A Novel Approach for Detecting Unique Variations among Infectious Bacterial Species in Endocarditic Cardiac Valve Vegetation

Infectious endocarditis (IE) remains one of the deadliest heart diseases with a high death rate, generally following thrombo-embolic events. Today, therapy is based on surgery and antibiotic therapy. When thromboembolic complications in IE patients persist, this is often due to our lack of knowledge regarding the pathophysiological development and organization of cells in the vegetation, most notably the primordial role of platelets and further triggered hemostasis, which is related to the diversity of infectious microorganisms involved. Our objective was to study the organization of IE vegetations due to different bacteria species in order to understand the related pathophysiological mechanism of vegetation development. We present an approach for ultrastructural analysis of whole-infected heart valve tissue based on scanning electron microscopy and energy-dispersive X-ray spectroscopy. Our approach allowed us to detect differences in cell organization between the analyzed vegetations and revealed a distinct chemical feature in viridans Streptococci ones. Our results illustrate the benefits that such an approach may bring for guiding therapy, considering the germ involved for each IE patient.


Introduction
While important progress has been made in the diagnosis and treatment of cardiovascular diseases, infectious endocarditis (IE) remains one of the deadliest heart diseases with a high death rate [1]. It is an infection of the endocardium and heart valves or prosthetic valve implants caused mainly by cocci gram positive bacteria [2]. In short, our knowledge of the pathogenesis of this disease remains limited to the formation of a fibrin-platelet matrix that serves as a platform to receive the pathogen, followed by valve inflammation, all of which constitutes endocarditic vegetation [3]. Only a few studies that are available in the literature have shown interest in the mechanism of the development of the vegetation.

Patients and Histological Analysis
Our analysis was performed on specimens obtained from six patients after cardiac surgery. One patient was a negative control with a degenerative valve without IE, presenting with severe mitral regurgitation related to mitral prolapse with flail leaflets, and five other patients diagnosed for IE, based on modified Duke criteria and due to one different and unique bacterial species (Table 1) [1]. Classical histological analysis was carried out during the diagnosis, using Giemsa, Grocott, Gram, periodic acid Schiff and Warthin-Starry staining.

Vegetation Preparation
Whole-cardiac valves pieces were fixed with glutaraldehyde 2.5% in 0.1 M sodium cacodylate buffer for at least 1 h. In order to access the depth of the tissue, vegetations were cut transversally with a razor blade in sterile conditions and then they were rinsed with 0.1 M sodium cacodylate buffer for 10 min and with distilled H 2 O for 10 min. The vegetations were gradually dehydrated with: ethanol 20% for 20 min; ethanol 50% for 10 min; ethanol 70% for 10 min; ethanol 85% for 20 min; ethanol 95% for 20 min; ethanol 100% for 30 min. Then the vegetations were incubated with ethanol 100%/hexamethyldisilazahne 100% in 1:2 ratio for 30 min (HDMS; Sigma-Aldrich; Ref. 379212) and with HDMS 100% for 10 min. A drop of HDMS 100% was dropped on top of the specimen and the vegetation was air-dried while protected from dust in a microbiological safety cabinet overnight. All solutions were 0.2 µm filtrated before use. The vegetations were mounted with their transverse cut facing upward on double-sided tape on clean glass slides and copper-coated with a MC1000 sputter coater (Hitachi) for 40 s at 10 mA.

Scanning Electron Microscopy and Energy-Dispersive X-ray Spectroscopy
Observation by SEM was either performed in a SU5000 scanning electron microscope (Hitachi High-Technologies, Tokyo, Japan) in low-vacuum at 15 keV with a BSE detector and observation mode (spot size 30) or a TM4000Plus table-top SEM (Hitachi High-Technologies, Tokyo, Japan) operated at 15 keV, lens mode 4 with a BSE detector. Magnification ranged between ×90 and ×11 000. A mean number of 173 images for each vegetation and 68 images for the degenerative valve were acquired. For semi-quantitative analysis (see results in Table 2), the relative abundance of each component (cocci, platelets, erythrocytes, leukocytes or fibrin) was determined by estimating the ratio of its presence with regard to the total number of images acquired for each valve, as follow: (−) Absence: no image found; (+) Rare: less than 10% of images; (++) Moderate: between 10% and 80% of images; (+++) Abundant: more than 80% of images. EDX was performed with an INCA X-Stream-2 detector (Oxford Instruments) attached to the TM4000 SEM and AztecOne software (Oxford instruments) at 10mm working distance. Blind chemical mapping (all elements considered) on random regions of the valves were performed in the first round of observations, and selected elements (carbon, nitrogen, oxygen, calcium and phosphorus) were mapped in the second time on selected regions (bright versus non-bright regions). Weight and atomic percentages were retained for quantitative study.

Statistical Analysis
GraphPad (Prism) for Windows was used for statistical analysis of the EDX data. As samples were normally distributed, significant differences were determined using the two-tailed, unpaired student's t test. Statistical significance was set at p < 0.05.

Histological and Immunohistochemical Observations
Six valves were considered for analysis. One valve served as a negative control provided from a patient without IE, who had severe mitral regurgitation related to mitral prolapse with flail leaflets. The five other valves corresponded to IE cases with vegetation. Out of these five vegetations, histological analysis reported an inflammatory state typical of infectious endocarditis for three of them (Enterococcus faecalis, Staphylococcus aureus and Streptococcus oralis). A discrete inflammation was reported for Streptococcus agalactiae and Streptococcus gallolyticus. The presence of bacteria was reported for S. oralis and S. aureus, but not for E. faecalis, S. agalactiae and S. gallolyticus. Blood culture was used to identify the bacterial species involved in each IE case [15].

Scanning Electron Microscopy
We used SEM to understand the cellular organization on top and inside the analyzed vegetations. We found various tissue and cellular organizations depending on the bacterial species involved by comparing the amount of platelets, fibrin, bacteria and leukocytes, and the respective cellular arrangements of the different elements within the vegetation ( Table 2).

Presence and Organization of Bacteria in the Vegetative Tissue
First, we made sure of the absence of bacteria in the degenerative valve serving as a negative control ( Figure S1). Second, we looked for the presence of bacteria in the endocarditic vegetations. Bacteria appeared as round objects, isolated or arranged in clusters. The mean maximal diameter of bacteria ranged between 750 to 800 nm, which is consistent with their morphological characteristics [9].
In contrast to the cases of E. faecalis, where the bacteria were present in low amounts (less than 10% of our images) (Figure 1), S. aureus, S. oralis, S. agalactiae and S. gallolyticus (Figures 2-5) were present in larger numbers in the other vegetations. We also noticed a significant number of dividing bacteria in the case of S. oralis, where the bacteria presented typical proliferation features such as double cell bodies and division septa ( Figure 4B). Because the SEM screening of large regions of the surface and the depth of the IE cardiac vegetations comes directly from the operative bloc, we were able to detect bacteria in all the five IE vegetations analyzed. Even if the histological analysis had been negative for finding such objects in the cases of E. faecalis, S. agalactiae and S. gallolyticus, SEM proved to be efficient in finding infectious bacteria in all five vegetations.
Cells 2020, 9, x FOR PEER REVIEW 5 of 12 [9]. In contrast to the cases of E. faecalis, where the bacteria were present in low amounts (less than 10% of our images) (Figure 1), S. aureus, S. oralis, S. agalactiae and S. gallolyticus  were present in larger numbers in the other vegetations. We also noticed a significant number of dividing bacteria in the case of S. oralis, where the bacteria presented typical proliferation features such as double cell bodies and division septa ( Figure 4B). Because the SEM screening of large regions of the surface and the depth of the IE cardiac vegetations comes directly from the operative bloc, we were able to detect bacteria in all the five IE vegetations analyzed. Even if the histological analysis had been negative for finding such objects in the cases of E. faecalis, S. agalactiae and S. gallolyticus, SEM proved to be efficient in finding infectious bacteria in all five vegetations.

The Different Extent of Inflammation
The extent of the inflammation was estimated by the abundance of leukocytes, which likely corresponded, based on morphology, to polymorphonuclear neutrophils and macrophages. These cells were identified by their globular shape and diameters ranging from 12 to 20 µm. The inflammatory state was more pronounced in the case of E. faecalis, S. agalactiae and S. aureus, (between 10% and 80% of analyzed images in the case of E. faecalis and S. agalactiae and more than 80% in the case of S. aureus vegetation) (Figures 1-3) compared to the other vegetations. The inflammatory cells mostly appeared as individualized or embedded in the tissue. They were located in regions that also contained bacteria in the case of S. aureus and S. agalactiae ( Figure 2D,F and Figure 3D), and E. faecalis, despite a lower level of bacteria in the latter case. In the vegetations with S. gallolyticus and S. oralis, a few inflammatory cells were observed, and these were mainly isolated.

Platelet Recruitment and Extent of the Fibrin Network
Firstly, the overall appearance of the vegetation was mainly amorphous and filamentous in the case of the Streptococcus and S. aureus species. The filaments had the appearance of a fishing net in the latter case ( Figure 2C). The vegetation with E. faecalis as well as with the degenerative valve had a stratified appearance ( Figure S1 and Figure 1) and was devoid of filaments. Platelets appeared with a diameter between 1.5 and 3 µm, with or without protrusions. They were observed in large numbers with the Streptococci and S. aureus vegetation, (between 10% and 80% of analyzed images in the case of S. agalactiae and more than 80% in the case of S. aureus, S. oralis and S. gallolitycus vegetation). In these vegetations, platelets were either individual, caught in the fibrin network, or rather aggregated in clusters, with loss of cellular integrity and an amorphous platelet matrix (Figures 2-5). The platelet/bacteria ratio was approximately 1:1 in all vegetations, except in the case of S. oralis, where this ratio was approximately one platelet per 10 bacteria. We noticed that in the case of E. faecalis, erythrocytes were much more abundant than platelets (Figure 1). Erythrocytes were recognized based on their mean diameter of 7 µm and their characteristic biconcave donut shape.
There seemed to be a correlation between the abundance of the fibrin network and the number of platelets in the different vegetations. Indeed, large networks of fibrin were observed in S. aureus (Figure 2), S. agalactiae ( Figure 3) and S. oralis ( Figure 4D,E) vegetation, surrounding both bacteria and platelets, and also erythrocytes in the case of S. agalactiae and S. aureus vegetation. Fibrin filaments were less abundant in the case of S. gallolyticus vegetation. Regarding E. faecalis vegetation, the fibrinous networks were very reduced and limited only to a few narrow regions ( Figure 1D).
Finally, we also noticed regions with a remarkably bright appearance when imaging the S. oralis and S. gallolyticus vegetation ( Figure 4A, Figure 5C and Figure S2). When zooming-in, these bright regions under the SEM electron beam had well defined outlines and were mainly composed of platelets and bacteria. In some of these regions, cocci-like ghost shapes were detected.

Energy Dispersive X-Ray Spectroscopy
To decipher the nature of the bright regions observed in the S. oralis and S. gallolyticus vegetation, we proceeded to EDX analysis. Surprisingly, EDX revealed that the bright regions emitted strong signals for calcium (Ca) and phosphorus (P) compared to the other regions ( Figure 6A). This resulted in significant peaks for these two elements in the EDX spectra ( Figure 6B). As shown in Table 3, which provides the values obtained following the analysis of the S. gallolyticus vegetation, significant changes in the weight and atomic percentage of C, O, Ca and P were observed between non-bright (n = 5 for each element) and bright regions (n = 9 for each element). Indeed, the measurements carried out on the bright regions showed a decrease in the abundance of carbon and an increase in the abundance of oxygen, phosphorus and calcium. There was no difference in nitrogen between the two types of region ( Table 3). Analysis of the other vegetation as well as the degenerative valve using the EDX tool showed a low signal emission with regard to calcium and phosphorus ( Figure S3). Table 3. Results (mean ± SD) of weight and atomic percentages of C, O, N, P, Ca in the bright and non-bright regions of the S. gallolyticus vegetation. ** p < 0.01; *** p < 0.001.   To better understand the possible nature of the compound(s) responsible for such an increase in Ca and P in these regions, we calculated the average weight and atomic Ca/P ratios in the bright regions of interest in S. gallolyticus vegetation (n = 9). We found a mean Ca/P weight percentage ratio of 2.18 ± 0.29 and a mean Ca/P atomic percentage ratio of 1.52 ± 0.34, both values that are close to those calculated for hydroxyapatite compounds in biological samples [16]. Similar values were obtained from the S. oralis vegetation analysis (Table S1).

Discussion
Our study reported different cell organization between the analyzed endocardial vegetations ( Table 2). These differences involved, on the one hand, the hemostatic state represented by the abundance of platelets and the extension of the fibrin network, and on the other hand, the extension of the inflammatory state represented by the abundance of leukocytes. To better understand the possible nature of the compound(s) responsible for such an increase in Ca and P in these regions, we calculated the average weight and atomic Ca/P ratios in the bright regions of interest in S. gallolyticus vegetation (n = 9). We found a mean Ca/P weight percentage ratio of 2.18 ± 0.29 and a mean Ca/P atomic percentage ratio of 1.52 ± 0.34, both values that are close to those calculated for hydroxyapatite compounds in biological samples [16]. Similar values were obtained from the S. oralis vegetation analysis (Table S1).

Discussion
Our study reported different cell organization between the analyzed endocardial vegetations ( Table 2). These differences involved, on the one hand, the hemostatic state represented by the abundance of platelets and the extension of the fibrin network, and on the other hand, the extension of the inflammatory state represented by the abundance of leukocytes.
First, the influence of the bacterial species on the ultrastructure of platelet aggregates has been previously demonstrated in vitro, by light and electron microscopy [9]. E. faecalis induced loose aggregates, with platelets that fully retained their cellular integrity while S. sanguinis and S. aureus were associated with very dense aggregates and coalesced platelets. Accordingly, we showed here that the vegetation induced by E. faecalis was formed by a few platelets and fibrin, which suggested low engagement in the hemostatic process, compared to the vegetation due to S. aureus and streptococci species, which were associated with many more platelets and fibrin.
Second, we showed that the inflammatory state of the vegetations was different, with very inflammatory vegetations, such as the ones associated with S. aureus, E. faecalis and S. agalactiae and less inflamed vegetations associated with S. gallolyticus and S. oralis. Accordingly, group B streptococci (S. agalactiae) have been shown to activate neutrophils [17]. In another comparative study, it was reported that the activation of neutrophils induced by S. aureus and E. faecalis was relatively greater than that induced by viridans streptococci [18].
Third, EDX analysis showed some unexpected results regarding the chemical composition of endocarditic vegetation. Viridans streptococci vegetations (S. gallolyticus and S. oralis) were associated with a bright deposit, composed mainly of Ca and P, with a ratio close to that of hydroxyapatite. Such calcified deposits in cardiac valves have been previously reported [19], especially in the vegetation due to S. salivarius, which is also part of viridans streptococci [20]. Normally, Ca-P deposits occur in the bones and teeth, but also they can be found abnormally in other soft tissues, and heart and urinary stones [21,22]. Interestingly, in such stones, bacteria ghosts were found, such as the ones we imaged in our samples [23]. Although calculogenic substances in the body fluids and/or lipids may be causally linked to the Ca-P deposits we observed [21,24], the presence of bacteria proliferation features and of bacteria ghosts inside the Ca-P deposits tends to link the bacteria to the development of these deposits in the case of IE vegetation. We hypothesize that the development of Ca-P deposits may primarily be an infectious bacteria species-dependent phenomenon, due to the specific activity of the viridans streptococci bacteria. It may ultimately influence the consistency of the vegetation and may have a role in further appearance of emboli.
In conclusion, the use of SEM on whole tissues made the detection of bacteria relatively fast with no need for special staining or time-consuming tissue embedding/ultramicrotomy methods. EDX provides a chemical signature for the tissues infected with specific species, which eventually will help to define new strategies for rapid diagnosis of the bacterial infection in the heart.
We are aware that the analysis of a single vegetation per bacterial species cannot establish a firm relationship between ultrastructure-bacteria, something that would be interesting to study further. Nonetheless, through this study we are able to highlight that the composition of endocardial vegetation is far from being universal, with infected and inflamed fibrin-platelet deposit, but rather needs to be analyzed considering both the bacterial species involved and the characteristics of the patient. Indeed, we suggest a divergent pathogenesis for IE, which could act as a guide for personalized therapy depending on the infectious agent involved. Meanwhile, the patient's history may also influence the vegetation development/organization, for example, pre-operative antibiotics and anti-thrombotic treatments. Also, here we worked with vegetations recovered from mitral valves and aortic valves, and this may also be a factor that differentially influences the development of the vegetation.
Supplementary Materials: The following are available online at http://www.mdpi.com/2073-4409/9/8/1899/s1, Figure S1: Scanning electron microscopy of control uninfected valve., Figure S2: Scanning electron microscopy of S. gallolyticus infected vegetation showing bright regions., Figure S3: Scanning electron microscopy coupled to energy-dispersive X-ray spectroscopy of representative regions from the degenerative control valve and E. faecalis, S. aureus and S. agalactiae infected vegetations. Table S1  conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript.