Pyrosequencing-based expression profiling and identification of differentially regulated genes from Manduca sexta, a lepidopteran model insect

Abstract

Although Manduca sexta has significantly contributed to our knowledge on a variety of insect physiological processes, the lack of its genome sequence hampers the large-scale gene discovery, transcript profiling, and proteomic analysis in this biochemical model species. Here we report our implementation of the RNA-Seq cDNA sequencing approach based on massively parallel pyrosequencing, which allows us to categorize transcripts based on their relative abundances and to discover process- or tissue-specifically regulated genes simultaneously. We obtained 1,821,652 reads with an average length of 289 bp per read from fat body and hemocytes of naïve and microbe-injected M. sexta larvae. After almost all (92.1%) of these reads were assembled into 19,020 contigs, we identified 528 contigs whose relative abundances increased at least 5- and 8-fold in fat body and hemocytes, respectively, after the microbial challenge. Polypeptides encoded by these contigs include pathogen recognition receptors, extracellular and intracellular signal mediators and regulators, antimicrobial peptides, and proteins with no known sequence but likely participating in defense in novel ways. We also found 250 and 161 contigs that were preferentially expressed in fat body and hemocytes, respectively. Furthermore, we integrated data from our previous study and generated a sequence database to support future gene annotation and proteomic analysis in M. sexta. In summary, we have successfully established a combined approach for gene discovery and expression profiling in organisms lacking known genome sequences.

1. Introduction

Insects possess an effective defense system to control pathogen invasion, which includes a physical barrier intertwined with biochemical and cellular mechanisms to block penetration and proliferation of infectious agents (Gillespie et al., 1997; Strand, 2008; Lemaitre and Hoffmann, 2007). These mechanisms are mediated by molecules that recognize pathogens, relay or modulate immune signals, and kill the invading pathogens. Most of the molecules are proteins in body fluids (e.g. plasma) and tissues/cells (e.g. fat body and hemocytes), either constitutively produced for responses occurring within minutes or induced within hours to days after the initial encounter of pathogens. Fat body is a major source of plasma proteins, some of which participate in humoral immunity, whereas hemocytes are mainly involved in cellular responses such as phagocytosis and encapsulation.

The tobacco hornworm, Manduca sexta, has been used extensively as a model species to study the biochemical basis of insect immunity (Jiang, 2008; Regan et al., 2009), as well as other physiological processes. Pathogen recognition proteins, hemolymph proteinases (HPs), serpins, phenoloxidases (POs), and antimicrobial peptides (AMPs) have been isolated from larval hemolymph of this insect for functional analysis. A differential expression study uncovered 120 expressed sequence tags (ESTs) identical or similar to immunity-related genes (Zhu et al., 2003). Pyrosequencing of cDNA fragments using the RNAseq approach (Morin, et al. 2008; Mortazavi et al., 2008) from a mixture of eight total RNA samples revealed 218 new EST contigs coding for defense proteins (Zou et al., 2008). Additional immunity-related genes were identified in a gut EST project that combined Sanger and 454 sequencing technologies (Pauchet et al., 2010). Sequences provided by these studies, albeit useful, are limited by the methods used to obtain them, such as low throughput (Zhu et al., 2003), high rate of indels (Zou et al., 2008), gene discovery solely based on homology (Pauchet et al., 2010), and lack of information on relative gene expression levels in all the cases. While these problems can be overcome by the genome sequence and microarray analysis yet to come, is it possible to efficiently discover genes along with their expression profiles using next-generation RNA-Seq technologies without resorting to the reference genome and thereby directly uncover process-related gene expression in non-model organisms whose genome sequences are not yet determined?

Here, we report the results of our ongoing studies aimed at discovering alterations in gene expression in M. sexta larvae before and after a bacterial injection and characterize genes based on their tissue-preferential expression patterns in fat body and hemocytes using the RNAseq approach (Morin, et al. 2008; Mortazavi et al., 2008). Therefore, instead of relying on a priori knowledge of the genome, our approach contributes to future genome annotation, cDNA cloning, and protein identification in this insect and, through extremely deep RNA-Seq studies, reveals novel genes that likely play a role in insect defense, and provides useful leads for functional elucidation of unknown defense proteins in this biochemical model insect. More importantly, this method is applicable to gene discovery and study of process/tissue-related transcriptome changes in all non-model species with no known genome sequences.

2. Methods and materials

2.1. Insect rearing, bacterial injection, RNA isolation, and library construction

M. sexta eggs, purchased from Carolina Biological Supply, were hatched and reared on an artificial diet as described by Dunn and Drake (1983). Each of day 2, 5^th instar larvae (60) was injected with a mixture of Escherichia coli (2×10⁷ cells), Micrococcus luteus (20 μg) (Sigma-Aldrich), and curdlan (20 μg, insoluble β-1,3-glucan from Alcaligenes faecalis) (Sigma-Aldrich) in 30 μl H₂O. Total RNA samples were extracted from induced hemocytes (IH) and fat body (IF) 24 h later using TRIZOL Reagent (Life Technologies Inc.). Control hemocyte (CH) and fat body (CF) RNA was prepared from day 3, 5^th instar naïve larvae (60). PolyA+ RNA was separately purified from the total RNA samples (1.0 mg each) by binding to oligo(dT) cellulose twice in the Poly(A) Purist^™ Kit (Ambion). First strand cDNA was synthesized using mRNA (5.0 μg), random dodecanucleotides (100 pmol), and SuperScript^™ III reverse transcriptase (1000 U, Life Technologies Inc.). RNase H treatment, second strand synthesis, and gap joining were performed according to the published protocol (Zou et al., 2008). After shearing via nebulization, the four samples were end-repaired (Roe, 2004) and ligated to double-stranded adaptor A and biotinylated adaptor B (Margulies et al., 2005).

2.2. PCR amplification, pyrosequencing, and sequence assembly

The cDNA with adaptor B attached on one or both ends was isolated using streptavidin-coated magnetic beads, end repaired, and quantified on an Agilent 2100 Bioanalyzer (Agilent Technologies). Diluted DNA molecules, individually captured by beads, were amplified using emulsion PCR with the two primers complementary parts of A and B adaptors (Margulies et al., 2005). After removal of the second strand and empty beads, the sequencing primer identical to another part of A adaptor was used for sequencing. Two full plates were run with one-half plate for each library on a 454 GS-FLX pyrosequencer (Roche Applied Science) using long-read GS-FLX Titanium chemistry. Reads were assembled separately for each library (CF, CH, IF, IH) and collectively (CIFH) using Newbler Assembler (Roche Applied Science) into five datasets: CF, CH, IF, IH, and CIFH (Fig. 1). To improve coverage and quality of the sequence sets, data from our previous run on a 454 GS20 (Zou et al., 2008) were assembled into two datasets (06 for the 2006 data and 06CIFH for the 2006 and 2009 data) using the updated Newbler software. The resulting contigs and singletons from the seven datasets were compared against the NCBI nr/nt and KEGG databases using BLASTN, BLASTP, and BLASTX with a maximum E-value of 1×10⁻⁵. For the combined library CIHF, numbers of CH, CF, IH, and IF reads assembled into each contig were extracted from the standard Newbler Assembler output and tabulated using Microsoft Excel.

Fig 1. Scheme of library sequencing, dataset assembling, read normalization, contig categorization, and function prediction.

Five cDNA libraries (CF, CH, IF, IH, and 2006) were assembled into seven datasets, one of which (#5: CIFH) was further analyzed by extracting numbers of CF, CH, IF and IH reads assembled into each contig. As described in Section 2.3, read numbers were calibrated using library normalization factors (LNFs) for the calculation of relative abundances (RAs) or adjusted read numbers (ARNs). Based on thresholds set arbitrarily, contigs were categorized into four groups: UP and DN for up- and down-regulated; HC and FB for hemocyte- or fat body-specific.

2.3. Read normalization and ratio calculation

Based on frequencies of several commonly used standards in each of the four libraries (e.g. number of rpS3 reads in CH ÷ number of total reads in CH), a set of six ribosomal protein genes were selected as internal standards, which had high total read numbers and low coefficients of variation (i.e. SD/mean) in their frequencies. The sums of their read numbers for specific libraries, or library normalization factors (LNFs), which already reflected the differences in library sizes, were directly used to calibrate other read numbers in the corresponding libraries. For a specific contig in CIFH, its relative abundance (RA) in libraries X and Y is defined as: RA_x/y = (actual read # in library X ÷ LNF_x)/(actual read # in library Y ÷ LNF_y). In case read # in library Y is zero, adjusted read number (ARN), instead of RA, is calculated as: ARN_x = actual read # in library X × LNF_y/LNF_x. Some of the contigs in CIFH, whose RAs or ARNs are above certain thresholds, are categorized into UP, DN, HC, and FB: UP for up-regulated genes (RA_IF/CF >5, RA_IH/CH >8, ARN_IF >10 when RN_CF =0, or ARN_IH >10 when RN_CH =0), DN for down-regulated genes (RA_CF/IF >10, RA_CH/IH >10, ARN_CF >20 when RN_IF =0, or ARN_CH >20 when RN_IH =0), HC and FB for genes preferentially expressed in hemocytes (RA_IH/IF >40, RA_CH/CF >40, ARN_IH >80 when RN_IF =0, or ARN_CH >80 when RN_CF =0) and fat body (RA_IF/IH >100, RA_CF/CH >100, ARN_IF >200 when RN_IH =0, or ARN_CF >200 when RN_CH =0), respectively.

2.4. Sequence extension, database search, and domain prediction

CIFH contigs in UP, DN, HC, and FB categories were used as queries to search local databases of 06CIFH_contigs/singletons, UK_gut_contigs by BLASTN (http://darwin.biochem.okstate.edu/blast/blast.html). The M. sexta midgut ESTs (i.e. UK_gut_contigs) (Pauchet et al., 2009) were kindly provided by Dr. Yannick Pauchet at University of Exeter, UK. The search results were used to extend the CIFH contigs or, in some cases, fill a gap between two contig sequences. The extended sequences were searched against NCBI using BLASTX as described above. For UP CIFH contigs lacking BLAST hits, a set of more stringent conditions was applied to select sequences for further analysis: a) RA_IF/CF >15, RA_IH/CH >15, ARN_IF >30 when RN_CF =0, or ARN_IH >30 when RN_CH =0, b) total read number >70, and c) GC content ≥35% (i.e. coding region-including). Open reading frames in a chosen contig were examined for leader peptide using SignalP 3.0 (http://www.cbs.dtu.dk/services/SignalP/), which is commonly found in proteins highly induced upon immune challenge (Jiang, 2008; Ragan et al., 2009). The polypeptide sequences were then analyzed to detect conserved domain structures by SMART (http://smart.embl-heidelberg.de/smart/set_mode.cgi).

3. Results

3.1. Identification of differentially regulated genes in M. sexta by pyrosequencing

In order to find immunity-related genes expressed in fat body or hemocytes based on their expression profiles, we isolated mRNA of these two tissues from naïve and bacteria-injected larvae of M. sexta, a lepidopteran insect whose genome sequence has not yet been determined. Using random dodecanucleotide primers that annealed to different regions of mRNA molecules, we generated four cDNA libraries: CF, CH, IF and IH. To facilitate assembly and ORF identification, we adopted long-read Titanium chemistry to sequence these libraries on a 454 GS-FLX pyrosequencer and obtained a total of 227,302 reads from CF, 647,587 reads from CH, 405,739 reads from IF, and 541,024 reads from IH (Table 1). The total number of reads from two plates (0.5 plate per library) was 1,821,652, which was 19.1-fold higher than that from one plate (95,358 reads) sequenced on a 454 GS20 in 2006 (Zou et al., 2008). There also was a substantial increase in average read length from 185 bp to 289 bp, but that was still much lower than what the manufacturer claimed (>400 bp) (http://454.com/about-454/index.asp).

Table 1.

Summary statistics for pyrosequencing analysis of M. sexta ESTs

	06 a	CF	CH	IF	IH	CIFH b	06CIFH c
Total number of reads	95,458 (95,358)	227,302	647,587	405,739	541,024	1,821,652	1,917,110
Average reads length (bp)	185 (185)	296	287	293	287	289	284
Total number of contigs	1,471 (7,231)	2,118	11,540	4,063	10,600	19,020	19,504
Contigs size (avg./longest in bp)	391/3,552 (300/3,909)	770/12,740	827/11,667	764/8,482	832/10,591	923/23,095	911/23,097
Total assembled reads	64,874 (69,429)	191,156	561,054	349,028	465,561	1,677,738	1,757,333
Singlet reads	28,518 (25,929)	32,518	68,861	49,444	61,108	108,587	120,670
Singlet length (avg. in bp)	179	244	245	235	254	209	200
Total BLASTable sequences	29,989	34,636	80,401	53,507	71,708	127,607	140,174
Orphan sequences (no BLAST match, #/%)	19,963/67	17,982/52	51,968/65	28,649/54	46,521/65	73,915/58	89,948/64
Contigs and reads with functional assignment	10,026	16,654	28,433	24,858	25,187	53,692	50,226

^aResults from reanalysis of the 2006 sequence data. The numbers in parentheses (adopted from Zou et al., 2008) are listed for comparison with the new results.

^bAnalysis of the 2009 EST sequences of control fat body (CF), control hemocytes (CH), induced fat body (IF), and induced hemocytes (IH) from M. sexta larvae.

^cAnalysis of the combined reads of 2006 (raw flow signals interpreted with the up-graded software) and 2009 (CF, CH, IF, and IH).

We assembled the reads into five datasets: CF, CH, IF, IH, and CIFH (Fig. 1). The first four each came from its respective library, whereas the 5^th dataset was assembled from the 1,821,652 reads in the four libraries sequenced in 2009. In CF, CH, IF and IH, 84.1~86.6% of the total reads were incorporated into contigs at average sizes of 764~832 bp; In CIFH, 1,677,738 (92.1%) of the 1,821,652 reads were assembled to 19,020 contigs at 923 bp per contig (Table 1). These assemblies were better than the previous one, that integrated 69,429 (72.8%) of the 95,358 reads into 7,231 contigs at an average length of 300 bp (Zou et al., 2008). To improve the transcriptome coverage, we used the latest version of Newbler to re-analyze the previously generated flowgrams, assembling 64,874 of the reads into 1,471 contigs with an average of 391 bp per contig in the 6^th dataset (“06”) (Table 1). Finally we assembled all the source libraries (2006, CF, CH, IF, and IH) into “06CIFH”, which contained 19,504 contigs (average size: 911 bp) and 120,670 singletons.

We used numbers of CH, CF, IH, and IF reads for each CIFH contig to identify differentially regulated genes. Since read numbers depended on library sizes and needed to be normalized against control genes, we compared frequencies of commonly used internal standards in each of the four libraries and found that six ribosomal protein genes (rpS2–rpS5, rpL4 and rpL8) showed low coefficients of variation (<30%) and high total read numbers (>1,000). So, we used the sums of their read numbers 825 (CF), 3,980 (CH), 1,618 (IF), and 3,352 (IH) as library normalization factors (LNFs) to calibrate read numbers and calculate relative abundances (RAs) (Fig. 1). Based on the RA values, 920 or 4.84% of the 19,020 contigs in CIFH were categorized into four groups: UP (Table S1) and DN (Table S2) for up- and down-regulated genes upon immunization; HC (Tables S3) and FB (Table S4) for genes preferentially expressed in hemocytes and fat body, respectively.

3.2. Sequence analysis and function prediction of UP genes

We discovered 528 CIFH contigs whose RA_IF/CF or RA_IH/CH was greater than 5 and 8, respectively, or whose adjusted number of IF (or IH) reads (ARN) was >10 when the CF (or CH) read was zero – the adjustment for IF was read # × 825/1618 and that for IH was read # × 3980/3352 (Table S1). As we anticipated, these contigs encoded polypeptides either identical to immunity-related proteins previously isolated from M. sexta (e.g. hemolin), or similar in sequence or domain structure to defense factors found in other insects (e.g. Spodoptera frugiperda X-tox), or related to proteins previously not known to play a role in immune responses (e.g. carboxylesterase), or having no significant sequence similarity to known proteins. In the following, we describe these contigs in the order of their putative immune functions.

A. Recognition of molecular patterns associated with microbes

To reinforce detection of invading organisms, certain pattern recognition receptors (PRRs) are synthesized in insects at higher levels after the initial encounter of foreign entities or abnormal host components. For instance, we found an Ig-domain protein (contig 03442) had an RA_IF/CF of 748.5 (Table 2). This protein, M. sexta hemolin, was reported previously as a highly inducible PRR that recognizes LPS of Gram-negative bacteria (Ladendorff and Kanost, 1991). Other PRRs included M. sexta immulectin-2 (contig 04775, RA_IF/CF: 45.4), immulectin-4 (contig 04808, ARN_IF: 217.2), peptidoglycan recognition protein-1 (PGRP1) (contig 13190, ARN_IH: 10.7; contig 14104, RA_IF/CF: 6.3; ARN_IH: 15.4), PGRP2 (contig 14700, residues 1–96, ARN_IF: 93.3; contig 14752, residues 98–196, ARN_IF: 60.2), β-1,3-glucan recognition protein-2 (βGRP2) (contig 01326, RA_IF/CF: 9.7; RA_IH/CH: 9.2). These data not only confirmed the published PRR sequences but also provided information on fold increases in their transcript abundances. Contig 06630 (RA_IF/CF: 11.2), 58% identical to M. sexta immulectin-3 (Yu et al., 2005) in residues 1–276, represented a previously unknown immulectin discovered based on its induced expression as well as sequence similarity. Newly identified PRRs also included PGRP3 (contig 00575, RA_IF/CF: 44.0), homologs of Bombyx mori PGRP5 (contig 11845, RA_IH/CH: 10.1) and PGRP-S6 (contig 08467, ARN_IF: 57.6), homologs of B. mori CTL10 (contig 14515, residues 54–182, RA_IF/CF: 8.7; contig 15639, residues 233–308, RA_IF/CF: 5.6; contig 11458, residues 54–306, ARN_IF: 28.0), homolog of B. mori Gram-negative binding protein (contig 08247, RA_IH/CH: 10.7) (Tanaka et al., 2008), LPS-binding leureptin (contig 15857, RA_IH/CH: 10.7) (Zhu et al., 2010), Ig domain-containing hemicentin-1 (contig 00131, RA_IF/CF: 6.4) and -2 (contig 14278, RA_IF/CF: 8.7) (Vogel and Hedgecock, 2001). Therefore, expression profiling and sequence similarity together provided a powerful tool to discover process-related genes without a priori genome sequence.

Table 2.

A list of 19 UP CIFH contigs with similarity to pattern recognition receptors^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	IF/CF	IH/CH
00131	11	41	137	61	250	6.4	1.8	gi|198430641|ref|XP_002123478.1| hemicentin 1, Ig domains [Ciona intestinalis]
00575	3	0	259	5	267	44.0	5.9	gi|154240658|dbj|BAF74637.1| peptidoglycan recognition protein-D [Samia cynthiaricini]
01326	1	9	19	70	99	9.7	9.2	gi|52782739|sp|Q8ISB6.1|BGBP2_MANSE β-1,3-glucan recognition protein 2
03442	1	12	1468	40	1521	748.5	4.0	gi|511297|gb|AAC46916.1| hemolin [Manduca sexta]
04775	1	0	89	0	90	45.4	0.0	gi|237869126|gb|AAF91316.3|AF242202_1 immulectin-2 [Manduca sexta]
04808	0	0	426	2	428	217.2	2.4	gi|237861314|gb|AAV41237.2| immulectin-4 [Manduca sexta]
06630	2	40	44	77	163	11.2	2.3	gi|55139125|gb|AAV41236.1| immulectin-3 [Manduca sexta]
08247	27	2	122	18	169	2.3	10.7	gi|208972535|gb|ACI32828.1| β-1,3-glucan recognition protein 3 [Helicoverpa armigera]
08467	0	0	113	0	113	57.6	0.0	gi|112983866|ref|NP_001036858.1| peptidoglycan recognition protein-6 [Bombyx mori]
11458	0	0	55	0	55	28.0	0.0	gi|148298818|ref|NP_001091784.1| multi-binding protein [Bombyx mori]
11845	0	2	9	17	28	4.6	10.1	gi|18202160|sp|O76537.1|PGRP_TRINI peptidoglycan recognition protein
13190	15	0	117	9	141	4.0	10.7	gi|27733423|gb|AAO21509.1|AF413068_1 peptidoglycan recognition protein 1A [Manduca sexta]
14104	14	0	173	13	200	6.3	15.4	gi|27733423|gb|AAO21509.1|AF413068_1 peptidoglycan recognition protein 1A [Manduca sexta]
14278	1	34	17	179	231	8.7	6.3	gi|83583693|gb|ABC24706.1| hemicentin-like protein 1, Ig domains [Spodoptera frugiperda]
14515	2	0	34	0	36	8.7	0.0	gi|148298818|ref|NP_001091784.1| multi-binding protein [Bombyx mori]
14700	0	0	183	2	185	93.3	2.4	gi|260765453|gb|ACX49764.1| peptidoglycan recognition protein 2 [Manduca sexta]
14752	0	0	118	2	120	60.2	2.4	gi|260765453|gb|ACX49764.1| peptidoglycan recognition protein 2 [Manduca sexta]
15639	10	0	109	0	119	5.6	0.0	gi|148298818|ref|NP_001091784.1| multi-binding protein [Bombyx mori]
15857	0	1	0	9	10	0.0	10.7	gi|27733411|gb|AAO21503.1|AF413062_1 leureptin, LPS-binding [Manduca sexta]

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RAIF/CF >5, RAIH/CH >8, ARNIF >10 when RNCF =0, or ARNIH >10 when RNCH =0. RAIF/CF and RAIH/CH values are shown in red if they are greater than 5 and 8, respectively. ARNIF and ARNIH values are shown in blue if they are higher than 10. In the columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. The complete list of 528 UP CIFH contigs is in Table S1.

B. Extracellular signal transduction and modulation

Hemolymph proteinases (HPs) in insect plasma form enzyme cascades to detect pathogen-PRR complexes and activate precursors of defense proteins (e.g. PO, spätzle, serine proteinase homolog (SPH), and plasmatocyte-spreading peptide (PSP) by limited proteolysis (Jiang and Kanost, 2000). We found eight HPs in the UP list: M. sexta HP7 (ARN_IF: 11.2), HP9 (RA_IH/CH: 28.5), HP17 (ARN_IH: 15.4), HP18 (RA_IH/CH: 40.4), HP19 (RA_IF/CF: 7.1), HP22 (RA_IF/CF: 5.1), proPO-activating proteinase-2 (PAP2) (ARN_IF: 50.0), and PAP3 (ARN_IF: 22.9) (Table 3). Expression profiles associated with the immune inducibility agreed well with the RT-PCR and northern blot results published earlier (Jiang et al., 2003a, 2003b, 2005). We also found six contigs encoding isoforms of a strongly inducible protein (scolexin) that contained all three catalytic residues of S1A proteinases but did not display any amidase activity (Finnerty et al., 1999). The high ratios and read numbers of these contigs (RA_IF/CF: 338.6 and 551.2; ARN_IF: 70.9, 129.5, 145.3, 169.8) suggested that primer binding and reverse transcriptase pausing were biased at certain sites of the template because, otherwise, there should not have been any gap for such a short ORF of ~1.36 kb. The exact role of scolexin in defense is still unclear.

Table 3.

A list of 40 UP CIFH contigs with similarity to extracellular signal modulators^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	IF/CF	IH/CH
00915	0	21	21	26	68	10.7	1.5	gi|91084647|ref|XP_966816.1| AGAP002414-PA, Zn protease [Tribolium castaneum]
00940	0	0	209	7	216	106.6	8.3	gi|1352212|sp|P48861.1|DDC_MANSE dopa decarboxylase DDC
02023	1	0	33	7	41	16.8	8.3	gi|148611442|gb|ABQ95973.1| tyrosine hydroxylase isoform A [Manduca sexta]
01667	0	7	98	33	138	50.0	5.6	gi|26006435|gb|AAL76085.1| proPO-activating proteinase-2 [Manduca sexta]
01818	0	26	45	66	137	22.9	3.0	gi|60299972|gb|AAX18637.1| proPO-activating proteinase-3 [Manduca sexta]
02361	7	4	70	1	82	5.1	0.3	gi|56418425|gb|AAV91020.1| hemolymph proteinase 22 [Manduca sexta]
02382	0	2	109	69	180	55.6	41.0	gi|4090964|gb|AAD09279.1| immune-related Hdd1 [Hyphantria cunea]
02693	21	7	310	19	357	7.5	3.2	gi|27733415|gb|AAO21505.1|AF413064_1 serpin 3a [Manduca sexta]
02813	108	9	313	72	502	1.5	9.5	gi|242351233|gb|ACS92763.1| serine proteinase-like protein 1b [Manduca sexta]
02985	3	0	158	0	161	26.9	0.0	gi|56418466|gb|AAV91027.1| serine proteinase-like protein 4 [Manduca sexta]
03018	0	54	22	79	155	11.2	1.7	gi|56418395|gb|AAV91005.1| hemolymph proteinase 7 [Manduca sexta]
03778	0	11	192	28	231	97.9	3.0	gi|74813957|sp|Q86RS3.1|DFP_MANSE immune-related Hdd11, precursor
03989	0	1	8	24	33	4.1	28.5	gi|56418399|gb|AAV91007.1| hemolymph proteinase 9 [Manduca sexta]
05186	0	0	8	13	21	4.1	15.4	gi|56418413|gb|AAV91014.1| hemolymph proteinase 17 [Manduca sexta]
05606	1	0	19	4	24	9.7	4.7	gi|4090968|gb|AAD09281.1| immune-related Hdd13 [Hyphantria cunea]
05831	3	8	97	25	133	16.5	3.7	gi|45594232|gb|AAS68507.1| serpin-5A [Manduca sexta]
06149	21	22	686	32	761	16.7	1.7	gi|27733421|gb|AAO21508.1|AF413067_1 serine protease-like protein [Manduca sexta]
06215	29	1	108	8	146	1.9	9.5	gi|112983872|ref|NP_001036857.1| serpin-like protein (SEP-LP) or serpin-12 [Bombyx mori]
06581	0	0	13	10	23	6.6	11.9	gi|4090970|gb|AAD09282.1| immune-related Hdd23 [Hyphantria cunea]
07639	651	0	1237	14	1902	1.0	16.6	gi|134436|sp|P14754.1|SERA_MANSE serpin-1
08231	0	1	0	34	35	0.0	40.4	gi|56418417|gb|AAV91016.1| hemolymph proteinase 18 [Manduca sexta]
10791	1	0	1081	1	1083	551.2	1.2	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]
10792	0	0	333	0	333	169.8	0.0	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]
13453	5	4	58	7	74	5.9	2.1	gi|45594232|gb|AAS68507.1| serpin-5 [Manduca sexta]
13454	0	1	17	10	28	8.7	11.9	gi|45594232|gb|AAS68507.1| serpin-5 [Manduca sexta]
14093	1	0	14	0	15	7.1	0.0	gi|56418419|gb|AAV91017.1| hemolymph proteinase 19 [Manduca sexta]
14248	0	6	0	196	202	0.0	38.8	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
14393	2	4	132	11	149	33.7	3.3	gi|27733421|gb|AAO21508.1|AF413067_1 serine protease-like protein [Manduca sexta]
14456	0	0	1	52	53	0.5	61.7	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
15055	1	1	16	0	18	8.2	0.0	gi|112983896|ref|NP_001037394.1| paralytic peptide binding protein 1 [Bombyx mori]
15111	1	48	8	800	857	4.1	19.8	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
16520	1	0	664	1	666	338.6	1.2	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]
16917	0	40	2	519	561	1.0	15.4	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
17048	0	1	0	95	96	0.0	112.8	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
17058	0	32	4	545	581	2.0	20.2	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
17751	0	24	1	269	294	0.5	13.3	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
18441	0	1	0	65	66	0.0	77.2	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta]
18669	0	0	285	0	285	145.3	0.0	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]
18670	0	0	139	0	139	70.9	0.0	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]
18963	0	0	254	0	254	129.5	0.0	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta]

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RAIF/CF >5, RAIH/CH >8, ARNIF >10 when RNCF =0, or ARNIH >10 when RNCH =0. RAIF/CF and RAIH/CH values are shown in red if they are greater than 5 and 8, respectively. ARNIF and ARNIH values are shown in blue if they are higher than 10. In the columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. The complete list of 528 UP CIFH contigs is in Table S1.

In the reaction of proPO activation, a high molecular weight complex of SPH1 and SPH2 has to be present along with PAP and proPO to generate active PO (Gupta et al., 2005). In this study, we identified SPH1 (contig 02813, RA_IH/CH: 9.5) and SPH2 (contig 6149, RA_IF/CF: 16.7; contig 14393, RA_IF/CF: 33.7) and confirmed their induced expression (Yu et al., 2003). Contig 02985 (RA_IF/CF: 27) contained a complete ORF coding for a regulatory clip domain followed by a serine proteinase-like domain. The protein, designated M. sexta SPH4, is 49% and 92% identical to SPH1 in the amino- and carboxyl-terminal domains, respectively. Such a disparity in sequence alterations suggests that the selection pressures or structural constraints for these two regions differ dramatically.

Functions of serine proteinases are modulated not only by SPHs but also by their inhibitors. Particularly, some members of the serpin superfamily regulate serine proteinase activities by forming covalent complexes with their cognate enzymes (Kanost, 1999). We have identified six serpins in the UP list (Table 3), five of which are known as M. sexta serpin-1 (contig 7639: ARN_IH: 16.6), serpin-2 (four contigs, ARN_IH: 61.7, RA_IH/CH: 13.3, 15.4, 20.2), serpin-2 homolog (four contigs, RA_IH/CH: 19.8, 38.8, 77.2, 112.8), serpin-3 (contig 2693, RA_IF/CF: 7.5), serpin-5 (three contigs, RA: 5.9, 11.9, 16.5). We have found a new serpin (contig 6215, RA_IH/CH: 9.5) and its ortholog in B. mori, SLP or serpin-12. The silkworm serpin was expressed in fat body of bacteria-injected larvae but not in fat body of naïve ones (Zou et al., 2009). Its transcription in hemocytes also was similar to that of the M. sexta serpin: the mRNA was low in naïve larvae and became higher in induced ones.

Besides serine proteinases, SPHs and serpins, we also have found other proteins that either mediate or regulate immune responses in M. sexta or other moths (Table 3). These include: tyrosine hydroxylase (contig 2023, RA_IF/CF: 16.8) (Gorman et al., 2007), dopa decarboxylase (contig 00940, ARN_IF: 106.6) (Noguchi et al., 2003), PSP-binding protein (contig 15055, RA_IF/CF: 8.2) (Matsumoto et al., 2003), and Zn proteinase (contig 0915, ARN_IF: 11) (Altincicek and Vilcinskas, 2008). Four immunity-related proteins, Hdd1, Hdd11, Hdd13, and Hdd23 (Shin et al., 1998), are included here even though their functions remain unknown.

C. Intracellular signaling pathways and their components

Pathogen recognition and signal transduction can either go through a PRR-SP system in insect plasma (e.g. spätzle processing for Toll activation) or directly binds to PRRs on the surface of immune tissues/cells (e.g. PGRP-LC binding for Imd activation in Drosophila). After that, intracellular proteins are mobilized to relay signals into the cell nucleus where transcriptional regulation occurs. As shown in Table 4, we have detected increase in transcript levels of the putative pathway members: Toll-like receptors (contigs 06893 and 18001, 68% and 94% similar in amino acid sequence to ABO21763) (Ao et al., 2008), cactus (contig 01044) (Furukawa et al., 2009), relish (contigs 04802 and 15532) (Tanaka et al., 2007), and eiger (contig 01020, a membrane-bound TNF homolog) (Kauppila et al., 2003). Other intracellular proteins possibly involved in signal transduction or modulation include a Ser/Thr protein kinase, GTP/GDP exchange factors, a receptor Tyr phosphatase, a protein phosphatase 2c, ankyrin repeat proteins, and vrille transcription factor.

Table 4.

A list of 18 UP CIFH contigs with similarity to intracellular signal transducers^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	IF/CF	IH/CH
00461	1	48	14	108	171	7.1	2.7	gi|47217104|emb|CAG02605.1| unnamed protein product, integrin β6 precursor [Tetraodon nigroviridis]
00537	1	32	10	32	75	5.1	1.2	gi|270009406|gb|EFA05854.1| TcasGA2_TC008649 Tyr protein kinase [Tribolium castaneum]
00671	1	12	10	46	69	5.1	4.6	gi|189235637|ref|XP_967498.2| ral guanine nucleotide exchange factor [Tribolium castaneum]
01020	42	4	63	27	136	0.8	8.0	gi|91082721|ref|XP_972476.1| ~ eiger CG12919-PA, JNK [Tribolium castaneum]
01044	9	70	163	105	347	9.2	1.8	gi|289629214|ref|NP_001166191.1| cactus [Bombyx mori]
01313	2	52	33	35	122	8.4	0.8	gi|242009174|ref|XP_002425367.1| Ser-Thr protein kinase, plant-type [P. humanus corporis]
01390	1	31	14	19	65	7.1	0.7	gi|46403173|gb|AAS92609.1| vrille transcription factor [Antheraea pernyi]
01970	1	16	12	29	58	6.1	2.2	gi|157118595|ref|XP_001659169.1| guanine nucleotide exchange factor [Aedes aegypti]
04802	2	42	25	66	135	6.4	1.9	gi|157412326|ref|NP_001098704.1| Relish2 [Bombyx mori]
05836	2	1	0	7	10	0.0	8.3	gi|189235110|ref|XP_971078.2| receptor tyrosine phosphatase type r2a [Tribolium castaneum]
06304	1	0	11	1	13	5.6	1.2	gi|170038257|ref|XP_001846968.1| dipeptidyl peptidase 4, apoptosis, immunity [Culex quinquefasciatus]
06868	0	1	1	11	13	0.5	13.1	gi|193713771|ref|XP_001946690.1| ankyrin repeat domain 54 [Acyrthosiphon pisum]
06893	0	1	1	20	22	0.5	23.7	gi|126635756|gb|ABO21763.1| Toll receptor [Manduca sexta]
11311	0	1	3	9	13	1.5	10.7	gi|189237512|ref|XP_972880.2| protein phosphatase type 2c [Tribolium castaneum]
11356	0	1	4	7	12	2.0	8.3	gi|156551808|ref|XP_001603899.1| arf6 guanine nucleotide exchange factor [Nasonia vitripennis]
13966	0	1	0	9	10	0.0	10.7	gi|190570736|ref|YP_001975094.1| ankyrin repeat protein [Wolbachia of C. quinquefasciatus Pel]
15532	1	19	12	9	41	6.1	0.6	gi|157412326|ref|NP_001098704.1| Relish2 [Bombyx mori]
18001	0	1	0	7	8	0.0	8.3	gi|126635756|gb|ABO21763.1| Toll receptor [Manduca sexta]

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RAIF/CF >5, RAIH/CH >8, ARNIF >10 when RNCF =0, or ARNIH >10 when RNCH =0. RAIF/CF and RAIH/CH values are shown in red if they are greater than 5 and 8, respectively. ARNIF and ARNIH values are shown in blue if they are higher than 10. In the columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. The complete list of 528 UP CIFH contigs is in Table S1.

D. Antimicrobial peptides/proteins

Overproduction of effector proteins that immobilize pathogens, block their proliferation, or directly kill them is a hallmark of insect immunity (Bulet et al., 2004). Consistent with this notion, we have detected 65 UP contigs encoding: A) antimicrobial peptides, B) low molecular weight proteinase inhibitors, C) lysozymes, and D) transferrins (Table 5). In group A, twenty-five contigs (06782, 07203, 08902, 11040, 11711, 13563, 14343, 14380, 14641, 15159, 15732, 15744, 15953, 15997, 16129, 16150, 16576, 17135, 17304, 17350, 17632, 17705, 18324, 18814, 18977) code for at least six attacins, eight (03746, 14568, 15998, 16292, 17184, 18150, 18699, 18819) for at least three X-tox (Girard et al., 2008), six (04903, 07116, 10853, 13916, 17301, 17434) for four lebocin-related proteins (Rayaprolu et al., 2010), four (12151, 13894, 14997, 15041) for three cecropins (Zhu et al., 2003), two (09484, 17439) for two moricins (Dai et al., 2008), and one (02067) for gloverin (Zhu et al., 2003). Group B consists of eight contigs (03142, 03674, 04175, 05197, 08286, 10722, 13936, 16018) encoding proteinase inhibitor-like proteins which may block proteinases released by bacteria, fungi, or parasites (Armstrong, 2006; Zang and Maizels, 2001). Group C has three contigs (08421, 15931, 16133) coding for two lysozymes (Mulnix and Dunn, 1994) that hydrolyze bacterial peptidoglycans. Group D includes seven contigs (02145, 11027, 14937, 16606, 17206, 18239, 18308) encoding at least two transferrins that may sequester iron and, by doing so, prevent bacteria from proliferation (Nichol et al., 2002).

Table 5.

A list of 65 UP CIFH contigs with similarity to antimicrobial proteins^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	IF/CF	IH/CH
02067	1	0	280	82	363	142.8	97.4	gi|110649240|emb|CAL25129.1| gloverin [Manduca sexta]
02145	0	15	20	95	130	10.2	7.5	gi|157134051|ref|XP_001663123.1| transferrin [Aedes aegypti]
03142	1	7	420	121	549	214.2	20.5	gi|33860163|sp|P82176.2|IMPI_GALME inducible metalloproteinase inhibitor IMPIα precursor
03674	1	0	5	21	27	2.5	24.9	gi|110347837|gb|ABG72720.1| protease inhibitor-like protein [Antherae amylitta]
03746	0	7	55	389	451	28.0	66.0	gi|148298709|ref|NP_001091749.1| possible antimicrobial peptide [Bombyx mori]
04175	0	7	40	45	92	20.4	7.6	gi|114052803|ref|NP_001040277.1| salivary Cys-rich peptide [Bombyx mori]
04903	0	0	279	6	285	142.3	7.1	gi|187281722|ref|NP_001119732.1| lebocin 3 precursor [Bombyx mori]
05197	0	0	20	1	21	10.2	1.2	gi|115392217|gb|ABI96910.1| brasiliensin precursor, thrombin inhibitor [Triatoma brasiliensis]
06782	0	0	102	17	119	52.0	20.2	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
07116	1	4	902	3	910	459.9	0.9	gi|171262319|gb|ACB45566.1| lebocin-like protein [Antheraea pernyi]
07203	2	3	312	22	339	79.5	8.7	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
08286	0	0	139	23	162	70.9	27.3	gi|56462340|gb|AAV91453.1| protease inhibitor 6 [Lonomia obliqua]
08421	4	2	28	99	133	3.6	58.8	gi|7327646|gb|AAB31190.2| lysozyme [Manduca sexta]
08902	0	0	164	14	178	83.6	16.6	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
09484	1	0	134	56	191	68.3	66.5	gi|29469961|gb|AAO74637.1| antimicrobial peptide moricin [Manduca sexta]
10234	0	1	249	7	257	127.0	8.3	gi|169264911|dbj|BAG12297.1| gallerimycin [Samia cynthia ricini]
10722	9	3	102	3	117	5.8	1.2	gi|110347833|gb|ABG72718.1| protease inhibitor-like protein [Antherae amylitta]
10853	0	0	113	1	114	57.6	1.2	gi|171262319|gb|ACB45566.1| lebocin-like protein [Antheraea pernyi]
11027	59	0	694	0	753	6.0	0.0	gi|136206|sp|P22297.1|TRF_MANSE transferrin precursor
11040	0	4	51	249	304	26.0	73.9	gi|29469969|gb|AAO74640.1| antimicrobial protein attacin 2 [Manduca sexta]
11711	0	7	85	1317	1409	43.3	223.4	gi|29469969|gb|AAO74640.1| antimicrobial protein attacin 2 [Manduca sexta]
12151	0	0	153	0	153	78.0	0.0	gi|116084|sp|P14665.1|CEC5_MANSE bactericidin B-5P, cecropin-like
13563	0	0	657	0	657	335.0	0.0	gi|110347786|gb|ABG72695.1| attacin-like protein [Antheraea mylitta]
13894	0	0	48	29	77	24.5	34.4	gi|112984238|ref|NP_001037460.1| cecropin B precursor [Bombyx mori]
13916	1	0	741	0	742	377.8	0.0	gi|219958086|gb|ACL68097.1| lebocin-related protein precursor [Manduca sexta]
13936	0	0	25	0	25	12.7	0.0	gi|123725|sp|P26227.1|HTIB_MANSE trypsin inhibitor B, BPTI-type
14343	0	0	186	7	193	94.8	8.3	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
14380	0	0	106	0	106	54.0	0.0	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
14568	0	0	2	68	70	1.0	80.7	gi|148298709|ref|NP_001091749.1| possible antimicrobial peptide [Bombyx mori]
14641	0	0	157	0	157	80.1	0.0	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
14937	13	0	164	0	177	6.4	0.0	gi|136206|sp|P22297.1|TRF_MANSE: transferrin precursor
14997	0	0	34	10	44	17.3	11.9	gi|29469965|gb|AAO74638.1| antimicrobial peptide cecropin 6 [Manduca sexta]
15041	0	0	36	0	36	18.4	0.0	gi|116084|sp|P14665.1|CEC5_MANSE bactericidin B-5P, cecropin-like
15159	0	0	0	15	15	0.0	17.8	gi|15963410|dbj|BAB69462.1| attacin [Samia cynthia ricini]
15732	0	1	253	43	297	129.0	51.1	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
15744	0	0	0	35	35	0.0	41.6	gi|29469969|gb|AAO74640.1| antimicrobial protein attacin 2 [Manduca sexta]
15931	40	37	1504	364	1945	19.2	11.7	gi|7327646|gb|AAB31190.2| lysozyme [Manduca sexta]
15953	1	0	43	6	50	21.9	7.1	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
15997	0	0	142	4	146	72.4	4.7	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
15998	0	0	1	10	11	0.5	11.9	gi|73921456|gb|AAZ94260.1| immune related protein X-tox [Spodoptera frugiperda]
16018	0	0	40	12	52	20.4	14.2	gi|116833115|gb|ABK29470.1| immune reactive putative protease inhibitor [Helicoverpa armigera]
16129	1	0	212	35	248	108.1	41.6	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
16133	47	57	1719	440	2263	18.6	9.2	gi|233964|gb|AAB19535.1| lysozyme
16150	0	1	145	3	149	73.9	3.6	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
16292	0	0	1	34	35	0.5	40.4	gi|148298709|ref|NP_001091749.1| possible antimicrobial peptide [Bombyx mori]
16576	0	0	0	18	18	0.0	21.4	gi|74767320|sp|Q5MGE6.1|DFP3_LONON defense protein 3 precursor, attacin E
16606	8	0	164	0	172	10.5	0.0	gi|136206|sp|P22297.1|TRF_MANSE transferrin precursor
17135	0	9	103	1157	1269	52.5	152.6	gi|110649242|emb|CAL25130.1| attacin II [Manduca sexta]
17184	0	11	76	449	536	38.8	48.5	gi|73921456|gb|AAZ94260.1| immune related protein, X-tox [Spodoptera frugiperda]
17206	3	0	136	0	139	23.1	0.0	gi|136206|sp|P22297.1|TRF_MANSE transferrin precursor
17301	1	0	272	0	273	138.7	0.0	gi|219958086|gb|ACL68097.1| lebocin-related protein precursor [Manduca sexta]
17304	0	1	412	13	426	210.1	15.4	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
17350	0	0	205	0	205	104.5	0.0	gi|29469969|gb|AAO74640.1| antimicrobial protein attacin 2 [Manduca sexta]
17434	1	0	314	0	315	160.1	0.0	gi|219958086|gb|ACL68097.1| lebocin-related protein precursor [Manduca sexta]
17439	0	0	98	31	129	50.0	36.8	gi|110649236|emb|CAL25127.1| moricin [Manduca sexta]
17632	0	0	83	6	89	42.3	7.1	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
17705	0	0	36	0	36	18.4	0.0	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
18150	0	0	0	18	18	0.0	21.4	gi|148298709|ref|NP_001091749.1| possible antimicrobial peptide [Bombyx mori]
18239	3	0	67	0	70	11.4	0.0	gi|136206|sp|P22297.1|TRF_MANSE transferrin precursor
18308	15	0	169	0	184	5.7	0.0	gi|136206|sp|P22297.1|TRF_MANSE transferrin precursor
18324	0	0	25	0	25	12.7	0.0	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
18699	0	1	26	114	141	13.3	135.4	gi|148298709|ref|NP_001091749.1| possible antimicrobial peptide [Bombyx mori]
18814	0	0	235	29	264	119.8	34.4	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]
18819	0	5	59	405	469	30.1	96.2	gi|73921456|gb|AAZ94260.1| immunity-related protein X-tox [Spodoptera frugiperda]
18977	0	1	20	2	23	10.2	2.4	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta]

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RA_IF/CF >5, RA_IH/CH >8, ARN_IF >10 when RN_CF =0, or ARN_IH >10 when RN_CH =0. RA_IF/CF and RA_IH/CH values are shown in red if they are greater than 5 and 8, respectively. ARN_IF and ARN_IH values are shown in blue if they are higher than 10. In the columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. The complete list of 528 UP CIFH contigs is in Table S1.

E. Other up-regulated genes

Among the 528 UP contigs, 177 did not have any BLAST hits (Table S1), indicating that some of them may encode polypeptides previously not known to be involved in immunity. To ensure these sequences are indeed up-regulated, we selected contigs with RA >15 (or ARN >30) and total read numbers >70. We then extended these contigs, if possible, with sequences in dataset “06” (Table 1) and in the M. sexta gut EST dataset (Pauchet et al., 2010). After eliminating the contigs with GC-contents <35% (hence, likely representing 5′ or 3′ AT-rich untranslated regions of up-regulated genes), we examined the remaining ones in greater detail (Table 6). Contigs 00327, 01714, 04720, 05532, and 07536 contain ORFs with a secretion signal peptide. The putative mature proteins (41, 61, 37, 86, 179 residues long) could be novel AMPs or in other ways involved in immunity. Contig 02467 encodes a secreted protein containing ten Cys that may tether the 139-residue polypeptide into a stable domain functioning as a proteinase inhibitor or an antifungal protein (Kanost, 1999). Contigs 15852 and 17316 encode proteins with 2 and 3 Kazal-type proteinase inhibitor domains, respectively. Contigs 17537 and 17568 encode proteins with a DM9 domain. Contigs 03381 and 15910, after extension, are found to be a part of cactus and serpin-2 transcripts. The other contigs encode sequences similar to B. mori heat shock protein 25.4, SPH, and esterases.

Table 6.

A list of 22 UP CIFH contigs without BLAST hit^*

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RA_IF/CF >15, RA_IH/CH >15, ARN_IF >30 when RN_CF =0, or ARN_IH >30 when RN_CH =0. Contigs with total read numbers lower than 70 or GC content lower than 35% are not listed. Some of the contig sequences have been extended using sequences in dataset “06” (Table 1, Zou et al., 2008) and in the M. sexta gut EST dataset (Pauchet et al., 2009). RA_IF/CF and RA_IH/CH values are shown in red if they are greater than 15, while ARN_IF and ARN_IH values are shown in blue if they are higher than 30. In the two columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. The complete list of 528 UP CIFH contigs is in Table S1. The contigs labeled with the same letter (a to e) in superscript indicate high sequence similarity between them, as highlighted with different colors at certain key sites of the protein sequences. Underlined sequences represent putative signal peptides. The * indicates the end of protein sequence (stop codon).

3.3. Sequence analysis and function prediction of DN genes

The analysis of down-regulated genes yielded results that surprised us at first: among the 53 DN CIFH contig groups with BLAST hits, ten were closely related to immune responses (Table 7). A contig group represents a single contig in most cases but, in other times, has multiple contigs with the same BLAST hit, which may come from different genes. They include lectins (06497, 07642, 11280, 13813, 14570, 14760), lacunin (00015), HP1 (16288), and proPOs (17085 and 17958). A closer inspection of the data indicated that the decreases in mRNA levels seem to always occur in fat body instead of hemocytes. Since these genes were all expressed at much higher levels in hemocytes than fat body (RA_{CH/CF or IH/IF} >40), we suggest the apparent down regulation in fat body were caused by unequal contamination of fat body tissue by hemocytes: somehow there was much less contamination in induced fat body of these hemocyte-specific transcripts. In hemocytes, their average RA_CH/IH was only 2.1 – no major down-regulation was observed for these immunity-related genes in cells mainly expressing them. It is likely that similar contamination of fat body tissue by hemocytes also resulted in the observation of genes not known to be directly related to immunity, which includes 11 contig groups (00010, 00248, 00379, 00623, 00628, 03286, 03654, 07139, 08686, 10124, 13842) with RA_{CH/CF or IH/IF} >40 (hemocyte-specific) and RA_CF/IF >10 (fat body DN) but RA_CH/IH <3.

Table 7.

A list of DN CIFH contigs with BLAST hits^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	CF/IF	CH/IH
00010	29	464	3	286	782	19.0	1.4	gi|242005387|ref|XP_002423550.1| cAMP-dependent protein kinase subunit [Pediculus humanus corporis]
00015 &	112	4705	17	3918	8752	12.9	1.0	gi|6164595|gb|AAF04457.1|AF078161_1 lacunin [Manduca sexta] [00015, 02717]
00248	7	200	1	155	363	13.7	1.1	gi|157113908|ref|XP_001657920.1| n-acetyllactosaminide β-1,3-NAG transferase [Aedes aegypti]
00379	10	308	1	184	503	19.6	1.4	gi|170037242|ref|XP_001846468.1| Leu-rich repeat-containing protein 1 [Culex quinquefasciatus]
00623	12	527	1	443	983	23.5	1.0	gi|157132531|ref|XP_001656056.1| odd Oz protein [Aedes aegypti]
00628	7	38	1	39	85	13.7	0.8	gi|170030982|ref|XP_001843366.1| rho/rac/cdc GTPase-activating protein [Culex quinquefasciatus]
00773	49	12	93	1	155	1.0	10.1	gi|157103945|ref|XP_001648193.1| dihydropyrimidine dehydrogenase [Aedes aegypti]
00851	6	42	1	26	75	11.8	1.4	gi|158300087|ref|XP_320080.3| AGAP009284-PA [Anopheles gambiae]
01289	7	45	1	31	84	13.7	1.2	gi|187281809|ref|NP_001119723.1| kinesin-like protein Ncd [Bombyx mori]
02637	5	12	9	1	27	1.1	10.1	gi|116789445|gb|ABK25249.1| unknown [Picea sitchensis]
02730	8	15	8	1	32	2.0	12.6	gi|2970687|gb|AAC06038.1| β-glucosidase precursor [Spodoptera frugiperda]
03286 etc.	62	1976	7	586	2631	17.4	2.8	gi|254746344|emb|CAX16637.1| C1A Cys protease [Manduca sexta] [03286, 05560, 15201, 17978]
03654	21	686	2	647	1356	20.6	0.9	gi|157134123|ref|XP_001663157.1| atlastin [Aedes aegypti]
03792	7	20	1	5	33	13.7	3.4	gi|91090218|ref|XP_968156.1| E1a binding protein P400 [Tribolium castaneum]
03996	6	6	1	6	19	11.8	0.8	gi|170052039|ref|XP_001862040.1| small GTP-binding protein [Culex quinquefasciatus]
05824	8	0	1	4	13	15.7	0.0	gi|116326818|ref|YP_803355.1| hypothetical TNAV2c gp132 [Trichoplusia ni ascovirus 2c]
06497 etc.	225	10451	12	4266	14954	36.8	2.1	gi|217262|dbj|BAA03124.1| lectin [Bombyx mori] [06497, 15047, 15764, 16677, 16801, 16877, 16886, 17700]
06713	0	12	0	1	13	0.0	10.1	gi|193613364|ref|XP_001943860.1| limkain b1 [Acyrthosiphon pisum]
06902	12	3	2	0	17	11.8	2.5	gi|114050917|ref|NP_001040414.1| 3-hydroxyacyl-CoA dehydrogenase [Bombyx mori]
07139	21	767	2	262	1052	20.6	2.5	gi|110649216|emb|CAL25117.1| dVA-AP3 [Manduca sexta]
07515	7	1	1	0	9	13.7	0.8	gi|158295141|ref|XP_316035.4| AGAP005993-PA [Anopheles gambiae]
07642	9	601	1	153	764	17.7	3.3	gi|55139125|gb|AAV41236.1| immulectin-3 [Manduca sexta]
07754	0	12	1	1	14	0.0	10.1	gi|71895231|ref|NP_001026433.1| coiled-coil domain containing 93 [Gallus gallus]
08686 &	21	854	3	680	1558	13.7	1.1	gi|82880638|gb|ABB92836.1| scavenger receptor C-like protein [Spodoptera frugiperda] [08686, 15116]
08705	8	10	1	5	24	15.7	1.7	gi|224084416|ref|XP_002192181.1| selenium binding protein 1 [Taeniopygia guttata]
08707	6	9	1	13	29	11.8	0.6	gi|24585081|ref|NP_609923.2| CG10639 [Drosophila melanogaster]
08801	1	14	1	1	17	2.0	11.8	gi|91081401|ref|XP_972667.1| exosome component 8 [Tribolium castaneum]
09847	0	13	0	1	14	0.0	10.9	gi|194745608|ref|XP_001955279.1| GF16313 [Drosophila ananassae]
10124 etc.	115	4638	8	2848	7609	28.2	1.4	gi|114050871|ref|NP_001040411.1| carboxylesterase [Bombyx mori] [10124, 16922, 17330, 18860]
10316	0	13	1	1	15	0.0	10.9	gi|157106599|ref|XP_001649397.1| hypothetical protein AaeL_AAEL004554 [Aedes aegypti]
10439	12	0	1	0	13	23.5	0.0	gi|183979241|dbj|BAG30782.1| cuticular protein CPR41B [Papilio xuthus]
11030	13	0	2	0	15	12.7	0.0	gi|3121953|sp|Q25504.1|CU16_MANSE larval cuticle protein 16/17 precursor
11098	40	0	3	0	43	26.1	0.0	gi|159526|gb|AAA29320.1| Met-rich storage protein 1 [Manduca sexta]
11161	0	12	1	1	14	0.0	10.1	gi|125808686|ref|XP_001360831.1| GA18253 [Drosophila pseudoobscura]
11280 etc.	143	7866	11	2589	10609	25.5	2.6	gi|91090548|ref|XP_971239.1| hemolectin CG7002-PA [Tribolium castaneum] [11280, 15506, 15594, 18551]
12095	10	0	1	0	11	19.6	0.0	gi|194741936|ref|XP_001953465.1| GF17208 [Drosophila ananassae]
12848	0	16	0	1	17	0.0	13.5	gi|2822109|sp|P14730.2|EXPI_RAT extracellular peptidase inhibitor, WDNM1 precursor
13013	7	1	1	0	9	13.7	0.8	gi|189031278|gb|ACD74812.1| cuticle protein 1 [Helicoverpa armigera]
13094	15	10	1	5	31	29.4	1.7	gi|183979298|dbj|BAG30762.1| similar to CG5304-PA [Papilio xuthus]
13813	31	2398	4	848	3281	15.2	2.4	gi|110758905|ref|XP_395067.3| hemolectin CG7002-PA [Apis mellifera]
13842	14	677	2	228	921	13.7	2.5	gi|138601|sp|P19616.1|VITM_MANSE microvitellogenin precursor
14129	7	0	1	0	8	13.7	0.0	gi|91078692|ref|XP_971204.1| phospholipase A2, grp VI (cytosolic, Ca-independent) [Tribolium castaneum]
14570 etc.	559	28386	29	10677	39651	37.8	2.2	gi|162462371|ref|NP_001104817.1| lectin [Bombyx mori] [14570, 15250, 15380, 15792, 16289, 16291, 16594, 16842, 17159, 17421, 17471, 17732, 17769, 18032, 18067, 18097, 18286, 18326, 18719, 18721, 18794, 18997]
14760 etc.	57	3372	3	1184	4616	37.3	2.4	gi|156545430|ref|XP_001606650.1| CG7002-PA [Nasonia vitripennis] [14760, 18045]
14781	28	0	3	1	32	18.3	0.0	gi|114052677|ref|NP_001040269.1| phosphoserine aminotransferase 1 [Bombyx mori]
15132	9	0	1	0	10	17.7	0.0	gi|112984526|ref|NP_001037199.1| promoting protein [Bombyx mori]
15465	6	0	1	1	8	11.8	0.0	gi|170574840|ref|XP_001892989.1| hypothetical protein Bm1_07595 [Brugia malayi]
16105	10	23	1	42	76	19.6	0.5	gi|91087179|ref|XP_975411.1| CG9471-PB [Tribolium castaneum]
16288 etc.	63	3044	4	1126	4237	30.9	2.3	gi|2738863|gb|AAB94557.1| hemocyte protease-1 [Manduca sexta] [16288, 16719, 17102]
17085 etc.	236	11035	27	7455	18753	17.1	1.2	gi|74763772|sp|O44249.3|MANSE proPO-p1 [17085, 17315, 17420, 17612, 17629, 18065, 18463, 18887]
17958 etc.	130	5309	19	3669	9127	16.3	1.2	gi|75038472|sp|Q25519.3|MANSE proPO-p2 [17958, 18004, 18516]
18482	11	0	0	0	11	21.6	0.0	gi|114240|sp|P14296.1|ARYA_MANSE arylphorin α subunit precursor
18611	0	12	4	1	17	0.0	10.1	gi|12585261|sp|Q9U639.1|HSP7D_MANSE heat shock 70 kDa protein cognate 4 (Hsp70-4)

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RA_CF/IF >10, RA_CH/IH >10, ARN_CF >20 when RN_IF =0, or ARN_CH >20 when RN_IH =0. RA_CF/IF and RA_CH/IH values are shown in red if they are greater than 10, whereas ARN_CF and ARN_CH values are shown in blue if they are higher than 20. In the two columns of RA or ARN, cells shaded yellow and blue represent fat body- and hemocyte-specific gene expression, respectively. Contigs with identical BLAST results are combined, with their average RAs or ARNs calculated based on the sums of original reads in CF, CH, IF, and IH for each group. Contigs with no BLAST hit can be found in Table S2, a complete list of 148 DN CIFH contigs.

After eliminating contigs whose RA_{CH/CF or IH/IF} calculated from low read numbers, we have found four DN contigs: 02730 encodes a β-glucosidase, 11098 a Met-rich storage protein, 12848 a proteinase inhibitor, and 14781 a phosphoserine amino transferase. Follow-up studies are needed to confirm their down-regulation and explore physiological relevance of the decrease in transcript levels.

3.4. Tissue-specifically regulated genes in larval hemocytes

Using the same set of read numbers in CIFH, we found 45 contig groups representing genes preferentially expressed in hemocytes. Interestingly, this tissue-specific pattern (RA >40 or ARN_IH >80) was only found in the induced samples but not in the control ones (Table 8). A closer examination of the data uncovered the possible reason for this bias: although fat body was collected under the same conditions, more hemocytes attached to the control fat body tissue than the induced one. Consequently, higher read numbers from contaminating hemocytes in control fat body led to much lower RA_CH/CF values than their corresponding RA_IH/IF’s. While the same reason caused wrong identification of some contigs as down-regulated ones (Table 7), the skewing of RAs against the control samples (i.e. lower RA_CH/CF) did not seem to affect the correct calling of hemocyte-specificity in a qualitative term. For the entire contig groups, the sums of CF and CH reads were 2173 and 105143, respectively. The average RA_CH/CF of 10.0 was much lower than the cutoff value of 40 but still substantially higher than 2–5, thresholds commonly used in microarray or qPCR studies to assess differential expression. In comparison, the sum of IF and IH reads were 302 and 62907, respectively, and their average RA_IH/IF was 100.5.

Table 8.

A list of HC CIFH contigs with BLAST hits^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	CH/CF	IH/IF
00010	29	464	3	286	782	3.3	46.0	gi|242005387|ref|XP_002423550.1| cAMP-dependent protein kinase catalytic subunit [Pediculus humanus corporis]
00015 etc.	119	5073	20	4227	9439	8.8	102.0	gi|6164595|gb|AAF04457.1|AF078161_1 lacunin [Manduca sexta] (00015, 02717, 15269)
00028	13	958	4	754	1729	15.3	91.0	gi|91081003|ref|XP_975140.1| odd Oz protein [Tribolium castaneum]
00248	7	200	1	155	363	5.9	74.8	gi|157113908|ref|XP_001657920.1| n-acetyllactosaminide β-1,3-NAG transferase [Aedes aegypti]
00379	10	308	1	184	503	6.4	88.8	gi|170037242|ref|XP_001846468.1| Leu-rich repeat-containing protein 1 [Culex quinquefasciatus]
00541	14	567	7	760	1348	8.4	52.4	gi|170029717|ref|XP_001842738.1| Leu-rich repeat-containing G-protein coupled receptor 4 [Culex quinquefasciatus]
00569	4	182	1	176	363	9.4	85.0	gi|283135216|ref|NP_001164363.1| homeobox protein prospero [Nasonia vitripennis]
00623	12	527	1	443	983	9.1	213.8	gi|157132531|ref|XP_001656056.1| odd Oz protein [Aedes aegypti]
00752	0	38	1	164	203	7.9	79.2	gi|194859640|ref|XP_001969420.1| GG23966 [Drosophila erecta]
00802	3	203	3	253	462	14.0	40.7	gi|260840271|ref|XP_002613791.1| hypothetical BRAFLDRAFT_85332 [Branchiostoma floridae]
00839	3	340	1	226	570	23.5	109.1	gi|242021897|ref|XP_002431379.1| conserved hypothetical protein [Pediculus humanus corporis]
00882	7	268	0	261	536	7.9	126.0	gi|112983326|ref|NP_001037620.1| ras-related GTP-binding protein Rab3 [Bombyx mori]
01064	5	134	1	116	256	5.6	56.0	gi|48095930|ref|XP_394560.1| Jagged-1 precursor (Jagged1, hJ1, CD339 antigen) [Apis mellifera]
01429 &	27	924	4	827	1782	7.1	99.8	gi|157134123|ref|XP_001663157.1| atlastin [Aedes aegypti] (01429, 03654)
01609	1	71	1	144	217	14.7	69.5	gi|134001247|gb|ABO45233.1| reverse transcriptase [Ostrinia nubilalis]
02159	3	101	1	144	249	7.0	69.5	gi|114052056|ref|NP_001040346.1| septin [Bombyx mori]
02473	10	255	2	382	649	5.3	92.2	gi|281362668|ref|NP_651533.2| eater [Drosophila melanogaster]
02852	23	1128	7	885	2043	10.2	61.0	gi|66391199|ref|YP_239364.1| hypothetical protein [Microplitis demolitorbracovirus]
03225	1	25	1	143	170	5.2	69.0	gi|195445668|ref|XP_002070431.1| GK11035 [Drosophila willistoni]
03246 &	4	182	2	245	433	9.4	59.1	gi|83583697|gb|ABC24708.1| G protein-coupled receptor [Spodoptera frugiperda] (03246, 06319)
03287	7	493	0	237	737	14.6	114.4	gi|114052174|ref|NP_001040228.1| aminoacylase [Bombyx mori]
04085	0	34	3	268	305	7.0	43.1	gi|206725499|ref|NP_001128673.1| cathepsin L like protein [Bombyx mori]
04278	3	141	1	154	299	9.7	74.3	gi|270001550|gb|EEZ97997.1| hypothetical TcasGA2_TC000395 [Tribolium castaneum]
04746 etc.	0	0	16	1939	1955	0.0	58.5	gi|195486646|ref|XP_002091593.1| GE13745 [Drosophila yakuba] (04746, 13353, 14100)
05560	24	965	4	440	1433	8.3	53.1	gi|254746344|emb|CAX16637.1| putative C1A Cys protease precursor [Manduca sexta]
05577	4	157	22	1895	2078	8.1	41.6	gi|254746342|emb|CAX16636.1| putative C1A Cys protease precursor [Manduca sexta]
05933 etc.	39	1862	8	1395	3304	9.9	84.2	gi|82880638|gb|ABB92836.1| SRC-like protein [Spodoptera frugiperda] (05933, 08686, 13271, 15116, 15350, 15564)
06497 etc.	237	11297	15	4531	16080	9.9	145.8	gi|217262|dbj|BAA03124.1| lectin [Bombyx mori] (06497, 15047, 15764, 15986, 16677, 16801, 16877, 16886, 17700)
07139	21	767	2	262	1052	7.6	63.2	gi|110649216|emb|CAL25117.1| dVA-AP3 [Manduca sexta]
07199	2	73	1	102	178	7.6	49.2	gi|110649250|emb|CAL25134.1| immulectin III [Manduca sexta]
07480	3	248	2	193	446	17.1	46.6	gi|91086517|ref|XP_971701.1| NtR CG6698-PA [Tribolium castaneum]
07642 etc.	17	1246	3	562	1828	15.2	90.4	gi|55139125|gb|AAV41236.1| immulectin-3 [Manduca sexta] (07642, 13452, 14991)
07883	0	0	3	792	795	0.0	127.4	gi|157128533|ref|XP_001661472.1| hypothetical protein AaeL_AAEL011180 [Aedes aegypti]
08524 etc.	74	3481	7	1984	5546	9.8	136.8	gi|2738863|gb|AAB94557.1| hemocyte protease-1 [Manduca sexta] (08524, 12527, 16288, 16719, 17102)
10124 etc.	162	6970	18	4204	11354	8.9	112.7	gi|114050871|ref|NP_001040411.1| carboxylesterase [Bombyx mori] (10124, 15112, 16627, 16922, 17330, 18860)
11280 etc.	143	7866	11	2589	10609	11.4	113.6	gi|91090548|ref|XP_971239.1| hemolectin CG7002-PA [Tribolium castaneum] (11280, 15506, 15594, 18551)
13813	31	2398	4	848	3281	16.0	102.3	gi|110758905|ref|XP_395067.3| ~ hemolectin CG7002-PA [Apis mellifera]
13842	14	677	2	228	921	10.0	55.0	gi|138601|sp|P19616.1|VITM_MANSE microvitellogenin precursor
14248 etc.	1	150	15	2329	2495	31.1	74.9	gi|2149091|gb|AAB58491.1| serpin-2 [Manduca sexta] (14248, 15111, 16917, 17058, 17751)
14570 etc.	562	29402	26	11144	41134	10.8	206.9	gi|162462371|ref|NP_001104817.1| lectin [B. mori] (14570, 15250, 15380, 15792, 16278, 16289, 16291, 16594, 16842, 17159, 17421, 17471, 17732, 17769, 18032, 18067, 18073, 18089, 18097, 18286, 18326, 18719, 18721, 18794)
14760 &	57	3372	3	1184	4616	12.3	190.5	gi|156545430|ref|XP_001606650.1| ~ CG7002-PA [Nasonia vitripennis] (14760, 18045)
14811	5	136	1	121	263	5.6	58.4	gi|221055473|ref|XP_002258875.1| hypothetical protein, conserved in Plasmodium [Plasmodium knowlesi]
15584	3	241	1	202	447	16.7	97.5	gi|66535330|ref|XP_623280.1| atlastin CG6668-PA, isoformA [Apis mellifera]
16815 etc.	208	9161	39	6243	15651	9.1	77.3	gi|75038472|sp|Q25519.3|PRP2_MANSE proPO-2 (16815, 17417, 17958, 18004, 18516, 18811)
17085 etc.	261	12058	33	8286	20638	9.6	121.2	gi|74763772|sp|O44249.3|PRP1_MANSE proPO-1 (17085, 17315, 17420, 17612, 17629, 17562, 18065, 18463, 18887)

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RA_IH/IF >40, RA_CH/CF >40, ARN_IH >80 when RN_IF =0, or ARN_CH >80 when RN_CF =0. RA_IH/IF and RA_CH/CF values are shown in red if they are greater than 40, whereas ARN_IH and ARN_CH values are shown in blue if they are higher than 80. In the columns of RA or ARN, cells shaded green and orange represent down- and up-regulated gene expression, respectively. Contigs with identical BLAST results are combined, with their average RAs or ARNs calculated based on the sums of original reads in CF, CH, IF, and IH for each group. Contigs with no BLAST hit can be found in Table S3, a complete list of 161 HC CIFH contigs.

The hemocyte-specific gene expression is, in several cases, supported by previous studies on M. sexta defense proteins such as lacunin (Nardi et al., 1999), HP1 (Jiang et al., 1999), serpin-2 (Gan et al., 2001), and proPO (Jiang et al., 1997). Lacunin is an extracellular matrix protein responsible for transforming circulating non-adhesive hemocytes to adhesive ones that aggregate on foreign surfaces (Nardi et al., 2005). Contigs 16288, 16719 and 17102 encodes clip-domain HP1; contigs 08524 and 12527 encode an HP1 homolog ~97% identical in sequence to the published one (Jiang et al., 1999). HP1 may be involved in a serine proteinase cascade that proteolytically activates proPO in plasma. Hemolymph proPO is synthesized in oenocytoids only (Jiang et al., 1997): 6 contigs encode proPO subunit-1 and 9 encode proPO subunit-2.

Based on sequence homology, we also discovered 51 contigs that were not known to be related to hemocyte-mediated immunity in M. sexta (Table 8). Contigs 11280, 13813, 15506, 15594, and 18551 probably encode parts of hemolectin or hemocytin, a >300 kDa protein participating in hemolymph coagulation (Lesch et al., 2007; Kanost and Nardi, 2010). As many as 37 contigs encode multiple lectins that bind to carbohydrates. Contigs 05933, 08686, 13271, 15116, 15350, and 15564 encode scavenger receptor C-like proteins that could also recognize carbohydrates. Apparently, hemocytes play critical roles in the recognition of pathogens that are covered with polysaccharides on the surface. Contig 02473 encodes a protein homologous to Drosophila eater that mediates bacteria phagocytosis by hemocytes (Kocks et al., 2005). Contigs 03287 and 07139 may be related to antiviral and antiparasitoid responses, respectively (Abdel-latief and Hilker, 2008; Liu et al., 2010).

Inside hemocytes, proteins may relay signals in a cell-specific manner. These include contigs 00541, 00752, 03246, 06319 (G-protein coupled receptors), 00882 (GTP-binding protein) 00010 (cAMP-dependent kinase), 00839 (receptor-type Tyr-protein phosphatase), 02159 (septin for ubiquitination), 15584 (GTPase atlastin), 14248, 15111, 16917, 17058, and 17751 (serpin-2 and 2′). It is unclear how these two highly inducible, intracellular serpins may inhibit a proteinase during apoptosis (Bird, 1998). Nor is it known how the other proteins may transduce signals dependent on the immune status of hemocytes.

3.5. Specific gene expression in fat body from feeding larvae

Because hemocyte samples collected through cut prolegs of feeding larvae were unlikely contaminated with fat body tissue, the 132 fat body-specific (i.e. FB) contig groups had high RA_{CF/CH or IF/IH} values (Table 9). Moreover, since chances for such contamination were equal for hemocytes from naïve and challenged M. sexta larvae, there was no globally uneven distribution of RAs or ARNs between the CF/CH and IF/IH groups. In other words, the data on fat body-specific gene expression were unbiased and reliable.

Table 9.

A list of FB CIFH contigs with BLAST hits^*

CIFH contig #	Original read #					RA or ARN		BLAST results
	CF	CH	IF	IH	Total	CF/CH	IF/IH
00051	291	1	329	0	621	1403.9	681.6	gi|183979376|dbj|BAG30740.1| muscle myosin heavy chain [Papilio xuthus]
00153 etc.	2069	4	2563	1	4637	2495.3	5309.8	gi|2498144|sp|Q25490.1 apoLp (00153 02405 02406 03748 04510 06831 06834 07770 14087 14589)
00194	37	0	81	1	119	178.5	167.8	gi|48476133|gb|AAT44358.1| calcium-activated potassium channel α subunit [Manduca sexta]
00285 &	298	23	921	5	1247	62.5	381.6	gi|73921301|gb|AAG42021.2|AF327882_1 JHE precursor [Manduca sexta] (00285, 00859)
00409	168	0	216	0	384	810.5	447.5	gi|110750043|ref|XP_394261.3| plexin A CG11081-PA, isoform A [Apis mellifera]
00414	58	1	50	0	109	279.8	103.6	gi|195382713|ref|XP_002050074.1| GJ21937 [Drosophila virilis]
00423	149	0	220	0	369	718.8	455.8	gi|158295580|ref|XP_316291.4| AGAP006225-PA [Anopheles gambiae]
00465	134	1	230	0	365	646.4	476.5	gi|149755131|ref|XP_001491560.1| hemicentin 1 [Equus caballus]
00535	67	1	100	0	168	323.2	207.2	gi|242015135|ref|XP_002428229.1| thrombospondin-3 precursor [Pediculus humanus corporis]
00575	3	0	259	5	267	14.5	107.3	gi|154240658|dbj|BAF74637.1| peptidoglycan recognition protein-D [Samia cynthiaricini]
00609	324	0	762	0	1086	1563.1	1578.6	gi|225542786|gb|ACN91276.1| dentin sialophosphoprotein precursor [Bos taurus]
00737	2	4	131	2	139	2.4	135.7	gi|198466442|ref|XP_002135189.1| GA23919 [Drosophila pseudoobscura]
00748	131	4	118	2	255	158.0	122.2	gi|29346557|ref|NP_810060.1| glycine dehydrogenase [Bacteroides thetaiotaomicron]
00766	45	0	74	1	120	217.1	153.3	gi|158293377|ref|XP_314728.3| AGAP008632-PA [Anopheles gambiae]
00773	49	12	93	1	155	19.7	192.7	gi|157103945|ref|XP_001648193.1| dihydropyrimidine dehydrogenase [Aedes aegypti]
00785	120	2	139	2	263	289.5	144.0	gi|193795848|gb|ACF21977.1| paramyosin [Bombyx mandarina]
00884	39	1	23	0	63	188.1	47.6	gi|156553304|ref|XP_001599652.1| GA21752-PA [Nasonia vitripennis]
00960	52	2	99	1	154	125.4	205.1	gi|157107996|ref|XP_001650030.1| sarcosine dehydrogenase [Aedes aegypti]
01095	64	0	99	1	164	308.8	205.1	gi|169639235|gb|ACA60733.1| venom acid phosphatase [Pteromalus puparum]
01097	134	2	436	5	577	323.2	180.7	gi|55139125|gb|AAV41236.1| immulectin-3 [Manduca sexta]
01127	41	1	52	1	95	197.8	107.7	gi|189491898|gb|ACE00761.1| adipokinetic hormone receptor [Manduca sexta]
01454	599	3	1337	3	1942	963.2	923.3	gi|91082539|ref|XP_973726.1| inter-α (globulin) inhibitor H4 (Kallikrein-sensitive) [T. castaneum]
01480	211	0	729	0	940	1017.9	1510.3	gi|183979392|dbj|BAG30748.1| hypothetical protein [Papilio xuthus]
01601	60	1	79	0	140	289.5	163.7	gi|270005801|gb|EFA02249.1| hypothetical protein TcasGA2 TC007912 [Tribolium castaneum]
01742	65	0	75	0	140	313.6	155.4	gi|283100192|gb|ADB08386.1| sugar transporter 4 [Bombyx mori]
01743	27	0	112	0	139	130.3	232.0	gi|134252572|gb|ABO65045.1| β-hexosaminidase [Ostrinia furnacalis]
01870	184	0	323	0	507	887.7	669.2	gi|242010783|ref|XP_002426138.1| conserved hypothetical protein [Pediculus humanus corporis]
01892	82	0	108	0	190	395.6	223.7	gi|158289807|ref|XP_311448.4| AGAP010734-PA [Anopheles gambiae]
01915	85	2	275	0	362	205.0	569.7	gi|110757936|ref|XP_623940.2| peroxidase precursor (DmPO) [Apis mellifera]
01956	127	0	99	0	226	612.7	205.1	gi|156551746|ref|XP_001602035.1| ENSANGP00000015052 [Nasonia vitripennis]
01972 etc.	383	0	3327	0	3710	1847.7	6892.5	gi|136206|sp|P22297.1|transferrin (01972 10382 11027 14937 17193 17206 17395 16606 18234 18308)
02101	51	0	75	0	126	246.0	155.4	gi|186909546|gb|ACC94296.1| glucose oxidase-like enzyme [Helicoverpa armigera]
02104	59	1	67	1	128	284.6	138.8	gi|91079628|ref|XP_967731.1| AGAP002355-PA [Tribolium castaneum]
02137	101	0	24	0	125	487.2	49.7	gi|91084191|ref|XP_967340.1| AGAP002557-PA [Tribolium castaneum]
02144	82	0	132	3	217	395.6	91.2	gi|62002223|gb|AAX58711.1| pheromone-degrading enzyme 1 [Antheraea polyphemus]
02166	60	0	57	0	117	289.5	118.1	gi|193876254|gb|ACF24761.1| lipid storage droplet protein 1 [Manduca sexta]
02184	53	2	111	1	167	127.8	230.0	gi|226342886|ref|NP_001139705.1| serpin 13 [Bombyx mori]
02219	454	3	971	3	1431	730.1	670.5	gi|219815604|gb|ACL36977.1| putative ecdysone oxidase [Helicoverpa zea]
02329	143	0	411	0	554	689.9	851.5	gi|112984054|ref|NP_001037422.1| yellow1 [Bombyx mori]
02337 &	107	2	170	7	286	258.1	50.3	gi|91079867|ref|XP_967070.1| AGAP005945-PB [Tribolium castaneum] (02337, 15796)
02361	7	4	70	1	82	8.4	145.0	gi|56418425|gb|AAV91020.1| hemolymph proteinase 22 [Manduca sexta]
02393	45	0	77	5	127	217.1	31.9	gi|156545523|ref|XP_001607196.1| dihydroxyacetone kinase-2 homolog (yeast) [Nasonia vitripennis]
02394	28	1	23	0	52	135.1	47.6	gi|91077746|ref|XP_966706.1| conserved hypothetical protein [Tribolium castaneum]
02409	113	0	187	0	300	545.1	387.4	gi|109502352|gb|ABE01157.2| carboxylesterase [Spodoptera litura]
02482	63	0	85	1	149	303.9	176.1	gi|66519258|ref|XP_625210.1| ~ CG6188-PA [Apis mellifera]
02609	97	0	146	2	245	468.0	151.2	gi|156968285|gb|ABU98614.1| α-amylase [Helicoverpa armigera]
02638 &	241	0	206	0	447	1162.6	426.8	gi|41016826|sp|Q27772.3|C1TC_SPOFR C-1-THF synthase, cytoplasmic (02638, 07658)
02651	24	0	124	0	148	115.8	256.9	gi|5326830|gb|AAD42058.1|AF122899_1 plasmatocyte-spreading peptide precursor [Manduca sexta]
02800	28	0	97	0	125	135.1	201.0	gi|260765449|gb|ACX49762.1| β-fructofuranosidase 1 [Manduca sexta]
02847	33	0	103	0	136	159.2	213.4	gi|114051702|ref|NP_001040423.1| zinc-containing alcohol dehydrogenase [Bombyx mori]
02931 &	187	0	429	0	616	902.1	888.8	gi|1658003|gb|AAB18243.1| microsomal epoxide hydrolase [Trichoplusia ni] (02931, 04388)
02947	518	21	981	56	1576	119.0	36.3	gi|259493819|gb|ACW82749.1| hemocyte aggregation inhibitor protein precursor [Manduca sexta]
02979	49	0	92	4	145	236.4	47.6	gi|52782757|sp|Q9NJ98.1|BGRP1_MANSE β-1,3-glucan recognitionprotein 1 βGRP-1
02985	3	0	158	0	161	14.5	327.3	gi|56418466|gb|AAV91027.1| serine proteinase-like protein 4 [Manduca sexta]
03185	106	0	234	10	350	511.4	48.5	gi|157117489|ref|XP_001658792.1| 3-hydroxyacyl-CoA dehyrogenase [Aedes aegypti]
03224	98	0	477	0	575	472.8	988.2	gi|226342906|ref|NP_001139715.1| serpin 22 [Bombyx mori]
03226	222	0	663	0	885	1071.0	1373.5	gi|153791757|ref|NP_001093275.1| myo-inositol oxygenase [Bombyx mori]
03395	22	1	24	0	47	106.1	49.7	gi|157908523|dbj|BAF81491.1| juvenile hormone epoxide hydrolase [Bombyx mori]
03415	190	0	216	1	407	916.6	447.5	gi|2708688|gb|AAB92583.1| acyl-CoA 9 desaturase [Trichoplusia ni]
03434	1	0	387	0	388	4.8	801.7	gi|189234566|ref|XP_001815977.1| Kaz1-ORFB CG1220-PE [Tribolium castaneum]
03454	28	0	102	0	130	135.1	211.3	gi|6560669|gb|AAF16712.1|AF117590_1 unknown [Manduca sexta]
03483	280	0	374	0	654	1350.8	774.8	gi|283558277|gb|ADB27116.1| aliphatic nitrilase [Bombyx mori]
03712	49	2	157	5	213	118.2	65.1	gi|170779021|gb|ACB36909.1| glutathione S-transferase θ [Antheraea pernyi]
03737	167	1	197	0	365	805.6	408.1	gi|56462300|gb|AAV91433.1| putative serine protease-like protein 2 [Lonomia obliqua]
03776 etc.	204	8	960	51	1223	123.0	39.0	gi|112983872|ref|NP_001036857.1| serpin-like protein [Bombyx mori] (03776, 06215, 06531, 17814)
04012 &	167	3	727	11	908	268.5	136.9	gi|27733411|gb|AAO21503.1|AF413062_1 leureptin, LPS binding [Manduca sexta] (04012, 08453)
04413	69	1	133	1	204	332.9	275.5	gi|194743582|ref|XP_001954279.1| GF18195 [Drosophila ananassae]
04424	72	0	64	0	136	347.3	132.6	gi|114052020|ref|NP_001040445.1| tropomyosin 1 [Bombyx mori]
04430	74	0	68	0	142	357.0	140.9	gi|114052573|ref|NP_001040481.1| phosphoribosyl pyrophosphate synthetase [Bombyx mori]
04498	46	0	115	0	161	221.9	238.2	gi|90025232|gb|ABD85119.1| JH epoxide hydrolase [Spodopteraexigua]
04504	53	0	135	0	188	255.7	279.7	gi|7239259|gb|AAF43151.1|AF226857_1 hemolymph JHBP precursor [Manduca sexta]
04722 &	578	0	861	0	1439	2788.4	1783.7	gi|116791778|gb|ABK26104.1| unknown [Picea sitchensis] (04722, 04994)
04781	56	0	237	0	293	270.2	491.0	gi|118359591|ref|XP_001013035.1| PHD-finger family protein [Tetrahymenathermophila]
04786	61	0	62	0	123	294.3	128.4	gi|219686082|emb|CAW30924.1| putative aldo-ketose reductase 1 [Papilio dardanus]
04791	144	0	200	0	344	694.7	414.3	gi|116788175|gb|ABK24783.1| unknown [Picea sitchensis]
04806	518	1	372	0	891	2499.0	770.7	gi|157122933|ref|XP_001659963.1| actin [Aedes aegypti]
04808	0	0	426	2	428	0.0	441.3	gi|237861314|gb|AAV41237.2| immulectin-4 [Manduca sexta]
04830 etc.	59	2	755	6	822	142.3	260.7	gi|169646838|ref|NP_001112375.1| heat shock protein 25.4 [Bombyx mori] (04830, 04887, 05717)
05038 &	101	0	175	1	277	487.2	362.5	gi|110759694|ref|XP_394781.3| rTS β protein [Apis mellifera] (05038, 05832)
05136	1074	11	1041	37	2163	471.0	58.3	gi|114051966|ref|NP_001040198.1| mitochondrial aldehyde dehydrogenase [Bombyx mori]
05324	68	0	88	0	156	328.0	182.3	gi|225346695|gb|ACN86370.1| troponin I transcript variant C [Bombyx mandarina]
05348	50	0	67	0	117	241.2	138.8	gi|189234391|ref|XP_974849.2| GA16498-PA [Tribolium castaneum]
05417 etc.	273	0	917	0	1190	1317.0	1899.7	gi|260907784|gb|ACX53694.1| alcohol DH [Heliothis virescens] (05417, 05461, 07389, 07432)
05984	89	0	97	0	186	429.4	201.0	gi|56462260|gb|AAV91413.1| myosin 3 light chain [Lonomia obliqua]
06175	11	0	52	1	64	53.1	107.7	gi|170070451|ref|XP_001869584.1| conserved hypothetical protein [Culex quinquefasciatus]
06227	251	1	715	0	967	1210.9	1481.3	gi|124527|sp|Q00630.1|ICYB_MANSE insecticyanin-B, blue biliprotein
06251	66	2	57	7	132	159.2	16.9	gi|158289206|ref|XP_310956.4| AGAP000179-PA [Anopheles gambiae]
06394	51	0	228	0	279	246.0	472.3	gi|110611262|gb|ABG77980.1| alanine-glyoxylate transaminase 1 [Glossinamorsitans morsitans]
06588	60	0	75	0	135	289.5	155.4	gi|56462256|gb|AAV91411.1| myosin 1 light chain [Lonomia obliqua]
06597	60	0	200	0	260	289.5	414.3	gi|56462320|gb|AAV91443.1| secreted peptide 30 [Lonomia obliqua]
06732	115	1	244	0	360	554.8	505.5	gi|25090512|sp|Q25513.1|HGLY_MANSE 27 kDa hemolymph glycoprotein
06789 &	159	0	460	0	619	767.1	953.0	gi|156968291|gb|ABU98617.1| unknown [Helicoverpa armigera] (06789, 06876)
06975 &	106	2	212	2	322	255.7	219.6	gi|189237651|ref|XP_001813448.1| N-acetyl neuraminatelyase [Tribolium castaneum] (06975, 14637)
07116 &	1	4	1015	4	1024	1.2	525.7	gi|171262319|gb|ACB45566.1| lebocin-like protein [Antheraea pernyi] (07116, 10853)
07565	24	1	14	0	39	115.8	29.0	gi|7862150|gb|AAF70499.1|AF255341_1 3-dehydroecdysone 3α-reductase [Spodoptera littoralis]
07608 etc.	353	3	3931	0	4287	567.7	8143.9	gi|159526|gb|AAA29320.1| Met-rich storage protein 1 (07608, 07975, 08141, 14688)
07629	65	0	82	0	147	313.6	169.9	gi|77415676|emb|CAJ01507.1| hypothetical protein [Manduca sexta]
07639 &	811	0	1616	18	2445	3912.5	186.0	gi|134436|sp|P14754.1| alaserpin or serpin-1 (07639, 15891)
07671	227	3	450	3	683	365.0	310.8	gi|195164814|ref|XP_002023241.1| GL21066 [Drosophila persimilis]
08076 &	47	3	115	2	167	75.6	119.1	gi|226342878|ref|NP_001139701.1| serpin 7 [Bombyx mori] (08076, 14528)
08224 etc.	7528	8	10093	0	17629	4539.6	20909.2	gi|1168527|sp|P14297.2| arylphorin β subunit (08224, 16474, 16501, 16664, 16715, 16764, 18695)
08467	0	0	113	0	113	0.0	234.1	gi|112983866|ref|NP_001036858.1| T7 lysozyme-like protein 1 (BTL-LP1) [Bombyx mori]
08500	138	0	407	0	545	665.7	843.2	gi|156406857|ref|XP_001641261.1| predicted protein [Nematostella vectensis]
08821	246	0	436	2	684	1186.8	451.6	gi|112983550|ref|NP_001036879.1| fibrillin-like protein [Bombyx mori]
08845	27	0	130	0	157	130.3	269.3	gi|195029763|ref|XP_001987741.1| GH19797 [Drosophila grimshawi]
08854 &	302	5	5234	0	5541	291.4	10843.3	gi|5869985|emb|CAB55603.1| moderately Met-rich storage protein [Spodoptera litura] (08854, 15324)
09928	30	0	106	0	136	144.7	219.6	gi|242090851|ref|XP_002441258.1| hypothetical SORBIDRAFT_09g023310 [Sorghum bicolor]
10071 &	15	1	1243	0	1259	72.4	2575.1	gi|228382|prf||1803340A Met-rich storage protein SP1A (10071, 17516)
10326	284	4	299	11	598	342.5	56.3	gi|56462160|gb|AAV91363.1| hypothetical protein 10 [Lonomia obliqua]
10791 etc.	2	0	2756	2	2760	9.6	2854.8	gi|4262357|gb|AAD14591.1| scolexin A [Manduca sexta] (10791, 10792, 16520, 18669, 18670, 18963)
11039 etc.	13962	11	19836	0	33809	6123.3	41094.2	gi|114240|sp|P14296.1| arylphorin α subunit (11039 16171 16537 16814 17492 18240 18257 18556)
11830	26	1	33	0	60	125.4	68.4	gi|260780799|ref|XP_002585527.1| hypothetical protein BRAFLDRAFT_89257 [B. floridae]
11922 &	901	12	1052	3	1968	362.2	726.5	gi|114058|sp|P13276.1| apoLp-III (11922, 13093)
12005	154	0	2177	0	2331	742.9	4510.1	gi|2625150|gb|AAB86646.1| moderately Met-rich hexamerin precursor [Hyalophora cecropia]
12151	0	0	153	0	153	0.0	317.0	gi|116084|sp|P14665.1|bactericidin B-5P, cecropin-like peptide precursor
12749	135	0	1462	0	1597	651.3	3028.8	gi|159530|gb|AAA29322.1| Met-rich storage protein 3 [Manduca sexta]
13563	0	0	657	0	657	0.0	1361.1	gi|110347786|gb|ABG72695.1| attacin-like protein [Antheraea mylitta]
13916 etc.	3	0	1327	0	1330	14.5	2749.1	gi|219958086|gb|ACL68097.1| lebocin-related protein precursor [M. sexta] (13916, 17301, 17434)
13994	57	0	62	0	119	275.0	128.4	gi|112983654|ref|NP_001036872.1| bombyrin [Bombyx mori]
14173	45	0	32	0	77	217.1	66.3	gi|153792114|ref|NP_001093267.1| phosphatidylethanolamine binding protein [Bombyx mori]
14375 etc.	400	0	681	0	1081	1929.7	1410.8	gi|400673|sp|P31420|OMBP ommochrome-binding protein precursor (14375, 14659, 17494, 17813)
14380 etc.	0	1	408	3	412	0.0	281.8	gi|67906420|gb|AAY82587.1| attacin-1 [Manduca sexta] (14380, 14641, 16150)
14700 &	0	0	301	4	305	0.0	155.9	gi|260765453|gb|ACX49764.1| peptidoglycan recognition protein 2 [Manduca sexta] (14700, 14752)
15089	271	0	194	0	465	1307.4	401.9	gi|158293921|ref|XP_315269.4| AGAP011516-PA [Anopheles gambiae]
15639	10	0	109	0	119	48.2	225.8	gi|148298818|ref|NP_001091784.1| multi-binding protein [Bombyx mori]
16000	61	0	138	0	199	294.3	285.9	gi|109458629|ref|XP_001073545.1| hypothetical protein [Rattus norvegicus]
16223	22	1	47	2	72	106.1	48.7	gi|242003442|ref|XP_002422733.1| bifunctional purine biosynthesis protein [Pediculus corporis]
16281 &	358	0	541	0	899	1727.1	1120.8	gi|134103857|gb|ABO60878.1| cationic peptide CP8 precursor [Manduca sexta] (16281, 17312)
16849	134	0	541	0	675	646.4	1120.8	gi|114051738|ref|NP_001040426.1| alcohol dehydrogenase [Bombyx mori]
17199	42	2	33	4	81	101.3	17.1	gi|3108073|gb|AAC15763.1| putative multifunctional protein ADE2 [Manduca sexta]
17350	0	0	205	0	205	0.0	424.7	gi|29469969|gb|AAO74640.1| antimicrobial protein attacin 2 [Manduca sexta]
18797	9	0	549	0	558	43.4	1137.4	gi|39843367|gb|AAR32136.1| VHDL receptor [Helicoverpa zea]

^*RA and ARN are calculated using original read numbers as described in Section 2.3. Listed here are contigs with RA_IF/IH >100, RA_CF/CH >100, ARN_IF >200 when RN_IH =0, or ARN_CF >200 when RN_CH =0. RA_IF/IH and RA_CF/CH values are shown in red if they are greater than 100, whereas ARN_IF and ARN_CF values are shown in blue if they are higher than 200. In the columns of RA or ARN, cells shaded green and orange represent down- and up-regulated gene expression, respectively. Contigs with identical BLAST results are combined, with their average RAs or ARNs calculated based on the sums of original reads in CF, CH, IF, and IH for each group. Contigs with no BLAST hit can be found in Table S4, a complete list of 250 FB CIFH contigs.

Insect fat body, analogous to combined mammalian liver and adipose tissue, is the site where most intermediary metabolism takes place (Arrese and Soulages, 2010). It also is the principal source of plasma proteins, including those participating in innate immune responses (Jiang, 2008; Ragan et al., 2009). These notions are strongly supported by the identification of FB contigs and BLAST search: 61 or 46% of the 132 FB contig groups are metabolism-related, whereas 32 or 24% are immunity-related (Table 9). Since metabolism-related genes and their transcript level changes after the immune challenge will be reported elsewhere, we only discuss fat body-specific gene expression involved in antimicrobial defense responses and the UP contigs covered in Section 3.2 are not repeated here. β-1,3-glucan recognition protein-1 (02979) (Ma and Kanost, 2000), immulectin-3 (01097) (Yu et al., 2005), and leureptin (04012 and 08453) (Zhu et al., 2010) are pattern recognition receptors binds fungi and bacteria (Table 9). HAIP (02947), a chitinase-like protein, inhibits hemocyte aggregation (Kanost et al., 1994). Contig 05348 encodes a protein with at least three Ig domains. Contig 00535 encodes a thrombospodin-like protein with eight EGF-like domains and one coiled coil for protein-protein interaction. Contig 07671, after extension, is found to encode a >60 kDa protein with at least four EGF domains. Hemicentin (00465) is a cell adhesion protein containing a von Willebrand A domain (Vogel and Hedgecock, 2001). Contig 08821 encodes a fibrillin-like nimrod B which may play a role in pathogen recognition and phagocytosis (Kurucz et al., 2007).

We have found six proteinase inhibitor-like proteins, including homologs of B. mori serpin12 (or SLP: 03776, 06215, 06531, 17814), serpin13 (02184) and serpin22 (03224) (Zou et al., 2009), two Cys-rich secreted protein (06175, 06597), and cationic protein-8 (16281, 17312) (Ling et al., 2009). Contig 02651 encodes three cytokines that may regulate cellular immune responses (Kanamori et al., 2010).

4. Discussion

Next-generation sequencing has been increasingly used for profiling gene expression in the past few years (Costa et al., 2010; Marguerat and Bähler, 2010). While its advantages over microarray analysis are obvious in some aspects, this technique has been so far, to the best of our knowledge, only applied to species with known genome sequences for transcript profiling. (It has also been used in other species for general transcriptome analysis but not for systematically studying mRNA level changes.) In this study, we have extended massive sequencing and data analysis to a new dimension in which gene discovery, expression profiling, and function prediction are done at the same time in a lepidopteran insect lacking known genome sequence. This technical improvement, mathematically simple, generated a wealth of information in a cost-effective manner and opens a door for similar studies in non-model organisms of practical importance.

Using numbers of reads assembled into specific contigs to calculate RAs provides a genome- independent way of looking at gene expression in relation to specific physiological processes and tissue/cell types. This perspective becomes more relevant when homology-based search results are also taken into consideration, even though expression pattern by itself is an autonomous parameter for gene discovery (Table 6). Systematic examination of transcript level changes in conjunction with sequence comparison can be extremely powerful in terms of gene discovery and functional prediction. It is our general impression that details of microarray data were, in many cases, overlooked when they were merely considered as spots rather than specific genes associating with functions. This tends to be true, especially when expression of genes does not strongly correlate with a treatment due to biological or technical reasons. For example, real expression changes of certain genes are sometimes too small to be distinguished from background caused by nonspecific binding in hybridization-based methods. In contrast, since read numbers for individual contigs are “digital”, their corresponding RAs or ARNs have a greater dynamic range in measurement of relative transcript abundance and do not carry noise or artifact and can, thus, be calculated and compared with high confidence. In this study, we set cutoff RA or ARN values more or less empirically based on preliminary tests with different thresholds, presence or absence of immunity-related hits in the results, and number of contigs appropriate for manual checking and tabulation. If necessary, however, we can conveniently use lower RA cutoffs to increase detection sensitivity to discover subtle changes in transcript levels. The depth of our datasets surely allows us to explore in greater details the immunity-relatedness and tissue specificity of gene expression. For instance, some metabolic enzymes do not change much in their transcript levels. By mapping their contigs onto metabolic pathways with close-to-one RA values, we may detect trends from multiple members of specific pathways, which reveal impacts of immune response on general metabolism of the insect.

In essence, we have projected RA values of our dataset (“CIFH”) onto four surfaces of a tetrahedron: two for UP and DN, two for HC and FB (Fig. 2). If each contig is represented by a point, the data we examined are located far away from the axis of immune inducibility (RA_{IF/CF or IH/CH} =1) or tissue specificity (RA_{IF/IH or CF/CH} =1). The remaining >95% of the 19,020 data points are densely packed along these two axes. The large number of such contigs reflects the depth of our sequence data, as also revealed by studying a number of selected contigs. For example, we have found in our dataset four lebocin-related sequences: the first one encodes a precursor protein which is processed by an intracellular processing proteinase into 4 peptides and 2 are antibacterial (Rayaprolu et al., 2010). The other three lebocin precursors are anticipated to be processed in the same manner, based on their sequences and conserved cleavage sites (data not shown). Further evidence in support of sequence depth came from the detection of highly similar contigs assembled from large numbers of reads, such as attacins (data not shown).

Fig 2. Analysis of the CIFH dataset in terms of immune inducibility and tissue specificity.

Examination of the RA values is a process of projecting the CIFH dataset onto four surfaces of a tetrahedron. Two of them represent deviations from 1 (brown line) regarding mRNA level changes before and after the immune challenge: UP (red) for RA_{IF/CF or IH/CH} above a set value of greater than 1; DN (green) for RA_{IF/CF or IH/CH} below a set value of smaller than 1. The other two surfaces represent deviations from 1 (green line) regarding abundance changes in the two tissues: FB (yellow) for RA_{IF/IH or CF/CH} above a set value of greater than 1; HC (blue) for RA_{IF/IH or CF/CH} below a set value of smaller than 1. In this study, we used RA_IF/CF >5 and RA_IH/CH >8 as cutoff values for UP (red line), RA_{CF/IF or CH/IH} >10 for DN (dashed green line), RA_{IH/IF or CH/CF} >40 for HC (blue line), and RA_{IF/IH or CF/CH} >100 for FB (dashed yellow line). The regions above these lines contain 528 UP, 148 DN, 161 HC, and 250 FB contigs (accounting for 2.8%, 0.8%, 0.8%, and 1.3% of the 19,020 contigs, respectively), whereas the regions below represent >94.3% of the total contigs, not analyzed in this study.

Our data also indicate that not all immunity-related genes are up-regulated after the injection of microbes – some of them do not increase much and others may even decrease. For instance, HP6 and HP8 mRNA levels did not significantly change (RAs for CIFH 00540, 05370, and 09086: 1.6, 2.3, and 1.3, respectively) after the challenge but they are involved in spätzle and proPO activation (An et al., 2009 and 2010). In order to identify gene products that may be involved in pathogen recognition or signal transduction but not significantly increase in mRNA levels, we have searched the entire dataset (“06CIFH”) using the B. mori sequences (Tanaka et al., 2008; Zou et al., 2009) in conjunction with M. sexta HP1 through HP24 and PAPs. Of the 534 contigs identified, 368 do not belong to the highly induced (UP: RA_IF/CF >5 or RA_IH/CH >8) or suppressed groups (DN: RA_{CF/IF or CH/IH} >20): 119, 193, 62, and 8 contigs have their RA values falling into moderately induced (RA_IF/CF: 2~5, RA_IH/CH 2~8, or ARN_{IF or IH}: 4~10), no change (RA_{IF/CF or IH/CH}: 2~0.5 or ARN_{IF, IH, CF, or CH} <4), slightly suppressed (RA_{CF/IF or CH/IH}: 2~6 or ARN_{CF, or CH}: 4~12), and moderately suppressed (RA_{CF/IF or CH/IH}: 6~20 or ARN_{CF, or CH}: 12~40) groups, respectively. While most of the highly suppressed group (“DN”) (RA_{CF/IF or CH/IH}: >20 or ARN_{CF, or CH}: >40) turned out to be false positive (Table 7), mRNA levels of 48 and 3 contigs were indeed reduced slightly and moderately (data not shown), respectively.

Beyond method development, this study has established a foundation for genome annotation, especially for genes that are induced or suppressed in response to the immune challenge and for genes preferentially expressed in fat body or hemocytes. As a new version of the transcriptome analysis, it provided two high-quality cDNA datasets (“CIFH” and “06CIFH”) that can be used for identification of plasma proteins from naïve and induced feeding larvae. We detected a lot fewer indels in these contigs than those from our previous work (Zou et al., 2008). The coverage of our datasets, as well as expression profiles from the read comparisons, is anticipated to not only facilitate proteomic analysis but also assist cDNA cloning, recombinant expression, and functional elucidation of immunity-related genes in this biochemical model insect. It is our sincere hope that similar experiments including data processing will be performed in non-model organisms to discover genes of critical importance.

Abbreviations

CF, IF, CH and IH

control and induced fat body or hemocytes

RA

relative abundance

HP

hemolymph proteinase

PO and proPO

phenoloxidase and its precursor

AMP

antimicrobial peptides

EST

expressed sequence tag

rpS/Lx

ribosomal small or large subunit protein X

LNF

library normalization factor

ARN

adjusted read number

UP and DN

up- and down- regulated

FB and HC

fat body- and hemocyte-specific

ORF

open reading frame

PRR

pattern recognition receptors

LPS

lipopolysaccharide

PGRP

peptidoglycan recognition protein

βGRP

β-1,3-glucan recognition protein

CTL

C-type lectin

SPH

serine proteinase homolog

PSP

plasmatocyte-spreading peptide

PAP

proPO-activating proteinase

HAIP

hemocyte aggregation inhibitor protein

EGF

epidermal growth factor

RT-PCR

reverse transcription-polymerase chain reaction