Transglutaminase activity arising from Factor XIIIA is required for stabilization and conversion of plasma fibronectin into matrix in osteoblast cultures

Bone 59 (2014) 127–138

Contents lists available at ScienceDirect

Bone journal homepage: www.elsevier.com/locate/bone

Original Full Length Article

Transglutaminase activity arising from Factor XIIIA is required for stabilization and conversion of plasma fibronectin into matrix in osteoblast cultures Cui Cui a, Shuai Wang a, Vamsee D. Myneni a, Kiyotaka Hitomi b, Mari T. Kaartinen a,c,⁎ a b c

Division of Biomedical Sciences, Faculty of Dentistry, McGill University, Montreal, QC, Canada Department of Applied Molecular Biosciences, Graduate School of Pharmaceutical Sciences, Nagoya University, Chikusa, Nagoya, Japan Division of Experimental Medicine, Department of Medicine, Faculty of Medicine, McGill University, Montreal, QC, Canada

a r t i c l e

i n f o

Article history: Received 22 August 2013 Revised 8 November 2013 Accepted 10 November 2013 Available online 15 November 2013 Edited by: J. Aubin Keywords: Plasma fibronectin Extracellular matrix Transglutaminase Factor XIIIA Collagen type I Osteoblasts

a b s t r a c t Circulating plasma fibronectin (pFN), produced by hepatocytes, is a major component of the noncollagenous bone matrix where it was recently shown in vivo in mice to control the biomechanical quality as well as the mineral-to-matrix ratio in bone. FN fibrillogenesis is a process generally requiring FN binding to cellular integrins, and cellular tension to elongate and assemble the molecule. Whether soluble pFN undergoes cell-mediated assembly in bone is not fully established. FN is a well-known substrate for transglutaminases (TGs), which are protein-crosslinking enzymes capable of stabilizing macromolecular structures. The role of this modification regarding the function of FN in bone matrix has remained unknown. Osteoblasts express two TGs— transglutaminase 2 and Factor XIIIA—and we have shown that Factor XIIIA is the main TG active during osteoblast differentiation. In the present study, conducted using MC3T3-E1 osteoblast cultures and bone marrow stromal cells, we demonstrate that pFN requires a TG-mediated crosslinking step to form osteoblast matrix in vitro. This modification step is specific for pFN; cellular FN (EDA-FN) does not serve as a TG substrate. Inhibition of pFN assembly using a TG inhibitor, or depletion of pFN from cell culture serum, dramatically decreased total FN matrix assembly in the osteoblast cultures and affected both the quantity and quality of the type I collagen matrix, and decreased lysyl oxidase and alkaline phosphatase levels, resulting in decreased mineralization. Experiments with isozyme-specific substrate peptides showed that FXIIIA is responsible for the crosslinking of pFN. Addition of exogenous preactivated FXIIIA to osteoblast cultures promoted pFN assembly from the media into matrix. Exogenous TG2 had no effect. Analysis of pFN and EDA-FN fibrils by immunofluorescence microscopy demonstrated that they form distinct matrix network, albeit with minor overlap, suggesting different functions for the two FN forms. Further analysis using EDA-FN blocking antibody showed that it regulated preosteoblast proliferation whereas pFN depletion from the serum had no effect on this process. In conclusion, our study shows that pFN assembly into bone matrix in vitro requires FXIIIA transglutaminase activity making pFN assembly an active, osteoblast-mediated process. © 2013 Elsevier Inc. All rights reserved.

1. Introduction The extracellular matrix of bone is generated by osteoblasts and consists primarily of collagen type I (COL I), noncollagenous proteins and proteoglycans. While COL I provides tensile strength and a mineralization scaffold for bone matrix, other protein components are thought to contribute to the regulation of cellular activity, and matrix quality and integrity, as well as to the control of mineralization [1,2]. Fibronectin (FN) is a major component of the noncollagenous bone matrix [3] and number of studies have demonstrated its importance for osteoblast

⁎ Corresponding author at: Faculty of Dentistry, McGill University, 3640 University Street, Rm. 72, Montreal, QC H3A 0C7, Canada. Fax: +1 514 398 8900. E-mail address: [email protected] (M.T. Kaartinen). 8756-3282/$ – see front matter © 2013 Elsevier Inc. All rights reserved. http://dx.doi.org/10.1016/j.bone.2013.11.006

function and bone matrix deposition [4–7]. FN is a ubiquitous extracellular matrix glycoprotein required for many cellular activities, including adhesion, proliferation, migration, and differentiation [8–12]. It exists as a soluble, circulating plasma FN (pFN) made by hepatocytes, and as cellular FN (cFN) synthesized by tissue-resident cells. cFN is a product of alternative splicing of precursor mRNA and contains additional type III domains A and/or B (EDA and/or EDB) which are not found in pFN. Conditional FN-knockout models have demonstrated that pFN and cFN have different roles in bone cell biology [13]. Circulating pFN constitutes the majority (90%) of the bone FN matrix and its conditional deletion in hepatocytes leads to decreased bone biomechanical properties and an altered mineral-to-matrix ratio demonstrating its importance for bone quality. Conditional deletion of cFN in osteoblasts affects osteoblast numbers and function, but has no effect on biomechanical properties of bone [13].

128

C. Cui et al. / Bone 59 (2014) 127–138

Both pFN and cFN undergo assembly into extracellular matrix fibrils. Generally the fibrillogenesis process requires FN binding to cell-surface integrins and cytoskeletal tension to induce extension and opening of FN structure which exposes FN self-assembly sites [9]. FN assembly leads to formation of deoxycholic acid-soluble fibrils, which are subsequently converted into a dense, detergent-insoluble fibrillar network. This fibrillar FN matrix has been shown to assist in the deposition of several other matrix components including COL I and III, fibrinogen, fibrillin, fibulin, laminin and tenascin-C as demonstrated in cultures of smooth muscle cells, megakaryocytes, endothelial cells, osteoblasts and fibroblasts [6,7,14–18]. FN fibrils can also bind and regulate bone morphogenic protein-1 (BMP-1) [19], and sequester growth factors such as transforming growth factor-β (TGFβ) via binding to latent transforming growth factor (TGF)-β binding protein-1 (LTBP-1) [20]. FN matrix has also been shown to activate lysyl oxidase (LOX) [21], emphasizing the important role of FN as a regulator of bone matrix quality and in the elaboration of an appropriate fibrillar COL I network. FN is a substrate of transglutaminases (TGs) which are a family of calcium-dependent enzymes that catalyze formation of ε-(γglutamyl)-lysine isopeptide crosslinks between glutamine (Q) and lysine (K) residues from specific substrate proteins [22–24]. The presence of isopeptide crosslinks in a protein structure can alter protein conformation [25], increase its interaction with other proteins or cells [26,27], and change the biomechanical stability or solubility of the substrate protein [28]. The TG enzyme family consists of TGs 1–7 and Factor XIIIA (FXIIIA), as well as enzymatically inactive erythrocyte band 4.2 [22,24]. In vitro, it is known that FN can be crosslinked by FXIIIA to fibrin [29], to itself [30] and to COL I [31]. The TG-reactive glutamine residues (Q3, 4, 6, 7, 9) are located in the N-terminal 27 kDa fragment [32], and Q246 in the N-terminal Type I module of the FN molecule [33]. Very little is known about the role of this modification with regard to FN assembly or function in bone. We have previously demonstrated that osteoblasts express two TG family members—TG2 and FXIIIA; of these two, only FXIIIA is secreted by osteoblasts into the extracellular space [34–39]. FN is a major extracellular TG substrate during osteoblast differentiation and matrix assembly [33,38]. Inhibition of TG activity in osteoblast cultures leads to defective FN matrix assembly and decreased COL I deposition, and this results in delayed osteoblast differentiation and mineralization in vitro [34]. In this study we have investigated the role of TG crosslinking activity in FN matrix assembly, and we demonstrate that TG activity arising from FXIIIA is specifically required for stabilization of pFN into the osteoblast matrix. pFN, but not cFN, acts as a FXIIIA substrate and requires a TG-mediated insolubilization step for its fibrillogenesis. A lack of pFN or TG activity in osteoblast cultures results in decreased COL I deposition, an altered COL I fibril network, low alkaline phosphatase activity, decreased lysyl oxidase levels and decreased mineralization in vitro. Our results bring additional evidence the importance of the role of TGs and pFN in bone matrix formation, and demonstrate that pFN assembly in bone is an active, osteoblastmediated event. 2. Materials and methods 2.1. Antibodies, proteins and peptides Rabbit anti-FN antibody and normal rabbit IgG were purchased from Millipore (Temecula, CA, USA). Monoclonal anti-cellular FN (clone FN-3E2, clone FN-15), monoclonal anti-vimentin (clone vim 13.2), anti-actin antibody, bovine plasma FN, and monodansyl cadaverine were obtained from Sigma Aldrich (St. Louis, MO, USA). Mouse monoclonal anti-FN (IST-9) was from Abcam (Cambridge, MA, USA). Anti-dansyl antibody was purchased from Molecular Probes (Eugene, OR, USA). Rabbit anti-biotin antibody was from Rockland Immunochemicals (Gilbertsville PA, USA), horseradish peroxidase linked antirabbit IgG and anti-mouse IgG were obtained from Cell Signaling (Boston, MA, USA) and GE Healthcare Life Sciences (Baie d'Urfe, QC,

Canada), respectively. Anti-lysyl oxidase (LOX V-20) antibody was from Santa Cruz (Dallas, TX, USA). The Alexa Fluor® secondary antibodies 488 (green) and 568 (red) and DAPI (4′, 6-diamidino-2-phenylindole) were from Thermo Scientific (Rockford, IL, USA). Biotinylated peptides (F11: Biotin-DQMMLPWPAVAL, F11QN: Biotin-DNMMLPWPAVAL, T26: Biotin-HQSYVDPWMLDH, T26QN: Biotin-HNSYVDPWMLDH) were synthesized by Biologica Co. (Nagoya, Japan) and by Biomatik Corp (Wilmington, DE, USA). Preactivated human FXIIIA enzyme was purchased from Zedira (Darmstadt, Germany) and guinea pig liver TG2 was obtained from Sigma. All other reagents, if not specified below, were purchased from Sigma or Fisher Scientific. 2.2. Cell culture, differentiation and treatments MC3T3-E1 preosteoblasts (subclone 14) were cultured in α-MEM (Gibco, Burlington, ON, Canada) supplemented with 10% fetal bovine serum (FBS), 1% penicillin–streptomycin and incubated in a humidified atmosphere with 5% CO2 at 37 °C. Cell differentiation into osteoblasts was induced by media containing 50 μg/ml ascorbic acid and 10 mM β-glycerophosphate, which is referred to hereafter as differentiating media (DM). Control cells were treated with media alone (M). All treatments were begun 24 h after plating and cells were treated every other day until the end time point. COL I levels were visualized and quantified on day 12 of culture using the Picrosirius Red staining method [40]. Mineralization was visualized by von Kossa staining and alkaline phosphatase activity assay was performed as described previously [34]. For TG substrate labeling and TG inhibition experiments, cells were cultured as above. For labeling experiments, monodansyl cadaverine (MDC) (100 μM) was added to cell culture media on day 4 for 24 h which was followed by immunofluorescence staining or protein extraction on day 5. When used as an inhibitor, MDC was continuously applied to cell culture for 5 days at a 100 μM concentration. NC9, an irreversible TG activity inhibitor [41] (synthesized by CGene Tech, Indianapolis, IN, USA) was administered to the cells from day 0 to day 5 at a 25 μM concentration. Peptides F11, F11QN, T26 and T26QN (10 μM concentration) were given to cells on day 4 and protein was extracted 24 h later. In experiments where the effects of FXIIIA and TG2 on FN assembly were examined, both enzymes were added to the cultures for 24 h at a concentration of 0.1 μg/ml. 2.3. Bone marrow stromal cell isolation and culture C57BL/6 mice (from Jackson Laboratories) (6–10 wks old) were used for isolating bone marrow stromal cells (BMSCs)[42]. Femurs and tibias were collected and bone marrow was flushed with 2% FBS in PBS. The cell suspension was filtered, and centrifuged at 1000 rpm at 4 °C for 5 min. Cells were seeded into a T-25 cell culture flask, and incubated in a humidified atmosphere with 5% CO2 at 37 °C. Cells were cultured in DMEM (Gibco) supplemented with 20% FBS, 1% penicillin– streptomycin. Passage 3–5 BMSCs were used for experiments. Osteoblast differentiation was induced by the addition of 50 μg/ml ascorbic acid, 10 mM β-glycerophosphate and 10 nM dexamethasone. DOCsoluble and DOC-insoluble protein extraction, Picrosirius Red staining and von Kossa staining were performed on day 18 of culture. 2.4. Protein extraction and Western blotting (WB) Total cellular protein preparations were obtained using buffer containing the following components: 150 mM NaCl, 10 mM Tris–HCl pH 7.2, 5 mM EDTA, 0.1% sodium dodecyl sulphate (SDS), 1% Triton X-100, 1% sodium deoxycholate (DOC). Cytoplasmic, membrane, nuclear, cytoskeletal and matrix fractions were prepared with ProteoExtract® Subcellular Proteome Extraction Kit (Calbiochem, Gibbstown, NJ, USA). DOC-soluble and DOC-insoluble fractions were extracted as described previously [43]. Briefly, cultured osteoblasts were lysed with 2% DOC lysis buffer and the extract was

C. Cui et al. / Bone 59 (2014) 127–138

centrifuged at 14,000 g 4 °C for 20 min. The supernatant was collected and identified as the ‘DOC-soluble’ extract. The remaining pellet was solubilized in 1% SDS lysis buffer and identified as the ‘DOC-insoluble’ extract. Protein concentrations were determined using the BCA protein assay kit (Pierce, Rockford, IL, USA). Proteins (10 μg/lane) were resolved by 8.5% SDS-PAGE gel electrophoresis under reducing conditions with SDS-sample buffer, and transferred to PVDF membranes (BioRad, Mississauga, ON, Canada). Membranes were blocked with 5% non-fat milk powder in Tris-buffered saline (TBS)-Tween buffer, and individual proteins were detected with specific antibodies followed by corresponding horseradish peroxidase conjugated secondary antibodies. Protein bands were visualized with Amersham ECL Prime Western Blotting Detection Reagent (GE Healthcare).

A kDa

M

M + MDC

C Me Ck/m

C Me Ck/m

DM

DM + MDC C Me Ck/m

C Me Ck/m

129

2.5. Immunoprecipitations Proteins were extracted using subcellular proteome extraction kit (Calbiochem) or by preparing DOC-soluble or DOC-insoluble fractions (as above). Protein extracts (500 μg total) were first pre-cleared with 100 μl of a Protein G Plus Agarose bead slurry (Pierce) at 4 °C for 30 min under agitation. Beads were then centrifuged at 2500 g for 3 min and the supernatant collected. One (1) μg of antibody was added to the supernatant and the mix was incubated overnight at 4 °C. The protein–antibody mix was added to 100 μl Protein G Plus Agarose and the mix was incubated for 2 h at room temperature. The supernatant was removed after centrifugation, and beads were washed four times with the following buffer: 150 mM NaCl, 10 mM Tris–HCl PH 7.2, 5 mM EDTA, 0.1% SDS, 1% Triton X-100, and 1% DOC. The final

E

M + MDC Dansyl

DM + MDC 20µm Dansyl

20µm

250 75

vimentin

IgG M DM

B

IgG M

M insol

Merge

Merge

DAPI

DAPI

Ck/m

IP: Anti-FN WB: Anti-dansyl

DOC sol

FN

DM

Ck/m

C

FN

IP: Anti-dansyl WB: Anti-FN

M+MDC sol

insol

DM sol

insol

DM+MDC sol

insol

FN

dansyl vimentin

DOC-insoluble FN /vimentin ratio

D 8

*

*

6 4 2 0

M

M+MDC

DM

DM+MDC

Fig. 1. Fibronectin (FN) is major TG substrate in MC3T3-E1 osteoblast cultures—TG activity affects FN solubility. (A) Analysis of MDC labeling in subcellular protein fractions of MC3T3-E1 osteoblasts. Cells were cultured for 5 days with medium only (M) or with differentiating media (DM) together with 100 μM monodansyl cadaverine (MDC) to detect its incorporation into substrates in the presence of TG activity. WB analysis using anti-dansyl antibody shows MDC label incorporation only into cytoskeletal and matrix (Ck/m) fractions in both treatment groups. Less labeling is seen in differentiating osteoblasts. Cytosolic (C) and membrane (M) fractions are devoid of labeling. Vimentin was used as loading control. (B) Immunoprecipitation of FN and the dansyl group of MDC from labeled Ck/m fractions demonstrating that dansyl-labeled material is FN. Normal rabbit IgG was used as negative control. (C) Solubility of FN in the presence of MDC. Osteoblasts were treated with MDC (100 μM) continuously for 5 days to examine its effect on FN solubility. WB analysis of FN from DOC-soluble (sol) and DOC-insoluble (insol) extracts shows less DOC-insoluble FN in differentiating osteoblasts (DM) compared to medium-only treated cells (M). Dansyl detection of MDC by WB shows that only the DOCinsoluble fraction is labeled by MDC suggesting that only DOC-insoluble FN is crosslinked. (D) Quantification of FN in DOC-soluble (sol) and DOC-insoluble (insol) extractions shows that MDC treatment significantly increases FN solubility in SDS (i.e. DOC-insoluble). Quantification was done by Image J software with 3 separate blots and vimentin was used for normalization. Data are presented as mean values ± SEM. *p b 0.05. (E) Immunofluorescence microscopy and MDC co-localization with FN in M- and DM-treated cells. MDC (red) co-localizes with FN matrix (green) in clusters in the matrix network (merge in yellow). Visibly more FN matrix was seen in differentiating cells.

130

C. Cui et al. / Bone 59 (2014) 127–138

sample was mixed with SDS-sample loading buffer and boiled at 95 °C for 5 min to elute protein from the beads. Samples were centrifuged and the supernatant was collected and analyzed by WB. Normal rabbit IgG was used as a negative control.

2.10. Immunofluorescence microscopy

2.6. Preparation of plasma FN-depleted serum pFN was eliminated from FBS with Gelatin Sepharose 4B beads (GE Healthcare) according to a previously published protocol [44]. Successful deletion (90% reduction) of pFN was confirmed by FN ELISA assay using rabbit anti-FN antibody (Millipore). Cells were treated with media containing pFN-depleted serum, and 25 μg/ml bovine plasma FN was added as exogenous plasma FN into cell culture to demonstrate rescue. 2.7. Biotinylation of plasma FN pFN was labeled by biotin following a published protocol [43]. Briefly, bovine pFN (1 mg) was dialyzed with Slide-A-Lyzer dialysis cassettes (Pierce) against PBS overnight at 4 °C. Dialysis buffer was changed and dialysis was continued for an additional 1 h at room temperature. Forty (40) μl of 1 mg/ml EZ-link sulfo-NHS-biotin (Pierce) was added to the FN solution and incubated for 30 min at room temperature on a shaker. Biotinylated pFN (bpFN) was then dialyzed against PBS, and centrifuged at 14,000 g for 15 min. This supernatant, containing the bpFN, was used in experiments. 2.8. Biotinylated plasma FN ELISA Ten (10) μg/ml bpFN was added to the cell culture media and given to the cells overnight. The media was collected from cell cultures and bpFN was detected by ELISA assay by the following protocol. Fifty (50) μl of media from each sample was coated onto each well of a 96-well cell culture plate (in triplicate) and incubated overnight at 4 °C, then blocked with 3% BSA for 1 h at room temperature. This was followed by a washing step and the captured bpFN was detected with streptavidin horseradish peroxidase conjugate. Visualization was done by applying TMB (3,3′,5,5′-Tetramethylbenzidine) substrate solution to the wells for 10 min, after which the reaction was stopped by addition of 50 μl 2 M H2SO4. The absorbance was measured at 450 nm using an ELISA microplate reader. 2.9. Lysyl oxidase ELISA Lysyl oxidase was analyzed from conditioned media of MC3T3-E1 osteoblast cultures and assessed by ELISA assay where 50 μl media was coated onto each well in 96-well cell culture plates (in triplicate) and incubated overnight at 4 °C. Plates were washed and blocked with 3% BSA, incubated with anti-lysyl oxidase antibody, followed by incubation with anti-goat horseradish peroxidase conjugate for 1 h at room temperature. Visualization was done by adding TMB (3,3′,5,5′-

A

B Day 0

bpFN + MDC

Day 4 Day 5

Extract / IP

Tetramethylbenzidine) substrate solution to the wells for 10 min after which the reaction was stopped by addition of 50 μl 2 M H2SO4. The absorbance was measured at 450 nm using an ELISA microplate reader.

MDC bpFN

Cells were plated on Nunc® Lab-Tek® chamber slides (Fisher) and cultured as described above. On day 5, cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS) for 10 min at room temperature. Slides were washed three times with PBS and blocked with 2% bovine serum albumin (BSA) in PBS. This was followed by primary antibody incubation for 2 h at room temperature, and a washing step with 0.1% BSA in PBS and incubation with AlexaFluor® secondary antibody conjugates. Nuclei were stained by 4′,6-diamidino-2-phenylindole (DAPI). Samples were mounted with Prolong Gold Anti-Fade medium (Invitrogen) and dried overnight at room temperature. Slides were analyzed using an Axioskop 2 upright fluorescence microscope equipped with an AxioCam MRm camera and AxioVision 4.8 imaging software from Carl Zeiss. The black and white images were generated by Adobe Photoshop from original color images. 2.11. Cell proliferation assay Cell proliferation was assessed using a BrdU Cell Proliferation ELISA kit from Roche Diagnostics. MC3T3-E1 osteoblasts were plated at a density of 2.5 × 104 cells/cm2 and cultured for 20 h with the indicated treatments. Cells were then incubated with 10 μM BrdU solution and immunostained for BrdU 4 h later following the kit instructions. 2.12. Scanning electron microscopy (SEM) Cultured cells were fixed for 1 h at room temperature with 4% paraformaldehyde and 0.1% glutaradehyde in 0.1 M sodium cacodylate, pH 7.3, and then continued overnight at 4 °C. After rinsing with washing buffer (0.1 M sodium cacodylate, pH 7.3), cells were dehydrated sequentially in a 30%, 50%, 70% 80%, 90%, 95% graded ethanol series for 10 min each and as a final step with 100% ethanol for 10 min three times followed by drying with critical point dryer (EM CPD030 from Leica). SEM analysis of gold-coated samples was performed using an Environmental Scanning Electron Microscope (ESEM Quanta 200 FEG from FEI) under high-vacuum mode. COL I fibril thickness was quantified from 538 collagen fibrils taken from 5 to 6 images from different areas of each treatment group. Measurements were done using Image J software (NIH). 2.13. Image and statistical analysis WBs were quantified using Image J software (NIH). Error bars represent standard error of the mean (SEM) of three separate samples or experiments. Statistical significance was determined by the Student's t test.

Ck/m

DOC -insoluble

+ -

+ +

+ +

+ +

+ +

DM/IgG

M

DM

M

DM

Ck/m

+ DM IP: Anti-dansyl WB: Anti-biotin IP: Anti-dansyl WB: Anti-FN

Fig. 2. Plasma FN is a substrate for TG activity and becomes incorporated into osteoblast matrix. (A) Scheme describing the flow of the experiment. MC3T3-E1 osteoblasts were given biotinylated plasma FN (bpFN) together with monodansyl cadaverine (MDC) for 24 h on day 4. Cytoskeletal and matrix fractions (Ck/m) and DOC-insoluble fractions were prepared on day 5, and used for immunoprecipitation (IP). (B) IP of MDC-labeled material with anti-dansyl antibody and positive WB detection of the material with anti-biotin and anti-FN antibodies confirms that bpFN incorporates into the osteoblast matrix and that it is a substrate for TG activity.

C. Cui et al. / Bone 59 (2014) 127–138

mimicking a lysine residue and integrating into the TG-reactive Q residues. MDC incorporation into protein(s) was analyzed by WBs of subcellular protein fractions. Fig. 1A shows strong MDC-labeling of a 250 kDa protein in the cytoskeletal/matrix fraction prepared from both nondifferentiating and differentiating cells. Labeling was not found on the membrane or in the cytosol. Differentiating cells showed less WB-detectable labeling as compared to nondifferentiating osteoblasts. Immunoprecipitation of FN followed by dansyl detection of MDC by WB demonstrated that FN was a prominent TG substrate in both nondifferentiating and differentiating cells (Fig. 1B). Immunoprecipitation with dansyl-antibody followed by detection with FN antibody confirmed this result (Fig. 1B). FN matrix assembly and remodeling via endocytosis is a continuous process. In the assembly process, FN forms short deoxycholate (DOC) soluble fibrils which are further converted into a dense detergentinsoluble fibrillar network that is insoluble to DOC, but soluble to SDS.

3. Results 3.1. TG-mediated crosslinking of FN acts as an additional insolubilization step in FN matrix assembly To begin our investigation on how TG-mediated crosslinking of FN might affect its assembly and function in osteoblast cultures, we first confirmed that FN acts as a TG substrate and investigated whether the TG-mediated crosslinking affects its solubility, i.e. FN matrix conversion from detergent-soluble to detergent-insoluble matrix. For both experiments nondifferentiating (treated with medium only, M) and differentiating (treated with differentiating media, DM) MC3T3-E1 osteoblasts were labeled with monodansyl cadaverine (MDC) on day 4 of culture for 24 h. MDC is a dansyl-containing primary amine that covalently incorporates into TG-reactive Q residues of TG substrates. It also acts as a competitive inhibitor for protein–protein crosslinking by

A

131

M

DM+NC9

bpFN / tot. FN / merge

bpFN / DAPI

DM

20 µm

B

DM + NC9

DM

20µm

FN

bpFN in media

C

1.2

**

1.0

**

D

DOC-soluble DM M DM +NC9

0.8

DOC-insoluble M

DM DM +NC9

0.6 0.4 0.2 0.0 input

DM

bpFN

bpFN

actin

vimentin

DM+NC9

Fig. 3. TG activity is required for plasma FN assembly into osteoblast matrix. (A) Immunofluorescence (IF) microscopy of biotinylated plasma FN (bpFN) incorporation into MC3T3-E1 osteoblast matrix. bpFN was given to osteoblasts for 24 h on day 4. Immunofluorescence microscopy shows a strong detection of bpFN in osteoblast matrix (red) and a dramatic decrease in matrix assembly after NC9 treatment. Total FN (tot FN) matrix assembled (green) prior to bpFN addition shows that bpFN forms new fibrils and only partially integrates into existing fibrils (yellow, merge). (B) Higher-magnification, gray-scale image of the FN network shows that TG inhibition with NC9 results in a discontinuous FN network. (C) ELISA analysis of bpFN left in cell culture media after incubation with cells shows a significant increase in soluble bpFN in media upon NC9 treatment. Triplicates were tested for each treatment. Data are presented as mean values ± SEM. ** p b 0.01. (D) WB analyses of bpFN levels in the osteoblast cultures. NC9 treatment inhibits bpFN incorporation into DOC-soluble and DOC-insoluble protein fractions. Actin and vimentin were used as loading controls for DOC-soluble and DOC-insoluble fractions, respectively.

C. Cui et al. / Bone 59 (2014) 127–138

We next investigated in the cultures if TG-mediated crosslinking of FN is part of the insolubilization process and analyzed the effect of MDC on FN assembly into DOC-soluble and DOC-insoluble fractions. Analysis was done by WB followed by quantification of FN bands. As seen in Fig. 1C and D, differentiating osteoblasts had significantly less of both DOCsoluble and DOC-insoluble FN, although differentiating osteoblasts deposited visibly more FN as seen in the immunofluorescence images of Fig. 1E. This suggests that FN is converted into nonextractable matrix. FN co-localized with the dansyl-group of MDC in immunofluorescence analysis in both nondifferentiating and differentiating cells. This suggests that a TG-mediated FN insolubilization process occurs upon osteoblast differentiation. In general, an additional gelatinous pellet remains in extraction tubes after preparation of DOC-soluble and DOCinsoluble extracts. The pellet contains likely COL I and the insoluble, nonextractable FN (data not shown). The concept that TG-mediated FN insolubilization occurs upon differentiation is further supported by the following observations: i) MDC, which acts as a competitive inhibitor of protein–protein crosslinking, resulted in significantly (p b 0.05) more DOC-insoluble FN in differentiating cells, but not in nondifferentiating cells (Fig. 1C), ii) MDC had no major effect on DOC-soluble FN which is considered to precede the formation of the DOC-insoluble FN matrix (Fig. 1C), iii) the MDC probe only incorporated into the DOC-insoluble fraction, as seen by dansyl detection of fractions (Fig. 1C). Thus, we conclude that only DOC-insoluble FN is crosslinked by TG activity and that the co-localization of MDC and FN by immunofluorescence at specific sites in the matrix likely represents TGcrosslinked DOC-insoluble FN (Fig. 1E). 3.2. Plasma FN constitutes the majority of FN matrix in osteoblast cultures and is crosslinked by TG activity It has been shown that many cell types assemble the majority of their FN matrix from plasma FN (pFN) which is abundant in the

A

DOC-soluble M

DM DM +NC9

bovine/fetal serum routinely added to culture media [45,46]. We therefore examined whether pFN acts as a TG substrate. For this, we added exogenous biotinylated pFN (bpFN) together with MDC to cell cultures on day 4 for 24 h. At day 5, cytoskeletal/matrix and DOC-insoluble fractions were prepared and immunoprecipitated with anti-dansyl antibody followed by detection of biotin from bpFN (Fig. 2A). As presented in Fig. 2B, MDC immunoprecipitation with dansyl-antibody captured bpFN confirming that pFN acts as a TG substrate and becomes crosslinked into osteoblast matrix. We next investigated whether inhibition of TG activity is affected by bpFN assembly into matrix. bpFN was given to the cells for 24 h together with NC9, an irreversible TG inhibitor containing a dansyl-group. Immunofluorescence microscopy analysis (Fig. 3A) showed that the exogenous bpFN formed the majority of FN matrix in both nondifferentiating and differentiating osteoblasts. Staining for total FN (using a general FN antibody) (green) showed FN that was deposited before bpFN addition. Exogenous bpFN formed new fibrils (red) and integrated with the previously deposited FN network as shown by co-localization in yellow (Fig. 3A). Deposition of bpFN into matrix was dramatically inhibited by the NC9 treatment. Neither new bpFN matrix nor previously deposited FN network existed in long fibrillar structures. NC9 treatment resulted in a discontinuous FN network as seen at higher magnification and high contrast gray-scale immunofluorescence images (Fig. 3B). Furthermore, ELISA analysis of bpFN levels in the cell culture media after incubation showed that the NC9-treated group had significantly more bpFN left in the media after incubation with cells (Fig. 3C) compared to incubation with differentiating control cells. Simultaneously, a decrease in bpFN in both DOC-soluble and DOC-insoluble fractions was observed upon NC9 treatment (Fig. 3D). The effect of NC9 on total FN matrix assembly was confirmed in MC3T3-E1 osteoblast and bone marrow stromal cell (BMSC) cultures (Figs. 4A–D). In both culture models, NC9 had a major effect on both the DOC-soluble and the DOC-insoluble FN matrix. NC9 dramatically decreased the levels of DOC-soluble

C

DOC-insoluble M

DM DM +NC9

1.2

total FN

total FN

vimentin

actin dansyl

dansyl

actin

bpFN in media

132

**

1.0 0.8

*

0.6 0.4 0.2 0.0

vimentin

M MC3T3-E1

D M

DM DM +NC9

DOC-insoluble M

DM+NC9

D18 BMSC

B DOC-soluble

DM

DM DM +NC9

M

DM

DM +NC9

COL I

total FN

total FN

actin

vimentin

Mineral

D18 BMSC

D18 BMSC

Fig. 4. Inhibition of TG activity attenuates plasma FN assembly from media into matrix and changes FN solubility in the matrix of MC3T3-E1 osteoblast and bone marrow stromal cell cultures. (A) WB analysis of FN in DOC-soluble and DOC-insoluble material from cells treated with medium only (M) and with differentiating media (DM) and DM + NC9. Inhibition of TG activity with NC9 leads to less DOC-soluble and dramatically more DOC-insoluble FN matrix. Dansyl detection of NC9 shows localization of TG enzyme in both DOC-soluble and DOCinsoluble fractions. (B) WB analysis of FN in DOC-soluble and DOC-insoluble protein material from day-18 bone marrow stromal cell cultures (BMSC) that were differentiated into osteoblasts. This data confirms that seen using MC3T3-E1 cells where NC9 treatment decreases levels of DOC-soluble FN and increases the solubility of FN to SDS (DOC-insoluble). Actin and vimentin were used as loading control. (C) Detection of bpFN in media of BMSC cultures after incubation with cells show increased, soluble bpFN left in the media compared to control cells. Data are presented as mean values ± SEM. * p b 0.05, ** p b 0.01. (D) Picrosirius Red staining and von Kossa staining of day-18 BMSC culture showing decreased COL I deposition and mineralization after NC9 treatment. COL I: type I collagen.

C. Cui et al. / Bone 59 (2014) 127–138

(in Figs. 3A and B). Adding exogenous pFN back to pFN-depleted serum/media rescued FN matrix assembly (Fig. 5A). To examine if NC9 treatment and pFN depletion have similar effects on in vitro bone formation, we grew osteoblasts under both conditions and investigated COL I deposition, mineralization and alkaline phosphatase levels. Figs. 5B and C show that pFN depletion and NC9 treatment significantly attenuated COL I deposition as quantified by Picrosirius Red staining and decreased mineralization as visualized by von Kossa staining; the effects were more pronounced by pFN depletion. As seen in Fig. 5D, alkaline phosphatase activity during osteoblast differentiation was decreased by both treatments. NC9 had an inhibitory effect on BMSC differentiation as shown earlier in Fig. 4D. It has been demonstrated that FN can act as a scaffold for COL I deposition and can affect the activity of factors relevant to COL I fibrillogenesis, such as BMP-1 and LOX [6,7,47]. Thus, we used scanning electron microscopy (SEM) to analyze the characteristics of COL I fibrils deposited under these conditions. As seen in Fig. 6A, NC9 treatment resulted in i) COL I fibrils that appeared thicker and more coalesced than COL I fibrils in control differentiating cells, ii) COL I fibrils that appeared shorter or fragmented. Similarly, pFN depletion led to visibly thicker COL I fibrils. Quantification of COL I fibril diameter by image analysis showed that COL I fibers were indeed 38.8% thicker in NC9-treated cultures as compared to COL I fibrils of the normally differentiated group (Fig. 6B). Analysis of LOX levels in conditioned medium (Fig. 6C) showed a dramatic, significant decrease after NC9 treatment indicating that pFN assembly and TG activity are important contributors to the COL I assembly machinery and affect overall quantity and quality of COL I matrix.

FN and increased the DOC-insoluble FN to the levels of control, nondifferentiating cells (Figs. 4A and B), i.e. NC9 increased FN solubility to SDS-buffer. The decrease of DOC-soluble FN was accompanied by increased FN in the media of MC3T3-E1 cells upon NC9 treatment (Fig. 4C) confirming that less FN was assembled into the matrix. Altogether, these data suggest that pFN is a major contributor to osteoblast matrix and requires a TG enzyme to capture it from the media into DOC-soluble matrix, and a TG-mediated, covalent stabilization step for its assembly into DOC-insoluble matrix as well as finally into non-extractable matrix. The sequential ‘capture and crosslink’ step is also supported by the following data: i) MDCmediated inhibition of crosslinking only affects DOC-insoluble FN matrix, ii) NC9, which binds to the active TG enzyme, inhibits the formation of both DOC-soluble and DOC-insoluble FN and blocks FN capture from the media, iii) MDC is found covalently bound only to DOC-insoluble matrix, whereas NC9 is found in both. 3.3. TG inhibition, or lack of plasma FN in serum, inhibits COL I deposition and ALP activity and delays osteoblast differentiation We have previously demonstrated that NC9 treatment decreases COL I deposition, alkaline phosphatase activity and mineralization which was related to poor FN assembly and decreased COL I secretion and assembly into matrix [34,35]. To further link the effects of TG inhibition to the lack of pFN assembly in osteoblast cultures, we compared effects of NC9 with effects of pFN-depleted FBS on osteoblast function. Results presented in Fig. 5A show that pFN depletion results in highly similar discontinuous FN matrix as observed after the NC9 treatment

A

DM

133

DM – pFN

DM – pFN + exog. FN

20µm

M

DM

DM +NC9

DM -pFN

COL I

Mineral

C COL I µg/µl

2.5

**

2.0

**

1.5

D DM DM + NC9 DM - pFN

16

ALP activity (10-4U/10 µg protein)

B

14

*** ***

12 10 8

**

6 4

1.0 2

0.5 0.0

0 M

DM

DM+NC9 DM-pFN

0

2

4

6

8

10

12

Day Fig. 5. Inhibition of TG activity as well as depletion of plasma FN from serum results in a delay in MC3T3-E1 osteoblast differentiation. (A) pFN depletion from FBS results in decreased and discontinuous FN matrix (DM-pFN) as analyzed by immunofluorescence microscopy (red). Cell density is represented by DAPI staining in the inset. Adding the exogenous pFN back to the depleted media (DM—pFN + exog.FN) rescues FN matrix assembly. (B,C) Culturing osteoblasts with TG inhibitor NC9 or with pFN-depleted serum significantly decreases COL I deposition, as assessed by Picrosirius Red staining, and affects mineralization as assessed by von Kossa staining of the mineral. COL I quantification data are presented as mean values ± SEM. ** p b 0.01. (D) Alkaline phosphatase activity (ALP) was measured spectrophotometrically at 405 nm using p-nitrophenyl phosphate as a colorimetric substrate. TG inhibitor NC9 and pFN depletion from serum both result in a significant decrease in ALP activity. Data are presented as mean values ± SEM. *** p b 0.001, ** p b 0.01.

134

C. Cui et al. / Bone 59 (2014) 127–138

3.4. EDA-FN is not a TG substrate: different functions for plasma FN and EDA-FN In addition to pFN, osteoblasts also secrete and assemble cellular FN (cFN) which is a specific splice variant of FN that contains either EDA and/or EDB domains [8]. cFN can be assembled into both DOC-soluble and DOC-insoluble fibrils. Conditional deletion of cellular FN affects mainly osteoblast numbers in mouse bone [13]. We therefore asked if EDA-FN was crosslinked by TG activity and whether it was required for osteoblast proliferation. As seen in Fig. 7A, EDA-FN co-localized with MDC in both nondifferentiating and differentiating osteoblasts at specific sites of the FN network. However, numerous attempts to detect EDA-FN together with MDC (dansyl) in immunoprecipitation experiments from cytoskeletal/matrix fractions and from DOC-insoluble extraction (Fig. 7B) failed to show that it was covalently labeled. This was likely not because of lower levels of EDA-FN in cultures; it was detected in abundance in these extracts as seen in the positive control of these immunoprecipitation experiments. Therefore, we conclude that EDA-FN is not crosslinked by TG activity and that the EDA-FN/MDC co-localization might represent EDA-FN co-localization with MDClabeled pFN. Indeed, although EDA-FN and bpFN show mostly distinct fibrillar networks (Fig. 7C), they do co-localize within the matrix (Fig. 7C). As expected from in vivo data showing the relevance of circulating pFN in matrix formation, EDA-FN also appears to form only small portion of the total FN matrix in MC3T3-E1 osteoblast cultures. Since EDA-FN has been linked to control of cell proliferation [48], we asked if EDA-FN could regulate osteoblast proliferation and whether pFN might have an effect. Assessment of cell proliferation was done for 24 h by BrdU labeling of preosteoblasts in pFN-depleted serum and in the presence of inhibitory EDA antibody. The results showed that only EDA-FN antibody decreased cell proliferation. A 59% and 51% reduction in proliferation compared to control was observed with 1:100 and

A

DM

DM + NC9

1:200 dilutions of the blocking antibody, respectively. pFN depletion had no effect on cell proliferation (Fig. 7E). We conclude that pFN and EDA-FN have different functions in osteoblast matrix. EDA-FN regulates osteoblast proliferation and pFN modulates matrix assembly. Only pFN is a TG substrate and can be covalently crosslinked.

3.5. Plasma FN is a specific substrate for FXIIIA which contributes to plasma FN matrix assembly in osteoblast cultures Our previous results have shown that osteoblasts express two TG enzymes—TG2 and FXIIIA—of which FXIIIA is the predominant, active crosslinking TG in osteoblast cultures. It is upregulated during osteoblast differentiation, localizes to the cell surface and is secreted to the matrix in the presence of extracellular COL I [39], while TG2 is localized to the cell surface having no crosslinking activity. Thus, to further confirm whether pFN is a specific substrate for FXIIIA in osteoblast cultures, we used highly reactive and isozyme-specific Q-donor substrate probes that have been developed from a phage display random peptide library [49–51]. The peptides—F11 and T26—are specific activity probes and incorporate into substrates of FXIIIA and TG2, respectively. Both probes have been tested in isozyme-specific assays in vitro to confirm their specificity as well as in in situ activity assays in tissue sections [52,53]. Cells were grown with biotinylated F11 and T26 peptide for 24 h. Biotinylated controls—F11QN and T26QN (reactive Q-residue is replaced by unreactive N-residue)—were used as negative controls. Labeled substrates were pulled down with Neutravidin beads from cell extracts and analyzed by WB. Fig. 8A shows that FN was pulled down only when F11 was used. T26 peptide was not found to be incorporated into FN. EDAFN was not labeled by either peptide. Therefore, we conclude that pFN is a specific substrate for FXIIIA and confirm that EDA-FN is not a substrate for either of the TGs.

DM – pFN

B

a)

COL I fibril diameters (nm)

2 µm

b)

500 nm

***

40 30 20 10 0 DM

DM +NC9

200 nm

LOX in conditioned media (OD value at 450nm)

C 0.2

**

0.2 0.1 0.1 0.0 M

DM

DM +NC9

Fig. 6. Inhibition of TG activity and depletion of plasma FN from the cultures leads to an altered COL I network and decreased lysyl oxidase activity. (A) Scanning electron microscopy (SEM) analysis of COL I matrix in MC3T3-E1 osteoblasts grown for 12 days in the presence or absence of TG inhibitor NC9 or in pFN-depleted media. Cells were fixed with 4% paraformaldehyde and 0.1% glutaraldehyde in 0.1 M sodium cacodylate. Compared to the DM group, NC9 treatment led to thicker collagen fibrils that appeared to coalesce together, and to a discontinuous COL I network. (B) Quantification of COL I fibril diameter by NIH Image J software shows that NC9 treatment leads to 38.8% thicker fibrils compared to normal differentiated cells. ***p b 0.001. (C) ELISA assay of lysyl oxidase (LOX) levels in conditioned media. NC9 treatment results in a significant decrease in LOX levels. Data are presented as mean values ± SEM. **p b 0.01.

C. Cui et al. / Bone 59 (2014) 127–138

We next asked if adding soluble FXIIIA or TG2 into MC3T3-E1 cultures would increase bpFN assembly from media to matrix. For this, we added recombinant, exogenous and soluble FXIIIA (preactivated), or TG2, to cultures together with bpFN for 24 h and analyzed the bpFN left in the media. As seen in Fig. 8B, analysis showed significantly less bpFN left in media treated with 0.1 μg/ml FXIIIA (−32%) compared to control. No change was observed with exogenous TG2 addition confirming that FXIIIA is responsible for promoting the assembly of soluble pFN into osteoblast matrix. 4. Discussion Bone matrix contains abundant noncollagenous proteins which control the quality of mineral and the quality of the bone matrix [1]. pFN produced by hepatocytes is a major bone matrix component which has recently been shown to control overall biomechanical quality of

A

M+MDC

EDA-FN

Dansyl

EDA-FN

bone as well as the mineral-to-matrix ratio in vivo in mice [13]. Circulating pFN adsorbs into bone in a globular and soluble form where it undergoes fibrillogenesis to form detergent-soluble and insoluble matrix networks—it is estimated that 90% of the FN found in bone matrix derives from circulating pFN. Although it is generally considered that FN fibrillogenesis is a cell- and integrin-dependent process, it is not clear whether the large quantity of adsorbed pFN could be assembled in vivo in this manner to induce its insolubilization and fibrillogenesis in a bone environment where the cell-matrix ratio is very low. FN is a well-known substrate for TG enzymes; however, the role of this modification in bone has remained unknown. Our study demonstrates for the first time that pFN fibrillogenesis requires a TG enzyme and a TGmediated crosslinking step to form osteoblast matrix in vitro. The pFN assembly is orchestrated specifically by FXIIIA transglutaminase that is synthesized and secreted by osteoblasts which makes pFN assembly an active bone cell-mediated process. It also represents an alternative

C

DM+MDC

20 µm

135

bpFN / EDA-FN / merge

20 µm

Dansyl

10µm Merge

Merge

D

EDA-FN

DAPI

DAPI

IgG

M

DM +Ctrl

IP: Anti-dansyl WB: Anti-FN WB: Anti-EDA-FN Ck/m

DOC-insol

1.2

BrdU incorporation / 24 h

IgG M DM +Ctrl

Merge

10 µm

E

B

tot.FN

1.0 0.8 0.6

**

**

Anti EDA FN 1:200

Anti EDA FN 1:100

0.4 0.2 0.0

M

-pFN

Fig. 7. EDA-FN is not a TG substrate and regulates MC3T3-E1 osteoblast proliferation. (A) Immunofluoresence microscopy of MDC co-localization with EDA-FN in differentiating (DM) and control cells (M) (treated with medium only). Staining with anti-EDA-FN (green) and anti-dansyl (red) shows partial co-localization. (B) Immunoprecipitation of MDC-labeled proteins from cytoskeletal/matrix (Ck/m) fractions and from the DOC-insoluble fraction failed to pull down EDA-FN. Immunoprecipitation was only successful with anti-FN antibody that detects all FN variants. Normal rabbit IgG was used as negative control, and the original protein extract (before immunoprecipitation) was used as positive control (+Ctrl). (C) EDA-FN and bpFN formed mostly separate matrix networks as analyzed by immunofluorescence microscopy. bpFN is seen in red and EDA-FN in green. Partial co-localization (yellow) was observed. (D) Co-localization of EDA-FN (green) with total FN (red) shows that EDA-FN constitutes only a small portion of total FN matrix in osteoblasts. (E) EDA-FN, but not pFN, regulates MC3T3-E1 osteoblast proliferation. MC3T3-E1 osteoblasts were cultured for 20 h with the indicated treatments. Proliferation was assessed with BrdU labeling for 4 h followed by BrdU immunostaining. Blocking antibodies against EDA-FN (dilutions of 1:100 and 1:200) decreased cell proliferation to 59% and 51% of control, respectively. pFN-depleted media allowed normal cell proliferation. Data are presented as mean values ± SEM. ** p b 0.01. Student's t-test relative to control.

136

C. Cui et al. / Bone 59 (2014) 127–138

A

DM

DM+ F11QN

Liver / hepatocytes

DM Total +F11 CE

IP: Neutravidin WB: Anti-FN

pFN serum/media

WB: Anti-EDA-FN Bone

FXIIIA DM+ DM Total T26QN +T26 CE

DOC-soluble FN Osteoblast

DM IP: Neutravidin WB: Anti-FN WB: Anti-EDA-FN

B

FXIIIA (crosslinking)

DOC-insoluble (SDS-soluble) FN

Day 5 FN in media/day 0 ratio

FXIIIA (crosslinking)

1.4 1.2

Insoluble FN – COL I matrix

**

1.0

*

0.8

LOX

0.6

ALP

0.4 0.2

COL I scaffold

0.0

Day 0

M

DM

DM + FXIIIA

DM + TG2

DM + NC9

Fig. 8. Plasma FN is a specific substrate for FXIIIA transglutaminase. (A) FN is specifically labeled by FXIIIA-specific F11 substrate peptide. F11 (and negative control peptide F11QN), or TG2-specific substrate peptide T26 (and negative control peptide T26QN), were given to MC3T3-E1 osteoblasts for 24 h prior to protein extraction. F11, T26 and control peptides (all biotinylated) were pulled down with Neutravidin beads and analyzed by WB using anti-FN and anti-EDA-FN antibodies. The results show clear incorporation of biotinylated F11 into FN, but not into EDA-FN. No T26 peptide was found to be incorporated into FN or EDA-FN. (B) FXIIIA promotes pFN assembly. Pre-activated FXIIIA or TG2 (both 0.1 μg/ml) were added to differentiating osteoblasts (DM) together with bpFN at day 4 for 24 h. bpFN remaining in the media was assessed by ELISA assay. Exogenous FXIIIA addition results in significantly decreased bpFN levels in the media compared to control cells, demonstrating that it promotes its assembly into matrix. There was no significant difference in bpFN levels in media after TG2 incubation compared to control cells. NC9 was used as a positive control showing that inhibition of TG activity decreased bpFN assembly. Data are presented as mean values ± SEM. *p b 0.05, ** p b 0.01.

mechanism by which pFN can be “captured” and insolubilized into bone matrix by crosslinking. Furthermore, our data suggests that FXIIIA may initially only bind pFN to promote formation DOC-soluble matrix from the media and subsequently crosslink the network to further insolubilize it. The proposed sequence of events is summarized in Fig. 9. pFN appears to form a separate fibril network from cFN (EDA-FN), not only suggesting separate assembly mechanisms but also separate functions. In fact, as demonstrated by our work here, cFN (EDA-FN) appears to regulate preosteoblast proliferation whereas pFN, together with extracellular FXIIIA, appears to modulate the quality of COL I matrix elaborated by mature osteoblasts. Our conclusions on pFN's role in bone matrix quality are strongly supported by recent in vivo data from a liver-specific FN-knockout mouse [13], however, the role of cFN/EDA-FN in regards to osteoblast function in vitro appears not to be supported by what was observed in vivo. Our data shows a clear proliferation effect with EDA-FN inhibition. The mouse model lacking osteoblast FN (created with Col1a1-Cre) showed altered osteoblast numbers; however, the authors reported no change in osteoblast proliferation [13]. Our work here also shows for the first time that TGmediated modification is pFN-specific. Despite a number of attempts and several approaches, we were not able to see any MDC-labeled

Mineralization

Bone quality Fig. 9. Proposed scheme of how liver-derived plasma FN and FXIIIA transglutaminase mediate bone quality. Assembly of circulating pFN, synthesized by hepatocytes in the liver, is promoted by Factor XIIIA (FXIIIA) transglutaminase which is synthesized by osteoblasts. First, a FXIIIA-mediated assembly step to form DOC-soluble matrix does not require crosslinking and may be mediated by FXIIIA binding to pFN. This is followed by a second step that involves crosslinking that assists in the formation of DOC-insoluble pFN matrix. It is likely that FXIIIA also promotes pFN crosslinking to COL I as a third step occurring during matrix maturation. The FN-COL I network regulates both lysyl oxidase (LOX) levels in the matrix and alkaline phosphatase (ALP) levels produced by osteoblasts which together affect the overall quality and mineralization of the collagenous matrix.

EDA-FN in osteoblast cultures during differentiation. As EDA-FN contains the same Q residues as pFN, but an extra EDA domain located between FN repeats III11 and III12, it is possible that this extra region blocks TG-mediated crosslinking via hindering the FXIIIA–FN interaction, presenting an interesting mode to regulate FN crosslinking. Future studies are directed towards examining this possibility. The FN matrix has been shown to be important for the assembly of many matrix components, the most relevant to bone being COL I matrix. We show here that deleting pFN from cultures or inhibiting TG activity decreases COL I levels and affects the quality of COL I network laid down under these circumstances. The effects seen here can result from a lack of FN and/or a lack of COL I processing, as the potential mechanisms on how FN network can regulate COL I assembly include: i) FN acting as a scaffold for COL I deposition [6,7], ii) FN regulating BMP-1 activity to stimulate COL I processing and fibrillogenesis [19], iii) FN regulating LOX activity to increase COL I fibril stabilization [21]. Indeed, it is likely that multiple mechanisms are involved. Our work supports the concept that pFN matrix assembly regulates COL I processing since its inhibition by blocking TG activity results in decreased LOX levels, a discontinuous COL I network and thicker COL I fibrils. Furthermore, TG activity may also be required for FN-COL I crosslinking and/or COL I crosslinking alone. This is supported by studies that have shown COL I crosslinking to FN by FXIIIA in vitro [31]. It has been considered that FN/COL I crosslinks involve K residues from COL I and Q residues from FN [32]. Experiments where heat-denatured collagen was labeled with MDC showed that half of the COL I crosslinking residues were located within the triple helical region [54], suggesting that TG-mediated crosslinking

C. Cui et al. / Bone 59 (2014) 127–138

could be needed to ‘tighten’ COL I molecules within the fibrils—the observation that COL I fibrils in TG inhibited cultures were thicker supports this concept. The role of TG activity in ‘tightening’ the COL I network was also demonstrated in in vitro-generated COL I scaffolds that were crosslinked with microbial TG [55]. We have shown previously both in vitro and in vivo that osteoblasts express two TG enzymes—TG2 and FXIIIA [34,37,38]—of which FXIIIA was shown to be the prominent crosslinking enzyme during osteoblast differentiation. TG2 was demonstrated to remain inactive on the cell surface during the matrix assembly phase, but released to the matrix as a smaller fragment at a later phase of differentiation, likely contributing to matrix mineralization [56,57]. Here, we provide further evidence using isozyme-specific substrate probes to demonstrate that FXIIIA is an important regulator of matrix assembly and is capable of crosslinking pFN in osteoblast cultures. Indeed, our previous results also show that FXIIIA gene expression and protein production in osteoblasts reacts immediately to ascorbic acid treatment that promotes COL I synthesis in cells. FXIIIA in osteoblasts is translocated onto the osteoblast surface as well as being secreted into the extracellular compartment in the presence of COL I in osteoblasts [39]. Based on the results our studies, we suggest that FXIIIA is an important part of the COL I assembly machinery and assists COL I deposition via promoting formation of a pFN scaffold and likely via crosslinking of COL I fibrils. The role of FXIIIA in COL I assembly is supported by the observed COL I defect after induced myocardial infarction in F13a1−/− mice where the defect site displayed less COL I deposition during healing and remodeling. This defect was not rescued by exogenous plasma FXIIIA therapy suggesting that FXIIIA produced locally by tissue-resident cells is responsible for the modulation of COL I levels [58]. The fact that circulating pFN is the main substrate for FXIIIA in osteoblast cultures, and thus also likely in bone in vivo, and that exogenous FXIIIA increases pFN assembly, presents the possibility that also circulating plasma FXIIIA (from FXIIIA2B2) could adsorb to bone and contribute to bone quality. However, we have not been able to detect FXIIIB in bone matrix as indicator that circulating heterotetrameric FXIII would be present (data not shown). In our cell culture experiments here, circulating FXIIIA is not involved in pFN stabilization as cell culture serum should not contain any coagulation factors. As outlined above, pFN is synthesized by hepatocytes and secreted into the circulation at high levels, resulting in a 300 μg/ml concentration in humans and close to 600 μg/ml in mice [59,60]. Patients suffering from chronic liver disease such as chronic persistent hepatitis, chronic active hepatitis and liver cirrhosis show significantly decreased plasma FN levels [61] as well as bone pain and bone loss. Indeed, osteoporosis affects up to 53% of patients with liver cirrhosis [62,63]. Although the etiological links between liver disease and osteoporosis are no doubt complex, it is possible that pFN represents one important factor influencing disease progression [62]. In conclusion, our results show that liver-derived pFN, which is a major component of bone matrix, requires FXIIIA transglutaminase activity for its assembly into a fibrillar network in osteoblast cultures. A lack of TG activity or lack of pFN in osteoblast matrix leads to decreased COL I deposition, a defective COL I network, reduced LOX levels, decreased alkaline phosphatase activity and mineralization in vitro, which are all major contributors to bone quality (Fig. 9). Our results bring further evidence towards the role of FXIIIA in the function of FN matrix formation in bone, and emphasize the importance of liver– bone organ crosstalk for maintaining bone health. Conflict of interest The authors declare that there are no conflict of interests. Acknowledgments We would like to thank Mrs. Aisha Mousa for her assistance in all phases of this study, and Dr. Marc McKee and Ms. Line Mongeon for

137

assistance with the SEM experiments. This study was supported by grants to MTK from the Canadian Institutes of Health Research (CIHR) (MOP-119403), and the CIHR Institute for Musculoskeletal Health and Arthritis (IMHA) (IMH-89827). KH was supported by a Grant-in-Aid for Scientific Research (B) (No. 23380200) from JSPS, Japan (KH). CC and SW were supported by stipends from China Scholarship Council, SW and VDM received support from Faculty of Dentistry at McGill University. VDM received a stipend from CIHR Systems Biology Training Grant.

References [1] McKee MD, Addison WN, Kaartinen MT. Hierarchies of extracellular matrix and mineral organization in bone of the craniofacial complex and skeleton. Cells Tissues Organs 2005;181:176–88. [2] Turner CH. Bone strength: current concepts. Ann N Y Acad Sci 2006;1068:429–46. [3] Young M. Bone matrix proteins: their function, regulation, and relationship to osteoporosis. Osteoporos Int 2003;14:35–42. [4] Moursi AM, Damsky CH, Lull J, Zimmerman D, Doty SB, Aota S, et al. Fibronectin regulates calvarial osteoblast differentiation. J Cell Sci 1996;109:1369–80. [5] Moursi AM, Globus RK, Damsky CH. Interactions between integrin receptors and fibronectin are required for calvarial osteoblast differentiation in vitro. J Cell Sci 1997;110:2187–96. [6] Sottile J, Hocking DC. Fibronectin polymerization regulates the composition and stability of extracellular matrix fibrils and cell-matrix adhesions. Mol Biol Cell 2002;13:3546–59. [7] Sottile J, Shi F, Rublyevska I, Chiang H-Y, Lust J, Chandler J. Fibronectin-dependent collagen I deposition modulates the cell response to fibronectin. Am J Physiol Cell Physiol 2007;293:C1934–46. [8] Pankov R, Yamada KM. Fibronectin at a glance. J Cell Sci 2002;115:3861–3. [9] Wierzbicka-Patynowski I, Schwarzbauer JE. The ins and outs of fibronectin matrix assembly. J Cell Sci 2003;116:3269–76. [10] Mosher D. Fibronectin: Access Online via Elsevier; 1989. [11] Magnusson MK, Mosher DF. Fibronectin: structure, assembly, and cardiovascular implications. Arterioscler Thromb Vasc Biol 1998;18:1363–70. [12] Hynes RO. Fibronectins. New York: Springer-Verlag; 1990. [13] Bentmann A, Kawelke N, Moss D, Zentgraf H, Bala Y, Berger I, et al. Circulating fibronectin affects bone matrix, whereas osteoblast fibronectin modulates osteoblast function. J Bone Miner Res 2010;25:706–15. [14] Sabatier L, Chen D, Fagotto-Kaufmann C, Hubmacher D, McKee MD, Annis DS, et al. Fibrillin assembly requires fibronectin. Mol Biol Cell 2009;20:846–58. [15] Malara A, Gruppi C, Rebuzzini P, Visai L, Perotti C, Moratti R, et al. Megakaryocyte– matrix interaction within bone marrow: new roles for fibronectin and factor XIIIA. Blood 2011;117:2476–83. [16] Pereira M, Rybarczyk BJ, Odrljin TM, Hocking DC, Sottile J, Simpson-Haidaris PJ. The incorporation of fibrinogen into extracellular matrix is dependent on active assembly of a fibronectin matrix. J Cell Sci 2002;115:609–17. [17] Godyna S, Mann DM, Argraves WS. A quantitative analysis of the incorporation of fibulin-1 into extracellular matrix indicates that fibronectin assembly is required. Matrix Biol 1995;14:467–77. [18] Chung CY, Zardi L, Erickson HP. Binding of tenascin-C to soluble fibronectin and matrix fibrils. J Biol Chem 1995;270:29012–7. [19] Huang G, Zhang Y, Kim B, Ge G, Annis DS, Mosher DF, et al. Fibronectin binds and enhances the activity of bone morphogenetic protein 1. J Biol Chem 2009;284:25879–88. [20] Dallas SL, Sivakumar P, Jones CJP, Chen Q, Peters DM, Mosher DF, et al. Fibronectin regulates latent transforming growth factor-β (TGFβ) by controlling matrix assembly of latent TGFβ-binding protein-1. J Biol Chem 2005;280:18871–80. [21] Fogelgren B, Polgár N, Szauter KM, Újfaludi Z, Laczkó R, Fong KSK, et al. Cellular fibronectin binds to lysyl oxidase with high affinity and is critical for its proteolytic activation. J Biol Chem 2005;280:24690–7. [22] Lorand L, Graham RM. Transglutaminases: crosslinking enzymes with pleiotropic functions. Nat Rev Mol Cell Biol 2003;4:140–56. [23] Folk JE. Transglutaminases. Annu Rev Biochem 1980;49:517–31. [24] Iismaa SE, Mearns BM, Lorand L, Graham RM. Transglutaminases and disease: lessons from genetically engineered mouse models and inherited disorders. Physiol Rev 2009;89:991–1023. [25] Kaartinen MT, Pirhonen A, Linnala-Kankkunen A, Mäenpää PH. Cross-linking of osteopontin by tissue transglutaminase increases its collagen binding properties. J Biol Chem 1999;274:1729–35. [26] Forsprecher J, Wang Z, Nelea V, Kaartinen MT. Enhanced osteoblast adhesion on transglutaminase 2-crosslinked fibronectin. Amino Acids 2009;36:747–53. [27] Forsprecher J, Wang Z, Goldberg HA, Kaartinen MT. Transglutaminase-mediated oligomerization promotes osteoblast adhesive properties of osteopontin and bone sialoprotein. Cell Adh Migr 2011;5:65–72. [28] Nelea V, Nakano Y, Kaartinen M. Size distribution and molecular associations of plasma fibronectin and fibronectin crosslinked by transglutaminase 2. Protein J 2008;27:223–33. [29] Mosher DF. Cross-linking of cold-insoluble globulin by fibrin-stabilizing factor. J Biol Chem 1975;250:6614–21. [30] Keski-Oja JF, Mosher D, Vaheri A. Cross-linking of a major fibroblast surfaceassociated glycoprotein (fibronectin) catalyzed by blood coagulation factor XIII. Cell 1976;9:29–35.

138

C. Cui et al. / Bone 59 (2014) 127–138

[31] Mosher DF, Schad PE. Cross-linking of fibronectin to collagen by blood coagulation Factor XIIIa. J Clin Invest 1979;64:781–7. [32] Mosher DF, Schad PE, Vann JM. Cross-linking of collagen and fibronectin by factor XIIIa. Localization of participating glutaminyl residues to a tryptic fragment of fibronectin. J Biol Chem 1980;255:1181–8. [33] Hoffmann BR, Annis DS, Mosher DF. Reactivity of the N-terminal region of fibronectin protein to transglutaminase 2 and Factor XIIIA. J Biol Chem 2011;286:32220–30. [34] Al-Jallad HF, Nakano Y, Chen JLY, McMillan E, Lefebvre C, Kaartinen MT. Transglutaminase activity regulates osteoblast differentiation and matrix mineralization in MC3T3-E1 osteoblast cultures. Matrix Biol 2006;25:135–48. [35] Al-Jallad HF, Myneni VD, Piercy-Kotb SA, Chabot N, Mulani A, Keillor JW, et al. Plasma membrane Factor XIIIA transglutaminase activity regulates osteoblast matrix secretion and deposition by affecting microtubule dynamics. PLoS ONE 2011;6: e15893. [36] Kaartinen MT, El-Maadawy S, Räsänen NH, McKee MD. Tissue transglutaminase and its substrates in bone. J Bone Miner Res 2002;17:2161–73. [37] Nurminskaya M, Kaartinen MT. Transglutaminases in mineralized tissues. Front Biosci 2006;11:1591–606. [38] Nakano Y, Al-Jallad HF, Mousa A, Kaartinen MT. Expression and localization of plasma transglutaminase Factor XIIIA in bone. J Histochem Cytochem 2007;55:675–85. [39] Piercy-Kotb SA, Mousa A, Al-Jallad HF, Myneni VD, Chicatun F, Nazhat SN, et al. Factor XIIIA transglutaminase expression and secretion by osteoblasts is regulated by extracellular matrix collagen and the MAP kinase signaling pathway. J Cell Physiol 2012;227:2936–46. [40] Tullberg-Reinert H, Jundt G. In situ measurement of collagen synthesis by human bone cells with a Sirius Red-based colorimetric microassay: effects of transforming growth factor β2 and ascorbic acid 2-phosphate. Histochem Cell Biol 1999;112:271–6. [41] Keillor JW, Chica RA, Chabot N, Vinci V, Pardin C, Fortin E, et al. The bioorganic chemistry of transglutaminase—from mechanism to inhibition and engineering. Can J Chem 2008;86:271–6. [42] Anjos-Afonso F, Bonnet D. Isolation, culture, and differentiation potential of mouse marrow stromal cells. Current Protocols in Stem Cell Biology. John Wiley & Sons, Inc.; 2007 [43] Pankov R, Yamada KM. Non-radioactive quantification of fibronectin matrix assembly. Current Protocols in Cell Biology. John Wiley & Sons, Inc.; 2001 [44] Pankov R, Momchilova A. Fluorescent labeling techniques for investigation of fibronectin fibrillogenesis (labeling fibronectin fibrillogenesis). In: Even-Ram S, Artym V, editors. Extracellular matrix protocols. Humana Press; 2009. p. 261–74. [45] Velling T, Risteli J, Wennerberg K, Mosher DF, Johansson S. Polymerization of type I and III collagens is dependent on fibronectin and enhanced by integrins α11β1and α2β1. J Biol Chem 2002;277:37377–81. [46] Corbett SA, Lee L, Wilson CL, Schwarzbauer JE. Covalent cross-linking of fibronectin to fibrin is required for maximal cell adhesion to a fibronectin–fibrin matrix. J Biol Chem 1997;272:24999–5005. [47] Shi F, Harman J, Fujiwara K, Sottile J. Collagen I matrix turnover is regulated by fibronectin polymerization. Am J Physiol Cell Physiol 2010;298:C1265–75.

[48] Manabe R-i, Oh-e N, Sekiguchi K. Alternatively spliced EDA segment regulates fibronectin-dependent cell cycle progression and mitogenic signal transduction. J Biol Chem 1999;274:5919–24. [49] Sugimura Y, Hosono M, Wada F, Yoshimura T, Maki M, Hitomi K. Screening for the preferred substrate sequence of transglutaminase using a phage-displayed peptide library: identification of peptide substrates for TGase 2 and Factor XIIIA. J Biol Chem 2006;281:17699–706. [50] Hitomi K, Kitamura M, Sugimura Y. Preferred substrate sequences for transglutaminase 2: screening using a phage-displayed peptide library. Amino Acids 2009;36:619–24. [51] Watanabe K, Tsunoda K, Itoh M, Fukui M, Mori H, Hitomi K. Transglutaminase 2 and Factor XIII catalyze distinct substrates in differentiating osteoblastic cell line: utility of highly reactive substrate peptides. Amino Acids 2013;44:209–14. [52] Hitomi K, Kitamura M, Alea MP, Ceylan I, Thomas V, El Alaoui S. A specific colorimetric assay for measuring transglutaminase 1 and factor XIII activities. Anal Biochem 2009;394:281–3. [53] Itoh M, Kawamoto T, Tatsukawa H, Kojima S, Yamanishi K, Hitomi K. In situ detection of active transglutaminases for keratinocyte type (TGase 1) and tissue type (TGase 2) using fluorescence-labeled highly reactive substrate peptides. J Histochem Cytochem 2011;59:180–7. [54] Stachel I, Schwarzenbolz U, Henle T, Meyer M. Cross-linking of type I collagen with microbial transglutaminase: identification of cross-linking sites. Biomacromolecules 2010;11:698–705. [55] Garcia Y, Collighan R, Griffin M, Pandit A. Assessment of cell viability in a threedimensional enzymatically cross-linked collagen scaffold. J Mater Sci Mater Med 2007;18:1991–2001. [56] Nakano Y, Addison WN, Kaartinen MT. ATP-mediated mineralization of MC3T3-E1 osteoblast cultures. Bone 2007;41:549–61. [57] Nakano Y, Forsprecher J, Kaartinen MT. Regulation of ATPase activity of transglutaminase 2 by MT1-MMP: implications for mineralization of MC3T3-E1 osteoblast cultures. J Cell Physiol 2010;223:260–9. [58] Nahrendorf M, Hu K, Frantz S, Jaffer FA, Tung C-H, Hiller K-H, et al. Factor XIII deficiency causes cardiac rupture, impairs wound healing, and aggravates cardiac remodeling in mice with myocardial infarction. Circulation 2006;113:1196–202. [59] Ejim OS, Fonseca V, Coumar A, Mathur S, Bell JL, Dandona P. Fibronectin concentrations in plasma in peripheral vascular disease. Clin Chem 1988;34:2426–9. [60] George EL, Georges-Labouesse EN, Patel-King RS, Rayburn H, Hynes RO. Defects in mesoderm, neural tube and vascular development in mouse embryos lacking fibronectin. Development 1993;119:1079–91. [61] Soresi M, Di Martino D, Montalto G, Carroccio A, Ruggeri MI, Bascone F, et al. Plasma fibronectin in chronic liver diseases. Recenti Prog Med 1993;84:602–7. [62] Nakchbandi IA, van der Merwe SW. Current understanding of osteoporosis associated with liver disease. Nat Rev Gastroenterol Hepatol 2009;6:660–70. [63] Gallego-Rojo FJ, Gonzalez-Calvin JL, Munoz-Torres M, Mundi JL, Fernandez-Perez R, Rodrigo-Moreno D. Bone mineral density, serum insulin-like growth factor I, and bone turnover markers in viral cirrhosis. Hepatology 1998;28:695–9.