Preferential binding of DAP-PGs by major peptidoglycan recognition proteins found in cell-free hemolymph of Manduca sexta

Abstract

Peptidoglycan recognition proteins (PGRPs) detect invading bacteria to trigger or modulate immune responses in insects. While these roles are established in Drosophila, functional studies are not yet achieved at the PGRP family level in other insects. To attain this goal, we selected Manduca sexta PGRP12 and five of the nine secreted PGRPs for recombinant expression and biochemical characterization. We cloned PGRP2–5, 12 and 13 cDNAs, produced the proteins in full (PGRP2–5, 13) or in part (PGRP3s, 12e, 13N, 13C) in Sf9 cells, and tested their bindings of two muramyl pentapeptides by surface plasmon resonance, two soluble peptidoglycans by competitive ELISA, and four insoluble peptidoglycans and eight whole bacteria by a pull-down assay. Preferential binding of meso-diaminopimelic acid-peptidoglycans (DAP-PGs) was observed in all the proteins containing a peptidoglycan binding domain and, since PGRP6, 7 and 9 proteins were hardly detected in cell-free hemolymph, the reportoire of PGRPs (including PGRP1 published previously) in M. sexta hemolymph is likely adapted to mainly detect Gram-negative bacteria and certain Gram-positive bacteria with DAP-PGs located on their surface. After incubation with plasma from naïve larvae, PGRP2, 3f, 4, 5, 13f and 13N considerably stimulated prophenoloxidase activation in the absence of a bacterial elicitor. PGRP3s and 12e had much smaller effects. Inclusion of the full-length PGRPs and their regions in the plasma also led to proHP8 activation, supporting their connections to the Toll pathway, since HP8 is a Spätzle-1 processing enzyme in M. sexta. Together, these findings raised concerns on the common belief that the Toll-pathway is specific for Gram-positive bacteria in insects.

1. Introduction

Recognition of conserved molecular patterns associated with microbes is the first step of a successful innate immune response against pathogen invasion. Pattern recognition receptors have evolved in vertebrates and invertebrates to specifically bind peptidoglycans, lipopolysaccharide, β-1,3-glucan, and other microbial surface molecules (Kanost et al., 2004). Peptidoglycan recognition by their recognition proteins (PGRPs) has been extensively studied in Drosophila but less in other insects (Kurata, 2014). Peptidoglycans of walled bacteria are high M_r polymers of unbranched glycan strands cross-linked by short stem peptides, which maintain the cell integrity and shape (Vollmer et al., 2008; Dramsi and Bierne, 2017). The glycan strands consist of alternating N-acetylglucosamine and N-acetylmuramate (NAM) residues linked by β-1,4-glycosidic bonds. A short peptide attached to NAM via lactate has a Lys at the third position in Lys-PGs of most Gram-positive bacteria and has a meso-diaminopimelic acid at the same location in DAP-PGs of Gram-negative bacteria and Gram-positive Bacillus and Clostridium species. The stem peptides are linked directly or indirectly through cross-bridges, which lead to the formation of a multilayer net wrapping the bacterial protoplast. Lys-PG layers constitute the cell wall of most Gram-positive bacteria whereas DAP-PG sheets of Gram-negative bacteria are covered by an outer membrane. Different bacteria vary in stem peptide, cross-bridge, and physical properties (e.g. thickness, elasticity, porosity) of their peptidoglycans (Vollmer et al., 2008; Kim et al., 2015). To recognize various peptidoglycans, each insect employs a set of PGRPs that exhibits a certain level of specificity toward Lys- and DAP-PGs (Werner et al., 2000; Tanaka et al., 2008; Zhang et al., 2015). Some PGRPs are present in hemolymph; others are cytosolic or bound to cytoplasmic membrane. Most of them are catalytically inactive whereas some possess an amidase activity that cleaves between lactate and l-alanine at the first position of the stem peptide. The Zn²⁺-dependent enzyme activity comes from a 160-residue peptidoglycan-binding domain homologous to the lysozyme of bacteriophage T7 (Ochiai and Ashida, 1999; Mellroth et al., 2003). As such, PGRPs sense bacterial presence, activate defense responses, and hydrolyze peptidoglycans to attenuate immune responses in Drosophila and other insects (Royet and Dziarski, 2007; Jiang et al., 2010). In Bombyx mori, PGRP-S1 (the founding member of this protein family) is responsible for initiating the proPO activation system after peptidoglycan binding (Yoshida et al., 1996). PGRP-S4 and -S5 hydrolyze peptidoglycans but stimulate melanization in the silkworm (Chen et al., 2014, Yang et al., 2017). The PGRP-S5 amidase activity is required for proPO activation and antimicrobial peptide suppression (Chen et al., 2016). In Tenebrio molitor, clustered PGRP-SA detects Lys-PG fragments to activate the proPO cascade and Toll pathway (Park et al., 2007). These findings, not without apparent contradiction, reflect the current state of our understanding of peptidoglycan recognition and hydrolysis in insects.

In the tobacco hornworm Manduca sexta, PGRP1 is a 19 kDa protein 54% identical in sequence to B. mori PGRP-S1. While PGRP1 level in larval plasma increased after an immune challenge, its role as a sensor of the immune system remained unclear until the bindings to various Lys- and DAP-PGs were correlated with proPO activation (Sumathipala and Jiang, 2010). X-ray crystal structure of the PGRP1 revealed the basis of its unusual binding properties (Hu et al., 2019). M. sexta PGRP1 and microbe binding protein (MBP) bind DAP-PG as well as Lys-PG to cause activation of hemolymph protease-14 precursor (proHP14) (Wang and Jiang, 2017). HP14 is the initiating enzyme of an extracellular serine protease system integrating PO-mediated melanin formation and Spätzle-triggered Toll pathway activation (Wang et al., 2020). RNA-Seq analyses yielded contigs of nine M. sexta PGRPs (Zou et al., 2008; Gunaratna and Jiang, 2013). After the M. sexta genome became available, we identified genes for eight short PGRPs (1–7, 9), four long PGRPs (8, 10–12) all with a transmembrane region, and PGRP13, fusion of a signal peptide, a peptidoglycan-binding domain, and a lipoprotein domain (Zhang et al., 2015). We profiled their transcript levels in 52 tissue samples from different life stages in the same study. Proteomics analysis detected four short PGRPs (1–3, 5) in cell-free larval hemolymph and their level increases after a bacterial challenge (He et al., 2016). Much lower levels of PGRP6, 7, and 13 proteins were detected in prepupal, pupal, and adult hemolymph (Cao et al., 2020). To study the binding specificity of major plasma PGRPs, we cloned M. sexta PGRP2–5, 12, and 13 cDNAs, expressed the proteins in full and in part in insect cells, tested their binding to peptidoglycan monomers, peptidoglycan polymers and whole bacteria, and explored their possible roles in proPO activation. Our results from the in vitro experiments suggested substantial differences between the moth and fly immune systems in terms of bacterial recognition. A phylogenetic analysis of the PGRP1-like sequences from twelve orders of insects and their predicted specificities of peptidoglycan binding raised a concern on applicability of the simple Drosophila model. These findings may profoundly impact future studies of innate immunity in insects.

2. Materials and methods

2.1. Insect rearing, hemolymph collection, and plasma screening

M. sexta eggs were purchased from Carolina Biological Supply and larvae were reared on an artificial diet (Dunn and Drake, 1983). Hemolymph samples were collected from a cut proleg of day 2, 5^th instar naïve larvae and centrifuged at 5000 × g for 5 min to remove hemocytes. The plasma aliquots (1.0 μl each) were screened for low basal PO activity and for high proPO activation after incubating with 1.0 μg M. luteus for 1 h at room temperature. Selected naïve larval plasma (NP) samples were aliquoted and stored at −80 °C for use within a month.

2.2. cDNA cloning and expression of M. sexta PGRP2–5, 12, 13, and their segments

A fragment of M. sexta PGRP2 cDNA was amplified from cDNA of induced fat body using primers j971 (5’-GGTGCAAATGCAAACCACACA) and j972 (5’-CATATCCTCACGAAGTTGTTCC), designed based on contig 3684 (Zou et al., 2008). After sequence confirmation, the 165 bp product was labeled with [α-³²P]-dCTP to screen 1.2 ×10⁵ plaques from M. sexta induced fat body cDNA library (Wang et al., 2011). Purification of positive plaques, in vivo excision of phagemids, plasmid isolation, and sequencing were performed. The longest full-length cDNA in clone P5 was used as a template to amplify a fragment using primers T3 (5’-GCAATTAACCCTCACTAAAGG) and j206 (5’-GTCCTCGAGTATCCTCACGAAGT). Following T/A cloning and sequence validation, the insert was retrieved by BamHI-XhoI double digestion and subcloned into the same sites of pFH₆. Baculovirus generation, Sf9 cell infection, and PGRP2 isolation from conditioned medium were performed as described previously (Wang et al., 2011). The purified PGRP2 (Y¹⁹PS…VRI¹⁹⁶LEHHHHHH) (0.5 mg) was used as an antigen to raise a rabbit antibody and the remaining protein was aliquoted and stored at −80 °C for functional assays.

M. sexta PGRP3 cDNA was isolated as a false positive clone from the fat body cDNA library. A dopachrome conversion enzyme cDNA fragment was amplified from induced fat body cDNA using primers j276 (5’-CTCTCCACTGCTGATGCCGTATCCA) and j277 (5’-GATCGCAGGC GGCTGGAGTTG), based on a singlet called EAY20JP04I8116 (Zou et al., 2008). After product labeling, screening, in vivo excision, and sequencing, one plasmid (NC12) was found to encode PGRP3. A cDNA fragment was amplified using primers j1301 (5’-TCATATGTTTCCATCATTATTTGCAG) and j1302 (5’-AAGCTTATCAAGAGCATAGTGTTG, G replacing CT due to a design error) for T/A cloning and sequence confirmation. The NdeI-HindIII fragment was inserted to the same sites in pSKB3, a derivative of pET-28a with a TEV cleavable affinity tag (MGSSHHHHHHDYDIPTTENLYFQ), to fuse with GHMF¹⁸PSL…PEST²¹⁸TLCS. The underlined region was later found to be a part of the mature PGRP3 (F¹⁸PS…PEST²¹⁸KTMLLN RRNNTTV²³¹. The shorter PGRP3 made in Escherichia coli BL21 (DE3) cells harboring PGRP3s/pSKB3 (“s” for short) was isolated from the supernatant of cell lysate by nickel affinity chromatography, separated by SDS-PAGE, and used as antigen to raise antibodies in a rabbit. The PGRP3s cDNA was also amplified from the induced fat body cDNA pool using primers j287 (5’-GGAATTCTTCCATCATTATTTGCA) and j290 (5’-CTCGAGAGAGCATAGTGTTGTG) for expression in insect cells. Following T/A cloning and sequence validation, the EcoRI-XhoI fragment was inserted to the same sites in pMFH6 (Lu and Jiang, 2008). In vivo transposition of the expression cassette, transfection of Sf9 cells with bacmid DNA, large-scale infection of insect cells, and purification of the secreted PGRP3s were performed as described before (Wang et al., 2011). According to the design, the PGRP3s has a sequence of GILP¹⁹SL…PEST²¹⁸TLCSLEHHHHHH. To express the full-length PGRP3, another cDNA was amplified using primers j287 and j1490 (5’-CTCGAGAGTGGTATTATTTCTGCG). After cloning and sequence confirmation, PGRP3f/pMFH6 (f for full-length) was constructed to yield a baculovirus to infect Sf9 cells and produce the secreted PGRP3f (GILP¹⁹SL…NNTT²³⁰LEHHHHHH).

Based on the gene models of PGRP4 and PGRP5 (Zhang et al., 2015), their cDNAs were amplified from induced fat body cDNA using primer pairs j1491-j1492 (5’-GAATTCGACCTAACTTTCACAGTG, 5’-CTCGAGTGTCTTTTTAATTTTGTCGA) and j1479-j1480 (5’-GAATTCATATGGACTGCGGCGTGGTCAGTAAG, 5’-AAGCTTACTCGAGCTGATCGACTTTATCCAGCCA), respectively. PGRP4/pMFH6, PGRP5/pMFH6, and corresponding viruses were generated to produce secreted PGRP4 (GIR²¹PNF…IKKT²⁰⁹LEHHHHHH) and PGRP5 (GIHMD¹⁹CGV…KVDQ¹⁸⁷LEHHHHHH).

The PGRP12 and PGRP13 cDNAs were amplified from induced fat body and nervous tissues using primer pairs j1475-j1476 (5’-GAATTCATATGGATTCAACAAGAGATGACA, 5’-AAGCTTACTCGAGTGTTTTTGAGACCATCTC) and j429-j430 (5’-GAATTCATATGGATTGTGACGTAATCGATAAG, 5’-AAGCTTACTCGAGATGAAAGATGCGCCAAC), respectively. To produce the N- and C-terminal domains of PGRP13, j429-j431 (5’-AAGCTTACTCGAGCCATTGAGGCCGT) and j430-j432 (5’-GAATTCCTCAATGGATAGAAAAC) primer pairs were used to amplify the cDNA fragments, respectively. After T/A cloning and sequence confirmation, NdeI-HindIII fragment encoding PGRP12e (“e” for ectodomain) was inserted to the same sites in pSKB3 to express the polypeptide MGSSHHHHHHDYDIPTTENLYFQGHMD¹⁷¹STRDD…EMVSKT³⁷⁴LEHHHHHH*. The PGRP12e made in E. coli BL21 (DE3) was isolated from supernatant of the E. coli cell lysate to raise antibodies in a rabbit as described above. The EcoRI-XhoI fragments were inserted to the same sites in pMFH6 to produce protein in baculovirus infected Sf9 cells as secreted PGRPs (12e: GIHMD¹⁷¹ST…SKT³⁷⁴LEHHHHHH, 13f: GIHMD²⁰CD…IFH⁴⁴⁶LEHHHHHH, 13N: GIHMD²⁰CD…PQW¹⁸²LEHHHHHH, and 13C: GIP¹⁸⁰QW…IFH⁴⁴⁶LEHHHHHH). In vivo transposition, transfection of Sf9 cells with bacmid DNA, large-scale infection of insect cells, and purification of the secreted PGRP4, 5, 12e, 13f, 13N, and 13C were performed as described before (Wang et al., 2011).

2.3. Measurement of DAP- and Lys-MPP binding to the recombinant PGRPs by surface plasmon resonance (SPR)

DAP- and Lys-muramyl pentapeptides (MPPs) were synthesized previously and stored at −20 °C (Hu et al., 2019). Binding interactions of the PGRP2 and PGRP3s with the ligands were examined on a Biacore T100 biosensor system (GE Healthcare) at University of Georgia, while association and dissociation kinetics of the PGRP1, 3f, 3s, 4, 5, 12e, 13f with DAP-MPP and Lys-MPP were studied on a Biacore T200 at Oklahoma Medical Research Foundation. As described before (Hu et al., 2019), the proteins were immobilized by standard amine coupling; HBS-EP (pH 7.4, 10 mM HEPES, 150 mM NaCl, 3 mM EDTA, 0.005% polysorbate 20) was used as running buffer. Analyte in the buffer was employed at a flow rate of 30 μl/min for association and dissociation at 25 °C. Using Biacore evaluation software, response curves of DAP- and Lys-MPPs at various concentrations were globally fitted to the 1:1 and two-state binding models to calculate dissociation constants, and the ones with lower χ² values (i.e. better fitting) were reported.

2.4. Enzyme-linked immunosorbent assay (ELISA) of PGRP binding to soluble peptidoglycans

Soluble peptidoglycans (InvivoGen) from E. coli and Staphylococcus aureus were used as ligands to test binding of M. sexta PGRP2–5, 3s, 12e, 13f, 13N, and 13C as described before (Sumathipala and Jiang, 2010). Briefly, each ligand (2 μg/well) was fixed on a 96-well microplate by drying overnight and baking at 60 °C for 30 min. After blocking with 200 μl of 1 mg/ml bovine serum albumin (BSA) in Tris buffered saline (TBS) for 2 h and after washing with 200 μl TBS four times, PGRP samples (0, 200, 300, 400, and 500 ng for each) in 50 μl TBS containing 0.1 mg/ml BSA were added to the wells and incubated for 3 h at room temperature. A competition experiment was performed to test binding specificity by pre-incubating PGRP samples (200, 300, 400 and 500 ng for each PGRP) with 20 μg of a soluble peptidoglycan in 50 μl TBS with 0.1 mg/ml BSA for 1 h. The mixtures were added to the wells coated with the same ligand and incubated for 3 h at room temperature. After washing with TBS four times, 1:1000 diluted anti-(His)₅ monoclonal antibody (GenScript) in 100 μl TBS with 0.1 mg/ml BSA was added to the wells and incubated for 2 h at 37 °C. Following a washing step, 1:1500 diluted goat-anti-mouse IgG conjugated to alkaline phosphatase (Bio-Rad) in 100 μl TBS with 0.1 mg/ml BSA was added to the wells and incubated overnight at room temperature. After washing with TBS four times and 200 μl of 0.5 mM MgCl₂, 10 mM diethanolamine once, 50 μl aliquots of p-nitrophenyl phosphate (1.0 mg/ml) in the diethanolamine buffer were added to the wells. Absorbance at 405 nm was monitored for 20 min in the kinetic mode on a microplate reader and one unit of activity is defined as the amount of phosphatase causing an increase of 0.001 absorbance unit per minute at room temperature.

2.5. Binding of the recombinant PGRPs to insoluble peptidoglycans and bacterial cells

Insoluble peptidoglycans were isolated from the Gram-positive bacteria, B. megaterium, B. subtilis, Micrococcus luteus and S. aureus in the previous study (Sumathipala and Jiang, 2010). One milligram of each peptidoglycan was mixed with 10 μl of a PGRP (0.2 μg) and 40 μl of buffer A (20 mM NaCl, 20 mM Tris-HCl, pH 8.0). After incubation for 2 h at 4 °C with mixing, 10 μl of the suspension was treated with 2.5 μl 5 × SDS sample buffer as total. The remaining mixture (40 μl) was centrifuged at 6000×g for 15 min. Ten microliters of the supernatant was treated with 2.5 μl 5 × SDS buffer as unbound. The pellet was washed three times with buffer A (200 μl each) and then treated with 20 μl of 2 × SDS buffer for 5 min at 95 °C as bound. The total, unbound, and bound fractions (all equivalent to 8 μl of 1:5 diluted PGRP) were subjected to 15% SDS-PAGE and immunoblot analysis using anti-(His)₅ monoclonal antibody (GenScript), goat-anti-mouse IgG conjugated to alkaline phosphatase, BCIP and NBT substrate solutions (Bio-Rad).

Single colonies of Bacillus megaterium, Bacillus subtilis, M. luteus, S. aureus, Klebsiella pneumoniae, E. coli, Pseudomonas aeruginosa, and Salmonella typhimurium were grown in 3 ml LB medium at 37 °C until A₆₀₀ was close to 0.6. After centrifugation at 10,000×g and washing with buffer A twice, the bacteria were resuspended in 40 μl of buffer A. PGRPs (10 μl, 3 μg) were separately added to the suspensions and incubated for 2 h at 4 °C with mixing, and then 10 μl each of the suspensions was treated with 2.5 μl 5 × SDS sample buffer as total. After centrifugation at 6000×g for 15 min, 10 μl each of the supernatants was treated with 2.5 μl 5 × SDS buffer as unbound. The cell pellet was washed 3 times with 200 μl of buffer A, suspended with 20 μl of 2 × SDS buffer, and heated at 95 °C for 5 min to obtain the bound. The total, unbound, and bound fractions (all equivalent to 8 μl of 1:5 diluted PGRP) were separated by 15% SDS-PAGE followed by immunoblot analysis as described above.

2.6. Elicitor-independent proPO and proHP8 activation in plasma caused by the exogenous PGRPs

Stored cell-free hemolymph from naïve larvae was thawed and diluted ten times with buffer B (0.001% Tween-20, 1 mM CaCl₂, 20 mM Tris-HCl, pH 8.0). Then, 5 μl aliquots of the sample were added to 20 μl of the purified PGRPs or fragments (0, 200, 400, 600, 800, 1000 ng) in buffer B. As a negative control, same amounts of BSA (0–1000 ng) in 20 μl buffer B were mixed with the diluted plasma (5 μl). After 60 min incubation at room temperature, PO activities in the test and control groups were measured using dopamine as a substrate on a 96-well microplate reader (Wang et al., 2011). Absorbance at 470 nm was monitored in the kinetic mode and one unit of activity is defined as 0.001 ΔA_{470 nm}/min at 25 °C. The activity data were plotted as mean ± SD (n = 3) against concentrations of the proteins added (Sumathipala and Jiang, 2010).

Similarly, 1.0 μl aliquots of a selected plasma from naïve larvae (NP) were separately incubated with 1.0 μl buffer, BSA (0.5 μg/μl), and the recombinant proteins (PGRP1, 2, 3f, 3s, 4, 5, 12e, 13f, 13N, and 13C, 0.5 μg/μl). After total volume of the control and treatment groups was adjusted to 10 μl with buffer (20 mM Tris-HCl, pH 7.5, 0.001% Tween-20, 1 mM CaCl₂, 0.01% 1-phenol-2-thiourea), the mixtures were incubated for 30 min at room temperature. SDS-PAGE and immunoblot analysis was performed using 1:1000 diluted HP8 antiserum as the primary antibody. The antibody-antigen complexes were detected using 1:3000 diluted goat-anti-rabbit antibodies conjugated to alkaline phosphatase (Bio-Rad). The experiment was repeated twice using different NP samples.

3. Results and Discussion

3.1. Rationale for characterizing M. sexta PGRP2–5, 12, and 13

We systematically evaluated the genome, transcriptome, and proteome data to select major or interesting PGRPs in cell-free hemolymph for functional assays. Of the thirteen genes in the genome, nine encode short-type PGRPs (i.e. 1–7, 9, 13), which contain a signal peptide but no transmembrane region (Zhang et al., 2015), and their mRNA level peaked at 2,245, 1,273, 58, 181, 3,265, 28, 15, 23, and 230 FPKM (Fig. S1), respectively. PGRP1 and PGRP5 were detected in plasma of naïve larvae and at higher levels in hemolymph from larvae challenged with bacteria (He et al., 2016). PGRP2−4 mRNA levels greatly increased in fat body and hemocytes but their protein levels (spectral counts: 9, 11, and 0) were still much lower than PGRP1 (126) or PGRP5 (40) in induced plasma. Low levels of PGRP6, 7, and 13 (0.1–5.2 ppm) were detected in plasma during metamorphosis, contrasting high levels of PGRP1 (213.1–611.1 ppm) and PGRP5 (19.2–105.0 ppm) (Cao et al., 2020). Therefore, PGRP1–3 and 5 are major sensors or potential destructors of bacterial peptidoglycans in M. sexta plasma. PGRP1 was studied before (Sumathipala and Jiang, 2010; Hu et al., 2019). PGRP2, 3 and 4 were chosen because of their putative N-acetyl-muramoyl-l-alanine amidase activity. PGRP5 was selected due to its second highest abundance. The ectodomain of PGRP12, a membrane-bound protein predicted to be a receptor for DAP-PG and Imd pathway, was adopted as a control for comparison with the plasma PGRPs. Owing to its unique domain structure, M. sexta PGRP13 was examined as a full-length protein (13f), N-terminal PGRP domain (13N), and C-terminal lipoprotein domain (13C, as a negative control).

3.2. cDNA cloning, structural features, and expression profiles of M. sexta PGRP2–5, 12 and 13

M. sexta PGRP2 and PGRP3 cDNAs were cloned by library screening and amplification by PCR. Based on the genome and transcriptome data (Zhang et al., 2015), we directly amplified PGRP4 and PGRP5 cDNAs from fat body of induced larvae. They encode 188–231-residue PGRP2–5 including 17–20-residue signal peptide (Table S1). The four mature proteins have calculated M_r’s of 19.3–24.1 kDa and isoelectric points (pI) of 6.20–8.47. PGRP3’s M_r (24.1 kDa) and PGRP4’s pI (8.47) values are higher than the others’ (19.3–21.1; 6.05–6.64). PGRP3 includes full-length (f) and short (s) versions, and the latter was a result of design error (Section 2.2) but yielded valuable functional insights. The PGRP12 and 13 cDNA fragments were amplified from induced fat body and nervous tissues, respectively, for cloning and recombinant expression.

We compared the peptidoglycan-binding domains with their counterparts in the other insect PGRPs, human PGRP1αC, and T7 lysozyme to reveal key structural features (Fig. 1). Multiple sequence alignment revealed that M. sexta PGRP2–4 possess four of the five residues (His¹⁸, Tyr⁴⁷, His¹²³, Lys¹²⁹, Cys¹³¹, T7 lysozyme numbering) that interact with a catalytic zinc ion in the phage enzyme. Since substitution of Lys¹²⁹ with Thr in D. melanogaster PGRP-SB1 and -SC1a did not abolish the N-acetyl-muramoyl-l-alanine amidase activity (Mellroth et al., 2003) and PGRP2–4 have a Thr at the equivalent position, M. sexta PGRP2–4 may also hydrolyze peptidoglycans. In contrast, M. sexta PGRP1, 5, 12, and 13 lack His¹⁸, C¹³¹, and/or Tyr⁴⁷ (Fig. 1) and may not possess the amidase activity. The equivalents of Cys¹⁹, Ser²⁰, Gly⁴¹, Trp⁴², Tyr⁴⁷, Arg⁶¹, Asn⁷⁴, His¹²³, and Cys¹³¹ in PGRPs are involved in peptidoglycan binding, based on the crystal structures of Drosophila and human PGRPs (Kaneko et al., 2004; Guan et al., 2004). In particular, residues 41, 42 and 61 participate in distinguishing the two types of peptidoglycans. Human PGRP1αC has Asn, Phe and Val residues (NFV) at these sites to bind Lys- and DAP-PGs equally; T7 lysozyme has Gly, Trp and Arg residues (GWR) to hydrolyze E. coli DAP-PGs (Guan et al., 2004; Kaneko et al., 2004). The positively charged Arg may enhance DAP-PG binding through electrostatic interaction with the carboxyl group of DAP. This residue is conserved in all known Drosophila and human PGRPs that recognize DAP-PGs but seldom found in PGRPs that bind Lys-PGs (Onoe et al., 2007). Therefore, Manduca PGRP2–4 (QWR) may preferentially bind and hydrolyze DAP-PGs, whereas PGRP12 (QWR) may only do the first on the surface of immune cells. Analogous to Drosophila PGRP-SB1 (NFR) that hydrolyzes DAP-PGs, M. sexta PGRP1 (NYS) preferentially binds to DAP-PGs (Sumathipala and Jiang, 2010; Hu et al., 2019). Like human PGRP1αC (NFV), M. sexta PGRP5 (NFV) may recognize Lys- and DAP-PGs equally well.

Fig. 1. Multiple sequence alignment of peptidoglycan binding domains in the PGRPs and T7 lysozyme.

Amino acid sequences of the domains in M. sexta PGRPs (Ms1, 2, 3, 4, 5, 12, and 13), D. melanogaster PGRP-SB1 (DmSB1), PGRP-SC1a (DmSC1a), PGRP-LE (DmLE), PGRP-SA (DmSA), PGRP-SD (DmSD), B. mori PGRP1 (Bm1), T. ni PGRP (Tn), S. cynthia ricini PGRP-A (ScA), Homo sapiens PGRP1αC (Hs1αC), and T7 lysozyme are aligned. The names in bold red are shown or predicted as N-acetylmuramoyl-l-alanine amidases with the conserved His¹⁸, Tyr⁴⁷, His¹²³, K¹²⁹/T¹²⁹ and C¹³¹ (in bold red font) all present for binding the catalytic zinc ion in T7 lysozyme as well as the five insect PGRPs. Numbers on the right indicate positions of the residues in the complete PGRP sequences. For the sixteen insect PGRPs, positions with 70, 90 and 100% identities are marked with “.”, “:” and “*”, respectively, and residues different from the consensus are shaded grey. Residues highlighted yellow represent the conserved residues for peptidoglycan binding, some of which are responsible for differential recognition of DAP-PG (bold back) and Lys-PG (bold green). The predicted or confirmed specificity of peptidoglycan binding is shown on the right as D (for DAP-PG), K (for Lys-PG), D>K, D=K, or K>D.

The expression profiles of M. sexta PGRP1–7, 9, 13 in the 52 tissue samples in various life stages were examined to infer functions (Fig. S1, Zhang et al., 2015). The average FPKM values of PGRP1 (140), PGRP2 (75) and PGRP5 (132) were much higher than those of PGRP3 (7) and PGRP4 (8). PGRP3 mRNA level was 4.8-fold lower than PGRP1’s (Gunaratna and Jiang, 2013) in naïve larval fat body (C for control) and became 44-fold higher after an immune challenge (I for induced) (Table S1). The I/C ratios of PGRP1, 2, 4 and 5 transcripts were 5.1, 1,818, 58, and 54, respectively. While the infection-dependent up-regulation of PGRP1–5 indicated their relatedness to immunity, the major increase in PGRP1 and PGRP5 expression in fat body of early pupae (Fig. S1) suggested a developmental control. Infection-independent up-regulation was also observed in other immunity-related genes (e.g. antimicrobial peptides) (He et al., 2015). The PGRP2, 4 and 5 mRNA peaked in midgut of early 5^th instar larvae, wandering larvae and young pupae, respectively, supported a spatiotemporal regulation and diverse functions of these PGRPs in the digestive tract. Co-transcriptional regulation of PGRP9 and PGRP13 in head and muscle were apparent (Fig. S1), and these two genes are closely located on scaffold 00320 (Zhang et al., 2015).

The induced PGRP synthesis was observed at the protein level (Zhang et al., 2014; He et al., 2016). In contrast to the high I/C ratios of PGRP2–5 transcripts, elevations of their protein abundances were not dramatic (Table S1). The I/Cs of PGRP1 and PGRP5 were 1.2- and 2.8-fold, respectively, which are likely related to their high accumulation in control plasma and some consumption and induced synthesis during the immune response to injected bacteria. The lack of good correlations in mRNA and protein levels was documented previously (He et al., 2016). PGRP2 and PGRP3 were only detected in induced larval hemolymph, but their spectral counts were less than one tenth of PGRP1’s. In another proteomics analysis, we detected high-to-moderate levels of PGRP1 (2^8–9 ppm) and PGRP5 (2^5–7 ppm), low levels of PGRP6, 7 and 13 (2^0–2 ppm), and no PGRP2–4 in hemolymph of M. sexta prepupae, pupae, and adults (Cao et al., 2020). Together, these data indicated that PGRP1 and PGRP5 are predominant in plasma of naïve and induced larvae and may play major roles in preferential recognition of peptidoglycans.

3.3. Recombinant production of M. sexta PGRP2–5, 13, and their fragments (3s, 12e, 13N, 13C)

We selected the baculovirus-insect cell system to produce the PGRPs to avoid E. coli cell wall components that cause insect immune responses. The cDNA fragments were subcloned into pFH₆ or pMFH₆, allowing secretion of the mature proteins into conditioned media. The proteins, fused with a hexahistidine tag at the carboxyl-terminus, were captured by cation exchange chromatography on a dextran sulfate-Sepharose column and further purified by affinity chromatography on a Ni-nitrilotriacetic acid agarose column. From one liter of conditioned medium for each protein, we obtained 2.0 mg PGRP2, 3.1 mg PGRP3f, 2.0 mg PGRP3s, 2.4 mg PGRP4, 1.5 mg PGRP5, 1.5 mg PGRP12e, 3.0 mg PGRP13f, 1.0 mg PGRP13N, and 3.3 mg PGRP13C. A polyclonal antiserum was raised against purified PGRP2, and the antibodies cross-reacted with PGRP1, 3 and 4 (data not shown). The purified recombinant PGRP2. 3f, 3s, 4, 5, 12e, 13f, 13N, and 13C migrated as a single band at 19, 24, 19, 19, 20, 27, 50, 19, and 31 kDa positions (Fig. 2), close to their calculated masses (20,770, 25,230, 24,191, 22,312, 20,706, 24,722, 50,578, 19,759, and 32,484 Da), respectively. While N- and O-linked glycosylation sites are predicted (Table S1), treatment of the proteins with N- and O-glycosidases did not increase their gel mobility (data not shown), except for PGRP12e (Fig. 2E). Since these sites reside in regions flanking the binding domain, no direct impact is expected on binding to peptidoglycans or their monomers.

Fig. 2.

SDS-PAGE and immunoblot analysis of the purified M. sexta PGRP2 (A), PGRP3f and 3s (B), PGRP4 (C), PGRP5 (D), PGRP12e (E), PGRP13f, 13N and 13C (F). The PGRPs (2 μg for staining and 0.2 μg for blotting) were separated on 15% polyacrylamide gels, stained with Coomassie Brilliant Blue (left) or detected using anti-(His)₅ primary antibody (right). Sizes and positions of the pre-stained M_r standards are indicated on the left, with the 50 kDa marker highlighted red. Glycosylation of PGRP12e (1 μg, in E right) was tested by incubating it with buffer (−) or PNGase F (+) for 1 h at 37 °C, followed by SDS-PAGE and immunoblotting. The major band of PGRP12e before and after treatment is marked with arrowhead.

3.4. Binding properties of the PGRPs and their fragments

To study associations of the PGRPs and peptidoglycans, we employed an ELISA-based, semiquantitative binding assay, and detected preferential bindings of soluble DAP-PG from E. coli by PGRP2, 3f, 3s, 4, 5, 12e, 13f, and 13N in the range of 0.2–0.5 μg (Fig. 3, A–F). No specific binding was detected with PGRP13C, a negative control that lacks a peptidoglycan binding domain. In comparison, associations of the PGRPs with S. aureus Lys-PG yielded signals 2–10 fold lower than those with E. coli DAP-PG (Fig. 3, A’–F’). While total bindings of the soluble Lys-PG by PGRP4, 12e, and 13N in the range of 0.2–0.5 μg were significantly higher than those after competition, such differences for PGRP2, 3f, 3s, 5, and 13f were in most cases small or insignificant, indicating specific bindings to the Lys-PG were limited. PGRP5, with its level in plasma only lower than PGRP1, also strongly and favorably bound to the DAP-PGs. Thus, the prediction of equal binding of PGRP5 to Lys- and DAP-PGs (Section 3.2) does not happen in reality. In summary, the repertoire of soluble PGRPs in the cell-free hemolymph of M. sexta has evolved for favored detection of DAP-PG from E. coli rather than Lys-PG from S. aureus.

Fig. 3.

Binding of M. sexta PGRP2 (A, A’), PGRP3f (left), 3s (right) (B, B’), PGRP4 (C, C’), PGRP5 (D, D’), PGRP12e (E, E’), PGRP13f (left), 13N (middle), and 13C (right) (F, F’) to soluble peptidoglycans from E. coli (A–F) and S. aureus (A’–F’). As described in Section 2.4, the purified proteins (0.2–0.5 μg) were pre-incubated with buffer or 20 μg of soluble peptidoglycan, added to microplate wells immobilized with 2.0 μg of the same ligand, and incubated with primary and secondary antibodies. Binding was determined using an alkaline phosphatase substrate and shown as bars (mean ± SD, n = 3). White, dark blue, maroon, diagonal maroon stripes, green, brown, pink, cyan, diagonal cyan stripes, horizontal cyan stripes, and light purple represent BSA/− (negative control), PGRP2, 3f, 3s, 4, 5, 12e, 13f, 13N, and 13C without competitor, and all PGRPs with competitor, respectively. An asterisk (*) indicates that total binding is significantly higher (p < 0.05) than binding after competition in each pair.

We used SPR to further confirm interactions of PGRPs with chemically synthesized monomeric peptidoglycans. Since linking DAP- or Lys-MPPs through its amino groups is expected to affect binding due to improper orientation and steric hindrance (Hu et al., 2019), the purified proteins were immobilized on CM5 sensor chips, allowing the ligands to flow through the chip and freely interact with the PGRPs. As shown in Table 1, K_d values for Lys-MPP complexed with the PGRPs (8.77 to >100 μM) were 4.9 to >200 times as high as those for DAP-MPP (0.42 to 4.70 μM). In other words, DAP-MPP bound to the PGRPs stronger than Lys-MPP did, consistent with the ELISA data (Fig. 3). Association of DAP-MPP and PGRP3s (K_d: 1.75 μM) was higher than that of DAP-MPP and PGRP3f (K_d: 4.70 μM). In comparison, total and specific bindings of PGRP3f to E. coli DAP-PG were higher than those of PGRP3s, suggesting that the C-terminal change markedly affected recognition of the polymeric peptidoglycan (Fig. 3B). Both forms hardly bound S. aureus Lys-PG in a specific manner (Fig. 3B’).

Table 1.

Binding constants (K_d, μM) of PGRPs with peptidoglycan monomers^*

protein	Lys-MPP		DAP-MPP
	Kd (μM)	χ2 (RU2)	Kd (μM)	χ2 (RU2)
PGRP1	37.95 (45.58 a)	0.041 (0.184 a)	2.013 (0.57 a)	0.898 (0.503 a)
PGRP2	(>100)	n/a	(0.49)	(0.210)
PGRP3f	29.25	1.140	4.697	0.826
PGRP3s	23.81 (64.03)	0.467 (0.288)	1.749 (0.42)	1.170 (0.981)
PGRP4	79.83	0.313	4.061	2.800
PGRP5	37.80	0.292	2.010	0.979
PGRP12e	45.10 b	0.066 b	2.635	0.212
PGRP13f	8.77	0.928	1.798	0.932

^*The values in parentheses were determined using freshly prepared Lys-MPP and DAP-MPP on a Biacore T100 biosensor system at University of Georgia, whereas the other data for M. sexta PGRP1, 3f, 3s, 4, 5, 12e and 13f were obtained on a Biacore T200 system at Oklahoma Medical Research Foundation using the compounds stored at −20 °C.

^a:The data for M. sexta PGRP1 were adopted from Hu et al. (2019).

^b:Results from steady-state affinity analysis of the weak binding of Lys-MPP to PGRP12e.

The PGRP bindings were further tested using insoluble peptidoglycans from B. megaterium, B. subtilis, M. luteus, and S. aureus in a pulldown test. After the PGRPs had been separately incubated with DAP-PGs from B. megaterium and B. subtilis, more than half of PGRP2–5, 12e, 13f, and 13N were present in the bound fraction (Fig. 4A). Order of binding completeness was B. megaterium > B. subtilis > M. luteus > S. aureus peptidoglycan. Not detected was the binding of PGRP13C to any of the four peptidoglycans, PGRP13f and 13N to M. luteus and S. aureus peptidoglycans, or PGRP3f, 3s, and 12e to M. luteus peptidoglycan. These data are consistent in the most part to the ELISA results using the soluble peptidoglycans (Fig. 3). M. luteus Lys-PG appears to be an exception. Its partial or complete binding of PGRP1, 2, 4, 5, and 13N (Sumathipala and Jiang, 2010; Fig. 4A) should have a special structural reason, which is unclear currently. But we notice that, unlike most Gram-positive bacteria, the cross-bridge connecting stems of M. luteus Lys-PG is identical in sequence to the stem peptide (Vollmer et al., 2008). As such, it may not be proper to extrapolate experimental results obtained using M. luteus or its peptidoglycan to other Gram-positive bacteria with typical cross-bridges in their Lys-PGs.

Fig. 4.

Binding of M. sexta PGRPs (2, 3f, 3s, 4, 5, 12e, 13f, 13N, and 13C) to insoluble peptidoglycans (A) from B. megaterium, B. subtilis, M. luteus, and S. aureus or whole bacteria (B) of B. megaterium, B. subtilis, M. luteus, S. aureus, K. pneumoniae, E. coli, S. typhimurium, and P. aeruginosa. As described in Section 2.5, binding experiments were performed using the purified protein and peptidoglycans or cells. The total (T), unbound (U), and bound (B) fractions were separated by 15% SDS-PAGE followed by immunoblot analysis using anti-(His)₅ as the primary antibody.

PGRP bindings of the eight whole bacteria resembled bindings of the isolated peptidoglycans but displayed more complex patterns (Fig. 4B). In general, order of binding completeness of PGRP2–5, 13f, and 13N remained the same: B. megaterium > B. subtilis > M. luteus > S. aureus. This is likely because their peptidoglycans are exposed on the surface and the PGRPs mainly recognize DAP-PGs of Bacillus, a genus of Gram-positive bacteria. In contrast, as DAP-PGs are hidden under outer membrane of the Gram-negative bacteria, no or low bindings by PGRP2–5, 12e, 13f, and 13N were common. The complete binding of PGRP2 to K. pneumonia and partial association of PGRP3s with the Gram-negative bacterium are interesting exceptions. M. sexta PGRP2 and PGRP3s could be employed to probe what structural moieties of the Gram-negative bacterial surface are responsible for the observed bindings.

3.5. Possible roles of M. sexta PGRP2–5, and 13 in proPO and Toll pathway activation

Knowing that the recombinant PGRP1 alone caused proPO activation in hemolymph from naïve larvae (Sumathipala and Jiang, 2010), we performed the same experiment by mixing plasma with different amounts of the purified PGRP2–5, 12e, and 13, and observed concentration-dependent PO activity increases in the absence of a bacterial elicitor. At 40 ng/μl, PGRP2, 3f, 4, 5, 13f, and 13N induced 10.5, 13.3, 11.8, 15.4, 10.6, 10.4 U of PO activity (Fig. 5), respectively. The PO activity levels were significantly higher than the negative controls of BSA (0.1 U) and PGRP13C (0.2 U). Low but substantial PO activity increases were observed after PGRP3s or 12e (≥24 ng/μl) had been incubated with the diluted plasma and, at 40 ng/μl, PO activities reached 5.2 and 3.1 U, respectively. Therefore, the recombinant PGRP1–5, 12e, 13f, and 13N (but not 13C) have an inherent tendency to trigger the proPO activation system, even though PGRP12, a membrane-bound receptor, is not anticipated to be a part of the humoral immune response. The C-terminal truncation of PGRP3f considerably impaired this tendency, despite that the altered region is 33 residues away from the peptidoglycan-binding domain. Unlike the other secreted PGRPs, PGRP3 has a 46-residue tail, a lot longer than the other six’s (range: 16–30, average: 22).

Fig. 5.

Elicitor-independent proPO activation in hemolymph from naïve larvae after incubating with the recombinant M. sexta PGRPs. As described in Section 2.6, diluted plasma (5 μl) was incubated at room temperature with 0, 0.2, 0.4, 0.6, 0.8, and 1.0 μg of BSA or purified PGRPs (2, 3f, 3s, 4, 5, 12e, 13f, 13N, and 13C) in a final volume of 25 μl. PO activity was measured after 60 min and plotted as mean ± SD (n = 3) against amount of the proteins added.

Since endogenous PGRPs do not lead to spontaneous melanization, proPO activation elicited by the exogenous PGRPs alone may represent an artifact to an extent, yet this phenomenon yields mechanistic insights. It indicates that recognition of peptidoglycans by PGRP1–5 or 13 may enhance immune responses through serine proteases, at least for the proteolytic activation of proPO. In M. sexta, melanization is mediated by an integrated serine protease system that also induces Toll pathway via HP14-HP21-HP5-HP6-HP8-Spätzle (Wang et al., 2020). Because M. sexta HP14 is the only proven initiating protease, it is worth testing if proHP14 autoactivation is due to the PGRPs’ binding to peptidoglycans, as shown in the case of PGRP1 and microbe binding protein (Wang et al., 2017). In particular, DAP-PG recognized by PGRP1–5 or 13 may induce the Toll pathway, and this would radically differ from the prevailing model derived from Drosophila research, in which Lys-PGs of Gram-positive bacteria lead to the Toll pathway activation.

To test the hypothesis, we examined if there is an increase in proHP8 proteolytic activation in hemolymph from naïve larvae after incubation with the PGRPs or their fragments in the absence of peptidoglycan. HP8 cleaves proSpätzle-1 to form Spätzle-C108 dimer, a likely ligand of a Toll receptor in M. sexta (An et al., 2009 and 2010). Inclusion of PGRP2, 3f, 4, 5, and 13N led to complete HP8 activation, while BSA had no effect (Fig. 6). In comparison, PGRP1, 3s, and 13f caused lower conversion of proHP8. Surprisingly, PGRP13C didn’t led to proPO activation (Fig. 5) but somehow caused significant proHP8 cleavage activation. PGRP12e treatment led to a low PO activity but HP8 activation was complete (Fig. 6), and we observed two additional bands at about 70 and 30 kDa on the blot, which were not found in the other ten reactions. Because the experiment was repeated twice using different plasma samples, further studies are demanded to explore mechanisms underlying the interesting differences between proPO and proHP8 activation. In summary, these results suggest connections between PGRPs and Toll pathway activation in M. sexta. The roles of abundant PGRP1 in plasma, C-terminal extension of PGRP3, and the lipoprotein domain of PGRP13 in affecting HP8 activation are worth exploring in the future, and so is PGRP12’s predicted function in activation of an Imd pathway in M. sexta (Cao et al., 2015).

Fig. 6.

Cleavage activation of M. sexta proHP8 in larval hemolymph after incubation with the recombinant PGRPs. As described in Section 2.6, cell-free hemolymph samples from naïve larvae (NP) were separately incubated for 30 min with buffer, 0.5 μg of BSA or purified PGRPs. The control, BSA-, and PGRP-treated samples were analyzed by 12% SDS-PAGE and immunoblotting using HP8 antibody. The proHP8 and HP8 doublets are labeled by asterisk (*) and arrow, respectively. Uncharacterized 70 and 30 kDa bands in the PGRP12e lane are marked by “?”.

3.6. Peptidoglycan recognition and immune signaling

Genetic research of innate immunity in Drosophila has greatly stimulated immunological studies in other insects and beyond. Identification of the Toll/TLR and Imd/TNF-α pathways in the fly and human demonstrated the evolutionary conservation of innate immunity in animals (Hoffmann and Reichhart, 2002). Extracellular serine protease networks are also conserved to some extent in holometabolous insects (Cao and Jiang, 2018), which are analogous to the blood clotting and complement activation systems in mammals. Upon recognition of microbial surface molecules, a protease network is activated in insect plasma to generate POs, Spätzle, and stress responsive peptides via sequential limited proteolysis (Kanost and Jiang, 2015; Schrag et al., 2017). While POs catalyze melanin formation, the cytokines may induce the intracellular Toll and JNK pathways (Tsuzuki et al., 2012). In Drosophila, likely because PGRP-SA is a major sensor of Lys-PGs in plasma, recognition of Gram-positive bacteria leads to the Toll pathway activation (Michel et al., 2001). Similarly, GNBP3 binding to fungal β-1,3-glucan triggers the serine protease system to produce Spätzle for Toll signaling (Buchon et al., 2009). While these results explain why Gram-positive bacteria and fungi favorably trigger the Toll pathway in Drosophila, it is unclear whether or not this is true in other insects. Nevertheless, it is frequently claimed that, in insects, Gram-positive bacteria and fungi induce the Toll pathway, whereas Gram-negative bacteria induce the Imd pathway. To clarify the situation in M. sexta, we first examined the structure and function of PGRP1 and found that the most abundant PGRP preferentially binds to DAP-PGs (Hu et al., 2019; Sumathipala and Jiang, 2010). In this study, we systematically characterized the other soluble PGRPs (2–5, 13), PGRP variants (3s, 13f, 13N), and controls (PGRP12e, 13C) and obtained consistent results that lead to the following questions. With the major PGRPs favorably binding to DAP-PGs, shouldn’t Gram-negative and DAP-type Gram-positive bacteria preferentially trigger the Toll pathway in M. sexta? How about insects in Lepidoptera and other orders with a similar repertoire of DAP-PG-specific PGRPs?

We searched GenBank using M. sexta PGRP1 and PGRP5 sequences as queries and identified 43 nonredundant hits from 33 species in the order of Lepidoptera (Table S2). Only Antheraea mylitta, Antheraea pernyi, Helicoverpa armigera, and Ostrinia furnacalis each has two paralogs for certain. Suppose the close homologs of PGRP1 are also most abundant in the other 29 insects, as shown in B. mori (Ochiai and Ashida, 1999), Trichoplusia ni (Kang et al., 1998) and Samia cynthia ricini (Onoe et al., 2007), their binding specificities may predict how Toll pathways in these insects respond to Lys-PGs vs. DAP-PGs. Because M. sexta PGRP1 (NYS) and PGRP5 (NFV), T. ni PGRP (NYA), and S. cynthia PGRP-A (KYS) preferentially bind DAP-PGs (Kang et al., 1998; Onoe et al., 2007), we speculate on the basis of their specificity determinants that 37 of the 38 PGRPs favors DAP-PGs. Only O. furnacalis XP_028160363.1 (DWA) may better bind to Lys-PGs. An extended search of Insecta using D. melanogaster PGRP-SA (DFT) yielded 360 hits in twelve orders (Table S2). In the orders of Hymenoptera (93/103), Lepidoptera (59/61) and Blattodea (7/7), most of the hits are predicted to better bind DAP-PGs. Three bee PGRP-SAs indeed bind to DAP-PGs preferentially (Liu et al., 2019). In Diptera (62/121) and Coleoptera (27/45), about half of the hits have D/E at the first site which may favorably bind Lys-PGs through charge interaction. Therefore, it seems appropriate to suggest that Toll pathways are mainly induced by Lys-PGs from most Gram-positive bacteria in some insects, such as Drosophila, and by DAP-PGs in other species including M. sexta. Future studies are needed to test this new hypothesis in the Class level.

The prevailing model represents a simplification of the experimental data mainly from Drosophila (Hultmark, 2003; Kurata, 2014). The preferential or discriminatory activation of the Toll and IMD pathways has a limit apparently. Note that Drosophila PGRP-SA also binds to DAP-PGs of E. coli and Pseudomonas aeruginosa, although less than Lys-PGs of M. luteus and Enterococcus faecalis (Chang et al., 2004). DAP-PGs and Gram-negative bacteria do stimulate the Toll pathway and expression of Drosomycin, yet at a lower level than Gram-positive bacteria (Leulier et al., 2003). Genetic removal of teichoic acids in cell wall of S. aureus and B. subtilis may have reduced the structural distinction between Lys- and DAP-PGs sensed by PGRP-SA and -LC (Vaz et al., 2019). PGRP-SA seemed to recognize DAP-PGs equally well whereas PGRP-LC also strongly responded to Lys-PGs. In this context, we hope our studies of the major soluble PGRPs in M. sexta hemolymph could stimulate research into the preferential binding of bacterial peptidoglycans in other insect species.

Supplementary Material

FigS1

Fig. S1. Expression of the nine short-type PGRPs in 52 tissue samples of M. sexta. Relative mRNA levels of the PGRP1–7, 9, and 13 (Zhang et al., 2015) are shown in the bar graphs. Based on the highest value of fragments per kilobase of transcript per million mapped reads (FPKM) in each, the graphs are divided into categories of high (0–3000, purple), middle (0–300, green), and low (0-30, grey) levels of expression. The cDNA libraries (1 through 52) were constructed using tissues at the following stages: head [1. 2^nd (instar) L (larvae), d1 (day 1); 2. 3^rd L, d1; 3. 4^th L, d0.5; 4. 4^th L, late; 5. 5^th L, d0.5; 6. 5^th L, d2; 7. 5^th L, pre-W (pre-wandering); 8. P (pupae), late; 9. A (adults), d1; 10. A, d3; 11. A, d7], fat body (12. 4^th L, late; 13. 5^th L, d1; 14. 5^th L, pre-W; 15. 5^th L, W; 16. P, d1-3; 17. P, d15-18; 18. A, d1-3; 19. A, d7-9), whole animals [20. E (embryos), 3h; 21. E, late; 22. 1^st L; 23. 2^nd L; 24. 3^rd L), midgut (25. 2^nd L; 26. 3^rd L; 27. 4^th L, 12h; 28. 4^th L, late; 29. 5^th L, 1-3h; 30. 5^th L, 24h; 31. 5^th L, pre-W; 32–33. 5^th L, W; 34. P, d1; 35. P, d15-18; 36. A, d3-5; 37. 4^th L, 0h), Malpighian tubules (MT) (38. 5^th L, pre-W; 39. A, d1; 40. A, d3), muscle (41. 4^th L, late; 42–43. 5^th L, 12h; 44–45. 5^th L, pre-W; 46–47. 5^th L, W), testis (Ts, 48. P, d3; 49. P, d15-18; 50. A, d1-3), and ovary (Ov, 51. P, d15-18; 52. A, d1). Four long-type PGRPs (8, 10–12) all contain a transmembrane region.

The abbreviations used are:

BSA

bovine serum albumin

DAP

meso-diaminopimelic acid

ELISA

enzyme-linked immunosorbent assay

MBP

microbe binding protein

MPP

muramyl pentapeptide

NAM

N-acetylmuramate

NP

cell-free hemolymph (i.e. plasma) from näive larvae

PG and PGRP

peptidoglycan and its recognition protein

PO and proPO

phenoloxidase and its precursor

SDS-PAGE

sodium dodecyl sulfate-polyacrylamide gel electrophoresis

SPR

surface plasmon resonance

TBS

Tris buffered saline