Collagen-hydroxyapatite composites for hard tissue repair

Collagen-hydroxyapatite composites for hard tissue repair

(Parte 1 de 5)

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue RepairEuropean Cells and Materials Vol. 1. 2006 (pages 43-56) ISSN 1473-2262

Abstract

Bone is the most implanted tissue after blood. The major solid components of human bone are collagen (a natural polymer, also found in skin and tendons) and a substituted hydroxyapatite (a natural ceramic, also found in teeth). Although these two components when used separately provide a relatively successful mean of augmenting bone growth, the composite of the two natural materials exceeds this success. This paper provides a review of the most common routes to the fabrication of collagen (Col) and hydroxyapatite (HA) composites for bone analogues. The regeneration of diseased or fractured bones is the challenge faced by current technologies in tissue engineering. Hydroxyapatite and collagen composites (Col-HA) have the potential in mimicking and replacing skeletal bones. Both in vivo and in vitro studies show the importance of collagen type, mineralisation conditions, porosity, manufacturing conditions and crosslinking. The results outlined on mechanical properties, cell culturing and denovo bone growth of these devices relate to the efficiency of these to be used as future bone implants. Solid free form fabrication where a mould can be built up layer by layer, providing shape and internal vascularisation may provide an improved method of creating composite structures.

Keywords: Collagen Type I, hydroxyapatite, composite scaffolds, biocompatible devices, bone substitute, tissue engineering

*Address for correspondence: Denys A Wahl, Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK E-mail: denys.wahl@materials.ox.ac.uk

Introduction

Bone tissue repair accounts for approximately 500,0 surgical procedures per year in the United States alone (Geiger et al., 2003). Angiogenesis, osteogenesis and chronic wound healing are all natural repair mechanisms that occur in the human body. However, there are some critical sized defects above which these tissues will not regenerate themselves and need clinical repair. The size of the critical defect in bones is believed to increase with animal size and is dependent on the concentration of growth factors (Arnold, 2001). In vivo studies on pig sinus (Rimondini et al., 2005) and rabbit femoral condyles (Rupprecht et al., 2003) critical size defects of 6x10mm and 15x25mm respectively were measured. These defects can arise from congenital deformities, trauma or tumour resection, or degenerative diseases such as osteomyelitis (Geiger et al., 2003). Bone substitutes allow repair mechanisms to take place, by providing a permanent or ideally temporary porous device (scaffold) that reduces the size of the defect which needs to be mended (Kohn, 1996). The interest in temporary substitutes is that they permit a mechanical support until the tissue has regenerated and remodelled itself naturally. Furthermore, they can be seeded with specific cells and signalling molecules in order to maximise tissue growth and the rate of degradation and absorption of these implants by the body can be controlled.

Bioresorbable materials have the potential to get round the issues that occur with metallic implants, such as strain shielding and corrosion. Titanium particles produced from wear of hip implants, were shown to suppress osteogenic differentiation of human bone marrow and stroma-derived mesenchymal cells, and to inhibit extra cellular matrix mineralisation (Wang et al., 2003). Furthermore, these materials should help to reduce the problems of graft rejection and drug therapy costs, associated with for example the use of immunosuppressants (e.g. FK506) after implantation of bone grafts (Kaihara et al., 2002).

When using a biodegradable material for tissue repair the biocompatibility and/or toxicity of both the material itself and the by-product of its degradation and subsequent metabolites all need to be considered. Further, at the site of injury, the implant will be subjected to local stresses and strains. Thus, the mechanical properties of the implant, such as tensile, shear and compressive strength, Young’s modulus and fracture toughness need to be taken into consideration when selecting an appropriate material. However, given a bone analogue is ideally resorbable, these properties are not as important as for an inert implant which does not (intentionally) degrade. It is important for the bioresorbable material to be osteoconductive and osteoinductive, to guide and to encourage de novo tissue formation. The current aim of the biological implant is to be indistinguishable from the surrounding host bone

DA Wahl and JT Czernuszka Department of Materials, University of Oxford, Parks Road, Oxford, OX1 3PH, UK

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue Repair

(Geiger et al., 2003). It is self evident that creating new tissue will lead to the best outcome for the patient in terms of quality of life and function of the surrounding tissue.

Synthetic polymers are widely used in biomaterial applications. Examples in tissue engineering include aliphatic polyesters [polyglycolic acid (PGA) and poly- L-lactic acid (PLLA)], their copolymers [polylactic-coglycolic acid (PLGA)] and polycaprolactone (PCL). However, the chemicals (additives, traces of catalysts, inhibitors) or monomers (glycolic acid, lactic acid) released from polymer degradation may induce local and systemic host reactions that cause clinical complications. As an example, lactic acid (the by-product of PLA degradation) was found to create an adverse cellular response at the implant site by reducing the local pH, in which human synovial fibroblasts and murine macrophages released prostaglandin (PGE2), a bone resorbing and inflammatory mediator (Dawes and Rushton, 1994). Nevertheless, a potential way to stabilise the pH is by the addition of carbonate to the implant (Wiesmann et al., 2004). Some polymeric porous devices also have the disadvantages of not withstanding crosslinking treatments such as dehydrothermal treatment (DHT) and ultraviolet (UV) irradiation (Chen et al., 2001). The drawback of requiring chemical crosslinking (glutaraldehyde) is the formation and retention of potential toxic residues making these techniques less desirable for implantable devices (Hennink and van Nostrum, 2002). The reader is referred to Athanasiou et al. (1996) for a review in the biocompatibility of such polymeric materials.

Ceramics [eg HA, tricalcium phosphate (TCP) and/or coral] have been suggested for bone regeneration. Bone substitutes from these materials are both biocompatible and osteoconductive, as they are made from a similar material to the inorganic substituted hydroxyapatite of bone. However, the ceramic is brittle (Kc<1MPam-1) (Dewith et al., 1981) and does not match alone the mechanical properties of cortical bone (Kc~2-12MPam-1) (Bonfield, 1984). Therefore, calcium phosphates have been used in areas of relatively low tensile stress such as bone/ dental fillings or as coatings on implanted devices (Vallet- Regi and Gonzalez-Calbet, 2004).

Collagen, as a natural polymer, is increasingly being used as a device material in tissue engineering and repair. It is, for example, found in bone (Type I), cartilage (Type I) and in blood vessel walls (Type II) and has excellent biocompatible properties. Collagen is easily degraded and resorbed by the body and allows good attachment to cells. However, its mechanical properties are relatively low (E ~100 MPa) in comparison to bone [E ~2-50GPa (Clarke et al., 1993)] and it is therefore highly crosslinked or found in composites, such as collagen-glycoaminoglycans for skin regeneration (O’Brien et al., 2004), or collagenhydroxyapatite for bone remodelling (Kikuchi et al., 2004b). Both collagen and hydroxyapatite devices significantly inhibited the growth of bacterial pathogens, the most frequent cause of prosthesis-related infection, compared to PLGA devices (Carlson et al., 2004).

Other approaches of bone repair have been to use autografts, allografts and xenografts. Although very successful in many operations, autografts have the disadvantages of insufficient supply and morbidity, as well as increasing surgery times and donor site pain (Uemura et al., 2003). Allografts and xenografts are associated with infection and inflammation and have perceived ethical disadvantages. Xenografts also carry the risk of speciesto-species transmissible diseases (Meyer et al., 2004).

Careful consideration of the bone type and mechanical properties are needed for bone substitutes. Indeed, in high load-bearing bones such as the femur, the stiffness of the implant needs to be adequate, not too stiff to result in strainshielding, but stiff enough to present stability. However, in relatively low load-bearing applications such as cranial bone repairs, it is more important to have stability and the correct three dimensional shapes for aesthetic reasons. One of many approaches of tissue engineering is to create a device of similar mechanical and biological properties to the one of the substituted tissue. Therefore, this review will focus on the engineering of a bone substitute, from the understanding of the individual and main components of bone to the creation of a collagen-HA composite. It will bring forward the idea that the manufacturing process of such biocompatible device defines its final microstructure, which in turn will determine its mechanical and biological response, and therefore its efficiency in repairing a hard tissue defect.

A Composite of Collagen and Hydroxyapatite Skeletal bones comprise mainly of collagen (predominantly type I) and carbonate substituted hydroxyapatite, both osteoconductive components. Thus, an implant manufactured from such components is likely to behave similarly, and to be of more use than a monolithic device. Indeed, both collagen type I and hydroxyapatite were found to enhance osteoblast differentiation (Xie et al., 2004), but combined together, they were shown to accelerate osteogenesis. A composite matrix when embedded with human-like osteoblast cells, showed better osteoconductive properties compared to monolithic HA and produced calcification of identical bone matrix (Serre et al., 1993; Wang et al., 1995). In addition, Col-HA composites proved to be biocompatible both in humans and in animals (Serre et al., 1993; Scabbia and Trombelli, 2004).

These composites also behaved mechanically in a superior way to the individual components. The ductile properties of collagen help to increase the poor fracture toughness of hydroxyapatites. The addition of a calcium/ phosphate compound to collagen sheets gave higher stability, increased the resistance to three-dimensional swelling compared to the collagen reference (Yamauchi et al. 2004) and enhanced their mechanical ‘wet’ properties (Lawson and Czernuszka, 1998). This happened even when the collagen was highly crosslinked.

Collagenase digestion can represent an in vitro measure of the rate of degradation and resorption of a biological implant. Uncrosslinked collagen and hydroxyapatitecollagen gel beads were analysed by collagenase digestion. HA-containing gel beads were less prone to collagenase and degraded more slowly than collagen gel beads. The

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue Repair

improved resistance of the composite material to degradation was explained by a potential competition of the hydroxyapatite to the collagenase cleavage sites, or by the absorption of some collagenase to the surface of HA (Wu et al., 2004).

The direct comparison of other materials compared with Col-HA composites for bone substitutes have yet to be clearly investigated. However, increasing the biomimetic properties of an implant may reduce the problems of bacterial infections associated with inserting a foreign body (Schierholz and Beuth, 2001). Evidence of the biological advantage compared to artificial polymeric scaffolds have been further demonstrated in cartilage regeneration (Wang et al., 2004). Polymeric scaffolds can take up to 2 years to degrade whilst Col-HA have a more reasonable degradation rate with regards to clinical use of 2 months to a year (Johnson et al., 1996). Furthermore, osteogenic cells adhered better in vitro to collagen surfaces compared to PLLA and PGA implants (El-Amin et al., 2003).

When comparing ceramic scaffolds and ceramic composite scaffolds, it was shown that Col-HA composites performed well compared to single HA or TCP scaffolds (Wang et al., 2004). The addition of collagen to a ceramic structure can provide many additional advantages to surgical applications: shape control, spatial adaptation, increased particle and defect wall adhesion, and the capability to favour clot formation and stabilisation (Scabbia and Trombelli, 2004).

In summary therefore, combining both collagen and hydroxyapatite should provide an advantage over other materials for use in bone tissue repair. However, the manufacturing of such composites must start from an understanding of the individual components.

Collagen The natural polymer collagen that represents the matrix material of bone, teeth and connective tissue can be extracted from animal or human sources. This may involve a decalcification, purification and modification process. This discussion will focus on collagen type I because it is by far the most abundant type used in tissue engineering and its use is widely documented (Friess, 1998).

Collagen type I: extraction from animal or human tissue Skin, bones, tendons, ligaments and cornea all contain collagen type I. The advantage of using tendon or skin is that it eliminates the decalcification process of the bone mineral component. The removal of all calcium phosphate from a calcified tissue can be achieved through immersion in an Ethylenediaminetetracetic acid (EDTA) solution (Clarke et al., 1993). This process can take as long as several weeks depending on the size of the specimen but it does not remove all antigens from the bone.

Collagen can be extracted and purified from animal tissues, such as porcine skin (Kikuchi et al., 2004a) rabbit femur (Clarke et al., 1993) rat and bovine tendon (Hsu et al., 1999; Zhang et al., 2004) as well as ovine (Damink et al., 1996) and human tissue, such as placenta (Hubbell,

2003). The possible use of recombinant human collagen (although more expensive) could be a way of removing concerns of species-to-species transmissible diseases (Olsen et al., 2003). Freeze- and air-dried collagen matrices have been prepared from bovine and equine collagen type I from tendons and the physical and chemical properties have been compared with regards to the potential use in tissue engineering scaffolds. The matrices of different collagen sources (“species”) showed no variations between pore sizes and fibril diameters but equine collagen matrices presented lower swelling ratio and higher collagenase degradation resistance (Angele et al., 2004).

Collagen type I separation and isolation The separation of collagen requires the isolation of the protein from the starting material in a soluble or insoluble form. Soluble collagen can be isolated by either neutral salts (NaCl), dilute acidic solvents (acetic acid, citrate buffer or hydrochloric acid) or by treatment with alkali (sodium hydroxide and sodium sulphate) or enzymes (ficin, pepsin or chymotrypsin) (Friess, 1998; Machado-Silveiro et al., 2004). The addition of neutral salts can decrease protein solubility (salting out) (Martins et al., 1998). The type of solvent required to isolate collagen will depend greatly on the tissue from which it is extracted, the amount of crosslinking present (maturity of the tissue) and whether decalcification is required. Other separation methods include gel electrophoresis (Roveri et al., 2003) (SDSPAGE and/or Western blotting). Collagen is then recuperated usually by centrifugation. Insoluble collagen can be isolated by modifying its structural configuration (mild denaturation agents) and by mechanical fragmentation (Friess, 1998).

Collagen modification and purification Collagen type I can be modified chemically to achieve a polyanionic protein or a purified protein, known as atelocollagen. Polyanionic chemical modification can be achieved by selective hydrolysis of the asparagine (Asn) and glutamine (Gln) side chains of the collagen type I molecule and have the characteristic of having higher carboxyl group content (Bet et al., 2001). Polyanionic collagen type I was found to improve cell adhesion by 1.5 times compared to native collagen type I (Bet et al., 2003).

The purification of collagen is required to eliminate the antigenic components of the protein. These are mainly the telopeptide regions of collagen type I that can be most efficiently treated by enzymatic digestion. Pepsin is a widely used enzyme for the elimination and digestion of this immunogenic peptide (Rovira et al., 1996; Zhang et al., 1996; Hsu et al., 1999; Kikuchi et al., 2004a). As an example, rat tendon collagen type I was extracted and purified in 0.5mg/ml Pepsin in 0.5M acetic acid for 24 hours (Hsu et al., 1999). However, complete immunogenic purification of non-human proteins is difficult, which may result in immune rejection if used in implants. Impure collagen has the potential for xenozoonoses, the microbial transmission from the animal tissue to the human recipient (Cancedda et al., 2003). Furthermore, Wu et al. (2004) reported that pepsin treatment of impure collagen could

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue Repair

result in the narrowing of D-period banding. However, although collagen extracted from animal sources may present a small degree of antigenicity, these are considered widely acceptable for tissue engineering on humans (Friess, 1998). Furthermore, the literature has yet to find any significant evidence on human immunological benefits of deficient-telopeptide collagens (Lynn et al., 2004).

Commercial collagen Native collagen will have passed many extraction, isolation, purification, separation and sterilisation processes before they have been allowed to be used as biomaterial implants. Commercially available collagen type I can come either in insoluble fibril flakes (Sachlos et al., 2003), in suspension (Muschler et al., 1996; Miyamoto et al., 1998; Goissis et al., 2003), sheets or porous matrices (Du et al., 1999; Du et al., 2000). Many researchers use these collagens directly without further processing.

Hydroxyapatite Compound Calcium phosphates are available commercially, as hydroxyapatite extracted from bones or they can be produced wet by the direct precipitation of calcium and phosphate ions.

Commercial Calcium Phosphate Powders Hydroxyapatite powders can be obtained commercially with different crystal sizes. Unfortunately, such products may not be free of impurities. As examples, some commercially available HA particle sizes ranged between 10-40µm, averaged 5.32µm with a Ca/P ratio of 1.62 (Hsu et al., 1999), while other sources had values of 160-200µm with a Ca/P ratio range of 1.6 to 1.69 (Scabbia and Trombelli, 2004). Most manufacturers produce sintered components which differ chemically from the biological carbonate apatites (Okazaki et al., 1990). Sintering of HA (depending on stoichiometry) produces decomposition of the calcium phosphate phases to oxyapatite and possibly, tetracalcium phosphate and tricalcium phosphate (TCP). It has been found that stoichiometric HA is much less biodegradable than substituted HA and TCP (Kocialkowski et al., 1990).

HA extraction from bone Bone powders or hydroxyapatite have been extracted from cortical bovine bone (Rodrigues et al., 2003). The bone was cleaned, soaked in 10% sodium hypochlorite for 24h, rinsed in water and boiled in 5% sodium hydroxide for 3h. It was then incubated in 5% sodium hypochlorite for 6h, washed in water and soaked in 10% hydrogen peroxide for 24h. The material was subsequently sintered at 1100°C and pulverised to the desired particle size (200-400µm). Grains of different crystal size could be separated by sieving. The final stages included sterilisation of the HA particles at 100-150°C.

In vitro Hydroxyapatite (HA) powders Hydroxyapatite can also be precipitated in vitro through the following chemical reactions:

Ammonia was used to keep the solution basic (pH 12) (Sukhodub et al., 2004). Hydroxyapatite precipitates were then extracted by heating the mixture to 80°C for about 10 minutes and incubating them at 37°C for 24h. Bakos et al. (1999) kept the reaction at pH 1, filtered the precipitate, washed it in distilled water and dried the solution at temperatures of 140-160°C. The dried material was then sintered at 1000°C for 2h before being crushed in a mortar. Only HA particles of 40-280µm were used for their composites. Alternatively, by using a different ammonium phosphate as a countercation for the phosphate ligand, non-stoichiometric hydroxyapatite powders have been filtered to an average particle size of 64µm, and then dried at 90°C. The cake is then ground and particles of 60- 100 µm were used for the composites (Martins and Goissis, 2000).

The ceramic compound was synthesised at 60-80°C and at pH 7.4 (Okazaki et al., 1990; Okazaki et al., 1997). The apatite was then extracted by filtration, washed with distilled water and dried at 60-80°C. This method was further used to create FgMgCO3Apatite for composite substitutes (Yamasaki et al., 2003).

In this reaction, chloride and potassium have been used as the counteranions and countercations respectively in order to form hydroxyapatite. As well as creating in vitro hydroxyapatite particles with controlled crystal size, this is a direct route for producing Col-HA composites by directly mineralising collagen substrates (Lawson and Czernuszka, 1998; Zhang et al., 2004). The actual composite method of production will be reviewed later; however, it is important to be aware of differences in ion solutions used for the different experiments. For biomaterials, purity and sterility of all excipients is a key to favourable cellular response. Therefore, reaction 3 is recommended as it does not make use of calcium nitrate and ammonia. The purity of calcium nitrate was found to be directly linked to the purity of the precipitated calcium phosphate, whilst cytotoxic ammonia and its ammonium products are hard to remove (Kweh et al., 1999).

Low temperature methods of HA processing have been proposed to avoid high crystallinity of HA due to sintering at high temperatures, resulting in similar carbonate substituted bone hydroxyapatite. These include colloidal processing, uniaxial and cold isostatic pressing, starch consolidation and a combination of gel casting and foaming (Vallet-Regi and Gonzalez-Calbet, 2004).

The influence of HA properties HA implants or coatings are valuable because they provide a good adhesion to the local tissue due to their surface chemistry and have been shown to enhance osteoblast proliferation and differentiation (Xie et al., 2004). In bone filling applications, bulk material is clinically harder to insert into a complex defect compared to injectable HA particles. Although particles provide an advantage of having a higher surface area, they are hard to manipulate alone and secure at the site of the implant. Therefore, they

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue Repair

have been mixed with biodegradable matrices, such as collagen and PGA (Vallet-Regi and Gonzalez-Calbet, 2004).

The cellular response to HA particles, incorporated into a matrix or as coating, has been shown to depend on properties such as particle size and morphology (needle like, spherical or irregular plates), chemical composition, crystallinity and sintering temperatures. Due to such variability in HA properties, contradictions arise in the literature on the influence of one property over another and a need for a more systematic research was proposed by Laquerriere et al. (2005). However, it is generally accepted that needle shaped particles produce deleterious cellular response compared to spherical shaped particles. Indeed, macrophages have been found to release higher levels of inflammatory mediators and cytokines such as metalloproteinases (MMPs) and Interleukine-6 respectively (Laquerriere et al., 2005).

In the case of collagen-HA implants, the size and crystallinity of the HA particles will have an importance to its stability and inflammatory response. In skeletal bones, carbonate substituted HA crystals are mineralised within small gaps of the collagen fibrils and have been quoted as 50nm×25nm×2-5nm in length, width and thickness respectively (Vallet-Regi and Gonzalez-Calbet, 2004). They provide a local source of calcium to the surrounding cells as well as interacting with collagen fibrils in order to achieve the relatively high mechanical properties of bones. However, small sintered particles of less than 1µm have been cautioned against in bone implants due to their increased inflammation response (Laquerriere et al., 2005) and cell toxicity in vitro (Sun et al., 1997). In contrast, Lawson and Czernuszka (1998) have shown that smaller plate-like particles of the order of 200nm×20nm×5nm produced an enhanced osteoblastic adhesion and proliferation compared to HA particles of an order of magnitude larger. These were carbonate substituted HA particles and produced at physiological temperatures (unsintered).

Collagen-Hydroxyapatite Composite Fabrication This section will summarise the different methods for collagen-hydroxyapatite composite formation. It will include the production methods of composite gels, films, collagen-coated ceramics, ceramic-coated collagen matrices and composite scaffolds for bone substitutes and hard tissue repair.

In vitro collagen mineralisation Direct mineralisation of a collagen substrate involves the use of calcium and phosphate solutions. Collagen can either be a fixed solid film through which calcium and phosphate ions diffuse into the fibrils (Lawson and Czernuszka, 1998), or as a phosphate-containing collagen solution (Kikuchi et al., 2004a), or an acidic calcium-containing collagen solution (Bradt et al., 1999). The advantage of using the first method is that the orientation of the collagen fibres can be controlled (Iijima et al., 1996). Indeed, it has been shown that the c-axis of HA crystals can be made to grow along the direction of collagen fibrils if the right conditions of mineralisation are met. These conditions (pH 8-9 and T =40°C) promote calcium ion accumulation on the carboxyl group of collagen molecules, leading to HA nucleation (Kikuchi et al., 2004a).

The formation of HA is temperature and pH dependent as well as molar dependent (Ca/P ratio). Figure 1 shows an experimental set-up for calcium and phosphate diffusion and apatite crystallisation onto collagen substrates. Undenatured collagen films can be obtained from an acidic collagen suspension by air dehydration at different temperatures (4-37°C) or by applying cold isostatic pressure (200MPa) for 15h (Kikuchi et al., 2001).

Figure 1. An experimental set-up for the direct mineralisation of a collagen sheet (Modified from Lawson and Czernuszka, 1998)

DA Wahl et alCollagen-Hydroxyapatite Composites for Hard Tissue Repair

Thermally-triggered assembly of HA/collagen gels Liposomes have been used as drug delivery system due to their ability to contain water-soluble materials in a phospholipids layer (Ebrahim et al., 2005). Pederson et al. (2003) have combined the direct mineralisation method with the ability of liposomes to exist in a gel for the potential use in injectable composite precursors. They reported a method whereby calcium and phosphate ions where encapsulated within the liposomes and the latter inserted into a collagen acidic suspension. After injection into a skeletal defect, the increase in temperature due to the body heat would start a gelation process, forming a collagen fibrous network where mineralisation occurs after reaching the liposome’s transition temperature (37ºC) (Pederson et al., 2003).

Vacuum infiltration of collagen into a ceramic matrix Multiple tape casting is a method of ceramic scaffold production: an aqueous hydroxyapatite slurry containing polybutylmethalcrylate (PBMA) spheres is heated to high temperatures to burn out the BMPA particles, forming a porous HA green body (Werner et al., 2002). Collagen infiltration was performed under vacuum, and the collagen suspension filled in the gaps of the porous matrix. The final composite was then freeze dried to create microsponges within. Variation of the final product was dependent on process time and flow resistance during filtration (Pompe et al., 2003).

Enzymatic mineralisation of collagen sheets Figure 2 shows a method of enzymatic loading of collagen sheets and the following cycle of mineralisation. The collagen containing an alkaline phosphatase was allowed to be in contact with an aqueous solution of calcium ions and phosphate ester. The enzyme provided a reservoir for

(Parte 1 de 5)

Comentários