The development and application of a molecular community profiling strategy to identify polymicrobial bacterial DNA in the whole blood of septic patients

Abstract

Background: The application of molecular based diagnostics in sepsis has had limited success to date. Molecular community profiling methods have indicated that polymicrobial infections are more common than suggested by standard clinical culture. A molecular profiling approach was developed to investigate the propensity for polymicrobial infections in patients predicted to have bacterial sepsis.

Results: Disruption of blood cells with saponin and hypotonic shock enabled the recovery of microbial cells with no significant changes in microbial growth when compared to CFU/ml values immediately prior to the addition of saponin. DNA extraction included a cell-wall digestion step with both lysozyme and mutanolysin, which increased the recovery of terminal restriction fragments by 2.4 fold from diverse organisms. Efficiencies of recovery and limits of detection using Illumina sequencing of the 16S rRNA V3 region were determined for both viable cells and DNA using mock bacterial communities inoculated into whole blood. Bacteria from pre-defined communities could be recovered following lysis and removal of host cells with >97% recovery of total DNA present. Applying the molecular profiling methodology to three septic patients in the intensive care unit revealed microbial DNA from blood had consistent alignment with cultured organisms from the primary infection site providing evidence for a bloodstream infection in the absence of a clinical lab positive blood culture result in two of the three cases. In addition, the molecular profiling indicated greater diversity was present in the primary infection sample when compared to clinical diagnostic culture.

Conclusions: A method for analyzing bacterial DNA from whole blood was developed in order to characterize the bacterial DNA profile of sepsis infections. Preliminary results indicated that sepsis infections were polymicrobial in nature with the bacterial DNA recovered suggesting a more complex etiology when compared to blood culture data.

Fig. 1

The addition of 0.85 % saponin to bacterial cells does not impact viability. The average CFU/ml and the standard deviation were plotted for each sample prior to and after the incubation with 0.85 % saponin (represented by “sap” in the figure label) for 1 h at room temperature either with ambient air or in an anaerobic chamber for F. necrophorum and P. melaninogenica. Each experiment was done in triplicate with samples recorded in duplicate. There were no statistically significant differences found in the viable cell count of the bacteria (two-tailed Students t-test, p > 0.05 % significance)

Fig. 2

Enzymatic digestion requirement for bacterial DNA extraction of whole blood. Whole blood was spiked with a CBA plate pool from ASN087 BAL sample. Following a treatment with saponin, the resulting pellet was either digested with no enzyme (black), lysozyme (grey), or lysozyme and mutanolysin (blue). More T-RFs were present when enzymatic digestion was done with lysozyme (10, grey) and lysozyme plus mutanolysin (24, blue) identified, respectively, when compared to no enzymatic treatment with only 7 T-RFs present

Fig. 3

Limit of detection for synthetic communities of bacteria spiked into whole blood. The CFU/ml of each bacterium in the community was determined prior to blood spiking (solid black bars). The CFU/ml of bacteria recovered after each step in the saponin-blood treatment protocol was determined; addition of 0.85 % saponin with no further washes (solid grey bars), addition of saponin at 0.85 % with the addition of a 1 ml sterile double distilled water wash (diagonal lined bars), or addition of saponin at 0.85 % with two washes with 1 ml sterile double distilled water (solid white bars). Each organism was spiked into whole blood alone to verify the limit of detection observed in Table 3 (D)

Fig. 4

OTU abundance of 16S rRNA Illumina sequenced DNA from synthetic communities. Taxonomic summaries for the synthetic community samples after each step in the saponin blood-treatment protocol were compared. Each bar represents the total PCR amplified DNA sequenced for the sample and the relative abundance of each OTU in the molecular profile. The representative sequence for each OTU was aligned to both the NCBI and HOMD 16S databases in order to determine what synthetic community organism they represented. E. hormaechei sequences were represented by the Enterobacteriaceae OTU, S. aureus was represented by the Staphylococcus OTU, K. pneumoniae by the Klebsiella OTU, E. coli by the Escherichia OTU, S. pneumoniae and S. intermedius by the Streptococcus OTU, N. flava by the Neisseria OTU, M. luteus by the Actinomycetales OTU, F. necrophorum by the Fusobacterium OTU, and P. melaninogenica by the Prevotella OTU. All OTUs with sequence alignments that could not be correlated to the bacteria spiked into the synthetic community were combined into “Other OTUs”, which accounted for 20–40 % of the total OTU abundance

Fig. 5

The bacteria DNA profiles of healthy blood. Whole blood was collected from 10 adult donors that worked in a health care setting but were healthy at the time of sampling. Two negative template controls (NTC) and sterile PBS were included for comparison. Taxonomic summaries for the blood samples were compared. Each bar represents the total DNA sequenced for the sample and the percent relative abundance of each OTU identified (a). The bacterial DNA profile profiles of the HB samples were similar to each other but distinct from the NTCs and PBS samples. The letter in front of each taxonomic group indicates the level of taxonomic depth with p__ representing phyla, f __representing family, o__ representing order, and g__ representing genus (a). Principal coordinates analysis (PCoA), based on weighted UniFrac, indicated the healthy blood samples (red) clustered separately from septic blood samples (green) (b)

Fig. 6

Bacterial DNA profiles of case studies from septic ICU patients. Taxonomic summaries of the bacterial DNA extracted from the primary infection sample and saponin treated whole blood from each case study patient. Each bar represents the combined results from two separate experiments with each PCR sample amplified in triplicate. The size of the bar indicated the percent relative abundance with the taxonomic identification labeled for each major group. The letter in front of each taxonomic group indicates the level of taxonomic depth with p__ representing phyla, f __representing family, o__ representing order, and g__ representing genus. The table indicated the comparison of the clinical diagnostic culture results to our culture results