Transcriptome-targeted analysis of human peripheral blood-derived macrophages when cultured on biomaterial meshes

Abstract

Surgical meshes are commonly used to repair defects and support soft tissues. Macrophages (Mϕs) are critical cells in the wound healing process and are involved in the host response upon foreign biomaterials. There are various commercially available permanent and absorbable meshes used by surgeons for surgical interventions. Polypropylene (PP) meshes represent a permanent biomaterial that can elicit both inflammatory and anti-inflammatory responses. In contrast, poly-4-hydroxybutyrate (P4HB) based meshes are absorbable and linked to positive clinical outcomes but have a poorly characterized immune response. This study evaluated the in vitro targeted transcriptomic response of human Mϕs seeded for 48 h on PP and P4HB surgical meshes. The in vitro measured response from human Mϕs cultured on P4HB exhibited inflammatory and anti-inflammatory gene expression profiles typically associated with wound healing, which aligns with in vivo animal studies from literature. The work herein provides in vitro evidence for the early transcriptomic targeted signature of human Mϕs upon two commonly used surgical meshes. The findings suggest a transition from an inflammatory to a non-inflammatory phenotype by P4HB as well as an upregulation of genes annotated under the pathogen response pathway.

1. Introduction

Suture or mesh biomaterials are used to augment surgical repairs and usually should remain relatively inert, causing minimal inflammation and allowing tissue healing while performing their stabilizing function [1]. Multiple studies have shown that permanent biomaterials such as polypropylene (PP) meshes, while effective for tissue support, can have long-term complications often attributed to inflammation due to a chronic foreign body response [2–4]. Absorbable polymeric scaffolds are alternative types of biomaterials that can provide the same function, but may reduce these complications by slowly degrading, allowing the host tissue to remodel and generating new tissue in situ.

Absorbable polymeric scaffolds aim to reduce the magnitude of foreign body response and the potential for long-term complications such as pain, contraction, extrusion, and infection [1]. Several recent clinical studies have documented the success of absorbable scaffolds made from poly-4-hydroxybutyrate (P4HB) as an alternative to PP meshes or animal-derived scaffolds to successfully treat ventral hernias, incisional hernia, reduce donor site bulge, prevent ptosis, and as an alternative to acellular dermal matrices in breast reconstruction [5–7]. A better understanding of how absorbable biomaterials interact with the host response is critical in order to improve patient outcomes and develop improved solutions for wound healing and soft tissue support.

Macrophages (Mϕs) play a critical role in the host response and are among the earliest immune cells to respond to tissue damage and surgical repair sites. In combination with tissue-resident Mϕs, newly recruited Mϕs secrete signals to attract appropriate cell types and orchestrate inflammatory and healing events [8]. During the initial inflammatory phase, neutrophils and Mϕs begin the healing process by removing damaged tissue [9], destroying microorganisms, phagocytizing foreign material and debris, and recruiting inflammatory cells and other cell types needed for repair, angiogenesis, and remodeling [10]. Mϕs are crucial contributors to each of these phases of healing and play important roles in orchestrating the body’s response to an injury or surgical insult [11, 12].

This initial inflammatory response can last several days, followed by a Mϕ-centered response to stabilize the tissue and guide the healing process [13, 14]. During the healing phase, local cells begin to produce the needed extracellular matrix and, over a period of weeks to months, the matrix matures and remodels as tissue repair occurs. If initial inflammation caused by the insult or the biomaterial does not resolve, a chronic inflammatory response or other adverse outcomes may result [15]. A biomaterial can provide the necessary mechanical support for surgical intervention without long-term, chronic inflammation, while an absorbable biomaterial can slowly degrade allowing healing and site-specific tissue production by promoting the optimal transition of Mϕs.

Mϕs can be induced in vitro into distinct, well-characterized phenotypes ranging from the classically-activated, pro-inflammatory M1-like phenotype to the alternatively-activated, anti-inflammatory M2-like phenotype [16]. In an attempt to classify the Mϕs present during different stages of repair, M2-like Mϕs have been further classified into subtypes such as pro-healing M2a, immunoregulatory M2b, immunosuppressive M2c, and tumor-associated M2d [17, 18]. However, Mϕs in the body exist on a spectrum of phenotypic traits due to the dynamic concentration of cytokines present across the host tissues [19]. In vitro, the different phenotypes can be induced or approximated experimentally with specific stimuli, thus providing a convenient experimental cell model that can help study the role of Mϕs in homeostasis, inflammation, remodeling, and healing [20], or in the case of this study, their response to different biomaterials. While gene expression assays have been used to characterize the different Mϕs phenotypes, few studies have been performed to elucidate the targeted transcriptome gene expression of human Mϕs in response to in vitro culture on different biomaterial meshes and how this might translate in vivo.

We hypothesized that an absorbable P4HB mesh will modulate a particular targeted-transcriptome gene expression in Mϕs isolated from human peripheral blood mononuclear cells (PBMCs) that differs from the response to a permanent PP scaffold. In this work, we investigated the response of non-polarized Mϕs (which we identify as ‘M0s’) and polarized Mϕs (‘M1’, ‘M2a’, and ‘M2c’ like-phenotypes) to absorbable and permanent meshes typically used in hernia repair and plastic and reconstructive surgical applications. Our goal was to determine how the materials affect the Mϕ’s gene expression by contrasting the transcriptomic targeted cell response when seeded on different biomaterials and compared to four in vitro control phenotypes (M0, M1, M2a, and M2c).

In this study, we used a commercially available, 740 gene panel targeted towards innate immunity to compare gene expression profiles for non-polarized and polarized Mϕs. We identified genes with significant changes in their expression in Mϕs cultured on commonly used meshes. The changes in the transcriptome response for Mϕs seeded on P4HB mesh provide clues into a potential transition from an inflammatory gene expression profile to a more anti-inflammatory profile. Since the biomaterial-cell interaction with the host immune system is critical for wound healing applications, these results offer insight into the observed in vivo responses to these materials and how we can use such information to guide their clinical use.

2. Materials and methods

2.1. Materials and reagents

Human buffy coats from five unidentified donors were purchased at different times from the New York Blood Center (New York, NY). The EasySep^™ Human CD14 Positive Selection Kit was purchased from Stem Cell Technologies (Vancouver, Canada). Heat-inactivated human serum (Human Serum-H) derived from male AB plasma (Cat. #H3667) was purchased from MilliporeSigma (Burlington, MA). Corning^® Ultra-Low Binding plastic (UBP), well plates and cell culture flasks, 4-hydroxybutyric acid (4HB) as the sodium salt, and Lipopolysaccharide (LPS) were purchased from Sigma-Aldrich (St. Louis, MO). Cytokines used for the differentiation and polarization of Mϕs, Monocyte Colony-Stimulating Factor (M-CSF), Interferon-gamma (IFNγ), Interleukins 4, 10, and 13 (IL4, IL10, and IL13, respectively), were obtained from Peprotech (Rocky Hill, NJ).

GalaFLEX^® P4HB mesh and fatty acid (FTY) residues extracted from the P4HB were obtained from Tepha, Inc. (Lexington, MA). Prolene^® PP-mesh, Vicryl^®, and Mersilene^® were manufactured by Ethicon (Somerville, NJ). TIGR^® mesh was manufactured by Novus Scientific (Uppsala, Sweden). SERI^® was manufactured by Sofregen Medical (Medford, MA) and Alloderm^® was manufacture by Allergan (Dublin, Ireland).

The use of the NanoString Technologies Prep Station and Digital Analyzer instruments was provided by the Lineberger Comprehensive Cancer Center (LCCC) at the University of North Carolina-Chapel Hill School of Medicine.

2.2. Scanning electron microscopy (SEM)

Samples from P4HB and PP meshes were sputter coated with a Denton Vacuum Desk V sputter coater using a gold/palladium target for a total of 90 s (3 × 30 s intervals to avoid raising the temperature of the samples) and on 45 kVp intensity. Samples were imaged with a Hitachi S-3400 N-II SEM with Secondary Electron Detector, Accelerating Voltage 10 kV, under high vacuum <1 Pa.

2.3. Isolation of monocytes and Mϕs differentiation

CD14⁺ monocytes were isolated from human PBMCs and differentiated into Mϕs as previously described [21, 22].

Heparinized blood was diluted with 1 mM EDTA in 1× Dulbecco’s PBS (DPBS), carefully overlaid onto Lymphocyte H Cell Separation Media (Cedarlane, Burlington, NC) and spun down at 400 g-forces for 20 min. The PBMC layer was removed with a transfer pipette, washed three times with EDTA/DPBS and centrifuged for 150 g-forces for 10 min each. Using the Human CD14 Positive Selection Kit, CD14⁺ cells were isolated from the PBMCs according to the manufacturer’s instructions. Briefly, the cell density was adjusted at 1.0 × 10⁸ cells ml⁻¹ and EasySep^™ positive selection cocktail was added at 100 μl ml⁻¹ for 15 min. EasySep^™ magnetic particles were then added at 50 μl ml⁻¹ and incubated for 10 min. Using ‘The Big Easy’ EasySep^™ Magnet (Stem Cell Technologies), the CD14⁺ cells were isolated by incubating the cell suspension three times for 5 min each and gently removing the supernatant.

After isolation, 0.5–1.0 × 10⁶ cells per antibody were fixed in FACS buffer (0.5% bovine serum albumin (BSA) in 1 mM EDTA in 1× PBS) for 15 min. The cells were then stained (1:100) for 15 min at 4 °C with either a mouse anti-human CD14:APC antibody (BioRad, Hercules, CA, Cat #MCA596APCT), CD34 mouse anti-human:PE (Fisher Scientific, Cat. #BDB560941), or CD68 mouse anti-human:FITC (Fisher Scientific, Cat. #BDB562117). Isotype controls included anti-IgG1 κ Mouse:PE (Fisher Scientific, Cat. #BDB555749), anti-IgG2b, κ Mouse:FITC (Fisher Scientific, Cat. #BDB555057) and mouse IgG2a:APC (BioRad, Cat #MCA929APC) antibodies. Samples were then washed and fixed with 4% formaldehyde for 15 min, washed and suspended in FACS buffer. Flow cytometry was performed on a BD^™ LSR II (Becton, Dickinson and Company, Franklin Lakes, NJ) and analyzed using BD FACSDiva^™ software, see supplemental figure s1 (available online at stacks.iop.org/BMM/16/025006/mmedia).

The isolated CD14⁺ cells were cultured at 1.0 × 10⁶ cells ml⁻¹ in low attachment flasks (Corning, Corning, NY) [23, 24]. Media used to culture the monocytes and Mϕs was composed of RPMI Media 1640 with GlutaMAX^™ supplemented with 10% Human Serum-HI and 1% Penicillin/Streptomycin (Thermo Fisher Scientific, Waltham, MA) (‘human Mϕ media’). M-CSF was added at 20 ng ml⁻¹ to the media to push the CD14+ cells into non-polarized Mϕs (‘M0’) for 5 d with a media change on Day three of culture.

2.4. Mϕ polarization

On Day five of culture, Mϕs were seeded at 1.0 × 10⁶ cells ml⁻¹ with 20 ng ml⁻¹ M-CSF into UBP six-well plates. Mϕs that did not receive any other cytokines (non-polarized), only fresh media change, are continually referred to as M0. Mϕs that were polarized with 100 ng ml⁻¹ of LPS and 100 ng ml⁻¹ of IFNγ are referred to as ‘M1’. Mϕs that were polarized with 40 ng ml⁻¹ of IL4 and 20 ng ml⁻¹ of IL13 were designated ‘M2a’. Mϕs polarized with 40 ng ml⁻¹ of IL10 were referred to as ‘M2c’.

2.5. Seeding of the biomaterials

2.5.1. Biomaterial screening attachment

7 mm diameter disks of each biomaterial were made with a sterile biopsy punch (Acuderm, Fort Lauderdale, FL) and attached to 96 well plates (Corning). For the material screening experiments, biomaterials were attached using a 15 μl fibrin clot comprised of 2% fibrinogen in 1 × DPBS and thrombin (Sigma-Aldrich, St. Louis, MO) at 100 U ml⁻¹ at a 5:1 ratio as reported by [25]. A fibrin clot alone was used as a control. The attached scaffolds were stored in 100 μl of 1 × DPBS at room temperature until ready for cell seeding.

2.5.2. PP and P4HB mesh preparation test

The fibrin clot was not used in any of the subsequent tests for the direct comparisons of PP (PROLENE Mesh, Ref PMII, 3 × 6”, Lot LAH068, Exp. 2021–12-31), P4HB (GalaFLEX Scaffold, Ref GP0408, 10 × 20 cm, Lot 200 354, Exp. 2023–05-31), and UBP. Rather, two layers of 12 mm diameter disks of the meshes (to provide a higher surface area available for cell attachment) were arranged and placed in a 48 well plate. The materials were stored in 400 μl of 1 × DPBS at room temperature until cells were ready for cell seeding.

2.5.3. Mϕ seeding and polarization

Mϕ were detached from the UBP flasks with Accutase (Thermo Fisher Scientific) for 5 min at 37 °C on Day 5 of culture, collected, and counted using Countess^™ II Automated Cell Counter (Thermo Fisher Scientific). Cells were suspended at a cell density of 1.0 × 10⁶ cells ml⁻¹, the DPBS was removed from the scaffolds, and 2.0 × 10⁵ cells were seeded onto each scaffold in 200 μl of human Mϕ media to each well with 20 ng μl⁻¹ of M-CSF. Non-polarized M0s (Mϕ + M-CSF) or Mϕs exposed to the cytokines required for the polarization controls (i.e. M1-, M2a-, and M2c-like phenotypes) were also seeded onto UBP alone.

The plates were kept at 37 °C with 5% CO₂ concentration for 48 h until Day 7 of culture. The sample collection was done on Day 7 of culture by carefully removing the scaffolds from the media container and immersed twice into DPBS to wash out non-attached Mϕs. Cells attached to P4HB and PP were stained using CellMask^™ Deep Red Plasma Membrane Stain at 5 μg ml⁻¹ for 30 min following manufacturer protocol. The samples were gently washed in DPBS 1× twice before imaging using fluorescence microscopy with a Revolve microscope (Echo, San Diego, CA). Separate scaffolds with adherent cells were used for RNA isolation.

2.5.4. Mϕ seeding on UBP upon 4HB and FTY exposure

Since P4HB is absorbable and may hydrolyze to release low molecular weight species, M0s were also seeded onto UBP along with the FTY monomer 4HB and extractable FTY residues from the P4HB polymer at the following concentrations 4HB (5 mM) and FTY (5 μg ml⁻¹). Samples were analysed on Day 7 n using 350 μl of TRK Lysis Buffer provided by the E.Z.N.A. Total RNA Kit I (Omega Bio-tek, Norcross, GA) and stored at −80 °C.

2.6. RNA isolation and nanostring gene expression

After a gentle immersion in DPBS to remove non-attached cells as described in section 2.4.3, the scaffolds with adherent cells were placed into microcentrifuge tubes containing 350 μl TRK lysis buffer and stored at −80 °C for no longer than a week. RNA was isolated via the E.Z.N.A. Total RNA Kit I according to the manufacturer’s instructions. The quality of the RNA recovered was assessed by using Agilent Tapestation 2200TM Quality Assessment. Gene expression was measured using nCounter^® Myeloid Innate Immunity Gene Expression Panel V2.0 (Seattle, WA), using at least 100 ng of RNA per sample. Briefly, the RNA was hybridized overnight at 65 °C with reporter and capture probes, including probes for several positive controls and housekeeping genes. The RNA and probes were then washed and attached to a nCounter cartridge in the nCounter^®-NanoString Technologies Prep Station. The cartridge was then read in the nCounter Digital Analyzer.

2.7. Data analysis

NanoString data was imported into the nSolver^™ software V 4.0 (NanoString Technologies, Seattle WA) and analyzed using the nCounter Advanced Analysis plugin. Normalization, differential expression, and gene set analysis (GSA) were conducted on the raw data using nSolver [26]. Briefly, the software normalizes the data to the geometric mean of housekeeping genes selected using the geNorm algorithm [27]. The data was sorted by using housekeeping genes with counts larger than 100. The differential expression calculated the Log2 fold change (FC) values of the normalized data and the p-values adjusted using the Benjamini-Yekutiel method, which estimates the false discovery rate assuming there may be some biological connection between genes. GSA computed the directed global significance scores relative to a selected control. This value represented the overall differential expression of a particular gene set calculated using the signed square root of the mean squared t-statistic of genes. Full details of these algorithms can be found in the nCounter Advanced Analysis 2.0 Plugin for nSolver Software User Manual. The principal component analysis (PCA) was conducted with a custom R script (R version 3.6.1, R Studio version 1.2.5001) using the ‘prcomp’ function and normalized nSolver data [28]. Graphs were prepared using the ‘ggplot2’ and ‘Complex Heatmap’ packages in R [29, 30].

3. Results

3.1. M0s (Mϕ + M-CSF) response to biomaterials

The targeted transcriptional response of Mϕs to several different biomaterial scaffolds (i.e. UBP, GalaFLEX^® (P4HB), Prolene^® (PP), TIGR^®, Vicryl^®, Mersilene^®, SERI^®, and Alloderm^®), was determined using the experimental design shown in figure 1(A). Human CD14+ were isolated from PBMCs and then cultured in media supplemented with M-CSF to differentiate them towards Mϕ, see supplemental figure s1 for CD14+ FACS identification. After 5 d of incubation time, M0s (Mϕs + M-CSF) were seeded onto the biomaterials for 48 h (endpoint of cell culture 7 d). The RNA was then isolated and processed via NanoString sequencing with a panel for innate immunity to evaluate the targeted transcriptional response.

Figure 1.

The experimental design approach and material screening. (A) Timeline for cell isolation, culture, seeding, and transcriptome analysis. (B) Images representing the microview of P4HB (absorbable) and PP (permanent) meshes. (i),(ii),(iv), and (v) SEM Images. (iii) and (vi) Plasma membrane fluorescently stained images of M0-like phenotype cultured on P4HB and PP. Plasma membranes were stained with cellMask^™ deep red. (C) PCA of M0s (Mϕ + M-CSF) cultured on various biomaterials. Each point represents the mean coordinates of the individual samples for each biomaterial. Each sample is the biological average from one reading of three replicates pooled together (n = 3).

Figure 1(B) shows representative SEM images for P4HB (absorbable) and PP (permanent) meshes. Notably, the knitted meshes show smooth monofilament fibers with a diameter around 150–175 μm. SEM images show a mainly uniform filament with similar substrate topography, curvature and contact areas. The Mϕs cultured on these materials were not subjected to any external loads or mechanical stimuli.

The other biomaterials tested, table 1, include biomaterials of different compositions (TIGR, SERI, Mersilene, and Vicryl) and decellularized human tissue (Alloderm). PCA using the population Z score of the normalized transcriptome data (figure 1(C)), was conducted to evaluate the M0s transcriptome response of the biomaterials relative to UBP control. Overall, most mesh materials localized to the middle of the dimension (Dim) 1—Dim 3 coordinate plane, which describes 60.5% of the transcriptome data variance (supplemental figure s3, other dimensions from the PCA).

Table 1.

A summary of the biomaterials used in the materials screening of this study [31–3,7].

Biomaterial	Material composition	Shape	Degradation ability
GalaFLEX®	Poly-4-hydroxybutyrate (P4HB)	Monofilament mesh	Absorbable
PROLENE®	Polypropylene (PP)	Monofilament mesh	Permanent
Alloderm®	Acellular dermal matrix	Sheet	Absorbable
Mersilene®	Polyethylene terephthalate polyester	Multifilament mesh	Permanent
SERI®	Silk fibroin	Multifilament mesh	Absorbable
TIGR®	Two copolymers based on glycolide, lactide, and trimethylene carbonate	Multifilament mesh	Absorbable
VICRYL®	Polyglactin	Multifilament mesh	Absorbable

Alloderm, a human-derived acellular dermal matrix, had a differential response corresponding to a large increase in the Dim 3 axis with little change in Dim 1. Two absorbable meshes, TIGR (made from two copolymers based on glycolide, lactide, and trimethylene carbonate) and Vicryl (polyglactin 910), also saw an increase in Dim 3 with little change in Dim 1, but to a much lower extent.

On the other hand, transcriptome data for M0s cultured on SERI (absorbable silk-based mesh) and Mersilene (permanent polyester mesh) were located in a common region with an observable change in Dim 1. M0s cultured on UBP localized to quadrant III with only absorbable P4HB located closely in the same quadrant. PP mesh, a permanent biomaterial with a similar monofilament fiber structure (figure 1(B)) to P4HB mesh, in the PCA plot separated to the bottom right of the IV quadrant. This PCA suggests that P4HB and UBP are most similar on these axes, and a large differential transcriptome response was seen in M0s cultured on the absorbable P4HB compared to the M0s on PP.

3.2. M0 response to P4HB mesh

A more in-depth evaluation of the targeted transcriptome for M0s cultured on the absorbable P4HB mesh referenced to M0s cultured on UBP via NanoString mRNA expression profile is depicted in the volcano plot in figure 2(A). Here, a group of 133 genes showed a shift to the right demonstrating upregulation by the P4HB mesh compared to UBP (adj. p-value < 0.05 and Log2 (P4HB vs UBP) > 0). From this significantly upregulated group, 27 genes (represented in yellow) have a FC larger than three. CXCL8 and RASAL1 were two genes with the highest statistically significant FC observed. Meanwhile, from the 77 genes significantly downregulated by the material, 12 genes (represented in blue) presented a FC reduction equal or lower than a third compared to the UBP condition. To better understand the effect of this differential expression, a GSA was conducted to determine how P4HB-mesh affected each annotated pathway. This analysis is shown in figure 2(B), where the directed global significance scores are listed with a color scale to highlight the up or downregulation (yellow or blue color, respectively) of the annotated pathway (see supporting file LBL-10 397-01_nCounter_Hs_Myeloid_Innate_I, used with permission from NanoString Technologies, Inc).

Figure 2.

M0 (Mϕ + M-CSF) response to P4HB mesh. (A) Volcano plot representing the -Log 10 adjusted p-value and Log2 FC gene expression for Mϕs cultured on P4HB vs UBP. (B) GSA based on the differential expression testing of P4HB (column 1), the 4HB monomer (column 2), and fatty acid residues extracted from the P4HB (FTY, column 3). (C) Volcanos plots referencing the annotated pathways (i) TH1 Activation and (ii) Pathogen Response from the GSA table. The data represents n = 3 independent biological replicates. ‘- - -’ and ‘…’ represent the alpha for -Log10 adj. p-values equal to 0.01 and 0.05 respectively.

Overall, P4HB led to major changes in the differential expression, with all but two pathways upregulated. TH1 activation annotated pathway was the most significantly downregulated pathway. Along with this pathway, the extracellular matrix (ECM) remodeling pathway was also downregulated. A volcano plot for the genes annotated in the ECM remodeling pathway can be found in the supplemental figure s4(A).

On the other hand, genes within the cell migration and adhesion pathway were upregulated in P4HB compared to the UBP reference (figure 2(B) column 1, Volcano plot for Cell migration and adhesion pathway in supplemental figure s5). Similarly, the pathogen response pathway showed an upregulation of the genes clustered in this pathway. The pathogen response is of particular interest as recent reports show P4HB meshes elicit a pro-healing immune response and increased expression of antimicrobial peptides in animal models and with isolated macrophages [34]. Thus, the TH1 activation and pathogen response annotated pathways, enclosed in red, were selected for further analysis.

Next, the low molecular weight species 4HB and extracted FTY were tested with M0s to determine if the observed response resulted from the degradation product of the polymer P4HB or FTY residues present in the P4HB polymer from its production. GSA, seen in figure 2(B), columns 2 & 3, revealed similar trends with TH1 activation associated genes generally downregulated, and genes associated with pathogen response generally upregulated. However, there were small changes in the overall gene expression between 4HB and FTY under the conditions tested when compared to the full effect of the P4HB mesh.

The volcano plot in figure 2(C(i)) shows the genes associated with the TH1 activation annotated pathway for M0 cultured on P4HB mesh. The three most significantly downregulated genes (HAVCR, STAT1, and CCR1) had a Log2 FC of approximately −1.9 to −1.7. It is noticeable that IL10 was downregulated, however, it was not considered significant with an adjusted p-value of 0.0535. Figure 2(C(ii)) presents the Pathogen response pathway with the significant upregulation of 15 genes, including CXCL8, STAT6, and IL1B contrasted to 9 significantly downregulated genes such as CCL2, PYCARD, and ZMPSTE24.

3.3. Mϕs polarization on P4HB mesh upon polarizing stimuli

The effect of P4HB on Mϕ polarization was evaluated by performing an analogous GSA with polarized M1, M2a, and M2c like-phenotypes in reference to the M0s cultured on UBP.

Figures 3(A(i) and (ii)) illustrate cell culture conditions and cytokine supplementations to the culture media to polarize the Mϕs (described in section 2.5.3 Mϕ Seeding and Polarization), which were then transferred to the biomaterial or UBP as control. Mϕs polarized on UBP used during this work were compared across three different donors (A, B, and C) and the expression of representative gene markers from each like-phenotypes are shown in figure 3(B). As expected, all representative gene markers had higher expression in their respective Mϕ like-phenotypes across all three donors shown in figure 3(B) as the diagonal yellow-orange color pattern. Additional genes identified from this analysis that are highly expressed in Mϕs polarized on UBP can be found in supplemental figure s2.

Figure 3.

M0 (Mϕ + M-CSF), M1 (LPS, IFNγ), M2a (IL4, IL13), and M2c (IL10) Mϕ response to P4HB (A) Schematic for (i) Mϕ polarization (ii) interaction between Mϕ and biomaterial (B) Polarization markers for M1, M2a and M2c-like phenotypes in Mϕ cultured on UBP across three donors. Colors represent the Log2 FC value referenced to non-polarized M0 UBP control corresponding to the same donor and gene. (C) Fluorescent images of M1, M2a and M2c-like phenotype Mϕs cultured on P4HB. For each condition the top left image is stained for the plasma membrane (Deep Red), the bottom left is transmitted light, and the right is a merge of the two. Scale bar = 130 μm. (D) Directed global significance scores for polarized Mϕs. Scores calculated for M0, M1, M2a, M2c like-phenotype in Mϕs on P4HB and UBP referencing a baseline of M0 on UBP.

Mϕs, seeded and polarized on P4HB and PP meshes, are shown in figure 3(C) shown in red at Day 7. The images show that polarized Mϕs attach to the biomaterials at the experimental endpoint, which is the same time used for RNA isolation.

Figure 3(D) shows the directed global significance scores of both P4HB and UBP for each polarization normalized to M0s cultured on UBP. Overall, the P4HB absorbable material did not interfere with the polarization behavior of Mϕs cultured except for specific pathways and conditions.

M1 like-phenotype controls (UBP_M1) clustered with P4HB_M1, with most pathways upregulated in both conditions. However, the genes found in the ECM remodeling pathway were downregulated by P4HB (volcano plot is shown in supplemental figure s4(b)). In the case of the M2a like-phenotype, the genes annotated in the TH1 activation pathway in P4HB_M2a were downregulated, contrasting the upregulation pattern typically seen in UBP_M2a. This effect can also be seen in the P4HB_M2c condition whose annotated genes show an overall downregulation compared to the upregulated reference, UBP_M2c. The downregulation of the TH1 activation pathway by P4HB was also seen in M0s, indicating that the change in gene expression relative to UBP is conserved even in the presence of polarizing cytokines. Meanwhile, the pathogen response pathway was upregulated in P4HB_M2a conditions contrary to what was observed for its control UBP_M2a where the genes annotated into the pathway were downregulated. The heat map also shows other annotated pathways such as toll-like receptor (TLR) signaling, cytokine signaling, and Fc receptor signaling that behave in a different manner than the UBP control.

3.4. Pathogen-related gene transcriptome between P4HB vs PP

The Log2 FC for the genes annotated in the pathogen response pathway was extracted and contrasted as follows: P4HB vs PP and P4HB vs UBP using a dendrogram analysis presented in figure 4. CXCL8 was less expressed in P4HB than in PP, but it was significantly upregulated for Mϕs cultured and polarized on P4HB than on UBP. Interestingly, CAMP was upregulated, but these changes were not considered before due to the low statistical significance in the general data analysis. Also, IL1B, P2RX1, and NLRP3 were upregulated regarding both P4HB and polarization. In the middle section of the dendrogram, CCL5 was upregulated for both references under M2a-like conditions. Similarly, located in the bottom section of the dendrogram, CCL4 shows the same trend of upregulation in P4HB under M2a cytokine polarization conditions.

Figure 4.

Heatmap for P4HB vs PP and P4HB vs UBP. The heat map shows the pathogen annotated genes. M0s and Mϕs upon polarization signaling are presented in the columns. M0 (Mϕ +M-CSF), M1 (LPS, IFNγ), M2a (IL4, IL13), and M2c (IL10).

At the bottom, CCL2 was downregulated for all the conditions tested. The pathogen response profile under various polarization conditions prompted further global evaluation of PP and P4HB meshes.

3.5. Particularities on Mϕ response between P4HB and PP

To compare the Mϕ response to absorbable P4HB with a physically comparable permanent mesh, the differential expression of the M0s transcriptome in P4HB vs PP was analyzed (Volcano plot shown in figure 5(A)). 154 significantly upregulated genes were identified (adj. p-value <0.05), with 45 genes having a 3-FC or greater. There were also 77 significantly downregulated genes, with 14 genes having an FC reduced to 1/3x or lower. Similar to the M0 response seen in P4HB vs UBP, a trend towards upregulation was observed when contrasting P4HB vs PP. To assess how these changes may affect polarization, a PCA was performed with M0, M1, M2a, and M2c-like phenotypes cultured on either a UBP control, PP, or P4HB seen in figure 5(B).

Figure 5.

Macrophage (Mϕ) response to P4HB mesh compared to PP. (A) Volcano plot representing the Log2 FC gene expression for M0s cultured on P4HB vs PP. The data represents n = 3 independent replicates. ‘- - -’ and ‘…’ represent the threshold for −Log10 adj. p values equals 0.01 and 0.05 respectively. (B) Biplot with dimensions 1 and 3 of PCA of non-polarized and polarized Mϕ (Shape: M0, M1, M2a, and M2c) on various materials (Color: UBP, P4HB, and PP). (C) FC differences of statistically significant genes with an FC greater than 3 in M2a-like Mϕs cultured on UBP referenced to M0s culture on UBP. The expression of M2a-like Mϕ cultured on P4HB and PP for the same genes are also presented. The data represents the mean ± SEM. M0 (Mϕ + M-CSF), M1 (LPS, IFNγ), M2a (IL4, IL13), and M2c (IL10).

Dim 1 of this PCA describes 26.9% of the variance and separates the M1s (circle shapes) to the left, indicating that polarization strongly affects the gene expression of Mϕ cultured on all three materials. The M1s cultured on P4HB further separates into the upper half of the plot in Dim 3. The M0s and M2c-like phenotypes (square and diamond shapes respectively) of each material group together, though in distinct locations. This suggests that the polarization to M2c does not have a large effect on gene expression selected in this innate immunity panel compared to the variance caused by each material. Dim 3 separates the polarized phenotypes cultured on P4HB to the top half of the plot (quadrant I), while PP and UBP group in quadrant IV.

M2a-like phenotype (triangle shape) cultured on all three materials is located in the upper half of Dim 3, specifically in quadrant I alongside the M0s and M2cs cultured on P4HB. In P4HB and UBP, polarization towards an M2a-like phenotype results in a large shift up or increase in Dim3 compared to M0s of the corresponding material. However, this increase is not seen when polarizing to M2as on PP. Indeed, PP seems to reduce the change in gene expression associated with Dim 3 when polarizing to M2as, resulting in less separation from the corresponding M0s. Notably, it groups closely with M0s and M2cs cultured on P4HB.

Further analysis into the potential genes that are generating this difference in M2as cultured on UBP, P4HB, and PP was performed and presented in figure 5(C). Genes that were significantly changed (p-value < 0.05) with a FC greater than three when polarized to M2a-like phenotype cultured on UBP (M2a UBP ref. M0 UBP) were selected and plotted for all three materials. In M2as cultured on UBP, 29 genes were upregulated vs 28 downregulated (gold and blue color, respectively). Of the 29 upregulated genes, all genes were similarly upregulated in P4HB; however, seven genes were downregulated in PP. Of the 28 downregulated genes in M2as cultured on UBP, 8 genes were upregulated by P4HB and PP upregulated 10 genes. Overall, gene expression in P4HB tracks more closely to UBP than PP.

4. Discussion

Surgical meshes are commonly used to support tissue recovery by providing mechanical and structural support for breast reconstruction, hernia repair, reconstruction of the pelvic floor, and other applications [35, 38]. The local tissue response towards surgical meshes (i.e. absorbable or permanent) remains important for proper wound healing or optimal material integration [39–41]. Any biomaterial implanted into a site of injury will interact with the host’s dynamic wound environment where Mϕs are critical cells with roles in inflammation, angiogenesis, extracellular matrix remodeling, and cell recruitment [42]. As Mϕs adhere and interact with the surface of an implanted biomaterial, it is crucial to understand the Mϕ’s phenotypical changes when exposed in vitro to commercially available meshes [43]. The present study focused on two clinically available biomaterials: one absorbable (P4HB) and the other permanent (PP) (using ultra-low binding tissue culture plastic (UBP) as a control).

In addition to the material’s composition, Mϕ polarization can also be affected by material properties such as contact area, stiffness, surface chemistry, and material topography [44–46]. As the Mϕs cultured on the materials were not subjected to macroscopic mechanical stimuli, the mechanical properties of the meshes fell out of the scope of this work but some reports show potential similarities that may be of interest in future studies [47–49].

Initially, to understand how the Mϕs respond to different types of biomaterials, we measured the expression profile of targeted genes by M0s (Mϕs + M-CSF) when cultured in vitro on the surface of several clinically available biomaterials (biomaterials summarized in table 1) [31–37]. The screenings presented in figure 1(C) and supplemental figure s3 show a difference between permanent, absorbable, and naturally derived biomaterials. In order to understand the response of Mϕs towards different types of biomaterials, we first focused on biomaterials with varying surfaces such as multifilament biomaterials knitted into meshes (Mersilene^®, SERI^®, TIGR^® and Vicryl^®) and monofilament meshes such as P4HB and PP (figure 1(B) and table 1) [50]. Our screening also included representative biomaterials from synthetics (PP, P4HB, Mersilene^®, TIGR^®, and VICRYL^®) and biological sources (Alloderm^® and SERI^®). P4HB is a polyester obtained from a bacterial organism and PP is a polyolefin produced using catalytic reactions [51]. In general, these thermoplastic materials are extruded into fibers and converted into meshes, while silk-derived fibroin fibers are sourced from cocoons of the Bombyx mori silkworm, cleaned and converted into meshes. Alloderm^® is sourced from cadaveric human skin and was chosen as a representative absorbable acellular matrix [52]. Previous studies have reported, to some extent, the Mϕ response towards these biomaterials, but the in vitro response of human Mϕs towards P4HB based biomaterials remains mostly under-explored [34, 53–59].

Targeted transcriptome comparison using PCA of non-polarized Mϕs (figure 1(C)) showed that multifilament materials clustered together, Alloderm^® and PP clustered away from the rest of the materials, and P4HB clustered with UBP. Interestingly, PP showed a significant phenotype change, separating from both mono- and multi-filament mesh materials. P4HB meshes also have absorbable and degradative properties, while PP is permanent and non-degradable (table 1). While some material properties were controlled, such as shape and topography, the differential Mϕ transcriptome changes between P4HB and PP could be attributed to the polymer composition, substrate stiffness, or surface chemistry. A previous study showed how P4HB-based meshes could promote neovascularization and reduce inflammatory and fibrotic responses when implanted in a porcine hernia model [60]. Our in vitro characterization results provide further evidence that the absorbable P4HB scaffold could influence the expression of pro- and anti-inflammatory genes in M0s suggesting potential immunomodulatory effects (figure 2(A)). However, this remains to be entirely determined by direct comparison with in vivo responses.

The transcriptome-targeted analysis performed suggests that P4HB stimulates the M0s towards a co-expression of genes associated with both pro-inflammatory and anti-inflammatory phenotypes. P4HB upregulated a group of genes associated with inflammatory phenotypes (e.g. ITGFB7, IL1A) as well as genes related to anti-inflammatory phenotypes such as MMP12, FCGR2B, and KLF4. Interestingly, P4HB has been shown to upregulate KLF4, which is a critical transcription regulator factor of Mϕ polarization, suggesting that P4HB could stimulate a transition to an anti-inflammatory state [61]. This anti-inflammatory effect by P4HB resulted in the downregulation of genes annotated as part of the TH-1 Activation group, where HAVCR2, CCR1, and STAT1 were downregulated ~3–4 fold (figure 2(C(i))). Generally, STAT1 and STAT6 are thought to be antagonistic, with a downregulation of STAT1 helping to shift the Mϕ phenotype from M1-like to M2-like phenotype [62–65]. This gene expression profile is reflected in the transcriptome data with the downregulation of STAT1 and upregulation of STAT6 by P4HB, as shown in figures 2(C(i)) & (C(ii)) for the TH-1 activation and pathogen response genes, respectively.

Clinical studies as well as in vivo animal studies have suggested that the use of P4HB may enhance the host defense response against microbial infections [6, 66–68]. In our in vitro platform, P4HB mesh affected the expression of genes annotated in the pathogen response pathway for all Mϕ polarizations, as seen in figure 3(C), but the specific mechanism has yet to be determined. STAT6 and CXCL8 (IL-8), genes annotated in the pathogen response pathway, were significantly upregulated in M0s by P4HB. Furthermore, IL-8 is a chemokine involved in the recruitment of neutrophils and other leukocytes in response to pathogens, which may provide insights on the potential mechanism of the antimicrobial effects shown by P4HB. The data in figure 4 also showed an upregulation of CAMP (Cathelicidin also called LL-37), an antimicrobial peptide, whose upregulation in mice (Cramp gene) provides resistance to bacterial contamination [58, 66, 69]. CCL5, another gene in the pathogen response that acts as a chemokine for lymphocytes recruitment as well as activates antigen-specific T-cells, was upregulated in M2a P4HB compared to PP or UBP, suggesting a unique response to P4HB [70]. The co-expression of genes related to inflammatory, pathogen response, and the anti-inflammatory phenotypes may suggest a unique in vitro polarization state elicited by the P4HB mesh under these culture conditions.

In wound healing, the M1-like phenotype is important during the acute phase. The transition to anti-inflammatory phenotypes (M2a-, M2c-like) help to advance the wound healing response towards tissue restoration and ECM deposition [71]. Therefore, the comparison between P4HB and PP and how this could affect the transcriptome signaling in in vitro-derived Mϕs is critical to understanding the in vivo response and potential benefits. Mϕs cultured on P4HB showed a distinct gene expression profile when the polarizing cytokines were added. A particular transcriptome response was seen in stimulated M2a-like macrophages cultured on the materials, as seen in figures 5(B) and (C). The Mϕs cultured on PP and stimulated with IL4 and IL13 were not able to fully match the M2a-like Mϕs in UBP. PP promoted a gene expression profile closer to M0 or M2c-like phenotypes when compared to P4HB (based on the genes targeted in the panels). Even though an in vivo study of P4HB and PP in a murine model showed M2-like cells around both biomaterials, a larger M2-like population surrounding P4HB based scaffolds was reported [34]. Our results not only support this M2-like shift by P4HB, but the data suggest that the subpopulation transcriptome closest resembles the M2a-like gene profile.

It is important to note that the human wound healing response is complicated and involves many diverse cell types, cytokines, and intricate signaling cascades. Although the present study considered a wide range of representative gene expression markers for Mϕs, the study is limited to one immune cell under laboratory conditions. Also, the specific cytokine concentrations used to stimulate Mϕ polarization in vitro and the short culture duration may not fully recapitulate the phenotype and time scale typically observed in vivo. Even with these limitations, identifying large-scale transcriptome changes in key cells such as Mϕs can enhance our understanding of biomaterial-cell interactions and help guide future use and material design. This study allowed us to further understand previously published in vivo animal studies by targeting the in vitro human Mϕ gene expression. The targeted gene expression panel used provided a good representation of genes associated with inflammatory and anti-inflammatory responses. Additionally, while the Mϕs are a crucial component in wound healing, crosstalk between cell types affects the host response and should be considered in future studies. The overall outcomes in this study aligned to the response observed in clinical and in vivo studies. Moreover, the present study offers a new look into the in vitro responses of human Mϕ and valuable insights into the potential clinical response.

5. Conclusions

This work elucidates the changes in the targeted transcriptomic response of human Mϕs to absorbable P4HB and permanent PP meshes in vitro. M0s cultured on P4HB resulted in a phenotype with both pro-inflammatory and anti-inflammatory transcriptome signature. Further, the targeted transcriptome profile from human Mϕs aligns with the pathogen response and increased M2a-like Mϕ population as reported in the literature for other models.