Rapid DNA visual detection of polymicrobial bloodstream infection using filter paper

In this study, we optimized the length of probe immobilized on the filter paper surface; designed specific primers of one bacterium (E. coli), one fungus (S. cerevisiae), and one virus (HCMV); and confirmed the corresponding optimal concentration of each pair of primers and the universal annealing temperature in one multiplex PCR system. The functionalized filter paper can simultaneously detect the three amplified products from different species of DNA (E. coli, S. cerevisiae, and HCMV) in one test, which are the common pathogens of polymicrobial BSI in patients with neutropenia^{29,30,31,32,33,34,35,36}.

Investigation of printed probe length

The goal of printing a probe on the filter paper is to capture a specific complementary target in detective solution and hybridize with it to form a double helix. Stability and specificity of DNA hybridization are impacted by several factors, such as hybridization temperature, time, platform, complexity of the probe or target, washing solution composition, washing conditions, among others^37,38. The hybridization in our method occurred at room temperature, on the surface of a cellulose filter paper, and in a short period of time. We have previously optimized washing solution and conditions²³, and investigated the hybridization specificity based on the different mutation rates of a target sequence²⁸. In this study, we compared the hybridization effect based on four different lengths of S. cerevisiae specific probe sequence. All detection showed target-specific visual signals. Both visual detection and quantity analysis based on the 45-base probe displayed the strongest signal intensity (Fig. 1). ANOVA revealed a significant difference in the average intensities between these four groups (p < 0.05). Post-hoc analysis revealed that the signal intensity captured by the 45-base probe was significantly stronger than that of the 35-based probe. One of the most likely reasons would be that the length of 45 bases helps the cooperative transient hybridization on the functionalized filter paper surface to form the most stable conformation. However, the complexity of longer probes may challenge this cooperativity by increasing thermodynamic driving forces of hybridization at room temperature, for example, by extending the formation time of the correct initial base-pair and complicating the following events of stacking and base pairing influenced by neighboring pairs^39,40. Considering hybridization efficiency in the given condition and cost of detection, we selected the length of 45 bases to design all the probes in the downstream work. To avoid background noise resulting from primer dimers, all the probes for subsequent experiments were designed to differ from primer sequences.

Detection effect comparison based on the length of printed probe sequence. The strongest signal was captured by the 45-base probe. *p ≤ 0.05/comparison times (post hoc test).

Optimization of PCR assay and construction of standard curve

Blood culture is suboptimal to identify polymicrobial BSI for various reasons, such as cultivability of pathogens, the low quantity of pathogens, or the different stages of infection^41,42. To simultaneously identify multiple pathogens from one sample in one test, we developed a multiple detection model using activated cellulose filter paper. E. coli, S. cerevisiae, and HCMV are pathogenic microorganisms commonly occurring in the neutropenic patients with BSI and are associated with a significant increase in mortality rate^29,31,34. In our amplification system, the optimal concentrations of these three pairs of primers were 200 nM (E. coli), 100 nM (S. cerevisiae), and 300 nM (HCMV). The universal annealing temperature of these three pairs of primers was 60 °C. All three standard curves based on a tenfold serial dilution of gDNA showed negative linear correlations between the copies of initiated DNA and cycle threshold (C_T) (Fig. 2). R squared values for E. coli, S. cerevisiae, and HCMV were calculated as 0.98, 0.99, and 0.99, respectively, to verify the reliability of the reactions. We also investigated accuracy of multiplex real-time PCR assays based on these three gDNA. Both duplex and triplex real-time PCR assays showed target-specific amplification based on melting curve analysis (Fig. S1).

Standard curves of three qPCRs (Target: (a) Escherichia coli, (b) Saccharomyces cerevisiae, and (c) Human cytomegalovirus; Primer: (a) specific primers of Escherichia coli, (b) ITS3/4 fungal primers, and (c) specific primers of Human cytomegalovirus) display negative linear relationships between the initiated copies of each pathogen and the corresponding C_T values. R² values indicate the sufficient quality of primers to yield the reliable experimental data.

Investigation of singleplex, mixed, and multiplex amplicon detection

PCR products amplified either from gDNA or sDNA of E. coli (B), S. cerevisiae (F), and HCMV (V) were detected using the corresponding probe (Table S2) printed on filter paper. The target was 500 ng of each singleplex PCR product (F, B, or V), or of the mixture of singleplex PCR products (FV, VB, FB, or FVB), or of the multiple targets from multiplex PCR (FV, VB, FB, or FVB). All the amplicon detection from both gDNA (Fig. 3a) and sDNA (Fig. 3b,c) displayed clear and specific signals from a singleplex target to a mixed target, or to a multiplex target. Figure 3 shows that the difference of intensity between each target signal and the negative control of APT₂ on the same paper is significant. All the specific single or multiple target signals were possible to be identified on site by the naked eye as shown in Fig. 3c.

Detection with different types of targets. (a) Signal intensities of multiplex PCR products (FV, VB, FB, and FVB) amplified from genomic DNA. (b) Signal intensities from a single PCR product (F, V, and B) to a mixture of two or three single PCR products (FV, VB, FB, or FVB) amplified from spiked DNA. (c) Visual detection using multiplex PCR products amplified from spiked DNA. All the differences of intensities between the specific and the NTC signals on the same filter paper were significant. X axes of 3a and 3b represent the targets of detection. Y axes display the signal intensities. F: Saccharomyces cerevisiae. V: human cytomegalovirus. B: Escherichia coli. NTC: negative control. *p ≤ 0.05 using student t-test (F, V, and B) or p ≤ 0.05/comparison times using post hoc test (FV, VB, FB, or FVB).

Evaluation of limit of detection

PCR products amplified from gDNA were used to estimate the limit of detection using the activated filter paper tool (Figs. 4 and S2). Seven printed filter paper slides, three different singleplex amplicons (F, V, B), and four different multiplex amplicons (FV, VB, FB, and FVB) were used to complete the detection. The detection signals of four target quantities—10 ng, 50 ng, 100 ng, and 500 ng—were compared. Increasing signal intensity was displayed as the increase in target quantity (Fig. S2). For a singleplex PCR product detection, by comparing the average specific intensity to that of NTC on the same paper, 2 (F and V) of 3 (F, V, and B) comparisons showed statistically significant differences (p < 0.05) when 10 ng of each target was used; however, all three differences were significant when 50 ng of each target was used (p < 0.05). For a multiplex PCR product detection, by comparing the average specific intensity to that of NTC on the same paper, only three (FV_V, VB_V, and FVB_V) differences were statistically significant after post hoc analysis when 10 ng of each multiplex amplicon was used, while all the differences were statistically significant only except that of VB_B when 50 ng of each multiplex target was used (Fig. 4). The current detection limit of this tool was between 0.1 ng/µL (10 ng of each target) and 0.5 ng/µL (50 ng of each target) based on the manual probe printing, which may be further improved when robotic printing is used⁴³.

Detection limit (DL) analysis of the functionalized filter paper based on PCR products amplified from genomic DNA. F, V, or B represents a single PCR product. FV, VB, or FB is the corresponding duplex PCR product. FVB means the triplex PCR product. The signal intensities of blue columns represent detection outcome when 10 ng of each target was used, and the pink ones represent the results when 50 ng of each target was used. Every signal intensity was compared to that of NTC on the same paper slide. The work was done in triplicate. The DL was between 0.1 ng/µL and 0.5 ng/µL. F: Saccharomyces cerevisiae. V: human cytomegalovirus. B: Escherichia coli. NTC: negative control. *p ≤ 0.05 using student t-test (F, V, and B) or p ≤ 0.05/comparison times using post hoc test (FV, VB, FB, or FVB). NS: no significant difference.

Rapid and accurate pathogen diagnosis is crucial in initiation of appropriate antimicrobial treatment. The main advantage of NAAT-based diagnosis of BSI is to reduce turnround time and to increase the sensitivity and specificity of pathogen diagnosis even after antibiotic treatment to guide the early BSI regimens²⁰. To develop a rapid, portable, and cost-efficient pathogen detection tool, NAATs can be combined with some cost-efficient materials such as paper^23,44,45,46. Cellulose filter paper composed of cellulose fibers possesses a three-dimension porous structure and can be functionalized with chemicals, which enable a stable immobilization of a higher concentration of probe DNA, a higher flux of targets for pathogen detection, and an automatic detection via capillary force created by the paper micropores when compared to the glass surfaces⁴⁷. Cellulose filter paper also has more options for pore size than nitrocellulose, making it suitable for molecular target binding 1 µm magnetic beads in diameter. Moreover, the high porosity and multiple pore size options of the cellulose filter paper enable it to produce rapid multiplex detection through instantaneous wicking force^46,48.

NAAT techniques are more sensitive than blood culture, however, diagnosis provided by NAATs must exclude any possible contamination from degraded or inactivated pathogens²⁰. This suggests that, without increasing the cost, the current NAAT-based and blood culture-based diagnostic methods of BSI are mutually complementary, particularly for polymicrobial BSI. An ideal BSI regimen might be based on rapid results from NAATs and physicians’ clinical evaluation, and can be further optimized by integrating the results from blood culture²⁰.

We have verified that the developed diagnostic tool using cellulose filter paper is suitable for rapid identification of multiple pathogens’ DNA simultaneously. The tool is cost-efficient (approximately $3 per triplex assay), easy to manufacture, easy to carry, easy to operate, and biodegradable^23,28. All the detection steps (from printing to detection) included in this method were operated manually and have potential to be integrated into an automatic sample-to-result portable equipment for polymicrobial BSI diagnostics. Emerging technologies in point-of-care biosensing systems such as described here and by other groups^49,50,51 have potential application in point-of-care medical diagnosis, which is in huge demand^52,53,54,55.

In the future, this work can also be expanded to different microorganisms and their antimicrobial susceptibility testing. Utilizing isothermal amplification in this system would allow instrument-free detection.