Gene expression profiles underlying alternative caste phenotypes in a highly eusocial bee, Melipona...

Insect Molecular Biology (2006)

15

(1) 33ndash44

copy 2006 The Royal Entomological Society

33

Blackwell Publishing Ltd

Gene expression profiles underlying alternative caste phenotypes in a highly eusocial bee

Melipona quadrifasciata

C C Judice M F Carazzole F Festadagger M C Sogayardagger K HartfelderDagger and G A G Pereira

Departamento de Geneacutetica e Evoluccedilatildeo Instituto de Biologia Universidade Estadual de CampinasndashCampinas Brazil

dagger

Departamento de Bioquiacutemica Instituto de Quiacutemica Universidade de Satildeo Paulo Satildeo Paulo Brazil and

Dagger

Departamento de Biologia Celular e Molecular e Bioagentes Patogecircnicos Faculdade de Medicina de Ribeiratildeo Preto Universidade de Satildeo Paulo Ribeiratildeo Preto Brazil

Abstract

To evaluate caste-biased gene expression in

Meliponaquadrifasciata

a stingless bee we generated 1278 ESTsusing Representational Difference Analysis Mostannotated sequences were similar to honey bee genesof unknown function Only few queen-biased sequenceshad their putative function assigned by sequence com-parison contrasting with the worker-biased ESTs Theexpression of six annotated genes connected to castespecificity was validated by real time PCR Interestinglyqueens that were developmentally induced by treatmentwith a juvenile hormone analogue displayed an expres-sion profile clearly different from natural queens forthis set of genes In summary this study represents animportant first step in applying a comparative genomicapproach to queenworker polyphenism in the bee

Keywords representational difference analysis sub-tractive hybridization stingless bee juvenile hormonecaste development

Introduction

The queen and worker castes of social insects are primeexamples of polyphenism generated by divergent

developmental pathways In the social Hymenoptera castepolyphenism occurs exclusively in the female of the speciesas is attributed to haplodiploid sex determination resultingin asymmetries in genetic relatedness amongst offspring(Wilson 1971) In most social insects the developmentalmechanisms underlying this complex divergence in mor-phological characteristics and behavioural patterns cangenerally be summarized as a specific set of environmentalstimuli that trigger a threshold response in the endocrinesystem (Hartfelder amp Emlen 2005) The latter then takes overthe lead during the metamorphic stages and drives caste-specific patterns in gene expression (Evans amp Wheeler 2001)

Most gene expression studies on caste development insocial insects have been performed in the honey bee

Apismellifera

(Evans amp Wheeler 2000 Hepperle amp Hartfelder2001 Guidugli amp Hartfelder 2004) These studies focusedon differential gene expression during critical larval stagesusing suppression subtractive hybridization and differentialdisplay PCR strategies The changes in gene expressionpatterns that determine division of labour and taskperformance in adult worker bees have been revealed ina microarray system generated from a large-scale ESTsequencing approach on a brain cDNA library (Whitfield

et al

2002) As a result of these previous studies thecomplex social behaviour of bees can be theoreticallyinterpreted as changes in gene regulation (Robinson ampBen-Shahar 2002 Whitfield

et al

2003) evident as age-and task-related changes in gene expression in the brainof adult bees (Kucharski amp Maleszka 2002 Takeuchi

et al

2002 Whitfield

et al

2002 2003 Grozinger

et al

2003)Recent advances in the functional genomics of honey beecaste development and division of labour are now accom-panied by a full-scale genome sequencing project (httphgscbcmtmceduprojectshoneybee) and by further ESTlarge-scale sequencing (Nunes

et al

2004)Both genomic and expressed sequence information

should provide tools for the investigation of regulatorynetworks underlying the connection between the endocrinesystem developmental programs and behaviour in theevolution of insect sociality

Unraveling such networks and especially unveiling the sofar unknown function of conserved genes is still an arduous

doi 101111 j1365-2583200500605x

Received 1 February 2005 accepted after revision 1 August 2005 Corre-spondence Gonccedilalo Amarante Guimaratildees Pereira Laboratoacuterio de Genocircmicae Expressatildeo Departamento de Geneacutetica e Evoluccedilatildeo Instituto de BiologiaUniversidade Estadual de Campinas 13083ndash970 Campinas SP BrazilTel 0055 193 7886 237 fax 0055 193 7886 235 e-mail goncalounicampbr

34

C C Judice

et al


Insect Molecular Biology

15

33ndash44

task in social insects they are not easily amenable to genetictests or to transgenic analyses To date the only successfulstrategy to test gene function in the honey bee is by RNAinterference protocols (Beye

et al

2002 Amdam

et al

2003) Comparative analysis of gene expression in closelyrelated species of similar social status could therefore beexpected to provide information on possible conservedcircuits involved in developmental pathways and in divisionof labour between adult colony members We undertooka study on differential gene expression in the castes ofa stingless bee

Stingless bees (Meliponini) are a large tribe of highlyeusocial bees comprising over 300 species in 14 recog-nized genera (Camargo amp Pedro 1992) The Meliponini areclosely related to the honey bees (Apini) (Michener 2000)which they parallel with respect to social organization (Engelsamp Imperatriz-Fonseca 1990) The queen and workerphenotypes of stingless bees are strongly divergent in theirmorphology and behaviour and in most of the stingless beegenera the development of these caste differences is initiatedby a nutritional signal that is queen larvae are provided witha much larger quantity of larval food than worker larvae

An exception to this rule is the stingless bee genus

Melipona

where brood cells are all of the same size and allcontain similar quantities of larval food The large percent-age of queens emerging from the brood combs is apuzzling phenomenon that has inspired a model for castedetermination by genetic predisposition in a two-locustwo-allele system (Kerr 1950) After the allelic difference isthought to have set the primary developmental switch forcaste development the endocrine system is then calledinto action to subsequently direct differentiation into thequeen or worker pathways during metamorphosis In thisaspect of endocrine regulation we can thus see a generalconvergence in caste development of the social Hymenoptera(Hartfelder amp Rembold 1991 Bonetti

et al

1995 Pinto

et al

2002) The genus

Melipona

therefore differs fromthe other highly social bees primarily in the initial trigger ofcaste development but not so much in the actual differen-tiation processes during metamorphosis These featuresmake

Melipona

an interesting model for comparativestudies on caste development in bees especially on thepossible consequences resulting from the transition froma nutritional determination of caste as observed in themajority of social bees to a supposed genetics-basedmechanism in

Melipona

(Kerr 1950) In general termssuch a transition is reminiscent of the frequent switchesfrom environmental sex determination (ESD) to genetic sexdetermination (GSD) frequently observed and describedin reptiles (Greenbaum amp Carr 2001)

To investigate the molecular mechanisms underlying castedevelopment in

Melipona

we recently undertook a compara-tive study on caste-biased gene expression in


by a differential display RT-PCR approach

Most of the differentially expressed genes that wereannotated with respect to function turned out to be over-expressed in the worker caste (Judice

et al

2004) Theoverexpression observed for the putative Rab10 Sin3Aand cytochrome P450 homologs in

Melipona

workers isparalleled by their similar pattern of overexpression in honeybee foragers (Whitfield

et al

2003 Tsuchimoto

et al

2004)suggesting common patterns of caste-specific gene expres-sion in highly eusocial bees In this study we extendedthe approach by establishing a representational differenceanalysis (RDA) methodology With this aim we did not includedeterminants of caste that act in the premetamorphic andmetamorphic stages and focused on gene expression injuveniles that had not yet started to exhibit any caste-specific behaviour

Because the method calls for high-throughput sequenc-ing it was also expected to generate a large set of ESTsfor comparative analyses with the honey bee databaseFrom the set of assembled and functionally annotatedsequences we selected a set of six clones for which wedesigned specific primers to confirm their differentialexpression by real-time RT-PCR This validation analysiswas performed not only on naturally produced queens andworkers but also included queens that were obtained bytreatment of larvae with the juvenile hormone analoguepyriproxifen (PPN) Treatment with authentic juvenile hormone(JH) or with its analogs has long been shown to inducethe expression of morphological queen characteristicsin emerging adults of stingless bees (Campos

et al

1975Bonetti

et al

1995) Consequently we determined whetherqueens induced by PPN treatment express this set ofgenes at the same level as naturally produced queens

Results

Differentially expressed genes library sequencing EST quality and redundancy analysis

Sequencing of cDNA subtractive libraries clones fromnewly emerged bees (worker vs queen) generated usingthe RDA methodology (Pastorian

et al

2000) resulted in atotal of 1510 sequences (Table 1) composed of 658 readsfrom the queen minus worker (QU) library and 860 readsfrom the worker minus queen (WO) library The sequenceswere analysed and 228 reads were filtered as low quality(Phred lt 20 82 sequences from the QU library and 146from the WO library) or as contaminants (12 representedribosomal RNA vector or poly A sequences two QU and10 WO) The set of 1278 high quality ESTs consisted of 574sequences (45) from QU and 704 sequences (55) fromWO After assembly using the CAP3 program these readswere clustered into 337 unique sequences from which 14were subsequently removed from further analysis becausethey lacked a clear coding region (non-coding RNA) Theremaining unique sequences were distributed into 134

Caste-biased gene expression in

Melipona quadrifasciata 35



15

33ndash44

contigs (contiguous sequences originating from 1005 ESTs)and 189 singlets Table 1 summarizes these data The uniquesequences were deposited in G

EN

B

ANK

under accessionnumbers CO729449-CO729473 CO776957 and CO578497-CO578779

The high efficiency of the RDA methodology to extractdifferentially expressed transcripts is illustrated by the factthat 84 of the contigs were composed of sequenceswith exclusive occurrence in the WO (92) library or the QUlibrary (21) Only 21 contigs were represented by ESTsfrom both libraries (MIX contigs) and from these eightshowed a clear tendency to be predominantly representedby ESTs from one of the libraries For example contigs119 114 and 139 were formed with a queen-to-workerESTs ratio of 224 11 70 2 and 56 2 respectively

Analysis of assembled sequences

The unique sequences were submitted to similarity searchesusing the direct nucleotide comparison (B

LAST

N) or com-parison after dynamic translation (B

LAST

X and TB

LAST

X)against non-redundant (NR) and

Apis

EST and genomicdatabases (

Apis

Unigene NCBI and Honeybee genomeproject version 12 NHGRIUSDA) Approximately half ofthe assembled sequences (160 474) showed similarityto sequences in the non-redundant andor

Apis

databases

(B

LAST

X and TB

LAST

X respectively E

le

10

minus

5

) while for theremaining sequences (177 526) we did not obtain B

LAST

hits (Fig 1)To check the protein coding potential in the set without

B

LAST

matches sequences were submitted to searches forputative open reading frames using the ORF Finder pro-gram (NCBI) (httpwwwncbinlmnihgovgorfgorfhtml)As mentioned above (Table 1) 14 sequences did not showany clear ORF The other 163 sequences yielded a positiveresult for possible open reading frames and were classified aslsquounknown ORFsrsquo (Fig 1) In this figure the unique sequenceswere also organized according to their identity as followsgroup I (NR minus

Apis

) consists of four sequences thatmatched in the NR but not in the

Apis

database group II(NR and

Apis

) has 22 sequences with matches in both NRand

Apis

databases group III (

Apis

only) has 134 uniquesequences with matches only in

Apis

database Group IV(no hits ORF) contains the unknown ORFs The relativefrequencies of the unique sequences and the ESTs in theirdifferent categories were observed for each caste andidentity group (Table 2) showing that the ESTs are not equallydistributed between the castes and possibly reflects differ-ences in the physiology of workers and queens

For all sequences of group I and for 22 sequences of groupII putative protein functions could be assigned (Table 3)most of which based on their

Drosophila melanogaster

orthologs Interestingly one sequence of group II codingfor a putative serinethreonine kinase showed homology to

Apis

and mammals (

Mus musculus

) but not to anotherinsect sequence In the queen caste only five genes couldbe annotated three of which were related to gene expres-sion control (transcription and signalling) and two to proteinbiosynthesis For the worker caste a much larger numberof genes were annotated and these covered a much widerspectrum of molecular function andor biological processcategories such as learning memory behaviour defenceprocesses cell signalling growth and maintenance andgenes involved in transcriptional regulation

Table 1 Results of base quality analysis and CAP3 assembly for Melipona quadrifasciata ESTs sequenced from clones obtained in the cDNA-RDA screen

Total QU WO MIX

Total ESTs 1510 658 860Excluded sequences 228 84 156Total High Quality ESTs 1278 574 704Total Unique Sequences 337 85 229 23Non Coding RNA (NCR) 14 (84) 7 5 2Total contigs minus NCR 134 21 92 21( ESTs) (1005) (71) (412) (427 95)Total singlets minus NCR 189 57 132

Figure 1 Distribution of CAP3 assembled Melipona ESTs after similarity searches employing BLASTX and TBLASTX tools on non-redundant and honey bee databases The number of sequences present in each group is shown in parentheses The lsquoBLAST matchrsquo results (white area) were considered as significant at an e-value le 1E-05 These are represented by 474 of the assembled sequences The lsquono BLAST matchrsquo sequences (black area) comprise 526 of the assembled ESTs After screening for possible technical artifacts 8 of the lsquono BLAST matchrsquo sequences were considered as non-coding RNA and 92 were classified as lsquounknown ORFsrsquo (left pie graph) which consist of 163 ESTs shorter than 350 bp and 3 ESTs of at least 350 bp (ORF finder minimum length set to 33 amino acids) The significant lsquoBLAST matchrsquo results are shown relative to the Apis databases and the non-redundant database 83 of the Melipona sequences showed similarity to Apis genes 14 exhibited similarity to genes from Apis and other insects 2 matched with genes from insects other than Apis and 1 matched with Apis and mammal (right circle)

36

C C Judice

et al



15

33ndash44

Gene expression and juvenile hormone action

We used quantitative real-time PCR to assess transcriptlevels for six genes with a predicted function that were uniqueto either the worker (

n

= 5) or queen (

n

= 1) libraries

This set included the predicted transcripts for fatty acidsynthase from the QU library and for amino acid-polyaminetransporter diacylglycerol kinase groucho dunce andparamyosin from the WO library (Fig 2) For the genesselected from the WO library the relative quantification oftranscripts in newly emerged queens and workers showedthat genes coding for paramyosin diacylglycerol kinase(DGK) dunce groucho and amino acid-polyamine trans-porter are significantly more expressed in the worker castewhen compared with natural queens (

P

lt 005 for aminoacid-polyamine transport and

P

lt 001 for the other genes)Of the six predicted genes only the one from the QU librarynamely fatty acid synthase (FAS) did not display the expectedhigher expression in the queen RNA sample

In this real-time PCR study we compared transcriptabundance not only between naturally raised queens andworkers but in addition we included RNA samples fromqueens that were induced by PPN treatment of larvae Withthis comparison we wanted to test to which extent thephenotypic plasticity in morphological characteristicscorresponds to gene expression patterns (Fig 2) The real-

Table 2 Relative frequencies of unique sequences (contigs and singlets) and ESTs for each caste according to their identity groups and source library (QU queen-library WO worker library US unique sequences)

Relative EST frequencies in relation to caste

Identity groups QU WOContigs US EST US EST(I) NR minus Apis 00 00 09 06(II) NR and Apis 26 07 54 116(III) Apis only 115 162 192 310(IV) No Hits ORF 128 728 156 362Total 269 897 411 793Singlets(I) NR minus Apis 00 00 04 02(II) NR and Apis 38 05 27 09(III) Apis only 231 32 254 89(IV) No Hits ORF 462 65 296 106Total 731 103 581 207

Table 3 Putative Biological Process or Molecular Function for the unique sequences (from identity groups I and II) showing significant similarity to known sequences The reads were organized according to their caste-specific occurrence in the RDA libraries and by the number of ESTs in contigs after assembly (EST)

Id_GENBANK EST

Apis matchBest match in non-redundant database_BLASTX

OrganismPutative biological process or molecular functionGEN EST Gene name

Genes identified in queensCO729469 2 e-26 Grainy head (e-10) D melanogaster Transcription factorCO729464 2 e-31 Fatty acid synthase (e-16) A gambiae Fatty acid biosynthesisenzymeCO729457 1 e-29 CG13880-PA (e-22) D melanogaster Protein biosynthesisCO729454 1 e-60 Glutamate receptor (e-22) D melanogaster Metabotropic glutamate receptorCO578657 1 e-09 CG2839 (e-06) D melanogaster Nucleic acid bindingGenes identified in workersCO729459 48 e-23 JhI-21 (e-10) A gambiae Amino acid-polyamine transporterCO729460 7 e-87 Roundabout (e-20) D melanogaster Axon guidanceCO729458 6 e-35 e-50 Paramyosin (e-21) D melanogaster Structural constituent of muscleCO729466 5 e-39 e-13 Diacylglycerol kinase (e-13) D melanogaster Protein C kinase activationCO729449 5 NM ENSANGP00000016183 (e-11) A gambiae Pol proteinCO729467 4 e-26 Scavenger receptor (e-13) D melanogaster Defense responseCO729471 4 e-31 e-26 CG2909-PA (e-07) D melanogaster UnknownCO729470 3 e-30 e-13 ENSANGP00000024457 (e-07) A gambiae PDZ DomainCO729461 3 NM Pol polyprotein (e-18) A gambiae Pol proteinCO729450 2 e-12 e-52 Dunce (e-39) D melanogaster cAMP-specific phosphodiesteraseCO729452 2 e-96 Serinethreonine kinase (e-23) M musculus Serinethreonine kinase activityCO729453 2 e-57 MICAL (e-24) D melanogaster MagnetoreceptionCO729463 2 e-17 Sulphate permease (e-16) D melanogaster Sulphate transportCO578625 2 e-10 TPA HDC16673 (e-14) D melanogaster ATPase activityCO729456 2 e-81 CG32542-PA (e-24) D melanogaster Epigenetic control of protein activityCO729472 1 e-22 Receptor type 99 A precursor (e-07) D melanogaster Protein tyrosine phosphataseCO729462 1 e-108 e-117 Tryptophan hydroxylase activator (e-18) A gambiae Tryptophan hydroxylase activatorCO729455 1 e-07 e-83 PK61-C (e-23) D melanogaster Protein amino acid phosphorylationCO729468 1 NM Receptor guanylyl cyclase (e-09) A gambiae ATP bindingCO729451 1 e-48 Na+K+ ATPase (e-36) D melanogaster Na+K+ exchanging ATPaseCO729473 1 e-18 ENSANGP00000015979 (e-10) A gambiae UnknownCO729465 1 e-29 Groucho (e-14) A gambiae Transcription corepressor

Id GENBANK accession number Genes for which higher expression in the predicted caste was confirmed by real time PCR are shown with the respective level of statistical significance (P lt 005 P lt 001) (see Fig 2) BLASTX results show the best hits against the non-redundant (nr) NCBI database including E-score value and organism The TBLASTX results of Melipona sequences against Apis mellifera are represented by their score values (Apis match) against genome (GEN) or EST database (NM) indicates a no match for the Apis database

Caste-biased gene expression in Melipona quadrifasciata 37

copy 2006 The Royal Entomological Society Insect Molecular Biology 15 33ndash44

time PCR analysis revealed that only the FAS gene showeda similar expression between natural and artificial queensand this was also the gene for which we could not confirma differential expression between queens and workersFor all other genes the expression levels in PPN-inducedqueens differed from natural queens For the DGK encodinggene the PPN-induced queens showed an intermediateexpression level when compared to natural queens andworkers For the genes encoding putative amino acid-polyamine transporter paramyosin groucho and duncehomologs the situation was even more extreme as inthese cases the expression levels for PPN-induced queensby far exceeded even those of worker samples This clearlyindicates a divergence in hormone-induced morphologicalqueen characters and the caste-biased gene expressionpattern at the transition from pupal to the adult stage

Discussion

EST analysis and annotation

Although we obtained similar numbers of high quality ESTsfrom both the queen and the worker libraries (574 and 704respectively) we noted a divergent distribution patternfor these sequences once they were clustered (85 uniquesequences for queens vs 229 for workers) Conversely90 of the queen ESTs became assembled into only 27of the queen unique sequences while 79 or the workerESTs formed 41 of the unique sequences of this caste(Table 2) In particular we detected that 427 ESTs from thequeen library were concentrated in 21 MIX contigs (with379 ESTs in only four contigs) vs 95 ESTs from the sub-tracted worker library (Table 1) These data suggest thatamong the differentially expressed genes the queen librarypresents some highly transcribed sequences Finally weobserved that almost 60 of the queen unique sequences(contigs and singlets) were classified as no hits vs 45 inthe worker unique sequences Taken together these results

suggest that the transcriptional pattern of a worker bee is morecomplex in the sense of a higher diversity of genes expressed

With respect to the results of local alignment matcheswe noted that 526 of the ESTs did not yield a BLAST

match At first glance this large number of no-match ESTsis somewhat striking since these ESTs cannot all representspecies-specific transcripts An ORF analysis revealed that92 of these no-match ESTs contained a recognizableprotein coding region whereas for 14 sequences no ORFcould be detected but could represent technical artifactsHowever recent studies on honeybee transcripts also iden-tified ESTs that contained no significant ORF but that wereclearly expressed in different organs and castes and arethought to have a potential function as nuclear RNAs(Sawata et al 2002 Sawata et al 2004) Potentially thisset could also include microRNAs (Bartel 2004) There-fore the no-ORF Melipona sequences should be analysedfurther to identify possible functions especially in light ofthe fact that they were obtained from subtracted libraries

When compared to other high-throughput EST sequenc-ing projects on bees (Whitfield et al 2002 Nunes et al2004) the number of no-match ESTs with a recognizableORF in these is much lower This difference could be attrib-uted to a number of factors Firstly the RDA methodologystarts out with a reverse transcription reaction that uses anoligo dT primer and thus many of the no-match ESTs mayactually represent 3prime-untranslated regions of expressedgenes A second reason may also be attributable to theRDA protocol (as this methodology inherently normalizeslibraries) but also to highly enrich subtracted transcripts(transcripts that may normally be under-represented inEST library sequencing projects) The most importantreason however seems to be that this class of ESTs containan exceedingly large number of short ESTs ndash a problemencountered not only here but also in previous studieson the honey bee transcriptome This is detailed in Fig 1where 160 of the 163 ESTs with an ORF but no match in

Figure 2 Relative gene expression in newly emerged females of Melipona quadrifasciata for genes with an RDA-predicted occurrence in either the queen or the worker caste The Y-axis is on log10 scale indicating the gene expression levels (2DDCt method) and the calculated errors (plusmn SE) for newly emerged workers natural queens and for queens obtained by juvenile hormone (PPN) treatment of last instar larvae (1 microgmicrol acetone) and for workers that were obtained from the acetone control treatment Different letters over bars (a b c) indicate a significant difference (P lt 005 for aminotransporter and DGK genes P lt 001 for paramyosin and groucho genes and P lt 0001 for dunce gene) The table under the real-time PCR graph shows the number of ESTs representing each of the genes in the RDA libraries

38 C C Judice et al


the nr database (unknown ORFs) are shown to be shorterthan 350 bp and only three ESTs are equal to or larger than350 bp The problem and effect of EST size also becomesapparent when comparing the respective percentages ofno-hit ORFs in the contigs (128 and 156) and the singlets(462 and 296) that were formed after CAP3 assembly ofthe transcripts (see Table 3)

The next point to determine is whether the 126ndash156no-hit ORFs in the contigs actually represent Melipona-specific genes In a recent comparison we estimated that226 of a total of 5000 honey bee ESTs may actuallyrepresent species-specific transcripts (Nunes et al 2004)when compared to Drosophila and Anopheles ndash the onlyfully sequenced insect genomes (Adams et al 2000 Holtet al 2002) The estimated divergence time for Hymenop-tera and Diptera is set at over 280 million years Consider-ing the more recent divergence between the two tribes ofhighly social bees ndash they are clearly separated in Baltic amber(Engel 2001) ndash the above figure of a 12ndash15 transcriptspecificity for Melipona may indeed be a realistic estimatefor transcriptome differences due to phylogenetic distance

With respect to ESTs representing conserved sequencesthe majority of the putative genes listed in Table 3 showshigh similarity to sequences from honey bee workersand the expression of some of these have already beenreported for A mellifera For example the transcriptspreviously identified in newly emerged Melipona workersas homologs of the Sin3A transcriptional regulator (Judiceet al 2004) diacylglycerol kinase RNA binding proteinand Na+ K+-ATPase have been described as overexpressedin foraging workers of A mellifera (Tsuchimoto et al 2004)Genes such as dunce and one encoding a protein kinaseC (PKC)-like kinase were shown to be expressed in adulthoney bee workers (Whitfield et al 2002) and their coun-terparts are involved in behavioural traits of Drosophila(Davis amp Dauwalder 1991 Tinette et al 2004) Thisindicates that highly eusocial bees utilize a common set ofgenes in the molecular network underlying their physiologicaland behavioural phenotype

The other set of conserved sequences contains theunknown yet supposedly conserved sequences that dis-played significant matches with the Apis database but notwith other insect species It consists of 363 ESTs (108 fromthe QU and 255 from the WO subtractive cDNA libraryrespectively) The higher number of transcript matchesfrom the Melipona worker library may represent a biasdue to the fact that the honey bee EST database consistsmainly of sequences from worker brain cDNAs (Whitfieldet al 2002) or from worker-derived open reading frameESTs (Nunes et al 2004) and also could be related to oursmall number of sequences that matched against ESThoney bee sequences Interestingly this group of ESTs thatare conserved between Apis and Melipona contains 83 ofthe unique sequences with blast matches We take this to

mean that the group of social bees may have a set of exclu-sive genes that may be related to its specific life historytraits and that these genes may not have close matches inthe two fully sequenced dipteran genomes or in the nearcomplete silkworm genome (Mita et al 2004) These beesequence tags may therefore contain interesting candi-dates for further investigations on processes that may bespecific to caste development and division of labour insocial insects A similar picture can be drawn for the annot-ation result of a Melipona gene a serinethreonine kinasewith a mammalian gene as its closest match Such best-match relationships for cell signalling-related genes havealso been described for honey bee ESTs with human genesand were thought to reflect certain higher-order organismicfunctions (Whitfield et al 2002 Nunes et al 2004)

Functional annotation of queen- and worker-specific transcripts

Among the conserved sequences isolated during thisscreen for caste-biased genes only a small set was func-tionally interpretable (59 and 68 of the queens andworkers unique sequences respectively) according to theirsimilarity to genes of known function In the queen casteonly five genes could be annotated Two of these arerelated to gene regulation a predicted protein similar to ahoney bee gene with a LIM domain a motif found inmany regulators of developmental pathways (Kadrmas ampBeckerle 2004) and the grainy head like gene a D mela-nogaster homolog of a transcription factor involved in Wntsignalling (Lee amp Adler 2004) Two other genes that wereidentified in the queen library are apparently involved inprotein biosynthesis and in fatty acid synthesis The fattyacid synthase gene encodes a multifunctional enzymerelated to fatty acid and cholesterol metabolism (Bennettet al 1995) Itrsquos differential expression could be related tothe higher accumulation of fat in queens However the realtime PCR results indicate that this gene is also expressedin workers at a similar level (Fig 2) Another gene showedsimilarity to a metabotropic glutamate receptor which cou-ples with various intracellular effectors through G-proteinsto mediate and modulate synaptic transmission (Holscher1999) In the honey bee this gene was also found and hasbeen reported to be more expressed in brain cells (Funadaet al 2004)

In the worker caste only one transcriptional regulatorhomolog was identified the groucho gene which encodesa long-range corepressor that silences transcriptionof linked promoters in a relatively indiscriminate manner(Barolo amp Levine 1997 Courey amp Jia 2001) Most of thefunctionally annotated genes in this caste are related to cellsignalling We could detect a unique sequence composedof 48 ESTs from WO representing an amino acid-polyaminetransporter similar to D melanogaster JhI-21 which is presentin several insects In Drosophila this gene is induced by JH



and is involved in vitellogenesis (Dubrovsky et al 2002)Another cell signalling candidate is a putative Meliponahomolog of a gene that encodes diacylglycerol kinase Thisenzyme is critically involved in depleting the diacylglycerol-generated signalling complex (Kanoh et al 2002) In insectsthe inhibition of DGK activity has been shown to elevate theconcentration of DG inside the corpora allata cells resultingin a significant dose-dependent decrease in JH biosynthesis(Rachinsky et al 1994) Therefore the worker caste-biasedexpression of DGK could also provide a link between cellsignalling and the JH-related regulation of behavioural andphysiological caste characters in adult bees (for review seeHartfelder amp Emlen 2005) The higher expression of DGKin workers was confirmed by real time PCR

As another putative member in the DG metabolicpathway we isolated from the WO library a tryptophanhydroxylase activator annotated by GO as diacylglycerol-activatedphospholipid dependent protein kinase C inhibitorpresenting a 14-3-3 protein domain-specific binding activityIts D melanogaster homolog was initially described assimilar to a family of mammalian proteins called 14-3-3which activate tyrosine and tryptophan hydroxylases Theseare two key enzymes that regulate the biosynthesis ofbiogenic monoamine neurotransmitters (Swanson amp Ganguly1992) More recently studies have reported this gene to berelated to learning and memory (Broadie et al 1997)

Three other annotated genes of the worker library canalso be related to nervous system organization and functionnamely multidomain flavoprotein oxidoreductase (MICAL)that encodes a product with an actin binding activity whichacts in repulsive axon guidance (Terman et al 2002) theroundabout gene product which functions as an axonguidance (Kidd et al 1998) and the protein-tyrosine phos-phatase 99 A precursor that is a component of the plasmamembrane and has a protein tyrosine phosphatase activity(Tian et al 1991) In addition we isolated two putativereceptors that could also be related to nervous systemfunctions a guanylyl cyclase which is the main receptor fornitric oxide in vertebrates and invertebrates (Bicker 2001Ott et al 2004) and a scavenger receptor which interactswith a peptide binding domain named PDZ (Ikemoto et al2000) Interestingly two PDZ domain-containing proteinswere found in libraries generated as caste-specific transcriptsof M quadrifasciata workers one of them in the presentwork and the other one reported in a previous study (Judiceet al 2004)

The detection of dunce as a worker-expressed generepresents an interesting finding which based on itsDrosophila homolog may have direct implications in divi-sion of labour of social bees Its expression was confirmedas being higher in worker than in natural queens but it wasoverexpressed in artificial queens when compared tonatural queens This gene was recently found to be essen-tial for a form of cooperative behaviour in Drosophila called

lsquosearch-aggregationrsquo (Tinette et al 2004) This consists ofa two-step process initially the flies land randomly ondifferent food patches in a second phase the remaininggroup of flies move to the most favourable food source andaggregate there Dunce mutant flies were deficient in thissearch-aggregation behaviour Thus if this role of Duncecan be functionally confirmed in bees the higher expressionlevel of this gene in workers would indicate that a molecularprogram for co-operative behaviour may be expressedin newly emerged bees ndash before they actually becomeengaged in colony division of labour

Two other genes were similar to genes encoding polymer-ases (POL proteins) involved in retrotransposition (Shankaret al 2004) The Melipona sequences matched with coun-terparts of Anopheles gambiae Bombyx mori and even withSchistosoma mansoni but not with Apis where they have notbeen reported This kind of incongruent phylogeny is commonfor these elements which are widely spread and seem to besubject to horizontal transfer (Alberola amp de Frutos 1996)

The presence of metabolism-related gene expressionappears as an emergent feature already present in veryyoung queens that could be interpreted as priming orpreparation of the fat body for the future synthesis of pro-teins and lipids required for the high rates of egg production(Engels amp Engels 1977) In workers the high percentage(12 out of 22) of signalling and signal transduction-relatedtranscripts in the putatively annotated Melipona ESTs sug-gests that the groundwork for the performance of complextasks and behavioural shifts observed in Melipona workers(Waldschmidt et al 2002) is being laid down at this stage

Caste divergence between alternative phenotype morphology and gene expression

A selected set of the conserved and functionally annotatedMelipona transcripts was submitted to real-time PCRanalysis in order to validate the RDA results This analysiscarried out for paramyosin amino acid-polyamine transporterdiacylglycerol kinase dunce and groucho transcriptsconfirmed the caste-specificity of expression for five of thesix genes revealed in the RDA screen providing a furtherevidence for the efficiency of our subtractive libraries toreliably reveal differentially expressed genes in a highthroughput protocol Yet validating the RDA results was notthe sole purpose of this quantitative RT-PCR analysis wealso sought to investigate the strength of the connectionbetween the morphological phenotype (queen or worker)and gene expression patterns For this purpose we includedin the analysis a third set of cDNA samples derived fromqueens that were obtained by PPN treatment of spinning-stage larvae (Campos 1978)

The outcome of this comparison between natural queensand PPN-induced queens was quite surprising as it showedremarkable differences in expression levels of caste-specific genes In the expression of three out of five genes

40 C C Judice et al


that were drawn from the worker-specific RDA clones weobserved that PPN-induced queens were significantlydifferent from natural queens In this analysed set of genesthey were either intermediate between queens and workeror attained expression levels that were even significantlyhigher than those detected in newly emerged workers Thissurprising result could be tentatively interpreted under thehypothesis of a genetic predisposition to caste determina-tion in the genus Melipona (Kerr et al 1975) evidencing acertain degree of independence between the morphologicalcaste phenotype and gene expression patterns in youngadult bees The first step in caste determination is thoughtto involve a genotype-environment interaction in the earlylarval stages which then installs caste-specific programsof hormone synthesis during metamorphosis that directs theestablishment of the distinct morphologies (Pinto et al 2002)

From this point of view we interpret the current resultsas a relative independence between caste genotype andendocrine regulation of caste differentiation at least inrespect to the role of JH That is when a larva of the workerbut not of the queen genotype is treated with PPN duringthe critical larval instar thus mimicking the queen endocrinesituation in stingless bees (Hartfelder amp Rembold 1991) itmay actually develop the morphological queen phenotypeor atleast overexpress worker-biased genes in the adultstage

The results on a divergence between caste morphologyand adult gene expression draw new attention to the pro-posed genetic predisposition for caste in Melipona (Kerr1950) and more recent discoveries of further examplesof genetic caste determination in social Hymenopterareinforce this hypothesis (Julian et al 2002 Volny amp Gordon2002 Fraser et al 2000 Winter amp Buschinger 1986) Inlight of the present results the interaction of different alleleswith the endocrine system may explain why the experimentaldesign suggested by Velthuis (1976) to test genetic castedetermination in Melipona bees has failed This designrelied on the induction of queen development by JH andproposed the evaluation of caste ratios in the offspring ofthese queens Most of the JH-induced queens howeverfailed to establish colonies (Campos et al 1979) possiblybecause JH alone was not sufficient to establish the actualqueen transcriptome as shown in the present study It isimportant to note that our data are still too limited topropose a comprehensive model on caste differentiationHowever there is sufficient evidence to affirm that anartificial queen is not equal to a natural queen althoughwe can not estimate the extension of this difference We arecurrently investigating this issue using the microarray tech-nology on a larger set of ESTs including also no-match tags

In conclusion the RDA methodology generated a totalof 1278 high quality ESTs that represent 134 contigs and189 singlets with caste-biased expression in the highlyeusocial bee M quadrifasciata Although this total number

of sequences certainly provides only a first glimpse into thegeneral picture of caste transcriptomes this study representsan important first step in applying a comparative genomicsapproach to queenworker polyphenism in social bees

Experimental procedures

Samples and total RNA extraction

The queen and worker bees were obtained from Melipona quadri-fasciata anthidioides colonies kept at the Genetics DepartmentUniversity of Satildeo Paulo (Ribeiratildeo Preto campus) Entire broodcombs with late pupal stages were removed and maintained inan incubator at 28 degC Queens and workers were sampled 0ndash12 hafter emergence from the brood cells and were snap-frozen in liq-uid nitrogen and stored at minus80 degC The bee samples (10 workersand 10 queens) were homogenized in TRIzol reagent (Invitrogen)and total RNA was extracted and treated with DNase (RNasefreeDNaseI AmpGrade Invitrogen Carlsbad CA USA) RNA purityand integrity were checked spectrophotometrically and by electro-phoresis in 12 agarose gels under denaturing conditions

Induction of queen development by a juvenile hormone (JH) analogue

The application of juvenile hormone (JH) and JH analogs suchas pyriproxifen (PPN) to spinning-stage larvae has been shown toefficiently induce queen development in M quadrifasciata (Campos1978 Bonetti et al 1995) In the present study we induced queenformation by application of PPN dissolved in acetone (1 mgml)One microgram of PPN was topically applied on the cuticle ofspinning-stage larvae Control larvae received an acetone applica-tion After application the larvae were kept in an incubator untileclosion The queen characteristics induced by PPN were used torecognize queens at eclosion

Subtractive cDNA libraries

Two subtractive cDNA libraries for newly emerged stages weregenerated one representing the genes that are overexpressedin workers when compared to queens (named WO) and the otherrepresenting genes overexpressed in queens (QU) The methodologyemployed was the cDNA representational difference analysis(RDA) (Pastorian et al 2000) The WO library was establishedwith the cDNA of newly emerged workers as the tester populationhybridized against the cDNA of queens as the driver populationFor the QU library the tester was the queen cDNA and the driverwas the worker cDNA

In brief one microgram of total RNA isolated from both casteswas submitted to reverse transcription and long distance PCRusing the SMART PCR cDNA synthesis kit (Clontech LaboratoriesMountain View CA USA) The double-stranded cDNAs (15 microg)were digested with MboI (Fermentas Vilnius Lithuania) ligated toR-adapters (Table 4) and amplified by PCR These productsshowed a length of 02ndash15 kbp and were isolated in an exponentialPCR cycle to generate the respective cDNA representations Theenrichment of differentially expressed transcripts was achieved intwo successive rounds of PCR amplification that employed differentadapters (J and N Table 4) and subtractive hybridization of testerand driver populations in ratios of 1 100 and 1 800 respectivelyas previously described (Hubank amp Schatz 1999) The commontranscripts between driver and tester populations were removed



through the rounds of subtractive hybridization and the librarieswere enriched for differential transcripts DNA purification wascarried out using the GFX PCR purification kit (AmershamBiosciences Little Chalfont Bucks UK) The final differenceproducts were cloned into pGEM-T Easy vector (PromegaMadison WI USA) Competent DH10B Escherichia coli cells(Invitrogen) were transformed and positive recombinants submittedto plasmid extraction (Ausubel et al 1993) Inserts were sequencedusing a DYEnamic ET terminator kit and MegaBACE sequencer(Amersham Biosciences)

Quality analysis of sequences assembly and BLAST searches

The ESTs were filtered for ribosomal RNA contaminants andpoly (A) tail sequences as previously described (Telles amp da Silva2001) and the vector sequences were trimmed using the Cross-match program The quality of sequences was analysed by thePHRED program (Ewing amp Green 1998) and sequences with atleast 250 high quality bases (Phred gt 20) were considered aslsquohigh-qualityrsquo Searches for contaminant sequences such asbacterial DNA were performed using BLASTN (Altschul et al1997) against the non-redundant nucleotide sequences (nt)database at GENBANK (wwwncbinlmnihgov) The high qualitysequences were submitted to assembly by CAP3 (Huang ampMadan 1999) The assembled sequences were dynamicallytranslated by BLASTX and the deduced polypeptides were com-pared against the non-redundant database In addition we com-pared the M quadrifasciata sequences with those of the closelyrelated honey bee Apis mellifera by running them against thetranslated honey bee EST database (TBLASTX) from a honey beebrain cDNA library (wwwlifeuiucedurobinson) and against thehoney bee genome database (wwwhgscbcmtmceduprojectshoneybee) We also analysed the similarity of the respectivenucleotide sequences using the BLASTN tool on the GENBANK

(nr) and Apis databases (ESTs and genomic sequences) Matcheswere considered significant at a cutoff probability score E-value of10minus5 For the Melipona ESTs annotated as lsquono hits foundrsquo a searchfor open reading frames was performed using the ORF Finder

program (httpwwwncbinlmnihgovgorfgorfhtm) and ORFswere considered when they covered the majority of the ESTsequence (minimum length set to 33 amino acids) Molecular func-tion and biological process assignment was performed with GeneOntology set at level 3

Real-time PCR validation

To validate the caste-biased expression of genes detected by theRDA screen we selected a set of ESTs that showed similarity topredicted proteins with known function These were submitted toreal-time PCR analyses The selected ESTs represented genesannotated as fatty acid synthase from the QU library and as aminoacid-polyamine transporter diacylglycerol kinase groucho dunceand paramyosin from the WO library Gene-specific real-timeprimers were designed using the lsquoPrimer Expressrsquo software(Applied Biosystems Foster City CA USA) and the respectivetemplate sequences were from clones of the libraries except forthe 28S rRNA gene The PCR primer sequences used are listedin Table 4

For the real-time PCR analysis new cDNAs were synthesizedde novo from RNA extracted from newly emerged queens andworkers The extracted total RNA (1 microm) was treated with 10 UDNaseI (Invitrogen) and 40 U RNAse OUT inhibitor (Invitrogen)upon incubation at 37 degC for 30 min to remove contaminating DNAReverse transcription was performed using 200 U SuperScript10 mM dNTP mix oligo dT and random primers (50 ng) and thecDNA synthesis was carried out at 42 degC for 2 h RNA-cDNAhybrids were dissolved using 5 U RNAse H and incubated at 37 degCfor 30 min Real-time PCR assays were carried out using thecDNA products as templates in 96-well plates by incorporation ofthe intercalating dye SYBRGreenI (kit SYBRGreen PCR mastermix- Applied Biosystems Foster City CA USA) and the ABI Prism5700 detection system (Applied Biosystems) for fluorescencereading The amplification protocol used for all genes was 95 degCfor 10 min 40 cycles of 95 degC for 15 s and 60 degC for 1min

As a control for RNA concentration and amplification dynamicswe used a 28S rRNA sequence which is frequently used as

Adapters ndash RDA Symbol Sequence

R12 5prime-GATCTGCGGTGA-3primeR24 5prime-AGCACTCTCCAGCCTCTCACCGCA-3primeJ12 5prime-GATCTGTTCATG-3primeJ24 5prime-ACCGACGTCGACTATCCATGAACA-3primeN12 5prime-GATCTTCCCTCG-3primeN24 5prime-AGGCAACTGTGCTATCCGAGGGAA-3prime

Target gene ndash real-time PCRFatty acid synthase FASF 5prime-GATCGCGGGATTGATACCTACT-3prime

FASR 5prime-TCGACGGTAACAAAAGTCAAGGA-3primeDunce DunF 5prime-AGCCGACCTGCGACTTCTC-3prime

DunR 5prime-ACATGGACATTAGCCCAATGTG-3primeGroucho GrouF 5prime-CGGCGGACGGTTCGA-3prime

GrouR 5prime-GATCCCACGAACGCACTGT-3primeParamyosin PmyosF 5prime-ATCCGAGGGAAGATCCAGGTA-3prime

PmyosR 5prime-TGCCTCTTGTAGATGCTCATTTTC-3primeAmino acid-polyamine transporter AmitranspF 5prime-AGGGAAGATCCCGTCAAGAA-3prime

AmitranspR 5prime-GGGTCGTGTAAAATGCCATGT-3primeDiacylglycerol kinase DGKF 5prime-CTTCGTATCGATGCCAGCAA-3prime

DGKR 5prime-TTTTGTTGTTCGTCAATCCGTTT-3prime28S rRNA r28SF 5prime-ACGGTACTCGCACGGTATCG-3prime

r28SR 5prime-CACTTTCATTTCGCCTTTAGGTTT-3prime

Table 4 RDA adapters and real-time PCR primer sequences and their respective target genes

42 C C Judice et al


calibrator gene in real-time PCR studies (Pfaffl et al 2003) Therespective forward and reverse primers were derived from Amellifera and were designed to cover two different exons of the 28SrRNA gene (GENBANK accession number AJ302936) This enabledus to simultaneously detect 28S rRNA and amplification of genomiccontaminant visualized by their different PCR product size Forrelative quantification of gene expression we employed the com-parative Ct method (Pfaffl et al 2003) using the 28S rRNA Ct ofthe worker sample as a reference

The gene expression results were statistically analysed byANOVA and differences between means by a post hoc test (Tukeyrsquosat P lt 005 P lt 001 and P lt 0001) using the VarPC software(httpwwwwestatcomwesvar)

Acknowledgements

We would like to thank Anderson C Ferreira and CamilaOresco for help with the experimental protocols andWeyder C Santana for help in the collection of bees Thiswork was supported by the Satildeo Paulo State Foundation forResearch (FAPESP) and by the Brazilian federal agenciesCoordenaccedilatildeo de Aperfeiccediloamento de Ensino Superior(CAPES) and Conselho Nacional de Pesquisa e Desen-volvimento (CNPq)

References

Adams MD Celniker SE Holt RA Evans CA GocayneJD Amanatides PG et al (2000) The genome sequence ofDrosophila melanogaster Science 287 2185ndash2195

Alberola TM and de Frutos R (1996) Molecular structure of agypsy element of Drosophila subobscura (gypsyDs) constitut-ing a degenerate form of insect retroviruses Nucleic Acids Res24 914ndash923

Altschul SF Madden TL Schaffer AA Zhang J Zhang ZMiller W and Lipman DJ (1997) Gapped BLAST and PSI-BLAST a new generation of protein database search programsNucleic Acids Res 25 3389ndash3402

Amdam GV Simotildees ZLP Guidugli KR Norberg K andOmholt SW (2003) Disruption of vitellogenin gene functionin adult honeybees by intraabdominal injection of double-stranded RNA BMC Biotechnol 3(1) 8pp httpwwwbiomedcentralcom1472-675031

Ausubel FM Brent R Kingston RE Moore DD SeidmanJG Smith JA and Struhl K (1993) Current Protocols inMolecular Biology John Wiley amp Sons Inc New York USA

Barolo S and Levine M (1997) Hairy mediates dominant repres-sion in the Drosophila embryo EMBO J 16 2883ndash2891

Bartel DP (2004) MicroRNAs genomics biogenesis mecha-nisms and function Cell 116 281ndash297

Bennett MK Lopez JM Sanchez HB and Osborne TF(1995) Sterol regulation of fatty acid synthase promoter Co-ordinate feedback regulation of two major lipid pathways J BiolChem 270 25578ndash25583

Beye M Hartel S Hagen A Hasselmann M and Omholt SW(2002) Specific developmental gene silencing in the honey beeusing a homeobox motif Insect Mol Biol 11 527ndash532

Bicker G (2001) Nitric oxide an unconventional messenger in thenervous system of an orthopteroid insect Arch Insect BiochemPhysiol 48 100ndash110

Bonetti AM Kerr WE and Matusita SH (1995) Effects of juve-nile hormones I II and III in single and fractionated dosage inMelipona bees Rev Bras Biol 55(Suppl 1) 113ndash120

Broadie K Rushton E Skoulakis EM and Davis RL (1997)Leonardo a Drosophila 14ndash3ndash3 protein involved in learningregulates presynaptic function Neuron 19 391ndash402

Camargo JMF and Pedro SRM (1992) Systematics phylog-eny and biogeography of the Meliponinae (HymenopteraApidae) a mini-review Apidologie 23 509ndash522

Campos LAO (1978) Sex determination in bees VI effect of ajuvenile hormone analog in males and females of Meliponaquadrifasciata (Apidae) J Kansas Entom Soc 51 228ndash234

Campos LAO Velthuis HHW and Velthuis-Kluppell FM(1975) Juvenile hormone and caste determination in a sting-less bee Naturwissenschaften 62 98ndash99

Campos LAO Kerr WE and Silva DLN (1979) Sex determi-nation in bees VIII Relative action of genes xa and xb on sexdetermination in Melipona bees Rev Bras General 2 267ndash280

Courey AJ and Jia S (2001) Transcriptional repression the longand the short of it Genes Dev 15 2786ndash2796

Davis RL and Dauwalder B (1991) The Drosophila dunce locuslearning and memory genes in the fly Trends Genet 7 224ndash229

Dubrovsky EB Dubrovskaya VA and Berger EM (2002)Juvenile hormone signaling during oogenesis in Drosophilamelanogaster Insect Biochem Mol Biol 32 1555ndash1565

Engel MS (2001) A monograph of the Baltic amber bees andevolution of the Apoidea (Hymenoptera) Bull Am Mus Nat Hist259 5ndash192

Engels W and Engels E (1977) Vitellogenin und Fertilitaumlt beistachellosen Bienen Insectes Soc 24 71ndash94

Engels W and Imperatriz-Fonseca VL (1990) Caste develop-ment reproductive strategies and control of fertility in honeybees and stingless bees In Social Insects ndash an EvolutionaryApproach to Castes and Reproduction (Engels W ed)pp 168ndash230 Springer-Verlag Berlin Heidelberg New York

Evans JD and Wheeler DE (2000) Expression profiles duringhoneybee caste development Genome Biol 2 6pp httpgenomebiologycom200021research00011

Evans JD and Wheeler DE (2001) Gene expression and theevolution of insect polyphenisms Bioessays 23 62ndash68

Ewing B and Green P (1998) Base-calling of automated sequencertraces using phred II Error probabilities Genome Res 8 186ndash194

Fraser VS Kaufmann BP OB and Crozier RH (2000)Genetic influence on caste in the ant Camponotus consobrinusBehav Ecol Sociobiol 47 188ndash194

Funada M Yasuo S Yoshimura T Ebihara S Sasagawa HKitagawa Y and Kadowaki T (2004) Characterization of the twodistinct subtypes of metabotropic glutamate receptors fromhoneybee Apis mellifera Neurosci Lett 359 190ndash194

Greenbaum E and Carr JL (2001) Sexual differentiation in thespiny softshell turtle (Apalone spinifera) a species with geneticsex determination J Exp Zool 290 190ndash200

Grozinger CM Sharabash NM Whitfield CW and RobinsonGE (2003) Pheromone-mediated gene expression in thehoney bee brain Proc Natl Acad Sci USA 100(Suppl 2)14519ndash14525

Guidugli KCH and Hartfelder K (2004) A member of the short-chain dehydrogenasereductase (SDR) superfamily is a targetof the ecdysone response in honey bee (Apis mellifera L)caste development Apidologie 35 37ndash47



Hartfelder K and Emlen DG (2005) Endocrine Control of InsectPolyphenism In Comprehensive Molecular Insect Science(Elsevier Amsterdam) (eds Gilbert LI Iatrou K and Gill S)Vol 3 Endocrinology pp 651ndash703

Hartfelder K and Rembold H (1991) Caste-specific modulationof juvenile hormone-III content and ecdysteroid titer in postem-bryonic development of the stingless bee Scaptotrigonapostica depilis J Comp Physiol B 160 617ndash620

Hepperle C and Hartfelder K (2001) Differentially expressedregulatory genes in honey bee caste development Naturwis-senschaften 88 113ndash116

Holscher C (1999) Synaptic plasticity and learning and memoryLTP and beyond J Neurosci Res 58 62ndash75

Holt RA Subramanian GM Halpern A Sutton GG CharlabR Nusskern DR et al (2002) The genome sequence of themalaria mosquito Anopheles gambiae Science 298 129ndash149

Huang X and Madan A (1999) CAP3 a DNA sequence assem-bly program Genome Res 9 868ndash877

Hubank M and Schatz DG (1999) cDNA representational differ-ence analysis a sensitive and flexible method for identificationof differentially expressed genes Meth Enzymol 303 325ndash349

Ikemoto M Arai H Feng D Tanaka K Aoki J Dohmae NTakio K Adachi H Tsujimoto M and Inoue K (2000) Iden-tification of a PDZ-domain-containing protein that interacts withthe scavenger receptor class B type I Proc Natl Acad Sci USA97 6538ndash6543

Judice C Hartfelder K and Pereira GAG (2004) Caste-specific gene expression profile in the stingless bee Meliponaquadrifasciata ndash are there common patterns in highly eusocialbees Insectes Soc 51 352ndash358

Julian GE Fewell JH Gadau J Johnson RA and LarrabeeD (2002) Genetic determination of the queen caste in an anthybrid zone Proc Natl Acad Sci USA 99 8157ndash8160

Kadrmas JL and Beckerle MC (2004) The LIM domain fromthe cytoskeleton to the nucleus Nat Rev Mol Cell Biol 5 920ndash931

Kanoh H Yamada K and Sakane F (2002) Diacylglycerolkinases emerging downstream regulators in cell signaling sys-tems J Biochem (Tokyo) 131 629ndash633

Kerr WE (1950) Evolution of the mechanism of caste determina-tion in the genus Melipona Evolution 4 7ndash13

Kerr WE Akahira Y and Camargo CA (1975) Sex determina-tion in bees IV Genetic control of juvenile hormone productionin Melipona quadrifasciata (Apidae) Genetics 81 749ndash756

Kidd T Brose K Mitchell KJ Fetter RD Tessier-Lavigne MGoodman CS and Tear G (1998) Roundabout controls axoncrossing of the CNS midline and defines a novel subfamily ofevolutionarily conserved guidance receptors Cell 92 205ndash215

Kucharski R and Maleszka R (2002) Evaluation of differentialgene expression during behavioral development in the honey-bee using microarrays and northern blots Genome Biol 38pp httpgenomebiologycom200332research0007

Lee H and Adler PN (2004) The grainy head transcription factoris essential for the function of the frizzled pathway in the Dro-sophila wing Mech Dev 121 37ndash49

Michener CD (2000) The Bees of the World John HopkinsUniversity Press Baltimore 19 913

Mita K Kasahara M Sasaki S Nagayasu Y Yamada T Kan-amori H Namiki N Kitagawa M Yamashita H YasukochiY et al (2004) The genome sequence of silkworm Bombyxmori DNA Res 11 27ndash35

Nunes FM Valente V Sousa JF Cunha MA Pinheiro DGMaia RM Araujo DD Costa MC Martins WK CarvalhoAF et al (2004) The use of Open Reading frame ESTs(ORESTES) for analysis of the honey bee transcriptome BMCGenomics 5 84

Ott SR Delago A and Elphick MR (2004) An evolutionarilyconserved mechanism for sensitization of soluble guanylylcyclase reveals extensive nitric oxide-mediated upregulation ofcyclic GMP in insect brain Eur J Neurosci 20 1231ndash1244

Pastorian K Hawel L III and Byus CV (2000) Optimization ofcDNA representational difference analysis for the identificationof differentially expressed mRNAs Anal Biochem 283 89ndash98

Pfaffl MW Gerstmayer B Bosio A and Windisch W (2003)Effect of zinc deficiency on the mRNA expression pattern inliver and jejunum of adult rats monitoring gene expressionusing cDNA microarrays combined with real-time RT-PCRJ Nutr Biochem 14 691ndash702

Pinto LZ Hartfelder K Bitondi MMG and Simotildees ZLP(2002) Ecdysteroid titers in pupae of highly social bees relateto distinct modes of caste development J Insect Physiol 48783ndash790

Rachinsky A Zhang J and Tobe SS (1994) Signal transductionin the inhibition of juvenile hormone biosynthesis by allatostat-ins roles of diacylglycerol and calcium Mol Cell Endocrinol105 89ndash96

Robinson GE and Ben-Shahar Y (2002) Social behavior andcomparative genomics new genes or new gene regulationGenes Brain Behav 1 197ndash203

Sawata M Yoshino D Takeuchi H Kamikouchi A Ohashi Kand Kubo T (2002) Identification and punctate nuclear locali-zation of a novel noncoding RNA Ks-1 from the honeybeebrain Rna 8 772ndash785

Sawata M Takeuchi H and Kubo T (2004) Identification andanalysis of the minimal promoter activity of a novel noncodingnuclear RNA gene AncR-1 from the honeybee (Apis melliferaL) Rna 10 1047ndash1058

Shankar R Grover D Brahmachari SK and Mukerji M (2004)Evolution and distribution of RNA polymerase II regulatorysites from RNA polymerase III dependant mobile Alu elementsBMC Evol Biol 4 37

Swanson KD and Ganguly R (1992) Characterization of a Dro-sophila melanogaster gene similar to the mammalian genesencoding the tyrosinetryptophan hydroxylase activator andprotein kinase C inhibitor proteins Gene 113 183ndash190

Takeuchi H Fujiyuki T Shirai K Matsuo Y Kamikouchi AFujinawa Y Kato A Tsujimoto A and Kubo T (2002) Iden-tification of genes expressed preferentially in the honeybeemushroom bodies by combination of differential display andcDNA microarray FEBS Lett 513 230ndash234

Telles GP and da Silva FR (2001) Trimming and clusteringsugarcane ESTs Genet Mol Biol 24 17ndash23

Terman JR Mao T Pasterkamp RJ Yu HH and KolodkinAL (2002) MICALs a family of conserved flavoprotein oxi-doreductases function in plexin-mediated axonal repulsionCell 109 887ndash900

Tian SS Tsoulfas P and Zinn K (1991) Three receptor-linkedprotein-tyrosine phosphatases are selectively expressed oncentral nervous system axons in the Drosophila embryo Cell67 675ndash680

Tinette S Zhang L and Robichon A (2004) Cooperationbetween Drosophila flies in searching behavior Genes BrainBehav 3 39ndash50

44 C C Judice et al


Tsuchimoto M Aoki M Takada M Kanou Y Sasagawa HKitagawa Y and Kadowaki T (2004) The changes of geneexpression in honeybee (Apis mellifera) brains associated withages Zool Sci 21 23ndash28

Velthuis HHW (1976) Environmental genetic and endocrineinfluences in stingless bee caste determination In Phase andCaste Determination (M Luumlscher ed) pp 35ndash53 PergamonPress Oxford

Volny VP and Gordon DM (2002) Genetic basis for queen-worker dimorphism in a social insect Proc Natl Acad Sci USA99 6108ndash6111

Waldschmidt AM Marco-Junior P Barros EG and CamposLA (2002) Genetic analysis of Melipona quadrifasciata Lep(Hymenoptera Apidae Meliponinae) with RAPD markersBraz J Biol 62 923ndash928

Whitfield CW Band MR Bonaldo MF Kumar CG Liu LPardinas J Robertson HM Soares B and Robinson GE(2002) Annotated expressed sequence tags and cDNA micro-arrays for studies of brain and behaviour in the honey beeGenome Res 12 555ndash566

Whitfield CW Cziko AM and Robinson GE (2003) Geneexpression profiles in the brain predict behavior in individualhoney bees Science 302 296ndash299

Wilson EO (1971) The Insect Societies Harvard University PressCambridge Mass

Winter U and Buschinger A (1986) Genetically mediated queenpolymorphism and caste determination in the slave-makingant Harpagoxenus sublaevis (Hymenoptera Formicidae) EntomGener 11 125ndash137

34

C C Judice

et al



15

33ndash44

task in social insects they are not easily amenable to genetictests or to transgenic analyses To date the only successfulstrategy to test gene function in the honey bee is by RNAinterference protocols (Beye

et al

2002 Amdam

et al

2003) Comparative analysis of gene expression in closelyrelated species of similar social status could therefore beexpected to provide information on possible conservedcircuits involved in developmental pathways and in divisionof labour between adult colony members We undertooka study on differential gene expression in the castes ofa stingless bee

Stingless bees (Meliponini) are a large tribe of highlyeusocial bees comprising over 300 species in 14 recog-nized genera (Camargo amp Pedro 1992) The Meliponini areclosely related to the honey bees (Apini) (Michener 2000)which they parallel with respect to social organization (Engelsamp Imperatriz-Fonseca 1990) The queen and workerphenotypes of stingless bees are strongly divergent in theirmorphology and behaviour and in most of the stingless beegenera the development of these caste differences is initiatedby a nutritional signal that is queen larvae are provided witha much larger quantity of larval food than worker larvae

An exception to this rule is the stingless bee genus

Melipona

where brood cells are all of the same size and allcontain similar quantities of larval food The large percent-age of queens emerging from the brood combs is apuzzling phenomenon that has inspired a model for castedetermination by genetic predisposition in a two-locustwo-allele system (Kerr 1950) After the allelic difference isthought to have set the primary developmental switch forcaste development the endocrine system is then calledinto action to subsequently direct differentiation into thequeen or worker pathways during metamorphosis In thisaspect of endocrine regulation we can thus see a generalconvergence in caste development of the social Hymenoptera(Hartfelder amp Rembold 1991 Bonetti

et al

1995 Pinto

et al

2002) The genus

Melipona

therefore differs fromthe other highly social bees primarily in the initial trigger ofcaste development but not so much in the actual differen-tiation processes during metamorphosis These featuresmake

Melipona

an interesting model for comparativestudies on caste development in bees especially on thepossible consequences resulting from the transition froma nutritional determination of caste as observed in themajority of social bees to a supposed genetics-basedmechanism in

Melipona

(Kerr 1950) In general termssuch a transition is reminiscent of the frequent switchesfrom environmental sex determination (ESD) to genetic sexdetermination (GSD) frequently observed and describedin reptiles (Greenbaum amp Carr 2001)

To investigate the molecular mechanisms underlying castedevelopment in

Melipona

we recently undertook a compara-tive study on caste-biased gene expression in


by a differential display RT-PCR approach

Most of the differentially expressed genes that wereannotated with respect to function turned out to be over-expressed in the worker caste (Judice

et al

2004) Theoverexpression observed for the putative Rab10 Sin3Aand cytochrome P450 homologs in

Melipona

workers isparalleled by their similar pattern of overexpression in honeybee foragers (Whitfield

et al

2003 Tsuchimoto

et al

2004)suggesting common patterns of caste-specific gene expres-sion in highly eusocial bees In this study we extendedthe approach by establishing a representational differenceanalysis (RDA) methodology With this aim we did not includedeterminants of caste that act in the premetamorphic andmetamorphic stages and focused on gene expression injuveniles that had not yet started to exhibit any caste-specific behaviour

Because the method calls for high-throughput sequenc-ing it was also expected to generate a large set of ESTsfor comparative analyses with the honey bee databaseFrom the set of assembled and functionally annotatedsequences we selected a set of six clones for which wedesigned specific primers to confirm their differentialexpression by real-time RT-PCR This validation analysiswas performed not only on naturally produced queens andworkers but also included queens that were obtained bytreatment of larvae with the juvenile hormone analoguepyriproxifen (PPN) Treatment with authentic juvenile hormone(JH) or with its analogs has long been shown to inducethe expression of morphological queen characteristicsin emerging adults of stingless bees (Campos

et al

1975Bonetti

et al

1995) Consequently we determined whetherqueens induced by PPN treatment express this set ofgenes at the same level as naturally produced queens

Results

Differentially expressed genes library sequencing EST quality and redundancy analysis

Sequencing of cDNA subtractive libraries clones fromnewly emerged bees (worker vs queen) generated usingthe RDA methodology (Pastorian

et al

2000) resulted in atotal of 1510 sequences (Table 1) composed of 658 readsfrom the queen minus worker (QU) library and 860 readsfrom the worker minus queen (WO) library The sequenceswere analysed and 228 reads were filtered as low quality(Phred lt 20 82 sequences from the QU library and 146from the WO library) or as contaminants (12 representedribosomal RNA vector or poly A sequences two QU and10 WO) The set of 1278 high quality ESTs consisted of 574sequences (45) from QU and 704 sequences (55) fromWO After assembly using the CAP3 program these readswere clustered into 337 unique sequences from which 14were subsequently removed from further analysis becausethey lacked a clear coding region (non-coding RNA) Theremaining unique sequences were distributed into 134





15

33ndash44


EN

B

ANK





LAST


LAST

X and TB

LAST


Apis


Apis


Apis

databases

(B

LAST

X and TB

LAST

X respectively E

le

10

minus

5


LAST


B

LAST


Apis


Apis


Apis


Apis


Apis


Apis





Apis

and mammals (

Mus musculus



Total QU WO MIX



36

C C Judice

et al



15

33ndash44



n

= 5) or queen (

n

= 1) libraries


P


P







Id_GENBANK EST








Discussion








38 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al














15

33ndash44


EN

B

ANK





LAST


LAST

X and TB

LAST


Apis


Apis


Apis

databases

(B

LAST

X and TB

LAST

X respectively E

le

10

minus

5


LAST


B

LAST


Apis


Apis


Apis


Apis


Apis


Apis





Apis

and mammals (

Mus musculus



Total QU WO MIX



36

C C Judice

et al



15

33ndash44



n

= 5) or queen (

n

= 1) libraries


P


P







Id_GENBANK EST








Discussion








38 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al










36

C C Judice

et al



15

33ndash44



n

= 5) or queen (

n

= 1) libraries


P


P







Id_GENBANK EST








Discussion








38 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al













Discussion








38 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al










38 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al






















40 C C Judice et al





































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al
































42 C C Judice et al




Acknowledgements


References




































































44 C C Judice et al















































44 C C Judice et al










Gene expression profiles underlying alternative caste phenotypes in a highly eusocial bee, Melipona...

Documents

Transcript of Gene expression profiles underlying alternative caste phenotypes in a highly eusocial bee, Melipona...