Development and Evaluation of a Comprehensive Functional Gene array for Environmental Studies Zhili He 1,2, C. W. Schadt 2, T. Gentry 2, J. Liebich 3, S.C. Song 2, X. Li 4, and J. Zhou 1,2 To detect and monitor functions of microbial organisms in their environments, functional gene arrays (FGAs) have been used as a promising and powerful tool. In this study, we have constructed the second generation of FGA, called FGA2.0 that contains 23,843 oligonucleotide (50mer) probes and covers more than 10,000 sequences of targeted genes, which are involved in nitrogen, carbon, sulfur cycling and metabolism, metal reduction and resistance, and organic contaminant degradation. Several new strategies have been implemented in probe design, array construction and data analysis. Gene sequences were automatically retrieved by key words. A newly developed oligonucleotide design program CommOligo was used to select gene-specific and group-specific probes, and multiple probes were designed for each gene sequence or each group of highly homologous sequences. All designed oligonucleotides were verified and output in a 96-well format for direct order placement of oligonucleotide synthesis. To ensure the array specificity, the array has been systematically evaluated using different targets and environmental samples. The results demonstrate that such an array can provide specific analysis of microbial communities in a rapid, high- through-put and cost-effective fashion. ABSTRACT EXPERIMENTAL DESIGN 1 The University of Oklahoma, Norman, OK, 2 Oak Ridge National Laboratory, Oak Ridge, TN, 3 Forschungszentrum Julich GmbH, Julich, Germany, 4 Perkin Elmer Life and Analyetical Sciences, Boston, MA RESULTS CONCLUSIONS This research was funded by the U.S. Department of Energy (Office of Biological and Environmental Research, Office of Science) grants from the Genomes To Life Program and ERSP Program. ACKNOWLEDGEMENTS Oligonucleotide design and synthesis. A computer program CommOligo (Li et al., 2005) was used to design gene-specific and group-specific probes based on the following criteria: (i) gene-specific probes: =-35 kcal/mol free energy; (ii) group-specific probes: >=96% sequence identity, >= 35-base continuous stretch, and <=-60 kcal/mol free energy (He et al., 2005a; Liebich et al., 2006). Each gene sequence or a group of homologous sequences had up to three probes. All verified probes were synthesized without modification by MWG Biotech, Inc. (High Point, NC) in a 96-well plate format with the concentration of 100 pmol/µl. Oligonucleotide target synthesis. 25 oligonucleotides were synthesized as gene-specific and group-specific targets to evaluate the FGA specificity (Table 1). 50 pg for each oligonucleotide was used for hybridizations with a single target or a mixture of multiple targets. Preparations of PCR-generated targets. 17 target genes were selected, and their PCR products (PCR-amplicons) were obtained using gene-specific primers and standard PCR methods (Table 2 and Table 3). Each PCR product had a minimal length to cover all available probes (1, 2 or/and 3 depending on probes selected) on the array. DNA labeling and hybridization. The PCR-amplicons were fluorescently labeled by random priming using Klenow fragment of DNA polymerase as described previously (He et al., 2005b). Hybridization was at 50 o C with 50% formamide. Fig. 1 Major steps for construction of a comprehensive 50mer oligo functional gene array. CommOligo is the core program to select gene-specific and group-specific oligonucleotide probes. GeneDownloader, ProbeChecker, and PlateProducer were Perl scripts to pre- process gene sequences or post-process oligonucleotide probes.  For gene-specific probes, Fig. 2 shows the distribution of maximal sequence identities (Fig. 2A), maximal stretch lengths (Fig. 2B), or minimal free energy (Fig. 2C) with their non-targets. Most of the probes (~70%) had maximal sequence identities 72%~84%, stretch lengths 12~15 bases, and 0~-30kcal/mol free energy.  For group-specific probes, Fig. 3 shows the distribution of minimal sequence identities (Fig. 3A), minimal stretch lengths (Fig. 3B), or minimal free energy (Fig. 3C) with their group members. Most of the probes (~92%) had maximal sequence identities 100%, stretch lengths 45~50 bases, and free energy values of -65 kcal/mol or smaller. Fig. 4 The FGA was hybridized with a mixture of 15 synthesized oligonucleotide targets at 42 o C, 45 o C, 50 o C and 60 o C. Balancing probe sensitivity and specificity, the optimal hybridization temperature was determined to be o C with 50% formamide, which is generally consistent with our previous results.  Signal intensities for probe B and C were normalized with probe A (100%), and there were 14, 12 and 10 probe A, B, and C, respectively (Table S2 and Table S3).  The average of relative signal intensities for probe A, B and C were 100%, 103.8%, and 97.6%, respectively, and similarly, the average of SNR values were 73.1, 67.2 and 65.3 for probe A, B, and C, respectively (Fig. 5).  The results suggest that three probes performed similarly with known targets. 1.An FGA2.0 has been constructed with more than 23,000 oligos covering more than 10,000 gene sequences. To our knowledge, this is the most comprehensive FGA for environmental studies. 2.To ensure the array specificity, several new features has been implemented in the probe design, and array construction. 3.The FGA2.0 has been systematically evaluated using oligonucleotide and PCR-amplicon targets, and demonstrates that it can be used as a powerful tool for a rapid, high-through-put and cost-effective analysis of microbial communities. 4.The array can be used to profile microbial community differences, to address specific questions and/or hypotheses related to microbial population dynamics, and analyses of functional gene expression in microbial communities. FGA II design strategies:  1. Using MSA to identify conserved regions for each functional gene.  2. Using experimentally established oligonucleotide design criteria and the novel software tool CommOligo.  3. Designing gene-specific and group-specific probes.  4. Multiple probes for each sequence or each group of sequences. 15.2% probes target carbon metabolism genes 22.2% probe target the genes involved in nitrogen cycling 6.8% probes for sulfur reduction genes 3.6% probes for methane reduction and oxidation 19.0% probes target genes involved in metal reduction and resistance 34.0% probes target genes involved in degradation of organic compounds Fig. 2 Fig. 3  For oligo targets, there were three false positives and two false negatives, and for PCR-amplicon targets, four false positives and no false negatives observed (Table 5).  Possible reasons include: (i) First, the amounts of some oligonucleotides or PCR-amplicons applied to the array was too high or too low; (ii) Probe design criteria used were not specific enough for excluding all non-specific probes, and that some additional criteria may need to be considered; (iii) an optimization of hybridization conditions may improve probe specificity; (iv) there may be errors in probe or/and gene sequences.  To tackle the problem of false positives, relative comparisons are needed. Fig. 5 Relative signal intensities and SNR values detected by probe A, B and C for PCR-amplicon targets. REFERENCES He Z, Wu L, Li X, Fields MW and Zhou J (2005a). Appl. Environ. Microbiol. 71: He Z, Wu L, Fields MW and Zhou J (2005b). Appl. Environ. Microbiol. 71: Li X*, He Z* and Zhou J (2005). Nucleic Acid Res. 33: (*Co-first authors). Liebich J, Schadt CW, Chong SC, He Z, Rhee SK and Zhou J (2006). Appl. Environ. Microbiol. 72: N125