Impact of monocytic cells on recovery of uncultivable bacteria from atherosclerotic lesions


Dr Emil Kozarov, Columbia University, 630 W 168 St, P&S Box 20, New York, NY 10032, USA.
(fax: 212-305-9313; e-mail:


Abstract.  Rafferty B, Jönsson D, Kalachikov S, Demmer RT, Nowygrod R, Elkind MSV, Bush Jr H, Kozarov E. (Columbia University Medical Center, New York, NY; and Weill Cornell Medical College, New York, NY; USA) Impact of monocytic cells on recovery of uncultivable bacteria from atherosclerotic lesions. J Intern Med 2011; 270: 273–280.

Objective.  Epidemiological evidence suggests that infections may contribute to atherogenesis. However, with the exception of Chlamydophila pneumoniae, cultivable bacteria have not been recovered from atherosclerotic lesions. Therefore, we aimed at developing an approach to recover uncultivable bacteria from atherectomy tissues.

Methods.  We cultured homogenates from atherectomy specimens from seven nonseptic patients undergoing surgery for arterial obstruction either alone or together with THP-1 monocyte-like cells. We performed 16S rDNA analysis, biochemical tests, random amplification of polymorphic DNA PCR analysis, quantitative polymerase chain reaction (qPCR) and immunohistofluorescence to identify the cultivated bacteria. Wilcoxon signed-rank tests were used to determine whether THP-1 treatment yielded a higher number of isolates than did the untreated controls.

Results.  We recovered more bacteria from cocultures of atherectomy specimens with THP-1 cells than atherectomy specimens cultured alone. On average, tissue homogenates incubated with THP-1 cells versus control yielded 124 vs. 22 colony-forming units, a median of 140 vs. 7, respectively (= 0.02). We recovered 872 isolates of limited number of species, including Propionibacterium acnes, Staphylococcus epidermidis and Streptococcus infantis and the fastidious anaerobe Porphyromonas gingivalis, and confirmed its presence in tissue using double immunofluorescence imaging. qPCR demonstrated the presence of ≥3.5 × 103P. gingivalis genomes per gram of atheromatous tissue.

Conclusions.  These results indicate that viable previously uncultivable bacterial species are present within atheromas. Our results suggest revisiting the hypothesis that infections may have a causative role in atherosclerotic inflammation and have implications for research regarding novel diagnostics and treatments for cardiovascular disease.


Atherosclerosis is a chronic inflammatory condition associated with hypercholesterolaemia, hypertension, diabetes, smoking and genetic factors. Although it has been shown that these pro-inflammatory risk factors are highly prevalent amongst individuals with cardiovascular disease (CVD), the incidence of atherosclerosis is not fully explained by them [1]. An abundance of serological and epidemiological evidence supports a bacterial and viral infectious contribution towards the initiation and/or progression of atherosclerosis [2]. For example, Chlamydophila pneumoniae was cultivated from atheromas [3, 4], and DNA data suggest that various pathogens are associated with atherosclerotic tissue [5, 6]. Until now, however, no other bacteria could be cultivated from atheromas [7].

It has been suggested that large randomized controlled clinical trials of antibiotics in patients with pre-existing CVD [8] may have failed, in part, because the tested antibiotics lacked activity against noncultivable bacteria. Chronic infection diseases such as periodontitis may underlie the failure of antibiotics to prevent acute ischemias. In concordance, clarithromycin treatment reduced recurrent cardiovascular events only in subjects without periodontitis, but failed in subjects with periodontitis most likely because it is not functional for periodontal therapy, lacking effect on dental biofilm microbiota [9]. Moreover, the focus on one organism (C. pneumoniae) was likely insufficient given the mounting evidence that multiple species are likely to coexist in atherosclerotic tissue [5, 6].

We hypothesized that oral flora bacteria may become uncultivable when they reside in cells of atheromatous tissues. Indeed, monocytes/macrophages, which are prominent residents of atheromatous tissues, harbour many facultative intracellular pathogens (e.g. Legionella, M. tuberculosis). Moreover, uncultivable Legionella sp. have been made cultivable by coincubation with amoeba [10], and Tropheryma whippelii has been grown in vitro by coincubation with human monocytes [11]. Therefore, we examined whether the human monocytic cell line THP-1 can facilitate recovery of uncultivable bacterial species from human atherectomy tissues.

Materials and methods

Processing of atheromatous tissue

We obtained carotid and femoral atherectomy tissues at surgery from seven patients (men and women aged 58–79 years, see Table 1), four women, three men, average age 67.5 years, undergoing this procedure to relieve vascular obstruction. The specimens were obtained at surgery according to a protocol approved by Columbia University’s Institutional Review Board. Carotid or femoral atherectomy specimens (∼100 mg wet weight) were obtained and immediately washed in sterile saline (4 °C) to remove contaminating blood, homogenized in 5 mL of RPMI medium (4 °C) using a glass homogenizer and sonicated for 30 s at 4 °C until no intact nucleated cells were visible by phase microscopic examination. A 625-μL aliquot of the homogenate was added to 1.375 mL of medium containing 105 THP-1 cells (ATCC, Manassas, VA, USA), the mixture was placed in a 15.6-mm-diameter tissue culture-treated 24-well plate and incubated for 3 h at 37 °C in a 95% air per 5% CO2 atmosphere. The suspension was centrifuged for 7 min at 800 g, and after concentration via high-speed centrifugation, the entire supernatant was plated on fastidious anaerobic agar (FAA) plates for 14 days at 37 °C under aerobic and anaerobic conditions. The pelleted THP-1 cells were re-suspended in 2 mL of PBS, washed with this volume of medium two times by centrifugation at 800 g for 7 min, re-suspended in 1 mL of water to lyse the THP-1 cells, and after concentration via high-speed centrifugation, the entire lysate was plated. Mendonca [12] reported that noncultivable Listeria could be recovered only under strictly anaerobic conditions. Accordingly, we plated the lysate on FAA plates and incubated them under anaerobic conditions at 37 °C for 14 days. As a control, a second 625-μL aliquot of the atherectomy specimen homogenate was added to 1.375 mL of PBS, incubated for 3 h at 37 °C in a 95% air per 5% CO2 and plated on FAA plates as described previously. As an additional control, THP-1 cells not incubated with homogenized atherectomy tissue were processed in parallel in the same manner. All procedures were carried out in an alcohol- and UV-light-sterilized laminar flow hood. We used the Wilcoxon signed-rank test to compare the numbers of colonies of bacteria recovered from coculture of atherectomy homogenate with pelleted THP-1 cells, the supernatant of the pelleted THP-1 cells and the homogenate mixed with PBS.

Table 1.   Impact of incubation of homogenates of atherectomy tissues with versus without THP-1 cells on recovery of bacteria from atherectomy specimens from seven different nonseptic afebrile patients
Patient no.Age, genderSource of atherectomy specimenCFU bacteria from homogenates incubated with THP-1 cellsCFU bacteria from homogenates incubated without THP-1 cells
158, maleCarotid artery2561
273, femaleCarotid artery711
379, maleCarotid artery23
467, femaleFemoral artery1406
566, femaleFemoral artery15913
666, maleCarotid artery132115
764, femaleFemoral artery1767
Average or total67.5 years 872156
Outline of the procedure described above
1. Plaque tissue homogenization. Plating nontreated controls on FAA plates, incubation.
2. Treatment: 3-h coincubation with THP-1 monocytes. Centrifugation at 800 g for 7 min.
3. Plating supernatant on FAA, incubation.
4. Lysis of cell pellet with water, plating on FAA, incubation.

Biochemical identification of the clinical isolates

Clinical microbiology analysis of bacterial strains was performed with Advanced Expert System software in conjunction with the VITEK 2 test system (BioMérieux Inc., Hazelwood, MO, USA) in Columbia University’s Clinical Microbiology Service.

Bacterial genomic DNA isolation, 16S rDNA gene amplification, GenBank and phylogenetic analysis

Genomic DNAs from twenty P. gingivalis and two P. acnes clinical isolates were isolated, and 1516-bp segments of 16S rDNAs were amplified using Platinum Blue PCR Supermix (Invitrogen, Carlsbad, CA, USA) with universal 16S rDNA primers 9F (5′-GAGTTTGATYMTGGCTCAG-3′) and 1541R (5′-AAGGAGGTGWTCCARCC-3′)[13] using initial denaturation at 94 °C for 2 min followed by 40 cycles (94 °C per 30 s, 52 °C per 30 s, 72 °C per 1 min) with a final primer extension for 5 min at 72 °C. The resulting 1.5-kb PCR fragments from two P. gingivalis isolates, CU1 and CU2, were sequenced on 3730xl DNA Analyzer (Applied Biosystems, Foster City, CA, USA) with the same primers and with an additional internal primer 530R (5′-GTGCCAGCAGCCGCGG-3′). The complete sequences of the 16S rDNA segments were assembled and further analysed using NCBI (, Ribosomal Database Project (RDP) 16S RNA classifier [14] at the Ribosomal Database project website ( and Green Genes at Lawrence Berkeley National Laboratory ( In addition, all P. gingivalis isolates were analysed using species-specific 16S rDNA PCR, as described [6].

RAPD-PCR analysis

For genetic polymorphism analysis of the P. gingivalis isolates, random amplification of polymorphic DNA (RAPD) PCR analysis was carried out as described [15], with primer #01-05 and 100-ng DNA template (seven isolates, the laboratory strain W83 and five strains from the collection: 33277, 381, AJW2, AJW4 and A7436).

Isolation of total genomic DNA from the tissue

Genomic DNAs from atheromas were isolated as previously described [6].

Tissue analysis for P. gingivalis DNA using quantitative polymerase chain reaction (qPCR)

We tested total DNA from the atheromatous tissue for P. gingivalis DNA using 16S rDNA and ragA gene amplification ( The primers (forward, 5′-CATAGATATCACGAGGAACTCCGATT-3′ and reverse, 5′-AAACTGTTAGCAACTACCGATGTGG-3′) [6] define a 436-bp amplicon of the 16S RNA sequence of P. gingivalis, and the primers (forward, 5′-TTTGCCTGAACACAGAGTCG-3′ and reverse, 5′-GACTGCTTTTCCCACGAGAG-3′) define a 369-bp amplicon of the ragA gene. A series of 10× dilution calibration samples were used starting at 500 pg μL−1 down to 0.5 fg μL−1 or ∼0.16 mol L−1 equivalents of target DNA per microlitre. Three replicates containing 3 μL (100 ng) of the DNA isolated from the atheromatous samples were amplified in real-time PCR against three replicates of serially diluted P. gingivalis calibration samples and six template-free controls using 17 μL of 1× Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen) and amplified for 40 cycles using an ABI 7000 Sequence analysis system with the default protocol. Upon the completion of the protocol, the levels of the positive signals obtained in the PCR with atheromatous samples were compared with tissue-free controls using Student’s t-test. The specificity of each amplicon was interrogated with melting curve analysis, with amplicon size analysis using Agilent Bioanalyzer and – finally – by sequencing.

Metagenomic screening of diseased and healthy control tissue pairs

As controls, five pairs of human diseased and matching adjacent healthy (D/H) post-mortem tissue were analysed for the presence of eubacterial DNA. The specimens were obtained according to approved institutional review board protocol. The analysis was performed using 16S rDNA amplification as follows. A 1516-bp segment of 16S rDNA was amplified from the genomic DNA templates using Advantage 2 Polymerase mix (Clontech, Palo Alto, CA, USA) with universal 16S rDNA primers 9F (5′-GAGTTTGATYMTGGCTCAG-3′) and 1541R (5′-AAGGAGGTGWTCCARCC-3′) [16]. A ‘no template control’ reaction was included in the experimental protocol to monitor for the external contamination of the samples.

Immunohistofluorescent detection of P. gingivalis in a plaque specimen

Carotid specimen no. 1 was embedded in OCT medium (Miles, Inc., Indianapolis, IN, USA) and frozen at −70 °C. Blocks were cut to 5-μm-thick pieces using a cryostat, and the ribbon of tissue was put on a poly-L-lysine-coated slide and dried. The slides were fixed in acetone and blocked overnight at 4 °C in 5% bovine serum albumin and 1% dry milk. For staining, mouse anti-P. gingivalis IgG1 antibody 61BG1.3 [17] (kindly provided by R. Gmür) was diluted 1 : 50. Slides were incubated at 37 °C for 1.5 h, washed in PBS and treated with the secondary Alexa Fluor 456-conjugated goat anti-mouse antibody (4 μg mL−1; Invitrogen) for 1 h. Unrelated mouse IgG1 isotype controls (Santa Cruz Biotechnology, Santa Cruz, CA, USA) were also used. To confirm the P. gingivalis staining specificity, the specimens were double stained using a rabbit anti-P. gingivalis serum [18] diluted 1 : 50 for incubation at 37 °C for 1.5 h. The slides were washed in PBS, incubated in the secondary goat anti-rabbit Alexa Fluor 488-conjugated antibody (A11008, 4 μg mL−1; Invitrogen) for 1 h and viewed in a fluorescent microscope Nikon (Melville, NY, USA) Eclipse 80i. As additional control, slides were processed without primary rabbit antibody.

In addition, immunohistochemical detection of P. gingivalis in the tissue was performed in a similar fashion using the monoclonal anti-P. gingivalis antibody as described previously, with a goat anti-mouse AP-conjugated secondary antibody (Zymed, San Fransisco, CA, USA). The slides were further incubated with 5-Bromo-4-chloro-3-indolyl phosphate (Sigma, St Louis, MO, USA) for 1 h, stained with Meyer’s Hematoxylin (Sigma) for 3 min and washed for 3 min in running water. As control, same isotype unrelated antibody was used.


We recovered 5.6-fold more bacterial colonies from homogenates of the seven atherectomy specimens incubated with THP-1 cells than from the same homogenates incubated in medium without THP-1 cells (872 vs. 156 colonies) (Table 1). Ninety-eight per cent of the recovered 872 bacterial colonies (from THP-1 cells that had been incubated with homogenates of atherectomy tissues) were from 5 of the 7 patients, and 74% of the 156 bacterial colonies recovered from controls (homogenates incubated in medium without THP-1 cells) were from 1 of the 7 patients (Fig. 1). This is consistent with previously observed inter-individual variation [5]. The ratio of bacterial colonies recovered from homogenates from 4 of the 7 patients incubated with versus without THP-1 cells was 27 : 1 (Table 1).

Figure 1.

 Left, number of colony-forming units isolated from each of seven surgical atherectomy specimens from seven different patients. THP-1 treated = homogenates of atherectomy tissues incubated with THP-1 cells for 3 h at 37 °C and then plated on fastidious anaerobic agar (FAA). Control = an equal volume of atherectomy tissue homogenate incubated without THP-1 cells for 3 h at 37 °C and plated on FAA agar. Right, mean (±SEM) number of isolates recovered with THP-1 cells versus without THP-1 cells (p for difference = 0.02 using Wilcoxon signed-rank test). Pt, patient.

To determine whether the recovered bacteria were intracellular, a control experiment, with an additional step, was carried out in parallel. Before lysis, the THP-1 cells were treated with antibiotics (gentamicin, 300 μg mL−1, and metronidazole, 200 μg mL−1) for 1 h to kill all bacteria that had not invaded. The resulting numbers of recovered intracellular bacteria were matching those of nonantibiotic-treated specimen, indicating that all were intracellular.

We used the Wilcoxon signed-rank test to determine whether the THP-1 cocultivation method increased the recovery as compared to the PBS suspension method. On average, atherectomy homogenates incubated with THP-1 cells versus in PBS without THP-1 cells yielded 124 vs. 22 colony-forming units (CFUs), a median of 140 vs. 7, respectively (= 0.02) (Fig. 1). THP-1 cells alone did not produce any bacterial colonies. THP-1 incubation with atherectomy homogenate from patient 1 followed by lysis and incubation in anaerobic environment produced 20 black-pigmenting colonies in addition to 236 other colonies of different morphologies. Rapid illumination of the black-pigmented colonies with long-wave UV demonstrated no fluorescence, suggesting the species was P. gingivalis.

Indirect immunofluorescence with both mouse and rabbit anti-P. gingivalis antibodies further confirmed the presence of P. gingivalis in sections of atherosclerotic plaque from patient 1 (Fig. 2). The preservation of the specimens was not of sufficient quality to determine whether these bacteria were contained within human cells within the plaques.

Figure 2.

 Detection of P. gingivalis in carotid atheromatous tissue sections using double immunohistofluorescence and two different antibodies. Images are representative; three separate slides yielded similar results. (a) Primary antibody, mouse anti-P. gingivalis IgG1 monoclonal antibody 61BG1.3; secondary antibody, Alexa Fluor 456-conjugated goat anti-mouse antibody (red). (b) Primary antibody, rabbit anti-P. gingivalis serum; secondary antibody, goat anti-rabbit Alexa Fluor 488-conjugated antibody (green). (c) Colocalization of both stains is represented as overlap of the preceding images. Negative controls (not shown): primary mouse unrelated IgG1 isotype control antibody and primary rabbit antibody (omitted). Scale bar, 10 μm.

Biochemical analysis of the isolates was performed with Advanced Expert System software in conjunction with the vitek 2 test system (BioMérieux Inc.) in the Columbia University Clinical Microbiology Service. Further, 16S rDNA sequencing of an additional 32 amplicons from nonpigmenting isolates from different phenotypes identified them as P. acnes, S. epidermidis and S. infantis. Amplification with universal primers of genomic DNA from six black-pigmenting isolates followed by sequencing and nucleotide database analysis confirmed they were P. gingivalis (see Materials and methods). To assess the genetic relatedness between the recovered P. gingivalis isolates, we used RAPD-PCR analysis [15] (Fig. 3). The analysed P. gingivalis isolates gave identical patterns and displayed on 1% agarose gel a pattern different from that of the W83 strain used in our laboratory, and from the other strains in the tested collection (data not shown). All 20 black-pigmenting isolates from patient 1 were positively identified as P. gingivalis with species-specific 16S rDNA PCR (data not shown).

Figure 3.

 RAPD-PCR profiles of amplification using DNA template from P. gingivalis isolates recovered form atheromatous tissue. Lanes 1 and 2, profile of P. gingivalis clinical isolates CU1 and CU2. Lane 3, control P. gingivalis W83. MW, molecular mass marker with major bands indicated.

Quantitative PCR analysis of total atheromatous tissue DNA of patient 1 demonstrated the presence of at least 3.5 × 103P. gingivalis genomes per gram tissue. We were unable to process for cultivation nondiseased tissue sample controls from patients for ethical reasons. We have, however, analysed arterial tissues from atherosclerotic lesions and normal arteries of nonsurvivors (obtained 24–48 h after death) using 16S rDNA PCR. The metagenomic analysis of the 16S amplicons from atheromatous tissue and matching healthy control from the same patient identified a higher presence of bacterial species in diseased tissues, as compared to healthy controls (15 vs. 7). These data are in line with the reported prevalence of bacterial DNA in diseased tissues compared with healthy control material from 26 individuals [5].


A large body of evidence links chronic infections to CVD [19, 20], Nevertheless, the evidence for causality is still lacking [21]. This has led to two hypotheses regarding the role of periodontal bacteria in atherogenesis. Our current findings advance this science considerably by informing a direct infection hypothesis. It posits that periodontal bacteria (e.g. P. gingivalis) are phagocytosed by monocytes in oral tissues and carried within them via the blood to atheromatous lesions (the ‘Trojan horse’ model). According to this hypothesis, these bacteria persist within lesional monocytes/macrophages and stimulate them to secrete pro-inflammatory substances (e.g. TNFα). Phagocytic cells have been shown to harbour viable C. pneumoniae whilst at the same time protecting the intracellular bacteria from antibiotic treatment [22]. A variant of the direct invasion hypothesis is that periodontal bacteria invade the bloodstream and metastasize without the aid of monocytes to atheromatous lesions.

The second, or indirect hypothesis, posits that chronic low-grade ‘systemic’ inflammation, induced by persistent periodontal bacterial infection, can cause endothelial dysfunction and initiate and perpetuate atherogenesis. Consistent with both hypotheses, a recent randomized clinical trial demonstrated sustained reductions in endothelial dysfunction after periodontal therapy despite limited long-term effects on systemic inflammatory markers [23].

The most direct way to distinguish between the direct and indirect hypotheses is to determine whether bacteria from the oral flora are present within atheromatous tissues and whether they remain viable and/or grow within cells that participate in atherogenesis. Accordingly, using quantitative PCR analysis, we, [6, 24–27] and others, [5, 28, 29] have identified bacteria or bacterial DNA in the atheromas, shown bacteria can persist in vascular cells for extended periods of time and demonstrated they directly initiate and/or exacerbate atherogenesis. Bacterial DNA was mainly found in diseased vascular tissues but not in healthy tissues [5]. Serum IgA antibody titres to Aggregatibacter actinomycetemcomitans or to C. pneumoniae heat shock protein 60 were associated with coronary obstruction [30]. Whilst these studies provide support for the direct hypothesis, they provide no insight into the underlying mechanisms.

The cultivation and identification of a number of bacterial species in diseased vascular tissue is a necessary advancement of the direct biological hypothesis in addition to fulfilling the Koch’s postulate. Admittedly, Koch’s postulates might not be entirely appropriate in the context of a hypothesis that allows for multiple causative agents and suggests that infections alone are unlikely to be either sufficient or necessary causes of CVD. Chronic respiratory, urinary tract, dental and other infections increased the risk of carotid atherosclerosis according to a large epidemiological study [31]. No single infectious organism has so far been identified as causative agent of atherosclerosis; therefore, summary measures of multiple chronic infectious exposures, or ‘infectious burden’, have been associated with the risk of stroke and atherosclerosis affecting the carotid arteries [32]. This burden includes C. pneumoniae or viral infections such as herpes simplex virus, hepatitis virus and influenza [30, 33]. Nevertheless, the isolation from diseased human vascular tissue of a microbial species a priori hypothesized to have atherogenic potential is an important step forward. Moreover, the specific periodontal organisms identified are noteworthy. Accumulating evidence has implicated specific infectious agents in the progression of atherosclerosis, including the oral pathogen P. gingivalis [34]. Accordingly, P. gingivalis is also strongly associated with periodontal inflammation, but not with healthy state [35]. P. acnes, also a member of the periodontal microbiota [36], is the most prevalent species in apical periodontitis lesions [37] and has been recovered from root canals and from blood samples taken during and after endodontic treatment [38, 39].

Five of the seven studied specimens yielded most isolates, which reflects previously observed inter-individual variation and the dynamic composition of human microbiota [5, 40]. Ott et al. reported high bacterial diversity in atheromas with a range of 5–22. This may also mean low bacterial survival or indirect infection as a cause of atherosclerotic inflammation in two of the patients.

However, unlike the findings with DNA-based detection methods, here, we report the recovery of only a small number of bacterial species. This indicates that the inflammation may be associated with limited spectrum of organisms; therefore, an additional benefit of our approach may be a reduction in false positive findings. Importantly, the present study, first of its kind, defines the possibility to design qualitative and quantitative studies of the association of specific bacteria with clinical measures of disease.

Whilst P. acnes, S. epidermidis and S. infantis could have originated from many bodily sites, including periodontium, the presence of P. gingivalis is consistent with systemic periodontal infection. Whatever their site of origin, these findings support the direct invasion hypothesis. They are consistent with our finding that ‘infectious burden’ is associated with stroke risk and carotid plaque thickness [41], with the report that systemic exposure to P. gingivalis increases the risk of stroke[42] and with reports that oral colonization with periodontal bacterial pathogens is associated with increased carotid artery intima–media thickness [43].

Conflict of interest statement

Authors have no conflict of interest to declare.


This work was supported by the Columbia University College of Dental Medicine, Section of Oral and Diagnostic Sciences and National Heart, Lung and Blood Institute, HL72002. We thank Dr Samuel C. Silverstein for critical analysis of these results and for editorial assistance in preparation of this manuscript.