SEARCH

SEARCH BY CITATION

Keywords:

  • Drosophila melanogaster;
  • fat body;
  • transcriptome;
  • self-organizing maps;
  • transcriptome mapping

Abstract

  1. Top of page
  2. Abstract
  3. Research Methods and Procedures
  4. Acknowledgement
  5. References

In Drosophila, the fat body is a collective name for the masses and sheets of adipose tissue that are distributed throughout the fly body. Thus far, >386, 000 Drosophila expressed sequence tags (ESTs) have been deposited to the GenBank database, including 10, 443 derived from fat body in flies (data accessed on October 7, 2004). The objective of this study was to map the transcriptome of the fat body in flies and thus provide genomics and bioinformatics tools for developing a Drosophila model for addressing the genetic complexity of obesity in humans. The gene—EST Basic Local Alignment Search Tool (BLAST) matches revealed that these ESTs could represent 12, 188 coding genes in the Drosophila genome. Among them, at least 2261 are expressed in the fat body, including 41 identified as preferentially expressed genes with logarithm of odds >3.0. Self-organizing map analysis revealed a cluster of 290 genes favorably expressed in the fat body compared with genes expressed in five other tissues. Mapping of the fat body transcriptome identified a 1.7-Mb domain on 3L containing 35 genes that were expressed at a much higher level than in other tissues (transcript density factor = 1.0 ∼ 2.3).

Obesity has become one of the most prevalent diseases of modern societies: 61% of adults in the United States are overweight, and 26% are obese (1). The major consequence of overweight and obesity in humans is that these traits are associated with more than 30 degenerative medical conditions (2). Although >400 genes, markers, and chromosomal regions have been identified as associated or linked with human obesity phenotypes (3), the responsible genes are still unknown in >95% of severe obesity (4). The fruit fly Drosophila melanogaster has been proposed as one of the model organisms for identification and validation of genes involved in the genetic complexity of human obesity, because most genes and functions were conserved in the genomes of these two species throughout evolution (5).

Obesity is a chronic disease caused by an imbalance between energy ingested and expended (2). When energy intake exceeds energy expenditure, the resulting imbalance may expand the size and increase the number of fat cells. In Drosophila, the fat body consists of masses and sheets of adipose tissue that are distributed throughout the fly body. Information on the transcriptome of fat bodies in flies will, therefore, provide an entry point into the identification of key genes for the control of energy metabolism and fat storage and key intracellular signaling pathways or networks that are involved in determining susceptibility to the development of obesity in D. melanogaster. Our objective was, therefore, to conduct an in silico analysis of transcriptomes of the fat body in Drosophila based on expressed sequence tags (ESTs)1 to understand gene functions, expression patterns, and the potential pathways related to fat deposition.

Gene—EST Match Analysis

The current Release 3.2 finished euchromatic sequence of the six D. melanogaster chromosome arms shows that the fly genome contains ∼14, 015 genes, including 12, 486 coding genes (http:www.ncbi.nlm.nih.gov). More than 386, 000 Drosophila ESTs have also been released to the public (Table 1) (6). Gene—EST Basic Local Alignment Search Tool (BLAST) match analysis revealed that these ESTs could represent 12, 188 genes, accounting for 97.61% of a total of 12, 486 coding genes determined in the Drosophila genome. Previously, ESTs were grouped into unigene sequences (7) or clustered into tentative consensus sequences (8), which were used in determination of transcriptional profiling of different tissues such as the “Digital Differential Display” analysis. However, because most genes in the Drosophila genome have been well annotated, we believe that it is time to conduct tissue transcriptome analysis based on genes rather than EST clusters. For example, the current Drosophila gene index at the Institute of Genome Research contains 36, 289 unique sequences, which is apparently far beyond the number of genes.

Table 1. . Summary of D. melanogaster ESTs used in the study
TissueLibrariesEST number
  NCBITIGR
EmbryoSubtotal101989101979
 RE Drosophila melanogaster normalized Embryo pFlc-16248462481
 LD Drosophila melanogaster embryo pOT22181320439
 LD Drosophila melanogaster embryo BlueScript1603417406
 CK Drosophila melanogaster embryo BlueScript16581653
Fat bodySubtotal1044310348
 Exelixis FlyTag ML01 pSport-Tag211044310348
Head-brainSubtotal8446784386
 RH Drosophila melanogaster normalized Head pFlc-15647556472
 GH Drosophila melanogaster head pOT22474124737
 HL Drosophila melanogaster head BlueScript28832785
 HL Drosophila melanogaster head pOT2294390
 Drosophila melanogaster adult brain cDNA7474
Mixed and whole bodySubtotal105745105721
 Exelixis FlyTag CK01 pCDNA-SK+8016180145
 Exelixis FlyTag CK02 pCDNA-SK+94159413
 LP Drosophila melanogaster larval-early pupal pOT21616916163
OvarySubtotal1151411288
 GM Drosophila melanogaster ovary pOT258275777
 GM Drosophila melanogaster ovary BlueScript56875511
Saliva glandSubtotal46884657
 ESG0146884657
TestisSubtotal3257232436
 AT Drosophila melanogaster adult testes pOTB72394523942
 UT Drosophila melanogaster adult testes pOTB713291329
 Drosophila melanogaster adult testis library72987165
Unclassified and othersSubtotal3509331185
 SD Drosophila melanogaster Schneider L2 cell culture pOT22139921395
 Exelixis FlyTag MN08 BlueScript80538043
 Drosophila 8–12 hours embryonic cDNA library12191170
 Drosophila melanogaster cDNA Library13367
 Drosophila melanogaster Uni-ZAP XR library104103
 Drosophila melanogaster brain Canton S adult6060
 All other libraries with <50 EST4125347
Σ 386511382000

Preferentially Expressed Genes in the Fat Body

We identified at least 2261 genes expressed in the fat body (Supplemental Table 1, available online at the Obesity Research web site, www.obesityresearch.org). EST abundance of these genes varied from 0.015% to 5.542% in the fat body. Furthermore, 41 genes were identified as preferentially expressed genes in the tissue with logarithm of odds >3.0 and Fisher's test <0.0001 (Table 2) compared with the genes expressed in embryos, head-brain, ovary, saliva gland, and testes. Fifty-four percent of these 41 genes, such as LSP1α, LSP1γ, LSP2, Fbp1, Fbp2, Sgs3, Sgs4, Sgs5, and Sgs7, have known functions and have been well characterized in the fat body (9, 10). Recently discovered genes, such as TPR domain protein (11), FOXO transcription factor (12), and PI3K signaling pathway (13) also appeared in our gene list, but with a low level of gene expression (Supplemental Table 1, available online at the Obesity Research web site, www.obesityresearch.org). Using DNA microarrays, Klebes et al. (14) revealed ∼1000 genes differentially expressed between fat body and wing with an intensity difference >1.74. These data indicate that the fat body in flies may have its own unique gene expression patterns. Increasing the number of ESTs will contribute to the generation of a comprehensive gene list expressed in the Drosophila fat body. The genes with identified functions actually raise several questions. These are not those traditionally linked with lipogenic pathways. Obviously, almost one-half of the genes remain unidentified. Further work comparing these sequences to known sequences for adipose tissue enzymes will help define this transcriptome in more detail.

Table 2. . Preferentially expressed genes in the fat body of D. melanogaster (logarithm of odds ≥ 3.0)
ChromosomeGeneEST frequency (%)Logarithm of oddsFisher'sDescription
  Fat bodyOther tissues   
3LFbp15.5420.416255.6<0.0001Fat body protein 1
3RSgs53.4000.074255.6<0.0001Salivary gland secretion 5
2LFbp22.1420.078136.2<0.0001Fat body protein 2
2LCG152811.9320.053135.2<0.0001 
XCG22331.6480.08391.9<0.0001 
3LSgs71.2280.04081.3<0.0001Salivary gland secretion 7
3LLsp1γ1.4380.12062.6<0.0001Larval serum protein 1 gamma
3LLsp21.7230.20161.0<0.0001Larval serum protein 2
3LCG115381.7230.20161.0<0.0001 
3LHsp230.8240.05938.8<0.0001Heat shock protein 23
3LCG123100.5240.03227.0<0.0001 
3LEig71Ee0.4640.03023.1<0.0001Ecdysone-induced gene 71Ee
3LSgs30.4490.02922.4<0.0001Salivary gland secretion 3
2LCG67700.6290.07921.2<0.0001 
XSgs40.4340.03021.1<0.0001Salivary gland secretion 4
XLsp1α0.2850.02512.1<0.0001Larval serum protein 1 alpha
3RCG83690.2700.02510.9<0.0001 
3RCREG0.2700.0349.1<0.0001 
2RTsp42Ek0.2250.0248.4<0.0001Tetraspanin 42Ek
2RCG169260.2400.0298.2<0.0001 
3RCG75870.2250.0268.0<0.0001 
2RCG150980.2550.0486.1<0.0001 
3LQm0.4940.1675.9<0.0001 
2LCG167040.1800.0265.4<0.0001 
4RfaBp0.3300.0915.2<0.0001Retinoid- and fatty-acid binding protein
3RLsd-10.1950.0335.2<0.0001Lipid storage droplet-1
3RTotA0.1650.0264.6<0.0001Turandot A
2LCG89970.1650.0274.5<0.0001 
3LLysD0.1500.0234.3<0.0001Lysozyme D
2LCG119120.1800.0354.2<0.0001 
2LRpL37A0.2700.0784.1<0.0001 
XCG24440.1500.0263.9<0.0001 
3RRpL320.3150.1043.9<0.0001Ribosomal protein L32
2Rtsr0.2550.0753.7<0.0001twinstar
2RδTry0.2700.0853.6<0.0001deltaTrypsin
2RγTry0.2700.0853.6<0.0001gammaTrypsin
2RCG300310.2700.0853.6<0.0001 
3LCG321920.1350.0243.4<0.0001 
2RLcp10.1650.0373.3<0.0001Larval cuticle protein 1
4RpS3A0.6590.3433.3<0.0001Ribosomal protein S3A
2RCG129340.1350.0263.2<0.0001 

Obesity research has usually identified alterations of expression in a few key controlling systems, such as food intake (leptin; fa/fa rat), sympathetic nervous activity (ob/ob mouse), or insulin activity (ob/ob mouse, db/db rat), and not as much change in enzymes of nutrient catabolism or anabolism (15). From one point, this makes sense, because it would take a major mutation (deletion/malformation) in several enzymes to alter flux toward lipid, because most enzymes are present in concentrations exceeding substrate concentrations (15, 16). Very early research in the molecular biology of metabolic control has shown through strict kinetic analysis that it is highly unlikely that changes in expression of one or even several enzymes in metabolic pathways would lead to changes in pathway flux, but that control of processes or factors (e.g., hormones) regulate transcription of enzyme genes or regulate enzyme activity (15, 16). Thus, involvement of transport and binding proteins in control of lipid deposition is not surprising at this early stage of research. We would say that obesity is a typical quantitative trait, which is controlled by many genes, including genes involved both in transporting nutrients and in nutrient metabolism.

Self-organizing Maps of Drosophila Transcriptomes

We focused on gene expression patterns in six tissues—embryos, fat body, head-brain, ovary, saliva gland, and testes (Table 1). Self-organizing map (SOM) clustering analysis (http:genome.tugraz.at) revealed a cluster of 290 genes (Figure 1) favorably expressed in fat body, which are marked with asterisks in Supplemental Table 1 (available online at the Obesity Research web site, www.obesityresearch.org). Interestingly, a cluster of eight genes was concertedly expressed in both fat bodies and ovaries, while a cluster of 61 genes was co-expressed in fat bodies and testes, with some genes also expressed in ovaries. A link between obesity and reproductive function has been proposed. Studies have found that women who are obese or extremely thin have higher rates of amenorrhea and infertility. In addition, obese women are more likely to develop hypertension and gestational diabetes during pregnancy, as well as experience delivery complications such as caesarean sections and prolonged delivery times (17). Further characterization of these coordinated genes between fat bodies and reproductive tissues will, therefore, help elucidate mechanisms by which obesity is related to infertility.

image

Figure 1. A cluster of 290 genes favorably expressed in fat body of D. melanogaster based on SOM clustering analysis. The tissues included in the analysis are as follows: I, embryos; II, fat body; III, head-brain; IV, ovary; V, saliva gland; VI, testes.

Download figure to PowerPoint

Mapping the Fat Body Transcriptome

To explore the relationship of genes and gene activity with their genome locations, we mapped the fat body transcriptome along chromosomes based on estimates of transcript density factor (TDF), which was defined as the logarithm of the fractional gene expression in fat body over that in all other tissues as the background (18). The analysis identified a 1.7-Mb chromosomal domain on 3L that contained 35 genes highly expressed in the fat body, with TDF being from 1.0 to 2.3 (Figure 2). Among them, 8 genes (Lsp1γ, CG13905, LysX, LysB, LysC, LysD, LysE, and LysP; Supplemental Table 1, available online at the Obesity Research web site, www.obesityresearch.org) are included in the 290-gene fat body cluster generated by SOM analysis, indicating significant contribution of this chromosome domain to the fat body transcriptome in flies compared with other genome regions (Fisher's exact test, p = 0.041861). Suggestive evidence of chromosomal domains of fat body genes was also observed on 2R (0–0.34 Mb) and 3R (0.7– 7 Mb; Figure 2). However, no chromosomal domain for fat body genes was observed on other chromosomes. These results provide additional evidence that the transcriptional activities of normal tissues could be orchestrated at the chromosomal level and that many highly expressed genes with relevant functionality could share physical proximity (18).

image

Figure 2. Mapping transcriptome of fat body on Drosophila genome. The x-axis represents length of Drosophila chromosomes in base pairs. The y-axis represents TDF.

Download figure to PowerPoint

In summary, Drosophila is arguably the most versatile and one of the most powerful eukaryotic genetic model systems. The fly genome is relatively compact (1.7 × 108 bp) and has been completely sequenced. Therefore, our study on genes, expression patterns, and transcriptome domains in the fat body provides an entry point for developing a Drosophila model to address the complexity of obesity in humans. In particular, the results from mapping the fat body transcriptome would increase our understanding of genome organization of adipose tissue genes and gene activities.

Research Methods and Procedures

  1. Top of page
  2. Abstract
  3. Research Methods and Procedures
  4. Acknowledgement
  5. References

Most of the fly ESTs deposited in the GenBank databases at the National Center for Biotechnology Informatics (NCBI) were generated by the Berkeley Drosophila Genome Project, which is a consortium of the Drosophila Genome Center funded by the National Human Genome Research Institute, National Cancer Institute, and Howard Hughes Medical Institute (http:www.fruitfly.org). A web robot program (19) was used to collect gene sequences and other information from the NCBI Drosophila melanogaster Genome Resources (Build 3.2). The TIGR Drosophila melanogaster Gene Index was downloaded, and their EST library information was extracted from the NCBI dbEST database using a search strategy “drosophila[orgn] AND gbdiv_EST[prop]” for the fly EST entries in the GenBank database (Table 1). The current version of Drosophila melanogaster Gene Index contains 24, 291 tentative consensus sequences, 11, 108 singleton ESTs, and 890 singleton ETs (expressed transcripts), which correspond to a total of 382, 000 D. melanogaster ESTs. Annotation of the tentative consensus and singleton EST sequences were based on the best match in the BLAST similarity search with minimum criteria of match length >200 bp and E < 10−9. Gene expression pattern and profiling were conducted in six single tissues with ESTs derived from non-normalized libraries. SOM maps were used to decompose the complex multiple tissue expression data into simple two-dimensional patterns using the Genesis package (http:genome.tugraz.at). Briefly, SOM is an unsupervised neural learning algorithm, which finds prototype vectors that represent the input data and at the same time realizes a continuous mapping from input space to a lattice of a defined number of neurons. Biologically, SOM mapping of EST data would reveal clusters of genes with similar expression behavior, possibly because of common regulatory factors or pathways. A detailed description of SOM clustering algorithm is available in Xu et al. (20). Preferentially expressed genes in fat bodies were identified with transcriptional logarithm of odds >3.0, which was developed by Wu et al (21). Finally, the fat body transcriptome was mapped along each chromosome based on a TDF (18), with a window of 0.1 Mb and a moving interval of 0.02 Mb.

Acknowledgement

  1. Top of page
  2. Abstract
  3. Research Methods and Procedures
  4. Acknowledgement
  5. References

This study was supported by the Agricultural Research Center, Washington State University.

Footnotes
  • 1

    Nonstandard abbreviations: EST, expressed sequence tag; SOM, self-organizing map; TDF, transcript density factor; NCBI, National Center for Biotechnology Informatics; BLAST, Basic Local Alignment Search Tool.

References

  1. Top of page
  2. Abstract
  3. Research Methods and Procedures
  4. Acknowledgement
  5. References