids accumulated in all hexamerins trended toward significant explanation of variation in oviposition timing only in the grasshoppers fed the Vg-balanced + 1 g lettuce (Pearson's r, d.f.=6, r=0.68, t=2.24, P=0.066). For all other diets, there was no association between the total amino acids accumulated in all hexamerins and the timing of oviposition (Pearson's r, d.f.=[6–8], |r|<0.26, t=[−0.744–−0.348], P>0.48).

Parameters of the Vg developmental profiles were strongly affected by diet (MANOVA; Pillai's Trace F_12,93=3.83; P<0.001; Fig. 6). MANOVA indicated that AUC55, AUC35 and maximum level of Vg were greater in the Buffer + ad libitum lettuce group than in the other three groups, which did not differ from each other (REGWQ post hoc test). The age at maximum Vg level varied such that Buffer + ad libitum lettuce<Vg-balanced + 1 g lettuce<Unbalanced + 1 g lettuce=Buffer + 1 g lettuce. Direct comparison of the isonitrogenous diet groups showed no significant difference for any of the four parameters for Vg. While not significant, all four parameters of Vg profiles trended toward greater accumulation in the Vg-balanced + 1 g lettuce group than in the Unbalanced + 1 g lettuce group.

Last, by competing models to explain variation in AUC55 for Vg, our best model included diet, total AUC55 for hexamerin-270 and total amino acids accumulated in all hexamerins (Table 3). In post hoc analysis, AUC55 for Vg significantly correlated with both AUC55 for hexamerin-270 (Pearson's r, d.f.=30, r=0.44, t=2.69, P=0.0112) and total amino acids accumulated in all hexamerins (Pearson's r, d.f.=30, r=0.51, t=3.27, P=0.00272).

The results of this study support our hypothesis that a diet with an amino acid composition that matches Vg is a high-quality protein diet that facilitates reproduction, in comparison to a diet where the composition of amino acids is unbalanced compared with Vg. This fine-scale difference in diet composition, namely the composition of amino acids but not the total quantity of amino acids, produced clear differences in reproductive output (16% more eggs laid 11% faster, and 58% larger ovaries developing for the next clutch). Only one previous study has shown increased reproduction in insects on high-quality dietary protein in comparison to isonitrogenous controls (Ma et al., 2020). However, that previous study did not examine storage.

Perhaps even more intriguing and novel, the Vg-balanced + 1 g lettuce diet permitted greater accumulation of hemolymph storage proteins than did the isonitrogenous Unbalanced + 1 g lettuce diet. This was true even though the Vg-balanced + 1 g lettuce diet was not a better match for the composition of amino acids required to build the hexamerins than was the Unbalanced + 1 g lettuce diet (see next paragraph). This serves as evidence that the composition of amino acids in the Vg-balanced + 1 g lettuce diet may better support multiple aspects of adult physiology, beyond just the production of Vg. Our data substantially extend the pattern previously shown for fruit flies (that diets matching the species-specific amino acids needed to build eggs can improve reproduction; Ma et al., 2020), to an additional physiological level, namely to accumulation of hemolymph proteins serially sampled from individuals through their first clutch cycle. Our work shows how dietary amino acid balance (not concentration) can impact protein storage that ultimately fuels reproductive allocation.

Even though storage proteins accumulated to a greater extent in the Vg-balanced + 1 g lettuce group, this diet is not a better match for the composition of amino acids for building these storage proteins, in comparison to the Unbalanced + 1 g lettuce diet. The methionine+cysteine in our Vg-balanced + 1 g lettuce diet is only 34% of that in hexamerin-270 (a methionine-rich storage protein). Further, this 34% for methionine+cysteine is lower than the most poorly represented essential amino acids in the Unbalanced + 1 g lettuce diet, which are arginine at 42% of hexamerin-270 and leucine at 45% of hexamerin-270 (Table 1). Therefore, accumulation of hexamerin-270 ostensibly represents a developmental event that should require materials that are supplied at least as well by the Unbalanced + 1 g lettuce diet as by the Vg-balanced + 1 g lettuce diet. Despite this, cumulative levels of hexamerin-270 were 80% greater in the Vg-balanced + 1 g lettuce group than in the Unbalanced + 1 g lettuce group.

Multiple components of complex developmental processes occur simultaneously. Therefore, the nutritional need to support these overlapping processes in the animal can occur simultaneously, and so they may compete. In our grasshoppers undergoing egg production, the timing of the production of hexamerins overlaps with the production of egg yolk proteins (i.e. Vg plus vitellin; Hatle et al., 2001). Hexamerin production occurs slightly earlier than Vg production, but the magnitude of Vg production is much larger. Because grasshoppers in the Unbalanced + 1 g lettuce group did not have the most closely matched ratio of amino acids for making egg yolk protein, this potentially could create competition between production of egg yolk proteins and production of hexamerins. That is, the lower cumulative levels of hexamerins in Unbalanced + 1 g lettuce grasshoppers may be an indirect effect of not having sufficient substrate to meet a simultaneous, more demanding process, namely production of Vg and vitellin. Whether the effect is direct or indirect, our data suggest that a diet lacking the proper species-specific composition of amino acids for reproduction can attenuate storage of amino acids.

Significant differences in the accumulation of storage proteins were detected between our two focal treatment groups, despite high variability between individuals within each group. For example, AUC35 (area under the curve, showing total cumulative protein up to age 35 days) for hexamerin-270 was 80% greater in the Vg-balanced + 1 g lettuce group than in the isonitrogenous group, which at the same time showed high inter-individual variability (s.e. of 22% for the Unbalanced + 1 g lettuce group). Previous studies on grasshopper hemolymph proteins (and hormones) also showed high inter-individual variability in the magnitude of hemolymph proteins, but great consistency in the maximum timing of those same proteins (Hatle et al., 2001). That is, all individuals showed hemolymph protein levels that rose steadily until a peak, and then levels fell for the final 2 weeks before oviposition, ending at an intermediate level. The absolute levels of those proteins, however, can differ greatly from one individual to another. Consistent with previous studies, in the present study the timing of the maximum levels was relatively consistent (e.g. hexamerin-270 had a s.e. of 10% for maximum age for the Vg-balanced + 1 g lettuce group). Also consistent with previous studies, the cumulative levels of these proteins varied at least 5-fold (e.g. hexamerin-270 had a s.e. of 66% for AUC35 for the Unbalanced + 1 g lettuce group). That our current study shows differences in cumulative levels of storage proteins, despite the high variability in levels of storage proteins across individuals within a diet group, suggests large changes in the commitment to storage by the individual animals.

Studies on dietary protein quality have mostly utilized diets low in a single amino acid (e.g. Srivastava and Auclair, 1975; Orentreich et al., 1993; Miller et al., 2005; Arganda et al., 2017; Csata et al., 2020; Yu et al., 2021). Such diets may alter development and even extend lifespan, but they are unlikely to result in vitality, full reproduction and full lifespan. Fewer studies have sought to test development on a diet with high-quality protein of known amino acid composition (e.g. Piper et al., 2014). Work with fruit flies has shown that reducing dietary levels of any essential amino acids reduces reproduction and extends lifespan similarly (Juricic et al., 2020). No previous study on dietary protein quality has measured protein storage throughout the developmental period, as done in the present study.

How is a high-quality protein diet identified? Our paper provides support for the Piper et al. (2017) approach to identifying the amino acids composing a high-quality protein for a specific species; namely, matching the dietary amino acids to the amino acids in eggs. Similar approaches have long been used in fish aquaculture, matching dietary amino acids to the somatic amino acid content (reviewed in Akiyama et al., 1997). The various levels of the 10 essential amino acids (including arginine for insects) could potentially be distributed in countless ways in complex diets. Classic studies on growth and nitrogen balance (i.e. measuring nitrogen consumed, growth and nitrogen excreted) in rodents and fishes have identified the amino acids required for a high-quality protein for these animal groups (e.g. Pencharz and Ball, 2003; Gurure et al., 2007). For other animal groups, there is much less information on what constitutes a high-quality protein diet. Our results show that matching the amino acid proportions of egg protein identifies a high-quality protein for grasshoppers.

Lubber grasshopper Vg has a similar amino acid profile to the amino acids coded for by the codon usage frequency for insects (https://www.genscript.com/tools/codon-frequency-table). That is, the profile of essential amino acids in lubber grasshopper Vg is similar to the profiles of amino acids represented by the usage frequency of codons across all insects (https://www.genscript.com/tools/codon-frequency-table; presumably largely somatic genes). The Vg-balanced + 1 g lettuce diet is 27% lower in lysine than is the amino acid profile from insect codon frequency, and no other essential amino acids are lower. These data suggest that the amino acids needed for Vg, or even the soma, may predict the amino acid composition of a high-quality diet.

How can the quality of proteins be quantified? Akiyama et al. (1997) created an index for quantifying the similarity of dietary protein and somatic protein across several fishes, which was successful in demonstrating that the composition of optimal dietary proteins varied more across species than did the composition of somatic proteins. We used a similar approach to Akiyama et al. (1997) to create an index to compare the dietary proteins and grasshopper proteins in this study (Supplementary Materials and Methods and Table S1). Further, we also used PCA to compare the dietary proteins and grasshopper proteins in this study (Fig. S1). We conclude that PCA is the better tool for comparing the relative quality of amino acid composition across proteins.

Some previous studies have shown that high dietary methionine can abrogate the health benefits of life-extending dietary restriction, in both flies and mice (Hine et al., 2015). It may be that the high levels of methionine in our Unbalanced + 1 g lettuce diet, selected because this amino acid is low in Vg, are partly responsible for the poorer physiological performance of this group. We hypothesize that animals on diets like the Unbalanced + 1 g lettuce diet used here, but with levels of methionine and cysteine that match the Vg-balanced + 1 g lettuce diet, will develop faster than animals on diets with high methionine, but not as fast as animals on high-quality protein diets.

The composition of amino acids in the Vg-balanced + 1 g lettuce diet is somewhat different from the composition of Romaine lettuce (the lab diet for grasshoppers; see Fig. S1). The relative balance of essential amino acids in Romaine lettuce is somewhat similar to that for egg yolk protein, with tryptophan at half the level found in Vg being the most deficient essential amino acid. In contrast, the levels of non-essential amino acids in Romaine lettuce do not closely match those in Vg. Romaine lettuce is very high in the non-essential amino acids aspartic acid, glutamic acid and glycine (Table 1). This is an example of animal diets not matching their dietary needs (Simpson et al., 2017).

We detected that AUC55 for hexamerin-270 (i.e. total cumulative amount of the storage protein through the entire egg production cycle) was a significant explanatory variable when examining ovipositional timing. In post hoc analysis, AUC55 for hexamerin-270 was tightly correlated with ovipositional timing in the Vg-balanced + 1 g lettuce group, but not the other dietary treatments. This result was surprising to us, and it does not fit with our standing hypothesis that a threshold level of hexamerin(s) initiates the commitment to oviposit (Juliano et al., 2004). The present data are consistent with a model in which: (1) hexamerin-270 accumulates continuously in the Vg-balanced + 1 g lettuce group until vitellogenesis, but in contrast hexamerin-270 accumulates sporadically in the other three groups; and (2) hexamerin-270 titer does not accelerate ovipositional timing. This would be consistent with the correlation between AUC55 for hexamerin-270 and age at oviposition only in the Vg-balanced + 1 g lettuce group. To address our hypothesis that a threshold level of hexamerin(s) initiates a commitment to oviposition, we need more frequent hexamerin measures around the time of commitment to oviposition. We estimated the age at commitment to oviposition by plotting the mass gain of each group (Fig. S4). From this plot, we estimated 40% mass gain as a threshold after which there is a commitment to oviposition. Indeed, the treatment groups differed in time from adult molt to 40% mass gain, but they did not differ in time from 40% mass gain to oviposition (see Fig. S5 legend for further details). More rigorous determination of a mass gain threshold will require a ‘broken stick’ model or experiments in which hexamerins are manipulated (e.g. by RNAi, as in Tetlak et al., 2015).

Somatic mass did not differ among the two treatment groups and one control group fed 1 g lettuce daily. Piper et al. (2017) used the fly exome to identify a high-quality protein for flies. Codon usage is similar to the exome. That is, the entire genome (the encoding portion of which is mostly somatic) has a similar profile of essential amino acids to that of reproductive tissues. This may help explain why the Unbalanced + 1 g lettuce diet was unable to support somatic storage any better than did the isonitrogenous diet, as it was not providing the correct balance of amino acids for either type of development.

Egg quality was the only developmental parameter that was better in the Unbalanced + 1 g lettuce group than in the Vg-balanced + 1 g lettuce group. Eggs from Unbalanced + 1 g lettuce grasshoppers were 15% higher in fat and 12% higher in protein, despite being 11% lighter, than eggs from the isonitrogenous Vg-balanced + 1 g lettuce group. It may be that low-quality protein diets are leading to the production of fewer but better eggs, and more slowly. This could be a successful plastic tactic for adjusting reproductive effort to poor quality environments, by better preparing the offspring (e.g. Harvey and Orbidans, 2011). We did not test viability of the eggs, which is a better metric of egg quality than egg composition alone. Future studies may benefit from assessing the effects of dietary protein quality on offspring number and survival.

In this present study, we have identified that the amino acid composition of the major protein for reproductive provisioning (Vg) provides a dietary amino acid balance that improves protein storage and reproduction, in comparison to an isonitrogenous diet. Future studies can test health on such diets by measuring protein carbonylation, lipid peroxidation, DNA double-strand breaks, telomere length and DNA methylation patterns, all of which are associated with chronic disease or lifespan.

We thank Juana Zargon for feeding grasshoppers, Thomas H. Q. Powell for the initial transcriptome alignment, Phil Hahn for valuable advice on running statistics, Emma Kordek for help with the indexes of protein quality, and two anonymous reviewers for their suggestions.