Collagen triple helix repeat containing 1 is a new promigratory marker of arthritic pannus

Biological samples were obtained under a protocol approved by the Institutional Research Ethics Committee (IREC) of Nazarbayev University, Astana, Kazakhstan. All subjects gave written informed consent. Patients coming to outpatient facility of the Republican Diagnostics Centre (RDC, Astana, Kazakhstan), who presented with clinically apparent synovial swelling were examined for RA symptoms. The final diagnostic outcome was based on persistent inflammatory arthritis, magnetic resonance imaging (MRI), radiographic analysis, disease duration, and number of tender and swollen joints, and resulted in the 28-joint-count disease activity score (DAS28). Clinical assessment was accompanied with peripheral blood analysis for complete blood counts with differential, C-reactive protein (CRP), erythrocyte sedimentation rate (ESR), rheumatoid factor (RF), and anti-citrullinated protein antibodies (ACPA). Venous blood was collected into heparinized tubes, and cells were removed by centrifugation at 1000 × g for 10 minutes. Plasma was stored at –80 °C.

Sandwich ELISA was used to quantify CTHRC1 in human plasma according to the manufacturer’s protocol (www.​mmcri.​org/​antibody, Maine Medical Center Research Institute, Scarborough, ME, USA). Briefly, 96-well plates (Maxisorp, Nunc) were coated overnight at 4 °C with capture antibody 13E09 at 1.8 μg/ml in carbonate-bicarbonate buffer pH 9.4. All subsequent procedures were performed at room temperature. The next day, wells were washed twice with PBS containing 0.1 % BSA and 0.1 % Tween 20 (buffer PBS-BT) and then blocked with PBS-BT for 1 hour. Human plasma or conditioned media were diluted at least 1:5 in PBS-BT and incubated with absorbed capture antibodies for 2 hours. Subsequently, the wells were washed and then incubated with biotinylated detection antibody Vli10G07 diluted 1:500 in PBS-BT for 1 hour. After washing, wells were treated for 1 hour with streptavidin conjugated with horseradish peroxidase (St-HRP) (Pierce High Sensitivity St-HRP, Thermo Fisher Scientific Inc., Waltham, MA, USA) diluted at 1:8,000 in PBS-BT. After the final wash, TMB (3,3′,5,5′-tetramethylbenzidine) chromogenic substrate (Amresco, Solon, OH, USA) was added, and the developed signal was measured at 450 nm using the Multiscan-FC plate reader (Thermo Fisher Scientific Inc., Waltham, MA, USA). Absorbance was converted to absolute concentration using rhCTHRC1 as a reference. ELISA was performed in triplicates.

Mice were housed in a specific pathogen-free environment in the Institute for Animal Studies at the Albert Einstein College of Medicine, Bronx, NY, USA. All animal experiments were approved by the Institutional Animal Care and Use Committee (IACUC) of the Albert Einstein College of Medicine. Inbred BALB/c female mice, 2–4 months old (Charles River, Wilmington, MA, USA) were injected with ArthritoMab antibody cocktail (MD Biosciences Inc., St.Paul, MN, USA) according to the manufacturer’s protocol. Briefly, mice were injected intraperitoneally (i.p.) with 2 mg/mouse of antibodies on day 0 followed by i.p. injection of 40 μg of lipopolysaccharide (LPS) on day 3. Arthritis in the (FVB/N × BALB/c) F1 hybrid mice was induced similarly, but using CIA-MAB-2C ArthritoMab Cocktail (MD Biosciences Inc.), which was developed for murine strains that are refractory to arthritis. Mice were observed once or twice a day for paw swelling and redness by two independent observers. The arthritis scoring system was based on the number of inflamed joints in each paw producing a total score of 0–60 per mouse, as described previously [9, 17].

Total RNA was isolated from mouse paws using the RNeasy Mini kit (Qiagen Inc., Valencia, CA, USA) according to the manufacturer’s instructions. Paw tissue was taken from digits to the ankle or to the distal radius/ulna. RNA integrity and quality was tested on an Agilent 2100 Bioanalyzer. The Superscript First-Strand Synthesis kit (Invitrogen Life Technologies, Grand Island, NY, USA) was used to generate cDNA that was assayed using an Applied Biosystems 7900HT Fast Real-Time PCR System. Expression was measured in triplicates and at least two repeats using the Fast SYBR Green system (Applied Biosystems by Life Technologies, Grand Island, NY, USA). Cthrc1-specific primers were forward 5′-GGGATGGATTCAAAGGGGAAA-3′ and reverse 5′-AGAACTCGCAGAGCACTGTT-3′. Gapdh-specific primers were forward 5′-ACCCAGAAGACTGTGGATGG-3′ and reverse 5′-ACACATTGGGGGTAGGAACA-3′. Relative RNA concentration was calculated using the delta-delta cycle threshold (Ct) method corrected for PCR efficiency, as described previously [9, 18].

Human RA-FLS derived from the inflamed synovial tissue of patients with RA (Cell Applications Inc., San Diego, CA, USA) were used at passages three to four. RA-FLS were maintained in Dulbecco’s modified Eagle’s medium (DMEM) with GlutaMAX-1, 4.5 g/l D-glucose, and 25 mM HEPES, antibiotic and antimycotic (Gibco by Life Technologies, Grand Island, NY, USA) supplemented with 10 % heat-inactivated FBS and growth supplements (Cell Applications Inc., San Diego, CA, USA). Mouse embryonic fibroblast cell lines C3H/10T1/2 and NIH 3T3, and human skin fibroblast line Hs68 (all lines from ATCC Inc., Manassas, VA, USA) were studied.

Mice were killed by CO₂ inhalation according to the approved IACUC protocol. For histopathological analysis, hind limbs from the digits to femur, including the knee joint, were collected. Tissues were fixed in 10 % neutral buffered formalin, decalcified using Immunocal reagent (Decal Chemical Corporation, Suffern, NY, USA) and embedded in paraffin. Tissues were sectioned at a thickness of 5 μm and stained with hematoxylin-eosin or alcian blue and nuclear fast red for histological confirmation of arthritis.

For immunohistochemical analysis, sections were deparaffinized and pretreated with 10 mM sodium citrate buffer at pH 6.0 heated to 96 °C for 20 minutes for antigen retrieval. Endogenous peroxidase activity was quenched by incubation in 0.3 % hydrogen peroxide in PBS for 5 minutes. Nonspecific binding was blocked for 1 hour at room temperature using 2 % BSA (Sigma-Aldrich, St. Louis, MO, USA) and 2 % normal goat serum (Vector laboratories Inc., Burlingame, CA, USA) in PBS and 0.05 % Tween-20 (PBS-T buffer). For detection of FLS-specific and macrophage-specific markers in arthritic pannus, we used rabbit anti-CTHRC1 antibodies (Vli-55, www.​mmcri.​org/​antibody, Scarborough, ME, USA), rabbit anti-IBA1 antibodies (#019-19741, Wako Chemicals USA Inc., Richmond, VA, USA), rabbit anti-CDH11 antibodies (WTID1, Invitrogen by Life Technologies, Grand Island, NY, USA), and rabbit antibodies to smooth muscle actin (α-SMA) (E184, Abcam, Cambridge, MA, USA). As a secondary antibody, donkey anti-rabbit antibodies conjugated to HRP (Abcam, Cambridge, MA, USA) were used. After three washes with PBST, slides were finally stained with ImmPact 3,3′-diaminobenzidine (DAB) and counterstained with hematoxylin (Vector laboratories Inc.) or alcian blue (Thermo Fisher Scientific Inc., Waltham, MA, USA). Sections were dehydrated and mounted with Permount (Thermo Fisher Scientific Inc.).

We used a full-length human Cthrc1 cDNA cloned in frame with turbo-GFP (clone RG204348, OriGene Technologies, Inc., Rockville, MD, USA) as a template to amplify a Cthrc1 sequence. High fidelity PCR using Herculase II fusion DNA polymerase (Agilent Technologies, Inc., Santa Clara, CA, USA) with forward 5′-ATTTGGTACC ATGCGACCCCAGGGCCCCGCCG-3′ and reverse 5′-AATAGCGGCCG C TA ATGATGATGATGATGATGTTTTGGTAGTTCTTCAAT-3′ primers/linkers to introduce sites for restriction endonucleases Kpn°I and Hind°III and 6xHis tag and stop codon (restriction sites are underlined; ATG and stop codons are boldfaced; 6xHis coding sequence is italicized and underlined) for cloning the amplicon into the pCMV6-K/N vector (OriGene Technologies, Inc.). The recombinant construct structure was confirmed with a complete sequencing (Thermo Fisher Scientific Inc., Fair Lawn, NJ, USA).

Transient transfection of the Chinese hamster ovary (CHO) cells with lipofectamine LFA2000 was performed according to the manufacturer’s protocol (Life Technologies, Grand Island, NY, USA). CHO cells were grown in Opti-CHO chemically-defined serum-free conditioned media Opti-CHO (Gibco Life Technologies, Grand Island, NY, USA) for 5 days. Conditioned media were collected and centrifuged at 1000 × g for 15 minutes to separate from floating cells.

Cells were lysed in Laemmli sample buffer (62.5 mM Tris-HCl, 2 % sodium dodecyl sulfate (SDS), 10 % glycerol, 0.05 % bromphenol blue, and 100 mM 2-mercaptoethanol). Whole cell/tissue lysates were separated on 4–15 % gradient SDS-polyacrylamide gels (Bio-Rad Laboratories Inc., Hercules, CA, USA) and transferred onto polyvinylidene fluoride membrane (Thermo Fisher Scientific Inc., Waltham, MA, USA). Membranes were blocked for 1 hour in 1 % (w/v) casein in 25 mM Tris pH 7.4, 150 mM NaCl, 0.1 % Tween 20 (TBS-T) (casein blocking buffer, Thermo Fisher Scientific Inc.). After blocking, the membranes were probed with rabbit antibodies to CTHRC1 (Vli-55, www.​mmcri.​org/​antibody, Scarborough, ME, USA), rabbit antibodies to CDH11 (Invitrogen Life Technologies, Grand Island, NY, USA), rabbit antibodies to α-SMA (Abcam, Cambridge, MA, USA), mouse antibodies to α-tubulin (Sigma-Aldrich, St. Louis, MO, USA) as protein load control. As secondary antibodies, HRP-conjugated donkey anti-rabbit, anti-mouse or anti-sheep antibodies (Abcam) were used. Enhanced chemiluminescent substrate (Thermo Fisher Scientific Inc.) and HyBlot-CL film (Denville Scientific Inc., Holliston, MA, USA) were used to detect a signal. Optical density was quantified by ImageJ analysis software [19].

Cell migration was assessed using transwell permeable supports with an 8-μm pore membrane (Corning Costar, Tewksbury, MA, USA). C3H/10 T1/2 murine embryonic fibroblasts or RA-FLS were seeded in the upper chamber in DMEM supplemented with 1 % BSA. The lower chamber was filled with DMEM containing 0 %, 1 %, 6 %, or 10 % FBS. rhCTHRC1 (Sino Biological Inc., Beijing, China) or rhCTHRC1 (OriGene Technologies Inc., Rockville, MD, USA) were added to the upper chamber at 500–2,000 ng/ml. After 18 hours at 37 °C in 5 % CO₂, cells were fixed with 0.3 % glutaraldehyde in PBS for 10 minutes at room temperature and then stained with toluidine blue. Upper surface of the membrane was cleaned to remove any non-migrated cells using a cotton swab. Membranes were peeled off, and mounted onto glass slides for microscopic examination. Transmigrated cells were counted at 200 × magnification in four to eight fields of view using bright field microscopy.

Murine NIH 3T3 fibroblasts were plated on collagen type I-coated μ-slides (IBIDI USA Inc., Madison, WI) and incubated for 4 hours at 37 °C in 5 % CO₂/95 % air in complete DMEM media supplemented with 10 % FBS to insure complete cell adhesion on a bridge. Subsequently, rhCTHRC1 protein (Sino Biological Inc., Beijing, China) was added at 1000 ng/ml, and imaging has been performed for 14 hours in a serum-containing medium at 37 °C using time-lapse phase contrast microscopy with a Cell Observer microscope (Carl Zeiss Microscopy GmbH, Göttingen, Germany). Recorded stacks of images were transferred to ImageJ [19] and further analyzed with the MTrackJ plugin and Chemotaxis Tool (IBIDI USA Inc., Madison, WI, USA). The trajectories of individual cells were evaluated for accumulated distance migrated (Ad) and for cell velocity. Directness was calculated as a ratio of Euclidean distance (Ed) to the accumulated distance (Ed/Ad) (Fig. 6). Additionally, elongated morphology cell shape was studied (horizontal polarization), which was defined as maximal cell dimension using linear approximation. In preliminary experiments, we found that pro-motile and polarization effects of the CTHRC1 treatment were pronounced in the time window from 4–12 hours, and thereafter all calculations were performed for this time period.

Data were analyzed using Student’s t test for unpaired samples, using SPSS statistical software (SPSS Inc., Chicago, IL, USA). The Mann–Whitney U test for unpaired samples was performed using GraphPad Prism6 software (GraphPad Software Inc., La Jolla, CA, USA). P-values less than 0.05 were considered significant. Quantitative measurement of brown-colored DAB product of IHC staining was performed using ImageJ software [19] by color channel separation and densitometry of microscopic images at × 200–400 magnification. On each slide, at least five positively stained areas were analyzed for integrated brown-colored density; at least five slides per mouse (n = 5 per group) were analyzed.

Earlier we reported low levels of Cthrc1 mRNA in the paws of naïve mice, but substantial upregulation of mRNA during inflammation [9]. To characterize which tissue express CTHRC1, we used IHC staining of synovial joints. We induced arthritis in BALB/c mice using injections of anti-collagen antibodies followed by LPS injections 3 days later. Using a standard protocol, we induced strong inflammation in the paws of inbred BALB/c mice and also in (FVB/N × BALB/c)F1 hybrids that are less susceptible to CAIA. Arthritis was observed in the form of swelling and redness of the fore and hind paws of mice, and the diagnosis was confirmed using histopathological analysis of hematoxylin-eosin-stained tissue sections (Fig. 1). A thin synovial membrane in naïve joints (Fig. 1a, c, green arrowheads) turned into a massively hypercellular pannus infiltrated with inflammatory cells (Fig. 1b, d, yellow arrowheads). Polymorphonuclear leukocytes were most abundant within the synovial cavity and at the side of the pannus that was facing the cavity (Fig. 1d, blue arrowheads), while pannus consisted of fibrous-like large cells in contact with the bone surface. Although the edema and redness was not visually observable in larger joints, like the knee joint, the histopathological examination showed the presence of hyperplastic synovium and leukocyte infiltration.

IHC staining of the naïve synovial joints of hind paws and knees was negative (Fig. 1e), while the amount of CTHRC1 was dramatically increased in arthritis (Fig. 1f). We studied mice with early arthritis that had been present for 2 days (5 days after mAb cocktail injection or 2 days post-LPS injection) and also in mice that had developed arthritis for 2 weeks. Joints of mice with early and 2-week inflammation were equally positive for the amount of CTHRC1 (Additional file 1). Protein was predominantly located in the hyperplastic pannus at the bone/cartilage surface (Fig. 1f, yellow arrowheads).

The inducible pattern of the gene expression was corroborated at the levels of RNA and protein. At the second day after LPS injection we observed acute arthritis in mice, and Cthrc1 mRNA was induced 11.9-fold (p < 0.0001), while the corresponding protein increased 10.9-fold (p < 0.0018) (Fig. 2a, b). Western immunoblotting of the total protein from naïve and arthritic paws confirmed that immune staining was specific for the CTHRC1 polypeptide of the predicted size (data not shown).

To corroborate the results based on murine arthritis, and to show evidence for a possible role of CTHRC1 in human RA, we collected clinical samples of blood plasma from healthy individuals and from patients with a diagnosis of RA. The average DAS28 in patients with RA was 3.61 ± 0.72 (average ± standard deviation), and the CRP concentration was 12.99 ± 14.91 mg/L (Fig. 2d, e). In healthy individuals, CRP was significantly lower (0.75 ± 0.86 mg/L) than in patients with RA (p < 0.0001, Mann–Whitney U test). Circulating CTHRC1 was measured in both healthy and RA cohorts (median 2.00 versus 8.50, respectively, p < 0.0044, Mann–Whitney U test) (Fig. 2c). Higher CTHRC1 in human plasma correlated with increased CRP and stronger DAS28. In both the murine arthritis model and in human arthritis, CTHRC1 protein was highly expressed compared to much lower expression in normal/basal conditions.

IHC staining revealed several sites of the protein location in arthritic synovium. CTHRC1 was expressed in cells located closer to bone, cartilage, and fibrocartilage surfaces (Fig. 3a–d). Intracellular protein expression was maximal in cells in close contact with the bone and gradually decreased towards the synovial cavity side of the pannus (Fig. 3b–d). A similar pattern was observed in the knee joints and in smaller joints of the hind paws. The intimal lining layer of the pannus that faced synovial cavities was frequently CTHRC1-positive (Fig. 3a, e, f). Some deposition of the protein was detected on the bone surface (Fig. 3c, d), in underlying synovial membrane connective tissue (Fig. 3e), and in the areas adjacent to the bone/cartilage surface.

Specificity of the IHC staining was confirmed using skeletal tissues of the Cthrc1-deficient mice generated earlier [20]. Tissue sections were stained using primary and secondary antibodies, and the complete IHC staining was negative (Fig. 3g). Similarly, negative staining was achieved when primary anti-CTHRC1 antibodies Vli-55 were substituted with normal donkey serum (Fig. 3h). IHC staining showed a similar pattern for synovial tissues of two different genetic backgrounds of the inbred BALB/c and (FVB/N × BALB/c)F1 hybrids.

CTHRC1-positive synoviocytes exhibited fibroblast-like morphology and were located at the sites that are characteristic for activated FLS (Figs. 1f, 3c–f). We further studied localization of cells expressing CTHRC1 and their co-localization with fibroblast, myofibroblast, and macrophage markers. For this purpose, we used median sagittal histological sections of the knee (Fig. 4). Despite the fact that larger knee joints are not the primary target of arthritis in either human patients or mice, multicellular pannus was easy to find within knee joint synovium upon routine histological examination (Fig. 4a, yellow arrowheads). We used the meniscus as a landmark to align and compare images (Fig. 4).

IBA1 is a calcium binding adaptor molecule also known as allograft inflammatory factor 1 that is involved in membrane ruffling and phagocytosis [21]. IBA1 is expressed in activated macrophages and also in inflamed synovium [22]. In mice with CAIA, IBA1+ macrophages were infiltrating the entire pannus, but were more abundant in synovial lining sides adjoining the synovial cavity (intimal lining), and were basically absent at the area of the bone-pannus interface (Fig. 4b, j). The size of the IBA1+ cells was much smaller than large CTHRC1+ fibroblast-like cells. Dispersed IBA1+ staining contrasted with CTHRC1 signal distribution.

Synovium was stained for α-SMA, which is expressed in vascular smooth muscle cells and in myofibroblasts. Expression of α-SMA is increased in the transition of normal FLS to activated FLS [23]. In our experiments, anti-α-SMA antibodies stained smooth muscle cells of the vasculature of the naïve non-inflamed synovium (data not shown), but IHC staining of synovial lining and pannus was found only in CAIA (Fig. 4c, d, yellow arrowhead). α-SMA+ cells were found mostly at the synovial intimal lining (Fig. 4c, d), and also at sites of the pannus contacting fibrocartilage of the meniscus (Fig. 4l). Vasculature was strongly positive for α-SMA (Fig. 4l).

CDH11 was previously established as one of the most specific markers for activated FLS [2, 6]. In this CAIA model, staining for CDH11 was most prominent in the intimal lining (Fig. 4g, h), although any part of the pannus, including the middle fibrous-like layer and the bone-pannus interface contained CDH11+ cells (Fig. 4i).

CTHRC1 IHC staining was found at the intimal lining and the bone-pannus interface, locations that greatly overlapped with sites of α-SMA and CDH11 expression (Fig. 4c-l). Taking together, we found CTHRC1 protein staining at several locations within the arthritic synovium. The strongest signal was found in inflamed pannus with some accumulation at the pannus borders represented by the intimal lining and the bone/cartilage–pannus interface. A much weaker and infrequent signal was found in meniscus fibrocartilaginous cells in contact with the pannus. Sometimes CTHRC1 positivity was found in ligaments and striated muscles, which could be remodeled due to inflammation. Practically no IHC staining was found in bone, articular cartilage, or growth plate chondrocytes (Figs. 1, 3, 4).

In brief, the conclusion of CTHRC1 synthesis by arthritic pannus FLS is supported by (1) fibroblast-like morphology of CTHRC1+ cells; (2) significant overlap between CTHRC1+, CDH11+ and α-SMA+ staining; and (3) no overlap with IBA1+ cell locations.

Overexpression of Cthrc1 transgene promoted migration of cells from several lineages, including primary mouse embryonic fibroblasts, smooth muscle and cancer cells [13, 14]. The effect of CTHRC1 upon the cell motility of activated synoviocytes and the more general effect in inflammatory conditions have never been studied.

We examined expression of CTHRC1 in murine fibroblasts and in RA-FLS using western immunoblotting. Two lines of mouse embryonic fibroblasts NIH 3T3 and C3H/10T1/2 were studied, and both were negative for CTHRC1 production (Fig. 5a). On the contrary, RA-FLS were positive for CTHRC1 expression, and for cadherin CDH11 and α-SMA production (Fig. 5b). This staining of human synoviocytes paralleled IHC staining of the mouse CAIA pannus (Fig. 4).

At IHC staining of inflamed synovial joints, some deposition of CTHRC1 into bony and fibrocartilage tissues was noticeable (Figs. 3, 4). Expression of CTHRC1 by transiently transfected CHO cells in vitro and secretion of the protein into media was confirmed (Fig. 5c). Recombinant human rhCTHRC1 is expressed intracellularly in adherent cells with a peak at day 3 post transfection. After centrifugation of the conditioned media at 1000 × g for 15 minutes, protein was detectable in the media supernatant as a single protein band at day 3 (Fig. 5c). CTHRC1 continued to accumulate in the media until at least day 5, and there was no proteolytic degradation of the secreted protein.

To study whether CTHRC1 has a promigratory effect upon activated FLS and embryonic fibroblasts, we treated cells with purified recombinant protein and assayed cell motility using transwell plates. Neither RA-FLS nor C3H/10T1/2 fibroblasts migrated in the absence of serum even if CTHRC1 was provided (Fig. 5d, e), hence, the CTHRC1 required serum components for promoting cell motility. Cells were able to move through the membrane in presence of 1–6 % FBS, and CTHRC1 demonstrated a positive effect on transwell cell motility (1.4–5-fold, p < 0.05) (Fig. 5d, e). As a positive control, we used serum taken from mice with CAIA. A mixture of equal parts of arthritic serum and FBS (5 % arthritic serum and 5 % FBS) in complete media showed the strongest promigratory effect (Fig. 5f). At the end of the experiment, transmigrated fibroblasts had a normal morphological spread (Fig. 5g).

To study detailed cellular mechanisms of motility promoted by CTHRC1, we used time-lapse microscopy of NIH 3T3 fibroblasts in specially designed μ-slide chambers (IBIDI USA Inc., Madison, WI, USA). The μ-slide bridge was coated with collagen type I. We treated fibroblasts with rhCTHRC1 at 200–1000 ng/ml in 10 % FBS and monitored cells at the bridge using time-lapse microscopy for up to 14 hours. In these conditions that were similar to those in the transwell assay experiment, CTHRC1 also demonstrated a statistically significant promigratory effect (Fig. 6). We traced trajectories of individual cells and combined them in Rose plots (Fig. 6a). In the presence of CTHRC1, fibroblasts migrated a notably longer distance on average, although the effect varied greatly for individual cells. The velocity of cell movement, directness and accumulated distance traveled by cells all increased by 40 % (p < 0.01). The Euclidean distance increased by 93 % (p < 0.05), which indicated that in the presence of CTHRC1, fibroblasts were more prone to maintaining a specific direction of migration once it was selected by the cell.

Another clear effect of the treatment was the elongation of cells or increase of cell size along the front-tail axis. Cell size was measured between the front of the lamellipodium and the end of the cell tail (Fig. 6e). Upon rhCTHRC1 treatment, NIH 3T3 fibroblast cell size was significantly increased at 32 %, p < 0.01 (Fig. 6d, e). The effect of cell elongation was obtained due to the protraction of the cell tail.