Specificity and Sensitivity Analysis of Oligonucleotide Probes for WholeGenome Microarrays of Desulf

1
Specificity and Sensitivity Analysis of
Oligonucleotide Probes for Whole-Genome
Microarrays of Desulfovibrio vulgaris and
Geobacter metallireducens Zhili He1, Liyou Wu1,
Rayford B. Payne2, Judy D. Wall2, Jizhong Zhou1
and Matthew W. Fields1 1Oak Ridge National
Laboratory, Oak Ridge, TN 2University of
Missouri, Columbia, MO
Abstract Microbial bioremediation is of
great scientific and practical interest. The
genetic and physiological capability of
microorganisms for the reduction and
transformation of environmentally toxic metals
and radionuclides is evident in nature however,
little is known regarding the molecular
mechanisms and regulatory networks controlling
these processes. Access to the genomic contents
of metal and radionuclide-reducing bacteria
combined with microarray technologies provides an
opportunity to elucidate metal/radionuclide
respiratory pathways, thereby providing avenues
for predictable and effective bioremediation
practices. Desulfovibrio vulgaris Hildenborough
has been the focus of biochemical and
physiological studies in the laboratory, and the
metabolic versatility of this organism has been
largely recognized. D. vulgaris is capable of
coupling the oxidization of a variety of electron
donors (e.g., lactate, pyruvate, succinate,
ethanol) to the reduction of many different
electron acceptors (e.g. sulfate, fumarate, iron,
uranium, chromium, potentially O2). The capacity
of this bacterium to reduce different metals and
radionuclides enzymatically has been
demonstrated, and the focus of the proposed work
is the identification and characterization of the
cellular mechanisms for these reductions.
Desulfovibrio vulgaris Hildenborough is a
d-Proteobacterium with a genome of 3.6 Mb that is
65 GC. The genome sequence is complete and the
gaps have been closed (http//www.tigr.org/tigr-sc
ripts/ufmg/ReleaseDate.pl). Desulfovibrio spp.
are relatively easy to culture, and significant
amounts of biomass can be harvested when grown in
the presence of sulfate and other metals. This
versatility clearly facilitates laboratory
growth, maintenance, and manipulation of D.
vulgaris. Geobacter metallireducens is a
d-Proteobacterium with a genome size of
approximately 3.5 Mb. Geobacter spp. can reduce
iron oxides as well as radionuclides. In
addition, some species can degrade organic
contaminants. We are currently conducting
experiments to compare the utility of cDNA-based
and oligonucleotide-based microarrays. Once the
comparison is complete, whole genome microarrays
for D. vulgaris and G. metallireducens will be
constructed.
Introduction DNA microarrays are a revolutionary
tool for parallel analysis of whole-genome data
and gene expression. Currently both oligo-based
and cDNA microarrays are increasing used.
Oligo-based microarrays are becoming more popular
because of a number of advantages over cDNA
microarrays 1) only sequence information is
needed and PCR can be avoided, 2) the oligo
approach entails systematic design of oligo
probes to screen a whole genome for gene
identification and discovery, and 3) more
flexibility to control specificity of
hybridization (Relogio et al., 2002).
Experimental results show that oligo microarrays
compare well with cDNA microarrays, and that a
single probe per gene is sufficient to monitor
gene expression (Kane et al., 2000, and Shoemaker
et al., 2001). The challenge is how to identify
the optimum probes for each gene, and to
determine appropriate conditions for specificity.
General considerations include specificity
algorithms, oligo Tm, free energy, G-C content,
secondary structure, gene families and
cross-hybridization behavioral. Specificity is
one of the most important elements in the
development of oligo probes for genome-scale
microarrays. Currently, there are several free
and commercially available programs that can be
used for oligonucleotide probe design (Table 1).
Signal Intensity Comparison of PCR Product Probes
to Oligo Probes of Different Lengths
Specificity and Sensitivity Analysis Using
Synthesized Oligo Probes and Targets
Probe Design and Preliminary Expression Analysis
of Desulfovibrio vulgaris and Geobacter
metallireducens
A. 50 pg target
B. 10 pg target
C. 2.0 pg target
D. 0.4 pg target
A. Genomic DNA
B. Total RNA
cy5
T1 T2 T3 T4
cy3
50-mer
70-mer
C. Genomic DNA
D. Total RNA
50-mer oligo arrays
70-mer oligo arrays
cy3 target
cy3 target
E. Genomic DNA
F. Total RNA
cy5 target
cy5 target
Genomic DNA (A, C and E) and total RNA (B, D and
F) hybridized with microarrays containing PCR
product, 70-, 60-, 50-, 40- and 30-mer oligos at
42oC, 45oC, 50oC and 60oC. PCR products produced
the strongest signals the majority of the time.
In some cases, 70mer oligos produced strong
signals, but further work is needed. Signal
intensity increased with an increase in oligo
length.
PCR
Table 2 and Table 3 show the selected probe
information for G. metallireducens and D.
vulgaris by three different programs,
respectively. A specific probe is unique to one
gene, and a non-specific probe may
cross-hybridize with one or more genes. A
rejected ORF sequence denotes predicted ORFs
where a probe could not be designed.
ArrayOligoSelector chose more specific probes
than the others, and it also designed almost all
probes for the whole genome. OligoArray2
selected too few probes in total or a number of
specific probes. OligoPicker is fast but
selected fewer probes than ArrayOligoSelector.
Therefore, based on data, we chose
ArrayOligoSelector to design probes. Designed
probes were also checked by BLAST (Altschul et
al., 1997) for specificity and MFOLD (Zuker et
al., 1999) for secondary structures. Currently,
designed probes for G. metallireducens only cover
about 75 of the whole genome. Probes on the
microarray were based upon predicted ORFs for
Geobacter metallireducens. The array was
hybridized with cDNA from Desulfovibrio vulgaris
cells in the logarithmic phase or stationary
phase of growth. Signal intensities differed
from spot to spot, but a majority of probes
displayed some degree of hybridization at 45C
with 50 formamide. The true extent of cross
hybridization is not yet known, but the results
suggested that higher hybridization temperatures
are required. Interestingly, enzymes indicative
of logarithmic growth had significantly increased
hybridization signals compared to stationary
phase cells. For instance, probes for aconitase,
fructose-1,6-bisphosphatase, and glycogen
synthase had 33-fold, 175-fold, and 24-fold
increases for logarithmic phase cells,
respectively (Table 4).
Oligo-60
Oligo-40
Four synthesized targets (T1, T2, T3 and T4) were
equally mixed and hybridized with 50-mer and
70-mer oligo microarrays at 45oC with 50
formamide. T1, T2 and T3 were labeled with cy5
dye and T4 with cy3 dye. Those oligo arrays were
constructed with synthesized oligo probes which
were complementary to one of those targets or
mismatched with different number ( 3 to 37) of
nucleotides. The results showed that high
concentrations of targets caused different
degrees of hybridization even at low probe-target
similarities. 70-mer oligo arrays produced
higher sensitivity but lower specificity than
50-mer oligo arrays. About 2.0 pg of a target is
specific for 50-mer oligos and around 0.4 pg is
specific for 70-mer oligos. Future work will use
genomic DNA and total RNA to further study oligo
hybridization behavioral, especially specificity
and sensitivity.
Hybridized at 45oC
Hybridized at 45oC
Oligo-70
Oligo-50
Oligo-30
Conclusions Oligo probe specificity covers at
least two aspects. One is to select specific
probes using genome and sequence data and robust
software, and the other is to obtain specific
hybridization signals by optimizing hybridization
conditions. Our preliminary data suggested that
probe length and concentrations of targets
greatly affected probe specificity. Longer oligos
produced stronger signals but also increased
cross-hybridization. Therefore, specificity and
sensitivity need to be balanced. We designed all
probes for the whole genome of D. vulgaris and
75 of probes for G. metallireducens using
ArrayOligoSelector compared with other
programs. Preliminary results suggest that 70mer
oligoarrays could perform in a similar fashion to
microarrays constructed with PCR products.
Hybridizations should be done with 50 formamide
and between 45C and 50C. The concentration of
target nucleic acids can affect hybridization
specificity. Once the microarrays are
constructed, we will use microarrays for the
identification of genes and pathways involved in
sulfate, uranium and chromium reduction. Future
specific tasks include ? Determine expression
profiles of D. vulgaris and G. metallireducens
cells when grown in the presence of sulfate-,
chromium, and uranium as electron acceptors ?
Identify specific systems of D. vulgaris that
respond to various levels of nitrate and
nitrite ? Determine the whole genome expression
patterns of cells exposed to different
environmental stresses
The results suggested that optimal hybridization
temperature was very critical for obtaining
specific and strong signals. Different
microarrays may need to hybridize at different
temperatures. For example, cDNA arrays were at
50oC, and 70-mer and 50-mer oligo arrays at 45oC.
However, actual hybridization temperature needs
to be considered for probe specificity.
Cy3 image
Cy3 images
References Altschul SF, Madden TL, Schaffer AA,
Zhang J, Zhang Z, Miller W and Lipman DJ (1997)
Nucleic Acids Res. 25 3389-3402 Kane MD, Jatkoe
TA, Stumpf CR, Lu J, Thomas JD, and Madore SJ
(2000) Nucleic Acids Res. 28 4552-4557. Li F and
Stormo G (2001) Bioinformatics 17(11)
1067-1076. Relogio A, Schwager C, Richter A,
Ansorge W and Valcarcel J (2002) Nucleic Acids
Res. 30 e51. Rouillard J-M, Herbert CJ and Zuker
M (2002) Bioinformatics 18(3) 486-487. Rouillard
J-M, Zuker M and Erdogan Gulari (2003) Nucleic
Acids Res. 31 3057-3062. Shoemaker DD, Schadt
EE, Armour CD, He YD, Garrett-Engele P, McMonagh
PD, Loerch PM, Leonardson A, Lum PY, Cavet G et
al. (2001) Nature 409 922-927. Wang X and Seed B
(2003) Bioinformatics 19 796-802. Xu D, Li G, Wu
L, Zhou J and Xu Y (2002) Bioinformatics 18
(11)1432-1437. Zuker M, Mathews DH and Turner DH
(1999) Algorithms and thermodynamics for RNA
secondary structure prediction A practical
guide, NATO ASI Series. Kluwer Academic
Publishers, Dordrecht, NL
Cy5 images
Cy5 image