DNA Nanomachine (DNM) Biplex Assay for Differentiating Bacillus cereus Species

Conventional methods for the detection and differentiation of Bacillus cereus group species have drawbacks mostly due to the complexity of genetic discrimination between the Bacillus cereus species. Here, we describe a simple and straightforward assay based on the detected unamplified bacterial 16S rRNA by DNA nanomachine (DNM). The assay uses a universal fluorescent reporter and four all-DNA binding fragments, three of which are responsible for “opening up” the folded rRNA while the fourth stand is responsible for detecting single nucleotide variation (SNV) with high selectivity. Binding of the DNM to 16S rRNA results in the formation of the 10–23 deoxyribozyme catalytic core that cleaves the fluorescent reporter and produces a signal, which is amplified over time due to catalytic turnover. This developed biplex assay enables the detection of B. thuringiensis 16S rRNA at fluorescein and B. mycoides at Cy5 channels with a limit of detection of 30 × 103 and 35 × 103 CFU/mL, respectively, after 1.5 h with a hands-on time of ~10 min. The new assay may simplify the analysis of biological RNA samples and might be useful for environmental monitoring as a simple and inexpensive alternative to amplification-based nucleic acid analysis. The DNM proposed here may become an advantageous tool for detecting SNV in clinically significant DNA or RNA samples and can easily differentiate SNV under broadly variable experimental conditions and without prior amplification.


Introduction
Among many bacteria, the Bacillus cereus group stands out for its diversity and ubiquitous spread along the biosphere. Certain species are pathogenic with foodborne and infectious potential [1][2][3]. For instance, Bacillus cereus (B. cereus) is a well-recognized pathogen that causes two different forms of food poisoning, including emesis and diarrhea [4]. In contrast to B. cereus, Bacillus thuringiensis (B. thuringiensis) is categorized as non-harmful for humans and exploited as a pesticide targeting insects [5][6][7]. Nonetheless, B. thuringiensis has also been reported in the context of foodborne outbreaks [8]. B. cereus, B. thuringiensis, Bacillus anthracis, and Bacillus mycoides (B. mycoides) are classified collectively as members of the B. cereus group based on their close similarity. The actual frequency of B. thuringiensis in foodborne outbreaks is still unknown due to the complexity of the genetic discrimination between the Bacillus cereus species. Genomes of B. cereus, B. thuringiensis, and B. mycoides are especially infused with each other [9][10][11]. DNA strand Dza and Dzb bind RNA analyte and form a catalytic core that cleaves fluorophore and quencher labelled F-sub. The design of B. mycoides-specific BiDz probe differs from the design of B. thuringiensis-specific BiDz probe in the reporter substrate used. The Cy-sub reporter substrate was used with the B. mycoides-specific BiDz probe. The difference between F-sub and Cy-sub sequence is shown in Table S1.
In this study, we demonstrate the general applicability of the DNM approach in targeting and SNV discrimination in another scoop of nucleic acids, natural 16S rRNA without prior amplification. We demonstrate the ability of the DNM approach to outcompete the intramolecular base pairing of the RNA target and to "open up" the targeted fragment. We developed six variations of DNA nanosensors with a different number of DNA-binding arms for the detection and differentiation of B. thuringiensis and B. mycoides. For each strain, three DNA nanosensors were developed: BiDz sensor (Figure 1), three-armed DNA nanomachine (B. thuringiensis-specific DNM3 and B. mycoides-specific DNM3, in Figure 2 A,B, top), and four-armed DNA nanomachine (B. thuringiensis-specific DNM4 and B. mycoides-specific DNM4, in Figure 2A,B, bottom), attributable to one or two additional unwinding arms in comparison to the BiDz sensor, respectively. Two variants of the reporter substrate were used: F-sub was used for detecting B. thuringiensis and Cy-sub was used for B. mycoides (see Table S1 for sequences). DNA strand Dz a and Dz b bind RNA analyte and form a catalytic core that cleaves fluorophore and quencher labelled F-sub. The design of B. mycoides-specific BiDz probe differs from the design of B. thuringiensis-specific BiDz probe in the reporter substrate used. The Cy-sub reporter substrate was used with the B. mycoides-specific BiDz probe. The difference between F-sub and Cy-sub sequence is shown in Table S1.
In this study, we demonstrate the general applicability of the DNM approach in targeting and SNV discrimination in another scoop of nucleic acids, natural 16S rRNA without prior amplification. We demonstrate the ability of the DNM approach to outcompete the intramolecular base pairing of the RNA target and to "open up" the targeted fragment. We developed six variations of DNA nanosensors with a different number of DNA-binding arms for the detection and differentiation of B. thuringiensis and B. mycoides. For each strain, three DNA nanosensors were developed: BiDz sensor (Figure 1), three-armed DNA nanomachine (B. thuringiensis-specific DNM3 and B. mycoides-specific DNM3, in Figure 2A,B, top), and four-armed DNA nanomachine (B. thuringiensis-specific DNM4 and B. mycoides-specific DNM4, in Figure 2A,B, bottom), attributable to one or two additional unwinding arms in comparison to the BiDz sensor, respectively. Two variants of the reporter substrate were used: F-sub was used for detecting B. thuringiensis and Cy-sub was used for B. mycoides (see Table S1 for sequences). In this study, we demonstrate the general applicability of the DNM approach in targeting and SNV discrimination in another scoop of nucleic acids, natural 16S rRNA without prior amplification. We demonstrate the ability of the DNM approach to outcompete the intramolecular base pairing of the RNA target and to "open up" the targeted fragment. We developed six variations of DNA nanosensors with a different number of DNA-binding arms for the detection and differentiation of B. thuringiensis and B. mycoides. For each strain, three DNA nanosensors were developed: BiDz sensor (Figure 1), three-armed DNA nanomachine (B. thuringiensis-specific DNM3 and B. mycoides-specific DNM3, in Figure 2 A,B, top), and four-armed DNA nanomachine (B. thuringiensis-specific DNM4 and B. mycoides-specific DNM4, in Figure 2A,B, bottom), attributable to one or two additional unwinding arms in comparison to the BiDz sensor, respectively. Two variants of the reporter substrate were used: F-sub was used for detecting B. thuringiensis and Cy-sub was used for B. mycoides (see Table S1 for sequences).   Table S1 for sequences). The correct assembly of DNM3 and DNM4 are indicated by the arrows.

Design of DNA-Nanosensors and Preliminary Efficiency Tests
Initially, BiDz sensors based on 10-23 deoxyribozyme were used, an analogue to that used earlier for detection of E. coli cells [45], quantification of mutant ribosomal RNA [48], and analysis of Mycobacterium tuberculosis RNA [53]. The sequences of Dza, Dzb, and the reporter substrates used in this study are listed in Table S1. Two BiDz sensors targeting B. thuringiensis ( Figure 2) and B. mycoides 16S rRNA were designed. The sequences of 16S rRNA were obtained from the work of Ash et al. [54] and aligned ( Figure S1). The sensor B. thuringiensis-specific BiDz was complementary to B. thuringiensis 16S rRNA (position  Table S1 for sequences). The correct assembly of DNM3 and DNM4 are indicated by the arrows.

Design of DNA-Nanosensors and Preliminary Efficiency Tests
Initially, BiDz sensors based on 10-23 deoxyribozyme were used, an analogue to that used earlier for detection of E. coli cells [45], quantification of mutant ribosomal RNA [48], and analysis of Mycobacterium tuberculosis RNA [53]. The sequences of Dz a , Dz b , and the reporter substrates used in this study are listed in Table S1. Two BiDz sensors targeting B. thuringiensis ( Figure 2) and B. mycoides 16S rRNA were designed. The sequences of 16S rRNA were obtained from the work of Ash et al. [54] and aligned ( Figure S1). The sensor B. thuringiensis-specific BiDz was complementary to B. thuringiensis 16S rRNA (position 172-203). The sensor B. mycoides-specific BiDz was complementary to B. mycoides 16S rRNA (position 169-206). The design of the sensors differs in the targeted SNV and the reporter that was used. Strands B. thuringiensis-specific Dz a and B. mycoides-specific Dz a were complementary to a fragment of 16S rRNA containing two SNVs (192 and 197), which are shown in bold sequence in Table 1. SNV 197 was used to differentiate between B. thuringiensis and B. mycoides, whereas SNS 192 was used to differentiate between two strains from B. cereus and Bacillus anthracis. The analyte-binding arm of the Dz a strand was designed to be short enough to form a stable hybrid only with the specific target, thus enabling the high selectivity of 16S rRNA recognition, according to the fundamental principles of the binary probe design [45]. The sensors were tested using the total RNA extracted from B. thuringiensis and B. mycoides ( Figure S1) and did not show any response above the background signal, both after 1 h (Figure in Section 2.4) and after 3 h of incubation ( Figures S2 and S3). This can be attributed to the fact that the RNA region complementary to B. thuringiensis-specific BiDz and B. mycoides-specific BiDz is involved in a stable stem-loop structure ( Figure S4).
To access the SNV site within the tight 16S rRNA structure, we designed the threeand four-armed DNM3 and DNM4, respectively. The B. thuringiensis-specific DNM4 is a four-armed machine, complementary to B. thuringiensis 16S rRNA (position 143-232) and B. thuringiensis-specific DNM3-thu, a three-armed machine for the same region. The B. mycoides-specific DNM4 is a four-armed nanomachine, complementary to B. mycoides 16S rRNA (position 143-232) and B. mycoides-specific DNM3, a three-armed machine complementary for the same region.

Characterization of BiDzs and DNMs
Initially, the DNA nanosensors were characterized using two 92 nt synthetic DNA oligonucleotides corresponding to 138-237 nt fragment of both B. thuringiensis and B. mycoides 16S rRNA (Table S1). The B. thuringiensis synthetic analyte folded into a stable secondary structure (∆G = −8.77 kcal/mol) ( Figure S5A) while the B. mycoides synthetic analyte folded into a more stable secondary structure (∆G = −9.35 kcal/mol) ( Figure S5B) under the assay conditions. The correct assembly of DNM3 and DNM4 were confirmed by electrophoresis in PAGE gel ( Figure 2C). It was demonstrated that all DNA nanosensors responded to the presence of the specific synthetic DNA analyte in a concentration-dependent manner ( Figure 3). The limit of detection (LOD) for the DNA nanosensors was found to be in the range of 20-110 pM as shown in Table 2 ( Figures S2 and S3).
These data indicate that the LOD for synthetic DNA analyte was only slightly improved by increasing the number of analyte-binding arms. The fluorescence signal at the background level was observed for all the DNA nanosensors targeting B. thuringiensis 16S rRNA in the presence of nonspecific B. mycoides synthetic analyte and vice versa. This is consistent with the high selectivity of the sensors reported earlier.  Table S2 and Table S3 contain data correspond to Figure 3A and Figure 3B, respectively. These data indicate that the LOD for synthetic DNA analyte was only slightly improved by increasing the number of analyte-binding arms. The fluorescence signal at the background level was observed for all the DNA nanosensors targeting B. thuringiensis 16S

Selectivity of the DNA-Nanosensors
The sequence of the targeted 16S rRNAs of the Bacillus cereus group species shows a similarity of >99% [54]. In fact, the 16S rRNA sequences of B. thuringiensis and B. mycoides differ from each other and from the sequences of B. cereus and B. anthracis by only four to nine nucleotides [54]. Since the mutations are scattered throughout the 16S rRNA sequence, the sensor should have single nucleotide selectivity. Splitting the functional roles between several components of the sensor allows for the fine tuning of the sensor's characteristics, including selectivity [35]. For example, to make the sensors more selective, it was necessary to shorten the analyte-binding arm of the Dz a strand from 19 to 14 nucleotides (nt) for the sensors targeting B. thuringiensis 16S rRNA and from 20 to 15 nt for the sensors targeting B. mycoides 16S rRNA, since the sensors with long analyte-binding arms could not reliably discriminate the perfectly complementary analyte from the analyte containing a single mismatch.
The selectivity of the truncated sensors was tested and adjusted using synthetic DNA oligonucleotide analytes mimicking the targeted fragment of 16S rRNAs (positions 138-237). It was shown that for all the sensors, the high fluorescent signal was observed only in the presence of the fully complementary analyte, while the analyte forming a singlebase mismatch with the sensor triggered the signal close to the background (Figure 4), even when it was tested at saturation levels. These data are in agreement with the high selectivity of binary probes reported earlier [45]. The DNMs successfully detected the corresponding total RNA extracted from B. thuringiensis, B. mycoides, and B. cereus cultures. B. thuringiensis-specific DNM4 emits a signal only in the presence of B. thuringiensis RNA, and B. mycoides-specific DNM4 in the presence of B. mycoides RNA. Neither B. thuringiensisspecific nor B. mycoides-specific DNM4 detected the B. cereus RNA (Figure 4).
taining a single mismatch.
The selectivity of the truncated sensors was tested and adjusted using synthetic DNA oligonucleotide analytes mimicking the targeted fragment of 16S rRNAs (positions 138-237). It was shown that for all the sensors, the high fluorescent signal was observed only in the presence of the fully complementary analyte, while the analyte forming a singlebase mismatch with the sensor triggered the signal close to the background (Figure 4), even when it was tested at saturation levels. These data are in agreement with the high selectivity of binary probes reported earlier [45].   Table S4 and  Table S5 contain data correspond to Figure 4A and Figure 4B, respectively.

Detection of the Bacterial 16S rRNAs Using the DNA-Nanosensors
The performance of B. thuringiensis-and B. mycoides-specific DNA nanosensors was evaluated in the presence of 10 ng of total RNA isolated from B. mycoides, B. thuringiensis, or B. cereus ( Figure S3). It was found that the BiDz sensors showed a signal as low as the background with the three different totals of RNA even after 3 h of incubation.
A 1.5-fold increase in the fluorescence was observed for the DNM3 but only after 3 h

Detection of the Bacterial 16S rRNAs Using the DNA-Nanosensors
The performance of B. thuringiensisand B. mycoides-specific DNA nanosensors was evaluated in the presence of 10 ng of total RNA isolated from B. mycoides, B. thuringiensis, or B. cereus ( Figure S3). It was found that the BiDz sensors showed a signal as low as the background with the three different totals of RNA even after 3 h of incubation.
A 1.5-fold increase in the fluorescence was observed for the DNM3 but only after 3 h of incubation ( Figures S6 and S7), whereas the DNM4 showed a 1.5-fold increase in the fluorescence in the presence of the target RNA after 1 h of incubation. The B. mycoidesand B. thuringiensis-specific DNA nanosensors showed fluorescence at the background level in the presence of non-target RNA (B. cereus) ( Figure 5). The LOD of the DNM4 was calculated since it showed the best performance out of the three variants. It was found to be in the range of 1-1.5 ng, which corresponds to~5 × 10 8 of 16S rRNA molecules of the target 16S rRNA after 1 h of incubation ( Figure S8).   Table S6 and Table S7 contain data correspond to Figure 5A and Figure 5B, respectively.

Detection of the 16S rRNAs Using the B. thuringiensis and B. mycoides in Crude Lysates
Relying on the high selectivity of the DNM4 sensor, we used it with crude c without isolating target RNAs. We tested serial dilutions of a B. thuringiensis a coides cell culture (OD600 of 1-1.2) to demonstrate the performance of the DNM in comparison with the BiDz sensor ( Figure 6). The sensors detected 2 × 10 4 cell

Detection of the 16S rRNAs Using the B. thuringiensis and B. mycoides in Crude Cell Lysates
Relying on the high selectivity of the DNM4 sensor, we used it with crude cell lysates without isolating target RNAs. We tested serial dilutions of a B. thuringiensis and B. mycoides cell culture (OD600 of 1-1.2) to demonstrate the performance of the DNM4 sensor in comparison with the BiDz sensor ( Figure 6). The sensors detected 2 × 10 4 cells with S/B of~3, which is good enough for distinguishing a true signal from the noise in fluorescent assays as was discussed earlier [43].
The fluorescence response of the sensor's DNM4 increased with increased cell numbers, reaching saturation when 8 × 10 8 cells per reaction volume or more were used. At the same time, the BiDz sensor showed a response as low as the background (Figure 6, light grey bars). The response of B. thuringiensis-specific DNM4 to nonspecific B. mycoides and B. cereus cells were close to the background ( Figure 4A), which also verifies the excellent selectivity of DNM. The response of B. mycoides-specific DNM4 to nonspecific B. thuringiensis and B. cereus cells was close to the background ( Figure 4B). The LOD of the DNM4 sensor was found to be 30 × 10 3 CFU/mL for B. thuringiensis and 35 × 10 3 CFU/mL for B. mycoides (Figures S9 and S10).
B. cereus cells were close to the background ( Figure 4A), which also verifies the ex selectivity of DNM. The response of B. mycoides-specific DNM4 to nonspecific B. giensis and B. cereus cells was close to the background ( Figure 4B). The LOD of the sensor was found to be 30 × 10 3 CFU/mL for B. thuringiensis and 35 × 10 3 CFU/mL mycoides (Figures S9 and S10).   Table S8 and Table S9 contain data correspond to Figure 6A and Figure

Discussion
BiDz technology was introduced in 2007 as a protein-free signal amplification technique with a potential for amplification-free detection of nucleic acids [42]. The major advantage of this technology was the unprecedented high selectivity of SNV detection due to the split design [45]. Since then, achieving the lowest possible LOD for BiDz has been an important goal in developing the technology [53]. With the discovery of 10-23 BiDz by Mokany et at. [43], the technology received attention due to achieving an impressive 5 pM LOD after 3 h of incubation at 55 • C with short synthetic DNA analytes. This LOD was reduced to 2 pM within a 2 h timeframe by the Dz cascade for signal amplification [55]. BiDz equipped with a DNA antenna tile for F-sub delivery enabled a LOD of~5 pM after only 1 h of incubation [56]. BiDz probe has been demonstrated to be efficient in detecting short DNA analytes [57][58][59], but the performance is less in case of long biological RNA. Being the most sensitive probe among protein enzyme-free approaches, BiDz is not sensitive enough to detect a majority of biological DNA or RNA analytes without amplification, including the mRNA responsible for bacteria drug resistance. Indeed, while the PCR-based method can detect down to 1-10 analytes due to exponential amplification, DNM4 detected only~5 × 10 8 of 16S rRNA molecules ( Figure S8) in this study due to just linear signal amplification. One important exception for this limitation is the detection of 16S rRNA, which is present in thousands of copies in a single bacteria cell. We believe that developing an amplification-free technique for detecting bacteria based on 16S rRNA-specific BiDz is feasible.
However, analysis of such a long biologically folded RNA as 16S rRNA by traditional hybridization probes has always been a major challenge [60]. The main problem is that the targeted fragments are involved in an intramolecular base paring and, therefore, they are hardly accessible. Secondary structure in analytes has been found to significantly reduce probe-target duplex formation and slow hybridization kinetics [61][62][63]. Amplification of such RNA samples requires a PCR instrument and specialized diagnostic laboratory, which makes the analysis procedure costly and time consuming. To enable amplificationfree detection of 16S rRNA, we equipped BiDz with additional RNA-binding arms. This modification resulted in a thermodynamically favorable RNA/DNM complex, which led to the unfolding of the RNA secondary structure, thus reducing the time of the assay and achieving the lowest possible LOD. The strategy reported here can be further adapted from biplex to multiplex system targeting more than two analytes, since the adaptation would require only a new fluorogenic substrate and modification of the substrate-binding arms.
The use of the 16S rRNA sequence as a reliable marker for the discrimination of B. cereus group members has failed in previous studies [54,64] due to the high nucleotide conservation in the sequence among the species of this group. Detection and differentiation of B. cereus group species in routine diagnostics can be difficult and time-and reagent-consuming since current species' designations are linked to the specific phenotypic characteristic or the presence of species-specific genes. To the best of our knowledge, this is the first attempt to differentiate B. cereus species with an amplification-free approach. The DNM approach uses inexpensive reagents and equipment; it does not require perishable reagents (e.g., protein enzymes) and provides impressive selectivity towards SNV. Therefore, DNM technology promises to be less expensive and more deliverable to point-of-care settings than current gold diagnostic standard qPCR and new generation sequencing (NGS). DNM4 detected~1.5 ng of total RNA ( Figure S6), which corresponds to~5×10 8 16S rRNA molecules. The same sensor detected~30,000 colony-forming units in crude cell extract (Figures S9 and S10). Wei el al. [65] reported a RT-qPCR-based method for the detection of B. thuringiensis with an LOD that can reach down to 10 3 CFU/g of B. thuringiensis in spiked food samples. There is no well-established infectious dose of B. thuringiensis or B. cereus partly due to the large differences in the amount of enterotoxin produced by different strains [66]. However, counts ranging from 200 to 10 9 g −1 (or mL −1 ) B. cereus have been reported in affected food after food poisoning. Thus, the total infective dose seems to vary between 10 5 -10 8 viable cells or spores [67]. Despite this, further improvement in DNM sensitivity is needed. It can include using multiple DNMs for binding the same RNA molecule or adapting the antenna tile design for facilitated substrate delivery to the activated Dz core [56] or adding more analyte-binding arms. The DNMs proposed here can become an advantageous tool for detecting SNV in clinically significant DNA or RNA samples and can easily differentiate SNV under broadly variable experimental conditions [68,69].

Materials
The fluorescence susbstrates were purchased in DNA-synthesis, Moscow, Russia. Other oligonucleotides were purchased in Evrogen, Moscow, Russia. DNAse/RNAse free water was purchased from Qiagen, Hilden, Germany. Salts, including MgCl 2 , NaCl, KCl, HEPES, and DEPC were purchased from CarlROTH, Carlsruhe, Germany. Acrylamide was purchased from Applichem, Illinois, USA. Other materials for electrophoresis, including agarose, tris, boric acid, EDTA, bisacrylamide, APS, TEMED, ethidium bromide, and SDS were purchased from Helicon, Moscow, Russia. Ladders and gel loading dyes were purchased from Evrogen, Moscow, Russia. Consumables for bacteria growth, including agar, tripton and yeast extract, were purchased in Dia-M, Moscow, Russia. Glass beads of technical grade was a gift from a technical department. Sodium acetate, phenol, and chlorophorm were purchased from Lenreactiv, Saint-Petersburg, Russia.

Assembling of DNMs
Stock solutions of DNM1_thu, DNM1-1_thu, DNM1_myc, and DNM1-1_myc were prepared by annealing 10 nM of (T1_thu and T2_thu), (T1_thu and T2   20 nM DZa_myc, and specific synthetic analyte in concentrations ranged from 0 pM to 800 pM. When the assay was performed with either B. thuringiensis, B. mycoides, or B. cereus total RNA, the concentration ranged from 0.1 ng to 20 ng. When the assay was performed with cell lysates, the cells were first pelleted by centrifugation and resuspended in the reaction buffer containing DNM and DZa. The samples were heated for 5 min at 95 • C to enable cell lysis and centrifuged to pellet the cell debris, and the supernatant was transferred to a clean tube before adding the reporter to minimize the background. All samples were incubated at 55 • C for 60 or 180 min followed by fluorescent measurement. This measurement was performed at Spark, Tecan, Switzerland fluorimeter at wavelengths (ex/em) 480 nm/525 nm for F-sub and 617/662 nm for Cy-sub.

Preparation and Analysis of Total RNA
Total RNA was isolated from Bacillus cereus ATCC 14579, Bacillus thuringiensis ATCC 10792, and Bacillus mycoides ATCC 6462 according to the method developed by Oh and So [70] ( Figure S1). Namely, the cells of B. thuringiensis and B. cereus were grown in 3 mL LB at 37 • C for 16 h while the cells of B. mycodies were grown in 3 mL LB at 30 • C. Then the cells were harvested by centrifugation at 10,000× g for 3 min. The cell pellets were resuspended in a lysis solution of 800 µL (sodium acetate 2.7 g, SDS 5 g, EDTA 0.34 g per liter of deionized water, pH 5.5). Unless otherwise specified, all procedures were conducted at room temperature. The acid pre-treated glass beads (0.8 g, 425-600 um in diameter) were added in the cell suspensions and the samples were vortexed for 2-3 min at 2000 rpm to break the cell wall. The samples were centrifuged at 10,000× g for 3 min and the supernatants (500 µL) were transferred to fresh 1.5 mL microtubes. The supernatants were mixed with 500 µL of saturated phenol (pH 5.5) and incubated at 68 • C for 5 min with periodic mixing. Samples were then centrifuged at 10,000× g for 3 min. The aqueous layer (450 µL) was transferred to a 1.5 mL fresh microtube and mixed with chloroform (450 µL) before being centrifuged at 10,000× g for 3 min. The aqueous layer (400 µL) was transferred to a 1.5 mL fresh microtube and mixed with 100% ethanol (800 µL) and 3 M sodium acetate (1/10 volume). Then RNA pellets obtained by centrifugation at 10,000× g for 10 min, were washed with ice-cold 70% ethanol, and dried in a speed vacuum. Dried RNA samples were resuspended in 45 µL of nuclease-free water. The samples were visualized by electrophoresis in 1% agarose at 100 V during 30 min followed by staining in Ethidium Bromide Stain for 15 min.

Preparation and Analysis of the Whole Bacterial Cells
B. thuringiensis was inoculated with 200 mL of Lysogeny broth (LB broth) and cultivated overnight at 37 • C. The cells were diluted to reach OD600 of~1.1. To perform the calibration up to higher OD, 1 mL of the culture was centrifuged, the supernatant was removed, and another 1 mL culture was added. The sample was centrifuged again, the supernatant was removed, and 1 mL saline was added, and the cells were re-suspended. This B. thuringiensis stock solution reached an OD600 of 2.2 AU. Serial dilutions were made from the stock solution. The serial dilutions were further diluted by a factor of 80,000 (4 times 10×, followed by a final 8× dilution). A total of 50 µL of the diluted B. thuringiensis serial dilutions were plated onto Petri dishes and were incubated overnight at 37 • C and the colony forming units were counted. Only plates between 30 and 300 CFU were used for the calculation of the cell numbers. The cell number as CFU/mL was calculated by multiplication of the counted CFU on the Petri dish with the dilution factor of 80,000 and multiplication of 20 to correct for the CFU per mL. The same procedure was repeated to calculate the number of B. mycoides cells.
For the assay, the serial dilutions were prepared in 1 mL per sample and the cells were collected by centrifugation at 900× g for 5 min. The supernatant was discarded, and the cell pellet was resuspended in 150 µL of the reaction buffer containing either B. thuringiensis-specific or B. mycoides-specific DNM1 and BiDz. To ensure the cell lysis and liberation of the RNA from the cells and to anneal it with the deoxyribozyme strands, the mixture was heated at 95 • C for 5 min, left to cool down at room temperature for 5 min, and then centrifuged at 900× g for 3 min. The supernatant (~140 µL) was transferred into a clean tube before adding 1 µL of RNase inhibitor and the reporter to minimize the background. All samples were incubated at 55 • C for 60 or 180 min followed by fluorescent measurement.

Statistical Analysis
A t-test was used to compare the differences between the B. cereus, B. mycoides, and B. thuringiensis. An ANOVA of variance followed by post hoc tests (Bonferroni) was used to compare differences between different groups. All values are presented as mean ± SD. A p-value < 0.05 was considered statistically significant.

Conclusions
DNA machines for fluorescence analysis of biologically folded RNA were designed and tested. The multi-armed DNM approach enabled the amplification-free detection and discrimination of B. thuringiensis and B. mycoides. The multi-armed approach does not require long analyte-binding fragments and so allows for the detection of targets with complex secondary and SNVs at the same time. The assay enabled expedite quantification of rRNA content in cell cultures starting from whole cells with LOD down to 30,000 bacterial cells. The method requires 1.5 h with a hands-on time of~10 min, which outperforms other commercially available tests. The approach is sensitive to SNVs. Further development of the technology should lower the LOD and shorten the assay time to 15-30 min, which could make it competitive with state-of-the-art detection techniques. The same approach is not limited to the Bacillus cereus species and can be later expanded to apply to other difficult-to-discriminate targets.