Asporin-deficient mice have tougher skin and altered skin glycosaminoglycan content and structure

The main structural component of connective tissues is fibrillar, cross-linked collagen whose fibrillogenesis can be modulated by Small Leucine-Rich Proteins/Proteoglycans (SLRPs). Not all SLRPs’ effects on collagen and extracellular matrix in vivo have been elucidated; one of the less investigated SLRPs is asporin. Here we describe the successful generation of an Aspn-/- mouse model and the investigation of the Aspn-/- skin phenotype. Functionally, Aspn-/- mice had an increased skin mechanical toughness, although there were no structural changes present on histology or immunohistochemistry. Electron microscopy analyses showed 7% thinner collagen fibrils in Aspn-/- mice (not statistically significant). Several matrix genes were upregulated, including collagens (Col1a1, Col1a2, Col3a1), matrix metalloproteinases (Mmp2, Mmp3) and lysyl oxidases (Lox, Loxl2), while lysyl hydroxylase (Plod2) was downregulated. Intriguingly no differences were observed in collagen protein content or in collagen cross-linking-related lysine oxidation or hydroxylation. The glycosaminoglycan content and structure in Aspn-/- skin was profoundly altered: chondroitin/dermatan sulfate was more than doubled and had an altered composition, while heparan sulfate was halved and had a decreased sulfation. Also, decorin and biglycan were doubled in Aspn-/- skin. Overall, asporin deficiency changes skin glycosaminoglycan composition, and decorin and biglycan content, which may explain the changes in skin mechanical properties.

In this paper we hypothesized that asporin functions in formation of a structurally coherent extracellular matrix in vivo. To test this hypothesis we generated an asporin knockout mouse model (Aspn -/-). We focused on analyzing the effect of asporin deficiency on skin extracellular matrix, where asporin has been detected [21,22] and where we, in the current study, observed a biomechanical phenotype. To elucidate the underlying cause of this phenotype we also analyzed skin collagen, proteoglycans and GAGs. This report contributes to the overall understanding of the unique functions of SLRPs in connective tissues.

Materials and methods
Targeted disruption of the mouse Aspn gene All animal experiments were conducted under approval of the regional animal ethical board committee (Malmö-Lunds djurförsöksetiska nämnd, Jordbruksverket, Sweden). Mice were euthanized by carbon dioxide followed by cervical dislocation. The BAC clone bMQ103e19 containing the 129/Sv mouse Aspn genomic material was obtained from Geneservice (UK). To construct the targeting vector a recombineering system described in [33] was used. The resulting targeting vector contained exons 2-5 (out of 8 present in the gene), where exons 2-3 were flanked with loxP sites, and the neo cassette, flanked with FRT sites, was placed upstreams of the second loxP site. Targeting vector also contained MCI-TK cassette for positive selection of the transformed 129/Sv ES cells that were screened for genomic integration of the vector with Southern blotting and PCR. The positive clone was injected into C57BL/6J blastocysts to generate chimeric mice that were later intercrossed to generate Aspn fneo/+ heterozygous animals. They were then crossed with FLPeR mice (Jackson Laboratory, Bar Harbor, ME) to remove the neo cassette and give Aspn f/+ mice. Finally, they were crossed with EIIa-Cre transgenic mice (Jackson Laboratory, Bar Harbor, ME) to remove exons 2-3 of the targeted Aspn allele to create Aspn -/+ mice ( Fig 1A). Primer sequences in

Skin histology and immunohistochemistry
Mouse skins were fixed in HistoChoice (Sigma) for 24 h, dehydrated, embedded in paraffin and sectioned. Sections were stained with Masson Trichrome (Sigma) or processed for immunohistochemistry: rehydrated sections were washed 2 x 5 min in Tris-buffered saline (TBS) plus 0.025% Triton X-100 and blocked with 10% goat serum and 1% bovine serum albumin (BSA) for 2 h. Sections were then incubated for 16 h at 4˚C with rat anti-mouse CD31 (Optistain, clone SZ31) or rat anti-mouse F4/80 (BioRad, clone CI:A3-1) diluted to 10 μg/mL in 1% BSA in TBS. Sections were rinsed 2 x 5 min with TBS plus 0.025% Triton X-100 and quenched with 0.3% hydrogen peroxide in TBS for 15 min. After rinsing with TBS, sections were stained with ultrasensitive ABC staining kit (ThermoFisher). Sections were washed 3 x 5 min with TBS and developed using diaminobenzidine peroxidase substrate (Vector Labs).

Immunoblots
Dorsal skin samples were boiled in SDS-PAGE loading buffer for 10 min. The proteins were run on 4-12% Bis-Tris reducing SDS-PAGE (ThermoFisher) and transferred to a nitrocellulose membrane. The membrane was blocked with 5% BSA in TBS, rinsed in wash buffer (TBS with 0.1% Tween-20) and immunoblotted in the incubation buffer (TBS with 0.1% Tween-20 and 1% BSA) using the following primary antibodies diluted to 1 μg/mL: anti-asporin (Abcam, ab31303), anti-decorin (kind gift from Professor Åke Oldberg, Lund University), anti-biglycan (LF106, kind gift from Professor Larry Fisher, NIH), anti-beta-actin (ab6276, Abcam). Membranes were washed 3 x 5 min and incubated with the corresponding secondary HRP-conjugated antibodies (all from GE Healthcare) diluted to 0.1 μg/mL for 1 h. Membranes were washed 3 x 5 min and developed with Luminata Forte (Millipore) using a CCD camera.

Electron microscopy
Dorsal skin samples were shaved and processed for transmission electron microscopy as described [34]. Collagen fibril thickness was quantified using images from three reticular dermis regions per mouse, at 50,000x magnification, using Adobe Photoshop.

Skin tensile strength determination
Mice were euthanized and kept on ice before shaving and gently removing the dorsal skin samples. Two parallel strips of skin (3 mm x 30 mm) were punched out in the craniocaudal direction, symmetrically on the left and right side of the spine. The samples were washed briefly in PBS, the central 5 mm were wrapped in PBS soaked gauze and the ends were allowed to dry for 90 min while keeping the central part moist with PBS. The dry ends were glued onto clamps using cyanoacrylate glue and after curing for 40 min the sample was placed in a PBS bath and mechanically tested (200-N tensile stage, petri dish version, Deben, Suffolk, UK). The test consisted of 4 preconditioning cycles to 15% strain, followed finally by a stretch to failure. All tests were at a rate of 6 mm/min with data recorded at 10 Hz. Sample width and length were measured by microscopy images. To avoid mechanically damaging the strips, thickness was measured with a constant-pressure caliper (model 293-334-30, Mitutoyo, Kawasaki, Japan) on a piece of skin adjacent to the tested strip. The failure test was analyzed for strain, stress and energy at both the yield point and at failure. The peak modulus was determined by a moving fit over a 10% strain range. Yield was defined as the first point where the modulus dropped below 50% of the peak value. For each animal, results for the two strips were averaged and groups were compared by unpaired two-tailed t-test with unequal variance.

Collagen molecular phenotype (mass spectrometry)
Dorsal skin collagen was defatted with chloroform/methanol (3:1 v/v). Collagen was solubilized by 3% acetic acid extraction at 4˚C for 24 hrs or heat denaturation in SDS-PAGE sample buffer. The extracts were run on 6% SDS-PAGE and stained with Coomassie Blue. Bands corresponding to collagen α1(I) and α2(I) chains were cut out and processed for mass spectrometry using trypsin in-gel digestion, without reduction or alkylation. Collagen peptides were analyzed with LC-MS using an LTQ XL ion trap mass spectrometer (ThermoFisher) equipped with in-line liquid chromatography using a C4 5um capillary column (300um x 150mm; Higgins Analytical RS-15M3-W045) eluted at 4.5ul min. The LC mobile phase consisted of buffer A (0.1% formic acid in MilliQ water) and buffer B (0.1% formic acid in 3:1 acetonitrile:n-propanol v/v). The LC sample stream was introduced into the mass spectrometer by electrospray ionization (ESI) with a spray voltage of 4kV. Proteome Discoverer search software (Thermo Scientific) was used for peptide identification using the NCBI protein database. Proline and lysine modifications were examined manually by scrolling or averaging the full scan over several minutes so that all of the post-translational variations of a given peptide appeared together in the full scan.

Hydroxyproline determination
Dorsal skin samples were weighed and hydrolyzed in 6 M HCl for 4 hours at 120˚C at 2 atm. Hydroxyproline was quantified as described earlier [35] and normalized to total tissue weight.

Reverse transcriptase-and real-time qPCR
Total RNA was isolated from skin mice using TRIzol reagent (ThermoFisher). 500 ng RNA was used for reverse transcription reaction using Superscript VILO (ThermoFisher). Real-time qPCR was conducted using Taqman probes and Taqman Gene Expression Master Mix (Ther-moFisher) in an Applied Biosystems 7300 qPCR equipment. Gene expression was normalized to Actb transcripts. The Taqman probe list is presented in S1 Table. Glycosaminoglycan analysis Skins were lyophilized, weighed, and processed for GAG analyses as described in [34]. The proteins were degraded for 16 h at 55˚C with 200 μg/ml pronase in 50 mm Tris/HCl, pH 8, 1 mM CaCl 2 , and 1% Triton X-100. After heat inactivation of the protease, MgCl 2 was added to 2 mM and benzonase was added to a final 20 sigma units/ml and incubated for 2 h at 37˚C. DNAase was heat-inactivated, NaCl was added to a final 0.1 M, and GAGs were purified by batch incubation with a DE52 anion exchange gel, followed by the transfer of the gel to disposable spin centrifuge tubes (Costar 8163). The gel was washed with buffer 1 (50 mM Tris/HCl, pH 8, 0.1 M NaCl, and 0.1% Triton X-100), buffer 2 (50 mM NaAc, pH 4, 0.1 M NaCl, and 0.1% Triton X-100), water, and 0.1 M NH 4 HCO 3 . Elution was achieved by 2 M NH 4 HCO 3 . After lyophilization, GAGs were provisionally and roughly estimated by carbazole [36]. Then, 500 ng GAGs were subjected to chondroitinase ABC (Sigma) overnight at 37˚C in 20 μl 50 mM NH 4 OAc and 0.1 mg/ml BSA, containing 10 mIU enzyme, or chondroitinase B (R&D Systems) overnight at 37˚C in 20 μl 50 mM Tris pH 7.5, 50 mM NaCl, 4 mM CaCl 2 , and 0.1 mg/ml BSA, containing 5 mIU enzyme, or with a mixture of heparinases (in-house preparation, purified from E.coli, stably singularly transfected with the pET-15b vector containing heparinase I, or vector pET-19b containing heparinase II or III, as provided by Professor Jian Liu (University of North Carolina; incubation overnight at 37˚C in 20 μl 40 mM NaAc pH 7.0, 2 mM CaAc, 0.1 mg/ml BSA, containing 2 mIU of each enzyme). The samples were boiled and the supernatant was dried. Fluorophore-labelling of the resulting disaccharides was performed by adding 10 μl of 20 mM re-purified 2-aminoacridone (AMAC, Sigma), followed by a 20 min incubation at room temperature before the addition of 10 μl of 1 M NaBH 3 CN and incubation at 45˚C for 16 h [37]. Pre-column AMAC-labeled disaccharides were analyzed with HPLCfluorescence as described previously, with slight modifications [37]. Briefly, 20 μl samples were diluted to 100 μl in running buffer (98% A: NH 4 OAc, 60 mM, pH 5.6, and 2% B: MeCN) and injected (2 μl) onto an XBridge BEH Shield RP18 (2.1x100 mm, 2.5 μm). Disaccharides were separated using a 39 min gradient run at 0.35 ml/min (0-1 min: 98% A, 1-3 min: 98-96% A, 3-26 min: 96-85% A, 26-28 min: 85-10% A, 28-32 min: 10% A, 32-34 min: 10-98% A, 34-39 min: 98% A) on a Thermo Scientific UltiMate 3000 Quaternary Analytical system with an FLD-3400RS fluorescence detector (excitation λ = 428 and emission λ = 525). The column was kept at 30˚C to improve performance and reproducibility. Quantification was done by comparison to known weight of standard disaccharides (Iduron, UK), mock-treated in the same buffers and enzymes as the samples in each series of runs.

Semi-quantitative analysis of decorin and biglycan protein cores
Proteins were extracted from two-month old mice skin (n = 4 per genotype) in a dissociating buffer (4 M guanidine chloride, 50 mM NaAc pH 5.5, 0.1% Triton X-100, complete protease inhibitor cocktail (Roche)) by consecutive use of a Polytron-and Potter-type homogenizers. After overnight incubation the extracts were spun down and the supernatant exchanged against a denaturating buffer (8 M urea, 50 mM NaAc pH 5.5, 0,2 M NaCl, 0.1% Triton X-100). Proteins were quantified by Bradford method (BioRad) and equal amount of proteins were used in anion-exchange purification of proteoglycans on DE-52 (Whatman), eluted by of 2M ammonium bicarbonate, 0.1% Triton X-100. The recovery from the DE-52 purification was assessed by co-incubation of 35 S-labeled GAGs, which were protease-treated, with the skin extracts. The recovery of the labeled internal control was very similar among the eight samples (coefficient of variation = 11%). SDS-PAGE electrophoresis was done using stain-free gels (BioRad) and total proteins were visualized directly using the Chemi Doc Touch Imaging System (BioRad). For immunoblotting the samples were digested with chondroitinase ABC (as described in Glycosaminoglycan Analysis). Immunoblots were performed as described above.

Generation of Aspn -/mice
Our gene-targeting construct was designed to remove exons 2-3 in the Aspn gene where exon 2 contains the translation start site and 23 bp of upstream promoter sequence. First, we flanked exons 2-3 with two direct repeats of loxP sites, where the second loxP site was preceded by a neo cassette flanked with FRT sites. The gene-targeted mice were then bred with FLPeR mice in order to remove the neo cassette, and then with EIIa-Cre mice to remove exons 2-3 ( Fig  1A). The resulting Aspn +/mice were backcrossed with C57BL/6 mice for ten generations. When intercrossed, Aspn +/mice generated Aspn +/+ , Aspn +/and Aspn -/pups at normal Mendelian ratio. Excision of exons 2 and 3 was verified by genotyping (Fig 1B). No full size or fragments of asporin protein were present in Aspn -/mice, confirming the functional ablation of the asporin gene ( Fig 1C). All mice had normal fertility and external appearance. Five Aspn -/mice were kept until 15 months of age and showed normal behavior. Histological and necroscopical analyses of 2 month-old Aspn -/mice showed no gross abnormities in aorta, esophagus, intestine, kidney, liver, skeletal muscle, spleen and trachea.

Skin mechanical testing
Next, the general mechanical properties of Aspn -/skin were assessed. In four-month old female Aspn -/mice the skin toughness, or energy required to achieve failure, was almost 75% greater than in wild-type mice (Table 1). Other mechanical parameters, including failure stress and strain, were not significantly different but also tended toward superior strength of the Aspn -/skin. Note that the parameters at yield are almost identical in both genotypes, indicating that most of the difference occurred after the yield point.

Skin collagen molecular phenotype
Because asporin is a collagen-binding protein we investigated if asporin deficiency could result in a changed molecular phenotype of collagen. Two-month old skin collagen was extracted with SDS or acetic acid and the collagen band pattern on Coomassie-stained SDS-PAGE was analyzed. We did not observe any differences between Aspn -/and wild-type collagen (S1 Fig). We also used mass spectrometry to assess lysyl oxidase-mediated and lysyl hydroxylase-mediated modifications of collagen lysines involved in cross-linking. We could not detect any differences in the oxidation status of α1(I) chain C-telopeptidal lysine, and no differences in the hydroxylation status of α1(I) helical Lys-87, or in α2(I) helical Lys-90 (S1 Fig).

Semi-quantitative assessment of decorin and biglycan
To assess the quantities of skin proteoglycans we extracted proteins from dorsal skin samples (n = 4 per genotype) and applied similar total protein quantities ( Fig 4A) to anion-exchange. The eluted total proteoglycans (Fig 4B) appeared twofold more abundant in Aspn -/skin. To verify our findings and to identify the different proteoglycans we digested the samples with chondroitinase ABC and immunoblotted for decorin and biglycan. Aspn -/skin contained approximately twofold more decorin (Fig 4C) and twofold more biglycan (Fig 4D).

Discussion
Collagens can be arranged and cross-linked in different ways, which is apparent from the variety of collagen fibril phenotypes in different tissues and organs. The question of mechanisms behind these processes remains intriguing, but several seemingly tissue-specific modulators of collagen fibrillogenesis have been identified in vitro and in vivo, especially among the SLRP family of proteins [1,2]. In order to examine the role of the less investigated SLRPs we have characterized the in vivo extracellular matrix phenotype in a novel, SLRP-deficient, Aspn -/mouse. In earlier studies asporin gene polymorphism has been associated with human knee osteoarthritis and lumbar disc degeneration [20,40]; we were therefore intrigued that Aspn -/- mice did not show any apparent skeletal abnormities, even at histological examination. We suggest that the polymorphisms described in the previous papers [20,40] have a different effect Equal protein quantities were applied on anion-exchange beads and proteoglycans were eluted and imaged on gel (B); Alb is albumin control. Next, proteoglycan samples were digested with chondroitinase ABC and immunoblotted for decorin (C) and biglycan (D) to assess the quantities of the respective protein cores; ND is non-chondroitinase ABC-digested sample control (not visible in biglycan samples probably due to low quantity and smeary distribution).
https://doi.org/10.1371/journal.pone.0184028.g004 than total gene deletion. More studies are needed to elucidate the phenotypic impact of the different ASPN gene mutations.
In this paper we focused on skin phenotype-this is mainly because we, in this study, and other groups earlier [21,22], detected asporin in dermis, and because we observed no abnormities in other major organs in Aspn -/mice. In addition, we observed a changed biomechanical phenotype in Aspn -/skin, which prompted the skin ECM analyses conducted in this study.
The skin extracellular matrix in Aspn -/and wild-type mice was comparable in histological examinations, i.e. the relative amount of dermis and hypodermis, and the amount of blood vessels appeared similar in both genotypes. We observed no signs of inflammation, judging from the unaltered macrophage amounts.
Since asporin interacts with collagen and inhibits its fibrillogenesis in vitro [17] we conducted a detailed investigation of the molecular collagen phenotype in Aspn -/skin. Collagen fibrils examined in an electron microscope were 7% thinner in Aspn -/skin; this change, however, was not statistically significant and the fibrils were morphologically regular, in contrast to the fibril phenotypes reported in other SLRP-deficient mice, e.g. Dcn -/and Lum -/- [6,9]. Furthermore, the collagen band pattern on Coomassie-stained SDS-PAGE appeared normal, suggesting no defects in collagen post-translational modifications, e.g. proline hydroxylation. Despite a decrease in the amount of lysyl oxidase, the collagen α1(I)-and α2(I)-chain lysines involved in cross-linking were not modified differently than in wild-type collagen, in contrast to the phenotype reported in Fmod -/mice [3]. Summarizing the collagen phenotype studies, we could detect only a small, statistically non-significant, difference in the fibril diameter in the Aspn -/skin collagen. It is possible that the homologue decorin, which we found approximately doubled in Aspn -/skin, binding to the same site on collagen as asporin [17] can compensate for the asporin deficiency, thus ensuring an essentially normal collagen fibril phenotype.
Despite the almost normal Aspn -/skin collagen fibril phenotype we observed a robust transcript increase of collagens (Col1a1, Col1a2, Col3a1), collagen secretion-related Serpinh1 and Mia3, collagen cross-linking-related lysyl oxidases (Lox, Loxl2), the collagen fibrillogenesismodulating SLRP Lum, and the collagenases Mmp2 and Mmp3. Altogether, the expression data could be explained by an overall augmented ECM turnover, whose effect, however, on the overall ECM structure observed by histology and electron microscopy appears minute.
While we did not detect any major changes in the overall skin and ECM structure or in collagen molecular phenotype we observed a major change in GAG content and structure. In the ECM a major component are chondroitin/dermatan (CS/DS) proteoglycans versican, decorin and biglycan. In Aspn -/skin, CS/DS was more than doubled and the overall CS/DS sulfation was increased. The doubling of CS/DS corresponds well with the increased amount of decorin and biglycan protein cores in Aspn -/skin. This could reflect a compensatory mechanism triggered by deficiency of one SLRP, as was the case in e.g. fibromodulin knockout mice [7]. Interestingly, HS amount decreased in Aspn -/skin, and was considerably less sulfated, which could impact the cellular response to different growth factors. Overall, these changes in GAG content and structure could reflect a general cellular response to a changed extracellular matrix microenvironment. Our attempts at finding the mechanisms behind this response have so far not been fruitful.
The above-described changes in the ECM phenotype translate into a mechanically tougher skin in Aspn -/mice. The strain-stress curves obtained from biomechanical analyses suggest that the difference between Aspn -/and wild-type skin did not manifest in the initial elastic region where the deformation is reversible, but rather in the plastic region between the yield and the failure point, where sliding and possibly rupture of collagen fibrils lead to irreversible damage. We suggest that a compound effect of slightly thinner collagen fibrils and the large difference in GAG composition could both contribute to this biomechanical phenotype. While perhaps counterintuitive, the correlation of thinner collagen fibrils and higher mechanical strength has been described before, e.g. where decorin inhibited fibril assembly but increased fibril strength [41], and where thinner fibrils in 3D gels conferred increased tensile strength [42], possibly through promoting more interconnected fibrillar network (increased surface-tovolume ratio) [43]. Regarding GAGs' influence on the mechanical properties, several papers argue against that possibility in a number of in vitro or ex vivo studies [41,44,45]. These studies, however, were performed on ligaments, where collagen fibers are largely aligned in parallel and have therefore limited compressive force components. Skin, unlike ligaments, will likely have more compression between fibers even under tensile load. This compression is likely to be influenced by the water-retaining GAGs, and in particular CS/DS. We suggest that the changed GAG composition and structure, as well as the increased amounts of decorin and biglycan, in Aspn -/skin provide an explanation for the altered biomechanical properties.