Immune and repair responses in joint tissues and lymph nodes after knee arthroplasty surgery in mice

Abstract

The importance of a local tissue immune response in healing injured tissues such as skin and lung is well established. Little is known about whether sterile wounds elicit lymph node (LN) responses and inflammatory responses after injury of musculoskeletal tissues that are mechanically loaded during the repair response. We investigated LN and tissue immune responses in a tibial implant model of joint replacement surgery where wounded tissue is subjected to movement and mechanical loading postoperatively. Draining inguinal and iliac LNs expanded postoperatively, including increases in regulatory T cells and activation of a subset of T cells. Thus, tissue injury was actively sensed in secondary lymphoid organs, with the potential to activate adaptive immunity. Joint tissues exhibited three temporally distinct immune response components, including a novel interferon (IFN) response with activation of signal transducer and activator of transcription (STAT) and interferon regulatory factor (IRF) pathways. Fibrovascular tissue formation was not associated with a macrophage type 2 (M2) reparative immune response, but instead with delayed induction of interleukin-1 family (IL-1β, IL-33, IL-36), IL-17, and prostaglandin pathway genes concomitant with transforming growth factor (TGF)-β and growth factor signaling, fibroblast activation, and tissue formation. Tissue remodeling was associated with activity of the HOX antisense intergenic RNA (HOTAIR) pathway. These results provide insights into immune responses and regulation of tissue healing after knee arthroplasty that potentially can be used to develop therapeutic strategies to improve healing, prevent arthrofibrosis, and improve surgical outcomes.

INTRODUCTION

The repair of mechanically injured tissues, including surgical and skin wounds and bone fractures, is characterized by an early inflammatory phase, followed by tissue formation and subsequent remodeling.^(1,2) The early “type 1” inflammatory response typically lasts for 3 to 4 days and is characterized by influx of neutrophils and monocytes, and expression of canonical proinflammatory cytokines including tumor necrosis factor (TNF) and interleukin 1 (IL-1). Type 1 inflammation serves to sterilize and “clean up” the wound site and initiate the subsequent and partially overlapping phase of tissue formation. The tissue formation phase that occurs 3 to 10 days postinjury is characterized by resolution of type 1 inflammation and concomitant activation and proliferation of mesenchymal cells and fibroblasts, myofibroblast induction, and angiogenesis. Together, these processes result in the formation of fibrovascular granulation tissue and deposition of provisional extracellular matrix, which is subsequently remodeled by differentiating mesenchymal cells into a more organized and dense tissue. Tissue formation and remodeling are promoted in many tissues by the emergence of a type 2 immune response mediated by “M2-like” macrophages and various T cell subsets.^(3-5) These cells produce anti-inflammatory (e.g., IL-10), angiogenic (vascular endothelial growth factor [VEGF], placental growth factor [PlGF]), and growth (insulin-like growth factor 1 [IGF1], platelet-derived growth factor [PDGF], Wnt family member 3A [WNT3A], fibroblast growth factor [FGF]) factors, and activators of myofibroblasts and extracellular matrix (ECM) production (TGF-β). The type 2 response is typically promoted by efferocytosis of apoptotic cells and the cytokines IL-4, IL-13, and TGF-β.

The balance between the magnitude of type 1 versus type 2 immune reactions, and the kinetics of the transition from type 1 to type 2 inflammation, are important for effective wound healing and return of tissue integrity.^(1,2,4) Excessive or sustained type I inflammation, as occurs in type 2 diabetes, autoimmune/inflammatory disorders such as ulcerative colitis or rheumatoid arthritis (RA), or with lack of adequate wound site stabilization and thus excessive motion after surgical repair, results in delayed healing or chronic wounds.^(6-9) On the other hand, an excessive type 2 response, as occurs with elevated and prolonged IL-4, IL-13, or TGF-β expression, can result in fibrosis and scarring.⁽⁴⁾ The factors that drive the transition and sustain the tissue formation program are not well understood. In surgeries with insertion of implants, the host’s response to the foreign body can affect the nature of the inflammatory response and thus the tissue healing process and surgical outcomes.^(10,11) One important complication of total knee arthroplasty (TKA) surgery is arthrofibrosis, which is associated with pain, stiffness, and decreased range of motion, and accounts for 10% of revision surgeries performed within 5 years following primary TKA.^(12-14) At the cellular level, arthrofibrosis is characterized by increased numbers of myofibroblasts associated with excessive accumulation of ECM components in the joint capsule and pathological scar formation. The etiology and pathogenesis of arthrofibrosis are not well understood.

Investigation of the immune/inflammatory response to sterile injuries and surgical implants has focused predominantly on the affected tissue and local innate immune response.⁽¹⁵⁾ A key component of immune responses to infection is activation of adaptive immunity, mediated by T and B cells, that occurs in lymph nodes (LNs) which drain the site of infection. Dendritic cells (DCs) migrate from infected tissues to draining LNs via afferent lymphatics where they present antigens to immune cells.⁽¹⁶⁾ A productive immune response results in massive LN expansion secondary to recruitment of immune cells from the blood and their local activation and proliferation. Adaptive immune cells that have been activated in LNs return to the circulation via efferent lymphatics, and then can traffic to the site of infection to exert their effector functions. Recent work has revealed a role for adaptive immune cells in repair of damaged tissues.⁽¹⁷⁾ For example, T helper 2 (Th2) cells that produce IL-4 and regulatory T cells (Treg) that produce growth factors promote repair of damaged muscle.^(18,19) In contrast, T helper 17 (Th17) cells can promote fibrosis.⁽²⁰⁾ A role for adaptive immune cells in wound repair raises the question of whether sterile inflammation can activate draining LNs where these cells can be “educated” and then recirculate to help repair damaged tissues.

Little is known about the nature and biological importance of inflammatory/immune responses to TKA, which involves skin incision, transection of soft tissues, drill hole injury to bone, insertion of a titanium implant, and subsequent movement and mechanical loading of joint tissues as patients resume ambulation. We investigated the immune response to TKA in a mouse model with the goal of identifying immune pathways that could be therapeutically modulated to improve healing of soft tissues and implant integration with bone, and to prevent arthrofibrosis. We performed a time course analysis of joint tissues, peri-implant bone, and draining LNs using histology, flow cytometry, and transcriptomics in a well-established clinically relevant murine tibial implant model in which animals resume activity and weight bearing within a day after surgery.^(21,22) We found that sterile surgical injury elicited a dramatic LN reaction, with potential for mobilizing a pro-repair immune response. Contemporaneously, joint tissues exhibited an evolving immune response with IFN signature and profibrotic inflammatory components, the latter associated with a transient fibrotic tissue reaction. These results suggest therapeutic strategies to modulate the postoperative immune response to improve tissue healing and prevent development of arthrofibrosis.

SUBJECTS AND METHODS

Animals and surgical protocol

All animal procedures used in this study were approved by the Institutional Animal Care and Use Committee (IACUC) at Weill Cornell Medicine (WCM). Sixteen-week-old female C57BL/6 mice (Jackson Laboratory, Bar Harbor, ME, USA) were used in all experiments and housed in standard specific-pathogen-free (SPF) conditions at the WCM animal facility. All procedures are reported following the Animal Research: Reporting of In Vivo Experiments (ARRIVE) guidelines.⁽²³⁾ Animals underwent surgical implantation of a titanium implant in the right tibia as described.⁽²¹⁾ Briefly, surgeries were carried out with animals under general anesthesia (2% isoflurane, 2 L/min) administered via a nose cone. Using sterile technique, an incision was made in the right knee and the quadriceps fibers divided medially. The anterior cruciate ligament and menisci were resected, and the patella dislocated to expose the tibial plateau. The tibial articular cartilage and proximal epiphysis were removed with a fine-tip burr. Using a high-speed drill, a 0.9-mm-diameter hole was created in the medullary canal and a three-dimensional (3D) printed titanium implant inserted in a press-fit model into the hole. A full range of motion of the knee was confirmed before the closure of the wound. The extensor mechanism was closed using resorbable suture and skin closed with nonabsorbable suture. Mice receiving sham surgery underwent the same procedure, except for removal and drilling of bone and insertion of implant. The mice were given analgesia (buprenorphine, 0.05 mg/kg subcutaneously) for the first 72 h postoperatively. Mice resumed ad lib weight bearing and ambulation within a day after surgery without clear limitations relative to non-operated mice as described.⁽²¹⁾

Lymph node cell isolation and quantitation

Lymph nodes were collected and digested as described⁽²⁴⁾ from surgical (right) and uninjured contralateral (left) sides of operated mice. As additional controls in the time course experiment shown in Figures 1 and 2, and in Supplemental Figures S4, S6, S8A, S9, and S10, lymph nodes were also obtained from non-operated mice at the first postoperative time point (day 3), and from an independent cohort of mice age-matched to mice that were euthanized at day 35. Briefly, lymph nodes were dissected from mice, minced into fine fragments, and digested in Roswell Park Memorial Institute (RPMI) medium + 0.5% bovine serum albumin (BSA) + 4.84 mg/ml collagenase type II (Worthington Biochemical, Lakewood, NJ, USA) + 40 μg/ml deoxyribonuclease I (DNase I) (Sigma-Aldrich, St. Louis, MO, USA) for 30 min at 37°C while shaking at 50 rpm (0.15 x g). Cell suspensions were triturated 40 times using a Pasteur pipette, and ethylenediamine tetraacetic acid (EDTA) was added to 10mM. Cells were then incubated at 37°C for 5 min, and the suspension was passed through a 70-μm filter to remove debris and counted using a Cellometer(Nexcelom Bioscience, Lawrence, MA, USA) or Coulter Counter (Beckman Coulter Life Sciences, Brea, CA, USA). Images of lymph nodes were taken at the time of collection and uploaded into ImageJ software (NIH, Bethesda, MD, USA; https://imagej.nih.gov/ij/) where length and width were measured, and area calculated. In the time course experiment described above in this paragraph, area measurements in the non-operated control obtained from an independent cohort of mice showed variability and discrepancy from cell counts and were not included because of questions about reliability of measurement.

FIGURE 1.

Tibial implant surgery induces expansion of draining lymph nodes. (A) Schematic representation of drainage pattern of inguinal and iliac lymph nodes from mouse knee. (B) Representative photographs (top) of inguinal and iliac lymph nodes 1 week postsurgery. Scale bar = 1 mm. H&E-stained sections of iliac lymph nodes 1 week postsurgery (bottom). Scale bar = 250 μm. (C) Quantitation using ImageJ of cross-sectional area of inguinal (left panel) and iliac (right panel) lymph nodes at 5 time points following surgery: 3, 7, 14, 21, and 35 days (n = 5–6/time point). (D) Total cell counts of inguinal (left panel) and iliac (right panel) lymph nodes at the indicated time points (n = 6/time point). (E) Total cell counts of inguinal (top) and iliac (bottom) lymph nodes 1 week after skin incision only, sham surgery, or implant surgery (n = 5/group). Data are represented as median with interquartile range, p values obtained by two-way ANOVA followed by Tukey’s posttest. Abbreviations: ANOVA, analysis of variance; Ctrl, non-operated mice; Ctrl2, non-operated mice from an independent cohort of mice age-matched to day 35 mice; H&E, hematoxylin and eosin; L, left (contralateral lymph node in mice receiving implant); R, right (lymph node on same side as limb receiving implant).

FIGURE 2.

Surgery-induced expansion of immune cells in draining lymph nodes. (A) Representative FACS plot of CD3+ CD4+ double-positive T cells from draining (right side) iliac lymph node from mouse receiving tibial implant surgery (for full gating strategy, see Supplemental Figure S1). (B,C) Total cell counts of CD11c+ MHCII+ DCs in iliac lymph nodes 1 week after incision, sham, or tibial implant surgery (n = 5/group) (B) and at the indicated time points following surgery (n = 6/time point) (C). (D,E) Total cell counts of CD3+ and CD3+ CD4+ T cells in iliac lymph nodes 1 week after incision, sham, or tibial implant surgery (n = 5/group) (D) and at the indicated time points following surgery. (n = 6/time point) (E). (F) Total cell counts (left panel) and frequencies (right panel) of CD3+ CD69+ cells in iliac lymph nodes 1 week after incision, sham, or tibial implant surgery (n = 5/group). (G) Total cell counts of CD3+ and CD4+ proliferating (Ki67+) cells in iliac lymph nodes at the indicated 5 time points. (n = 6/time point). In C, E, and G, Ctrl represents non-operated mice, and Ctrl2 represents non-operated mice from an independent cohort of mice age-matched to day 35 mice. Data are represented as median with interquartile range, p values obtained by two-way ANOVA followed by Tukey’s posttest. Abbreviations: ANOVA, analysis of variance; DC, dendritic cell; FACS, fluorescence-activated cell sorting.

Multicolor flow cytometry

Fluorochrome-conjugated monoclonal antibodies (mAbs) were purchased from BioLegend (San Diego, CA, USA; for list of antibodies used, see Supplemental Table S3). Single-cell suspensions (1 × 10⁶ cells per sample) were prepared in fluorescence-activated cell sorting (FACS) buffer (phosphate buffered saline [PBS], 2% fetal bovine serum [FBS], 1% sodium azide, and 0.4% EDTA) and Fc receptors were blocked by incubation with purified rat anti-mouse CD16/CD32 (mouse Fc block; BD Biosciences, San Jose, CA, USA; #553141 used at 1:50 dilution) before proceeding to surface marker staining. Surface markers were detected by incubating cells with respective antibody cocktails for 15 min at room temperature in the dark; intracellular markers were detected using a Fixation/Permeabilization kit (Thermo Fisher Scientific, Waltham, MA, USA) with incubation at 4°C for 30 min in the dark. Samples were fixed in 4% paraformaldehyde for 5 min before acquisition on a FACS Canto II cytometer (BD Biosciences). Data were analyzed and compensated using FlowJo software (version 9.9.6; FlowJo, LLC, Ashland, OR, USA) and using single-stained antibody-capture bead controls (UltraComp eBeads Compensation Beads; Thermo Fisher Scientific).

Histology and immunohistochemistry analysis

Lymph nodes were dissected, placed in cryomolds, submerged in TissueTek OCT Compound (Sakura_®; FineTek, Walnut, CA, USA), and immediately placed on dry ice until fully frozen. Seven-micron (7-μm) sections were cut and stored at −80°C. Slides were brought to RT 1 h before performing hematoxylin and eosin (H&E) staining. Operated limbs were collected from mice, fixed in 10% formalin for 5 h while shaking, washed in distilled water, and decalcified in 0.5M EDTA. After decalcification was complete, titanium implants were removed laterally using forceps. Samples were then embedded in paraffin and 7-μm sections were cut and stained with H&E and Masson’s trichrome stains. Histology slides were reviewed by a pathologist at the Hospital for Special Surgery (HSS) with training in musculoskeletal pathology in a blinded manner. H&E sections were scored for histological signs of inflammation, fibroblastic activity, and peri-implant ossification. Masson’s trichrome stain was performed to better assess maturation of fibroblasts and connective tissue. In this particular staining method, collagen appears blue, cytoplasm is red-pink, and nuclei are dark purple. Based on this staining we classified fibroblasts as immature (reddish, plump, round shaped) and mature (bluish, collagen rich, elongated). Fibrocartilage amounts were also scored by using a combination of H&E and Masson’s trichrome findings. Specifications for each score are listed in Supplemental Table S4.

For visualization of nuclear factor κB (NF-κB) p65 and CD206 by immunohistochemistry, paraffin embedded 7-μm sections were deparaffinized in xylene and rehydrated in an ethanol series. After quenching of endogenous peroxidases (3% hydrogen peroxide in PBS for 30 min at room temperature), heat-induced antigen retrieval (10mM citrate buffer, pH 6.0, for 15 min) and blocking, sections were incubated overnight with Anti-p65 (1:100; Cell Signaling Technology, Danvers, MA, USA; Cat# 9167) and anti-CD206 (1:1000; Abcam, Cambridge, MA, USA; ab64693). The signals were developed by avidin-biotin peroxidase complexes with a diaminobenzidine (DAB) substrate solution (Vector Laboratories, Burlingame, CA, USA) and the staining intensity was blindly scored by a pathologist at HSS.

Histomorphometry

All stained slides were slide scanned with Leica CS2 (Leica Microsystems, Inc., Buffalo Grove, IL, USA) at magnification ×40 and imported into QuPath (https://qupath.github.io/).⁽²⁵⁾ A custom machine vision pipeline was created adapting a previously used architecture.⁽²⁶⁾ In short, regions of interest (ROIs) (either the test or training sets) were segmented into simple linear iterative clustering (SLIC) superpixels with high irregularity and small spacing to follow the contours of the image.⁽²⁷⁾ The minimum, mean, maximum, and standard deviation of intensity features (optical density sum, deconvolved hematoxylin, deconvolved eosin, hue, saturation, and brightness) and shape features (area, length, circularity, solidity, maximum diameter, and minimum diameter) were extracted from each superpixel. Smoothed features were also calculated for each superpixel averaging the features from neighboring superpixels within 5 μm and 25 μm. For the training set, four slides were selected irrespective of time point that were determined to have representative tissue types of primary fibrovascular granulation tissue, secondary granulation tissue, and chondrogenic tissue. Primary granulation tissue was characterized by diffuse tissue that was lightly stained with eosin with both spindle and round cells. Secondary granulation tissues were characterized by a stronger eosin stain, with many more spindle-shaped cells. Chondrogenic tissue was characterized by dark eosin and hematoxylin-stained tissue with round cells with potentially vacant nuclear lacuna, or cells dividing in a column. ROIs comprised of >10,000 superpixels of these tissue types were then annotated and used as training sets for a decision tree machine learning algorithm implemented by OpenCV (DTrees) (http://opencv.org). ROIs of pathologic soft tissue were manually drawn on the anterior and posterior compartments and the classifier was applied. Only one slide at day 35 did not have sufficient anterior pathologic tissue because of histologic artifacts and was excluded from the histomorphometry and statistical analysis. Total areas of the ROIs and the classifications were exported to generate ratios used for statistical analysis.

RNA isolation and real-time quantitative polymerase chain reaction

Cancellous bone was collected as described.⁽²⁸⁾ Briefly, tibias were rapidly dissected and soft tissues, fibulas, and the distal ends of tibias were removed. For operated tibias, implants were removed with forceps before proceeding. Bone marrow was removed by centrifuging in nested microcentrifuge tubes for 30 s at 13,000 rpm (15,871 x g). Non-operated tibias were cut at the proximal end until cancellous bone was fully exposed, and a 1-mm biopsy punch was used to isolate and collect the cancellous core from the cortical shell. Joint soft tissue was collected by dissecting out the full lower limb, removing muscle tissue, and collecting tissue surrounding knee joint. All samples were flash frozen in liquid nitrogen and stored at −80°C. Cancellous bone and joint soft tissue samples were homogenized in 1 ml TRIzol Reagent (Invitrogen, Carlsbad, CA, USA; 15596018) using 2.4-mm metal beads (VWR, Radnor, PA, USA; 10158-604) for 3 min at 30 Hz, then spun down and the aqueous layer removed for RNA isolation. Lymph nodes were homogenized in RLT Buffer using 5-mm stainless steel beads (Qiagen, Valencia, CA, USA) in a TissueLyser II (Qiagen) for 2 min at 30 Hz. Total RNA for all tissues (lymph node, cancellous bone, joint soft tissue) was isolated using the RNeasy Mini Kit (Qiagen) and reverse-transcribed using the RevertAid First Strand cDNA Synthesis Kit (Fermentas, Waltham, MA, USA) following the manufacturer’s instructions. Real-time quantitative PCR (RT-qPCR) was performed in duplicate using Fast SYBR Green Master Mix and a 7500 Fast Real-Time Cycler (Applied Biosystems, Foster City, CA, USA). Expression was normalized relative to levels of hypoxanthine phosphoribosyltransferase (HPRT). Gene-specific primer sequences used are listed in Supplemental Table S5.

RNA sequencing

Quality of all RNA and library preparations was evaluated with BioAnalyser 2100 (Agilent Technologies, Santa Clara, CA, USA). Sequencing libraries were sequenced by the Genomics Facility at Weill Cornell Medicine using a NovaSeq600 (Illumina, San Diego, CA, USA), with 100-base pair (bp) single-end reads to a depth of at least 20 million uniquely mapped reads per sample. Read quality was assessed and adapters trimmed using fastp⁽²⁹⁾ and mapped to the mouse genome (mm10) and reads in exons were counted against Gencode M25⁽³⁰⁾ with STAR Aligner.⁽³¹⁾ Differential gene expression analysis was performed in R (R Foundation for Statistical Computing, Vienna, Austria; https://www.r-project.org/)⁽³²⁾ using edgeR.^(33,34) Genes with low expression levels (<3 counts per million in at least one group) were filtered from all downstream analyses. Benjamini-Hochberg false discovery rate (FDR) procedure was used to correct for multiple testing.⁽³⁵⁾ Genes with an FDR-corrected p value <0.05 and log₂ fold change >1.5 were considered differentially expressed. Downstream analyses were performed in R using a visualization platform built with Shiny⁽³⁶⁾ developed by bioinformaticians at the David Z. Rosensweig Genomics Center at HSS (RNAseq DRaMA; https://hssgenomics.shinyapps.io/RNAseq_DRaMA). RNA sequencing (RNAseq) data is deposited in GEO (https://www.ncbi.nlm.nih.gov/geo/) with accession number GSE174294.

RNA sequencing analysis

Additional downstream analyses were performed using IPA (QIAGEN Inc., https://digitalinsights.qiagen.com/products-overview/discovery-insights-portfolio/analysis-and-visualization/qiagen-ipa/) and gene set enrichment analysis (GSEA; https://www.gsea-msigdb.org/gsea/index.jsp). Heat maps were generated using Morpheus (https://software.broadinstitute.org/morpheus). Gene ontology (GO) annotations were determined using the ClusterProfiler tool (https://bioconductor.org/packages/release/bioc/html/clusterProfiler.html) on each of the six clusters.⁽³⁷⁾ Upset and Sankey diagrams were generated in R using the UpSetR and networkD3 packages.^(38-40)

Statistical analysis

Graphs in the figures are presented as a median, interquartile range, and the number of mice (n) per group. Statistical testing was performed in GraphPad Prism 8 for Windows (GraphPad Software, Inc., La Jolla, CA, USA). All data were tested for normality using the Shapiro-Wilk normality test. Student’s t test and Mann-Whitney test were used for parametric and nonparametric data, respectively. Ordinary two-way analysis of variance (ANOVA) with Tukey’s post hoc test was used for multiple comparisons. Significance is indicated in the box plots.

RESULTS

Tibial implant surgery induces expansion of draining lymph nodes

Whether TKA primarily induces a localized inflammatory response or whether tissue damage is sufficient to induce an LN response is not known. We investigated the effects of tibial implant insertion on the size and cellularity of draining LNs in our model.⁽²¹⁾ The knee joint and surrounding tissues drain to the inguinal and para-iliac LNs (Figure 1A).⁽⁴¹⁾ Tibial implant surgery increased the size of draining inguinal and iliac LNs on the operated (right, R) side relative to control contralateral (non-operated, L) LNs (Figure 1B, representative images); a massive expansion of draining iliac LNs was readily apparent 1 week postoperatively. A time course experiment revealed a significant increase in LN size on the operated side, relative to the contralateral side or to LNs from control non-operated mice, as early as 3 days postoperatively that was sustained for at least 2 weeks (Figure 1C and Supplemental Figure S2A). In line with increased size, LN cellularity increased on the operated (R) side, peaking on day 3 in inguinal, and day 7 in iliac LNs (Figure 1D and Supplemental Figure S2B). Cellularity of LNs on the operated side diminished overtime and by day 35 R iliac LN cell counts were similar to cell counts of contralateral LNs and LNs from non-operated mice; R inguinal LN cellularity also decreased over time but remained modestly elevated at day 35 (Figure 1D). Thus, tibial implant surgery induces a substantial but transient reaction in inguinal and iliac LNs, which can be explained by trafficking and/or proliferation of immune cells.

We wished to distinguish the relative contributions of three components of tibial implant surgery, namely skin incision, the surgical procedure, and bone injury/implant, to the LN reactions. Thus, we compared the effects of skin incision only, sham surgery that replicates tibial implantation but omits drilling of bone canal and implant insertion as detailed in Subjects and Methods, and implantation surgery. All three procedures elicited comparable inguinal LN expansion on the operated (R) side (Figure 1E, top panel). In contrast, skin incision had minimal effect on draining iliac LNs, whereas sham surgery and tibial implantation elicited similar amounts of iliac LN expansion (Figure 1E, bottom panel). These results suggest that inguinal LNs drain the skin and joint compartments whereas the iliac LNs sense the deep tissue damage to joint soft tissues that occurs during or after surgery. An additional contribution from tibial bone injury and implantation was not detectable by LN size and cell counts, which may be explained at least in part by lack of lymphatic drainage from bone or the bone marrow cavity.⁽⁴²⁾

Surgery-induced expansion of immune cells in draining lymph nodes

We analyzed various immune cell populations in draining LNs after tibial implantation using flow cytometry; the gating strategy is shown in Supplemental Figure S1 and a representative two-color flow cytometry plot showing CD3+ CD4+ T cells is depicted in Figure 2A. Dendritic cells (DCs; defined herein as CD11c+ MHC II+) migrate from tissues to LNs to induce expansion and present antigens to activate T cells. In accord with LN size expansion, DCs were increased in the operated (R) iliac LNs 1 week after sham surgery and tibial implantation (Figure 2B). Time course experiments revealed that iliac and inguinal LN DCs peaked 1 week after implantation surgery and subsequently returned close to the baseline observed in non-operated control mice (Figure 2C and Supplemental Figures S3 and S4). Similarly, total T cells (CD3+) and CD3+ CD4+ T cells increased after surgery, peaking at day 7 in both iliac and inguinal LNs and returned to baseline after 35 days (Figure 2D,E and Supplemental Figures S5-S7). A survey of additional immune cell types in draining iliac LNs after surgery also revealed increases in CD3+ CD8+ T cells, γδ T cells, regulatory T cells (Treg, CD3+ CD4+ CD25+ Foxp3+), neutrophils (CD11b + Ly6G+), and macrophages (Supplemental Figures S3A, S4A, S5A, and S6A). The inguinal LNs followed similar patterns of immune cell expansion after implantation surgery (Supplemental Figures S3B, S4B, S5B, and S6B,C), although there were minimal differences between the incision, sham, and implant groups (Supplemental Figure S7B). Although the absolute numbers of the various cell types increased, their relative percentages did not substantially change (Supplemental Figure S8). This indicates a proportionate influx and/or expansion of cells, including cells such as Tregs that can suppress immune responses, in expanding LNs.

Activation of T cells represents one hallmark of initiation of adaptive immune responses in LNs, and thus we measured cell surface expression of the T cell activation marker CD69 and stained for proliferation marker Ki67. Although the total number of CD69+ T cells increased approximately threefold 1 week after surgery (Figure 2F), these cells represented only about 10% of total CD3+ T cells in both control and draining iliac LNs (Supplemental Figures S5A, S6A, and S7A). The number of CD3+ CD69+ T cells peaked 3 to 7 days after surgery (Supplemental Figure S9). In accord with potential T cell activation as assessed by CD69 staining, CD3+ T cells stained positive for proliferation marker Ki67 1 to 2 weeks after surgery (Figure 2G, left panel); increased Ki67+ and CD69+ cells were detected in both CD3+ CD4+ and CD3+ CD8+ compartments (Figure 2G, right panel, and Supplemental Figure S10). These data suggest the possibility that LN expansion is associated with activation of a subset of LN cells. However, screening of expression of 12 activation markers by RT-qPCR did not reveal significant differences between draining and control contralateral LNs (not shown).

To gain additional insight into the effects of surgery on LN gene expression, we performed RNA-sequencing on draining iliac lymph nodes 1 and 2 weeks after surgery. Controls included lymph nodes from the contralateral non-operated side, and from mice that did not undergo surgery (control mice). Comparison of iliac lymph nodes draining the surgical site to LNs from control mice revealed 224 and 351 differentially expressed genes (DEGs; log₂ fold change >1, p < 0.05), respectively, 1 and 2 weeks after surgery (Supplemental Figure S11A,B). Pathway analysis using the Quantitative Set Analysis of Gene Expression (QuSAGE; http://bioconductor.org/packages/release/bioc/html/qusage.html) platform revealed that nine of the 14 most significantly enriched pathways (p < 0.01) were related to cell proliferation, further supporting cell activation in draining LNs (Supplemental Table S1). This analysis also revealed enrichment of genes in pathways related to immune activation and cell trafficking, including IL-2-STAT5 signaling that is associated with Tregs and effector T cell proliferation⁽⁴³⁾ (Supplemental Figure S11C); given the limited number of DEGs, the significance of pathway enrichment was relatively modest. Inspection of expression of genes in the IL-2-STAT5 pathway (Supplemental Figure S11D) revealed increased expression of Foxp3 and Il2ra, consistent with increased Tregs detected by flow cytometry,⁽⁴⁴⁾ and increased expression of cell cycle genes Cdk6 and Ccna2, consistent with cell proliferation. Overall, the results suggest that despite massive iliac LN expansion in response to surgery, only a small subset of immune cells is activated at the 1-week and 2-week time points postsurgery, and dynamic changes in gene expression in this subset of cells are obscured by static gene expression in the much larger number of non-activated cells.

Postoperative inflammatory reaction in periarticular soft tissues

We investigated whether iliac LN expansion was associated with damage and inflammation in joint tissues and/or peri-implant cancellous bone. Histological analysis of the operated knee joint 1 week after surgery showed increased cellularity and inflammatory cell infiltrates in periarticular soft tissues (Figure 3A; the white space corresponds to the location of the implant, which was removed during tissue processing). In contrast, inflammatory infiltrates were not observed in cancellous bone surrounding the implant (Figure 3A and data not shown). For more detailed analysis, periarticular soft tissues were divided into three regions: the anterior capsule (outlined in yellow), posterior capsule (outlined in black), and the suprapatellar pouch where the surgical incision was made (outlined in green). The three compartments showed similar changes after surgery: fibroadipose tissue was infiltrated by numerous, active-appearing fibroblasts, monocytes, and lymphocytes, scattered polymorphonuclear cells, dense ECM, and numerous blood capillaries. The articular cavity, as well as the suprapatellar pouch, contained edematous material and inflammatory cells. Few tracts of synovium were still present after surgery and showed signs of reactive hyperplasia.

FIGURE 3.

Postoperative inflammatory reaction in periarticular soft tissues. (A) Representative H&E image of an operated knee joint 1 week postsurgery (white space corresponds to location of removed implant). Regions of interest are outlined: green, suprapatellar pouch; yellow, anterior capsule; and black, posterior capsule. (B,C) RT-qPCR analysis of Il1b, Il6, Il10 and Ifng mRNA normalized relative to Hprt in (B) joint tissues from operated (R) and non-operated, control (L) limbs (n = 6/group), and (C) cancellous bone (n = 3/group) surrounding and distal to the implant. Data are represented as median with interquartile range, p values obtained by paired Student’s t test. Abbreviations: H&E, hematoxylin and eosin; mRNA, messenger RNA; RT-qPCR, real-time quantitative polymerase chain reaction.

In line with the inflammatory infiltrates observed by histology, periarticular soft tissues from operated (R) limbs showed dramatically elevated expression of canonical inflammatory Il1b and Il6 genes, which are expressed by innate immune cells, 1 week after surgery (Figure 3B). Expression of Il10, which encodes the potent anti-inflammatory cytokine IL-10, was only modestly increased at this time point. Expression of the T cell cytokine Ifng was also elevated after surgery but was several orders of magnitude lower than Il1b or Il6 when normalized relative to housekeeping gene Hprt (Figure 3B). In contrast to soft tissue, there was minimal induction of inflammatory and immune gene expression in cancellous bone surrounding the implant (Figure 3), which is also in accord with the histology. These results demonstrate an inflammatory reaction in articular tissues and further support that this reaction, rather than bone injury per se, is driving expansion of draining LNs.

Time course of inflammation and tissue remodeling after tibial implantation surgery

We performed histological analysis of operated joints using H&E and Masson’s trichrome staining to delineate the time course of inflammation and its relationship to tissue remodeling (Figure 4). Analysis and scoring of H&E sections obtained 3, 7, 14, 21, and 35 days after tibial implantation surgery by a pathologist revealed that inflammation peaked 3 to 7 days postoperatively and slowly resolved over 5 weeks (Figure 4A,B). In line with these findings, immunohistochemistry analysis showed staining for NF-κB subunit p65 on day 3, which decreased by day 21, whereas staining for CD206, which is associated with a “M2-like” response increased over time (Supplemental Figure S12). In contrast to the inflammatory reaction, onset of a fibroblastic reaction and appearance of immature fibrovascular (granulation) tissue was delayed until day 7, with resolution of the active fibroblastic reaction, decreased vascularity, and maturation of connective tissue by days 21 and 35 after surgery.

FIGURE 4.

Time course of inflammation and tissue remodeling after tibial implantation surgery. (A) Representative H&E images of joint soft tissue from operated limbs at 5 time points following surgery: 3, 7, 14, 21, and 35 days (n = 3/time point). For all images, left is anterior and right is posterior; blue arrow corresponds to position of removed implant. (B) Histological scoring of inflammation, fibrovascular tissue formation, and peri-implant ossification progression (n = 3/time point). (C) Representative Masson’s trichrome staining of joint soft tissue from operated limbs at 5 time points following surgery: 3, 7, 14, 21, and 35 days (n = 3/time point). (D) Histological scoring of immature and mature fibroblasts (n = 3/time point), and (E) expression of genes related to inflammation (Tnf) and remodeling (Mmp13) at 5 time points following surgery (n = 4/time point). Data are represented as median with interquartile range, values of p obtained by one-way ANOVA followed by Bonferroni posttest. Abbreviation: ANOVA, analysis of variance.

Combined analysis and scoring of H&E and Masson’s trichrome (which stains collagen blue) sections showed a tissue formation and remodeling process that was almost resolved by 5 weeks after surgery (Figure 4A-D). An initial infiltration of inflammatory cells and immature fibroblasts (large, round-polygonal cells, with light nuclei and abundant cytoplasm) at day 3 was followed on day 7 by formation of fibrovascular granulation tissue around the joint. The main components of granulation tissue were maturing fibroblasts (as shown by trichrome stain score, Figure 4D), inflammatory cells, numerous blood microvessels, and abundant ECM. Thus, the onset of fibrovascular tissue formation was delayed relative to inflammation but day 7 showed simultaneous presence of strong inflammatory and tissue formation/fibrotic responses. Days 14 and 21 showed reduction of cellularity and vascularity with continuing maturation of reparative tissue, and an increase in extracellular collagen content. Scoring of the Masson’s trichrome staining (Figure 4D) indicates a decrease in immature fibroblasts over time, with increasing maturation and production of collagen. By day 35, there were narrow bundles of dense collagen, containing few, fully mature, inactive-looking fibroblasts and with occasional inflammatory cells. As highlighted by trichrome staining, there was extensive fibrocartilage formation around femoral and tibial epiphyses, which would serve to stabilize the joint. Overall, the joint space appeared drier and less swollen, consistent with resolution of an acute repair reaction with subsequent remodeling and maturation of fibrovascular tissue and ECM. To corroborate the histological analysis performed by a blinded pathologist, we used unbiased machine vision-based histomorphometry to quantitate immature and mature fibrovascular tissue as described in Subjects and Methods. This approach similarly found that immature fibrovascular tissue reflective of tissue formation peaked early, whereas more mature tissue increased progressively over time (Supplemental Figure S13).

In line with the above findings, the time course of expression of the inflammatory gene Tnf followed similar kinetics as did the inflammation score, peaking 3 to 7 days after surgery and returning to baseline at 5 weeks (Figure 4E). In contrast, induction of tissue remodeling-associated gene Mmp13 was delayed until 7 to 14 days, but expression similarly returned to baseline by 5 weeks. Overall, the data support a sequential induction of inflammatory and tissue remodeling processes after surgery, which are self-limiting and resolve after 5 weeks.

Transcriptomic analysis reveals molecular pathways active during postoperative tissue repair

To gain greater insight into molecular pathways underlying tissue response to surgery, we performed RNAseq analysis of periarticular soft tissues 3, 7, 14, 21, and 35 days after tibial implantation surgery (n = 4/time point for a total of 24 samples; controls were non-injured contralateral knee joints from mice euthanized on day 35). Principal component analysis revealed that samples separated according to postoperative time point (Figure 5A), with samples obtained at adjacent time points being most closely related. This was corroborated by hierarchical clustering which showed that day 3 and day 7 samples were closely related, as were day 21 and day 35, with three of the four day 14 samples occupying their own branch within the day 21/35 branch and one sample on the day 7 branch (Figure 5B). These analyses support a progressive process over time with the greatest differences in gene expression from control occurring at 3 to 7 days, with a transition at 14 days followed by a return close to the non-injured state. In line with this idea, the number of DEGs (log₂ fold change >1.5, FDR < 0.05) comparing each time point to non-injured control was highest at 3 and 7 days postoperatively (1653 and 2103 DEGs, respectively) (Figure 5C) and subsequently diminished. The 15 most significantly activated pathways on Ingenuity Pathway Analysis (IPA) of each time point relative to control are depicted in Figure 5D. Strikingly, the top seven pathways, and 12 out of the top 15 most significantly activated pathways at day 3 are related to immune responses; these pathways are predominantly related to a proinflammatory innate immune response; for example, TREM1 signaling, DC maturation, pattern recognition receptor signaling, and inflammasome pathway. The temporal pattern of select pathway enrichment is depicted in a heat map in Figure 5E and shows an early peak of inflammatory pathways with resolution by day 21. Activation of molecular pathways related to adaptive immunity or an “M2-like” or type 2 response were not observed.

FIGURE 5.

Transcriptomic analysis of tissue response to tibial implantation. (A) PCA of RNA-seq data using top 500 genes. (B) Dendrogram representation of the full RNA-seq dataset using Euclidean distance and average linkage clustering. (C) Volcano plots of differentially expressed genes comparing each time point to control; colored dots correspond to genes with significant (FDR < 0.05) and >1.5 log₂-fold expression changes. (D) Ingenuity pathway analyses showing most significant upregulated canonical pathways at each time point postsurgery. (E) Heat map showing significance of various inflammatory and ECM remodeling pathways evolving over time. (F) Heat map representation of expression of representative genes associated with the HOTAIR pathway, inflammation, growth factors, collagens, and MMPs at the specified time points. Data was Z-score–normalized. Analysis is based on 4 biological replicates per time point for a total of 24 samples. Abbreviations: ECM, extracellular matrix; FDR, false discovery rate; HOTAIR, HOX antisense intergenic RNA; MMP, matrix metalloproteinase; PCA, principal component analysis; RNA-seq, RNA sequencing.

In contrast to the kinetics of inflammatory pathway expression, expression of genes in tissue remodeling pathways such as the hepatic fibrosis and osteoarthritis pathways was activated with delayed kinetics, with most significant expression peaking at 7 or 14 days (Figure 5D,E). The hepatic fibrosis pathway is comprised of a combination of inflammatory genes (in particular genes related to IL-1 signaling), genes related to TGF-β and growth factors (e.g., Tgfb1, Tgfbr2, Acta 2 [encoding α-smooth muscle actin {α-SMA}], Serpine1, Snail1, Spp1, Pdgfc, Pdgfrb, Gli3, Smo, Wnt16, Fzd1, Fzd2), and genes involved in sensing and response to ECM (Itga2, Itga5, Itgb3, Rac2, Rac3, Rhoc, Rhod, Rhoh) (Supplemental Figure S14A). The gene composition of the osteoarthritis pathway is similar but has increased representation of genes encoding proteases that are important for tissue remodeling (e.g., Mmp3, Mmp9, Mmp12, Mmp13, Adamts4) (Supplemental Figure S14B). The GP6 (glycoprotein VI platelet) signaling pathway, which was the most significantly induced pathway at days 14 and 21 (Figure 5D,E), is comprised of collagen and ECM signaling genes (Supplemental Figure S14C). To obtain additional insights into regulators of postoperative articular tissue gene expression we performed IPA upstream regulator analyses (Supplemental Figure S15), which highlight a role for TGF-β, cell cycle, and NF-κB in regulating the tissue gene response.

Interestingly, significant activation of the HOTAIR regulatory pathway was detected on day 21 and persisted on day 35, where it was the most significantly activated pathway (Figure 5D,E). HOTAIR is a long noncoding RNA that controls expression of various genes involved in ECM synthesis and remodeling, including Mmp2, Mmp3, Mmp9, Mmp12, Mmp13, Mmp14, Spp1, Cd44, Col1a1, Col1a2, and Wnt16 by epigenetic mechanisms (Figure 5F, left panel).⁽⁴⁵⁾ These results identify HOTAIR as a candidate regulator of the later phases of articular tissue remodeling after tibial implant surgery. We additionally mined the RNAseq data set for expression of marker genes of key immune cell types. In accord with the histology and gene expression patterns (Figures 3-5), markers of innate immune myeloid lineage cells were readily detectable and peaked at days 3 to 14 (Supplemental Figure S16A). Analysis of macrophage polarization markers along the M1 (inflammatory) to M2 (promote tissue repair) axis showed a mixed M1/M2 pattern at early time points, with a second wave of late expression of Retnla, which is important for collagen crosslinks and maturation.⁽⁴⁶⁾ In contrast to myeloid cell marker genes, expression of marker genes of natural killer (NK) cells and adaptive immune cells (T and B cells) was below our stringency cutoffs, although inspection of raw counts suggests infiltration by small numbers of T and NK cells that is detectable by day 3 and peaks at days 7 to 14 (Supplemental Figure S16B). This low-level gene expression was concordant with histology showing minimal lymphocytic infiltration (not shown). The results highlight temporally related cascades of immune and tissue repair gene expression after surgery.

Analysis of the kinetics of expression of the genes within the pathways described in this section revealed an additional level of complexity. Genes in inflammatory pathways clearly segregated into subsets that were induced rapidly with a peak at 3 days or more slowly with peak expression at 7 days postoperatively (representative genes shown in Figure 5F, second panel). In contrast, expression of growth factor–related genes (Figure 5F, third panel) and collagen genes (Figure 5F, fourth panel, and Supplemental Figure S17A) peaked at 7 days and persisted at 14 days postoperatively, in line with the active fibrotic and tissue formation phase observed by histology. In contrast, matrix metalloproteinases (MMPs), which are involved in tissue remodeling, showed more prolonged kinetics of expression Figure 5F, fifth panel). Overall, the results reveal pathways activated as injured joint tissues progress through overlapping stages of inflammation, proliferation, and ECM synthesis and remodeling. The active tissue formation phase on days 7 and 14 is characterized by simultaneous activation of NF-κB, growth factor, and TGF-β pathways.

Temporal evolution of tissue immune response after surgical injury

The analysis in the immediately preceding section, which compared each time point to control, supported the importance of the temporal pattern of gene expression after tibial implant surgery, but focused on analysis of genes in defined pathways. Thus, we next performed an unbiased genomewide categorization of genes based on expression kinetics using k-means clustering of all differentially expressed genes in the dataset set (4807 DEGs). The data were best represented by six clusters that distinguished genes based upon kinetics of activation and resolution (Figure 6A). GO analysis of these clusters (Figure 6B) further supported early activation of a tissue immune and acute phase response on day 3 (clusters I and V) followed by tissue formation and ECM synthesis that peaks on day 7 and extends into day 14 (clusters III and IV) (Figure 6B and Supplemental Figure S17B). This analysis also revealed that the clusters of genes whose expression transiently decreased at days 3 to 14 (clusters II and VI) before returning to non-injured baseline are associated with oxidative phosphorylation and cellular metabolism (Figure 6B), and their return to baseline likely represents a return toward a homeostatic resting state. A return toward a homeostatic state was further supported by an additional analysis that partitioned genes based on both differential expression relative to control as well as relative to the previous time point (Supplemental Figure S17C,D).

FIGURE 6.

Kinetic analysis of tissue immune response after surgical injury. (A) K-means clustering (K = 6) of 4807 genes (one per row) induced more than two-fold with an FDR < 0.05 in the specified conditions; n values indicate total genes (among the 4807) in each cluster. Data was Z-score–normalized. (B) Dot plot representation of enriched GO terms for various biological processes associated with genes in each cluster using enrichGO function in cluster-Profiler. The x axis denotes the number of genes in each cluster; dot size represents gene ratio and color scale represents adjusted p value. (C) Expression of representative genes with a sharp peak of expression at day 3. (D) GSEA showing significant enrichment of ISGs at day 3 postsurgery. (E,F) Expression of representative genes associated with IFN and pro-fibrotic responses. Data were Z-score–normalized. Abbreviations: FDR, false discovery rate; GO, gene ontology; GSEA, gene set enrichment analysis; IFN, interferon; ISG, interferon-stimulated gene.

To gain greater insight into the relationship of the immune response with the active tissue formation phase on day 7 we performed a focused analysis of immune genes in clusters I, III, IV, and V (Supplemental Table S2). Cluster V genes, which are transiently expressed on day 3, were associated with an acute phase response (Figure 6B) and included potential regulators Il6ra and Stat3 (IL6-Stat3 signaling is a major driver of the acute phase response) and noncanonical NF-κB genes Nfkb2 (encoding p100) and Relb (Figure 6C and Supplemental Table S2). In contrast, cluster I genes, whose expression is induced on day 3 and at least partially sustained on day 7, showed a strong “IFN signature” (Figure 6D, Supplemental Figure S18, and Supplemental Table S2). Potential regulators of the IFN response include various components of the IFN-Jak-STAT-IRF signaling pathway such as Ifnar1, Ifnar2, Stat1, Stat2, Ikbke, Irf7, Irf8, and Irf9 (Figure 6E). Although cluster 1 contained a few chemokine genes (Supplemental Table S2), there was surprising lack of canonical inflammatory cytokines and canonical NF-κB target genes. Instead, these inflammatory genes and cytokines were contained in clusters III and IV and thus induced with delayed kinetics to coincide with peak induction of the fibrotic gene response (Figure 6F and Supplemental Table S2). Notably, these genes included known inducers of profibrotic reactions in musculoskeletal tissues including IL-1 family genes Il1b, Il1rl1 (encoding IL-33R), and Il1rl2 (encoding IL-36R),⁽⁴⁷⁾ prostaglandin pathway genes such as Ptgs1, Ptgs2, Ptges, Ptger4, Ptgis, Ereg, and Plaur,^(48-50) and Il17d.⁽²⁰⁾ These results show an evolving immune response after surgical injury, with an unexpected early IFN response and delayed induction of proinflammatory genes with pro-fibrotic functions.

DISCUSSION

In this study we investigated the tissue and LN immune responses to a surgical wound in a tibial implant model where wounded tissue is subjected to movement and mechanical loading postoperatively. One striking finding was induction of a massive expansion of draining iliac LNs, indicating that tissue injury was actively sensed in secondary lymphoid organs, with the potential to activate adaptive immunity. Our findings also revealed a temporally distinct acute phase, IFN-STAT-IRF, and inflammatory profibrotic components of the immune response after surgery, the latter of which overlapped with an intense fibrotic reaction with fibroblast activation and ECM deposition (Figure 7). This tissue formation phase was not associated with emergence of a type 2 or “M2” immune response, but instead with concomitant induction of IL-1, prostaglandin, and TGF-β pathway genes, and NF-κB and growth factor signaling. The subsequent tissue remodeling phase showed resolution of inflammatory gene expression and sustained expression of tissue-remodeling MMPs and other genes regulated by the HOTAIR pathway. These results provide insights into immune responses and regulation of tissue healing after TKA that potentially can be used to develop therapeutic strategies to improve healing, prevent arthrofibrosis, and improve surgical outcomes.

FIGURE 7.

Schematic representation of the activation and resolution of immune and tissue remodeling responses over time in our tibial implant model. Abbreviations: DC, dendritic cell; IFN, interferon; IL-1, interleukin 1; NF-κB, nuclear factor κB; PDGF, platelet-derived growth factor; PGE2, prostaglandin E2; TGFβ, transforming growth factor β.

Local activation of innate immune cells such as DCs and macrophages at sites of sterile tissue injury, which is induced by tissue and cell damage products (known as damage-associated molecular patterns [DAMPs]), has been extensively studied.^(15,51) However, relatively little is known about whether tissue injury and associated sterile inflammation are sensed by draining LNs; recently, changes in draining LNs suggestive of a higher percentage of T helper type 1 (Th1) cells were described at 4 weeks in the destabilization of the medial meniscus mouse model of osteoarthritis.⁽⁵²⁾ We found that tissue damage associated with tibial implant surgery, under sterile experimental conditions where infection does not occur,⁽²¹⁾ was sufficient to induce massive expansion of draining iliac LNs. Tissue damage in this model is related not only to the surgical procedure, but to mechanical loading of a surgically altered joint, which can contribute to tissue damage as animals resume ambulation. These results highlight the vigor of immune surveillance for potential microbial pathogens after surgery and deep tissue wounds. Interestingly, our results showed activation and proliferation of a small subset of LN T cells, including expansion of Tregs. This suggests the interesting possibility that in addition to surveillance of wounds for possible infection, adaptive immune cells with reparative functions such as Tregs or Th2s^(4,19) can be activated or “educated” in draining LNs in response to sterile wounds and potentially migrate to injured tissues to promote repair. Although we did not detect Th2 cells or type 2 responses in LNs or tissues, a potential role for Tregs in regulating the distinct components of joint inflammation, and the effects on tissue repair, are worth pursuing in future work.

Musculoskeletal tissues are subject to dynamic mechanical loading and remodel in response changes in load⁽⁵³⁾; injured musculoskeletal tissues sense the mechanical environment and need to rapidly gain biomechanical properties sufficient to withstand loads and stabilize structures such as joints. One advantage of our tibial implant model^(21,22) is that animals resume weight-bearing and ambulation within a day of surgery, and we were able to analyze wound healing under conditions of mechanical loading as occurs after TKA. The wound healing process in our surgical injury model differed from wound healing in non-weight-bearing tissues such as skin or lung where tissue repair has been extensively studied.⁽²⁾ Most notably, the immune response to joint injury showed three distinct components, including a novel IFN response and delayed induction of a subset of inflammatory genes to coincide with tissue formation. The delayed response included genes encoding proteins such as IL-1β, IL-33R, IL-36R, IL-17D, COX1, and COX2 that have well established profibrotic functions including in arthrofibrosis models.^(20,47,54) Very little is known about IFN responses in wound healing, and inflammatory genes such as TNF, IL-1, and prostaglandin pathway genes are typically induced during the first 1 to 4 days and their expression resolves with the emergence of a type 2 IL-4/13-mediated repair response. Instead, inflammatory gene expression was delayed, and a type 2 response was notably missing, in our model. It is possible that these differences are related to a need to rapidly increase the biomechanical properties of healing load-bearing tissues. Moreover, the importance of mechanical forces in modulating immune responses is being increasingly appreciated,⁽⁵⁵⁾ including in tissue repair models.^(56-60) It will be important in future work to assess the interaction between loading and inflammation in the tibial implant model, and the effects of joint stabilization on the inflammatory response.

Relatively little is known about IFNs in repair of injured musculoskeletal tissues. These cytokines have been implicated in bone remodeling with effects on both osteoclasts and osteoblasts,⁽⁶¹⁾ and a human study linked a low IFN response with implant loosening after knee replacement.⁽⁶²⁾ IFNs have also been linked to suppression of fibrosis in various models.^(63,64) Several mechanisms by which IFNs can suppress fibrosis have been described, including direct inhibition of collagen gene expression, inhibition of cell proliferation including fibroblasts, suppression of endothelial cells and blood vessel formation, and antagonism of TGF-β and IL-4/13-Stat6 signaling.^(61,65,66) Our laboratory and others have described antagonism between IFNs and IL-1 and prostaglandin pathways.^(67-69) In addition, a recent report has proposed that IFN-mediated induction of the histone demethylase Setdb2 limits the expression of inflammatory genes including Il1b in a skin wound healing model.⁽⁷⁰⁾ Conversely, IL-1 and prostaglandins (PGs) can suppress IFN responses.⁽⁷¹⁾ In the context of this literature, it is tempting to propose that the IFN response restrains and delays the IL-1–related and prostaglandin-related components of the immune response after surgery, and that the balance between these opposing IFN-STAT-IRF and IL-1-PG pathways modulates the nature of the inflammatory and linked tissue repair processes (Figure 7).

Although active fibrosis resolves and joint tissues remodel in our system, we believe that our findings have relevance for understanding the pathogenesis and developing therapeutic strategies for preventing arthrofibrosis, an important complication of TKA.^{(12,14,72-75)} The etiology and pathogenesis of arthrofibrosis after TKA are poorly understood but have been proposed to be linked with joint instability (and thus abnormal loading), inflammation, a persistent active fibrotic response, and insufficient remodeling into well-organized tissue with good biomechanical properties. We present a clinically relevant model that likely can be extended into a model of pathological arthrofibrosis with restricted joint mobility by using genetic or pharmacological interventions to manipulate pro-fibrotic pathways. Such a model could provide additional insights into mechanisms that drive and sustain active fibrosis and promote the switch from tissue formation to remodeling. Our results to date support further investigation of the therapeutic utility of modulating IFN, IL-1-NF-κB, and prostaglandin pathways, pharmacologically or by altering physical therapy and loading regimens, to limit the magnitude or duration of the active fibrotic response after surgery.

Another challenge is to understand what drives the transition from active fibrotic tissue formation to remodeling into well-organized functional connective tissue. In our system this transition begins approximately 14 to 21 days postoperatively and is associated with fibroblast maturation, changes in the pattern of collagen gene expression (Figure 5F), emergence of the GP6 and HOTAIR pathways and expression of Retnla. The GP6 (glycoprotein VI platelet) pathway is characterized by signaling by integrins that sense ECM and drive collagen production, whereas Retnla encodes Resistin-like alpha that promotes collagen maturation.⁽⁴⁶⁾ HOTAIR is a long noncoding RNA that epigenetically silences its target genes by recruiting the polycomb repressive complex 2 (PRC2) that deposits negative histone marks and the histone lysine-specific demethylase 1 (LSD1) that erases positive histone marks.⁽⁴⁵⁾ If a novel role for HOTAIR in driving the transition from fibrovascular tissue formation to remodeling can be corroborated, this would open up the possibility of repurposing already available PRC2 and LSD1 inhibitors to prevent arthrofibrosis.

Our study has several limitations. A limitation is that only female mice were used and in the future this model will be extended to male mice. Bulk RNA sequencing of total unfractionated LNs corroborated flow cytometric detection of an activated small subpopulation of T cells but did not provide insights into the functions of these cells because of limitations in detecting gene expression changes when the majority of cells were not activated. This can be addressed in future work by flow-sorting T cells based upon activation and proliferation markers and performing gene expression and functional analyses. The resolution of the active fibrotic reaction and effective tissue remodeling in our tibial implant model did not provide an opportunity to study the pathogenesis of arthrofibrosis. This can potentially be addressed by using mouse lines or mutants that are more susceptible to fibrosis and scarring (such as Balb/C or gain-of-function IL-4 transgenic mice) or by modulating the immune response described in Figures 5 and 6 to shift balance from type 1 to type 2 immune responses.^(3,4) In the RNAseq experiments we used a contralateral non-injured joint as a control, which may exhibit modest gene expression differences from joint tissues in non-operated mice. It will also be important to attempt to compare our findings in the murine tibial implant model to tissue and LN reactions after TKA in human patients. This approach is severely limited because of inaccessibility of LNs during currently used surgical approaches, and ethical issues with sampling joint tissues within the 1-week to 4-week time-frame after surgery that our work suggests corresponds to a “decision point” about resolution/remodeling versus sustained active fibrotic reaction leading to arthrofibrosis. It may be possible to address this issue by analyzing tissues from arthrofibrosis patients who come to early revision (approximately 2 years after surgery) for ongoing activity of the immune and fibrotic pathways we have described. This will require expansive sampling of tissues from various joint locations, because many arthrofibrosis patients’ samples exhibit dense fibrotic tissue that has not remodeled.^(72,73,75)

In summary, we have identified LN reactions and distinct components of the immune response related to distinct phases of joint tissue repair after tibial implantation in a mechanically loaded and clinically relevant mouse model. This work sets the stage for developing approaches to therapeutically modulate loading and inflammation to enhance healing and prevent arthrofibrosis.

DATA AVAILABILITY STATEMENT

All the data are available within the article or its supplemental materials. The RNA-seq datasets that were generated by the authors as part of this study have been deposited in the Gene Expression Omnibus database with the accession number GSE174294.