Factors Affecting the Evaluation of Collagen Deposition and Fibrosis In Vitro

Abstract

Immune responses to biomedical implants, wound healing, and diseased tissues often involve collagen deposition by fibroblasts and other stromal cells. Dysregulated collagen deposition can lead to complications, such as biomaterial fibrosis, cardiac fibrosis, desmoplasia, liver fibrosis, and pulmonary fibrosis, which can ultimately result in losses of organ function or failure of biomedical implants. Current in vitro methods to induce collagen deposition include growing the cells under macromolecular crowding conditions or on fibronectin-coated surfaces. However, the majority of these methods have been demonstrated with a single cell line, and the combined impacts of culture conditions and postculture processing on collagen deposition have not been explored in detail. In this work, the effects of macromolecular crowding versus fibronectin coating, fixation with methanol versus fixation with paraformaldehyde, and use of plastic substrates versus glass substrates were evaluated using the WI-38 human lung fibroblast cell line. Fibronectin coating was found to provide enhanced collagen deposition under macromolecular crowding conditions, while a higher plating density led to improved collagen I deposition compared with macromolecular crowding. Collagen deposition was found to be more apparent on plastic substrates than on glass substrates. The effects of primary cells versus cell lines, and mouse cells versus human cells, were evaluated using WI-38 cells, primary human lung fibroblasts, primary human dermal fibroblasts, primary mouse lung fibroblasts, primary mouse dermal fibroblasts, and the L929 mouse fibroblast cell line. Cell lines exhibited enhanced collagen I deposition compared with primary cells. Furthermore, collagen deposition was quantified with picrosirius red staining, and plate-based drug screening through picrosirius red staining of decellularized extracellular matrices was demonstrated. The results of this study provide detailed conditions under which collagen deposition can be induced in vitro in multiple cell types, with applications including material development, development of potential antifibrotic therapies, and mechanistic investigation of disease pathways.

Impact Statement

This study demonstrated the effects of cell type, biological conditions, fixative, culture substrate, and staining method on in vitro collagen deposition and visualization. Further the utility of plate-based picrosirius red staining of decellularized extracellular matrices for drug screening through collagen quantification was demonstrated. These results should provide clarity and a path forward for researchers who aim to conduct in vitro experiments on collagen deposition.

Introduction

Collagen deposition by fibroblasts and other stromal cells is a physiologic event that is involved in a variety of biologic processes, including soft tissue fibrosis, response to multicellular parasites, encapsulation of medical device implants, and wound healing. During the remodeling stage of wound healing, myofibroblasts secrete Type III collagen and later Type I collagen, after which collagen reorganization and crosslinking leads to increased fiber strength in the wound.¹ Although collagen deposition is critical for wound healing, dysregulated collagen deposition can also lead to complications, such as biomaterial fibrosis,² cardiac fibrosis,³ desmoplasia,⁴ liver fibrosis,⁵ and pulmonary fibrosis.⁶ In fibrosis, the excessive accumulation of extracellular matrix proteins can lead to losses in organ function⁷ or failure of biomedical implants.⁸ Recapitulating the factors that can induce dense fibrillar collagen deposition by fibroblasts is of broad interest for fields spanning material development, mechanistic investigation of disease pathways, basic biology of cell–extracellular matrix (ECM) interactions, and evaluation of potential antifibrotic therapies. Specifically, in vitro models can enable rapid and high-throughput evaluation of fibrillar collagen deposition in response to physical or biological stimuli; however, the widespread adoption of in vitro collagen deposition models is prevented by a lack of understanding of each model’s biological relevance or limitations, as well as the potential for synergy between different approaches that have previously been treated as separate. It is important to develop standardized in vitro assays to enable detailed evaluation of collagen deposition for a variety of applications.

Two in vitro approaches that have been shown to be successful in inducing fibroblast collagen deposition include macromolecular crowding^9–11 and fibronectin surface coating.^12,13 In macromolecular crowding, it is proposed that a pseudo-3D environment is needed to induce collagen deposition. This “crowded” environment is achieved by dissolving macromolecules, such as Ficoll, a hydrophilic polysaccharide, in the culture medium. Fetal bovine serum (FBS) is thought to make a small contribution to macromolecular crowding,¹⁴ but some reports have found that it has a significant impact on collagen deposition,¹⁵ and macromolecular crowding experiments typically involve low concentrations of FBS in the treatment medium, although the mechanism of this importance is unknown. Collagen deposition through macromolecular crowding in conjunction with transforming growth factor (TGF)-β1 stimulation and l-ascorbic acid treatment has been demonstrated for WI-38 human lung cells and primary human lung cells from patients with idiopathic pulmonary fibrosis in “Scar in a Jar” studies.^10,11 These experiments have ranged from 3 to 6 days of treatment with “crowded” media, after which cells are fixed and stained for markers, such as collagen I, collagen IV, fibronectin, and alpha smooth muscle actin (α-SMA), a marker for activated and myofibroblasts. Importantly, these experiments appear to have exclusively been conducted with cells cultured on plastic surfaces, such as 96-well or 24-well plates.

In a different approach that does not rely on macromolecular crowding, simply growing WI-38 cells on a fibronectin-coated surface has been demonstrated to lead to the deposition of collagen and other ECM proteins. In these experiments, cells have been cultured on fibronectin-coated surfaces for 7 to 10 days, after which the resulting cell–ECM layers are decellularized and then stained for extracellular matrix components fibronectin, collagen VI, collagen I, collagen IV, decorin, and versican. Inclusion of l-ascorbic acid in the culture medium was found to lead to increased collagen I content in the resulting ECM matrices.¹² The role of the fibronectin coating in inducing fibrillar collagen deposition is not entirely clear. In vivo, fibronectin is involved in the process of hemostasis,¹⁶ which can be considered as one of the early stages of wound healing. Thus, it is reasonable that the presence of fibronectin in in vitro cultures may trigger fibroblasts to deposit ECM proteins, such as collagen I.

Several other factors also need to be considered in in vitro collagen deposition experiments. First, the fixative used plays a role in the detection and visualization of collagen deposition. In immunostaining experiments, Chen et al. found that methanol fixation led to improved collagen I preservation compared to fixation with 4% paraformaldehyde (PFA).¹¹ Second, the culture surface appears to be given little consideration in most experiments. As culture surfaces are themselves made of a variety of materials (e.g., glass versus plastic) with different coatings and chemical modifications (e.g., plasma treatment), their influence on the cell behavior cannot be ruled out without exploration. Third, the majority of these studies have been restricted to a single cell line, WI-38 cells (a human lung fibroblast cell line), which limits the potential for adapting these models for applications outside of lung fibrosis. Additionally, the functional behavior of primary cells versus cell lines must be explored. The development of in vitro models that are compatible with both mouse cells and human cells is also of utmost importance, especially because in vivo experiments are typically done with rodent models, and in the case of any potential disagreements between in vitro human cell behavior and in vivo mouse cell behavior, it would not be clear whether the differences are due to in vitro versus in vivo cell behavior or mouse versus human cell behavior. Furthermore, the use of alternative stains, for example picrosirius red (PSR), which chemically stains collagen regardless of type, can enable the visualization of ECM proteins of interest without such artifacts or the need for antibody-dependent staining. This can also potentially facilitate faster and cheaper scaling of these experiments for greater throughput.

Although a number of reports have explored various thickening agents and culture conditions to optimize macromolecular crowding conditions,¹⁵ a direct comparison between macromolecular crowding and fibronectin coating protocols has not been made. In this work, we compare the two major in vitro fibrillar collagen deposition methods by evaluating the effects of various tissue culture parameters, including macromolecular crowding, fibronectin coating, cell plating density, cell type (primary vs. cell lines; lung vs. skin fibroblasts), cell species (human vs. mouse), culture surface (glass vs. plastic), decellularization, and staining method on the observation of in vitro collagen deposition. To this end, we conducted experiments in which collagen deposition by macromolecular crowding and fibronectin coating methods were compared for WI-38 cells. We further compared collagen deposition by WI-38 cells cultured in 96-well plastic plates and 8-chamber glass slides, as well as collagen quantification with methanol fixation and PFA fixation. We then expanded these studies to use two different methods (antibody-based staining and PSR staining) to compare collagen deposition by 6 different cell types (WI-38 cells, human primary lung fibroblasts [HLFs], human primary dermal fibroblasts [HDFs], L929 cells, mouse primary lung fibroblasts [MLFs], and mouse primary dermal fibroblasts [MDFs]) when grown on fibronectin-coated surfaces. Lastly, we evaluated the ability of PSR staining to be used in a well plate format for high-throughput screening. An overview of these experiments is provided in Figure 1.

FIG. 1.

Overview of experiments.

Method

Materials

l-ascorbic acid (BP351-500), Rapamycin (AAJ62473MF), and Galunisertib (69–565) were purchased from Fisher Scientific (Hampton NH). Human TGF-β1 (100-21-10UG) was purchased from PeproTech (Cranbury NJ) and dissolved according to the manufacturer’s instructions. Ficoll PM70 (F2878-50G, Fisher Scientific), Ficoll PM400 (F4375-10G), fibronectin bovine plasma (F1141), Tween 20 (P9416), Triton X-100 (X100), and ammonium hydroxide solution (221228-25ML-A) were purchased from Millipore Sigma. 200 proof ethanol (CAS # 64-17-5) was purchased from the Warner Graham Company (Cockeysville, MD). Methanol (A452-4) was purchased from Fisher Scientific. Phosphate-buffered saline (PBS), pH 7.4, 1X (114-058-131) was purchased from Quality Biological.

L929 mouse fibroblasts (ATCC CCL-1), WI-38 human lung fibroblasts (ATCC CCL-75), HDFs (ATCC PCS-201-012), and HLFs (PCS-201-013) were purchased from ATCC. MLFs and MDFs were isolated from C57BL/6 mice. Lot-specific information for the HDFs and HLFs is provided in Table 1.

Table 1.

Donor Information for HDFs and HLFs

Cell type	Donor tissue source	Donor gender	Donor age	Donor race
HDF	Skin	Female	39	Caucasian
HLF	Lung	Female	29	White

HDF, primary human dermal fibroblasts; HLF, primary human lung fibroblasts.

Antibodies used to stain mouse cells included Proteintech (Rosemont NY) collagen Type I rabbit anti-human, mouse, porcine, rat, polyclonal antibody (50–172-9648, Fisher Scientific), Invitrogen donkey anti-rabbit IgG (H + L) highly cross-adsorbed secondary antibody, Alexa Fluor Plus 555, (PIA32794, Fisher Scientific), Invitrogen α-smooth muscle actin monoclonal antibody (1A4), and Alexa Fluor 488 (50–112-4634, Fisher Scientific). Antibodies used to stain human cells included monoclonal anti-collagen, Type I antibody produced in mouse (C2456, Millipore Sigma, Burlington MA), anti-mouse IgG (H + L), highly cross-adsorbed, CF 488A antibody produced in goat (SAB4600043, Sigma); and anti-actin, α-smooth muscle-Cy3 antibody, mouse monoclonal (C6198, Sigma). NucBlue Fixed Cell ReadyProbes Reagent (DAPI) (Thermo Fisher, R37606) was used for the nuclear counterstain for all cell types.

PSR solution (ab246832) was purchased from Abcam (Boston, MA). 96-well black-walled clear-bottom tissue culture-treated plates (Corning, 3603, Corning NY) and 8-chamber tissue culture-treated glass slides (Corning 354118) were purchased from Corning and Millipore Sigma, respectively.

Primary mouse fibroblast isolation

Mouse dermal fibroblasts

C57BL/6 mice (Jackson Labs, Bar Harbor ME) were humanely euthanized with carbon dioxide gas followed by cervical dislocation (NIH ACUC protocol # NIBIB23-01). After euthanization, gauze pads and cotton swabs were used to apply Nair hair removal cream to the mice dorsal skin. After a few minutes, gauze pads were used to wipe off the Nair hair removal cream and any removed hair. This process was repeated until the dorsal skin was sufficiently cleared of hair. The dorsal skin was then sterilized through the application of povidone iodine prep solution (Dynarex, Orangeburg NY), followed by 70% ethanol. This process was repeated twice, for a total of 3 times. Afterward, dorsal skin was lifted using forceps, and cut away from the mouse. The skin was immediately placed in a sterile solution of ice-cold 1% penicillin–streptomycin (ThermoFisher, Waltham MA in PBS). Afterward, the skin was patted dry with Kimwipes, weighed, and placed in a fresh tube containing a sterile ice-cold solution of 1% penicillin–streptomycin in PBS. The remainder of this protocol was adapted, with some modifications, from Judson et al.’s protocol for fibro–adipogenic progenitor isolation.¹⁷ The tube containing the skin and penicillin–streptomycin solution was incubated on ice for 10 min. In a biological safety cabinet, the skin was transferred to a well of a 6-well ultra-low-attachment plate. Surgical scissors were used to cut the skin into small pieces, after which it was digested with 2 mL warm Liberase solution (05401127001, Roche, dissolved at a concentration of 0.5 mg/mL in phenol red-free Dulbecco’s modified Eagle’s medium (DMEM)/F12 medium, ThermoFisher, Waltham MA). The plate was incubated at 37°C for 30 min without shaking under standard atmospheric conditions. After incubation, syringe plungers from 6 mL syringes were used to gently and thoroughly mash the tissue. Afterward, 2 mL sterile ice-cold PBS was added to the well, and the tissue sludge was transferred to a 50 mL conical tube kept on ice. The conical tube was topped off with ice-cold PBS for a final volume of approximately 40 mL. The tissue was then centrifuged at 130g for 5 min at 4°C, after which the supernatant was discarded. The tissue pellet was then digested with a further 2 mL warm Liberase solution on a shaker incubator for 30 min at 37°C with shaking set to 100g. The tube was gently vortexed for 10 s after the first 15 min of incubation, before allowing the remainder of the incubation to continue. The tissue homogenate was then resuspended in 10 mL ice-cold sterile fluorescence activated cell sorting (FACS) buffer (Ca²⁺/Mg²⁺-free PBS supplemented with 2% FBS and 2 mM EDTA, pH-adjusted to a pH of 7.9 before sterile filtration), mixed, and placed on ice. The tissue homogenate was then filtered through a 40 µm cell strainer into a fresh sterile 50 mL conical tube. The tube was topped up with ice-cold FACS buffer for a final volume of approximately 40 mL in the tube and centrifuged at 520g for 5 min at 4°C. Afterward, the cell pellet was resuspended in 0.5 mL to 1 mL of complete culture medium and counted. Cells were plated in tissue culture–treated flasks at a density of 50,000 live cells/mL (5 mL per T-25 flask, 15 mL per T-75 flask) and incubated under standard growth conditions (37°C, 5% CO₂). The media were replaced 4 days after initial culture, and then every 2–3 days afterward.

Mouse lung fibroblasts

C57BL/6 mice were humanely euthanized with carbon dioxide gas followed by cervical dislocation (NIH ACUC protocol # NIBIB23-01). After euthanization, an incision was made along the entire ventral side, cutting through all layers until the ribcage was visible. The skin was pinned aside, and an incision was made along the length of the ribcage, after which the lungs were located and cut from the surrounding tissues. The tissue was placed in a fresh tube containing a sterile ice-cold solution of 1% penicillin–streptomycin in PBS and was processed in the same manner as described for the MDFs above.

Cell culture

L929 cells, MLFs, and MDFs were cultured in DMEM (Quality Biological 112-014-101, Gaithersburg MD) completed with heat-inactivated FBS (Quality Biological 110-001-101HI) and 100X penicillin–streptomycin solution (30-002-CI, Corning) for a final concentration of 10% FBS and 1% penicillin–streptomycin solution by volume. WI-38 cells were cultured in Eagle’s minimum essential medium (EMEM; ATCC 30–2003, Manassas VA) completed with heat-inactivated FBS (Quality Biological 110-001-101HI) and 100X penicillin–streptomycin solution for a final concentration of 10% FBS and 1% penicillin–streptomycin solution by volume. HDFs and HLFs were cultured in fibroblast basal medium (ATCC PCS-201-030) completed with fibroblast growth kit low serum (ATCC PCS-201-041) and 0.5 mL penicillin–streptomycin–amphotericin B solution (ATCC PCS-999-002).

Before plating, cells at passage 10 or lower were cultured using the corresponding media composition described above for each cell type and grown to 70–100% confluence in tissue culture treated flasks (CellTreat Scientific Products, Pepperell MA). Cell layers were rinsed with sterile PBS, trypsinized (0.25% Trypsin-EDTA [25200-056, ThermoFisher] for WI-38, L929, MLFs, and MDFs; Trypsin-EDTA for primary cells [PCS-999-003, ATCC] for HLFs and HDFs), and then neutralized with culture medium (WI-38, L929, MLFs, and MDFs) or trypsin neutralizing solution (ATCC PCS-999-004, HLFs and HDFs). After centrifugation, cells were suspended in medium, counted on an Invitrogen Countess II cell counter, and were further diluted in the appropriate medium before being plated at the desired plating density. The number of live cells per milliliter was used for all cell plating calculations.

Comparison of biological conditions for different human cell types

WI-38 cells, HDFs, and HLFs were cultured in glass 8-chamber slides in the presence or absence of a fibronectin coating using a high cell plating density, as well as with a lower cell plating density under macromolecular crowding conditions. The day of experiments, chambers that were to be fibronectin-coated were incubated with fibronectin (50 µg/mL in PBS; 300 µL for each chamber in 8-chamber glass slides) at 37°C, 5% CO₂ for 1 h. After 1 h of incubation, the fibronectin solution was discarded and cells were plated.

To replicate the macromolecular crowding protocol,¹¹ WI-38 cells, HDFs, or HLFs were plated in complete culture medium, without any further additions, at a concentration of 26,000 cells/cm² (low plating density). The following day, this medium was replaced with complete culture medium with 0.5% FBS, in addition to 37.5 mg/mL Ficoll PM-70, 25 mg/mL Ficoll PM-400, 50 µg/mL l-ascorbic acid, and 5 ng/mL TGF-β1). Cells were cultured for a further 6 days for a total of 7 days in culture.

To replicate the fibronectin coating protocol,¹² WI-38 cells, HDFs, or HLFs were plated at a concentration of 140,000 cells/cm² (high plating density) in the complete culture medium with the addition of 50 µg/mL l-ascorbic acid. This medium was unchanged throughout the experiment and cells were cultured for a total of 10 days after plating.

To prepare the media containing l-ascorbic acid, the appropriate volume of l-ascorbic acid stock solution (25 mg/mL in sterile molecular biology grade water; prepared on the day that cells were to be plated) was diluted with the appropriate cell culture medium.

At the end of the experiment, cells were fixed with ice-cold methanol before immunostaining.

Comparison of culture substrate

Glass 8-chamber slides (Nunc, Roskilde Denmark) were coated with fibronectin as described above. Then, 96-well plates were coated with fibronectin in the same manner but using 100 µL of fibronectin solution for each well. WI-38 cells were plated in culture medium containing 50 µg/mL l-ascorbic acid at the high plating density in fibronectin-coated 8-chamber glass slides or 96-well plates and cultured for 10 days without medium replacement. Cells were fixed with ice-cold methanol and then immunostained.

Comparison of fixative

WI-38 cells were plated in culture medium containing 50 µg/mL l-ascorbic acid at the high plating density in fibronectin-coated 8-chamber glass slides for 10 days without medium replacement. Cells were fixed with ice-cold methanol or with room temperature 4% PFA. For PFA-fixed cells, cells were further permeabilized with 0.5% Triton X solution (See “Immunostaining” section for full details).

Comparison of murine cell types

L929 cells, MDFs, and MLFs were plated in culture medium containing 50 µg/mL l-ascorbic acid at the high plating density in fibronectin-coated 8-chamber glass slides for 10 days without medium replacement. Cells were then fixed with ice-cold methanol before immunostaining.

Comparison of collagen I antibodies used to stain human and murine cells

WI-38 cells were cultured in culture medium containing 50 µg/mL l-ascorbic acid at the high plating density in fibronectin-coated 96-well plates for 10 days without medium replacement. Cells were then fixed with ice-cold methanol before immunostaining. Four wells were stained with the primary and secondary antibodies that were typically used for staining human cells, while four wells were stained with the primary and secondary antibodies that were typically used for staining murine cells.

Detection of collagen deposition by picrosirius red staining

WI-38 cells, HDFs, HLFs, L929 cells, MDFs, and MLFs were plated in culture medium containing 50 µg/mL l-ascorbic acid at the high plating density in fibronectin-coated 8-chamber glass slides for 10 days without medium replacement. Wells were then decellularized, fixed, and stained with PSR (see “Decellularization and Picrosirius Red staining” section for full details).

Collagen quantification with picrosirius red staining

WI-38 cells were plated in culture medium containing 50 µg/mL l-ascorbic acid in fibronectin-coated 96-well plates for 10 days without medium replacement. Plating densities were varied from 100% (i.e., high plating density) to plating densities equivalent to 75%, 50%, 25%, and 12.5% of the high plating density. Wells were then decellularized, fixed, and stained with PSR.

Drug screening experiment

WI-38 cells were plated in complete culture medium in fibronectin-coated 96-well plates for 2 days. Afterward, the medium was replaced with medium containing 5 ng/mL TGF-β1 and 50 µg/mL l-ascorbic acid, in addition to 250–4000 ng/mL Rapamycin or Galunisertib. Cells were cultured for a further 8 days without medium replacement. Wells were then decellularized, fixed, and stained with PSR.

Immunostaining

Cells were rinsed once with 1X PBS pH 7.4, and then fixed with methanol (ice-cold; 2 min incubation on ice) unless otherwise noted, and then rinsed 3 times with PBS. Afterward, the cells were incubated with collagen I primary antibody working solution (anti-collagen Type 1 antibody diluted in PBS) for 1.5 h at room temperature, rinsed 3 times with PBS-Tween (PBS-T, 0.05% Tween in PBS), incubated with secondary antibody working solution (anti-IgG antibody diluted in PBS) for 1 h while protected from light, rinsed 3 times with PBS-Tween, incubated with α-SMA antibody working solution (anti-α-SMA antibody diluted in PBS) for 1.5 h while protected from light, rinsed 3 times with PBS-T, and then incubated with NucBlue working solution (NucBlue diluted in PBS) for 10 min at room temperature while protected from light. Images were then acquired on an EVOS Cell Imaging System. In cases where fixation with paraformaldehyde was used, after the initial rinse with 1X PBS pH 7.4, cells were fixed for 10 min with 4% PFA (room temperature), then rinsed 3 times with PBS, after which they were permeabilized with 0.5% Triton X solution (Triton X-100 diluted in PBS) for 10 min (room temperature) and rinsed 3 times further with PBS. The immunostaining then proceeded as described above.

The staining volumes and antibody dilutions are provided in Table 2 and Table 3, respectively. Images of secondary-only controls (stained with only the collagen I secondary antibody) are provided in Supplementary Figures S1-S3.

Table 2.

Volumes Used for Fixation, Rinses, and Immunostaining of Each Well in the Culture Vessels

	96-well plates (culture area = 0.32 cm2)	8-chamber glass slides (culture area = 0.7 cm2)
Collagen I Primary Antibody Working solution	100 µL/well	250 µL/well
Collagen I Secondary Antibody Working solution	100 µL/well	250 µL/well
α-SMA Antibody Working solution	100 µL/well	250 µL/well
NucBlue Working Solution	100 µL/well	250 µL/well
Methanol Fixation	100 µL/well	700 µL/well
PBS Rinses	100 µL/well per rinse	700 µL/well per rinse
0.05 % PBST Rinses	100 µL/well per rinse	700 µL/well per rinse

α-SMA, alpha smooth muscle actin; PBS, Phosphatebuffered saline; PBST, Phosphate buffered saline +0.05% Tween20.

Table 3.

Antibody Dilutions Used to Prepare Antibody Working Solutions

Cell type	Antibody	Dilution	Channel for visualization
Mouse	Collagen Type I rabbit anti-mouse	1:100	N/A
	Donkey anti-rabbit IgG, Alexa Fluor Plus 555	1:200	RFP
	α-SMA antibody, Alexa Fluor 488	1:100	GFP
Human	Collagen Type 1 mouse antihuman	1:1000	N/A
	Goat anti-mouse IgG, CF 488A	1:500	GFP
	Mouse anti-human α-SMA antibody, Cy3	1:1000	RFP

The NucBlue working solution (nuclear counterstain) for all conditions was prepared using 2 drops NucBlue solution per milliliter of PBS.

RFP, red fluorescent protein; GPF, green fluorescent protein.

Decellularization and picrosirius red staining

After 10 days of culture, media were discarded, and decellularization was conducted according to established protocols.¹² Briefly, decellularization was initiated by adding Triton X/NH₄OH decellularization solution (50 mM NH₄OH and 0.05% Triton X solution in water) to the cell layer. Cells were incubated with this solution for 5 min at room temperature. Decellularized matrices were rinsed once with NH₄OH solution (50 mM in water), rinsed once with PBS pH 7.4, and then fixed with 4% PFA in water (10 min at room temperature). Following this, decellularized matrices were rinsed once with PBS, then stained with PSR solution (1 h, room temperature), after which they were rinsed twice with 0.5% acetic acid and then rinsed twice with ethanol. The liquid was then discarded, and stained decellularized matrices were allowed to air-dry before image acquisition. Images were collected on an EVOS cell imaging system using the RGB channel. Owing to the known red fluorescence of PSR stain,¹⁸ images were also acquired on the red fluorescent protein (RFP) channel.

The reagent volumes for each step of the decellularization and PSR staining process are provided in Table 4.

Table 4.

Volumes Used for Decellularization, Rinses, and Fixation of Each Well in the Culture Vessels

	96-well plates (culture area = 0.32 cm2)	8-chamber glass slides (culture area = 0.7 cm2)
4 % PFA Fixation	100 µL/well	700 µL/well
PBS Rinse	100 µL/well	700 µL/well
Picrosirius Red	100 µL/well	250 µL/well
Ethanol Rinse	100 µL/well per rinse	250 µL/well per rinse
50 mm NH4OH Solution Rinse	100 µL/well	700 µL/well
Triton X/NH4OH Decellularization Solution	100 µL/well	700 µL/well
0.5 % Acetic Acid Rinse	100 µL/well per rinse	250 µL/well per rinse

PFA, paraformaldehyde; PBS, Phosphate-buffered saline.

Image quantification

All image quantification was conducted in ImageJ. The average fluorescence intensity of each channel was determined using the “Measure” tool. The PSR intensity was determined by first inverting the image and then using the “Measure” tool. All quantification was done with images taken at 4× magnification to enable incorporation of a large field of view that is representative of each well. In cases where the cell layer was peeling, quantification was conducted only on the nonpeeling portion of the cell layer.

Well plate absorbance and fluorescence measurements

Absorbance and fluorescence measurements were collected on a BioTek Synergy H1 plate reader. Absorbance was measured at 540 nm, while fluorescence was measured using an excitation wavelength of 540 nm and an emission wavelength of 580 nm.

Statistical analysis

Statistical analysis was performed in GraphPad Prism 9 using a combination of one-way analyses of variance (ANOVAs) assuming unequal variances with corrections for multiple measurements and t-tests assuming unequal standard deviations.

Experiment

Effects of biological conditions on human cell collagen I and α-SMA expression

To evaluate the effects of biological conditions on in vitro collagen I deposition and α-SMA expression, WI-38 cells, HLFs, and HDFs were cultured in glass 8-chamber slides under the following four conditions: high plating density (with and without fibronectin coating) and lower plating density (with macromolecular crowding, with or without fibronectin coating) (Fig. 2A–C). For WI-38 cells cultured with a high plating density, the fibronectin coating increased collagen I and α-SMA expression, although it had no significant effect on collagen I and α-SMA expression for cells cultured under macromolecular crowding conditions (Fig. 2D). Furthermore, the collagen I deposition was higher for WI-38 cells cultured with a high plating density than for WI-38 cells cultured under macromolecular crowding conditions. For HLFs, fibronectin coating had no significant effect on collagen I or α-SMA expression, regardless of culture at a high plating density or under macromolecular crowding conditions. Here again, high plating density led to increased collagen I deposition compared with macromolecular crowding conditions. For HDFs, both collagen I and α-SMA expression slightly decreased in the presence of a fibronectin coating for cells cultured with a high plating density, but had no significant change for those cultured under macromolecular crowding conditions. As with the other human cell types, HDFs exhibited increased collagen I deposition when cultured with a high plating density compared with HDFs cultured under macromolecular crowding conditions. HDFs also exhibited increased α-SMA expression when cultured with a high plating density compared with HDFs cultured under macromolecular crowding conditions.

FIG. 2.

Effects of biological conditions on collagen I deposition and α-SMA expression by WI-38 cells, primary human lung fibroblasts, and primary human dermal fibroblasts. (A–C) Fluorescence microscopy images of WI-38 cells (A), primary human lung fibroblasts (B), and primary human dermal fibroblasts (C) immunostained after 10 days of culture without medium replacement or after 7 days of culture with medium replaced with macromolecular crowding medium on the second day. Images are representative images from n = 4 collected at 20× magnification for each condition. (D) Quantification of average fluorescence intensity of images taken at 4× magnification for each marker (n = 4 for each condition). Bars represent average values, and error bars represent standard deviation. Statistical significance was determined using a one-way ANOVA assuming unequal variances and with correction for multiple comparisons. High plating density leads to greater levels of collagen I and α-SMA than macromolecular crowding conditions. Fibronectin coating increases the expression of α-SMA for WI-38 cells and HDFs cultured with the high plating density. α-SMA, alpha-smooth muscle actin; HDFs, primary human dermal fibroblasts; HD, high plating density; MMC, macromolecular crowding; FN, fibronectin. (E) Scatter plots of α-SMA expression versus collagen I expression. For some cell types, cells grown on fibronectin-coated surfaces form distinct clusters compared with those grown without fibronectin. The different biological conditions also lead to the development of distinct clusters, especially for cells plated at a high density on fibronectin-coated surfaces.

The effects of fibronectin coating for all three human cell types are also apparent when the α-SMA expression is plotted against collagen I deposition (Fig. 2E). Here, distinct clusters based on the presence and absence of fibronectin are observed for WI-38 cells cultured with a high plating density. For all three cell types, separating the fibronectin-coated from nonfibronectin-coated wells enables observation of further distinct clusters based on culture with high plating density or macromolecular crowding conditions. In particular, when grown on fibronectin-coated surfaces, WI-38 cells exhibit both high collagen I and high α-SMA expression, whereas HDFs exhibit low collagen I and low α-SMA expression. Cells cultured with a high plating density generally exhibit increased collagen I expression than their counterparts cultured under macromolecular crowding conditions. The high plating density with fibronectin coating was selected for use in further experiments due to the combination of its simplicity and favorable expression of collagen I and α-SMA.

Effects of culture substrate on WI-38 cell collagen I and α-SMA expression

To evaluate the effects of culture substrate on in vitro collagen I deposition and α-SMA expression, WI-38 cells were cultured in fibronectin-coated glass 8-chamber slides and plastic 96-well plates at the high plating density (Fig. 3A). Collagen I deposition by the WI-38 cells was higher on the plastic substrate than the glass substrate (Fig. 3B), but the substrate had no significant effect on the expression of α-SMA.

FIG. 3.

Effects of culture substrate on collagen I deposition and α-SMA expression by WI-38 cells. (A) Fluorescence microscopy images of cells immunostained after 10 days of culture without medium replacement. Images are representative images from n = 4 images collected at 20× magnification for each condition. (B) Quantification of average fluorescence intensity of images collected at 4× magnification for each marker. For the glass substrate, three out of four wells were quantified because the 4× magnification image for one of the wells was not saved. Bars represent average values, and error bars represent standard deviation. Statistical significance was determined using unpaired t-tests not assuming equal standard deviations. The plastic substrate increases collagen I deposition by WI-38 cells but does not significantly alter α-SMA expression. α-SMA, alpha-smooth muscle actin.

Effects of fixative on collagen I and α-SMA visualization for WI-38 cells

To evaluate the effects of fixative on visualization of collagen I deposition and α-SMA expression, WI-38 cells were cultured in fibronectin-coated glass 8-chamber slides at the high plating density. Although the images acquired at 20× magnification with methanol fixation appeared to show a stronger signal than PFA-fixed images at the same magnification (Fig. 4A), quantification of images collected with a larger field of view (4× magnification) indicated no significant difference in collagen I and α-SMA detection between the two fixatives (Fig. 4B).

FIG. 4.

Effects of fixative on visualization of collagen I deposition and α-SMA expression by WI-38 cells. (A) Fluorescence microscopy images of cells immunostained after 10 days of culture without medium replacement. Images are representative images from n = 4 images collected at 20× magnification for each condition. (B) Quantification of average fluorescence intensity of images collected at 4× magnification for each marker (n = 4 for each condition). Bars represent average values, and error bars represent standard deviation. Statistical significance was determined using unpaired t-tests not assuming equal standard deviations. The plastic substrate increases collagen I deposition by WI-38 cells but does not significantly alter α-SMA expression. α-SMA, alpha-smooth muscle actin.

Effects of murine cell type on collagen I deposition and α-SMA expression

To evaluate the effects of murine cell type on collagen I deposition and α-SMA expression, L929 cells, MLFs, and MDFs were cultured in fibronectin-coated glass 8-chamber slides at the high plating density (Fig. S4A). As with human cells, the murine cell type had a clear impact on the expression of both collagen I and α-SMA (Fig. S4B). L929 cells had the highest collagen I expression, followed by MLFs and then MDFs. Importantly, however, the visualized collagen I did not appear entirely fibrillar in nature, with the most intense signal appearing to be concentrated in or around cell nuclei. MLFs had the highest α-SMA expression, followed by MDFs, and then L929 cells.

Effects of antibody on collagen I visualization

To evaluate the potential role of antibody selection on collagen I visualization, WI-38 cells were cultured under the high plating density condition in a fibronectin-coated 96-well plate. After 10 days of culture without medium replacement, cells were methanol-fixed and stained for collagen I with a nuclear counterstain. Half of the wells were stained with the collagen I antibodies used for the human cell immunostaining experiments described above, while the remaining wells were stained with the collagen I antibodies used for the murine cell immunostaining experiments described above (Fig. S5). Although the collagen I antibody used for the murine cells has both antihuman and antimurine reactivity and has previously been used in immunostaining murine cells,¹⁹ we observed a lack of fibrillar collagen I in WI-38 cells stained with this antibody, similar to what we observed for the murine cells.

FIG. 5.

PSR staining of decellularized extracellular matrices of human Cells and primary mouse cells. (A) Fluorescence and brightfield microscopy images of matrices decellularized and stained with PSR after 10 days of culture without medium replacement. Images are representative images from three to four images collected at 20× magnification for each condition. (B) Quantification of average fluorescence and brightfield intensities of images collected at 4× magnification. Statistical significance was determined using a one-way ANOVA assuming unequal variances and with correction for multiple comparisons. ECM deposited by L929 cells could not be visualized due to peeling of the ECM layers during decellularization. One well was excluded from the analysis of the WI-38 ECM and MLF ECM due to a completely peeled ECM layer and an error in image storage, respectively. (C) Quantification of fluorescence and absorbance signals from PSR-stained ECM deposited by WI-38 cells in a 96-well plate assay. (D) Drug screening assay conducted using PSR stain after decellularization. WI-38 cells were treated with varying concentrations of Rapamycin and Galunisertib for 8 days. PSR, picrosirius red; ANOVA, analysis of variance; ECM, extracellular matrix; MLF, mouse primary lung fibroblasts.

Picrosirius red staining for collagen visualization in all cell types

To evaluate the feasibility of using PSR for collagen visualization in decellularized extracellular matrices, WI-38 cells, HLFs, HDFs, L929 cells, MLFs, and MDFs were cultured in fibronectin-coated glass 8-chamber slides at the high plating density for 10 days without medium replacement. After decellularization and matrix fixation, the decellularized matrices were stained with PSR and visualized with microscopy. Decellularized matrices from L929 cells could not be visualized due to peeling of the cell layer during the decellularization process. PSR staining enabled collagen visualization with both brightfield and fluorescence microscopy (Fig. 5A). Quantification of brightfield and fluorescence images revealed similar trends from both image types, with MLFs having the highest collagen deposition, and significantly higher deposition than HDFs (Fig. 5B). Additionally, HLFs were found to have higher collagen deposition than HDFs. Based on the analysis of brightfield images, the MLFs exhibited higher collagen deposition than the HLFs, and HLFs exhibited greater collagen deposition than the MDFs.

Well plate–based quantification of collagen deposition by WI-38s

To illustrate the utility of using PSR staining of decellularized extracellular matrix in a well plate–based assay for evaluation of collagen deposition, WI-38 cells were plated at varying concentrations in a 96-well plate, after which they were decellularized and stained with PSR. Based on the spectral characteristics of PSR, the absorbance (540 nm) and fluorescence (excitation wavelength of 540 nm, emission wavelength of 580 nm) were collected for each well (Fig. 5C). For low cell plating densities, both the absorbance and fluorescence measurements exhibited a trend of increased intensity with an increase in cell plating density. At higher cell plating densities, the signals appear to reach saturation, likely indicating saturation in the ability of cells beyond a certain plating density to deposit collagen.

Picrosirius red–based drug screening

To illustrate the utility of plate-based PSR staining of decellularized extracellular matrix for drug screening, WI-38 cells were plated under the high plating density condition, with the addition of TGF-β1 and an antifibrotic drug (Rapamycin or Galunisertib) after 2 days. These drugs were chosen due to their inhibition of TGF-β. A concentration-dependent reduction in collagen deposition is observed for both Rapamycin and Galunisertib (Fig. 5D). Additionally, lower concentrations of Rapamycin appear more effective at reducing collagen deposition than low concentrations of Galunisertib, although their efficacy appears more similar at higher concentrations.

Discussion

Overall, we have demonstrated clear differences in collagen deposition and α-SMA expression results based on culture conditions. For WI-38 cells, HLFs, and HDFs, collagen I expression improved with the use of a higher plating density versus a combination of macromolecular crowding and lower plating density. This indicates potential involvement of cell–cell proximity in inducing collagen deposition. Additionally, for WI-38 cells, the fibronectin coating was found to lead to improvement in collagen I and α-SMA expression for high plating density conditions. Although this trend was not observed for the HLFs and HDFs, and in fact the opposite effect was observed for the HDFs, previous studies have reported a role for fibronectin in collagen deposition.^20,21 It is possible that fibronectin serves both a biological role (e.g., contributing to deposition of collagen fibers) and a mechanical role (i.e., adhesion of the collagen fibers) for collagen deposition by some cell types. Interestingly, although the HLFs did not exhibit the same increased collagen I deposition in the presence of the fibronectin coating as exhibited by the WI-38 cells, their behavior was not as different from the WI-38 cells as the behavior of the HDFs, which exhibited an opposite effect. WI-38 cells are sourced from the lungs, and it is thus possible that the origin of HLFs also being from the lungs played a role in this observation. WI-38 cells are commonly used in collagen deposition experiments, but the use of cell lines versus primary cells should be considered in cases where obtaining the most accurate biological relevance is important. Since lung fibroblasts are often used in models of fibrosis, the relevance of these models for other applications, for example, the evaluation of biomedical implants that interact with other tissues such as the skin, must be taken into account. Despite this, culturing human fibroblasts under the high plating density conditions in the presence or absence of fibronectin appears to be a robust method to obtain deposition of fibrillar collagen I.

Interestingly, we observed low collagen I deposition by some cells that have high α-SMA expression (e.g., HLFs) and high collagen I expression by some cells that have low α-SMA expression (e.g., HDFs). Although activated fibroblasts or myofibroblasts have traditionally been defined as being α-SMA positive, recent literature has shown that this is not necessarily the case. Hsia et al. identified amine oxidase copper-containing 3 (AOC3) as a myofibroblast marker and found that AOC3-sorted myofibroblasts exhibit heterogeneous α-SMA staining.²² Sun et al. reported that collagen-producing cells that coexpress α-SMA were a minority in the fibrotic lung and kidney, and that the role of α-SMA-expressing fibroblasts in fibrosis differed based on the organ fibrosis model.²³ Zhao et al. used a murine muscular dystrophy model to demonstrate that α-SMA was not a functional marker of fibrogenic cells in skeletal muscle fibrosis.²⁴ In the context of these reports, especially taking into consideration Sun et al.’s finding that α-SMA-positive collagen-producing cells were in the minority in the fibrotic lung, our observation of low collagen production in the α-SMA-positive primary human lung cells appears reasonable.

The use of a plastic substrate versus a glass substrate led to increased collagen I deposition by the WI-38 cells. This confirms the importance of considering the culture substrate for in vitro collagen deposition studies, such as those conducted for screening antifibrotic compounds, as well as for the development of more complex tissue culture models of fibrosis and collagen deposition in the future. Although methanol fixation appears to lead to more visually appealing images and has previously been demonstrated to lead to better collagen I retention than fixation with 4% PFA,¹¹ we did not observe statistically significant differences in the average collagen I signal for methanol-fixed versus PFA-fixed WI-38 cells. However, the relative simplicity of methanol fixation compared with a longer fixation time and the need for additional permeabilization steps for PFA-fixed cells makes methanol fixation appealing for practical purposes.

Of the murine cells evaluated, L929 cells, which are a murine cell line originating from subcutaneous areolar and adipose tissue, appeared to have little to no α-SMA expression. Despite successful staining for α-SMA, we primarily observed intracellular collagen I staining for the murine cells, which indicated a potential role of the specific antibody used on these results. Using the same antibody, which also has antihuman reactivity, to stain WI-38 cells confirmed that the antibody primarily stained intracellular collagen. Thus, simple factors such as the choice of antibody can play a role in the observed collagen deposition. This further cements the need for alternatives to antibody-based stains to minimize variability based on the choice of reagents. We have also observed that the variability of TGF-β1 from one vendor to another can impact the production of both collagen I and α-SMA, which can lead to further challenges in inducing in vitro collagen deposition.

PSR staining, however, provided confirmation that the murine fibroblasts indeed also deposited fibrillar extracellular matrix. Here again, the differences in cell type became even more pronounced, with primary murine lung fibroblasts exhibiting very high amounts of ECM deposition compared with other cell types, and with the primary human and murine lung fibroblasts generally exhibiting greater ECM deposition than their primary dermal fibroblast counterparts. The overall differences in ECM deposition by murine and human fibroblasts further highlight the need to consider the species in the evaluation of both in vitro and in vivo results. Furthermore, although the cell lines exhibited greater collagen I deposition than primary cells for both murine and human cells, PSR staining revealed no statistically significant difference in collagen deposition between WI-38 cells and primary human cells. This is potentially a result of the staining of multiple collagen types with PSR stain, rather than staining of only collagen I with immunostaining. In wound healing, collagen III is typically deposited in the early stages, after which it is replaced with collagen I.²⁵ The higher expression of collagen I, but not increased PSR signal, for the WI-38 cells indicates that this cell line potentially deposits more mature collagen compared with the primary cells. Additionally, the processing of samples for each staining type must be taken into account. While the antibody-based staining was conducted directly on cells after fixation, the PSR staining was conducted after decellularization and fixation. Thus, potential differences in the content of decellularized versus cellularized matrices, as well as potential changes as a result of the decellularization process, must be taken into account.

Lastly, we demonstrated the utility of PSR staining to efficiently stain and quantify ECM deposition by WI-38 cells, with the signal intensity corresponding to the plating density. Importantly, this provides a relatively simple, inexpensive, and fast method for collagen visualization compared with immunostaining. Although previous reports of PSR-based well plate assays for collagen quantification exist,^26–28 our method is unique in using decellularized extracellular matrix, and does not depend on dissolving the collagen-bound dye. We demonstrated the successful use of this approach to screen antifibrotic drugs, Rapamycin and Galunisertib, for inhibition of collagen deposition. This expands on the many ways in which PSR has been successfully used for quantification of in vitro collagen content.

There are a number of considerations that must be taken into account for future work. First, this work directly compared the results of culture under macromolecular crowding conditions with culture under high plating density conditions. In line with the published versions of these methods, there were differences between the cell plating densities, treatment medium, and experiment time for these two methods in our experiments. In future work, multiple other variables should be examined in more detail. For example, the high cell plating density condition should be repeated over a span of 6–7 days to match with the timeline of the macromolecular crowding experiment. Additionally, the application of macromolecular crowding medium to the higher cell plating density can be explored. In this work, our goal was to directly compare the results of published in vitro collagen deposition methods, and thus, the experimental conditions were determined by those of published methods, rather than trying to vary parameters to make one method more similar to the other.

Other considerations that can be explored in the future include those of race and sex as biological variables in these results. Using primary human cells from diverse sources, including male and female donors, as well as donors from a variety of races would enable the establishment of conclusions that are more broadly relevant for the diverse population, while also providing insight for potential patient-specific effects.

Authors’ Contributions

Conceptualization: K.S., P.F., and V.S. Formal analysis: P.F. Funding acquisition: K.S. Investigation: P.F., V.S., A.L.A., and A.R.V. Methodology: P.F., V.S., A.L.A., and K.S. Resources: K.S. Supervision: K.S and P.F. Validation: P.F. and V.S. Visualization: P.F. Writing––original draft: P.F. Writing––review and editing: K.S. and V.S.

Disclosure Statement

No competing financial interests exist.

Funding Information

This research was supported by the Intramural Research Program of the NIH, National Institute of Biomedical Imaging and Bioengineering. The contents of this publication are the sole responsibility of the authors and do not necessarily reflect the views, opinions, or policies of the NIH and the Department of Health and Human Services (HHS). Mention of trade names, commercial products, or organizations does not imply endorsement by the U.S. Government.