CD68+ macrophages as crucial components of the foreign body reaction demonstrate an unconventional pattern of functional markers quantified by analysis with double fluorescence staining

Funding information Federal Ministry of Education and Research, Germany, Grant/Award Number: FKZ 13GW0108B; Welcome Trust, Grant/Award Number: 103973/Z/14/Z Abstract Implants like meshes for the reinforcement of tissues implement the formation of a persistent inflammation with an ambient fibrotic reaction. In the inflammatory infiltrate several distinct cell types have been identified, but CD68+ macrophages are supposed to be most important. To investigate the collaboration among the various cell types within the infiltrate we performed at explanted meshes from humans double fluorescence staining with CD68 as a constant marker and a variety of other antibodies as the second marker. The list of second markers includes lymphocytes (CD3, CD4, CD8, CD16, CD56, FoxP3, and CD11b) stem cells (CD34), leucocytes (CD45, CD15), macrophages (CD86, CD105, CD163, and CD206); deposition of EC matrix (collagen-I, collagen-III, MMP2, and MMP8); Ki67 as a marker for proliferation; and the tyrosine-protein kinase receptor AXL. The present study demonstrates within the inflammatory infiltrate the abundant capability of CD68+ cells to co-express a huge variety of other markers, including those of lymphocytes, varying between 5 and 83% of investigated cells. The observation of co-staining was not restricted to a specific polymer but was seen with polypropylene fibers as well as with fibers made of polyvinylidene fluoride, although with differences in co-expression rates. The persisting variability of these cells without the functional reduction toward differentiated mature cell types may favor the lack of healing at the interface of meshes.

within fat, muscle, or fibrous tissues is hampered by chronic inflammation with concomitant fibrotic scar formation. This state-of-affairs favors a stiff-scar plate, nerve entrapments, chronic pain, migration, or erosion of adjacent structures (Glazener et al., 2017). The long-term safety of medical devices such as surgical meshes (Klosterhalfen & Klinge, 2013;Nolfi et al., 2016) or breast implants (Bachour et al., 2018) is decisively determined by the reaction of the host at the interface, in most cases forming a foreign body granuloma (FBG), with adjacent mononuclear inflammatory infiltrate and a surrounding fibrotic capsule (Figure 1). For decades, macrophages have been considered as the dominant immune cell subset in this chronic inflammation, usually being identified by their morphology and surface markers, such as CD68, CD86, or CD206 (Kunisch et al., 2004). Macrophages display tissue-specific diversity and a plastic ability to change their phenotype for production of cytokines to promote either chronic inflammation or wound healing (Krzyszczyk, Schloss, Palmer, & Berthiaume, 2018). These phenotypes are governed primarily by the local tissue microenvironment (Davies & Taylor, 2015). To characterize the inflammatory infiltrate in more detail, in 2014 we had studied the expression of various markers in the FBG of meshes by immunohistochemistry (Klinge, Dietz, Fet, & Klosterhalfen, 2014). The abundant expression of CD68, CD45, as well as CD8 within the FBG, suggested the possibility that macrophages alone may not be playing the decisive part. Consistently, Tennyson et al. (2018) analyzed tissue reactions to explanted pelvic floor meshes and confirmed the presence of abundant lymphocytes, for example, CD4+ helper T cells, CD8+ cytotoxic T cells, or FOXP3+ regulatory T cells. Furthermore, our observation of widespread positive staining of different markers suggested that cells may co-express some of them.
In the present study at explants from human patients, we tested in the cells of the inflammatory infiltrate the expression of various cell markers by double fluorescence staining to identify cell types other than CD68+ and to look for possible co-expression in CD68+ cells. The list of second markers include CD3 (lymphocytes), CD4 (T-helper), CD8 (cytotoxic T), CD16 (NK-cell), CD56 (NK-cell), FoxP3 (Treg), CD11b (DC), CD34 (stem cell), CD45 (leucocytes), CD15 (granulocytes), macrophages CD86 (pro-inflammatory M1), CD105 (activated, part of the TGFreceptor complex), CD163 (part of the scavenger receptor complex), CD206 (anti-inflammatory M2); deposition of EC matrix: collagen-I, collagen-III, MMP2, MMP8; Ki67 served as marker for proliferation; and AXL (tyrosine-protein kinase receptor) as inhibitor of the innate immune response. Furthermore, we looked for the proliferative activity of CD68+ cells as indicative for a local self-renewal.

| MATERIALS AND METHODS
We analyzed 15 meshes made of polypropylene (PP), and five made of polyvinylidene fluoride (PVDF), which all had been used for abdominal wall hernia repair in humans, and were excised between 1999 and 2017 (approval by ethic committee EK 239/19). All the meshes of this study are made of monofilaments, either of PP or PVDF. In some of the samples the material could be clearly identified: seven large pore Ultrapro®, two Ventralex® with a layer of PTFE, two small pore plugs, one Vypro®, one Proceed®, and five Dynamesh®-IPOM. In two others we cannot be sure about the specific material used, other than a monofilament is used.
Human tissue samples of the spleen, liver, lymph node, and tonsil without gross pathology served as healthy control tissue. All samples were embedded in paraffin for subsequent investigations.
F I G U R E 1 Foreign body reaction: Foreign body giant cells, an inner inflammatory infiltrate of mononuclear cells with expression of various markers (e.g. 25.7% CD68+), and an outer fibrotic capsule with collagen deposits around two fibers, in contrast to collagen deposits in scar tissue without this accumulation of inflammatory cells Prior to immunofluorescence staining, mesh samples were checked for the presence of mesh and FBGs by hematoxylin and eosin (supplemental Figure 1) as well as diaminobenzidine immunohistochemical staining. All mesh samples showed the typical foreign body reaction around the mesh fibers with an inner layer of inflammatory infiltrate, followed by an outer fibrotic layer. Most specimens showed a varying number of lymphocytes and foreign body giant cells (FBGCs), as well as small vessels at the mesh-tissue interface. Eight of 15 mesh explants made of PP were explanted due to hernia recurrence and three because of low-grade infection, defined by an accumulation of at least 23 polymorphonuclear cells per high power field. The tissue reaction to PDVF meshes was quite similar to PP meshes but less pronounced. The granulomas were smaller in PDVF compared to PP, particularly their fibrotic tissue layer; and the inflammatory infiltrate consists of smaller numbers of lymphocytes and FBGCs.

| General
All steps were performed at room temperature. Serial 2 μm sections of each specimen were double-labeled with CD68 as the first marker in combination with various other markers (Table 1). CD68 was always stained with fluorescein isothiocyanate (FITC), and the second marker with cyanine-5 (Cy5). All antibodies used were monoclonal and diluted with Antibody Diluent (with Background Reducing Components, Dako, Germany). Secondary antibodies were applied with ImmPRESS™ HRP (Peroxidase) Polymer Detection Kit (Vector, Laboratories, USA). Fluorochromes were diluted with 1× Plus Amplification Diluent (PerkinElmer, USA).

| Protocol
Tissue sections were deparaffinized with xylene, rehydrated through graded alcohol and Milli-Q, before incubation in 3.5% formalin for 10 min. Sections were then placed in a cuvette filled with Milli-Q and

| Analysis of the fluorescence images/stainings
Fluorescence imaging was performed with an Axio Imager 2 microscope (20×, Zeiss, Germany) and the TissueFAXS PLUS system (TissueGnostics, Austria). Images were processed and quantitatively analyzed with StrataQuest Analysis Software (v6, TissueGnostics). Optimized DAPI images were used to detect and segment nuclei ( Figure 2). Nuclei areas were used to measure the mean staining intensities on FITC-and Cy5-shades in selected 1 mm 2 circular regions of interest (ROIs) that were placed around mesh fibers, each including about 2,000-5,000 nuclei Additionally, controls without primary antibody and controls with isotype antibodies were performed, the later with all dilutions and incubation times (Figure 4). With a cut-off value of 100 for the mean intensity in the nuclear area on average less than 2% of false-positive cells were detected (Table 2).

| Statistical analysis
Calculations were done with MATLAB® 9.1 and Image Processing Toolbox 9.5 (The MathWorks, USA). Statistical analysis was performed using Statistical Package for Social Sciences software (SPSS® v23, IBM, USA).
Differences between groups and statistical p were assessed by the nonparametric Mann-Whitney U test. The raw data required to reproduce these findings are available on request from the authors.

| RESULTS
In different healthy human control tissues we always found some cells stained positive for CD68 as well as for CD45 + , CD8 + , CD4 + , or CD3 + , but cells that co-expressed CD68 + CD45 + , CD68 + CD8 + , CD68 + CD4 + , and CD68 + CD3 + were almost absent (<2% of all cells) (Figure 5; Co-expressing CD68 + CD8+ cells are colored yellow. All images have been superimposed with the contours of the cell nuclei in solid green color F I G U R E 3 Regions of interest (ROI) selection. Tissue section of a human mesh explant with selected circular ROI. Each ROI has an area of 1 mm 2 , contains at least one mesh fiber, and 2,000-5,000 nuclei Table 3). In mesh-scar tissue, 19% of all cells were CD68 + , and about 10% were positive for the lymphocyte markers CD3, CD4, or CD8.
However, co-expression of CD68 with CD45, CD3, CD4, or CD8 still was rare (2.8-4.8% of all cells). In contrast, in the area of the FBG around PP meshes, all these markers were seen abundantly, and costaining with CD68 was frequently found. Between 30 and 50% of the CD68 + cells were positive for a second marker, and half of the cells being positive for CD45 + , CD8 + , CD4 + , and CD3 + were simultaneously positive for CD68 + (Figure 6).
To test whether or not this dual-labeling was restricted to CD45, CD8, CD4, and CD3 in PP meshes we expanded our analysis to a mesh of another polymer and to further markers, which are considered to spec-  Figure 7; Table 4). While positively stained cells were mainly observed at the mesh-tissue interface for most markers, this was not the case for CD34, CD86, and MMP8, which usually were located outside the FBG.
FBGCs were positive for most markers, though showing strong heterogeneous staining patterns even within their multinucleated cell bodies.
Markers with high numbers of positive-labeled cells within the inflammatory infiltrate usually showed higher cell densities closer to the fibers, with no distinct zones that separate positive-and negativelabeling. Most of these markers even displayed a local gradient with continuously decreasing mean intensities with increasing distance from the mesh fibers ( Figure 8).
All markers tested showed substantial co-expression with CD68 + within the inflammatory infiltrate of the FBG, ranging from 5% for CD68 + CD86+ to 83% for CD68 + AXL + cells, mainly preferential close to the mesh fibers (Figures 9 and 10a-d; Table 4).
To study whether or not the inflammatory infiltrate of the FBG derives from proliferating cells, we looked for the expression of Ki67 at meshes, which at PP was found on average in 5.8% of the cells.

| DISCUSSION
Mesh-related complications are usually attributed to the foreign body reaction characterized by chronic fibrotic inflammation at the mesh boundary. Inflammation is governed by the types of immune cells on F I G U R E 4 Cumulative mean intensity distributions do not indicate distinct subgroups. Cumulative mean intensity distributions for all markers (e.g. CD68 labeled with fluorescein isothiocyanate (FITC) and CD4 labeled with Cy5) at human mesh explants never showed any s-shape or step, which would have reflected a distinct cluster of positive-labeled cells with higher intensities. Instead, the continuous smooth rise confirms the presence of a continuum and a strong cellular heterogeneity, not indicating a distinct cut-off to differentiate between positive-and negative-labeled cells. The curves of controls are strongly left-shifted; the cut-off value of 100 detects true positive cells with high intensities without including many false positives as can be seen by the lowest percentage of 99.3% for controls (positive predictive value >99%) T A B L E 2 Calculation of the positive predictive value (PPV). The explanation for this extreme heterogeneity could be the radical adaptation of the immune system to eliminate an indigestible foreign object. Such a reaction is commonly identified in parasite infections, where the macrophages fuse and form giant cells (Ruckerl & Allen, 2014). An identical phenomenon has been observed at the interface to our meshes and has been reported even between T cells and macrophages in HIV infection (Bracq et al., 2017). However, multicell fusion cannot explain the heterogeneity we observe at the mesh boundary, as the cells we identify are widely mononuclear. An alternative mechanism could be emergency hematopoiesis, such as that occurring in cancer and infections, where immature immune cells are recruited from the bone marrow and display stem-like properties (Boettcher & Manz, 2017). The level of multipotency may allow expression of normally lineage-restricted genes and proliferation at the site of the implant. Our data revealed that the majority of Ki67 + cells (indicating proliferation potential) were also positive for CD68, confirming these CD68 + Ki67+ cells are not terminally differentiated and can proliferate in situ rather than relying on newly recruited bone marrow monocytes, as has been observed in monocyte-derived macrophages in mice (Davies, Jenkins, Allen, & Taylor, 2013). This state-of-affairs may explain the level of cross-lineage heterogeneity we observe, though a more thorough examination by methods of lineage tracing as well as extended multiplex stainings will help to unravel the underlying mechanism (Baron & van Oudenaarden, 2019). Co-expression of cells in healthy human tissues and human mesh explants.   In this study, we can demonstrate that the heterogeneity of immune cells is not restricted to meshes made of PP, but is apparent with meshes made of another polymer PVDF, as well, although with differences mainly for the lymphocytic and collagen markers.
Considering the many problems recorded after the use of meshes to reinforce tissues, it is the persistent local reaction, which underlies the majority of these. The host tissue is not able to effectively seal the foreign body in a stable wall of defending cells and extracellular  Note: Analysis of co-expression for CD68 + Ki67+, and co-expression in relation to either CD68+ or Ki67+ within the foreign body granuloma of human mesh explants. Data are presented in mean percentages of seven PP meshes and five PVDF meshes (11-18 regions of interest each). Positive cells have a mean staining intensity >100.
can affect the results. Extended inflammation with an ROI limited to the infiltrate of the FBG will lead to increased cell counts, whereas large ROIs that include wide areas of polymeric fibers, adjacent fat tissue, or excessive collagen deposits will show reduced numbers of nuclei. Correspondingly, the percentage of co-expressing cells may be affected. Our choice of an ROI of 1 mm 2 , which usually includes about 2,000-5,000 cells of the inflammatory infiltrate, covers most of the inflammatory infiltrate around a mesh fiber. Even if parts of the ROI include adjacent noninflammatory tissues, it is both the number of CD68+ cells and the number of cells being positively stained with a second marker, which are reduced. Subsequently, the percentage of co-expressing CD68+ cells remains widely constant, which therefore can serve as the robust but specific property of the mesh. The analysis of the staining intensity only in the area of the nuclei cannot exclude that despite thin sections of about 2 μm the positive staining might be based on some overlapping cytoplasm membrane, but it offers the option to define positively stained cells in an objective and reliable way. To assure ourselves that our method identified true co-expression, we applied a cut-off fluorescence value of 100. This approach ensured a high PPV, which provides us with the most confidence for true positive selection. However, in some cases, the use of a lower cut-off value could have been argued, which would lead to higher percentages of double and single "positive" cells, at the loss of objectivity.
It should be noted that lowering this cut-off only supports the conclusions of this study.

ACKNOWLEDGMENTS
We gratefully thank A. Fiebeler for her advice and always-helpful criti-

CONFLICT OF INTEREST
The support by the Federal Ministry of Education and Research (FKZ 13GW0108B) enabled the acquisition of the Tissue Gnostics system.
A.D. as PhD student is an employee of the FEG. U.K. and B.K. had research projects and consulting fees in collaboration with the mesh manufacturers Ethicon and FEG; expert testimony at lawsuits concerned with pelvic floor meshes; mesh patients with FEG. The financial disclosures listed result from their expertise, and none of them have tried to influence any part of the work for this manuscript.