Assessment of cry1 Gene Contents of Bacillus thuringiensis Strains by Use of DNA Microarrays

Abstract

A single Bacillus thuringiensis strain can harbor numerous different insecticidal crystal protein (cry) genes from 46 known classes or primary ranks. The cry1 primary rank is the best known and contains the highest number of cry genes which currently totals over 130. We have designed an oligonucleotide-based DNA microarray (cryArray) to test the feasibility of using microarrays to identify the cry gene content of B. thuringiensis strains. Specific 50-mer oligonucleotide probes representing the cry1 primary and tertiary ranks were designed based on multiple cry gene sequence alignments. To minimize false-positive results, a consentaneous approach was adopted in which multiple probes against a specific gene must unanimously produce positive hybridization signals to confirm the presence of a particular gene. In order to validate the cryArray, several well-characterized B. thuringiensis strains including isolates from a Mexican strain collection were tested. With few exceptions, our probes performed in agreement with known or PCR-validated results. In one case, hybridization of primary- but not tertiary-ranked cry1I probes indicated the presence of a novel cry1I gene. Amplification and partial sequencing of the cry1I gene in strains IB360 and IB429 revealed the presence of a cry1Ia gene variant. Since a single microarray hybridization can replace hundreds of individual PCRs, DNA microarrays should become an excellent tool for the fast screening of new B. thuringiensis isolates presenting interesting insecticidal activity.

The gram-positive bacterium Bacillus thuringiensis produces one or more insecticidal crystal proteins (Cry) in the form of an intracellular parasporal crystal (37). After ingestion by a susceptible insect, Cry proteins dissolve in the insect midgut, where most are subsequently activated by midgut proteases. The protease-resistant toxin binds to specific docking proteins on the microvillous surface of susceptible midgut epithelial cells and then oligomerizes (35). Finally, the oligomeric toxin inserts into the membrane, forming a pore (9). To date, the structure of the internal Cry pore in the membrane remains uncertain; however, studies with synthetic membranes show that during membrane integration, the toxin undergoes a conformational change in which the helix-rich domain I separates from domains II and III (38) and a hairpin, composed of helices α4 and α5 subsequently inserts into the membrane with α4 lining the pore lumen to create a functional ion channel (34). Toxin-exposed midgut epithelial cells eventually die by a colloid osmotic lysis mechanism (24).

Although many Cry proteins are structurally and functionally similar, the diversity of Cry toxins and their insecticidal spectra is immense (19, 20). More than 280 different Cry toxins are organized into 46 primary ranks based on amino acid similarities (19). With over 130 entries in the Cry databank, Cry1 toxins are the largest and best known family. Assessing the insecticidal potential of a single B. thuringiensis strain is complex, as strains typically harbor between one to six cry genes, some of which are known to be cryptic (19, 31). In addition to bioassays, other techniques like Southern blotting, PCR, and chromatography have been utilized for analyzing cry gene or toxin contents (4, 8, 14, 22, 23, 31). Unfortunately, these techniques share a similar drawback in that they are labor intensive. In comparison, the strength of DNA microarray technology lies in the ability to perform massive parallel event measurements simultaneously while requiring only small quantities of nucleic acid material per experiment. Although the majority of reports deal primarily with differential gene expression, studies describing the use of microarrays in environmental microbial studies are growing rapidly in number (27). They can be used for the detection of known DNA sequences, to reveal the presence of either microbial strains (2, 7, 12, 13, 16, 28, 41, 42) or functional genes (25, 36, 39, 40, 43) in a sample, or for the identification of isolates and their phylogenic relationships (11, 18, 29).

In this study, we have designed and assessed the feasibility of using an oligonucleotide-based DNA microarray (cryArray) to identify B. thuringiensis cry1 family genes. Additionally, oligonucleotides specific for several other cry genes at primary rank were also added. To ensure more reliable identification at the secondary- and tertiary-rank level, whenever possible, we used a consentaneous approach in which multiple hybridization positives are necessary before the presence of a gene can be ascertained. Our cryArray is able to rapidly and specifically detect and identify known cry1 genes and can provide information of the presence of novel cry genes.

MATERIALS AND METHODS

Strains and media.

Bacillus thuringiensis subsp. kurstaki strain HD-1 (4D1) cells were obtained from the Bacillus Genetic Stock Center maintained at the University of Ohio, Columbus, Ohio [http://www.bgsc.org]). A second B. thuringiensis subsp. kurstaki HD-1 strain came from a commercial formulation (Thuricide) and was shown previously to have lost a cry1Ab gene (30). Strain HD-73 (1), the HD-73 cry mutant (33), subsp. aizawai strain HD-133 (31), subsp. israelensis strain HD-500 (21), and the subsp. kenyae strain (32) have been described elsewhere. B. thuringiensis strains IB031, IB059, IB217, IB360, IB429, and IB585 were obtained from a Mexican strain collection (10). All strains were grown at 30°C overnight in Luria-Bertani broth (6) for genomic DNA extraction and purification.

Rapid genomic DNA extraction and purification.

Bacterial cells from 1.5 ml of an overnight culture were harvested by a 1-min centrifugation at 13,000 rpm at room temperature. The cell pellet was resuspended in 400 μl of sterile deionized water and placed in a boiling water bath for 15 min. After cooling at room temperature for 10 min, the tubes were centrifuged 15 min as before and 200 μl of the supernatant was transferred to a new tube and stored at −20°C until needed.

Probe DNA.

The cry genes from B. thuringiensis and their expressed gene products are classified in four ranks (see Fig. S1 in the supplemental material). The primary rank groups toxins together on a relatively broad-based amino acid similarity, which then gets refined by subdivision into a secondary rank (identified by uppercase letters, e.g., Cry1A), which in turn is further subdivided to form a tertiary rank (identified by a lowercase letter, e.g., Cry1Aa). Finally a quaternary rank regroups the most closely related toxins (e.g., Cry1Aa1, Cry1Aa2, Cry1Aa3, etc.). Our design focused on the largest primary toxin group or rank, Cry1, with the inclusion of a few other groups but only at the primary rank. The Cry1 group currently possesses 41 tertiary-ranked holotype genes. All 50-mer oligonucleotide cry probes were designed based on multiple cry gene sequence alignments using MULTALIN (17). All potential probes were verified for their specificity using nucleotide database query software Fasta3 (European Bioinformatics Institute, Cambridge, United Kingdom [http://www.ebi.ac.uk/fasta33/nucleotide.html]). Lack of potential hairpin formation within each probe was verified using mFold v.3.0 software (44). Based on oligonucleotide design criteria (26), 101 oligonucleotides were synthesized (IDT, Coralville, IA) for printing on the microarray. The melting point temperature (T_m) of each probe, as calculated according to the formula optimized for DIG Easy Hyb digoxigenin buffer, where T_m = 49.82 + 0.41 × (%G+C) − 600/probe length in base pairs, was kept between 50 and 58°C. A complete list of all printed oligonucleotide probe sequences is available in the supplemental material (Table S1).

Amplification and labeling of target DNA.

Different target DNAs used in our study underwent one-step linear amplification and were simultaneously labeled using a modified BioPrime DNA-labeling system (Invitrogen, Burlington, Ontario, Canada). Briefly, 22 μl of crude genomic DNA extract was mixed with 20 μl of random-octamers 2.5× solution. It was incubated 5 min at 95°C and immediately cooled on ice for 5 min. Five μl of a mixture containing 1.2 mM of dATP, dGTP, dTTP, 0.6 mM dCTP, 2 μl of 1 mM Cy5-dCTP (PerkinElmer, Wellesley, MA), and 1 μl of Klenow fragment at 40 U/μl were added and the entire mixture incubated at 37°C during 4 h. The reaction was terminated by adding 5 μl of 0.5 M EDTA, pH 8.0, and the unincorporated dye was eliminated by using a QIAquick purification kit (Qiagen, Mississauga, Ontario, Canada). Labeled target DNA was eluted in 1 mM Tris-Cl (pH 8.0)-0.1 mM EDTA (pH 8.0). All subsequent steps involving labeled DNA were carried out with minimal exposure to direct light. Quantification of Cy5 incorporation was done by scanning the DNA sample from 200 to 700 nm and subsequently inputting the data into the web-based cyanine dye percent incorporation calculator found at http://www.pangloss.com/seidel/Protocols/percent_inc.html.

Printing and processing of microarrays.

Lyophilized oligonucleotides were resuspended to a concentration of 250 pmol/μl in ultrapure water and stored at −20°C. For printing, oligonucleotide concentrations were adjusted to 25 pmol/μl in 50% dimethyl sulfoxide-0.05% sodium dodecyl sulfate (SDS) and 10 μl of each sample transferred to a 384-well microplate. To facilitate the analysis of scanned images, all probes identifying primary, secondary, and ternary cry gene ranks were clustered separately as illustrated in the microarray printing key shown in Fig. 1 to facilitate a quick visual identification of any given cry gene type. Each oligonucleotide was spotted in triplicate onto CMT-GAPS II glass slides (Corning Canada, Whitby, Ontario, Canada) with a Virtek ChipWriter (Virtek Vision International, Waterloo, Ontario, Canada) equipped with SMP3 pins (TeleChem International, Sunnyvale, CA). After printing, the DNA was immobilized to the slides by UV cross-linking at 1,200 μJ (UV Stratalinker 1800; Stratagene, La Jolla, CA), followed by heating at 80°C for 4 h. Printed slides were stored in the dark at room temperature until needed.

FIG. 1.

Schematic representation for the cryArray showing the relative position of all oligonucleotide probes. Each keyword represents a unique probe printed in triplicate spots, and each gray block represents all probes that must be positive if a given gene is present. The incompletely separated blocks surrounding cry1Ib/c #105 and cry1Ib #081 reflect the common cry1Ib overlap between the two probes. Primary-ranked probes other than cry1 are grouped in the lower left quadrant, and secondary-ranked cry1 probes are grouped in the lower right.

Microarray hybridizations and analysis.

Printed arrays were prehybridized with 12 μl of solution prepared by mixing 13 μl DIG Easy Hyb buffer (Roche Diagnostics, Laval, Quebec, Canada), 3 μl of 10% bovine serum albumin, and 2 μl of a denatured 10 mg/ml salmon sperm DNA solution (Invitrogen, Burlington, Ontario, Canada) at 37°C for 2 h. After prehybridization, the coverslips were removed by dipping the slide in 0.1× SSC (15 mM NaCl-1.5 mM trisodium citrate, pH 7.0) and the slides dried by centrifugation in 50-ml conical plastic tubes at 600 × g for 5 min. Before hybridization, 750 ng of labeled target DNA was combined with 3 μg of salmon sperm DNA and it was evaporated and brought to a final volume of 2 μl. After addition of 10.5 μl of DIG Easy Hyb buffer, the solution was incubated for 5 min at 95°C and immediately cooled on ice for 2 min. Subsequently, 11 μl was transferred to the microarray for hybridization in Corning slide hybridization chambers containing 50 μl of water to maintain hydration and submerged overnight in a 47°C water bath. Both prehybridizations and hybridizations were carried out with the solution kept under a 12- by 12-mm coverslip.

After hybridization, the coverslips were removed by dipping in 1× SSC-0.1% SDS and the slide incubated in the same solution for 10 min at 42°C with light agitation. The slide was then washed once in 0.1× SSC-0.1% SDS and then twice in 0.1× SSC at room temperature for 10 min. Finally, the slide was rinsed by dipping it six times in 0.05× SSC and dried as described previously. The slides were scanned between 90 and 100% laser power and at 5-μm resolution with a ScanArray Lite fluorescent microarray reader (Canberra Packard Canada, Montreal, Quebec, Canada). Since most positive hybridization signals were relatively saturated, the presence of cry genes was assessed by visual inspection of the scanned image. In the case of weak hybridization spots, a spot was considered positive if the signal-to-noise (background) ratio was >4. Each experiment was performed in triplicate.

PCR validation.

The results obtained by microarray hybridizations were compared against PCR amplification results. Primer pairs, specific to cry genes under scrutiny, with sequences different from those used as microarray probes, can be found in the supplemental material (Table S2). Annealing temperatures used in PCR correspond to the formula T_m − 5°C (salt adjusted for 50 mM Na⁺). Amplifications were performed in 50-μl reaction volumes with 10 mM Tris-HCl (pH 9.0), 1.5 mM MgCl₂, 50 mM KCl, 200 μM (each) deoxynucleoside triphosphates, 0.5 μM (each) primers, and 2.5 U of Taq DNA polymerase (Pharmacia). The temperature program consisted of an initial 3 min at 94°C, during which the genomic DNA template (1 to10 ng) was added, followed by 35 cycles (35 s) at 94°C and 40 s at the calculated annealing temperature and a 1-min fragment extension at 72°C. Each PCR run was terminated by a final 7-min extension at 72°C. The amplicons were resolved on 1% (wt/vol) agarose gel and revealed by ethidium bromide staining.

RESULTS

Assessment of the cry1 gene content in known B. thuringiensis strains.

Hybridizations performed with 750 ng (5 ng/mm²) of Cy5-labeled amplified genomic DNA gave clear results producing strong fluorescent signals (Fig. 2). To confirm immobilized probe specificity, initial hybridizations were carried out using well-characterized laboratory or commercial B. thuringiensis strains. The array was constructed so that the presence of any given gene was confirmed only if all the secondary-rank probe(s) and all higher-rank probes targeting different regions within the gene produced positive hybridization signals. Strain HD-73, generally considered a single-gene strain (1), was hybridized to cryArray. Although seven probes for HD-73 were found positive, only two probe groupings, two cry1A secondary-rank and three cry1Ac tertiary-specific probes, were positive (Fig. 2A), indicating the presence of a cry1Ac gene. Since only one of three specific probes for either cry1Ab (#030) or cry1Ae (#097) was positive, these genes were not considered present. Hybridizations done with a strain of HD-73 cured from its cry gene-bearing plasmid (33) resulted in a complete loss of all hybridization signals (data not shown). A B. thuringiensis subsp. kenyae strain obtained from a strain collection at NRCan, Sault St. Marie, Ontario, Canada, which contains a cry1E gene is another strain considered to possess a single cry gene (32). Hybridization with genomic DNA from this strain produced six positive signals, of which only the single cry1E secondary- and both tertiary-ranked probes for cry1Ea were positive (Fig. 2B). Since only two of three cry1Ab and one of three cry1Ae probes were positive and since none of the cry1A secondary-ranked probes were positive, these genes were not considered present.

FIG. 2.

Scanned images of microarrays hybridized with Cy5-labeled genomic DNAs from different known B. thuringiensis strains. Probes belonging to the same cry gene target are grouped inside the individual squares.

In complex strains (having more than one cry gene), hybridization patterns also correlated consistently with the known cry gene content. A B. thuringiensis subsp. kurstaki HD-1 strain (4D1) confirmed the presence of three cry1A genes (cry1Aa, cry1Ab, and cry1Ac) and a cry1Ia gene from the primary to the tertiary rank (Fig. 2C). The presence of a cry2Aa gene in this strain was also confirmed. Furthermore, hybridization with HD-1 genomic DNA produced from an industrial preparation of Thuricide had positive hybridization signals only for the cry1Aa, cry1Ac, and cry1Ia gene probes (Fig. 2D). Two of the three cry1Ab probes did not hybridize with Thuricide DNA, and thus the gene was scored as absent, a result which agrees with previously published work (30). Consistent with the known cry gene content of B. thuringiensis subsp. israelensis strain HD-500 (5), primary-ranked cry4 but not cry1 genes were detected after hybridization (Fig. 2E). However, only one of the two primary-ranked cry11 probes produced a positive signal. PCR amplification with cry11A-specific primers (Table 1) showed that the gene was present in HD-500 as expected (data not shown), suggesting that the probe cry11#138 was somehow defective. Hybridization with genomic DNA from B. thuringiensis subsp. aizawai HD-133 (Fig. 2F) supported the presence of cry1Aa, cry1Ab, cry1Ca, cry1Da, and cry1Ia genes as has been described previously (31). A positive signal for the primary-ranked probes of cry2 and cry9 genes were found, suggesting their presence in this strain, which is consistent with PCR results published by an independent group (3).

TABLE 1.

PCR validation of cry gene presence identified by microarray hybridization

Oligonucleotide primer sets	Gene target	IB217		IB360		IB585		IB059		IB429		IB031
		PCR	Array	PCR	Array	PCR	Array	PCR	Array	PCR	Array	PCR	Array
Aa-1020F/Aa-1375R	cry1Aa	NDa	−	ND	−	ND	−	ND	−	ND	−	ND	−
CJ1/CJ2b	cry1Aa cry1Ad	+	NAc	+	NA	−	NA	−	NA	+	NA	−	NA
Ab-924F/Ab-2405R	cry1Ab	−	−	+	+	ND	−	ND	−	+	+	ND	−
CJ4/CJ5	cry1Ab cry1Ac	−	−	+	NA	−	NA	−	NA	+	NA	−	NA
CJ6/CJ7	cry1Ac	−	−	−	−	−	−	−	−	−	−	−	−
Ad-995F/Ad-1201R	cry1Ad	+	+	+	+	ND	−	ND	−	+	+	−	−
Ae-856F/Ae-1109R	cry1Ae	−	+	−	+	ND	−	ND	−	−	−	ND	−
CJ8/CJ9	cry1Ba	−	−	−	−	−	−	?d	+	−	−	−	−
CJ10/CJ11	cry1C	+	+	+	+	−	−	−	−	−	−	−	−
Ca-1100F/Ca-1581R	cry1Ca	+	+	+	+	ND	−	ND	−	ND	−	ND	−
CJ12/CJ13	cry1D	+	+	+	+	−	−	−	−	−	−	−	−
Da-54F/Da-517R	cry1Da	+	+	+	+	ND	−	ND	−	−	−	ND	−
CJ16/CJ17	cry1F	+	+	+	+	−	−	−	−	+	+	−	−
Fa-318F/Fa-1416R	cry1Fa	+	+	+	+	ND	−	ND	−	+	+	ND	−
CJ18/CJ19	cry1G	−	−	−	−	−	−	−	−	−	−	−	−
V(+)/V(−)	cry1I	+	+	+	+	−	−	−	−	+	+	ND	−
la-841F/Ia-1149R	cry1Ia	−	−	−	−	−	−	−	−	−	−	ND	−
Ib-620F/Ib-1220R	cry1Ib	−	−	−	−	−	−	−	−	−	−	ND	−
Ic-944F/Ic-1390R	cry1Ic	−	−	−	−	−	−	−	−	−	−	ND	−
cry1Jdir/cry1Jrev	cry1Ja	−	−	−	−	+	+	−	−	−	−	−	−
CJII120/CJII121	cry3	−	−	−	−	−	−	−	−	−	−	?	−
Cry2gral	cry2A	+	+	+	+	−	−	−	−	+	+	−	−
Cry4A spe	cry4A	−	−	−	−	−	−	−	−	−	−	−	−
Cry4B spe	cry4B	−	−	−	−	−	−	−	−	−	−	−	−
C9-2975F/C9-3509R	cry9	+	+	+	+	ND	−	ND	−	+	+	ND	−
Cry11 gral	cry11a/cry11b	−	NA	−	NA	−	NA	−	NA	−	NA	−	NA

^aND, not done.

^bOther primer sets were reported elsewhere, as follows: V(+) /V(−) (23); Cry1 I gral, Cry2gral, Cry4A spe, and Cry4B (22); CJ1 to CJ9 (13); and CJ14/CJ5 and CJII120/CJII121 (15).

^cNA, not applicable.

^d?, positive result in PCR, but amplicon size deviates from predicted size.

Assessment of the cry1 gene content in unknown B. thuringiensis strains.

A variety of B. thuringiensis strains from a Mexican culture collection (10) whose cry contents are essentially unknown were also examined. Two strains, IB217 and IB360, generated complex but somewhat similar patterns (Fig. 3A and 3B). At the primary level, cry1, cry2, and cry9 genes were detected. Within the cry1 rank, cry1A, cry1C, cry1D, cry1F, and cry1I genes were found to be present. At the tertiary level, the two strains were found to be different. Although both strains have genes for cry1Ad, cry1Ae, cry1Ca, cry1Da, and cry1Fa, a ninth gene, cry1Ab, was found in IB360 but was absent in IB217. In contrast to a positive signal in both strains for a secondary-ranked cry1I gene, the identity of the cry1I gene could not be confirmed at the tertiary level, as no signals were observed for cry1Ib or cry1Ic genes and only one of two cry1Ia probes was positive. Another complex pattern was observed with strain IB429 (Fig. 3C), which, unlike IB217 and IB360, lacks the cry1C and cry1D genes but, like IB360, has the cry1 genes cry1Ab, cry1Ad, and cry1Fa, as well as a cry2 and a cry9 gene. Again, the identity of the cry1I gene could not be confirmed at the tertiary level since, as observed with IB217 and IB360, only the cry1Ia#107 probe but not the cry1Ia#078 was positive.

FIG. 3.

Scanned images of microarrays hybridized with Cy5-labeled genomic DNAs from different unknown or partially characterized B. thuringiensis strains. Probes belonging to the same cry gene target are grouped inside the individual squares.

The other three strains examined, IB059, IB585, and IB031, possessed fewer cry genes as detected by the cryArray. Hybridization of strain IB059 revealed the presence of a single cry1 gene (cry1Ba). Although various probes hybridized to the genomic DNA of IB585 and IB031, neither strain hybridized to all probes within a group to confirm the presence of a particular gene.

PCR validation of microarray data.

For the B. thuringiensis strains with partially characterized cry gene contents, the presence of microarray-detected cry genes was validated by gene-specific PCR (Table 1). Although cry1Ae was positive with all three probes in microarray hybidizations, cry1Ae-specific PCR performed on genomic DNA from IB217 and IB360 strains did not produce the expected 265-bp band, suggesting that a new variant of this gene was present in both strains. When the general cry1I family primers V(+) and V(−) were used, the expected 1,137-bp band was amplified in IB217, IB360, and IB429 strains, confirming the presence of a cry1I gene. However, performing cry1Ia-, cry1Ib-, and cry1Ic-specific PCRs on the cry1I-positive strains produced negative results, thus confirming in our microarray data that a new variant of cry1I is present in these strains. Direct confirmation of a new cry1I variant was obtained by amplifying a PCR fragment using two primers designed from two consensus sequences obtained by aligning known cry1I genes. Two probes, AF (5′-ATGAAACTAAAGAATCAAG) and ER (5′-TTATAGTCTAAGTCCTCTC), produced a 1.8-kb PCR fragment from strains IB360 and IB429 (data not shown) which was sequenced (GenBank accession no. AY959880). Both PCR amplicons (1760 nucleotides) were identical and were most similar to a cry1Ia gene showing a total of 46 nucleotide mismatches (2.6%), resulting in 19 amino acid differences (data not shown). The sequences related to the six cry1I-related immobilized 50-mers on the cryArray are shown in Fig. 4. Consistent with the array data, the one tertiary and two secondary positive probes, cry1Ia#107, cry1I#142, and cry1I#143, were 98 to 100% homologous to the cry1I gene from IB360 and IB429. However, the three tertiary probes, cry1Ia#078, cry1Ib#081, and cry1Ib/c#105, showed much larger sequence variations (between 14 and 20%), thus accounting for their lack of hybridization to the IB360 or IB429 cry1I PCR fragment.

FIG. 4.

Sequence comparison of the cry1I amplicon from B. thuringiensis to cry1I-specific immobilized cryArray probes. Differences between the cry1I gene amplified from strain IB360 and the immobilized specific secondary or tertiary cry1I -derived gene sequence are shown in large, boldface capital letters.

For all other positive microarray hybridization results, PCR yielded the expected bands with one exception. Although the cryArray was positive for cry1Ba for strain IB059, subsequent PCR revealed that the amplicon produced was of a size different from that normally expected, indicating a new variant of this gene. In contrast to IB059, IB031 was negative for a cry3 gene on the microarray but was found positive by PCR. However, rather than the cryArray producing a false negative, the PCR amplicon size was found to be different from the predicted size, suggesting yet another gene variant.

DISCUSSION

Within the 46 current primary Cry toxin ranks, nearly half of the 280 proteins registered in the databank belong to the Cry1 group. Although at the primary rank, these toxins share some amino acid similarity, this similarity increases at each subdivision until the quaternary rank where, in most cases, these proteins may be identical or separated by only a few amino acids. Paralleling these subdivisions, the genes encoding these Cry proteins also increase in homology at each subdivision. The goal of this study was to determine if the parallel processing power of DNA microarrays could be exploited to identify the cry gene content of individual B. thuringiensis strains at a relatively high rank. To minimize false-positive results, a consentaneous approach was incorporated where multiple gene probes must produce positive hybridizations to confirm the presence of that gene. Since cry genes at increasingly higher ranks possess similar sequences, it becomes increasingly difficult to design specific cry gene probes to differentiate between tertiary- and quaternary-level genes inasmuch as five or more mismatches distributed over a 50-base sequence are required to prevent cross-hybridization (26). For example, since one cannot find a specific probe for cry1Ic, two probes were designed. One probe, cry1Ib/c#105, is common to both cry1Ib and cry1Ic. If the target is cry1Ib, both probes #105 and cry1Ib#081 should be positive. However, if the target is cry1Ic, only #105 will be positive. It is clear that our design can be further improved by extending the number of consentient gene-specific probes, whenever possible, through a reduction in probe size, and consequently, more specific oligonucleotides per gene could potentially be identified. Considering that our current protocol produces strong positive fluorescent signals, the concomitant loss of signal intensity inherent with a reduction in probe size could be offset by an increase in labeled target DNA concentration.

Although this study focused on cry1 toxins, some probes for the cry2, cry3, cry4, cry9, and cry11 gene groups at the primary rank were also included. The expected presence of cry4 genes in the B. thuringiensis subsp. israelensis strain HD-500 (5) was supported by hybridization results. However, it is not clear why only the primary-ranked cry11#139 and not the cry11#138 probe produced a positive hybridization signal with HD-500 which harbors a cry11Aa gene. The cry11 probe (cry11#139) has between one and three mismatches with all cry11 genes and should hybridize similar to a homologous probe. The cry11#138 probe is specific to known cry11Aa and cry11Ba genes containing two mismatches to either gene and five mismatches with cry11Bb.

A potential disadvantage of using DNA microarray technology to assess cry gene content is similar to that when using PCR-based techniques, i.e., that probes (or primers) are designed against known genes, hence the potential for overlooking novel genes. However, by including secondary-ranked probes, a positive hybridization in the absence of any signal for the tertiary-ranked probes (or alternatively, a weak tertiary signal) indicates the presence of a novel gene in that primary-rank group. An example of the former was shown with the strains IB217, IB360, and IB429. In these relatively similar strains, hybridization signals were positive with both general cry1I-family probes (cry1I#142 and cry1I#143). No specific tertiary probes were positive, although one of two cry1Ia probes (cry1Ia#107) did produce a positive signal. Our data suggest that the specific cry1Ia, cry1Ib, or cry1Ic genes were not present, indicating that a new cry1I gene is present. Amplification and sequencing of the cry1I signal in IB360 indeed revealed the presence of a cry1Ia variant. High sequence variability at all the probe sites except cry1Ia#107 and the two cry1I general primers was consistent with the microarray data and underscores the ability of the cryArray to detect new genes. In the case of weaker positive hybridization spots, two explanations are possible. The first is that the signal is comparatively weak since the gene is either chromosomal or on a single-copy plasmid. In either case, the gene dosage will be much less than that for genes on higher-copy-number plasmids. Secondly, depending on probe placement, a weaker hybridization could suggest a partial hybridization to that particular probe, suggesting that the genes are nearly homologous 100%.

As is the case with most discovery tools, the combination of several techniques, for example, DNA microarrays with PCR, should provide a more powerful means of cry gene detection and analysis. As shown in this study using several B. thuringiensis strains with unknown or partially characterized cry1 gene content, our cryArray detected a cry1Ad gene that was shown elsewhere to be negative by PCR (10). By designing new cry1Aa and cry1Ad gene-targeting PCR primers, strain IB217 has now been shown to possess a cry1Ad gene but lack a cry1Aa gene. Since cry1Ae-specific PCRs were negative for strains IB217 and IB360, it was assumed that these strains lacked the target gene. Based on microarray hybridizations, however, cry1Ae was indeed found to be present in these strains. Because the cry1Ae#098 probe has only three mismatches to the equivalent cry1Fa region, cross hybridization to cry1Fa was expected. However, since cry1Ae#097 has at least six mismatches to any other cry1 gene and cry1Ae#100 has five mismatches to cry1Ja, six to cry1Ad, and more to other cry1 genes, no cross-hybridization is expected for these probes, indicating that cry1Ae was probably a genuine result. Our data also revealed various instances where the PCR data suggested the presence of novel cry genes undetected by cryArray. In a prior publication involving IB031 (15), PCR results showed the presence of two potentially novel genes. A cry1 gene was amplified using a forward cry1E primer and a reverse cry1G primer, yet neither gene was detected in the cryArray, presumably due to this recombinant cry1E/1G gene having deleted the region complementary to our specific cry1E or cry1G array probes. Also in this strain, a cry3C PCR fragment of an unexpected size was produced, yet none of the three primary-ranked cry3 array probes was positive after hybridization. Interestingly, earlier immunoanalysis of the protein crystal indicated the presence of a single protein that could cross-react with either a Cry1E or a Cry3A monoclonal antibody (15). When combined, these results suggest that a novel cry3/cry1E-like gene may be present in this strain. In a different example, even though a positive cryArray signal was observed for a cry1Ba gene on strain IB059, PCR results showed an anomalous amplicon size.

Although numerous cry genes can be found within a single strain, they may express and crystallize proteins at different levels or, alternatively, the genes may not be expressed at all (30, 31). With strain HD-133, it has been shown that the cry1Aa gene (positively identified by the cryArray) does not express due to an insertion sequence element in the coding sequence (31). This potential disadvantage in detecting cry gene expression in the current chip prototype could be circumvented by including a complementary (antisense) probe oligonucleotide in order to detect both the gene and the gene-specific mRNA simultaneously.

In summary, our limited model study shows that DNA microarrays have good potential for rapid cry gene screening in new B. thuringiensis isolates presenting interesting pesticidal activity. Our cryArray can discriminate between highly similar genes with the production of unusual secondary- and tertiary-rank hybridization patterns as an indicator of novel cry genes.