A Novel Protein That Binds Juvenile Hormone Esterase in Fat Body Tissue and Pericardial Cells of the Tobacco HornwormManduca sexta L.*

Juvenile hormone esterase degrades juvenile hormone, which acts in conjunction with ecdysteroids to control gene expression in insects. Circulating juvenile hormone esterase is removed from insect blood by pericardial cells and degraded in lysosomes. In experiments designed to characterize proteins involved in the degradation of juvenile hormone esterase, a pericardial cell cDNA phage display library derived from the tobacco hornworm mothManduca sexta L. was constructed and screened for proteins that bind juvenile hormone esterase. A 732-base pair cDNA encoding a novel 29-kDa protein (P29) was isolated. Western and Northern analyses indicated that P29 is present in both pericardial cell and fat body tissues and is expressed in each larval instar. In immunoprecipitation experiments, P29 bound injected recombinant juvenile hormone esterase taken up by pericardial cells and nativeM. sexta juvenile hormone esterase in fat body tissue, where the enzyme is synthesized. Binding assays showed that P29 bound juvenile hormone esterase more strongly than it did a mutant form of the enzyme with mutations that perturb lysosomal targeting. Based on these data, we propose that P29 functions in pericardial cells to facilitate lysosomal degradation of juvenile hormone esterase.

Juvenile hormone esterase degrades juvenile hormone, which acts in conjunction with ecdysteroids to control gene expression in insects. Circulating juvenile hormone esterase is removed from insect blood by pericardial cells and degraded in lysosomes. In experiments designed to characterize proteins involved in the degradation of juvenile hormone esterase, a pericardial cell cDNA phage display library derived from the tobacco hornworm moth Manduca sexta L. was constructed and screened for proteins that bind juvenile hormone esterase. A 732-base pair cDNA encoding a novel 29-kDa protein (P29) was isolated. Western and Northern analyses indicated that P29 is present in both pericardial cell and fat body tissues and is expressed in each larval instar. In immunoprecipitation experiments, P29 bound injected recombinant juvenile hormone esterase taken up by pericardial cells and native M. sexta juvenile hormone esterase in fat body tissue, where the enzyme is synthesized. Binding assays showed that P29 bound juvenile hormone esterase more strongly than it did a mutant form of the enzyme with mutations that perturb lysosomal targeting. Based on these data, we propose that P29 functions in pericardial cells to facilitate lysosomal degradation of juvenile hormone esterase.
Juvenile hormone esterase (JHE 1 ; EC 3.1.1.1) is critical to insect development through its action on JH, which regulates gene expression. In many insects, JHE is the predominant anti-JH enzyme found in the hemolymph (blood); JHE hydrolyzes JH to produce JH acid and thereby regulates the titer of circulating JH (1). The importance of precisely regulated JH and JHE in insect development has been demonstrated by topical application of JH analogs or the JHE inhibitor 3-noctylthio-1,1,1-trifluoro-2-propanone to Lepidoptera (butterflies and moths), which can result in production of giant larvae (2). Conversely, larval development is impeded by removal of the corpora allata, which synthesize JH.
During development of lepidopteran larvae, the titers of JHE and JH are inversely regulated; hemolymph JH titers are high when JHE titers are low and vice versa. The titer of circulating JHE is regulated in part by differential transcription rates in fat body tissue (3). JHE is also cleared from the hemolymph by pericardial cells (4,5) via receptor-mediated endocytosis and is degraded in lysosomes (6 -9). The molecular processes involved in the processing and degradation of JHE in pericardial cells are unknown.
In earlier work, immunoelectron micrographs showed that targeting of JHE to lysosomes in pericardial cells was perturbed when two lysine residues of JHE (Lys 29 and Lys 524 ) were mutated to arginines (10). The present study was undertaken to identify proteins that differentially bind JHE and the mutant enzyme JHE K29R/K524R and that may be involved in endocytosis, sorting, and trafficking to lysosomes. Here, the results of screening a Manduca sexta pericardial cell cDNA phage display library for proteins that bind JHE are described.

EXPERIMENTAL PROCEDURES
General Methods-Total RNA and mRNA were isolated using a guanidium-based method (11) and the Micro Poly(A) Pure mRNA purification kit (Ambion Inc.), respectively. All proteins blotted from SDSpolyacrylamide gel for Western analysis were transferred to Hybond-P membrane (Amersham Pharmacia Biotech), and the secondary antibody (HRP-conjugated IgG) was detected using one-step 3,3Ј,5,5Ј-tetramethylbenzidine (Pierce).
Construction of the cDNA Phage Display Vector pBJuFo-Plasmid pBJuFo is shown in Fig. 1. A DNA fragment encoding a Jun leucine zipper domain fused to fd phage coat protein gene III (GenBank TM /EBI accession number J02448) and a leader sequence fused to the Fos leucine zipper domain was a generous gift from R. Crameri (17,18). EcoRV and NotI sites were added to the 5Ј-and 3Ј-ends, respectively, by PCR using the primers JF5ЈRV (5Ј-GGGATATCTTCTATTCAAG-GAGACAGTCATAG-3Ј) and JF3ЈNot (5Ј-CCGCGGCCGCACCACCG-CAACCACCGTGTGCCGCC-3Ј) prior to cloning into pCR2.1TOPO (Invitrogen). The resulting insert was isolated by digestion with EcoRV and NotI and cloned into pcDNA2.1 (Invitrogen), which had previously been digested with KpnI, blunt-ended by end filling with Klenow, and digested with NotI. The sequence encoding the gene III leader was constructed using overlapping oligonucleotides and inserted 5Ј to the jun leucine zipper region at the HindIII site. This step replaced the pelB leader sequence that was present in the original fragment with the gene III leader sequence. Next, a V5 epitope tag with a small 3Ј-multiple cloning site was constructed using the same technique and inserted downstream from the fos leucine zipper sequence into the NotI site to produce pBJuFo (see Fig. 1). All constructs were confirmed by sequencing.
Construction and Enrichment of the Phage Display Library-Pericardial cell complexes (pericardial cells and associated dorsal aortas) were dissected from 50 M. sexta larvae at the fifth instar (day 2 or day 3). Total RNA and mRNA were extracted (see "General Methods"), and cDNA was synthesized (Smart PCR cDNA synthesis kit, CLONTECH). First-strand synthesis was conducted using reverse transcriptase (Promega) with the oligo(T) NotI primer (Invitrogen). Second-strand synthesis was conducted using the Capswitch primer (CLONTECH) for synthesis of full-length cDNAs and the Advantage PCR kit (CLON-TECH). The PCR products were treated with T7 DNA polymerase; ligated to BstXI adaptors (Invitrogen); size-selected for Ͼ400 bp (on Size-Sep400 spin columns, Amersham Pharmacia Biotech); digested with NotI; and ligated into the phage display vector pBJuFo, which had previously been restricted with BstXI and NotI. Escherichia coli strain XL-1 Blue (Stratagene) was transformed with the ligation mixture. An aliquot of the recombinant E. coli cells was plated on Luria broth/ ampicillin plates for overnight incubation. Recombinant plasmids were isolated and restricted with EcoRI and NotI to show the range of cDNA insert sizes in pBJuFo. The recombinant E. coli cells were then infected with the helper phage vector cloning system M13 (Stratagene) to generate a large-scale recombinant phage expression library, which was stored at Ϫ70°C.
The phage display library was enriched by biopanning as described (18). One well of a polystyrene 24-well microtiter plate (Falcon) was coated with JHE (3 g in 300 l of 0.1 M sodium bicarbonate, pH 8.6), and recombinant phage (ϳ2.5 ϫ 10 7 plaque-forming units in 250 l) were added. After binding of phage and removal of unbound phage by washing with TBST (25 mM Tris, 3 mM KCl, 150 mM NaCl, and 0.01% Tween 20, pH 7.4), bound phage were eluted. For the first three rounds of screening, phage were eluted in acidic buffer (300 l of 50 mM HCl/glycine, pH 2.2, per well). For the fourth round of screening, phage were eluted with JHE (7.5 g of JHE in 150 l of PBS for 15 min). Fifty l of recombinant phage eluted after the fourth round of enrichment were used to infect E. coli cells. After overnight incubation on Luria broth/ampicillin plates, individual colonies were picked to test binding of specific recombinant phage to JHE on 96-well plates by enzymelinked immunosorbent assay.
Screening of the Phage Display Library-JHE (1 g in 100 l of 0.1 M sodium bicarbonate, pH 8.6 per well) was adsorbed to the solid phase of alternate rows on 96-well microtiter plates. Recombinant phage isolated from individual E. coli colonies were added to adjacent wells with or without recombinant JHE and incubated (3-4 h). Unbound phage were removed by washing in TBST, and bound phage were detected by enzyme-linked immunosorbent assay using anti-M13 antiserum (Amersham Pharmacia Biotech) conjugated to HRP. HRP activity on the substrate ABTS (2,2Ј-azinobis(3-ethylbenzthiazoline-6-sulfonic acid) diammonium salt; Amersham Pharmacia Biotech) was quantified at 412 nm. Wells precoated with monoclonal anti-M13 antiserum or skimmed milk were used as positive and negative controls, respectively. Wells with optical density readings of Ͼ2ϫ background levels were considered to be positive, and these clones were screened by enzyme-linked immunosorbent assay a second time. Plasmids were then isolated from positive clones and subjected to restriction analysis with NotI, EcoRI, and HindIII. Clones with dissimilar restriction enzyme fragments were selected for DNA sequencing using Applied Biosystems 377 automated DNA sequencing technology. Sequences were compared with those in GenBank TM /EBI Data Bank using BLAST (19). Protein motifs were identified, and the isoelectric point was determined from the deduced amino acid sequence using the program MOTIF (Genome Net). Preparations of pericardial cell protein-gene III fusions were isolated from the periplasm of selected clones, separated by SDS-PAGE, transferred to membrane, and probed with anti-V5 antibody (Invitrogen; see "General Methods") to confirm the presence of larger proteins (rather than peptides) that bind JHE.
Expression and Purification of Recombinant Juvenile Hormone Esterase-binding Proteins-The insert from a selected clone (pBJuFo.56) was restricted with BstXI and NotI, directionally cloned into the T7 polyhistidine expression vector pRSET-JF (Invitrogen) to produce pR-SET-JF.56, and transformed into E. coli BL21(DE3). Transformants were induced for 3 h (0.5 mM isopropyl-␤-D-thiogalactopyranoside), harvested, and lysed, and the recombinant protein was bound to a nickel column. Protein was then eluted in 50 -200 mM imidazole according to the manufacturer's directions (Invitrogen). The purified recombinant binding protein (P29) was separated by SDS-PAGE, electroblotted onto membrane, and detected with anti-polyhistidine primary antibody (anti-Xpress, Invitrogen; see "General Methods"). Purified P29 was used for production of polyclonal antisera in mice as described (20).
Analysis of JHE Binding by P29 in Vitro-Purified JHE and P29 were labeled with biotin (biotin labeling kit, Roche Molecular Biochemicals), column-purified on streptavidin to eliminate non-biotinylated protein, and quantified (Bio-Rad protein assay). Biotinylated proteins were separated by SDS-PAGE, transferred to Hybond-P membrane, and examined using streptavidin-HRP conjugate and the ECL chemiluminescence substrate luminol (Amersham Pharmacia Biotech). Fluorescence was detected by film exposure (Eastman Kodak Co.).
The binding of JHE to pericardial cell proteins and recombinant P29 was examined by ligand blotting. Pericardial cell complexes were dissected from larvae of M. sexta (fifth instar, day 3); homogenized in PBS, pH 7.4, supplemented with 10 mM EDTA and 10 mM phenylmethylsulfonyl fluoride; and centrifuged at 5200 ϫ g for 10 min. The supernatant was used for ligand blot analysis. E. coli samples from recombinant BL21(DE3) cells transformed with pRSET-JF.56 were sonicated for 2 min in PBS, pH 7.4, supplemented with 10 mM EDTA and 10 mM phenylmethylsulfonyl fluoride and centrifuged at 5200 ϫ g for 10 min. Protein concentrations were determined (Bio-Rad), and proteins were separated by SDS-PAGE and electroblotted onto Hybond-P membrane. Blots were incubated for 4 h with biotin-labeled JHE (2 g/ml) in PBS, washed with PBS and 0.1% Tween 20, and then blocked with skimmed milk prior to detection with streptavidin-HRP conjugate and one-step 3,3Ј,5,5Ј-tetramethylbenzidine.
For immunoprecipitation experiments, biotin-labeled JHE (50 l, 3.3 g) and biotin-labeled P29 (50 l, 1.4 g) were mixed and incubated at 37°C for 2 h. Anti-JHE or anti-Xpress antiserum (2 l) was added; the reaction was incubated on ice for 2 h; and Affi-Gel-protein A (200 l: Bio-Rad) was added to precipitate immune complexes. The immune complexes were washed (2 ml of PBS); pelleted by centrifugation at 10,600 ϫ g for 10 min; and then treated with 0.1 M sodium citrate, pH 3.0, to release proteins from the affinity gel. Samples were pelleted at 10,600 ϫ g for 5 min, and proteins in the supernatant were separated by SDS-PAGE (12% gel) and transferred to Hybond-P membrane. Biotinylated proteins were detected as described above. For positive controls, purified JHE was immunoprecipitated with anti-JHE antiserum, and P29 was immunoprecipitated with anti-Xpress antiserum. In negative control reactions, immunoprecipitation reactions contained JHE with anti-Xpress antiserum or P29 with anti-JHE antiserum.
Analysis of Expression and JHE Binding of P29 in Vivo-Pericardial cell and fat body proteins were separated by SDS-PAGE, transferred to membrane, and probed with primary antiserum raised against P29 (see "General Methods"). Poly(A) ϩ mRNAs (see "General Methods") from third, fourth, and fifth instar larvae of M. sexta were separated on a 2.2 M formaldehyde-containing 1% agarose gel; transferred to nitrocellulose; and probed with a biotinylated P29 coding sequence under high stringency conditions (21). The 1.3-kilobase biotin-labeled probe was prepared from pBJuFo.56 template by PCR with primers flanking the P29 coding sequence (forward primer PhD, 5Ј-GCGGCACACGGTGGT-TGC-3Ј; and reverse primer T7, 5Ј-AATACGACTCACTATAG-3Ј). As a negative control, a second probe was amplified by PCR using the T7 and PhD primers and an irrelevant cDNA insert (including a poly(A) tail) in pBJuFo. Biotinylated probe bound to mRNA on the membrane was detected using streptavidin-HRP with the ECL chemiluminescence substrate.
Larvae of M. sexta (fifth instar, day 3) were cooled on ice and injected with 10 g of biotinylated JHE or 10 g of bovine serum albumin, and pericardial cell and fat body tissues were dissected 1 h after injection. Tissues were homogenized on ice in 20 mM Tris-HCl, pH 6.8, 150 mM NaCl, 1 mM EDTA, and 10 mM phenylmethylsulfonyl fluoride and centrifuged at 5200 ϫ g for 5 min. Anti-P29 antiserum (1 l) was added to proteins in the supernatant, followed by immunoprecipitation with Affi-Gel-protein A. Proteins in the immunoprecipitate were separated by SDS-PAGE and transferred to membrane for detection of biotinlabeled JHE. For tissue samples from insects injected with bovine serum albumin (n ϭ 3), precipitated native JHE was detected by radiochemical assay (15).
Binding of P29 to JHE Mutants-The degree of biotinylation of JHE K29R, JHE K524R, and JHE K29R/K524R (purified and biotinylated as described above) was quantified by colorimetric assay at 412 nm in a microtiter plate using streptavidin-HRP conjugate with ABTS substrate. All assays (50 l of 2 g of enzyme/ml of stock per well) were replicated four times. Data were analyzed by one-way ANOVA.
A competition experiment was conducted to quantify the extent of binding of the JHE mutants to P29. Purified P29 was attached to a microtiter plate at different concentrations (0.75, 1.5, and 3 g/well).
Biotinylated JHE or mutant JHE (200 ng) in PBS, pH 7.4, was added. Bound enzyme was detected using streptavidin-HRP with ABTS at 412 nm. Five replicate assays were carried out, and data were analyzed by one-way ANOVA and Tukey's test for pairwise comparisons.

Isolation of Juvenile Hormone Esterase-binding Proteins from the Phage Display Library-
The phage display vector pBJuFo ( Fig. 1) was constructed for expression of pericardial cell-derived proteins as recombinant proteins fused to phage coat protein gene III and displayed on the surface of recombinant phage as a result of the interaction of Fos and Jun. Total RNA (50 g) extracted from 50 pericardial cell complexes was used to produce cDNA for the phage display library. After amplification, the size of the phage display library was ϳ10 8 plaque-forming units/ml (5 ml). Purified recombinant JHE was used to enrich the pericardial cell cDNA phage display library for proteins that bind JHE. After five rounds of enrichment, 287 individual clones were screened by enzyme-linked immunosorbent assay for JHE binding. Of these, 46 clones (16%) were positive for apparent JHE binding and did not bind to wells that were blocked with skimmed milk in the absence of JHE.
Plasmids isolated from the 46 clones were subjected to restriction analysis. The pBJuFo cDNA inserts were from 400 to 1000 bp in size. Nine clones with dissimilar restriction patterns were sequenced. One clone, pBJuFo.56, contained an 830-bp insert (Fig. 2) that included a 732-bp open reading frame that codes for the protein P29. P29 is predicted to contain 243 residues and to have a mass of 28,450 Da and an isoelectric point of 8.72. This protein has six potential phosphorylation sites and one potential myristoylation site (Fig. 2). Searches of the protein and DNA data bases failed to identify sequences related to P29.
Of the eight remaining clones, the cDNA sequences of three clones shared homology with known M. sexta genes (glutathione S-transferase (22), cytochrome oxidase (23), and 16 S RNA). The sequences of three other clones contained no poly(A) sequences. The DNA sequences of the two remaining clones did not share homology with previously published sequences and did not contain open reading frames. Therefore, characterization of these proteins was discontinued.
Purification and Binding Characteristics of P29 -Recombinant His-tagged P29 migrated at 29 kDa and was purified from transformed E. coli on a nickel column with elution at 150 mM imidazole (Fig. 3A, lanes 2 and 3). Antisera raised against recombinant P29 detected the recombinant 29-kDa protein and showed low background cross-reactivity to other E. coli proteins (Fig. 3C, lanes 1-3). A 29-kDa protein was detected in both pericardial cell and fat body tissues by Western blot analysis using anti-P29 antiserum (Fig. 3C, lanes 4 and 5).
Biotinylated JHE bound both crude and purified recombinant P29 (Fig. 3B, lanes 2 and 3). Of greater biological importance, biotinylated JHE bound to a 29-kDa protein in pericardial cell extracts (Fig. 3B, lane 4) as well as in fat body tissue (data not shown). Biotinylated JHE also bound to pericardial cell proteins of 75, 125, and 240 kDa (Fig. 3B, lane 4). The 29-kDa protein was detected by ligand blotting in all five instars of M. sexta (data not shown).
Northern blot analysis of RNA derived from pericardial cells showed a P29 mRNA of 1.1 kilobases. This P29 mRNA was present in M. sexta pericardial cells during the third, fourth, and fifth instars (Fig. 4). No signal was detected for the control Proteins separated by SDS-PAGE and transferred to Hybond-P membrane were probed with biotin-labeled JHE (2 g/ml). C, Western blotting of samples with anti-P29 antiserum. Proteins were transferred to Hybond-P membrane and detected using anti-P29 antiserum and anti-mouse IgG conjugated to HRP with one-step 3,3Ј,5,5Ј-tetramethylbenzidine.
probe that was generated from pBJuFo with an irrelevant polyadenylated insert (data not shown).
The binding of P29 to JHE in solution was demonstrated by immunoprecipitation of the two biotinylated proteins using both anti-Xpress and anti-JHE antisera (Fig. 5A). In the control reactions, P29 was not immunoprecipitated by anti-JHE antiserum, and JHE was not immunoprecipitated by anti-Xpress antiserum (data not shown).
Following injection of biotinylated JHE into M. sexta larvae and immunoprecipitation with anti-P29 antiserum, biotinylated JHE and several proteins of lower relative molecular mass were precipitated from pericardial cells (Fig. 5B, lane 3), but not from fat body tissue (lane 1). The proteins with lower relative molecular mass are presumed to be fragments of JHE produced by degradation in lysosomes. Native M. sexta JHE was immunoprecipitated from fat body tissue following injection with bovine serum albumin and detected by radiochemical assay (15). Total activity detected in the immunoprecipitates from fat body tissue was 3.97 Ϯ 2.6 nM JH hydrolyzed per min (n ϭ 3). JHE activity in the immunoprecipitates from pericardial cell tissue of bovine serum albumin-injected larvae was not above background levels for the assay (15).
Binding of P29 to JHE Mutants-Mutants JHE K29R, JHE K524R, and JHE K29R/K524R were purified and biotinylated (Fig. 6). There were no significant differences between enzymes in the efficiency of biotinylation (p Ͼ 0.05; one-way ANOVA). P29 was attached at different concentrations to the wells of a microtiter plate, and biotin-labeled JHE or mutant JHE was added (Fig. 7). Analysis of the binding of biotin-labeled enzymes to P29 showed that binding of JHE K29R/K524R was significantly less than that of JHE at 1.5 and 3 g of P29 added per well (p Ͻ 0.05; one-way ANOVA and Tukey's pairwise comparisons). There were no significant differences between the binding of JHE and mutants JHE K29R and JHE K524R (p Ͼ 0.05; one-way ANOVA and Tukey's pairwise comparisons).

DISCUSSION
Lepidopteran insects regulate titers of JHE to achieve the regulation of JH in the hemolymph that is required for development. Hemolymph JHE titers are regulated in part through clearance via receptor-mediated endocytosis into pericardial cells. The endocytosed enzyme is targeted to lysosomes for degradation. In earlier work, we observed that mutation of two residues in JHE perturbed lysosomal targeting of the enzyme in pericardial cells (10). Furthermore, this perturbed protein targeting was shown to result in toxicity to the insect: a recombinant baculovirus expressing the mutant enzyme JHE K29R/ K524R killed the host insect significantly faster than a baculovirus expressing JHE (10,24). The present study was based on the hypothesis that disruption of lysosomal targeting of JHE K29R/K524R results from decreased affinity for a binding protein involved in protein sorting in the endocytotic pathway. Here, we report the characterization of P29, which is present in pericardial cell and fat body tissues and shows reduced binding to JHE K29R/K524R relative to JHE.
Previous research showed that targeting of JHE K29R/ K524R to lysosomes in pericardial cells was significantly less efficient than targeting of JHE (10). Our data on reduced binding of P29 to JHE K29R/K524R imply a role for P29 in JHE targeting or processing in lysosomes. P29 is also present in the fat body, which is not involved in uptake of JHE, but is involved in endocytosis of other hemolymph proteins (25). The fat body is also a site of synthesis of JHE (3). To deduce the function of P29, immunoelectron microscopy will be used to determine the intracellular location of P29 and the sites of colocalization with JHE in the two tissues.
Because M. sexta JHE had not been cloned at the beginning of this study, recombinant JHE derived from the tobacco budworm Heliothis virescens (26) was used for enrichment and screening of the M. sexta pericardial cell cDNA phage display library. Ligand blotting with H. virescens-derived JHE against H. virescens pericardial cell and fat body tissues showed the same profiles as blots with M. sexta tissues. 2 H. virescens JHE shares 54% identity with M. sexta JHE. 3 Based on these observations, we expect the processing of the two enzymes to be comparable.
Use of the phage display library enabled the simultaneous screening for JHE-binding proteins and isolation of the cDNAs encoding binding proteins. cDNAs encoding potential JHE-binding proteins were selectively enriched in the phage display library by interaction of the gene products with JHE. This is the first time that phage display has been used successfully for screening of a tissue-derived library for specific binding proteins. The results highlight the importance of eliminating false positives, which may result from frameshifting (27) or from production of artificial peptides. Proteins detected for clones with no open reading frame in the cDNA insert are likely to have resulted from the insert being out of frame or from an incomplete coding sequence. The phage display technique limited the clone insert size to ϳ1 kilobase. There are clearly larger proteins in the pericardial cell complex that bind to JHE and that were not isolated using this technique (Fig. 3B). We are currently using alternative means to isolate and to characterize these proteins.