Cultural Isolation and Characteristics of the Blood Microbiome of Healthy Individuals

Background: On the analogy of the non-pathogenic microbiota found in oral cavity, skin and gastrointestinal tract, existence of blood microbiota was confirmed by DNA sequencing, but never deeply characterized. Hypothesis for the existence of dormant blood microbiota in healthy humans have been arisen and single species have been isolated. The aim of our study was to resuscitate and investigate the biodiversity of bacterial and fungal dormant blood microbiota in healthy individuals by blood culturing and NGS DNA sequencing. Results: Twenty eight blood samples of healthy individuals, seven for each blood type, were studied. Several culture media were tested. Blood microbiota resuscitation was performed in BHI broth supplemented with vitamin K 1 mg/ml, 2% sucrose, 0.25% sodium citrate and 0.2% yeastolate at 43 ̊C for 72 h. All tested blood samples were culture positive, as confirmed by Gram staining and TEM. TEM images demonstrated well defined cell structures. Analysis for bacterial and eukaryotic species was performed by 16S rRNA and ITS2 targeted sequencing. The obtained sequences were clustered (≥97% identity) in Operational Taxonomic Units (OTUs). Among cultured and uncultured samples we identified OTUs similarity with 47 bacterial orders belonging to 15 phyla and 39 fungi orders blonging to 2 phyla. For the first time we demonstrated isolation and sequencing identification of fungal blood microbiota in healthy individuals. Blood-group differences were identified among the bacterial microbiome compositions. Conclusion: The dormant blood microbiome is innate of the healthy individuals. Interventional strategies to bind the host blood microbiome with the states of health and disease remain an unmet research goal.


Introduction
In the human and animal evolution many microbial species successfully adapted to the macroorganism. Most of them could not be cultured and are proven indirectly by DNA sequencing. On the bases of sampling 242 healthy adults at 18 different anatomical sites of the human body and sequencing the 16S rRNA genes the presence of 5177 microbial taxonomic profiles was proven, but only 800 of them could be cultured [1]. The observation of cell-free DNAemia is well described feature of the healthy blood [2] [3], but the presence of transient culturable blood microbiota in the blood of healthy individuals could also be supposed. We tested whether the presence of DNA in healthy blood is associated with DNAemia or it is due to existing blood microbiota. It is proven that in the blood of clinically healthy individuals microorganisms could persist for many years without causing illness. The most investigated examples being the latent tuberculosis.
Nevertheless the blood microbiome is still an enigma, its existence was proven during the last 50 years. Indirect evidences for existence of bacteria residing in erythrocytes have been predicted in the past by radiometric methods [4]. In 1969 Tedeschi et al. reported on incorporation of nucleosides in human erythrocyte attributed to the metabolic activity of mycoplasma or bacterial L-forms [4]. In 1977, Domingue and Schlegel identified in 7% of the blood specimens from supposedly healthy individuals novel bacterial structures [5]. In 1993 the Bulgarian scientist Emil Kalfin experimentally proved by culturing and electron microscopy that microorganisms are multiplying in the erythrocytes of healthy people [6]. Kalfin reported 100% positivity of the blood cultures. Subsequently several other authors questioned the existence of the blood microbiome or DNAemia in healthy individuals, Domingue (1977) [5], (1997) [7], Nikkari et al. (2001) [8], Mc Laughlin et al. (2002) [9], Moriyama et al. (2008) [10], Markova (2015) [11], Damgaard (2015) [12] Dimova et al. (2017) [13], Gosiewski et al. (2017) [2] and Kowarsky et al. (2017) [3]. The authors reported supporting electron microscopy, cultural and molecular data in favour of the existence of the blood microbiota in healthy individuals. Hypothesis on the bacterial structure of the blood microbiota in healthy humans have been arisen [5] [6] [12], but cultures for exhaustive microbiota analysis remain a target. Rich bacterial diversity in the blood of healthy individuals was confirmed by 16S rRNA genes sequencing [14] and total RNA sequencing [15]. In 2016 Paise et al. demonstrated that a diversified microbiome exists in healthy blood. Most of the blood bacterial DNA was found located in the BC (93.74%), while RBCs contain more bacterial DNA (6.23%) than the plasma (0.03%) [14]. A significant number of bacterial species were also detected in the blood of healthy chickens [16] and cats  [17]. Resuscitation of dormant blood microbiota in healthy individuals has been tested and single bacterial species have been isolated on agar plates [12]. However, presence of eukaryotic microbial cells in the blood of healthy individuals has never been challenged. Goal of our study was to resuscitate the blood microbiota of healthy individuals and apply cultural, microscopic and NGS targeted sequencing methods for their characterization.
Here, we demonstrate that bacteria and fungi constitute a rich microbial diversity in the blood of healthy humans by applying a successful culture resuscitation strategy, DNA isolation and NGS targeted sequencing analysis. First preliminary results of the study have been presented as poster at the ASM Microbe congress 2017 [18].

Materials and Methods
Ethical committee approval of the study (decision 38/14.07.2016) by the Institute of Neurobiology, Bulgarian Academy of Sciences (BAS) and individual written consent were obtained.

Culturing
Blood of 28 healthy volunteers was collected in Vacutainer tubes with K 3 EDTA as anticoagulant (Vacutainer K3E, BD, USA), 7 samples per blood group. The blood samples were divided in two parts. One part for culturing (3 mL) and another part for direct DNA isolation (7 mL). All blood samples were tested for sterility by growing on Sabouraud and blood agar.
We applied a modified resuscitation strategy previously developed by Emil Kalfin [6]. Three ml of blood sample were added to 22 ml of culture medium.
Culturing was performed in sterile 50 ml polypropylene Falcon tubes (Corning Inc, USA). The culture base medium was composed by Brain Hearth Infusion (BHI, Difco, USA) medium and 0.2% yeastolate (Difco) adjusted at pH 6.8 and sterilized. Sterile (D+) sucrose at 10% final concentration and water-soluble form of vitamine K 3 -menadione sodium bisulfite (Sigma-Aldrich, USA) in concentration of 1 mg/ml sterilized by filtration were added to the base medium.
Resuscitation growth was induced at 43˚C at 24 h, but 72 h was found to be optimal time for growing blood microbiota in liquid cultures. The agar medium for subculturing contained 1.2% Noble agar (Difco, BD, USA). Isolated blood microbiota was confirmed by Gram staining and 16S rRNA genes PCR analysis.
Gram staining was directly applied on the cultured sample.

Transmission Electron Microscopy (TEM)
Blood samples for TEM were processed within 1 hour after collection. Sterile  To test the level of the ionizing radiation that the microbiota are able to withstand we applied gamma irradiation with 60 Co (Isledovatel MPX-γ-25M, TENEX, Moscow, RU) at 15, 20 and 25 kGy to 5 ml of blood from three healthy donors and erythrocyte concentrate of blood group A, Rh(+). The erythrocyte concentrate was a sample produced by collecting blood from several blood donors at the National Center for Transfusiology and Hematology in Sofia, Bulgaria. In order to control the quality of irradiation two control blood samples were spiked with Candida guiliermondii and Candida albicans with 10 9 cells. After irradiation with dose of 15 kGy, fungi inoculated control samples were tested for growth on Sabouraud agar. Control plates were negative after two weeks of culturing at 30˚C. tions were obtained with NucleoSpin Soil and NucleoSpin DNA Stool kits (Macherey-Nagel, Germany) and RIBO-prep kit (Ecoli s.r.o., Slovakia) applying the following essential modifications: After 72 h of culturing the tubes were centrifuged at 5000 g for 15 min. The pellet was washed by vortexing with 40 ml of DNase I 1 U/ml (Thermo Fisher Scientific, USA) in ddH 2 O and incubated for 30 min at 37˚C in order to eliminate residual DNA contaminations associated with lysed nuclear blood cells, DNA from the culture medium or DNA traces in water. Centrifugation and washings were repeated three times. These steps lyse the blood cells, wash away medium and human blood cells DNA/RNA contaminations. The cell pellet was resuspended in 2 ml of TE buffer (pH 7.5). The same procedure was applied for direct microbiota isolation from whole blood. Briefly, 7 mL of blood were lysed by vortexing with 40 ml of DNase I 1 U/ml in ddH 2 O and incubated for 30 min at 37˚C. After centrifugation at 5000 g for 15 min the same procedure was repeated three times. The pellet from the lysed whole blood was resuspended in 400 µl of TE buffer. One hundred µl of cell suspension was processed according to the manufacturer instructions and modifications. The cell suspension was subjected eight times to freeze-thawing in liquid nitrogen and dry bath at 96˚C for 10 min combined with vigorous vortexing or optional homogenization with ceramic/zirconium 0.6 -0.8 mm beads on a bead beater for 5 min (Biospec Products, USA). The extracted DNA was resuspended in 100 µl of TE and repurified by addition of 5 µl of 20% Chelex 100 (Bio Rad, USA). Suspension was vigorously vortexed for 1 min and centrifuged for 10 min at 15.000 × g. The supernatant, ~95 µl was recovered in a new tube without touching the Chelex 100 pellet. DNA typical yield was >150 ng/µl with a 260/280 nm ratio of ~1.7. For DNA analysis we applied 16S rRNA genes and ITS2 targeted sequencing on Illumina MiSeq (Illumina Inc., USA). DNA sequencing was performed at IMGM Laboratories GmbH (Martinsried, Germany). For bacteria identification V3 -V4 hypervariable regions of the 16S rRNA genes were amplified with universal primers 314F and 805R. For fungi identification ITS2 region was amplified with universal primers ITS3 and ITS4. The obtained sequences were clustered (≥97% identity) into Operational Taxonomic Units (OTUs). OTUs were analyzed for sequence similarities against reference sequence databases; SILVA database for 16S rRNA gene (https://www.arb-silva.de/) and UNITE for ITS2 (https://unite.ut.ee/). Common sequences between the samples and the negative (dH 2 0) and culture medium DNA controls were excluded from analysis. These were considered as contaminants from reagents and culture medium. For taxonomic affiliation the interference created by the MiSeq sequencer as well as those created between samples was also considered.

Results and Discussion
All tested blood samples were culture positive, as confirmed by Gram staining and TEM (Figure 1). Optimal growing temperature was at 43˚C. At 37˚C growth of the blood microbiota was suppressed. However, to our knowledge  evidences completely lacking that microbiological isolation of the blood microbiome in healthy subjects is possible within 48 hours. The applied methodology of isolation in liquid culture media is technically affordable and repeatable, although we did not found appropriate conditions to maintain viable the blood microbiota in subcultures. In subcultures the blood microbiota revert to their dormant state. We did not identified the factors responsible of this process. We observed several morphological forms of Gram stained blood microbiota, previously described by Domingue [5] [7] as "dense body" of Gram positively stained microbial core structure surrounded by Gram negative coat and chains of Gram negatively stained "primary cells" (Figure 1). TEM images demonstrated well defined cell structures (Figure 1). TEM and culture results confirmed that the blood microbiome represents viable structures rather than debris resulting from degradation of blood elements, such as lipids or haemoglobin complexes. We suppose that the "dense body" is a compact mass composed by "primary cells". Microscopic evidences are shown on Figure 3(c). Microscopic slides have poor stainability for DNA with Acridine Orange (AO). Impermeability is probably due to difficult stain penetration even after 15 min of staining in 1 mg/ml of AO.
After optimizing growth media and temperature conditions for liquid cultures sheep blood. We found that sheep blood agar was not suitable for isolation of human blood microbiota. Growth was not visible with naked eye even after prolonged cultivation for more than two months. Isolation strategy was changed by preparing agar plates with homologous human blood. Agar plates prepared with 10% blood of the same individual used for inoculation of liquid culture gave partial growth. Plates were sealed in plastic bags to diminish drying of agar and culturing was performed at 43˚C. After 30 days of culturing brown colonies, with no odour and diameter of 0.5 -1 mm were observed ( Figure 2). Longer periods of culturing do not give increase in colony size, due to the fact that agar is drying and somehow growth seems inhibited or much slower. Agar plates with 10%, 20% and 80% of homologous blood were tested and there was no difference in growth efficiency. The bacterial colonies were confirmed visually and by Gram. Isolation of colonies on agar media of the blood microbiota from healthy individuals was partly successful and still challenging. Subcultures are unstable. Evaluation of microbial cell lysis or disintegration was performed microscopically after Gram staining. On the slides intact "dense bodies" were observed. Increasing the time for disintegration with ultrasound for 2 h brought to unexpected finding. Cells observed with yeast like shape on light microscopy and on electron microscopy as "dense bodies" at different stages of maturation were disintegrated to single "elementary bodies" with streptococcal appearance (Figures 3(a)-(c)). We conclude that the "dense bodies" give origin to elementary structures of "primary cells" with the size of 0.1 -0.2 µm. Integrity of the "dense body" cells in fresh cultures was tested by treatments with 4M guanidine thiocianate, 10% NaOH, 10% KOH, 10% CH 3 COOH, 10% HCl and 10% H 2 SO 4 for 30 min at 37˚C, 0.5 M EDTA for 4 days at room temperature and freeze-thawing  (Figures 3(g)-(k)).
Blood microbiota resisted to high dose of 25 + 25 kGy gamma irradiation with 60 Co (Figures 3(d)-(f)). Growth of radioresistant blood microbiota was detected after two weeks. This is not surprising, because the majority of soil bacteria are killed by 20 kGy, however dose higher than 70 kGy may be required to kill certain radio-resistant bacteria [19].
For DNA analysis we applied 16S rRNA gene targeted sequencing of V3-V4 hypervariable region on Illumina MiSeq (Illumina Inc., USA). DNA sequencing was performed at IMGM Laboratories GmbH (Martinsried, Germany). The resulting 2 × 300 bp reads were demultiplexed, quality controlled and merged into continuous reads. Further bioinformatics analysis including clustering, phylogenetic analysis and alpha and beta diversity calculation was performed with the CLC genomics workbench and its microbial genomics module. In order to correctly describe the microbial biodiversity on phylum and order level, four filtering steps were applyed: 1) only Operational Taxonomic Units (OTUs) exceeding similarity of >97% on genus level were included in the analysis, 2) only OTUs exceeding a combined abundance among all tested samples of 100 were included, 3) OTU reads per sample below 1% abundance at order level were excluded and 4) exclusion of common sequence reads originating from exogenous DNA. The sources of these contaminants are reagents used in DNA extraction, culture medium, PCR reagents, and next-generation sequencing library preparation. The sequences, available after trimming and merging showed a high amount of high quality reads, suitable for downstream analysis. Among the cultured and non-cultured samples sequencing analysis identified rich biodiversity of OTUs similarity (>97% on genus level) with 47 bacterial orders belonging to 15 phyla and 39 fungi orders blonging to 2 phyla. The bacterial orders were predominantly of phylum Proteobacteria. Among non-cultured samples Proteobacteria were 93%, Actinobacteria 2%, Planctomycetes 2% and Firmicutes 2%, while among cultured samples Proteobacteria were 46%, Firmicutes 25%, Actinobacteria 14%, Bacteroidetes 6%, Fusobacteria 3% and Cyanobacteria 2%. The fungi phyla Basidiomycota, Ascomycota and unidentified fungi were 64%, 22% and 14% respectively among the non-cultured samples, while among the cultured samples they were 27%, 31% and 41% respectively. We note, that some phyla are enriched by the applied resuscitation strategy according to the OTUs combined abundancy (Table 1). Other studies also identified Proteobacteria as predominant in the blood of healthy individuals [14], healthy broilers [16] and cats [17].
Our results demonstrate that isolation of the blood microbiota is technically affordable by conventional means. We identified, that in 100% of the blood samples the blood microbiota resuscitate in 24 -48 hours under stress by culturing at 43˚C in high concentrations of vitamin K. In vitro we identified by microscopy that the most common mechanism of formation and reproduction of the "dense body" is by production in chains Gram negatively stained "primary cells", as previously described by Domingue (1977) [5], (1997) [7]. We observed loss of the Gram negative component of the "dense body" after treatment with 10% NaOH. Evidences for intraerythrocytic presence of normal blood microbiota were reported by Kalfin in 1997 [6]. We identified free circulating microbiota, "primary cells", by electron microscopy in blood samples processed within the first hour after collection (Figures 1(e)-(g)). We suppose that the "electron dense body" is composed by Gram negative and Gram positive 'primary cells'. Currently applied approaches for biodiversity analysis of blood microbiota use microbe-derived DNA or RNA that may have crossed into the blood from the gut. The proposed by us culture resuscitation approach overcomes these limitations. We showed that blood DNAemia in healthy individuals is associated wtih the presence of culturable microbiota in the blood. To prove this, cell-free DNA circulating in the blood (DNAemia) was eliminated by  (Table 1). It could be concluded that the applied resuscitation culture conditions enrich specific orders while others remain unchanged. Moreover we predict that applying other growth conditions or media will change the composition of the cultured microbiota in the sample.
When grouping the samples according the blood type we identified significant differences. In non-cultured samples of individuals of blood type A we identified less number of reads but higher bacterial diversity at order level as compared to the other blood types ( Table 2).
The identified taxa in the blood have been reported in previous studies of the blood microbiota in healthy individuals or animals. Mangul et al. (2016) identified 23 bacterial taxa at pylum level applying total RNA sequencing [16].
In a recent study Chen et al. (2017) showed that the uterus and the fallopian tubes, which are generally believed to be sterile, are home of rich microbiota indicative of a non-sterile environment [20]. Table 2. Bacterial blood microbiota according to the blood group. (a) Number of reads according to blood group and order in non-cultured blood samples; (b) Graphical distribution of the bacterial orders.  the blood microbiome field will focus on defining the mechanisms underpinning the interactions between microbes and host, mechanisms influencing the initiation and progression of diseases with a view towards personalization of patient's diagnosis and treatment.