Movement and joints: effects of overuse on anuran knee tissues

Specimens

A total of 67 specimens of five frog species were analyzed: 10 juveniles and 10 adults of Leptodactylus mystacinus Burmeister 1861, seven juveniles and 12 adults of Rhinella arenarum Hensel 1867, seven adults of Melanophryniscus rubriventris Vellard 1947, six juveniles and five adults of L. latinasus Jiménez de la Espada 1875 and four juveniles and six adults of Phyllomedusa sauvagii Boulenger 1882. The juveniles were recognized as individuals who have completed the metamorphosis, because they exhibit traits that indicate the completion of the metamorphosis (See Gosner, 1960, characters of the mouth and complete tail reabsorption), but they have not reached the adulthood size or sexual maturity indicated by secondary sexual characteristics. Thus, assessment of sexual maturity and identification of adult males were based on the presence of secondary sexual characters (e.g., colored vocal sacs, nuptial excrescences); sexual maturity of females was based on examination of the gonads. Previous experimental studies were performed with laboratory animals (Vilarta & De Campos Vidal, 1989; Ni et al., 2015; Nagai et al., 2016), which allows the use of a high number of specimens. However, laboratory animal’s exhibit restricted movements during their lives due to generally being confined to a box. In the present study, we used animals collected in the field, thus allowing us to assume that their histomorphology was determined by their normal conditions of mobility. The disadvantage is that the number of collected specimens is restricted, thus deriving in a small sample size. Specimens were collected during summer in Tucumán (Res. No.13–16), Salta (Res. No. 0308/14) and Jujuy (Res. No. 21/2012) provinces, Argentina. They were housed at Instituto de Herpetología of the Fundación Miguel Lillo, in individual terrariums (30 × 20 × 25 cm), where they moved freely, and under laboratory controlled conditions (temperature 24–29 °C). They were fed ad libitum with living insects (ants, crickets, cockroaches and worms). All the specimens were healthy and without signs of previous injuries. Animals were weighed with a digital scale (±0.01 gr.; Cen-Tech, San Diego, CA, USA) and sized snout-vent length (SVL) with a digital caliper (±0.01 mm.; CD-30C and CD-15B; Mitutoyo, Kanagawa, Japan) before and after the experiments. Since both SVL and weight were similar before and after the trials (±6 mm), we inform only the initial data (Table 1). Specimens of each species with both weight and SVL similar to those used for experiments were selected as control.

Experimental design

To observe the effect of the mechanical stress provoked by overuse of knee tissues, trials were performed on a treadmill belt, following the current protocols designed to this end (Kovanen, Suominen & Peltonen, 1987; Birch et al., 1999; Ni et al., 2015; Gao et al., 2017; Thampatty & Wang, 2018). We defined “overuse” as the excessive use of the joint when the frog is “over-stimulated” to move. In nature, they often stay still (Reilly et al., 2015) unless they need to move to escape from predators, find food, defend territories or find couple to mate (Nauwelaerts, Stamhuis & Aerts, 2005). The treadmill belt is one meter long and exhibits a flat surface. It is covered with a transparent polycarbonate box to prevent the escape of animals, while its lateral wall is covered with scaling paper (Fig. 1). The specimens were kept in captivity for 2 days before the performance trials. Specimens were stimulated to jump (jumper species: L. latinasus and L. mystacinus, Jorgensen & Reilly, 2013; Fabrezi et al., 2014; specie: R. arenarum; Abdala et al., 2018) or walk (walker species: M. rubriventris, P. sauvagii, Manzano et al., 2013; Fabrezi et al., 2014) by the contact with an elastic band crossed on the treadmill belt (Fig. 1), avoiding the stress that human contact could generate. A total of 34 specimens were arbitrarily chosen to perform the trials and the remaining specimens were used as control (i.e., did not undergo the exercise routine, Table 1). Trials were performed twice a day, during up to 10 min unless the frog reached the fatigue earlier. Fatigue was defined as the failure to maintain force of a muscle that has been under load, and that is relieved by rest (Johnson et al., 1996). During muscular fatigue, a cascade of physiological mechanisms occurs (Gibson & Edwards, 1985), provoking muscular pain and dyspnea, which are the principal reasons to stop motion (Güell, Casan & Giménez, 1996). After a series of previous trials, the velocity of the treadmill belt was settled at 2.73 cm/sec (±0.25 cm/sec), being this the maximum value allowing normal movement of specimens. These preliminary trials showed that velocity could be kept constant for all the animals. Juveniles and adults were exercised at the same velocity allowing to evaluate the effect of the same mechanical stimulus over the joint at different stages. The time and distance of the trials are detailed in Table 1. Experimental and control animals were sacrificed with xylocaine viscous (Lidocaine Hydrochloride 10%), fixed in a 10% formaldehyde solution for 24 h, preserved in alcohol 70%. Selected species for our study are composed by organisms whose life history is subject to r selection, often referred to as r-strategists. They inhabit temporary ponds, which are unstable and unpredictable environments. They have ability to reproduce quickly, produce many offspring, each of which has a relatively low probability of surviving to adulthood. In the case of M. rubriventris, that left not so many offspring, we used fewer specimens. The actual population trends of the species used for this study are LESS CONCERN according to the IUCN red list of threatened species (www.iucnredlist.org, version 2017-3). Apart from softly inducing them to move twice a day, specimens in captivity were maintained clean, healthy, with enough water and food (according to the Amphibian Husbandry Resource Guide: Poole & Grow (2012)). In addition, we pay attention to ARRIVE guidelines and to Guidelines for Ethical Conduct in the Care and Use of Nonhuman Animals in Research (CARE). Experiments were approved by the Ethical Committee of Facultad de Medicina, Universidad Nacional de Tucumán (Res. No. 81962-2014).

Histological analysis

A total of 67 knees, 34 from treated and 33 from control specimens corresponding both to juveniles (n = 27) and adults (n = 40) (Table 1 and Supplemental Material S1) were extracted and decalcified with a 50% citrate sodium-50% formic acid solution. Samples were immersed in sodium sulfate for 24 h, and then immersed in a mixture of glycerin and acetic acid for 48 h. The material was afterward dehydrated in a graded ethanol series and embedded in Histoplast embedding medium. Seven μm thick serial sagittal sections were cut with a rotary microtome (Microm HM 325; ThermoFisher Scientific, Waltham, MA, USA) and stained with hematoxiline-eosine and Masson trichrome, the latter allowing to identify collagen fibers. Histological samples were observed under an optical microscope (ICC 50 HD; Leica, Wetzlar, Germany) and photographed with a Nikon Coolpix P6000 digital camera for the diagnosis. The focus was put on the following tissues that integrate the knee-joint: tendons, fibrocartilage, menisci and articular cartilage. Additionally, the growth zone in the diaphysis was examined.

Tissue alterations score

To assess the effect of the experiments on the connective tissues, five parameters were considered: (i) collagen fibers of the fibrocartilage (ii) collagen fibers of the tendons (iii) roundness of the nuclei of the fibrocartilage (iv) arrangement of the collagen fibers of the menisci and (v) hypertrophic chondrocytes. The structural changes observed were categorized in a scoring system. Histological Scoring is a technique widely used in orthopedic research and clinical veterinary (Movin et al., 1997; O’Driscoll et al., 2001; Pritzker et al., 2006; Maffulli et al., 2008; Pauli et al., 2011). It is commonly used to show structural qualitative changes of tissue owed to any factor (Movin et al., 1997; Ameye et al., 2002; Pritzker et al., 2006; Maffulli et al., 2008; Pauli et al., 2011). In this study, two levels of structural or morphological changes were identified, namely, Score 1 for slight changes and Score 2 for severe changes. Score 0 was assigned to those tissues that showed a normal morphology. The tissue scoring is described in Table 2. This grading system is arbitrary and it does not represent fixed stages, however, it is a simple and direct, way to represent the effect of the experiments (Pritzker et al., 2006). Since it is as accurately as possible, is one of the most used systems in this context (see quotations above). Tissues were considered normal following Carvalho (1995) and Franchi et al. (2007) for tendons; Benjamin et al. (1991), Benjamin, Qin & Ralphs (1995), Benjamin & Ralphs (1998) and Carvalho & Felisbino (1999) for fibrocartilages; Pauli et al. (2011) and Senan et al. (2011) for menisci and Pacifici et al. (1990) for hypertrophic chondrocytes. For an overview of the alteration state of each specimen, scores of each trait were summed up. Specimens with Score 10 were those whose connective tissues exhibited the highest levels of abnormality.

The association of the score of each trait with the treatment, the locomotor mode, the species identity and the stage of the individual were assessed by using multinomial ordinal logistic regression. The multinomial ordinal logistic regression is used for describing and testing hypotheses about associations between an ordered categorical variable (i.e., the alteration tissues) and one or more categorical or continuous predictor variables (i.e., treatment, locomotor mode, stages and species) to predict the probability of occurrence of each category (Peng, Lee & Ingersoll, 2002). In the case of the shape of the hypertrophic chondrocyte where only two classes were observed (0 and 2) we used a simple logistic regression that is adequate to model binomial responses. In all the cases we proposed a set of biologically sensitive models (e.g., lineal combinations of predictors) and we compared to what extent they were supported by our data. The set of competing models were increasingly complex; they included an intercept model, which assumes that the probability of observing any category is identical for all the individuals, a model that only considered the treatment, different lineal combinations of the treatment with the stage, the species and the locomotor mode and the interaction between treatment and the other predictors. Locomotor mode and species were not included in the same model due to their nestedness. The models were compared using their Akaike information criterion (AIC; Symonds & Moussalli, 2011). The AIC simultaneously evaluates the level of adjustment of a model (maximum likelihood) with the number of parameters (i.e., an indicator of the complexity of the model). The AIC is not informative by itself and only has utility for compare different models. ΔAIC (delta of AIC) is used as a measurement of the distance of all the models that explain a variable with respect to the model with the best AIC. Due to the nature of our data, we used an AIC corrected by small samples (Burnham & Anderson, 2002). The best logistic models were identified by minimum AIC (Burnham, Anderson & Huyvaert, 2011). All the statistics analyses were performed with the R studio software (version 0.99.903, 2016).

Menisci cell quantification

To analyze the density of cells in the menisci, an area of 100 μm² of the knees of 58 specimens (30 treated and 28 control specimens) was selected (Fig. 2). The number of cells was quantified with ImageJ software and the density was calculated (ρ = no of cells/area). The normality and homoscedasticity of the data were tested with Shapiro–Wilk and Levene’s tests. Differences in the density of cells between control and treated groups were calculated with a Mann–Whitney U-test using R studio software (version 0.99.903, 2016).

Articular cartilage

To analyze the density of cells in the superficial and tangential zones of the articular cartilage, the knees of 25 (13 treated and 12 control) juvenile specimens were used. Adult epiphyses were not analyzed since their articular cartilage was not visible in our samples. The zones of the epiphyses of the femur and tibia-fibula were divided in three areas: internal, medial and external (Fig. 2). The number of chondrocytes was counted in each region with ImageJ software, and the density was derived (ρ = no of chondrocytes/area). The normality and homoscedasticity of the data were tested with Shapiro–Wilk and Levene’s tests. Differences in the density of cells between control and treated groups were calculated with a Mann–Whitney U-test. A posteriori Kruskal–Wallis test was made to assess for differences among zones and areas. R studio software (version 0.99.903, 2016) was used for the statistical analyses.

Microanatomy of the knee-joint

The normal knee-joint of an anuran juvenile specimen (SVL 22.94 mm) is formed by the joint capsule, the menisci, the epiphyses of the femur and the tibia-fibula, muscles, ligaments and tendons (Fig. 3A). The joint capsule consists of fibrous and dense connective tissue, that is, fibrocartilage and tendons. The tendons exhibit parallel collagen fibers with abundant and round nuclei (Fig. 3B). A spindle-shape fibrocartilage is present in the external surface of the knee over the tibia-fibula epiphysis (Fig. 3A). The cells of the fibrocartilage present spherical nuclei, usually arranged in rows, and collagen fibers arranged in parallel (Fig. 3B). The meniscus is present over and between the two epiphyses; the zones of attachment with the epiphyses (enthesis) are usually fibrocartilaginous (Fig. 3C). The meniscus is fibrocartilaginous, with collagen fibers usually packed or showing a more disordered pattern. The nuclei of the collagen fibers of the meniscus are dispersed or arranged in rows (Fig. 3C). The epiphyses of the femur and the tibia-fibula are cartilaginous and covered by the articular (hyaline) cartilage. The chondrocytes of the articular cartilage are isolated or disposed in isogenous groups. The irrigated osteochondral ligaments are located between the lateral articular cartilage and the periosteal bone of the diaphysis (Fig. 3D). Internally, a cartilaginous graciella sesamoid is found between the femur and the tibia-fibula (Fig. 3E). The m. gracilis major (Fig. 3E), m. extensor cruris brevis, m. vastus internus and m. gastrocnemius are mature at this stage.

The normal knee of an adult specimen of L. latinasus (SVL 30.86 mm; Fig. 4) presents a joint capsule with a big fibrocartilage over the surfaces of the tibia-fibula and of the femur (Fig. 4A). The cells of the adult fibrocartilage show round nuclei in rows, and parallel collagen fibers (Fig. 4B), more ordered than in juveniles. Tendons are mature tissues, as evidenced by the parallels fibers and flat nuclei. The menisci are thicker than in juveniles and show areas with packed and laxer collagen fibers (Fig. 4C). The graciella sesamoid is cartilaginous with a center of endochondral ossification. The m. gracilis major is mature (Fig. 4D). The epiphyses are ossified, evidenced by the wide medular cavity with endochondral trabeculae, osteocytes, osteoclasts and blood vessels (Fig. 4E).

Tissue Alterations Scores

Scores are presented in Fig. 5 and a global overview of the altered tissues in juveniles and adults are showed in Fig. 6. The first column in Fig. 5 corresponds to the control specimens and the two right columns correspond to the overuse trials scores.

Normal fibrocartilages were composed by parallel packed collagen fibers between cells (Score 0, Fig. 5A). In treated specimens, there was a gradual separation of the collagen fibers, from a loose (Score 1, Fig. 5B) to a laxer arrangement (Score 2, Fig. 5C). Normal fibrocartilage cells showed round nuclei (Score 0, Fig. 5D). These exhibited a change to oval (Score 1, Fig. 5E) or flattening shape (Score 2, Fig. 5F). Tendons were formed by tightly packed collagen fibers (Score 0, Fig. 5G). Such as in fibrocartilages, tendons showed a gradual disarrangement of their collagen fibers from a loose (Score 1, Fig. 5H) to a very loose pattern (Score 2, Fig. 5I). Normal menisci showed a loose pattern with collagen fibers separated from each other (Score 0, Fig. 5J). The menisci of treated specimens showed a slight (Score 1, Fig. 5K) or a maximum packing (Score 2, Fig. 5E) of their collagen fibers. The hypertrophic chondrocytes of the growth area diaphyses of both femur and tibia-fibula, normally showed an oval or round shape (Score 0, Fig. 5M). The cells exhibited a drastic change of shape, adopting a flattened shape (Score 2, Fig. 5O). This last feature was observed only in juvenile specimens because in adults the diaphyses are already ossified.

In the joint capsule of experimental juveniles, the cells of the fibrocartilages were very affected (Fig. 7B), and showed a flat or oval nuclei (Table 3). The fibers of the tendon showed a very lax arrangement (Score 2, Fig. 7C) or a slight disarrangement (Score 1, Fig. 7C; Table 3). The menisci fibers presented a packed arrangement (Score 2, Fig. 7D). The hypertrophic chondrocytes of the growth area of the diaphyses showed a severe flattening (Score 2, Fig. 7E; Table 3).

Among adults, the fibrocartilaginous tissue showed more extreme changes (Fig. 7; Table 3). There was a noticeable separation of the collagen fibers of the fibrocartilage (Score 2, Fig. 7F) and the nuclei of the fibrocartilage cells were elongated (Score 2, Fig. 7G; Table 3). Tendons also showed a slight abnormal arrangement with a loose pattern of the collagen fibril (Score 1, Fig. 7H).

The highest total sums of scores were 5 and 7 for one specimen of L. mystacinus and one adult specimen of L. latinasus, respectively. Both species have a jumper locomotion mode. The walker species also show tissue alterations, but the sum of their scores was 4 in a juvenile and an adult specimen of P. sauvagii and 4 in adults of M. rubriventris. A juvenile of R. arenarum have a total score of 4 and an adult a total score of 6. Scores distribution among species is detailed in Table 3 and Fig. 7.

In all the analyses the models that included the treatment outperformed the intercept model (more than two units of difference between their AIC). Some of the better models included also the stage (in fibrocartilage collagen fibers), the species (in fibrocartilage nuclei and collagen fiber of the tendon), and the stage and species (in menisci fibers and hypertrophic chondrocytes) (Table 4 and Supplemental Material S2). Neither the locomotor mode nor the interactions between explanatory variables were included in the winning models in any of the analyses. The probabilities estimated through the best models are detailed in Table 5. The probability of finding an alteration in the arrangement of the fibrocartilage fibers (Score 1 or 2) increase after the treatment in juveniles and adults. The latter stage doubles the probability of presenting the Score 2, from 0.07 to 0.14 (Table 5). The probability of showing a shape change in the nuclei of this tissue varies among the species and the treatment. The probability of showing the highest alteration (Score 2) triples in L. latinasus, P. sauvagii and M. rubriventris. No differences in the probability were found between the stages. Regarding tendons, the probability of showing alteration after the treatment also varies between species, in L. latinasus and L. mystacinus the probabilities of showing the Score 1 doubles, and in R. arenarum the probability of showing the Score 2, triples. For menisci fibers arrangement there is a slight effect of the treatment and a strong effect of the stage (e.g., the probability of observing Score 2 in L. mystacinus doubles between stages, from 0.18 to 0.42). For hypertrophic chondrocytes shape the probability of present alteration after the treatment varies also between stages and species. It is higher in juveniles of L. mystacinus and adults of L. latinasus and P. sauvagii (Table 5).

Menisci cell quantification

The density of cells in the menisci was similar between experimental (i.e., over-exercised) and control individuals, both in juveniles (U = 136; p = 0.43) and adults (U = 280.5; p = 0.98, Fig. 8). Both in control and experimental specimens, the density of cells in the menisci was lower in adults than in juveniles.

Articular cartilage cell quantification

In the articular cartilage of juvenile stages, there were no significant differences in the density of chondrocytes between control and experimental (over exercised) individuals, neither in the femur (U = 2.578; p = 0.38) nor in the tibia-fibula (U = 3.269, p = 0.08) (Table 3).