Microarray design
We designed a 7,448 spot, dual-genome microarray containing 1,862 genes. The array includes 76 Buchnera genes representing the E. coli heat shock (HS) regulon (clpX, dnaJ, dnaK, grpE, hflB, hslU, hslV, htpX, htrA, ibpA, mopA and mopB), essential amino acid biosynthesis genes (argA, argB, argC, argD, argE, argF, argG, argH, aroA, aroB, aroC, aroD, aroE, aroH, aroK, carA, carB, cysD, cysG, cysH, cysI, cysJ, cysK, cysN, cysQ, dapA, dapB, dapD, dapE, dapF, glyA, hisB, hisC, hisD, hisF, hisI, ilvC, ilvH, ilvI, leuA, leuD, lysA, metE, pheA, thrA, thrB, trpA, trpC and trpG) and a number of other genes representative of various functional categories such as cold shock, cell wall biosynthesis and DNA repair (cspC, cysS, endA, lgt, mraY, mrcB, mrsA, murC, murE, murF, nlpD, prfB, serC, ung and yjeA). The array also includes, nine aphid developmental and heat shock genes (distal-less, eyeless, fused, nanos, vasa, vasa-like, wingless, wnt2 and hsp70) and 1777 A. pisum ESTs (GEO Accession: GSE3742). All Buchnera genes were amplified using gene-specific primers designed from the Buchnera of A. pisum genomic sequence (GenBank Accession: BA000003, [6]). Primers were designed, where gene length permitted, to give an optimal product of 1000 base pairs with a melting temperature of 55°C. If gene coding sequences were less than 1000 base pairs, primers were designed to maximize product length. We amplified Buchnera genes in two 50 μl PCR reactions from a laboratory line of A. pisum. This line, 7-2-1 was collected in August 2001 in Cayuga County, New York from alfalfa by Jacob Russell. Since its collection, 7-2-1 has been maintained in continuous parthenogenetic culture in the laboratory of NAM.
All A. pisum ESTs were derived from a cDNA library constructed in the laboratory of DLS (described in [34], GenBank Accessions: CF587442 - CF588411 and CN582088 - CN587684). ESTs were clustered in a unigene set and 1824 genes were selected for inclusion based primarily on significant tblastx scores to other eukaryotes when used as queries in searches of the GenBank nr database. The corresponding cDNAs were amplified using T3 and T7 primers to pBluescript.
Genes were arranged in duplicate spots on slides, with Buchnera genes grouped into the first and second of 24 subarrays. In addition to spot duplication, the entire set was printed twice on each slide thus; each spot was printed in quadruplicate.
Heat shock
A single isofemale line of A. pisum was used in all HS experiments. This clone, called "TUC", was collected by NAM from fava bean in a garden in Tucson, Arizona in June 1999 and has been maintained in continuous parthenogenetic culture on fava bean at 20°C in the laboratory of NAM since its collection. Clone TUC, in addition to harboring the primary aphid symbiont, Buchnera, is host to the secondary bacterial symbiont, Candidatus Serratia symbiotica, previously known as the "R-type" [35].
Six to ten adult A. pisum were placed on a single fava bean plant in a cup cage and allowed to reproduce at 20°C 16 L:8 D. After 24 – 48 hours the adult aphids were removed from plants and flash-frozen in liquid nitrogen. First generation aphids were then maintained for one week, the developmental period required to attain the adult stage. Heat shock experiments were carried out 1 – 2 days following ecdysis to adulthood. Prior to HS, 10 – 20 control aphids were removed from each plant, flash-frozen in liquid nitrogen and stored at -80°C until RNA extraction and microarray hybridization. Heat shock conditions were applied by ramping from 20°C to 36°C over a two hour period at a rate of 2°C/15 min and then maintaining aphids at 36°C for an additional 2 hours (4 hours in total). At 4 hours cultures were removed from the growth chamber and aphids were quickly collected into Eppendorf tubes, flash-frozen in liquid nitrogen and stored at -80°C until processing. Two biological replicates of clone TUC were exposed to heat shock in parallel.
RNA extraction and microarray hybridization
We extracted total RNA from control and HS samples using an RNeasy Mini Kit (Qiagen Cat. No.74104). RNA quantity and quality was assessed using an RNA LabChip® on the Agilent 2100 Bioanalyzer. We constructed cDNA from 10 μg of total RNA using Omniscript Reverse Transcriptase and reaction buffer (Qiagen), 6 ng random hexamer primers, 0.5 mM each dATP, dCTP, dGTP, 0.3 mM dTTP, and 0.2 mM amino-allyl linked dUTP. RNA was hydrolyzed, and purified cDNA populations were coupled to Cy3 or Cy5 fluors (Amersham). We flipped dye labels between treatments across experimental replicates. Therefore, results from 4 total slides were analyzed. Purified, labeled cDNA populations were suspended in 80 μl Sigma Hybridization Buffer with 20 μg Human COT1 DNA. Microarray slides were rehydrated, UV crosslinked, and denatured immediately before hybridization. Target samples were combined, denatured for 10 min at 94°C, and pipetted into sealed, microvolume slide manifolds. We ran hybridizations in a GeneTAC automated station at 47°C for 16 hours and programmed 3 washes of variable-stringency SDS/SSC solutions (Sigma).
Analysis
Signal intensities were measured using spot-finding software, softWoRx Tracker (version 2.20, MolecularWare Inc). Raw median spot intensity and median background intensity data were exported; median background signal intensities were subtracted from median signal intensities. A zero signal value was substituted for any negative signal intensities and all non-zero values were log2-transformed. In order to compensate for differences in expression levels between the two genomes represented on the array we analyzed each species separately, by normalizing across probes for each genome. For each species we imported log2-transformed data into SAS version 9 and used 'PROC MIXED' to model and remove systematic sources of variation contributed by Cy3 versus Cy5 dyes, slide and hybridization effects, slide × dye interactions and slide × treatment interactions [36, 37]. Statistical residuals generated from the normalization step were subjected to mixed-model analyses of variance (log2 normalized data for this experiment can be accessed at GEO Accession: GSE3742). We regarded as potentially meaningful any response that showed fold-change greater than four or that had P-values less than 0.0001.
Verification of expression with RT-qPCR
We verified microarray results with real-time quantitative PCR from cDNA from both biological replicates used for microarray hybridization for four of the Buchnera genes that appeared to be upregulated under HS conditions (dnaK, ibpA, mopA, and yjeA), two Buchnera genes that did not show a HS response (argG and cysG) and two Buchnera genes that appeared to be down-regulated under HS conditions (aroK and carA) as well as five aphid ESTs that appeared to be upregulated under HS conditions (EST5, EST6: hsp70, EST7, EST8 and EST10, Figure 1b), two aphid genes that did not show a heat shock response (EST13: ef1α and EST14), and one aphid EST that appeared to be down-regulated under HS conditions (EST12, Figure 1b) following the protocol described in [38]. Primers were designed to amplify short regions of these genes and were as follows:
argG: F 5' TTCACTATGTCATGGTGCTACTG 3' and R 5' TGCCAGGAATTTTCATCTTTAC 3'; aroK: F 5' CTGGGAAAAGCACTATTGGTC 3' and R 5' ACCTCCTCCTGTAGCAAGAAC 3'; carA: F 5' TTGCGATTACGCTATTCATGC 3' and R 5' CCTTAACAGGATGGTTGCCTC 3'; cysG: F 5' AGGTGGTGATCCCTTTATTTTC 3' and R 5' GAGTATTTGCGATGTGTCAGTG 3'; dnaK: F 5' ATGGGTAAAATTATTGGTATTG 3' and R 5' ATAGCTTGACGTTTAGCAGG 3'; ibpA: F 5' CCAATATCTGATACACCGAC 3' and R 5' TGAACAGATATATCTAATTCTTTTTC 3'; mopA: F 5' ATGTAAAAGACGGAAAAGG 3' and R 5' CCTGCTGGAGAAGAACTAG 3'; yjeA: F 5' CGAACAAAAAGGTTGATACCATTC 3' and R 5' TTATTAGCATCAGAAAGCGGATC 3'; EST13 – ef1α: F 5' CTGATTGTGCCGTGCTTATTG 3' and R 5' TATGGTGGTTCAGTAGAGTCC 3'; EST6 – hsp70: F 5' TAAGAGGAAAACTAAAAAGGACG 3' and R 5' GGAAACTCGGGTGTAGAAATC 3'; EST14: F 5' TATGTTGCAGCAGCCGATAC 3' and R 5' AATGCTGGTCCTTCAGATGC 3'; EST5: F 5' CAAGAAATTCTCCCCTCATTTG 3' and R 5' TGCTCTTGCTAACCCACCAC 3'; EST7: F 5' TTCAAATCGTCGAAGTGTGC 3' and R 5' AAGTCGTCGACCGTGATTTC 3'; EST8: F 5' GCTGTTCTGCGTTCACTTGTAC 3' and R 5' AGTCCGCTATTTTGGATATTGG 3'; EST10 F 5' TTTGGAAGGAATGGACTGTGG 3' and R 5' TTGAGCGTTGATTGTGTGGAC 3'; EST12: F 5' GAAAGAAGATTTGGTGCTTGG 3' and R 5' CTATTATGGGCACTTGCGATC 3'.
Verification of spot identity with select resequencing across the array
Spots that showed a > 4 fold-change (log2 fold-change = ± 2) and/or a significant fold-change (≥ -Log(P) = 5), together with the set of ESTs that we annotated through blast-searching as HS genes were resequenced using vector primer, T3, to confirm spot identity. The identity of the majority of resequenced spots (24/36 spots) was verified by a blastn search restricted to the NCBI "EST_other" database [39], which includes the EST sequences for the fragments on our array. A subset of resequenced spots (11/36 spots) returned poor sequence, as expected if more than one template was present. However, three of these, CN585306 (EST3, Buchnera mopA), CN586357 (aphid dnaJ) and CN586725 (EST6, aphid hsp70), were annotated as HS genes and showed large and significant upregulation under HS conditions. The correspondence between gene annotation and change in expression for these genes indicates that the expected gene is present on the array in the correct position. This observation indicates that most elements on the array contain the correct sequence.
Investigation of cross-hybridization of heat-shock gene products
We investigated cross-hybridization of the most upregulated heat shock genes on our array by spiking control RNA with labeled PCR products corresponding to HS genes. Four Buchnera genes (dnaK, ibpA, mopA and yjeA) and one aphid gene (hsp70) were amplified. PCR reactions were carried out for each gene in two 25 μl volumes containing 50 mM KCl, 10 mM Tris-HCl, 1.5 mM Mg2+, 2.5 mM of each dATP, dCTP and dGTP, 1 mM dTTP, 1.5 mM amino-allyl linked dUTP (Sigma A0410-1MG), 0.25 units of Eppendorf Taq DNA polymerase and 10 ng of DNA (Buchnera genes) or cDNA (EST6 – hsp70). Primers used to amplify genes selected for spiking were as follows: Buchnera of A. pisum dnaK: F 5' CACCAGAGAGAACTCCTCCC 3' and R 5' ATGGATGGCAATAAACCACG 3', Buchnera of A. pisum ibpA: F 5' TCAATTAACACTAAGTATTCCTGG 3' and R 5' TAGGTTTTTCCTCTTCCGG 3', Buchnera of A. pisum mopA: F 5' ACCACAACAGCAACATTATTAGC 3' and R 5' ACCTCCAGCAACTACACCTTC 3', Buchnera of A. pisum yjeA: F 5' GAAGCGTTTATTAGCATCAG 3' and R 5' TGAACAAGGAGGTAAGCC 3', A. pisum hsp70: F 5' GTGTTGATATTTGACCTGGGC 3' and R 5' AATGGTAGAGTTGCGTTCGAC 3'.
We performed a standard dye-flip hybridization experiment using RNA isolated from a control line and spiked with either 20 ng of the aphid hsp70 PCR product or 20 ng of each of the four Buchnera gene PCR products. Spikes were added to the RNA samples following generation of cDNA and prior to cy3 and cy5 labelling. Thus, for one hybridization, the aphid spike was cy3 labeled and the Buchnera spikes were cy5 labeled and for the second hybridization the dyes were reversed. In this way we were able to distinguish between spots that were cross-hybridizing with aphid hsp70 versus the Buchnera genes. Data from the spike array hybridization experiment were analyzed using the analysis pipeline described above for the heat shock experiments.