Interleukin 1β and lipopolysaccharides induction dictate chondrocyte morphological properties and reduce cellular roughness and adhesion energy comparatively

Osteoarthritis (OA), a common degenerative joint disease that affects the knee among other joints, is a leading cause of disability with more than 32 million adults in the United States affected.¹ OA is marked by the progressive loss of articular cartilage (AC), AC’s degradation, joint inflammation, and subchondral bone remodeling; all that contribute to chronic pain, stiffness in the joint, and loss of functionality.^2,3 The imbalance between the catabolic and anabolic processes of the cells primes the secretion of proinflammatory cytokines, which further result in chronic inflammation in OA that leads to further degradation of the cartilage matrix.⁴ Proinflammatory cytokines such as interleukin-1-beta (IL1β) and bacterial lipopolysaccharides (LPS) play major roles in OA’s progression and in the tissue’s wound healing response.⁴ 

IL1β is one of the major inducers used to form OA models in vitro.^5–7 It is expressed in response to injury by the chondrocytes and cells of the inflamed synovium. IL1β induces the expression of many nuclear factor kappa B (NF-κB) pathway markers that are proinflammatory and damaging mediators of OA such as inducible nitric oxide synthase (iNOS) and cyclooxygenase (COX), which, in turn, affect the production of prostaglandins (PGE) such as PGE2.^8,9 NF-κB is majorly involved in the catabolism of the AC tissue through the upregulation of matrix metalloproteinases (MMPs), such as MMP3 and MMP13. Reactive oxide species (ROS), which encompass molecular species such as hydrogen peroxide and nitric oxide (NO), have also been found to be major regulators of the metabolism of chondrocytes and other cellular events important during normal and diseased states.^10,11

LPS, proinflammatory bacterial metabolites, also have a role in the dysfunction of AC’s metabolism and are often used to induce an OA in vitro, being a toll-like receptor 4 (TLR4) agonist.^12–16 The connection between LPS and OA has been examined through several studies that quantified LPS serum levels and the presence of AC’s degeneration as well as cell apoptosis.^12,14,17 LPS has also been found to induce the activation of NF-κB and mitogen-activated protein kinase (MAPK) pathways, which have been found to have crosstalk.^18,19

In addition to IL1β and LPS, tumor necrosis factor alpha (TNF-α) has been widely investigated as an inflammatory mediator of OA and its effects were widely compared to those of IL1β.^{8,11,14,15,20,21} To our knowledge, the effects of IL1β and LPS in the initial inflammatory responses of chondrocytes were not compared before. With that in mind, we investigated the effects of each inflammatory mediator (IL1β and LPS) separately on several aspects of chondrocytes’ morphology and nanomechanical properties using Atomic Force Microscopy (AFM). Specifically, the effects of IL1β and LPS on chondrocytes’ nanomechanical properties (elasticity and adhesion energy) and roughness were quantified and compared. To complement the AFM studies, fluorescence microscopy was used to further assess morphological changes in chondrocytes’ morphology upon induction. Finally, colorimetric and imaging techniques were used to assess extracellular matrix (ECM) formation.

Bovine articular chondrocytes (bAChs) were isolated from the metacarpophalangeal joint from cow knees acquired locally. AC tissues were minced and washed with a dissection-medium composed of phosphate buffered saline (PBS) with 1% antibiotic-antimycotic (anti-anti, Gibco). After that, samples were digested overnight in a digestion-medium [0.1% collagenase B (Roche) in Dulbecco’s Modified Eagle’s Medium (DMEM), 2% v/v fetal bovine serum (FBS), 2% v/v anti-anti] at 37 °C and 125-rpm. To isolate bAChs, the cells and the digestion solution were passed through a 40-μm filter. The filtrate was centrifuged, and the resulting pellet underwent two washes and resuspension in DMEM for cell counts. Cell-counting was performed using 0.4% trypan blue (TB) exclusion. bAChs were resuspended in a cryopreservation-medium [FBS with 10% v/v dimethyl sulfoxide (DMSO)] and frozen under controlled decreased thermal conditions of 1 °C/min and preserved until use.

For cell culture, primary bAChs were thawed at 37 °C and added to DMEM and washed to remove residual cryopreservation media. Viability was assessed using TB exclusion. bAChs were seeded at 19 K/cm² in Petri dishes (Tissue Culture Dish 40, TPP, Switzerland) for AFM testing or in well-plates for other tests. Cells were supplemented with the culture medium [DMEM/F-12 GlutaMAX (Gibco), 10% v/v FBS and 1% v/v ciprofloxacin] and maintained in a humidified 5% CO₂ incubator at 37 °C overnight. The next day, wells were assigned randomly to three different groups: (1) negative control (NC): reduced serum media [0.25% FBS, DMEM/F-12 GlutaMAX (Gibco) and 1% v/v ciprofloxacin], (2) 10 ng/ml IL1β in NC, and (3) 10 ng/ml LPS in NC.

A reduced serum medium was used instead of a full serum medium because the latter represents a complex natural product with a poor composition that may have a varied composition between batches even from the same manufacturer.²² This makes it very difficult to maintain consistent cellular growth-defined conditions across different treatments. The lack of consistent growth conditions may lead to results that are incomparable. Thus, we opted to use a reduced serum medium to cutout interferences from serum proteins and to provide reproducible growth conditions.

We have chosen to use a 10 ng/ml IL-1β in our study for the following two reasons. First, the use of 10 ng/ml of IL-1β is commonly reported in the literature in induction studies.^23–25 As such, using this concentration will enable us to compare our findings to these in the literature. Second, the expression of the nonchondrocytic genes that are reported in osteoarthritis was reproduced in vitro by treating healthy chondrocytes with 1 ng/ml IL1β.²⁶ Furthermore, the use of 1 ng/ml IL1β on normal human chondrocytes showed that the cells decreased the density of the interleukin 1 (IL-1) cell surface receptor (IL-1R) by 78% versus the control.²⁷ The density of IL-1R on healthy chondrocytes ranges between 3000 and 5000 sites per cell, and these sites are thought to modulate the IL1β role in AC’s homeostasis. As such, using a concentration that is higher than that would ensure that responses that can mimic to an extent osteoarthritic status will be obtained. When LPS is concerned, fewer literature studies have investigated its role as an inflammatory mediator nor compared its effects on chondrocytes to these of IL-1β. As such and to compare its inflammatory effects on chondrocytes to those of IL-1β, we choose to use the same concentration of 10 ng/ml during induction studies. Cells were induced for 24 h and maintained in a humidified 5% CO₂ incubator at 37 °C, after which further testing occurred.

For immunofluorescence, cells were cultured in 96-well plates and were fixed with 4% paraformaldehyde for 30 min at room temperature (RT); the step that was followed by washing with PBS twice. For actin’s visualization, the permeabilization buffer (0.5% Trixton-X-100 in PBS) was added for 30 s and quickly removed. PBS was added, and one drop of ActinGreen 488 (Invitrogen) was added and incubated overnight. For DNA’s visualization, one drop of 4′,6-diamidino-2-phenylindole (DAPI) (NucBlue, Invitrogen) was added to each well for 5 min.

Viability of the cells was measured after 24-h induction with the use of the PrestoBlue (PB) assay (Invitrogen). Briefly, 20 μl of PB was added to the media if each of desired wells in 96-well plates that to a final volume of 200 μl. The wells were incubated for 15 min at 37 °C. After incubation, fluorescence intensity was read at 560/590 nm excitation/emission. For live-dead imaging, Calcein-green, AM 488 (Invitrogen), was used to observe living cells and 5 μM propidium iodide (Alfa-Aesar) was used to observe dead cells.

The Griess reagent assay was used to quantify NO in the cell medium.²⁸ Briefly, a 50 μl culture medium sample was taken to a 96-well plate. While protecting the sample from light, sulfanilamide solution (1% sulfanilamide in 5% phosphoric acid) was added for 5 min incubation followed by addition of 0.1% N-1-napthylethylenediamine dihydrochloride (NED) in water and incubated for five additional minutes. Absorbance was read at 540 nm. Values were reported in terms of sodium nitrite concentration in μM.

Samples from each group were analyzed for PGE2 using a commercially available kit (Cayman Chemicals) according to manufacturer’s protocol. Plates were read at absorbance of 450 nm. A best-fit third-order polynomial standard curve was developed (R² ≥ 0.99) and used to calculate PGE2 concentrations for samples.

MMP3 was detected using a commercially available kit (MyBioSource, Inc.) according to manufacturer’s protocol. Briefly, a 100 μl sample was taken from the culture media supernatant after centrifugation. Plates were read at absorbance of 450 nm, and a standard curve was used to calculate MMP3 concentrations.

Cultures were analyzed for ECM formation, represented by DNA, glycosaminoglycans (GAGs), and collagen content. Samples were digested overnight at 65 °C in the papain extraction reagent [0.1 mg/ml papain (≥16 units/mg protein)] in 0.2M sodium-phosphate/ethylenediaminetetraacetic acid (EDTA) aqueous buffer suspension. DNA was quantified using the Quant-Fluor dsDNA System (Promega) following manufacturer’s instructions. Fluorescence was measured at 504 nm_ex/531 nm_em using a Cytation-5 Microplate-Reader (Biotek instruments, Vermont, USA). Total GAG was measured using the dimethylmethylene blue (DMMB) assay using the DMMB reagent (1.6 mg DMMB, 3.04 g glycine, 1.6 g in 95 ml 0.1M acetic acid completed to 1L).²⁹ Total collagen was measured using Picrosirius Red Solution [0.1% Sirius Red (direct red) in saturated picric acid) using a modified protocol.³⁰ Briefly, for both protocols, a 100 μl sample of digest was taken and 1 ml of DMMB reagent or 1 ml of picrosirius red solution was added to samples and shaken for 30 min to form a protein-dye conjugate. Samples were then centrifuged at 12 000 rpm for 10 min, and the supernatant was decanted. For GAG’s quantification, 1 ml of decomplexation solution (4 M guanidine-HCl and 10% propan-1-ol in 50 mM sodium acetate) was added to resulting pellet and vortexed, from this a 200 μl sample was read. For collagen’s quantification, the resulting pellet was first washed with an acid-wash (5% acetic acid in water) and centrifuged again. Once the supernatant was decanted, 1 ml of the dye release reagent (0.1M NaOH) was added to the resulting pellet and vortexed, from this a 200 μl sample was read. GAG’s and collagen’s absorbances were measured at 656 and 555 nm, respectively, using a microplate-reader. Reported values were normalized by DNA in μg.

AFM imaging and indentation on cells were performed after 24 h induction period using the NanoWizard IV AFM (JPK Instruments, Bruker, Billerica, MA) coupled with the CellHesion module. This module allows for large vertical piezo-element displacements (≤100 μm). The AFM head used is mounted on a Zeiss Axio observer 3 microscope (Carl Zeiss, Goettingen, Germany). All measurements were taken in PBS at 37 °C using the temperature-controlled Petri-dish heater. Nonconductive silicon nitride tips of 18 kHz resonance frequency and a 0.06 N/m nominal spring constant (DNP-10, probe D, Bruker, Billerica, MA) were used. Cantilever spring constants were calibrated individually by the thermal oscillation method.³¹ QI™ Imaging mode was used to acquire an AFM height image (128 × 128 pixels) of each individual cell using a speed of 300 μm/s and a set point of 2 nN to avoid damage to the cell surface during scanning. After acquisition of a topographic image of the cell, the force mapping mode was used to apply a minimum of an 8 × 8 grid, going up to an 11 × 11 grid depending on the cell area, to collect force displacement curves over the whole cell body with a maximum applied load of 3 nN, a Z-length of 15 μm, and a speed of 5 μm/s. At least three independent samples were taken for each time point tested, and n = 5 cells were randomly chosen from each sample for AFM analyses.

For analysis purposes, AFM force maps were used for quantification of nanomechanical properties of cells (adhesion energy and elasticity). Only force-indentation or force-distance curves that were on the top of a cell surface were used. The adhesion energy was quantified from each retraction curve as the area under the curve using the available function in JPK data processing software. To estimate Young’s modulus (E) of elasticity of cells using JPK data processing software, areas from indentation curves limited to 150 pN were used for consistency. For that analysis, the Hertz model described by Eq. (1) for a pyramid indenter was used to fit the experimentally measured force (F)- indentation (δ) profiles where E is the only fitting parameter.³² In Eq. (1), α, the face angle of the pyramid indenter tip, was set to 18° as provided by the manufacturer, and the Poisson ratio (ν) was set to 0.5 as commonly used for cells. E is the only fitting parameter in the following equation:

Cell dimensions (lengths, widths, heights, and surface areas) were quantified using Gwyddion software (http://gwyddion.net/). The surface roughness values of the cells were analyzed by measuring square (sq) root mean square (RMS) roughness from AFM height images. For each time point, 10 cells were selected randomly. Mean RMS values measured from five different locations (6 × 6 μm²) on each cell surface were reported.

All experiments were performed in independent triplicates. One-way or two-way analysis of variance (ANOVA) was used to determine statistical significance (p ≤ 0.05) using GraphPad Prism 8 (San Diego, CA). Data are presented as mean ± standard error of the mean (SEM). Box-plots and histograms were obtained using Sigmaplot software (Systat Inc, San Jose, CA).

Representative AFM height images that feature the morphologies, shapes, and ultrananoscopic details of chondrocytes’ surfaces are shown in Figs. 1(a)–1(c). In the NC group, the chondrocytes revealed a spherical morphology and a lot of granular texture and roughness on their surfaces [representative image in Fig. 1(a)] compared to when cells were treated [Figs. 1(b) and 1(c)]. A clustered orientation of filaments was also observed on the surfaces of the cells in the NC group (Fig. 1(a)]. In the IL1β treated group, chondrocytes were still spherical in morphology with less apparent roughness on their surface compared to the NC group [representative in Fig. 1(b)]. In the IL1β-treated group, the beginnings of lamellipodium extension can be seen on the edges of the cell as it started to spread on the substrate [Fig. 1(b)]. In the LPS group, a more evident elongation in morphology of the chondrocytes can be observed. In this group, there was an apparent reduced roughness and texture on the cell surface [Fig. 1(c)].

The representative actin images shown in Figs. 1(d)–1(f) qualitatively reveal the effects of induced inflammation on the chondrocytes’ cytoskeleton. In the IL1β- and LPS-treated groups, stress fibers were present as can be seen by white arrows highlighting thicker fibers along the long axis of the cell. Lamellipodium extrusions and many more focal adhesion (FA) points as seen by orange arrows compared to the NC group were also present. FA points are formed by a cluster of integrins that connect the ECM to the actin cytoskeleton.³³ While we are qualitatively observing general locations where FA would be found, staining for the proteins’ vinculin and paxillin can confirm these locations. The spreading seen in AFM and actin images correspond to quantified dimensions shown in Fig. 2. Chondrocytes adapted to induced inflammation by elongating compared to the NC group as seen by the significant increases in lengths and insignificant slight increases in widths of cells in these groups. This adaptation was consistent with the insignificantly increased surface areas of the cells in both induced groups compared to NC. The heights of cells were not apparently affected by induced inflammation as seen in both Figs. 1(a)–1(c) and 2(a). A summary of the data is included in Table I.

Cell shape is not only vital to maintain the phenotype of the cell, but it also plays a major role in regulating cellular growth, apoptosis, and motility.³⁴ Changes in cell shape in the form of cell spreading occur with lamellipodium extensions on the surface first and then through the leading-edge the cell will extend protrusions of filopodia, finger-like extensions, which sense the environment. Stress fibers, in particular, are long, aligned, actomyosin filament bundles that connect FAs at the cell base, connect the cytoskeleton to the ECM, preserve the cell shape, and aid in resisting stresses.³⁴ In response to environmental stimuli such as the presence or the absence of growth factors and shear stresses, cells often respond through morphological changes, growth, or death.^35,36 Chondrocytes are no exception. As can be seen from Figs. 1(b) and 1(e), cells spread during culture with more elongated filopodia produced in response to IL1β induction than to NC or LPS induction. These changes in cellular shape were in part how chondrocytes responded to the inflammatory stress induced by IL1β and LPS treatments.

Membrane roughness plays an important role in depicting cellular health and can be used as a means of detecting a diseased or stressed state. AFM height images were used to assess the square (Sq) root mean square (RMS) roughness values of chondrocyte cellular surfaces from 10 cells selected at random for each group. As can be qualitatively seen from the AFM representative images [Figs. 1(a)–1(c)], apparent roughness and texture were highest on the NC group and were less in both IL1β and LPS groups. These apparent morphologies agreed with quantified RMS values (Fig. 3). NC cells had significantly higher RMS values than both induced groups by an average 1.3-fold. Interestingly, cells treated with IL1β had roughness values that are lower by 1.37-fold compared to NC. A study utilizing AFM to detect cellular apoptosis found that increased oxidative stress led to a decrease in membrane roughness.³⁷ Our results highlight that cells respond to induced inflammation by decreasing membrane roughness. This is also in agreement with a decrease in surface roughness of OA AC’s tissue compared to healthy AC.³⁸ 

Adhesion of cellular membrane molecules to constituents in their environment has a vital role in cell physiological functions as well as in diseased states and has been used to differentiate healthy cells from cancerous cells.^39,40 In this study, we measured the nonspecific adhesion energy between the AFM tip and the cellular membrane’s existing surface molecules using AFM. A significant decrease in adhesion energy in induced inflammation groups compared to the NC group was observed [Fig. 4(a)]. LPS-induced cells significantly had the least adhesion energy compared to NC by 1.52-fold and IL1β by 1.37-fold. Histograms of adhesion energies quantified for all groups were skewed to the left with one main population dominant in all groups [Figs. 4(b)–4(d)]. The least variation was observed in the LPS group compared to the distributions of energies quantified for NC and IL1β groups. An increase in adhesion energy has been found to correlate with higher deformability of cells.⁴⁰ Our results show that under induced inflammation, chondrocytes reduced the adhesion energy of their membranes in order to withstand the stress and not to deform as was evident by cytoskeleton rearrangements they made and the stress fibers that presented on their surfaces in both IL1β and LPS groups. In our study, the decrease in roughness corresponded to the decrease in adhesion energy. Reduced roughness implies the presence of a smaller number of molecules on surfaces of cells to interact with the AFM tip. In another study, roughness has been speculated to be involved in adhesion.⁴¹ IL1β seems to affect the cellular roughness while LPS induction had the most effect on adhesion energy reductions; suggesting that chondrocytes vary the method and magnitude of their retorts in response to two induction methods.

Chondrocytes are mechanosensitive cells that respond to stimuli through mechanotransduction and cytoskeleton arrangements that in part affect their elasticity.^42,43 Young’s moduli were used to differentiate between healthy and diseased cells.^44,45 Furthermore, induced inflammation in chondrocytes has been shown to affect the nanomechanical properties by a reduction in elasticity.^20,39 Interestingly, in our study, the mean E values were similar between all three groups. However, even though the mean values were very close among the three groups [Fig. 5(a)], through a mean rank difference test, the LPS-induced group was found to be significantly different from NC and IL1β groups. Inducing inflammation by LPS slightly lowered E significantly compared to NC by 0.3% while IL1β reduced E insignificantly compared to NC by 0.4%. We believe that a larger reduction in elasticity would have been prominent later on during culture. This would have been probable as well as if a full serum medium was used in experimentations as a NC as the reduced serum medium used can act as a stressor on cells.²² However, a reduced serum is commonly used in such studies to allow for the full effect of the inflammatory stimulant to be tested as well as to avoid interactions with serum proteins.^46–48 Remarkably, even with the slight variation of mean E values among groups, a strong significant correlation was found with mean roughness values (r = 1, p = 0.0005) [Figs. 3(a) and 5(a)]. The histograms of the moduli for the three groups showed a non-normal distribution with one dominant population in the data and were skewed to the left [Figs. 5(b)–5(d)].

It is important to evaluate the viability of the cells under induced stresses. Assessment in our study revealed ratios of live to dead cells of 77% to 23% for NC, 74% to 26% for IL1β, and 84% to 16% for LPS, respectively [Figs. 6(a)–6(c)]. This result was also confirmed with the presto blue viability assay where induced groups had similar viability as NC, though nonsignificant, LPS-treated cells have the highest quantified viability [Fig. 6(d)].

Pro-inflammatory expressions of PGE2 and NO are seen to be major contributors to OA.^49,50 To evaluate the effects of our induced inflammatory mediators on cells’ production of key inflammatory markers, we measured NO, ROS, PGE2, and MMP3 contents. Nitric oxide (NO) is an inflammatory mediator formed from inducible NO synthase enzymes, which gets stimulated by inflammatory cytokines.^51,52 Our results indicated that the NO content was slightly insignificantly higher in induced groups compared to NC by an average 1.04-fold [Fig. 7(a)]. IL1β group had the slightly highest insignificant NO content by 1.06-fold and 1.03-fold compared to NC and LPS, respectively [Fig. 7(a)]. LPS induction resulted in slightly insignificant higher NO content compared to NC by 1.02-fold. While the values in the induced groups were not considerably higher than those estimated for the NC group, we expect that a longer culture time would have stimulated and shown higher NO content especially as we saw higher PGE2’s expression with induction [Fig. 7(b)]. PGE2’s levels were significantly higher in IL1β-treated cells by 5.7-fold compared to NC and 9.4-fold compared to LPS-treated cells. Interestingly, PGE2’s levels in the NC group were higher than those quantified for the LPS group by 1.65-fold [Fig. 7(b)]. When ROS was concerned, the induced inflammation by LPS resulted in a similar content to that of NC [Fig. 7(c), Table I]. In comparison, the ROS content for the IL1β group was less than that quantified for the NC group by 1.07-fold. Interestingly, it has been found that chondrocytes upon in vitro monolayer culture expressed increasingly higher ROS from day 0 to day 14 under stress-free conditions.⁵³ As such and based on our data, we can assume that the chondrocytes’ responses to induction in the form of expressions of inflammatory markers would show up later in culture. MMP3 is an ECM degrading enzyme that reduces the production of proteoglycans, laminin, and collagens (types 3, 4, and 9).⁵⁴ MMP3 was significantly highest in the NC group by 1.76-fold compared to IL1β and was not detected in the LPS group [Fig. 7(d)]. While the homeostasis cycle of maintaining the tissue consists of the presence of both catabolic proteases such as MMPs and anabolic processes,^4,55 the evident MMP3 in the NC group compared to induced inflammation is interesting. Autophagy can be used as a protective mechanism by chondrocytes and has been shown to have an increase in ROS and MMP activity.⁵⁶ Our data suggest the possibility of LPS having a chondroprotective role in reducing MMP activity. Furthermore, MMPs such as MMP13 are typically upregulated in later stages of osteoarthritis.^55,57 This can possibly explain why we did not see an increase in MMP 3 in our study as we only induced cells for 24 h [Fig. 7(d)].

ECM formation was affected by induced inflammation, specifically the GAG’s content. Both induction groups had significant lower GAG than the NC, yet LPS-induction had significantly lower GAG’s content compared to NC by 1.59-fold [Fig. 8(a)]. While one study has found that silencing MMP3 can lead to an increase in GAG,⁵⁴ this was not the case in our study. In fact, no MMP3 was detected in the LPS treatment. Another study found that LPS induction of tissue explants led to an increase in GAG’s release to the media, compared to the noninduced groups, suggesting an increase in aggrecanase activity.¹⁵ An increase in the aggrecanase activity leads to the digestion of GAG that implies a reduction in the actual GAG’s content of the tissue as a function of induction, which supports our findings. It was also found that the activity of aggrecanase decreased upon induction with tumor necrosis factor alpha but not with IL1β induction.²¹ Interestingly, the reduction in GAG’s content correlated with the reduction seen in adhesion energy (r = 0.9, p = 0.28) (Table I). Aggrecan has been found to undergo nonspecific self-adhesion and to play a major role in maintaining ECM architecture.^58,59 As such, the reduction observed in the GAG’s content can in part explains the decrease in adhesion energy in the LPS-treated group. Similarly, aggrecan-collagen adhesion has been suggested to have a role in controlling ECM’s assembly as well as in governing cartilage deformability.⁵⁸ MMP3 has also been found to have multiple binding sites on collagen fibers that suggest a role for the enzyme in tissue remodeling.⁶⁰ Though nonsignificant, the total collagen’s content was lowest in the IL1β group by 1.2-fold and 1.28-fold compared to NC and LPS, respectively [Fig. 8(b)]. Our IL1β group had the highest PGE2 content as described earlier, and the expression of PGE2 has been found to decrease the collagen synthesis.^61,62

In this study, the effects of two commonly used inflammatory induction mediators that have been shown to contribute to OA’s progression, IL1β and LPS, on phenotypic mechanical and morphological responses of bovine chondrocytes to short-term induction were quantified. While there are various studies on the use of IL1β and LPS to develop OA disease models in vitro to test the efficacy of certain therapeutics,^6,7,13,15,63 or on the elasticity of chondrocytes under induced stress,^20,39 to our knowledge, we are the first to present a comparison of the effects of IL1β and LPS on mechanistic changes to phenotypes of chondrocytes through an AFM study. We have shown that chondrocytes responded to induced inflammation via morphological changes and cytoskeleton rearrangements in an effort to protect themselves against the induced stress. Cell shape is highly regulated by actin filaments.³⁴ In the IL1β group, we have shown the presence of abundant stress fibers and filopodia in the cell’s attempt to sense the environment and migrate. In contrast, in the LPS group, chondrocytes showed a more elongated morphology confirmed by lamellipodium extensions as well as the presence of stress fibers. Our results indicated that chondrocytes respond with reductions in cellular roughness and adhesion energies. While the noteworthy reduction was seen in both groups (Table I), IL1β had the least roughness compared to NC while LPS induction had the least adhesion energy compared to NC. These results suggest that both induction approaches act on different surface molecules to affect roughness or adhesion energy. Though our data suggest that both IL1β and LPS modified the cytoskeleton that may be responsible for the decrease observed in surface topology to the cells, further molecular events that induce this change in surface roughness and adhesion need to be explored in detail. Although, we did not detect significant differences in cell elasticity upon induction between the two mediators, we detected significant decrease in elasticity between the induced group by LPS and NC. Our results are consistent with prior studies that have shown a quick temporal response in relation to moduli values.^64,65 ECM synthesis, in terms of total collagen and GAG, was reduced in different ways to induced inflammation. IL1β induction had the least collagen content, which corresponded to the highest PGE2 content in that treatment. PGE2 has been shown to inhibit the collagen synthesis.^61,66 LPS induction resulted in the least GAG content, which corresponded to decreased adhesion energy quantified for cells in that treatment group, confirming the role aggrecan plays in cellular adhesion.^59,67 In summary, we have detected the initial responses of chondrocytes to induced inflammation by two different inflammatory mediators IL1β and LPS. The mechanisms by which chondrocytes responded to induction varied in how cells altered their morphologies. Cellular induction resulted in a reduction in cellular surface roughness and adhesion energies. These transformations led to a decrease in ECM composition that supports a healthy AC tissue suggesting that induced inflammation quickly results in an inferior ECM to that of healthy cartilage. These findings support the use of AFM as a tool to detect nanostructural changes due to disease onset and progression and to provide mechanistic details of how cells control their phenotypes within their microenvironments to inform their functions in response to stress.

This work was supported by the National Science Foundation (NSF) GOALI Grant No. CBET-1606226 as well as startup funds for N. I. Abu-Lail from the University of Texas at San Antonio UTSA. Alia Mallah was partially supported by the National Institute of Health (NIH) No. GM008336.

The authors have no conflicts to disclose.

Ethics approval is not required.

Alia H. Mallah: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (equal); Resources (equal); Validation (equal); Visualization (equal); Writing – original draft (equal). Mahmoud Amr: Conceptualization (equal); Methodology (equal). Arda Gozen: Funding acquisition (equal); Project administration (equal). Juana Mendenhall: Funding acquisition (equal); Project administration (equal). Bernard J. Van-Wie: Funding acquisition (equal); Project administration (equal). Nehal I. Abu-Lail: Conceptualization (equal); Data curation (equal); Funding acquisition (equal); Methodology (equal); Project administration (equal); Resources (equal); Supervision (equal); Validation (equal); Writing – review & editing (equal).

The data that support the findings of this study are available from the corresponding author upon reasonable request.