Molecular Detection of Bacterial Contamination in Blood Components Using Magnetic-based Enrichment

Bacterial contamination of blood products is a major problem in transfusion medicine, in terms of both morbidity and mortality. Platelets (PLTs) are stored at room temperature (under constant agitation) for more than 5 days, and bacteria can thus grow signicantly from a low level to high titers. However, conventional methods like blood culture and lateral ow assay have disadvantages such as long detection time, low sensitivity, and the need for a large volume of blood components. We used real-time polymerase chain reaction (PCR) assays with antibiotic-conjugated magnetic nanobeads (MNBs) to detect enriched Gram-positive and -negative bacteria. The MNBs were coated with polyethylene glycol (PEG) to prevent aggregation by blood components. Over 80% of all bacteria were captured by the MNBs, and the levels of detection were 10 1 colony forming unit [CFU]/mL and 10 2 CFU/mL for Gram-positive and negative bacteria, respectively. The detection time is < 3 h using only small volumes of blood components. Thus, compared to conventional methods, real-time PCR using MNBs allows for rapid detection with high sensitivity using only a small volume of blood components. 4 × 10 12 /mL (Supplementary Many bacterial species contaminate PLTs. IMS is not diagnostically useful; the species of bacteria present remain unknown until the results are conrmed. MNBs must be conjugated with receptors that bind to a broad spectrum of bacterial species. Vancomycin binds to peptides of Gram-positive bacteria that terminate in -Lys-D-Ala-D-Ala [41], and allantoin binds to the LPS of the outer membrane of Gram-negative bacteria [42]. The capture eciencies and SEM images showed that these compounds were immobilized on MNBs@PEGs that bound E. coli and S. aureus, respectively.


Introduction
Bacterial contamination of blood products is a major problem in transfusion medicine [1][2][3]. Especially, transfusion of contaminated platelets (PLT) may cause serious infections and septic reactions [1,4]. PLTs are stored at room temperature (under constant agitation) for 5 days; bacteria can thus easily multiply from low levels (< 1 colony forming unit [CFU]/mL) to high titers (< 10 8 CFU/mL) [5][6][7]. The US Food and Drug Administration (FDA) reported that transfusion of contaminated PLTs caused 51 deaths from 2001 to 2016 [8]. The European Commission reported that 43 transfusions were contaminated in the European Union from 2010 to 2013; 36 involved contaminated PLTs [9]. Although there have been a few cases, the morbidity and mortality rates are very high [10,11]. To reduce mortality, accurate and rapid bacterial detection is required [7]. The gold standard for the detection is blood culture, which is one of the oldest clinical techniques [12,13]. However, bacterial growth to detectable levels usually requires from 24 h [14] to several days [15]. Also, low bacterial titers and slow bacterial growth can cause false-negative results when automated blood culture systems are employed [16,17]. Much effort has thus been devoted to the rapid and sensitive detection of bacterial pathogens in blood [18][19][20][21]. The enhanced bacterial detection (eBDS) system (Haemonetics Corporation, Braintree, MA, USA) indirectly detects bacteria by measuring decreases in oxygen concentration over 24 h, but cannot detect anaerobic bacteria [22]. The Platelet Pan-Genera Detection (PGD) test (Verax Biomedical, Marlborough, MA, USA, USA) is a lateral ow immunoassay detecting lipoteichoic acid (LTA) and lipopolysaccharide (LPS) in aerobic and anaerobic Gram-positive and -negative bacterial species, respectively, within 30 min [23]. However, the sensitivity is low (approximately 10 4 CFU/mL) and the false-positive rate is high [24].
Nucleic acid (NA) ampli cation via polymerase chain reaction (PCR) sensitively and speci cally detects bacterial pathogens [25][26][27][28][29]. However, on the day of PLT production, the PCR sensitivity was only 12.8% that of the BacT/ALERT system (Organon Teknika Corp., Durham, NC, USA) because the bacterial loads were very low [30]. Also, PLTs contain many substances (such as immunoglobulin G) that interfere with NA ampli cation [31]. It is thus essential to enrich bacteria and prepare puri ed bacterial DNA for accurate and rapid detection. Many commercial kits are used to extract NA from blood products; most employ solid-phase extraction [32]. However, these cannot remove inhibitors or enrich pathogens and bacteria are not isolated. Immunomagnetic separation (IMS) using antibody-conjugated magnetic nanobeads (Ab-MNBs) is widely applied to isolate pathogens and thus eliminate inhibitory substances [33][34][35][36].
However, Ab-MNBs do not detect all bacteria that cause sepsis; the antibody is speci c for 1 species but at least 10 species of bacteria cause sepsis. Thus, MNBs must be conjugated with materials that bind a broad spectrum of bacterial species. Potential candidates include antibiotics and certain lectin-based natural compounds, although these have not yet been validated clinically [37,38].
Thus, we developed a method that enriches both Gram-positive and -negative bacteria using MNBs coated with two different materials; we then extracted bacterial DNA. Figure 1 shows two methods used for sample preparation (with and without specimen incubation). Performance was tested by spiking 2.5-mL amounts of apheresis plasma with Escherichia coli O157:H7 (E. coli O157:H7) and Staphylococcus aureus (S. aureus). Extracted bacterial DNAs were ampli ed via real-time PCR.  Table 1). E ciencies of bacterial enrichment by MNBs. As shown in Figure 3a Real-time PCR for the detection of enriched bacteria. DNA was extracted from the enriched Gram-negative and -positive bacteria and subjected to real-time PCR. As shown in Figure 4a, E. coli at 10 2 CFU/mL was detected after 12 h of specimen incubation, and Figure 4b shows that E. coli not captured by MNBs were detected at only 10 4 CFU/mL. As shown in Figure 4c, 10 3 CFU/mL of E. coli was detected without prior specimen incubation, but DNA extracted from Page 3/10 containing E. coli captured (Ct; 30.54 ± 0.48) and not captured (Ct; 31.95 ± 0.95) by MNBs at a level of 10 4 CFU/mL were compared, while MNBs@PEG-Al captured the bacteria effectively at a level of 10 3 CFU/mL ( Table 1). As shown in Figure 5a, S. aureus at 10 1 CFU/mL was detected after 12 h of specimen incubation. Figure 5b shows that S. aureus not captured by MNBs was detected at only 10 4 CFU/mL. 10 2 CFU/mL of S. aureus was detected without prior specimen incubation, but DNA extracted from 10 1 CFU/mL could not be detected (Figure 5c). Figure 5d shows that MNBs@PEG-Van captured the bacteria effectively at a level of 10 4 CFU/mL.

Discussion
When PLTs are stored for 5 days at RT with agitation, even if the initial suspension contains < 1 CFU/mL, the bacteria can proliferate [1,4]. Current culture methods do not yield contamination data during PLT storage. Culture requires 1-5 days and data are available only after the PLTs have been released.
Second, a large amount of blood (> 20 mL) is required for culture of both aerobic and anaerobic bacteria [23]. Molecular diagnostics requires small sample volumes (< 1 mL) and detects pathogens within 3 h. It is essential to reduce culture time by improving bacterial enrichment. Such enrichment by MNBs requires that they be dispersible in blood components, and that receptors such as vancomycin and allantoin bind to the bacterial pathogens. Plasma contains many proteins, clotting factors, and IgG [40]; these readily adsorb to non-PEG-coated MNBs. After a 20-min incubation, MNBs aggregated in plasma, but MNBs@PEG did not (Supplementary Figure 1). PEG prevented non-speci c binding to the surfaces of MNBs. The MNBs (4 × 10 9 to 4 × 10 12 beads/mL) were added to 1 mL amounts of apheresis plasma containing bacterial pathogens. Over 80% of all E. coli and S. aureus were captured at MNB levels of 4 × 10 11 to 4 × 10 12 /mL (Supplementary Table 2). Many bacterial species contaminate PLTs. IMS is not diagnostically useful; the species of bacteria present remain unknown until the results are con rmed. MNBs must be conjugated with receptors that bind to a broad spectrum of bacterial species. Vancomycin binds to peptides of Gram-positive bacteria that terminate in -Lys-D-Ala-D-Ala [41], and allantoin binds to the LPS of the outer membrane of Gram-negative bacteria [42]. The capture e ciencies and SEM images showed that these compounds were immobilized on MNBs@PEGs that bound E. coli and S. aureus, respectively.
Pathogen enrichment by MNBs prior to NA extraction reduced the levels of possible inhibitors and yielded more NA than commercial kits. The detection time was thus dramatically reduced (< 15 h). Both the E. coli and S. aureus enrichment rates were > 80% and levels of 10 1 and 10 2 CFU/mL, respectively, were detected after 12 h of specimen incubation. The levels of detection without incubation were 10 2 and 10 3 CFU/mL, respectively. Thus, our method is at least 100-fold more sensitive than lateral ow assay kits such as the Platelet PGD test (limit of detection = 10 4-5 CFU/mL) after 12 h of specimen incubation [24].
In addition, our method requires only small sample volumes (< 1 mL). However, we aim to further improve the sensitivity and devise a fully automated highthroughput system. Our current focus is on optimization of the sample preparation method. Bacterial enrichment by MNBs. Apheresis plasma containing bacteria (E. coli O157:H7 and S. aureus) at 10 1 -10 4 CFU/mL were incubated at RT for 12 h. One milliliter of plasma was mixed with 200 µL of either MNBs@PEG-Al or MNBs@PEG-Van (4 × 10 11 particles/mL, nal concentration) and the mixtures were incubated at RT for 20 min. Bacteria-MNB clusters were separated using a magnetic rack. The residues, which contained uncaptured bacteria, were microliters of lysis buffer were added to a suspension of BE-MNBs followed by incubation for 10 min at RT. The elution volume was 100 µL; the extracted DNA was stored at -80°C. DNA purity and yield were assessed by quantifying absorbance at 230, 260, and 280 nm using a Nano-200 spectrophotometer (Allsheng, Hangzhou, China). To detect Gram-positive bacteria, 10 µg lysostaphin (Sigma-Aldrich) was added to the BE-MNBs followed by incubation for 10 min at 37℃ prior to the addition of lysis buffer.

Methods
Real-time PCR assay. Real-time PCR was used to con rm the identities of the captured bacteria. The primers were designed using PrimerQuest (Integrated DNA Technologies Inc., Coralville, IA, USA): SA nuc_F (5'-TATGGACGTGGCTTAGCGTAT-3') and SA nuc_R (5'-GACCTGAATCAGCGTTGTCTT-3') for S. aureus; and EB tyrB_F (5'-AAGAGGATGCCTACGCCATT-3') and EB tyrB_R (5'-CTTGGCGGGCTGGAGTAGTT-3') for E. coli. Power SYBR Green PCR Master Mix (Applied Biosystems, Waltham, MA, USA) served as the PCR master mix; all primers were added to 0.2 µM. QuantStudio 3 (Applied Biosystems) was used to perform PCR. Positive and negative ampli cation controls were included in every run. The positive controls contained DNA directly extracted from S. aureus and E. coli, and the negative control was RNase-and DNase-free water. A result was considered positive when the threshold cycle (Ct) was > 37.0 and the melting temperature (Tm) was appropriate (76.0 ± 0.5°C).  Figure 1 Schematic illustration of pathogen enrichment by MNBs (magnetic nanobeads) and NA (nucleic acid) extraction from 1-mL samples of blood components spiked with bacteria (a) after 12 h of specimen incubation at RT (room temperature), and (b) without prior specimen incubation. The procedural steps were as follows: sampling, specimen preparation, pathogen enrichment by MNBs, and NA extraction.