A Multiplex Electrochemical Biosensor for Bloodstream Infection Diagnosis

Reagents and Samples

Six pairs of oligonucleotide probe pairs (defined in Table 1 ), including four specific probe pairs for EC, SA, PA, and PM; one Enterobacteriaceae (EB) probe pair for all EB; and a universal (UNI) probe pair for detecting the presence of all bacteria, were evaluated. ¹⁹ The probes were synthesized by Integrated DNA Technologies (Coralville, IA). The capture and detector probes were modified with thiol and fluorescein at the 5′ end and 3′ end, respectively ( Table 1 ). Four clinical isolates (EC, PA, PM, and SA) were obtained from the clinical microbiology laboratory at the Veterans Affairs Palo Alto Health Care System (VAPAHCS). The bacterial colonies were freshly incubated in Luria broth (LB) media and grown to logarithmic phase before the experiment. Human whole blood for spiking experiments was obtained from Innovative Research Inc. (Novi, MI). De-identified EC-positive blood culture samples were obtained from the clinical microbiology laboratory at the Penn State Hershey Medical Center.

OPEN IN VIEWER

Table 1.

Sequences of Capture and Detector Probe Pairs in This Study.

Probe a	Position (Length) b	Sequence (5′ to 3′)
Universal bacterial (UNI)
Capture	798 (22)	TCGTTTACRGCGTGGACTACCA
Detector	776 (22)	GGGTATCTAATCCTGTTTGCTC
Enterobacteriaceae (EB)
Capture	1275 (23)	ACTTTATGAGGTCCGCTTGCTCT
Detector	1252 (23)	CGCGAGGTCGCCTTCCTTTGTAT
Escherichia coli (EC)
Capture	471 (24)	CTGCGGGTAACGTCAATGAGCAAA
Detector	447 (24)	GGTATTAACTTTACTCCCTTCCTC
Proteus mirabilis (PM)
Capture	1019 (22)	AGCGTTCCCGAAGGCACTCCTC
Detector	997 (22)	TATCTCTAAAGGATTCGCTGGA
Pseudomonas aeruginosa (PA)
Capture	594 (23)	CCCGGGGATTTCACATCCAACTT
Detector	570 (23)	GCTGAACCACCTACGCGCGCTTT
Staphylococcus aureus (SA)
Capture	91 (22)	CCCGTCCGCCGCTAACRTCAGA
Detector	69 (22)	GRAGCAAGCTYCTCGTCYGTTC

a

The capture and detector probes were modified with 5′ thiol and 3′ fluorescein, respectively.

b

Position of the 5′ nucleotide in alignment with the bacterial 16S ribosomal RNA molecule.

Electrochemical Sensors

Gold electrodes for electrochemical sensing were fabricated by sputtering with a shadow mask on plastic substrates. Each chip consisted of 16 sensors. The thiolated capture probes were immobilized on the gold sensor surface by self-assembly. Briefly, 6 µL of 0.05 µM thiolated capture probe and 300 µM hexanedithiol (HDT) in TE buffer (Tris-EDTA, 1X) were spotted on the working electrode and incubated overnight at 4 °C. After washing with water and drying with nitrogen gas, 6 µL of 1 mM 6-mercapto-1 hexanol (MCH) solution was spotted on the working electrode and incubated for 50 min at room temperature. A laser machining system (VLS2.30; Universal Laser System, Scottsdale, AZ) was used to create plastic wells for the electrochemical biosensor. Each sensor received 50 µL of substrate solution for amperometric sensing. The current signal was measured by a multichannel potentiostat. A negative control, LB medium or blood, was included on each sensor chip.

Detection of Bacterial 16S Ribosomal RNA

Bacterial lysis was performed using 50 µL of 100 mg/mL lysozyme (Sigma, St. Louis, MO) and 25 µL of 100 mg/mL lysostaphin (Sigma, St. Louis, MO) in 425 µL of lysis buffer. The lysis solution (20 µL) was added to the sample (50 µL) and incubated for 5 min at room temperature. Then, 13 µL of 1 M NaOH was added to the sample and incubated for 5 min at room temperature. After incubation, 83 µL of the detector probe solution with 2.5% bovine serum albumin (Sigma, St. Louis, MO) and 1 M phosphate buffer (pH 7.4) were added. After hybridization for 10 min at 37 °C, 6 µL of the sample was placed on each sensor. The sample was incubated on the sensor for 15 min at 37 °C. The sensor was washed with deionized water and dried with nitrogen. Then, 6 µL of 0.5 U/mL antifluorescein horseradish peroxidase (HRP) Fab fragments (Roche Diagnostics, Indianapolis, IN) in 1 M phosphate buffered saline containing 0.5% casein was added to each sensor. The chip was incubated at room temperature for 15 min. Plastic wells were then placed on the sensor chips, and 50 µL TMB-H₂O₂ solution (K-Blue Low Activity TMB Substrate; Neogen, Lexington, KY) was added in each well. For amperometric sensing, a bias voltage of 200 mV was applied to the potentiostat to measure the amperometric signals. The assay required less than 1 h.

Pathogen Identification in Spiked Blood

For detection in whole blood, four times the volumes of each reagent and sample were used. The larger volume was required to generate adequate solution for pipetting onto the sensor. To spike bacteria in blood, the bacterial samples were centrifuged to remove the media and resuspended in blood. The concentration of the spiked samples was ~10⁸ colony-forming units (CFU)/mL. Each chip was immobilized with a capture probe. The samples with the detector probes were loaded to the sensor chips with the corresponding capture probes. To test the ability of the multiplex electrochemical sensor for bloodstream infection diagnosis, 40 blood samples were prepared by spiking different bacteria species into whole blood. The bacteria concentrations were on the order of 10⁷ and 10⁸ CFU/mL.

Amperometric Detection of 16S Ribosomal RNA

Electrochemical biosensing was performed using a portable, multichannel potentiostat ( Fig. 1A ). The multiplex electrochemical sensor chip consisted of 16 gold sensor electrodes with reference and auxiliary electrodes surrounding a circular working electrode ( Fig. 1B ). Each biosensor was coated with a self-assembled monolayer consisting of HDT, MCH, and a thiolated capture probe. ²⁰ The multiplex electrochemical sensor used a sandwich binding scheme with a capture probe and a detector probe for detecting the target 16S ribosomal RNA (rRNA) ( Fig. 1C ). The target 16S rRNA was immobilized by the capture probe and was detected by the detector probe bound with HRP. The HRP catalyzes the redox reaction to generate an amperometric signal with the addition of TMB and H₂O₂ substrates. An example of the current outputs of 16 electrochemical sensors simultaneously is shown in Figure 2A .

OPEN IN VIEWER

Figure 1.

A multiplex electrochemical sensor platform for detection of blood-borne pathogens. (A) A portable, multichannel potentiostat with a chip mount for amperometric sensing. (B) A disposable, 16-sensor chip (2.5 by 7.5 cm) fabricated by gold deposition on a plastic substrate. Each electrochemical sensor consists of a central working electrode, a circumferential reference electrode, and a short auxiliary electrode. (C) A schematic representation of the sandwich binding scheme on the central working electrode with the self-assembly monolayer. The target is captured on the sensor surface by the capture probe (CP) and detected using the detector probe (DP) amperometrically. HRP, horseradish peroxidase; rRNA, ribosomal RNA.

OPEN IN VIEWER

Figure 2.

Analytical performance of the multiplex electrochemical sensor. (A) A representative plot depicting the current outputs of 16 sensors over 60 s. (B) Analytical sensitivity of the electrochemical sensor for bacteria detection. The Escherichia coli sample was cultured to approximately ~10⁸ colony-forming units (CFU)/mL and diluted to the appropriate concentrations with Luria broth (LB) media. (C) Analytical specificity of the multiplex electrochemical sensor platform. Escherichia coli (EC), Pseudomonas aeruginosa (PA), Proteus mirabilis (PM), Staphylococcus aureus (SA), and a mixture of all four bacteria (Mix) were spiked in whole blood. Blank represents blood only.

Analytical Performance of the Electrochemical Biosensor

The analytical performance of the multiplex electrochemical biosensor was first investigated and optimized in media. A serial dilution experiment was performed to determine the signal levels of EC from 10⁴ to 10⁸ CFU/mL. The sensor signal correlated with the concentration of EC for over four orders of magnitude ( Fig. 2B ). Fitting the data with the power law revealed a limit of detection (blank + 3× standard deviation) of 290 CFU/mL in culture media.

The ability of the electrochemical biosensor for pathogen identification in whole blood was then examined. EC, PA, PM, SA, and mixtures of all four bacteria (MIX) were spiked in blood. The samples were detected using sensor chips with specific probe pairs (EC-chip, PA-chip, SA-chip, and PM-chip). A universal probe pair targeting the conserved region of the bacterial 16S rRNA was also included in the experiment (UNI-chip). Figure 2C shows the results for detecting different bacteria with each chip. The multiplex biosensor detected the target pathogens with signal levels significantly higher than the blank (blood only) and nonspecific bacteria, suggesting a low cross-reactivity between the probes. The background level of the blank (whole blood) was approximately 35 nA for target-specific chips (EC, PM, PA, and SA). The signal level for nonspecific target increased to ~66 nA. In this study, a threshold value of 100 nA was defined to identify the existence of a pathogen in the sample.

The UNI-chip successfully detected the existence of bacteria in all samples ( Fig. 2C ). In agreement, the highest signal was observed when all four bacteria were mixed in the sample. In a clinical setting, the UNI probe is beneficial for first determining whether a bacterial infection is present, in contrast to false-positive or fungal infection. More important, the UNI probe is designed based on the conserved region of the bacterial 16S rRNA and can indicate the existence of bacteria when a specific probe is not available.

Optimization for Bloodstream Infection Diagnostics

Several aspects were considered in the design of the multiplex biosensor for bloodstream infection diagnostics. The multiplex electrochemical biosensor is designed to detect both gram-negative and gram-positive bacteria. For instance, SA is commonly observed in bloodstream infection. ²¹ Without lysostaphin, a low signal was obtained due to the inefficiency in lysing the bacteria. Lysostaphin was, therefore, added in the lysis solution for cleaving the cell wall of staphylococci in all samples. Furthermore, a challenge of detecting bacteria in whole blood was coagulation and thickening of the sample after the addition of sodium hydroxide. Although an anticoagulant or a large volume of dilution buffer could be added to the sample, these procedures could reduce the signal-to-noise ratio of the biosensor. To facilitate pipetting of the sample, a larger volume was prepared in the protocol. At least four times the volumes of each reagent and sample were prepared to address the issue.

A test procedure was developed to demonstrate the ability of the multiplex electrochemical sensor for pathogen identification ( Fig. 3 ). UNI, EB, EC, PM, SA, and PA probes were precoated on the sensor chips. In the first step, the samples were tested with the UNI-chip to detect the existence of bacterial pathogens in the samples. If the samples were positive, the samples were tested with the EB-chip to determine if the bacteria was Enterobacteriaceae or not. If the samples were EB positive, the samples were tested with the EC-chip and the PM-chip. If the samples were EB negative, the samples were tested with the PA-chip and SA-chip.

OPEN IN VIEWER

Figure 3.

Strategy to identify bacteria species in blood samples. In step 1, unknown samples are first detected with the universal probes (UNI-chip) to confirm the existence of bacteria pathogens. In step 2, UNI-positive samples are then detected with the Enterobacteriaceae probes (EB-chip). In step 3, EB-positive samples are further detected using the EC-chip and PM-chip. EB-negative samples are detected using the PA-chip and SA-chip. EC, Escherichia coli; PA, Pseudomonas aeruginosa; PM, Proteus mirabilis; SA, Staphylococcus aureus.

Pathogen Identification with Clinical Isolates Spiked in Whole Blood

We designed a spiking study to test the ability of the electrochemical biosensor to detect bacteria in blood samples. Three common bacteria (EC, PA, and SA) were spiked into 30 blood specimens. Fifteen samples were spiked with EC, eight samples were spiked with PA, and seven samples were spiked with SA. In addition, 10 blood samples were included as a negative control. All 40 samples were first tested using the UNI-chip ( Fig. 4A ). The 30 positive and 10 negative samples were correctly identified using the threshold value. In the second step, the 30 UNI-positive specimens were detected using the EB-chip ( Fig. 4B ). The EB-chip had high signals for the 15 EC specimens and low signals for the eight PA and seven SA specimens. In the third step, the EC-chip and PM-chip were applied to detect the 15 EB-positive specimens, and the PA-chip and SA-chip were applied to detect the 15 EB-negative specimens ( Fig. 4C–F ). The chips correctly identified all EC, PA, and SA samples. For a nonspecific target, the signal levels were below 25 nA for all sensors, compared with the threshold value of 100 nA for positive detection. These results support the selectivity of the species-specific probes for bloodstream infection diagnostics.

OPEN IN VIEWER

Figure 4.

A pathogen identification experiment by spiking EC (15, three different concentrations), PA (eight, one concentration), and SA (seven, two different concentrations) in whole blood. Ten blood samples were also included as a negative control. (A) Detection with the universal probes (UNI-chip). Sample numbers are sorted for clarity. Samples 1 to 30 were spiked with bacteria and samples 31 to 40 were blood only. (B) Detection with the Enterobacteriaceae probes (EB-chip). Samples 1 to 15 were EC, samples 16 to 23 were PA, and samples 24 to 30 were SA. (C, D) Detection of EB-positive samples with the EC-chip and PM-chip. All samples were EC. (E, F) Detection of EB-negative samples with the PA-chip and SA-chip. Samples 1 to 8 were PA and samples 9 to 15 were SA. EC, Escherichia coli; PA, Pseudomonas aeruginosa; PM, Proteus mirabilis; SA, Staphylococcus aureus.

Electrochemical Biosensors for Positive Blood Cultures

To evaluate the feasibility of the multiplex electrochemical sensor for bloodstream infection diagnosis, samples from positive blood culture bottles were tested. The samples were confirmed to be positive for EC in the clinical microbiology laboratory. All samples were first tested with the UNI-chip ( Fig. 5A ). The signals were over 100 nA in all the sensors, suggesting the existence of the bacterial pathogens in the samples. The EB-chip was then applied and resulted in high signals for all samples ( Fig. 5B ). Since all samples were EB positive, only the EC-chip and PM-chip were used in the third step ( Fig. 5C ). In agreement with the microbiological analysis, all samples were positive for EC and negative for PA. Overall, the multiplex electrochemical sensor correctly identified EC in all blood cultures ( Fig. 6A ). Since a large variation was observed among different samples, a plate count was performed to estimate the concentration of bacteria in the samples ( Fig. 6B ). A positive correlation was obtained between the plate count and current signals from the UNI-chip and EC-chip (UNI-chip, n = 10, R² = 0.775, p = 0.0008; EC-chip, n = 10, R² = 0.654, p = 0.00462).

OPEN IN VIEWER

Figure 5.

A clinical feasibility study using positive blood cultures. All samples were positive for E. coli. (A–D) Detection with the UNI-chip, EB-chip, EC-chip, and PM-chip. PA-chip and SA-chip were not tested since all samples were EB positive. EB, Enterobacteriaceae; EC, Escherichia coli; PA, Pseudomonas aeruginosa; PM, Proteus mirabilis; SA, Staphylococcus aureus; UNI, universal probes.

OPEN IN VIEWER

Figure 6.

Comparison of the sensor performance with blood cultures. (A) Current outputs of different sensor chips. (B) Correlation of the current signal with the plate count of the positive blood cultures. EB, Enterobacteriaceae; EC, Escherichia coli; PM, Proteus mirabilis; UNI, universal probes.