Accelerated detection of Gram negative bacteria in blood culture by enhanced acoustic flow cytometry (AFC) following peptide nucleic acid fluorescence in situ hybridization (PNA-FISH)

Bacteraemia is a risk factor for subsequent clinical deterioration and death. Current reliance on culture-based methods for detection of bacteraemia delays identification and assessment of this risk until after the optimal period for positively impacting treatment decisions has passed. Therefore, a method for rapid detection and identification of bacterial infection in the peripheral bloodstream in acutely ill patients is crucial for improved patient survival through earlier targeted antibiotic treatment. The turnaround time for current clinical laboratory methods ranges from 12 to 48 hours, emphasizing the need for a faster diagnostic test. Here we describe a novel assay for accelerated detection of bacterial infection in blood culture (BC) using peptide nucleic acid fluorescence in situ hybridization enhanced acoustic flow cytometry (PNA-FISH-AFC). For assay development, we used simulated blood cultures (BCs) spiked with one of three bacterial species: Escherichia coli, Klebsiella pneumoniae or Pseudomonas aeruginosa at a low concentration of 10 CFU/mL. Under current clinical settings, it takes a minimum of 12 hours incubation to reach positivity on the BacTEC system, corresponding to a bacterial concentration of 107-109 CFU/mL optimal for further analyses. In contrast, our PNA-FISH-AFC assay detected 103 – 104 CFU/mL bacteria in BC following a much shorter incubation of 5 to 10 hours in culture. Using either PCR-based FilmArray® assay or MALDI-TOF for bacterial detection, it took 7-10 and 12-24 hours of incubation, respectively, to reach the positive result. These findings indicate a potential time advantage of PNA-FISH-AFC assay over currently used laboratory techniques for rapid bacterial detection in BC with significantly improved turnaround time.

138 flow cytometers can therefore be used successfully for small particle analysis such as 139 bacteria. 140 We recently reported a flow cytometry assisted susceptibility testing (FAST) assay 141 method to determine a carbapenem minimal inhibitory concentration in pure cultures of 142 carbapenem-resistant Klebsiella pneumoniae using the nucleic acid intercalating dye, SYTO® 7 144 dye SYTO® 9 in both unexposed and antibiotic-exposed bacteria in pure cultures. However, 145 we believed the same approach could not be reliably used to detect bacteria in BC due to 146 interference from blood-derived elements. Accordingly, we embarked on development of a Prior to hybridization, a lysis step was introduced for simulated BC to lyse blood cells 224 and concentrate bacteria. Briefly, 1 mL of inoculated BC was harvested and added to 1 mL 225 10% Triton X-100 for 5 minutes followed by centrifugation at 10,000 g for 5 minutes. The 226 pellet was re-suspended in 1 mL HBSS and centrifuged again followed by re-suspension of 227 the pellet in 20 µL in HBSS. The sample was mixed with 480 µL of hybridization buffer, 228 followed by vortexing and incubation at 40 o C for 15 minutes. The sample was centrifuged at 229 10,000 g for 5 minutes and supernatant removed. The pellet was re-suspended in 500 µL 230 wash solution and further incubated for 10 minutes at 40 o C. The sample was centrifuged 231 and washed once more in 500µL wash solution as described above. Samples were cooled to 232 room temperature before they were analyzed by AFC.

233
Uninfected BC were lysed as described in the hybridization protocol for simulated BC 234 above and stained with anti-CD 45 PerCP (ThermoFisher Scientific, Eugene, Oregon, USA) to 235 identify blood elements of leukocyte origin by AFC analysis.    (Fig 2A). Initially, a 275 combination of forward scatter (FSC) versus side scatter (SSC) analysis was used to identify 276 bacterial cells (so-called, population of interest, POI) based on their physical properties such 277 as size and density (Fig 2A, i). This was followed by application of the FSC-H/FSC-A plot to 278 distinguish intact single cells from doublets (Fig 2A, ii). Eventually, a subset of SYTO® 9 13 279 positive (SYTO® 9+) single cells was identified in a histogram (Fig 2A, iii) and used to pinpoint 280 intact bacterial cells on the scatter plot by back-gating. This permitted ultimate 281 identification of intact bacterial cells defined by their size and staining properties with 282 SYTO® 9 that were c alled a derived POI (dPOI) (Fig 2A, iv). When using direct staining with 283 fluorescent dye SYTO® 9 the limit of detection was determined as between 10 3 -10 4 CFU/mL 284 in pure culture (See Fig S1).

285
The next step was to test whether direct staining with SYTO® 9 could be applied to  (Fig 3; Fig S1). Optimizations of formamide 323 concentration and hybrization temperature were performed with SYTO® 9 using the gating 324 method that included both single cells and doublets in order to assess a recovery of the 325 overall bacterial population stained with SYTO® 9 (Fig S1). As shown in Fig 3, top two rows, 326 30% formamide in hybridization buffer and 40 o C hybridization temperature provided 15 327 optimal conditions for the best recovery of dPOI ( Fig S2) in comparison to lower and higher 328 formamide concentrations and hybridization temperatures, respectively.

329
The hybridization signal was optimised using PNA-FISH probe as described in Fig S1B, 330 altering two assay parameters, PNA-FISH probe concentration from 100nM to 300nM, and 331 hybridization duration, from 15 to 30 minutes (Fig 3). As shown in Fig 3 and  366 the bacterial population could be found in the BL1 channel (AlexaFluor 488) while the lysed 367 blood elements (autofluorescence) were observed in the BL2 channel (Fig 4, i). The bacterial 368 population was further refined by the FSC-H/FSC-A scatter plot that was used to gate out 369 the doublets (Fig 4, ii). Distribution of signal intensity of the AlexaFluor 488 on the histogram 370 indicated uniform probe hybridization (Fig 4, iii). This was later used as a bacteria-specific 371 signal to gate events representing pure bacterial cell population on the FSC-H/SSC-H scatter 372 plot that was labeled as dPOI (Fig 4, iv).

373
The validity of gating off blood debris using autofluorescence was confirmed in a 374 separate experiment in which the white blood cell-specific antibody anti-CD45 PerCP was 20 445 application of gating methods that removed the bacterial doublets during post analysis.

508
The newly developed PNA-FISH assay proved more sensitive than MALDI-TOF 509 analysis for detecting three bacterial species in BC, namely E. coli, K. pneumoniae and P.
510 aeruginosa, by four to five orders of magnitude. Here, we used a universal eubacterial 511 probe to optimize the assay. In future, we envisage the use of bacterial species-specific 512 probes for the most frequent bacterial species associated with sepsis.

513
The time to detection of FilmArray® was similarly effective as PNA-FISH for early 514 detection of K. pneumoniae and P. aeruginosa, however, this technology is PCR-based and 515 provides evidence only for the presence of bacterial nucleic acids. In the case of E. coli 516 detection, FilmArray® in common with other PCR-based assays, shows decreased sensitivity 517 due to a need to avoid false positives as it is a common contaminant of laboratory reagents. 518 A key advantage of the latter is the detection of bacterial nucleic acids within intact 519 bacterial cells despite a similarity in the time to detection between FilmArray® and PNA-520 FISH-AFC, and the ability of both methods to provide molecular identification.

521
A further advantage of our assay method is increased specificity and sensitivity of 522 detection of intact bacterial cells in BC that can reduce time to bacterial detection by 6 to 12 523 hours. The optimized protocol allows simultaneous set up of multiple samples with a simple 524 procedure for manual sample acquisition and processing. Overall, it takes 1.5 hours to 24 525 complete the FISH step with multiple samples, and an additional 3 minutes for AFC reading 526 per sample. In future, we plan to use multiple species-specific probes for early bacterial 527 identification in polymicrobial infections.

528
The novel diagnostic assay, PNA-FISH-AFC, described here has a potential to 529 significantly reduce the turnaround time of bacterial detection and identification in blood 530 culture in large clinical laboratories, and may allow blood culture autoanalyser systems to 531 use lower threshold values for earlier detection of bacteraemia. In septicaemic patients with 532 high bacterial loads in peripheral blood, this novel assay has the potential to identify the 533 causative agent(s) directly from the whole blood. Ultimately, the major potential of PNA-534 FISH-AFC is in early and enhanced bacterial detection in blood culture that will improve 535 survival and management of bacteraemic patients. 536 537