Changes in composition and levels of hemolymph proteins during metamorphosis of Manduca sexta

Abstract

The tobacco hornworm, Manduca sexta, is a lepidopteran model species widely used to study insect biochemical processes. Some of its larval hemolymph proteins are well studied, and a detailed proteomic analysis of larval plasma proteins became available in 2016, revealing features such as correlation with transcriptome data, formation of immune complexes, and constitution of an immune signaling system in hemolymph. It is unclear how the composition of these proteins may change in other developmental stages. In this paper, we report the proteomes of cell-free hemolymph from prepupae, pupae on day 4 and day 13, and young adults. Of the 1,824 proteins identified, 907 have a signal peptide and 410 are related to immunity. Drastic changes in abundance of the storage proteins, lipophorins and vitellogenin, for instance, reflect physiological differences among prepupae, pupae, and adults. Considerably more proteins lacking signal peptide are present in the late pupae, suggesting that plasma contains relatively low concentrations of intracellular components released from remodeling tissues during metamorphosis. The defense proteins detected include 43 serine proteases and 11 serine protease homologs. Some of these proteins are members of the extracellular immune signaling network found in feeding larvae, and others may play additional roles and hence confer new features in the later life stages. In summary, the proteins and their levels revealed in this study, together with their transcriptome data, are expected to stimulate focused explorations of humoral immunity and other physiological systems in wandering larvae, pupae, and adults of M. sexta and shed light upon functional and comparative genomic research in other holometabolous insects.

Graphical Abstract

1. Introduction

Cell-free hemolymph (i.e. plasma) of invertebrate animals bathes tissues, stores and transfers carbohydrates, lipids and proteins, and allows physiological processes to occur in the circulation (Kanost, 2009). Certain plasma proteins are responsible for nutrient storage and transport. For instance, hexamerins act in storage of amino acids for metamorphosis (Burmester, 1999; Telfer and Pan, 2003); vitellogenins synthesized in fat body enter ovaries to support embryonic development (Valle, 1993); lipophorins transport lipids that are hydrolyzed by lipases and oxidized in mitochondria to produce energy for flight (Arrese and Soulages, 2010). In addition, some plasma proteins cooperate with hemocytes to mount a local, graded defense against wounding and invading organisms (Jiang et al., 2010; Lemaitre and Hoffmann, 2007; Park et al., 2010; Strand, 2008). The surveillance system recognizes polysaccharides on the surface of bacteria, fungi or parasites, and relays the signal of tissue damage or microbial invasion. Other plasma proteins then function as effectors to stop bleeding and kill pathogens or as regulators to modulate the potency and duration of defense reactions.

We have investigated innate immunity of Manduca sexta (Cao et al., 2017; He et al., 2016; Hu et al., 2019; Kanost et al., 2016; Wang et al., 2020), a biochemical model representing lepidopteran insects, including many serious agricultural pests. Our research has mainly focused on serine proteases (SPs), non-catalytic SP homologs (SPHs), and serpins in hemolymph of the 5^th instar feeding larvae (Cao et al., 2015b; Li et al., 2018). Analogous to human blood coagulation and complement activation, some insect defense responses to wounds or invading microbes are mediated by a serine protease system regulated by serpins in plasma. Pattern recognition receptors (PRRs), SPs, SPHs, phenoloxidases, and antimicrobial peptides form high M_r complexes during defense responses (Zou and Jiang, 2005; He et al., 2016; Lu et al., 2008). While biochemical studies revealed the system constituents and their interactions, we also took advantage of the genome information to explore hemolymph proteomes of the larvae injected with buffer or a mixture of killed bacteria (He et al., 2016). Comparison of changes in protein and mRNA levels revealed a close correlation (0.75) for immunity-related genes expressed in fat body, a major source of hemolymph proteins. Comparative analyses of SPs, SPHs, and serpins have facilitated their function prediction and validation in holometabolous insects (Cao and Jiang, 2018; Meekins et al., 2017).

In this study, we extended the Manduca proteomics research to pharate, early, and late pupae, and adults to test a premise that the immune system varies in different developmental stages in terms of plasma protein composition, level, and activation state. Selection of the pupal and adult stages is based on the following reasons: 1) the pharate pupal plasma contains active SPs and SPHs for melanization around the larval-pupal molt (Jiang et al., 2003; Wang and Jiang, 2004), 2) infection-independent induction of immunity-related genes occurs in wandering larvae and early pupae (He et al., 2015; Zhang et al., 2015), 3) tissue remodeling is active in pupae during metamorphosis (Cao and Jiang, 2017; Kanost et al., 2016), 4) a proteome of the adult hemolymph forms a basis for reconstituting the extracellular signaling system, perhaps more similar to those in adult mosquitoes and Drosophila, the stage with extensive study of immunity in these dipterans, 5) identification of other plasma proteins in these stages may provide a suitable background for understanding features of physiology during metamorphosis and differences between larval and adult plasma proteomes, and 6) discrepancies between theoretical M_r’s of proteins and their gel mobilities should reveal immunity-related post-translational modifications (e.g. proteolysis) and protein interactions (He et al., 2017; He et al., 2016).

2. Methods and materials

2.1. Insect rearing and preparation of cell-free hemolymph samples

M. sexta eggs were purchased from Carolina Biological Supply, and larvae were reared on an artificial diet (Dunn and Drake, 1983). Pharate pupae with metathoracic brown bars (B), day 4 pupae (P4), day 13 pupae (P13), and day 1 adults (A) were chilled on ice for 10 min and dissected for hemolymph collection into separate tubes. Each tube contained a few crystals of p-aminobenzamidine and 1-phenyl-2-thiourea to block proteolysis and melanization. After centrifugation at 4000×g for 5 min at 4 °C, the supernatants were transferred to clean tubes, and equal volumes of the plasma samples from 3-4 insects at the same life stage were pooled. Two additional biological replicates at this stage were prepared similarly from 6-8 other insects. A total of three samples from each of the four developmental stages were stored at −80 °C.

2.2. SDS-PAGE separation, sample preparation, and LC-MS/MS analysis

The plasma samples (40 μl each, 1:5 diluted) were mixed with 6×SDS sample buffer (8 μl) and heated at 95 °C for 5 min. The samples equivalent to 1.0 μl B, P4 and P13, and 1.5 μl A cell-free hemolymph were loaded onto 4–15% precast gradient gels (Bio-Rad) and run along with M_r markers at constant current of 30 mA for 45 min. After brief staining with Coomassie blue, the gels were destained (Fig. S1), and each lane was cut into eight slices based on the banding pattern (Fig. 1). The 96 gel slices were separately diced, destained, and reduced with 10 mM tris(2-carboxyethyl)phosphine (He et al., 2016). Cys residues were then alkylated for 1 h at 25°C with 55 mM iodoacetamide. After rinsing, the gel pieces were dehydrated and then infiltrated with sequencing grade trypsin (Promega, 8 μg/ml). After digestion for 16 h at 37°C and extraction, tryptic peptides were dissolved in 50 μl of 0.1% formic acid in water (mobile phase A). Samples B-3 (Bar stage gel slice 3, same below), P4-3, P13-3, A-1, A-3, A-7, and A-8 were diluted 1:20 in the mobile phase to avoid overloading. Each of the 96 samples (10 μl) was injected onto a 75 μm×50 cm Acclaim PepMap RSLC C18 column (Thermo Fisher) using a vented trap-column. Peptides were separated via a gradient of 2–40% mobile phase B (0.1% HCOOH in 80% acetonitrile) developed in 80 min. Peptides were ionized in a Nanospray Flex ion source and analyzed by MS and MS/MS in an Orbitrap Fusion quadrupole mass spectrometer (Thermo Fisher) in OSU DNA/Protein Resource Facility. The Orbitrap Fusion analyses employed a “Top Speed” data-dependent MS/MS strategy, wherein survey scans were performed in FT mode (R = 120,000) at least every three seconds. In the interval between survey scans, up to 20 data-dependent MS/MS scans were conducted in the ion trap. The Fusion scan settings also included selection of precursor ions via the quadrupole mass filter, dynamic exclusion (n = 1), and monoisotopic precursor selection.

Fig. 1.

Overview of the proteomic analysis of cell-free hemolymph from M. sexta prepupae, pupae, and adults. A) Scheme of sample preparation and analysis. Twelve plasma samples from insects in the four stages (B, P4, P13, and A) were separated by SDS-PAGE on two gradient gels, which were later cut into lanes A–C (Fig. S1). Each lane was cut into eight slices (1–8) and each slice into pieces for in-gel trypsinolysis. The resulting peptides were extracted from each of the 96 tubes and subjected to LC-MS/MS separation, database search, and data analyses. B) Venn diagrams of the proteins identified in the four life stages. Proteins with or without a predicted signal peptide are shown on the left and right, respectively. C) GO categorization of the 1,824 identified proteins with names and protein counts shown for the major groups. The numbers in parentheses are for proteins lacking signal peptide.

2.3. Protein identification and quantification

For peptide identification, a database was assembled using M. sexta OGS 2.0 (Kanost et al., 2016) and MCOT 1.0 (Cao and Jiang, 2015) gene models. MaxQuant (version 1.6.0.16) (Cox and Mann, 2008) was used to process raw mass data using the default settings, with variable Cys modifications by acrylamide or iodoacetamide, Met oxidation, amino-terminal formylation or acetylation, and Gln cyclization to pyroGlu. Protein groups were assembled using principles of parsimony. The search parameters were: no charge state deconvolution or deisotoping, trypsin digestion, maximum missed cleavage of two, parental ion mass tolerance of 20 PPM, and fragment ion mass tolerance of 0.50 Da. False discovery rates were set at 1%, and proteins were considered as identified with one or more unique peptides. A name including life stage (B, P4, P13, or A), gel slice (1, 2, 3, 4, 5, 6, 7, or 8), and sample biological replicate ID (A, B, or C) was assigned to each of the 96 gel fractions. A protein’s intensity-based absolute quantification (iBAQ) in a gel lane was the sum of iBAQ values in slices from that lane where the protein was detected. In calculation, a dilution factor of 20 was taken into consideration for the diluted samples (e.g. P4-3-B, A-7-C). The iBAQ sum for a protein in each biological replicate or gel lane was divided by the total iBAQ of all non-contaminant, non-decoy proteins from that lane and then multiplied 1×10⁶ to obtain its parts-per-million (PPM) value.

2.4. Gene expression profiling, annotation, and categorization

Data from fifty-two cDNA libraries sequenced by Illumina technology, each representing a sample of whole larvae, body parts, or tissues in various life stages (Kanost et al., 2016), were used to analyze mRNA levels. The numbers of reads mapped onto each OGS 2.0 or MCOT 1.0 gene model identified in this study were used to calculate transcripts per kilobase million (TPM) values in these libraries using RSEM (version 1.2.12) (Li and Dewey, 2011). TPM values of proteins in the “Majority protein IDs” cell reported by MaxQuant were added together to represent the RNA level. Since fat body is a major source of hemolymph proteins, the PPM values of B, P4, P13, and A were compared with the TPM values of this tissue from feeding stage larvae and wandering larvae, early and late pupae, and young and old adults. Heatmaps were plotted using Pheatmap (version 1.0.12) in the R package (Kolde, 2012). Z-scores were calculated from the protein PPM values.

After removal of contaminants and decoys, the column of “Protein IDs” contained multiple IDs for a single gene in most cases, which cannot be differentiated by the identified peptides. The preferred IDs were selected for those with highest peptide counts, with a signal peptide, and with an OGS 2.0 ID. A note of “MCOT only” was added if all its related IDs in the “Protein IDs” column were from MCOT 1.0. Likewise, a note of “immunity-related” was added based on the previous studies (Cao et al., 2015a; Cao et al., 2015b; He et al., 2015; Rao et al., 2015; Zhang et al., 2015). The gene ontology annotation was performed using BLAST2GO (Conesa et al., 2005). GO enrichment analyses were performed using GOATools (Klopfenstein et al., 2018), with all the identified proteins as background group. Genes were named on the basis of their annotations in the Manduca genome project, comparisons with the models predicted by NCBI, and BLAST searches of homologous genes in the NCBI nr database. Signal peptides were predicted by SignalP 4.0 (Petersen et al., 2011) and Signal-3L (Shen and Chou, 2007). The sequences for some genes were manually improved as previously described (Miao et al., 2020).

3. Results

3.1. Overview of the experimental design and identified proteins

Along with major morphological changes during metamorphosis, composition and content of hemolymph proteins in holometabolous insects change greatly in response to altered physiological needs, for instance, for flight and reproduction in adults (Kanost et al., 1990). To better understand the developmental differences at the proteome level, we examined plasma samples from bar-stage pharate pupae (B), pupae at days 4 and 13 (P4, P13), and day 1 adults (A) of M. sexta. As illustrated in Fig. 1A and Fig. S1, we separated these samples (4 stages, 3 biological replicates each) by SDS-PAGE, cut the stained gels into 12 lanes, and then 96 slices (8 per lane) based on the banding patterns. This pre-fractionation step separated abundant proteins in gel slices 1, 3 and 7 from scarce ones in the other five slices, reduced shielding of weak signals from the latter, and hence increased the total number of identified proteins. We adopted a widely used method of intensity-based absolute quantification (iBAQ) (Schubert et al., 2017; Schwanhausser et al., 2011; Webb-Robertson et al., 2015) to quantify levels of the identified proteins using MaxQuant. The total intensities and iBAQ values in each gel slice were similar among the biological replicates (data not shown), which supports a reliable estimation of molar amounts of individual proteins in terms of PPMs of the total in one sample (Krey et al., 2018) for data comparison among the B, P4, P13, and A groups.

We identified a total of 1,824 proteins in the cell-free hemolymph samples from insects in the four stages, compared with 654 proteins we previously identified in the plasma of feeding stage fifth instar larvae (He et al., 2016). This increase is likely due in part to higher sensitivity of the Orbitrap Fusion system. 150 proteins were found only in the MCOT 1.0 gene set developed by integrated gene modeling (Cao and Jiang, 2015), whereas the other 1,674 were identified in both MCOT and OGS assemblies, consistent with previous finding that the MCOT gene set for M. sexta is more inclusive than OGS 2.0. We detected more proteins lacking a secretion signal peptide in the late pupae (854) than in other stages (Fig. 1B, right), but the differences in number of proteins with a signal peptide were less pronounced (Fig. 1B, left). We hypothesize that cell lysis in tissue remodeling during metamorphosis likely results in the release of intracellular proteins into the plasma of the late pupal samples. Nearly half (907) of the total proteins have a secretion signal, and 525 or 58% of these are present in all four stages examined and may perform functions in hemolymph common to all of these developmental stages. The other 917 proteins detected lack a signal peptide but, in each gel slice, their total mass (estimated by peptide intensity in MS) was only 0.04–9.7% (average: 2%) of the total mass of all proteins, confirming that secreted proteins are predominant in the plasma samples (Fig. 2A). Two proteins without a signal peptide known to be present in plasma are prophenoloxidase-1 and −2 (proPO1 and proPO2), released from ruptured oenocytoids (Jiang et al., 1997; Kanost and Gorman, 2008). Some of the other proteins lacking a signal peptide may be due to gene annotation errors that miss the presence of a signal peptide (e.g. Msex2.09436-RB) or unconventional secretion pathways (Nickel, 2010; Sitia and Rubartelli, 2020), but most of these may be due to cell lysis during tissue remodeling in the late pupal stage and subsequently cleared from plasma.

Fig. 2.

Counts, contents, and compositions of the proteins identified in individual gel slices. A) Counts (left y-axis) of proteins without (dark green bar) or with (light green bar) a predicted signal peptide and intensity content (right y-axis) of the proteins lacking signal peptide (black open bar) in each gel slice of different samples. Error bars represent standard deviation of the intensity contents of three biological replicates. A protein is considered as identified in a gel fraction if at least one unique peptide is detected among the replicates. B) Intensity contents of the most abundant proteins in each gel slice. For each fraction of a developmental sample, average intensity values of proteins from the three biological replicates were calculated as percentages. A protein is plotted if its percentage is >1% and ranked either as top-five in current fraction or greater than 5% of total intensity of any other fractions. INCYNB2, inseticyanin-B2; GP27, 27 kDa glycoprotein; AKR2E4-1, aldo-keto reductase-2 E4-1; SOD, Cu-Zn superoxide dismutase; MD2, myeloid differentiation-2. Abbreviations of the other proteins are listed in Table S2 or defined in the text.

A GO analysis of the 1,824 proteins revealed that 1,349 have GO annotations, including 757 of the proteins lacking a signal peptide and 592 secreted proteins (Fig. 1C). Major groups under “molecular function” include 272 secreted proteins and 390 intracellular proteins with catalytic activity, 256 secreted proteins and 407 intracellular proteins with binding activity, 151 secreted proteins and 177 intracellular proteins associated with cellular anatomical entities under “cellular component”, 102 secreted proteins and 324 intracellular proteins annotated as participating in cellular processes, and 248 secreted proteins and 411 intracellular proteins in metabolic processes under “biological process”. Because 62% of the GO annotations in Fig. 1C are from the 917 signal peptide-free proteins and 38% are from the 907 signal peptide-containing proteins, the GO analysis is biased towards intracellular roles of the scarce proteins. The observation that 315 of the 907 secreted proteins lack GO terms suggests that about a third of the secreted plasma proteins detected have unclear functions, with a need for research to reveal their roles in insect biology.

3.2. Changes in composition of abundant hemolymph proteins during development

Because an abundant protein is often detected in multiple gel slices with size ranges distinct from its calculated M_r, we used the sum of its iBAQ values from all the slices containing that protein to calculate its molar proportion of the total iBAQ of all proteins identified in the entire lane, to obtain PPM values for comparisons between stages. We selected proteins with PPM >1000 in at least one sample as highly abundant proteins. After removal of redundancy, a list of 98 abundant proteins were identified in the four stages (Table S1). To display their distributions in the 32 gel slices, we plotted intensity percentages of the 31 most abundant proteins in each gel slice (Fig. 2B). The only proteins in Fig. 2B that lack a signal peptide were proPO1 and proPO2.

The most dramatic changes in protein abundance occurred in the groups of hexamerins and lipid transport proteins. The first group includes arylphorin-α and -β, Met-rich and moderately Met-rich storage proteins (MRSPs) (Burmester, 2015), whose levels in adults decreased to 0.05–0.09% of their averages in B, P4, and P13 (Fig. 3A). Since production of the arylphorins and MRSPs ceases around pupation (Hiruma and Riddiford, 2010; Burmester, 2015) but the protein levels remain the same from B to P14, their disappearance in adults is consistent with their regulated uptake and breakdown to provide amino acids for adult tissues and reproduction (Telfer and Pan, 2003) rather than protein instability. In contrast, the second group includes apolipophorin-II-I (which is cleaved after synthesis to produce apoLp-II and apoLp-I) and apoLp-III, whose levels in adults increased 4.0–5.5 fold from their averages in B, P4, and P13 (Fig. 3B). Because their levels were already high (average: 3.2×10⁴ PPM) in B and became higher in P4 and P13, the 4.8-fold elevation on average makes apoLp-I (285 kDa), apoLp-II (75 kDa), and apoLp-III (19 kDa) the most prominent proteins in adult plasma (Fig. 1A and Fig. S1) (Shapiro and Law, 1983). ApoLp-II-I contains a RGRR site for intracellular processing into two subunits of a lipoprotein complex, and apoLp-III can significantly increase lipid content of the complex and thus its transport efficiency to provide energy for flight of adults (Arrese and Soulages, 2010). To support reproduction, vitellogenin level in adults was >1600 fold higher than the mean of B, P4 and P13, consistent with previous observations (Ismail et al., 1998). Vitellogenin is a female-specific protein synthesized in fat body, transported via hemolymph, and deposited in eggs (Mundall and Law, 1979). Abundance of microvitellogenin was highest in adults (Table S2). In addition, the levels of ferritin-1 and ferritin-2, immulectin-13, SPH2, SPH1b, and seven other proteins were much higher in adult than in the other three stages (Fig. 3B).

Fig. 3.

Relative abundances of the selected hemolymph proteins in the four life stages. Proteins whose levels greatly decreased (A) or increased (B) in adults. PPM values (mean ± SEM, n = 3) for a protein in L (for larval), B, P4, P13, and A proteomes are shown in as blue, gray, red, yellow, and black bars, respectively. The proteome data of day 2, 5^th instar naïve larvae are from the previous work (He et al., 2016), with “L” highlighted in blue bold font. The highest PPM value in each bar graph is shown above the top gridline. IML, immulectin; PAOX, polyamine oxidase; HTIC, hemolymph trypsin inhibitor C; ABP, antennal binding protein; golginL, goglgin-like protein; hypoPr1a, hypothetical protein-1a.

3.3. Developmental changes in hemolymph functions at the proteome level

The hierarchical cluster analysis and GO enrichment test provided new insights into the stage-specific changes in the proteomes. We identified six groups of proteins with signal peptide (a–f) and six without (g–l) (Fig. 4). In group-a, serine proteases and their homologs are abundant in the B proteome, consistent with our biochemical results showing the presence of an activated serine protease system in wandering larvae and prepupae (Jiang et al., 2003a, b; Wang and Jiang, 2004; He et al., 2018). In group-c, lipid transport and, possibly, hydrolysis of extracellular matrix proteins by metalloproteases are important biological processes in the adult stage. Consistent with high levels of the apolipophorins (Fig. 3B), lipid transport and metabolism are critical for flight and energy storage in eggs. Group-j proteins are enriched for the biological processes of protein transport and aromatic amino acid metabolism in young adults, which could be related to the uptake and degradation of the arylphorins and other storage proteins. Group-l is enriched by proteins involved in ubiquitin-dependent protein catabolism in proteasomes and ATP generation in the P13 and A stages.

Fig. 4.

Abundance changes of the identified plasma proteins with (A) or without (B) a predicted signal peptide. Normalized protein levels (i.e. PPM values) were used to calculate z-scores for samples from the four life stages, which were clustered into groups a–f (A) and g–l (B) for plotting the heat maps. Based on PPM values, a one-way ANOVA test was performed for each protein in three biological replicates from the four stages, with a short blue bar indicating p <0.05. GO enrichment analysis was performed for each of the 12 groups using GOATOOLS. Representative GO numbers and terms with significant enrichment (p <0.01) are listed on the right and, for those with false discovery rate (FDR) <0.05 in the Benjamini-Hochberg test, their GO terms are colored red.

3.4. Developmental changes in mRNA and protein levels of immunity-related genes

To understand the dynamics of defense proteins in these four stages, we compared the proteins identified in the MS analysis with the immunity-related genes found in the genome (Kanost et al., 2016). The comparison yielded a total of 410 proteins, most of which contain a signal peptide. We classified them into six functional groups and plotted heat maps of z-scores, PPM values, and TPM values of the proteins in each group (Fig. 5, A-F). The TPM values were from six RNA-seq datasets of fat body, a major source of hemolymph proteins. Stages of these fat body samples encompass those of the four hemolymph samples. In general, genes with high TPM values tend to have higher PPM values, but the correlation can be low (He et al., 2016).

Fig. 5.

Protein and RNA levels of genes related to immunity (A–F) or highly expressed (G). As shown in Fig. 4, a heatmap with columns B, P4, P13, and A represents z-scores calculated from the PPM values. A blue square on the right marks p <0.05 in the one-way ANOVA test. Columns 0–3 and 4–9 show log₂(PPM+1) values of proteins and log₂(TPM+1) values of their RNAs, respectively. A protein is considered as highly expressed if its average PPM in one of the four stages is greater than 1,000 (i.e. 2¹⁰ or >0.1% molar intensity). The gradient heat maps change from black (0) to red (>2¹⁰). The values of 0–0.49, 0.50–1.49, 1.50–2.49, … 8.50–9.49, 9.50–10.49, 10.50–11.49, … 15.50–16.49, and 16.50–17.49 are labeled as blank, 1, 2, … 9, A, B, … G, and H, respectively. As defined in our previous studies, defense proteins include pattern recognition receptors (PRRs) (A), serine proteases (SPs) and their homologs (SPHs) (B), serpins (C), signaling proteins (D), effectors (E), and other immune factors (F). Highly expressed genes unrelated to immunity (G) are also shown. B (“0”), prepupae with metathoracic brown bars; P4 (“1”) and P13 (“2”), pupae on days 4 and 13; A (“3”): day 1 adult; L5-preW (“4”): 5^th instar larvae in pre-wandering stage; L5-W (“5”): 5^th instar larvae in wandering stage; P1–3 (“6”): pupae on days 1–3; P15–18 (“7”): pupae on days 15–18; A1–3 (“8”): adults on days 1–3; A7–9 (“9”: adults on days 7–9. Columns “4”–“9” represent mRNA of fat body, a major source of hemolymph proteins. Abbreviations of the proteins are listed in Tables S1 and S2.

High protein levels (average PPM ≥2⁸) were observed in the four stages for fifteen pattern recognition receptors, including hemolin, β−1,3-glucan recognition proteins (βGRP1 and βGRP2), microbe binding protein, peptidoglycan recognition proteins (e.g. PGRP1), immulectin-2, 3, 7, 8, 10, 13, and Reeler-1 (Fig. 5A). Leureptin-9, 21, immulectin-4, 16, 17, and PGRP5 were moderately abundant (PPM: 2^5–7). Levels of βGRP2, immulectin-2, 4, 13, leureptin-1, and −9 decreased significantly from B to A stage. In contrast to A. gambiae thioester protein-1, a major PRR in hemolymph for parasite recognition (He et al., 2017), M. sexta thioester proteins were present at very low levels (PPM: 2^0–2).

Serine proteases (SPs) and their noncatalytic homologs (SPHs) coordinate several key immune responses (Fig. 5B) (Wang et al., 2020). In M. sexta larvae, at least thirteen proteins constitute an SP-SPH system to activate proPO and Toll pathway, including HP14, 21, 2, 5, 6, 8, 1a, 1b, PAP1–3, SPH1, and SPH2. All of them were detected in the larval, pupal, and adult hemolymph (Fig. 5B), suggesting that the system is likely functional through development, perhaps with some modifications brought by additional SP-related proteins. We detected as many as 41 other SP(H)s in the proteomes, whose functions have not yet been investigated. Their abundances, structural features, and phylogenetic relationships (Cao and Jiang, 2018) are helpful in candidate selection for future research on the SP-SPH system. Levels of SP32, SP34, SP60, SP112, HP16b, HP17a, HP19, HP20, HP25, HP28, HP29, SPH33, SPH101, and scolexins are similar to those of the known protease system components. SP112, HP25, and SPH33 contain Sushi domain(s) while HP17a, HP28, SP60, and SPH1b have a clip domain (Cao et al., 2015b). The peaking of HP1b, HP2, HP5, HP8, PAP1–3 mRNA levels in fat body of wandering larvae and young pupae and their subsequent decrease agree with the higher abundances or these proteins in the B and P4 stages. This suggests that increases in SP levels may induce antimicrobial peptide synthesis in an infection-independent manner, promoting accumulation of defense proteins that protect insects against pathogen infection as they enter the vulnerable wandering and pupal stages (He et al., 2015). Elevation of the HP6, HP19, HP20, SPH100, and SPH101 protein levels in newly emerged adults suggests their functional importance in adults, which is poorly understood. SPH1b, a paralog of SPH101, SPH1a, and SPH4, is mainly expressed in fat body, head, and muscle in the larval stages (Cao et al., 2015). In the B, P4, P13, and A samples, SPH1b and SPH101 (1,200–3,500 PPM total) is comparable in amount to SPH2 (800–3,800 PPM). Being 96.4% identical and 98.6% similar to SPH1b, SPH101 may form a high M_r complex with SPH2 in certain tissues or stages, similar in cofactor activity to the complex of SPH1b and SPH2 isolated from bar-stage hemolymph (Wang and Jiang, 2004).

As irreversible inhibitors, most serpins regulate protease-mediated physiological processes in insects. M. sexta serpin-1, 3, 4, 6, and 12 protein levels were high (PPM ≥2⁸) in all of the stages examined in this study; Serpin-5, 10, 11, 13, 15B, and 17 were moderate (PPM 2^4–7) (Fig. 5C). Fifteen serpins predicted to be non-inhibitory (Li et al., 2018) were either undetected or detected at low levels (PPM ≤2³). We detected 13 of the 14 splicing variants of serpin-1 (Table S2), with dramatic variations in protein levels. However, most of the serpins, including the total amount of serpin-1 splicing variants (Fig. S3), remained relatively constant in the four stages examined.

Precursors of some cytokines that are proteolytically processed and then function to stimulate immune responses were detected in the proteomes. The precursor of plasmatocyte spreading peptide was abundant (PPM: 2^11–13) in the post-larval proteomes. The protein levels of proSpätzle-7 and proSpätzle-2 were considerably higher in these post-larval stages than those of proSpätzle-1 (Fig. 5D, Table S2), which stimulates antimicrobial protein synthesis in feeding larvae (An et al., 2010). Their relative roles in activating Toll receptors in these stages are worth exploring. The putative co-receptor of Toll, MD2, was also highly abundant (PPM: 2^10–11).

Effectors of the M. sexta immune system include antimicrobial peptides, proPOs, and transferrins (He et al., 2015). Over 80 effector genes have been identified in the genome (He et al., 2015); about one third of their protein products were detected in the hemolymph samples of post-larval developmental stages not stimulated by infection (Fig. 5E, Section 4.1). Attacin-2 (average PPM: 2⁴, same below), attacin-7 (2⁶), diapausin-1 (2⁵), diapausin-3 (2⁴), and diapausin-4 (2¹¹) were more abundant than other paralogs of these antimicrobial proteins. Attacin-1, −3, −5, −9, lysozyme like protein-1 (LLP1) and LLP2 mRNA levels peaked in early pupal fat body (He et al., 2015), but their protein abundances peaked in the pharate pupae and decreased thereafter, suggesting consumption or degradation. In contrast, proPO1, proPO2 and transferrin-1 protein levels stayed high, even though their mRNA levels decreased significantly. These three proteins seem to be stable and not consumed in the later stages.

Some low M_r protease inhibitor domains were present in proteins detected at high levels in the plasma samples, including hemolymph trypsin inhibitors A–E (average PPM: 2¹¹ for HTIA, 2¹¹ for HTIB, 2⁹ for HTIC, 2⁸ for HTID, and 2¹⁰ for HTIE), inter-α-trypsin inhibitor heavy chain H4 like (ITIH4, 2¹²), Bombyx cysteine protease inhibitor homolog (BCPI, 2¹⁰), and Manduca cysteine protease inhibitor (CPI, 2⁹), and cationic protein-8 CP8 (2¹⁵) (Fig. 5F). Kunitz-type serine protease inhibitors HTIC and HTID were much more abundant in adults than in the earlier developmental stages.

4. Discussion

In this study, we have extended characterization of M. sexta hemolymph proteome from larval feeding stage (Furusawa et al., 2008; Zhang et al., 2014; He et al., 2016) to pharate pupal, early and late pupal, and young adult stages. While accurate quantification of protein levels in complex mixtures remains a challenge, our results provided a reliable overview of the hemolymph proteins and their abundances at these time points. Along with the fat body transcriptome data in similar life stages, the new proteomics information reflects changes in hemolymph functions underlying metamorphosis.

4.1. Proteomic changes in feeding larval and post-larval hemolymph

How do the proteomes change from larval feeding to prepupal stage? To address this question, we reanalyzed proteomics data of hemolymph samples from day 2, 5^th instar naïve larvae (L) (He et al., 2016). While sensitivity of the mass spectrometer used in that study was lower, the same method of data normalization yielded comparable PPM values (Supplemental Information), providing support for analyzing the larval hemolymph data together with the new results from other stages. To increase reliability of the comparison, we focused on proteins that had differences in level between stages that were statistically significant in ANOVA and Tukey’s HSD range tests. From these, we selected the ones with a signal peptide and a PPM value in one stage being at least 10-fold higher than the mean of the other four stages. There were 112, 8, 2, 43, and 29 such “stage-specific” proteins in the L, B, P4, P13, and A samples, respectively (Table S3). Among the 194 proteins, twenty-eight had PPM >1000 in larvae, one in late pupae, and five in adults.

We examined BLAST search results of these proteins and their stage specificity and made several observations. First, our knowledge about hemolymph proteins is quite limited, since 53 of the 194 “stage-specific” proteins with a putative signal peptide had no hit or are similar to uncharacterized proteins. Second, the number of proteins lacking a signal peptide in P13 is 2–4-fold higher than that in B, P4, or A (Fig. 2A), suggesting a release of cellular components. Third, no clear functional trend was apparent from examination of the stage-specific proteins, other than the abundant and immunity-related ones. We will have to rely on future studies of individual proteins to understand the global changes in the five life stages. Omics tools are powerful but cannot replace biochemical studies on individual proteins with no or limited functional information.

4.2. Functional insights from individual proteins

Comparison of the transcriptome and proteome data indicates that some hemolymph proteins may have complex tissue sources. For instance, Takeout is abundant in B and P4, but its mRNA levels are near zero in fat body (Fig. 5G). Because a single takeout mRNA peak (FPKM: 31,208) is found in head of pre-wandering larvae (Cao and Jiang, 2017), the 27 kDa takeout lipocalin is likely produced and released into hemolymph before the wandering stage begins. In Drosophila, starvation strongly induces the expression of takeout (a circadian clock-controlled gene) in brain. The protein affects feeding and locomotion behaviors, although its ligand and intracellular receptor are unclear (Sarov-Blat et al., 2000). In Manduca, takeout levels are high in B and P4 (PPM: 2¹³) and become lower in P13 (2⁸) and A (2⁴). As the decrease occurs along with a transition from the prepupal and pupal non-feeding stages to the adult stage, we suggest that takeout may perform a similar function in the moth, despite the low sequence identity of 26% with Drosophila. In another example, abundance of SP251 (an S1A serine protease) had a moderate peak (PPM: 2⁴) in P4, but its fat body mRNA level peaked at a later time (P15–18, TPM: 500). Higher expression occurred in midgut in the same period (TPM: 4,000) and reached maximum (TPM: 16,000) in adults (days 3–5) (Miao et al., 2020). It is worth investigating whether SP251 is involved in midgut and fat body tissue remodeling and whether hemolymph protease inhibitors regulate SP251 activity.

Besides lipophorins, other hemolymph proteins may also transport hydrophobic molecules. Lipoprotein D (lipoD) and its homolog (lipoDL) are lipocalins, which typically contain a binding pocket for a hydrophobic ligand, and both contain a secretion peptide. LipoD had a peak level in P13 (PPM: 2¹³), whereas lipoDL (2^10–12) is present at similar levels in the four stages (Fig. 5G). Insecticyanins B1 (PPM: 2^12–13) and B2 (2^14–16) are also lipocalins that bind biliverdin and provide camouflage to protect eggs and larvae (Riley et al., 1984). Juvenile hormone binding protein (JHBP) (PPM: 2^11–12), a takeout homolog (Sarov-Blat et al., 2000), protects JH from degradation by serine esterases (Kramer et al., 1976). While JHBP level reached 2¹⁴ in feeding larvae, the level of JH esterase (Hinton and Hammock, 2001) in B was 12-fold higher than the mean of C, P4, P13, and A. JHBP and low level of JHE together maintain a high JH level in feeding larvae. JH and ecdysone are probably responsible for many of the protein level changes reported in this study.

Both microvitellogenin (Wang et al., 1989) and PGRP14 (Zhang et al., 2015) contain a lipocalin-11 domain, as do the seven M. sexta homologs of Pseudaletia separata growth blocking peptide binding protein (GBP-BP) (Matsumoto et al., 2003). P. separata GBP induces the release of GBP-BP from ruptured oenocytoids to terminate its own function. The M. sexta ortholog of GBP is named paralytic peptide (PP) or plasmatocyte spreading peptide (Wang et al., 1999). Therefore, we have named the homologs of GPB-BP as paralytic peptide binding proteins (PPBP1–7). PPBP1 (PPM: 2⁸ in the four stages) is more abundant than PPBP2, PPBP5, PPBP6, and PPBP7 (2^0–5) (Fig. 5, A and D). However, it is unclear whether the PPBPs are released from hemocytes upon PP treatment to down-regulate this cytokine and its related peptides (Schrag et al., 2017). There are also members of the odorant binding protein (OBP) family in hemolymph, including ABP1, ABP7, ABP8, OBP20, OBP21, OBP25, OBP26, OBP30, OBP32, OBP35, and OBP36 (Vogt et al., 2015). Abundance of OBPs (ABP1, 2^5–9; ABP7, 2^7–11; OBP20, 2^9–11; OBP25, 2^8–11; OBP35, 2^3–9), ommochrome-binding protein (OmBP, 2^14–15), and chemosensory proteins (CSP1: 2^3–10; CSP2: 2^12–13; CSP3: 2^9–11; and CSP4: 2^5–10) are above 2⁹ PPM in one or more of the four stages we examined. Except for OmBP (Martel and Law, 1991), ligands and thus functions of these proteins are unclear, and this again reflects our lack of knowledge on hemolymph proteins in this well-established biochemical model.

4.3. Low M_r protease inhibitors in hemolymph

While serine protease inhibitors (SPIs) of the serpin superfamily have been quite extensively characterized in M. sexta (Li et al., 2018), other SPIs are also found, including hemolymph trypsin inhibitors A and B (Ramesh et al., 1988). Here we have identified HTIC–E and other small inhibitors of Ser, Cys, and metallo- proteases (Section 3.4). HTIA–E are Kunitz-type SPIs (Laskowski and Kato, 1980). We detected in hemolymph a splicing variant of lacunin, an extracellular matrix protein deposited in adult basal lamina by granular hemocytes (Nardi et al., 2001). Lacunin contains eleven Kunitz inhibitor domains, and this 3353-residue splicing variant had an extra immunoglobulin domain near the C-terminus. In contrast to the five low M_r Kunitz-type trypsin inhibitors (PPM: 2⁸–2¹¹), Kazal-type SPI-1–5 were present at a lower level, 2¹⁰, 2⁵, 2⁰, 2³, and 2⁸ PPM on average, respectively. Functions of the small SPIs are not well established and these canonical SPIs may be a part of the barrier against proteases released from pathogenic microbes.

We identified two cysteine proteases inhibitors (CPIs) that may regulate cathepsins 26/29-like-1 (Cath1), 5–8, B and L in hemolymph (Serbielle et al., 2009) or block cysteine proteases from invading organisms. Levels of the seven cathepsins are present at low-to-moderate levels, whereas the ortholog of B. mori cysteine protease inhibitor (BCPI) (Kurata et al., 2001) and M. sexta CPI are abundant (2⁹−2¹⁰) (Fig. 5F). The precursor of CPI contains four cystatin-like domains, nine MsCPI domains, and one cathepsin F domain (Miyaji et al., 2010). Intracellular proteolytic processing of the 2,676-residue precursor at RAKR yields nine nearly identical, low M_r inhibitors 105 residues long. The silkworm BCPI is a potent inhibitor of the cathepsin L-like Bombyx cysteine protease, with a K_i of 36 pM (Kurata et al., 2001), and the hornworm BCPI may perform a similar function. The mature MsCPI inhibited human cathepsins B (K_i: 6.8 nM) and L (K_i: 0.87 nM), but its target enzymes in vivo are unclear (Miyaji et al., 2007). While a role of cathepsins in tissue remodeling was suggested (Kanost et al., 2016), these cysteine proteases may participate in immune responses against parasitoids (Serbielle et al., 2009) or microbes as well. Notably, M. sexta Cath1 is orthologous to Drosophila 26-29-p, a circulating C1A protease that activates Persephone and triggers the Toll pathway (Issa et al., 2018).

Several other putative protease inhibitors were detected in the hemolymph proteomes. Inducible metalloprotease inhibitor (IMPI), containing two I8 domains, is a homolog of Galleria mellonella IMPI that inhibits thermolysin (Clermont et al., 2004). A 389-residue protein, designated 7PDP, is composed of seven I19 domains initially found in crayfish pacifastin (Liang et al., 1997). Cationic protein CP8 contains an I83 domain similar to a fungal protease inhibitor from Antheraes mylitta (Ling et al., 2009). Both WAP1 and WAP14 have an I17 domain first identified in human leukocyte peptidase inhibitor (Stetler et al., 1986).

4.4. Concluding remarks

Our combined approach yielded an overview of the composition and abundance of hemolymph proteins in M. sexta prepupae, pupae, and adults. Since most of the 1,824 proteins exist in all these stages, major functions of hemolymph are maintained, including storage, transport, and immunity. On the other hand, stage-specific changes in the proteome level may reflect physiological processes underlying metamorphosis. These include: 1) storage of amino acids in hexamerins, 2) intracellular proteins released during tissue remodeling, 3) transport of nutrients such as lipids by lipophorin and other lipid binding proteins, 4) innate immune proteins, including pathogen recognition receptors, proteases, protease inhibitors, signal transducers, effectors, and other proteins, in the absence of infection, perhaps protecting insects from microbial infection during metamorphosis. While functions of many proteins identified in the proteomes are unknown, comparison of the proteome and fat body transcriptome data provided functional insights into selected proteins. Future biochemical studies are expected to greatly enrich our knowledge on the protein molecules uncovered in this research.

Supplementary Material

Table S1

Table S1. Features of the 98 abundant proteins in M. sexta hemolymph*

*: OGS 2.0 and GenBank IDs of abundant proteins, defined as PPM >1000 in at least one samples, are enlisted for sequence retrieval on https://i5k.nal.usda.gov/ and https://www.ncbi.nlm.nih.gov/protein/, respectively. Modified NCBI name, short name, GO (for gene ontology), and KO (for KEGG ontology) are based on different BLAST search results. Z-score, PPM mean, PPM maximum, and p-value of ANOVA for each protein are shown. Specific stage (L, B, P4, P13, or A) denotes a protein’s PPM in that stage is at least 5-fold higher than the mean of the other four stages (>2.5 PPM, p-value of One-way ANOVA < 0.05 for the five stages).

Table S2

Table S2. Features of the 331 hemolymph proteins of interest in M. sexta*

*: Protein IDs, names, and specific stages are enlisted as described in footnotes of Table S1. Additional information about the 98 abundant proteins (e.g. PPM, TPM) is also included.

Table S3

Table S3. Level changes of the 194 stage-specific hemolymph proteins with a signal peptide*

*: Protein IDs, short names, NCBI annotation, PPM averages and SEMs, and specific stages (ratio >10) are enlisted as described in footnotes of Table S1.

5

Fig. S1. SDS-PAGE analysis of hemolymph proteins on 4–15% precast gradient gels. The samples from insects in the bar, P4, P13, and adult stages were separated by electrophoresis and stained with Coomassie blue. Sizes and positions of the M_r markers are shown on the left.

Fig. S2. Relative abundances of the selected hemolymph proteins in the four life stages. Proteins whose levels markedly increased in late pupae. PPM values (mean ± SEM, n = 3) for a protein in L (for larval), B, P4, P13, and A proteomes are shown in as blue, gray, red, yellow, and black bars, respectively. The proteome data of day 2, 5^th instar naïve larvae are from the previous work (He et al., 2016), with “L” highlighted in blue bold font. The highest PPM value in each bar graph is shown above the top gridline.

Fig. S3. Relative abundances of M. sexta serpin-1, OmBP, HAIP, CSP2, and GP27 in the six groups of plasma samples. PPM values (mean ± SEM, n = 3) for each protein in the B, P4, P13, A, C and I proteomes are shown in as gray, red, yellow, black, cyan, and green bars, respectively. The proteome data of control (C) and induced (I) plasma from day 2, 5^th instar larvae are from the previous work (He et al., 2016), with “C” and “I” highlighted in red bold font. The highest PPM value in each bar graph is shown above the top gridline. ***, **, and n.s. stand for P values <0.001, <0.01, and >0.05 in one-way ANOVA for the four stages (B, P4, P13, and A, in blue) and six groups (B, P4, P13, A, C, and I, in red), respectively. Results of Tukey tests for B, P4, P13, A, C, and I samples are labeled on top of bars, groups sharing a letter are statistically insignificant.

Highlights.

• Identification of 1,828 proteins in hemolymph of Manduca sexta in four different life stages

• Elucidation of the major functions of insect hemolymph, i.e. storage, transport, and immunity

• Detection of stage-specific proteomic changes by expression profiling and GO enrichment

• Examination of developmental changes in mRNA and protein levels of immunity-related genes

Abbreviations:

B

pharate pupae with metathoracic bars

P4/13

day 4 or 13 pupae

A

day 1 adults

CP

cuticle proteins

FDR

false discovery rate

βGRP

β−1,3-glucan recognition protein

iBAQ

intensity-based absolute quantification

IMPI

inducible metalloprotease inhibitor

MS

mass spectrometry

MRSP

Met-rich storage protein

proPO

prophenoloxidase

PAP

proPO-activating protease

PGRP

peptidoglycan recognition protein

PI

protease inhibitor

PPM

parts-per-million

PP

paralytic peptide

PRR

pattern recognition receptor

SP and SPH

serine protease and its non-catalytic homolog

TPM

transcripts per kilobase million