Lipid transfer particle from the silkworm, Bombyx mori, is a novel member of the apoB/large lipid transfer protein family.

Lipid transfer particle (LTP) is a high-molecular-weight, very high-density lipoprotein known to catalyze the transfer of lipids between a variety of lipoproteins, including both insects and vertebrates. Studying the biosynthesis and regulation pathways of LTP in detail has not been possible due to a lack of information regarding the apoproteins. Here, we sequenced the cDNA and deduced amino acid sequences for three apoproteins of LTP from the silkworm (Bombyx mori). The three subunit proteins of the LTP are coded by two genes, apoLTP-II/I and apoLTP-III. ApoLTP-I and apoLTP-II are predicted to be generated by posttranslational cleavage of the precursor protein, apoLTP-II/I. Clusters of amphipathic secondary structure within apoLTP-II/I are similar to Homo sapiens apolipoprotein B (apoB) and insect lipophorins. The apoLTP-II/I gene is a novel member of the apoB/large lipid transfer protein gene family. ApoLTP-III has a putative conserved juvenile hormone-binding protein superfamily domain. Expression of apoLTP-II/I and apoLTP-III genes was synchronized and both genes were primarily expressed in the fat body at the stage corresponding to increased lipid transport needs. We are now in a position to study in detail the physiological role of LTP and its biosynthesis and assembly.

To detect glycoproteins in LTP subunit proteins, following SDS-PAGE of 5 g of purifi ed larval LTP, the protein was electrophoretically transferred to a polyvinylidene difl uoride (PVDF) membrane (Bio-Rad). The membrane was stained with fl uorescein isothiocyanate conjugated to concanavalin A (FITC-Con A) (Calbiochem, San Diego, CA) according to Furlan,Perret,and Beck ( 24 ). Conjugates were visualized with ultraviolet light.

Estimation of LTP native molecular weight
Estimation of LTP native molecular weight was performed by blue-native PAGE ( 25,26 ) and Western blot analysis ( 27 ). Samples of hemolymph (1 l) from each fi fth instar larva at day 4, pupal stage at day 3, and adult stage at day 0 were electrophoresed on a 3 -10% blue-native acrylamide gradient gel at 4°C and a constant 150 V. A molecular weight marker set (Invitrogen, Carlsbad, CA) was used as the calibration standard. A rabbit polyclonal antibody was raised against purifi ed apoLTP-I. After proteins were separated by blue-native PAGE and transferred to a PVDF membrane, the membrane was incubated with the anti-apoLTP-I antibody and an alkaline phosphatase-goat anti-rabbit IgG conjugate (Jackson ImmunoResearch Laboratory, West Grove, PA). The signals were detected using an alkaline phosphatase-conjugate substrate kit (Bio-Rad).

Preparation of LTP apoproteins and amino acid sequence determination
Three apoLTP subunits were separated by SDS-PAGE of purifi ed LTP and transferred to a PVDF membrane. After Coomassie staining, membrane slices were excised. The N-terminal amino acid sequences of apoLTP-I and apoLTP-II were determined by Edman degradation ( 28 ), and the sliced membranes were incubated with lysyl endopeptidase. Digested peptides were separated by reverse-phase high-performance liquid chromatography (HPLC). Amino acid sequences of the peptides including the N-terminal amino acids were determined from six peptides of apoLTP-I, two peptides of apoLTP-II, and one peptide of apoLTP-III using a G1005A protein sequencing system (Hewlett-Packard, Palo Alto, CA).

Identifi cation and sequence analysis of apoLTP-I and apoLTP-II
We obtained candidate sequences containing apoLTP-I and apoLTP-II peptide sequences as described above by a TBLASTN search of the Silkworm Genome Database using the peptides as query sequences. A protein homology search using the candidate apolipoprotein sequences suggested that the B. mori apoLTP-I and apoLTP-II were encoded by one gene homologous to Drosophila melanogaster CG15828. To determine the full-length cDNA sequence of apoLTP-II/I , the 5 ′ and 3 ′ ends of the apoLTP-II/I sequence were obtained by rapid amplifi cation of cDNA ends (RACE) using a SMART RACE cDNA amplifi cation kit (Clontech, Mountain View, CA) with the following primers: 5 ′ -AGG C-TGGTGTCTTCTTGGCCCCGGACG-3 ′ for the first 5 ′ -RACE product, 5 ′ -GAGACGGCGCCTATGAATTTCTCCGCACG-3 ′ for the nested 5 ′ -RACE product, 5 ′ -CAGCTGGCAGGACTTCCT-CAAGACCCCG-3 ′ for the fi rst 3 ′ -RACE product, and 5 ′ -TGATCG-GCGAGGCCTTGAACACGATCGG-3 ′ for the nested 3 ′ -RACE product. Primers used for RACE were designed using the hypothetical partial cDNA sequences of the silkworm apoLTP-II/I gene, which are homologous to those of D. melanogaster CG15828. Total RNA for RACE was isolated from the fat body of fi fth instar larvae of strain N4 on day 4 using TRIzol reagent (Invitrogen). Because a guanine-cytosine-rich region occurs from positions involved in this process as well as in the transfer of lipids from the midgut to lipophorin ( 13 ). Lipid export from enterocytes does not involve de novo synthesis of lipophorin; rather, lipids are added directly to existing lipophorin particles in the hemolymph ( 14,15 ). Taken together, these reports indicate that LTP plays an important role in facilitating lipophorin function and may mediate the transfer of many lipids, including hydrocarbons ( 5 ) and carotenoids ( 15 ), with its specifi city being determined by the properties of putative lipid transfer factors in the target cell (16)(17)(18).
The mechanisms underlying LTP biosynthesis and assembly are virtually unknown. Due to its very large size, cloning and sequencing cDNA for LTP has been diffi cult, especially when apoLTP-I and apoLTP-II are both encoded on a single gene ( apoLTP-II/I ), as is the case with the precursor protein of lipophorin, apolipophorin-II/I ( 19,20 ). Recently, however, a whole genome sequence database for silkworm ( 21,22 ) has become available and genome annotation is ongoing. In this report, we present the cDNA and deduced amino acid sequences of apoLTP-I, apoLTP-II, and apoLTP-III from B. mori . Our results provide insights into the function of LTP as a novel member of the apoB/ large lipid transfer protein (APO) family and represent an important step in the study of LTP biosynthesis and assembly.

Isolation of LTP
The N4 strain of silkworm ( B. mori ) was maintained in a continuous laboratory colony and reared on an artifi cial diet. The larval LTP was isolated from the hemolymph of fi fth instar larvae on day 4 according to the method described by Tsuchida et al. ( 6 ). The adult hemolymph was collected from the abdomen of adults at day 0 by cutting with a needle and dropped into an icecold bleeding solution (20 mM sodium phosphate pH 6.8, 150 mM NaCl, 5 mM EDTA, 1 mM glutathione, and 1 mM 4-2-aminoethyl benzenesulfonyl fl uoride). The hemolymph was centrifuged at 800 g for 5 min to remove hemocytes. To the supernatant, 8.9 g potassium bromide (KBr) was added, and the volume was adjusted to 20 ml with bleeding solution. The solution was transferred to 36.2 ml OptiSeal TM centrifuge tubes (Beckman Coulter, Brea, CA) and overlaid with bleeding solution. The tube was centrifuged at 4°C and 50,000 rpm for 4 h in a VTi 50 rotor (Beckman Coulter). After centrifugation, LDLp, high-density lipophorin (HDLp), and LTP formed three yellow bands in the ultracentrifuge tube. The fractions from the middle yellow band, which contained mainly HDLp and LTP, were pooled, and after desalting, were applied to DEAE Bio-Gel (Bio-Rad, Hercules, CA) and eluted with 20 mM sodium phosphate pH 7.5, containing a linear NaCl gradient (20 -300 mM). The elutant was subjected to a Sephacryl S-300 column (GE Healthcare, Milwaukee, WI) and fractions were collected. Fractions containing LTP were examined by sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) ( 6 ).

SDS-PAGE and detection of protein glycosylation
SDS-PAGE was performed according to the method of Laemmli ( 23 ) in slab gels containing a 4 -15% linear gradient of using the same parameters as used for the apoLTP-II/I tree, except that the WAG (F+G) model was adopted as the best-fi tted amino acid substitution model.

Prediction of the amphipathic secondary structure in apoLTP-II/I
Amphipathic ␣ -helixes and ␤ -strand regions with high lipid affi nity were predicted using the computer program LOCATE, developed by Segrest et al. at the University of Alabama, Birmingham ( 33,34 ). All amino acid sequences, except that of B. mori apoLTP II/I, were obtained from the NCBI protein database. These included the Homo sapiens apolipoprotein B-100 precursor protein (AA35549.1), the D. melanogaster CG15828 protein (NP_995670), and the M. sexta apolipophorin precursor protein (AAB53254.1).

Northern blot analysis
The midgut, fat body, silk gland, Malpighian tube, testis, and ovary were dissected from fi fth instar larvae at day 0, and total RNA was extracted from each tissue separately using TRIzol reagent. To identify the expression of apoLTP-II/I and apoLTP-III , 20 g total RNA from each tissue was applied to a 0.7% agarose gel in 2% formaldehyde. After electrophoresis, RNA was blotted to a Hybond-N+ membrane (GE Healthcare) and hybridized in ULTRAhyb (Ambion, Austin, TX). Radiolabeled hybridization probes for apoLTP-I and a poLTP-II were generated from cDNA for each region (cloned into the pGem-T vector) using the Riboprobe system SP6 (Promega, Madison, WI) and [ ␣ -32 P]CTP; therefore, the pGem-T vectors contained nucleotides from positions 702 to 1,749 and 9,613 to 10,543 [the ATG start codon of each open reading frame (ORF) was indicated to be at positions 1 -3] of apoLTP-II and apoLTP-I , respectively. ApoLTP-III mRNA was detected using a [ ␣ -32 P]CTP-labeled single-stranded RNA probe synthesized from nucleotide positions 35 to1,083 (the ATG was indicated to be at positions 1 -3) of the apoLTP-III cDNA sequence using the method described above. Bound radioactivity was detected with a Typhoon FLA7000 image analyzer (GE Healthcare).

Developmental profi les of apoLTP-II/I and apoLTP-III mRNA expression in the fat body
Changes in apoLTP-II/I and apoLTP-III mRNA expression in the fat body were analyzed from the fourth instar larva to the adult stages. Total RNA was prepared from the fat body of fourth and fi fth instar larvae, pupae, and adults, and used for quantitative real-time PCR (qPCR) analyses of apoLTP-II/I and apoLTP-III gene expression. Single-stranded cDNAs were synthesized from total RNA using Superscript III reverse transcriptase (Invitrogen) and an oligo-dT primer, and then treated with RNaseH (Takara, Otsu, Japan). Quantifi cation of the transcripts was carried out by qPCR using the cDNAs as templates with LightCycler Fast-StartDNA MasterPLUS SYBR Green I reagent (Roche, Darmstadt, Germany) and a LightCycler DX400 thermocycler (Roche). The specifi c primer pairs used for apoLTP-II/I and apoLTP-III were 5 ′ -CTGACTGTCGATATGTTTGGCGAGT-3 ′ and 5 ′ -TTCA-TG TTCAAAGGCAAACCGCATCCG-3 ′ , and 5 ′ -TGTTCCAGT-TTAGGAACTGCC-3 ′ and 5 ′ -TGCATAGTTCCAAGAGTGAG-3 ′ , respectively. Transcript levels of the genes were normalized to the level of the ribosomal protein L3 ( rpL3 ) in the same samples ( 35 ) using the primer pair 5 ′ -TTCCCGAAAGACGACCCTAG-3 ′ and 5 ′ -CTCAATGTATCCAACAACACCGAC-3 ′ . mRNA levels were expressed relative to that found in fourth instar larvae at day 0.

Data deposition
The apoLTP-II/I and apoLTP-III cDNA sequences from the B. mori N4 strain were deposited in the DNA Data Bank of Japan with accession numbers AB700597 and AB700598, respectively. 12,203 to 12,342 bp in the B. mori apoLTP-II/I gene, full-length apoLTP-II/I cDNA was not amplifi ed by polymerase chain reaction (PCR) with a primer pair generated from the 5 ′ -and 3 ′ -untranslated regions (UTRs) obtained by RACE. Instead, two partial apoLTP-II/I cDNA fragments from the fat body were amplified by reverse transcription (RT)-PCR using KOD FX DNA polymerase (Toyobo, Tokyo, Japan) with the following primer pairs: 5 ′ -CGGTG G G CG AAA CGTTTGGACATGGATAT-3 ′ (forward) and 5 ′ -ACGATA TTTCTATTGGGTCAGT-3 ′ (reverse), and 5 ′ -CAC-TGACCCAATAGAAATATCG-3 ′ (forward) and 5 ′ -AATT ATCA AC-TAAGCGACGGTATGGTGGGG-3 ′ (reverse). These four primers were designed based on the partial sequences determined by 5 ′ -and 3 ′ -RACE. Sequences of the amplifi ed fragments were determined by direct sequencing. The determined sequences were then combined to obtain a full-length cDNA sequence of apoLTP-II/I encoding all eight of the apoLTP-I and apoLTP-II peptides determined by amino acid sequence analyses. The apoLTP-II/I cDNA cloning and sequence methods are diagramed in supplementary

Identifi cation and sequence analysis of apoLTP-III
The amino acid sequence of purifi ed apoLTP-III was determined as described above. We obtained candidate protein sequences of apoLTP-III containing the peptide sequence by a TBLASTN search of the Silkworm Genome Database using the peptide as a query sequence. A protein homology search with the candidate protein sequences revealed that B. mori apoLTP-III was encoded by one gene homologous to Tribolium castaneum (XP972731). To determine the full-length cDNA sequence of apoLTP-III , the 5 ′ and 3 ′ ends of the apoLTP-III sequence were obtained by RACE using a SMART RACE cDNA amplifi cation kit (Clontech) with primer 5 ′ -CCAATTTGTCGAGCTCCGAC-TGAAC-3 ′ for the 5 ′ -RACE product and primer 5 ′ -TACCC GA-GGAGGTGTCGAGTGAAG-3 ′ for the 3 ′ -RACE product. Both primers were designed based on the silkworm genomic sequence encoding the putative apoLTP-III protein. Total RNA was prepared using methods described previously. The determined sequences were combined to obtain the full-length cDNA sequence of apoLTP-III encoding the apoLTP-III peptide.

Phylogenetic analysis of deduced amino acid sequences
Amino acid sequences of 33 large lipid transfer proteins (LLTPs) were collected from the National Center for Biotechnology Information (NCBI) protein database. The deduced amino acid sequence of apoLTP-II/I was aligned with the 33 LLTP sequences using CLUSTALX2 ( 29 ). Alignments were edited and corrected manually with Genetyx version 9.0.1 software (Genetyx Corporation, Tokyo, Japan). Accession numbers of the LLTP sequences collected from the database are listed in supplementary Table I. Twenty-two N-terminal conserved motifs of the large lipid transfer (LLT) modules (N1 -N22) were extracted and aligned according to previous reports ( 30,31 ), and the conjugated sequences (supplementary Fig. II) were employed for subsequent phylogenetic analysis. A maximum-likelihood tree was constructed using the MEGA 5 (version 5.05) program ( 32 ). The RtREV (F+I+G) model was selected as the best fi tted amino acid substitution model, and the number of discrete ␥ categories was defi ned as fi ve. All positions containing gaps were eliminated. The bootstrap tests were replicated 1,000 times for each node.
The deduced amino acid sequence of apoLTP-III was used as the query in a BLASTP search against the NCBI protein database. Thirty-fi ve amino acid sequences corresponding to the best hits were aligned using CLUSTALX2, and the alignments were edited and corrected manually with Genetyx version 9.0.1 (supplementary Fig. III). The maximum-likelihood tree was constructed the prediction by the SignalP 4.0 program ( 36 ). Signal peptide cleavage may occur after residue 18 to produce the 4,103 amino acid apoLTP-II/I protein ( Fig. 1B ).
These results indicate that the apoLTP-II/I protein is arranged with apoLTP-II at the N-terminal end and apoLTP-I at the C-terminal end. The cleavage site was found between positions 720 and 721 in the precursor protein (the N terminus of apoLTP-II was indicated to be at position 1). The N-terminal amino acid sequence (SLNDTEDVRSK) from the purifi ed apoLTP-I was found at positions 721 -731 in the precursor protein following the RFAR amino acid sequence at the C terminus of apoLTP-II ( Fig. 1B ). The calculated molecular masses of apoLTP-I and apoLTP-II based on the total of the deduced amino acid sequences were 385,826 and 82,303 Da, respectively. Using NetNGlyc 1.0 software, apoLTP-I and apoLTP-II were predicted to contain 52 and 3 potential N-glycosylation sites, respectively (NXT/S, where X is any residue other than P).
The apoLTP-II/I gene for B. mori (strain p50), constructed from the Silkworm Genome Database, consisted of 68 exons separated by 67 introns and spanned more than 87 kb. The gene is located at position 3.2 Mb on chromosome 28 of the B. mori genome.

Homology search of apoLTP-II/I, phylogenetic analysis of the LLTP superfamily, and structural organization of apoLTP-II/I
The similarities between the deduced amino acid sequences of the B. mori apoLTP-II/I and H. sapiens apoB or RESULTS cDNA sequence and deduced amino acid sequence of apoLTP-I and apoLTP-II Three LTP subunits were separated by SDS-PAGE of purifi ed LTPs, and the amino acid sequence of eight peptides, including the N-terminal amino acid sequences of apoLTP-I and apoLTP-II, were determined. Based on these peptide sequences, we determined the full-length cDNA of apoLTP-II/I by searching the Silkworm Genome Database, 5 ′ -and 3 ′ -RACE, and RT-PCR. Methods for sequence determination of apoLTP-II/I cDNA are illustrated in supplementary Fig. I.
The full-length cDNA for apoLTP-II/I has a single 12,363 bp ORF, beginning with ATG at nucleotide positions 1 -3 and extending to a stop codon at positions 12,364 -12,366. The ORF is predicted to encode a 4,121 amino acid protein. Lengths of the 5 ′ -and 3 ′ -UTR were 63 bp and 92 bp, respectively.
Six peptides from apoLTP-I, SLNDTEDVRSK, YTTLALL-NFN, LVSGYLFLPP, WDINGSHFIDY, TNFIFDPRVGE, and VFTDPIEISS, and two peptides from apoLTP-II, SANSLKDPFI, and SDFQIIAAAPKT, which were determined by amino acid sequence analysis, were found in the apoLTP-I and apoLTP-II amino acid sequences deduced from cDNA ( Fig. 1A ). Because the N-terminal apoLTP-II (SANSLKDPFI) sequence was determined to begin at position 19 in the cDNA, the fi rst 18 amino acids were assumed to be the signal peptide, which was consistent with similar to CG15828, and the apoLTP-II/I subfamily consisting of these three proteins was separated from the apoB, apoCr, and apolipophorin-II/I subfamilies in the APO family ( Fig. 2 ).
The LLT domain comprises a large N-terminal domain of approximately 1,000 amino acids that is proposed to bind lipids or transfer lipids to apolipoproteins ( 30 ). An amphipathic ␣ -helix/ ␤ -strand region with high lipid affi nity was predicted using the LOCATE computer program ( 33,34 ). We analyzed the amphipathic clusters of apoLTP-II/I and found that it had a very similar arrangement as the lipophorin precursor protein ( 30 ), which contains three regions enriched in amphipathic ␣ -helical and amphipathic ␤ -strands organized as N-␣ 1 -␤ -␣ 2 -C ( Fig. 3A ). In apoLTP-II/I, the ␣ 1 , ␤ , and ␣ 2 domains were located between residues 1 to 800, 800 to 2,900, and 3,100 to 3,600, respectively. Predicted clusters of amphipathic secondary structure within the apoLTP-II/I L. migratoria apolipophorin-II/I were analyzed using the NCBI BLAST search, and the results showed that the fi rst 1,000 amino acid residues with the highest similarity were 43% similar to H. sapiens apoB and 44% similar to L. migratoria apolipophorin-II/I. Alignment of apoLTP-II/I with human and insect apolipoproteins showed that apoLTP-II/I belongs to the LLTP superfamily (supplementary Fig. II) ( 37,38 ). The LLTP superfamily contained three distinguishable groups: the APO family, which includes apoB, apolipophorin-II/I, and apolipocrustacein (apoCr); the Vtg/CP family, including vitellogenin (Vtg), melanine-engaging protein (MEP) and the crustacean clotting protein (CP); and the microsomal triglyceride transfer protein (MTP) family. The B. mori apoLTP-II/I fell within the APO family, although the bootstrap values were not high. The APO family also contains D. melanogaster protein CG15828 and Apis mellifera protein function from Danaus plexippus (EHJ73751.1) identifi ed in a NCBI BLAST database search. In addition, 34 similar proteins from insects and one protein from the water fl ea ( Daphnia pulex ; EFX77001.1), all of which are of unknown function, were found from a sequence homology search using NCBI BLAST. All of these proteins, including the B. mori apoLTP-III, had a putative conserved juvenile hormone-binding protein (JHBP) superfamily domain ( Fig. 4 ). A phylogenetic tree of the insect proteins was constructed based on their primary amino acid sequences ( Fig. 5 ). B. mori apoLTP-III, D. plexippus (EHJ73750.1), and D. plexippus (EHJ73751.1) proteins were clustered in a Lepidopteraspecifi c group.

Detection of apoLTP protein glycosylation
According to SDS-PAGE, both LTPs purifi ed from the hemolymph of larvae and adults contained three protein subunits; no protein other than the three subunits was observed ( Fig. 6A ). Additionally, the molecular masses of apoLTP-I, apoLTP-II, and apoLTP-III were estimated to be approximately 350, 85, and 60 kDa, respectively. Because apoLTP-I is extremely large and the largest protein used for a SDS-PAGE molecular mass standard was approximately 250 kDa, the molecular mass of apoLTP-I was not estimated precisely by SDS-PAGE.
Larval LTP was subjected to SDS-PAGE and transferred to a PVDF membrane. As shown in Fig. 6B , the three protein were found to share some similarity with H. sapiens apoB and M. sexta apolipophorin-II/I ( Fig. 3 ).

cDNA sequence and deduced amino acid sequence of apoLTP-III
The cDNA for apoLTP-III has a 1,368 bp ORF ( Fig. 4 ) and a deduced amino acid sequence with 456 amino acid residues and a calculated molecular mass of 51,074 Da. The 5 ′ -and 3 ′ -UTR lengths were 31 and 123 bp, respectively. The sequence included an 18 amino acid signal sequence predicted by the SignalP 4.0 program ( 36 ). Mature apoLTP-III is composed of 438 amino acids, corresponding to a calculated molecular mass of 48,969 Da. The existence of four potential N-glycosylation sites suggests the potential for an increase in the molecular mass of the protein. The amino acid sequence obtained following digestion of the purifi ed apoLTP-III (IDEVAGDLQF) was identical to a region of the translated deduced cDNA sequence ( Fig. 4 ). The B. mori apoLTP-III gene in the Silkworm Genome Database was located at position 14.2 Mb on chromosome 15 and consisted of three exons spanning more than 3 kb. The coding sequence of the apoLTP-III gene in strain p50 was identical to the N4 strain sequence.

Homology search and phylogenetic analysis of apoLTP-III
The deduced amino acid sequence of B. mori apoLTP-III shared 62% identity with a predicted protein of unknown instar larval to adult stages. qPCR analysis was performed to examine the expression levels of apoLTP-II/I and apoLTP-III mRNA. mRNA expression levels were expressed relative to that found in fourth instar larvae at day 0.
In the fat body, high levels of apoLTP-II/I mRNA were observed in fi fth instar larvae at day 0, immediately after initiation of feeding. Later, during the larval feeding stage and the pupal stage, levels gradually declined and remained relatively low and constant. From the late pupal stage to emergence, the transcript level increased sharply, after which it stayed relatively high until death ( Fig. 8A ).
The expression pattern of apoLTP-III mRNA was similar to that of apoLTP-II/I mRNA. In the fourth instar larvae the level was low, but increased and became high at day 0 of the fi fth instar. Following day 0 of the fi fth instar, the levels decreased gradually until the spinning stage. At day 0 of the spinning stage, apoLTP-III expression increased and remained at a low level during the early pupal stage. ApoLTP-III expression increased again from day 4 of the pupal stage to day 0 of the adult stage and then stayed relatively high until death ( Fig. 8B ).

DISCUSSION
Until now, apoLTP-II/I and apoLTP-III cDNAs have not been isolated from any species, and the gene structures of apoLTP-II/I and apoLTP-III were unknown. This is the fi rst study to report the LTP cDNA sequence, the gene structure of LTP, and the amino acid sequences of three apoproteins of LTP.
The three subunit proteins of B. mori LTP are coded by two genes, apoLTP-II/I and apoLTP-III . ApoLTP-II/I from the silkworm ( B. mori ) is a novel member of the APO family, which is similar to H. sapiens apoB and insect lipophorins. However, a major difference exists between apoB and apoLTP-II/I. ApoB is not cleaved during its association with lipids, but apoLTP-II/I is a proapoprotein that is cleaved to become two subunits (apoLTP-I and apoLTP-II) of LTP arranged with apoLTP-I at the C-terminal end and apoLTP-II at the N-terminal end. In H. sapiens apoB, subunits of LTP showed reactivity with FITC-Con A; therefore, the three subunits appear to contain carbohydrate chains.

Molecular mass of intact LTP
The molecular mass of intact LTP from B. mori was estimated by a combination of blue-native PAGE and Western blot analysis using anti-apoLTP-I rabbit serum. The molecular masses of the B. mori intact LTP from the hemolymph of the larval, pupal, and adult stages were estimated at approximately 620 kDa ( Fig. 6C ). In the hemolymph of adults at day 0, both the 620 kDa LTP and larger molecular mass LTP of about 800 kDa were found. The larger intact LTP was not found in the hemolymph of fi fth instar larvae at day 4 and pupae at day 3 ( Fig. 6C ).

Tissue-specifi c expression of apoLTP-II/I and apoLTP-III in fi fth instar larvae
Northern hybridization experiments were performed to identify tissue expression patterns of apoLTP-II/I and apoL-TP-III in fi fth instar larvae at day 0 ( Fig. 7 ). Both 32 P-labeled apoLTP-I -specifi c ( Fig. 7A ) and apoLTP-II -specifi c probes ( Fig. 7B ) detected a single product of over 12,000 nucleotides in the Northern blot analysis, whereby the size was estimated by comparison with RNA standards. As shown in Fig. 7A, B , the fat body, testis, and ovary accumulated the apoLTP-II/I transcript, whereas no expression was observed in the midgut, silk gland, or Malphigian tube. The highest levels of apoLTP-II/I mRNA were observed in the fat body.
One apoLTP-III transcript of approximately 2,300 nucleotides was detected in the fat body, ovary, and testis using a single-stranded RNA probe synthesized from a DNA fragment representing apoLTP-III . Northern blot analyses showed that the highest level of apoLTP-III mRNA was detected in the fat body with slightly lower, but still relatively high, levels in the ovary and testis ( Fig. 7C ).

Developmental changes in apoLTP-II/I and apoLTP-III expression
We analyzed changes in apoLTP-II/I and apoLTP-III mRNA expression in the fat body throughout the fourth enzyme for apoLTP-II/I and apolipophorin-II/I may be common because it is accordant to that of the limited endoproteolytic cleavage that occurs after a sequence containing two or more basic residues (K or R) (43)(44)(45)(46). The amino acid sequence RFAR was found at the C-terminal end of the putative B. mori apoLTP-II, and RGRR was also found in both M. sexta and B. mori apolipophorin-II/I ( 20,47 ). Our results presented here do not rule out a role for apoLTP-I and apoLTP-II. However, we completed sequencing cDNA for the 4,103 amino acid apoLTP-II/I and are now in a position to produce LTP protein with and without cleaved apoLTP-II/I to determine whether cleavage of apoLTP-II/I is required to produce a biologically active LTP.
Thirty-fi ve apoLTP-III protein homologs were found in insects and water fl eas from a sequence homology search using NCBI BLAST. All identifi ed proteins were of unknown function but all shared a putative conserved JHBP the whole ␤ ␣ 1 domain at the N-terminal end was proposed to be a lipid pocket for initiation of lipoprotein particle assembly ( 39,40 ), and the N-terminal sequence of apoB is known to be critical for TAG-rich lipoprotein assembly and secretion ( 41 ). ApoLTP-II is 720 amino acids long; the ␣ 1 domain covers the entire apoLTP-II sequence. Blacklock and Ryan ( 42 ) examined the catalytically important LTP apoprotein using LTP apoprotein-specifi c antibodies in a lipid transfer inhibition and reported that apoLTP-II plays a key role in lipid transfer activity. Posttranslational cleavage of apoLTP-II/I for building of the two subunits is assumed to be important for formation of the lipid-binding and transfer region required to recruit and pack DAG instead of TAG. Indeed, the two subunits of insect DAG-rich lipophorin, apolipophorin-I and apolipophorin-II, were also shown to result from cleavage of apolipophorin-II/I and are arranged in the same manner as apoLTP-II/I ( 19,20 ). In addition, the subunit cleavage between lipophorin as the functional activity of LTP have not been studied. Palm et al. ( 50 ) referred to the protein encoded by D. melanogaster CG15828 as apoLTP and observed its function using RNAi knockdown. Their results showed that the D. melanogaster CG15828-encoded protein facilitated lipid export from the gut to lipophorin, indicating a function similar to that of LTPs of other insects ( 13 ). Because CG15828 protein is equivalent to apoLTP-II/I, apoLTP-III may not function on lipid transfer in D. melanogaster . Palm et al. ( 50 ) suggested that if apoLTP-III exists in this fl y, it is not likely to be derived from the apoLTP precursor protein.
In D. melanogaster , three apoLTP-III homologs were found in NCBI BLAST database (accession numbers NP-608781.2, AAL68365.1, and AAL48116.1) that may be apoLTP-III. However, we did not address how apoLTP-III works within the LTP particle.
We determined that the molecular mass of intact B. mori LTP was approximately 620 kDa ( Fig. 6C ). Based on SDS-PAGE, LTP showed three subunits of molecular masses 350, 85, and 60 kDa, respectively ( Fig. 6A ), and all three apoproteins were glycosylated ( Fig. 6B ). However, based on the present results, the calculated molecular masses of apoLTP-I, apoLTP-II, and apoLTP-III proteins from their cDNA sequences were 385,826, 82,303, and 48,969 Da, respectively. Therefore, apoLTP-I was considered to be larger than 385,826 Da rather than 350,000 Da. Considering the total molecular mass of the three subunit proteins (>530 kDa) and the lipid composition of LTP (about 20%) together, we conjectured that intact B. mori LTP may consist of one apoLTP-I, one apoLTP-II, and one apoLTP-III molecule. Because we did not confi rm the molecular masses of the three subunits of B. mori LTP, quantitative carbohydrate Fig. 6. Estimation of the molecular mass of native LTP. A: Purifi ed LTP from the hemolymph of fi fth instar larvae and adults was subjected to SDS-PAGE and stained with Coomassie Brilliant Blue R-250. B: Purifi ed LTP from the hemolymph of fi fth instar larvae was subjected to SDS-PAGE and transferred to a PVDF membrane. The blot was incubated with FITC-Con A solution. Detection of glycoproteins bound to FITC-Con A was carried out under ultraviolet light. Arrows indicate apoLTP-I (I), apoLTP-II (II), and apoLTP-III (III) from the top, respectively. C: One microliter each of the hemolymph from fi fth instar larva, pupae, and adults was electrophoresed by 3 -10% blue-native PAGE. Separated proteins were transferred to a PVDF membrane. Native LTP was detected by Western blot analysis using the anti-apoLTP-I antibody. Arrows indicate the 620 kDa and 800 kDa LTP. Numbers on the left of each panel represent the molecular masses for protein standard.
superfamily domain ( 48 ). The JHs are acyclic sesquiterpenoids that regulate insect development and reproduction ( 49 ). The juvenile hormone (JH)-binding activity of apoL-TP-III and the JH transport or JH transfer/exchange activity Fig. 7. Tissue-specifi c gene expression of apoLTP-II/I and apoLTP-III from B. mori . Northern hybridization with apoLTP-I -specifi c probes (A) and apoLTP-II -specifi c probes (B) revealed transcripts for apoLTP-II/I in the fat body, ovary, and testis of fi fth instar larvae at day 0, which were larger than the 6,583 nucleotide RNA marker. ApoLTP-III-specifi c probes detected the approximately 2,300 nucleotide transcript in the fat body, ovary, and testis (C). No transcripts for apoLTP-II/I (A, B) or apoLTP-III (C) were detected in other tissues, including the midgut, silk gland, and Malpighian tube of fi fth instar larvae at day 0. The rRNA bands stained with ethidium bromide were used to monitor equal loading of the sample. Size of the RNA markers is shown on the left.
Northern blot analysis suggests that the apoLTP-II/I and apoLTP-III genes are more actively transcribed in the B. mori fat body than in the ovary and testis ( Fig. 7 ). Gene expression levels of both apoLTP-II/I and apoLTP-III were synchronized in the fat body ( Fig. 8 ). Both the apoLTP-II/I and apoLTP-III genes were strongly expressed at day 0 of the fi fth instar larval stage and at the early adult stage corresponding to increased DAG transport needs. ApoLTP-II/I and apoLTP-III expression might be enhanced by the onset of feeding. In adults, the high expression of apoLTP-II/I and apoLTP-III may correlate with LTP function during fl ight-related lipophorin conversion of HDLp to LDLp in response to adipokinetic hormone. These observations support the fi nding that the LTP concentration in the hemolymph is highest in adults ( 8,15 ).
While LTP plays an essential role in promoting lipid transfer from cells to lipophorin, and exchanges lipids between lipophorins, the present results suggest that LTP is a novel member of the APO family, and that LTP may play a major role as a lipid carrier in the hemolymph, similar to that of lipophorin. Finally, the molecular characterization of LTP reported here may not only open a new fi eld of research on the biosynthesis, lipid recruitment, and assembly of LTP, but may also allow its function in insect lipid metabolism to be clarifi ed. and other protein modifi cation analyses of separated subunits of B. mori LTP will be required to confi rm subunit composition. Ryan et al. ( 2,9 ) have extensively studied the properties of M. sexta LTP and reported that the molecular mass of LTP is greater than 500 kDa or over 670 kDa, which was estimated by gel permeation chromatography or pore-limiting gradient PAGE of purifi ed LTP, respectively, and that all three subunits were glycosylated proteins containing carbohydrate (5% by weight). Additionally, LTP becomes unstable and easy to aggregate when LTP purification is progressing. Particle aggregation might preclude an accurate molecular mass determination. To avoid this problem, we applied the hemolymph rather than purifi ed LTP on blue-native PAGE and detected LTP by Western blot analysis using an anti-apoLTP-I antibody.
We found a larger size of LTP (about 800 kDa) compared with the 620 kDa LTP in adult hemolymph ( Fig. 6C ). Although in the adult hemolymph, the larger size of LTP is assumed to have an increased amount of DAG, one cannot exclude that these molecules have another protein component other than the three subunits. Indeed, after KBr density gradient ultracentrifugation of adult hemolymph, the 800 kDa LTP coexists in the HDLp fraction, which has a density of 1.12 -1.15 g/ml, but the larval LTP (620 kDa) was recovered from the higher density fractions (1.20 -1.23 g/ml) than HDLp. Moreover, as indicated by SDS-PAGE, no protein other than the three subunits was observed in the LTP purifi ed from these fractions ( Fig. 6A ). In insects, during the fl ight activity of an adult, the amount of lipids mobilized in the hemolymph increases and forms HDLp to LDLp. LTP exchange or DAG transfer between two different density lipophorins progresses LDLp formation. During this process, larger LTP (800 kDa), which has an increased amount of DAG, may be formed without association of additional exchangeable apoproteins. The amount of lipid in LTP may increase up to approximately 30% to form the 800 kDa lower density LTP in adult hemolymph. Fig. 8. Developmental profi le of apoLTP-II/I (A) and apoLTP-III (B) gene expression. Total RNA was prepared from the fat body of fourth and fi fth instar larvae, pupae, and adults. Quantifi cation of transcripts was carried out by qPCR. The y-axis (relative amount) indicates the fold-increase in mRNA expression compared with that of fourth instar larvae at day 0 (L4-0). The y-axis bars on L4-0 in both (A) and (B) are set to one. Results shown are the mean ± SEM (n = 3). Stages are shown as L4, fourth instar larval stage; L5, fi fth instar larval stage; S, spinning cocoon stage; P, pupal stage; A, adult stage.